emo/procps
1
0
Fork 0
Code Issues Pull Requests Packages Releases Activity
c767adf974
procps/proc/pids.c

1487 lines
53 KiB
C
Raw Normal View History

library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
/*
* pids.c - task/thread/process related declarations for libproc
*
* Copyright (C) 1998-2005 Albert Cahalan
* Copyright (C) 2015 Craig Small <csmall@enc.com.au>
* Copyright (C) 2015 Jim Warner <james.warner@comcast.net>
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* This library is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with this library; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
//efine _GNU_SOURCE // for qsort_r
#include <ctype.h>
#include <errno.h>
#include <fcntl.h>
#include <limits.h>
#include <stdarg.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>
#include <sys/stat.h>
#include <sys/types.h>
library: final tweaks to code and/or comments, 3rd gen With the dust now settling on all those 3rd generation upgrades, this patch tries to provide some consistency among the separate modules involved. Someday we should consider a 4th generation where all redundant code has been removed and isolated in a new shared source file. Following is a summary of significant changes (if any) to each of these now upgraded 3rd gen library modules. <meminfo> ............................................ . strictly formatting/comment changes, code unaffected <pids> ............................................... . replaced a local mkSTR macro with existing STRINGIFY . added fetch narrative explaining duplicate addresses <slabinfo> ........................................... . rearranged some free logic for procps_slabinfo_unref . added fetch narrative explaining duplicate addresses <stat> ............................................... . added #define ENFORCE_LOGICAL, just as in <slabinfo> . replaced a local mkSTR macro with existing STRINGIFY . alphabetized the function declarations in the header <vmstat> ............................................. . made one coverity concession with read_vmstat_failed [ several of these changes may reflect this author's ] [ continuing pursuit of an unreasonable goal -- that ] [ of a 'perfect' (plus 'pretty') C language program! ] Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-09 10:30:00 +05:30
#include <proc/devname.h>
#include <proc/readproc.h>
library: add <pids> TIME_ALL/TIME_ELAPSED needed by ps Shoot, here's yet another bow to ps needs. But it's ok because it makes a lot of sense. Rather than force all users into their own calculations do but it once here. As an aside this need arose during ps testing when the sorts were using TIME_START or TICS_ALL. That was just fine for almost every need except 'etime' plus 'time'. That 'etime' was sorting the opposite of what's wanted when using TIME_START (of course) while 'time' yielded some weird ordering because TICS_ALL was too granular. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-09-30 10:30:00 +05:30
#include <proc/sysinfo.h>
#include <proc/uptime.h>
library: final tweaks to code and/or comments, 3rd gen With the dust now settling on all those 3rd generation upgrades, this patch tries to provide some consistency among the separate modules involved. Someday we should consider a 4th generation where all redundant code has been removed and isolated in a new shared source file. Following is a summary of significant changes (if any) to each of these now upgraded 3rd gen library modules. <meminfo> ............................................ . strictly formatting/comment changes, code unaffected <pids> ............................................... . replaced a local mkSTR macro with existing STRINGIFY . added fetch narrative explaining duplicate addresses <slabinfo> ........................................... . rearranged some free logic for procps_slabinfo_unref . added fetch narrative explaining duplicate addresses <stat> ............................................... . added #define ENFORCE_LOGICAL, just as in <slabinfo> . replaced a local mkSTR macro with existing STRINGIFY . alphabetized the function declarations in the header <vmstat> ............................................. . made one coverity concession with read_vmstat_failed [ several of these changes may reflect this author's ] [ continuing pursuit of an unreasonable goal -- that ] [ of a 'perfect' (plus 'pretty') C language program! ] Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-09 10:30:00 +05:30
#include <proc/wchan.h>
#include <proc/procps-private.h>
#include <proc/pids.h>
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
//#define UNREF_RPTHASH // report on hashing, at uref time
#define FILL_ID_MAX 255 // upper limit for pid/uid fills
library: miscellaneous tweaks to pid code and comments . traded a complex misaligned memory allocation scheme in the make_hist function for a simple aligned scheme. plus memory allocation increases are globally defined. . changed 1 parameter for procps_pids_stacks_sort() to better reflect the 'array of pointers', not an address of a pointer as is used with guys such as 'new/unref'. . the pids_reap struct was changed slightly to make it more reflective of it's actual implementation details. . the Item_table member .mustfree is now .needfree and that .makehist was now made .needhist for consistency. . reduced the number of separate 'return NULL;' source statements in that primary procps_pids_reap() routine. . ensured consistent reference to sizeof(void *) & not occasional reference to sizeof(void*) without a space. . rather than enable/disable validate_stacks via a #if in the function body, it is now handled via a #define. . some comments in the procps_pids_reset function were adjusted to reflect this current implementation. shown originally, they reflected an aborted attempt to avoid a testing aberration not fully understood at the time. . added a summary of the memory overhead cost of HST_t processing to that UNREF_RPTHASH output at unref time. . a 'PIDs at max depth:' portion of that UNREF_RPTHASH enabled #define is now published only when the maximum depth of hash table entry chains exceed depths of one. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-20 10:30:00 +05:30
#define MEMORY_INCR 128 // amt by which allocations grow
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
library: privatize two with <pids> select/fill changes After simplifying that select/fill interface, there is no longer a need for public 'alloc' & 'dealloc' stacks functions. There is now only one instance of stacks as an input parameter found in procps_pids_stacks_sort(). But sorting 'empty' stacks serves no possible purpose. So this commit retains both functions, since they will still be needed, but designates them private (static). Additionally, with their demise we will eliminate that pids_stacks structure from the header file, internally using what always was the true 'stacks_extent' struct. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-01 10:30:00 +05:30
struct stacks_extent {
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
struct pids_stack **stacks;
int ext_numstacks;
struct stacks_extent *next;
};
library: correct a flawed approach for PROCPS_FILL_UID Gosh, just because nobody uses some newlib provision I guess, since it is being offered, it ought to actually be tested at some point. Well, that point just arrived and guess what? A surprise: some bugs were discovered. The procps_pids_select function established a for loop wherein readproc is called until the passed 'maxthese' limit. Unfortunately this was incorrect for 2 reasons: 1. For PROCPS_FILL_PID results are limited by what the oldlib finds, having established the pid list at open. Total found already cannot exceed a passed 'maxthese'; 2. With PROCPS_FILL_UID, returned results could exceed a 'maxthese' thus making the for loop incorrect again. [ plus yours truly neglected to forward the required ] [ UIDs total to our old library, another oops biggie ] In summary: the loop should have been forever, exiting only when all those identified procs had been located. So, while addressing those bugs, I've consolidated all the retrieval code (initialize, iterate, summarize) in a single helper function which will now serve both the procps_pids_reap and select functions. And as a result those guys were reduced to quite trivial housekeeping. This patch, hopefully, completes the normalization for reap/select (fill), which began with references shown. Reference(s): commit 0c953eccc5fe7240be9d272e1b6c0ce8769d8ed2 commit 747dfc5987e6e91ea3a8575de307e2892790c598 Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-08 10:30:00 +05:30
struct fetch_support {
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
struct pids_stack **anchor; // reap/select consolidated extents
library: correct a flawed approach for PROCPS_FILL_UID Gosh, just because nobody uses some newlib provision I guess, since it is being offered, it ought to actually be tested at some point. Well, that point just arrived and guess what? A surprise: some bugs were discovered. The procps_pids_select function established a for loop wherein readproc is called until the passed 'maxthese' limit. Unfortunately this was incorrect for 2 reasons: 1. For PROCPS_FILL_PID results are limited by what the oldlib finds, having established the pid list at open. Total found already cannot exceed a passed 'maxthese'; 2. With PROCPS_FILL_UID, returned results could exceed a 'maxthese' thus making the for loop incorrect again. [ plus yours truly neglected to forward the required ] [ UIDs total to our old library, another oops biggie ] In summary: the loop should have been forever, exiting only when all those identified procs had been located. So, while addressing those bugs, I've consolidated all the retrieval code (initialize, iterate, summarize) in a single helper function which will now serve both the procps_pids_reap and select functions. And as a result those guys were reduced to quite trivial housekeeping. This patch, hopefully, completes the normalization for reap/select (fill), which began with references shown. Reference(s): commit 0c953eccc5fe7240be9d272e1b6c0ce8769d8ed2 commit 747dfc5987e6e91ea3a8575de307e2892790c598 Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-08 10:30:00 +05:30
int n_alloc; // number of above pointers allocated
int n_inuse; // number of above pointers occupied
library: misc tweaks for code and comments, <PIDS> api This commit simply tries to keep naming plus formating conventions on a par with the continuing climb up that learning curve. These changes were suggested following <slabinfo> sources upgrade from 2nd to 3rd generation. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-06 10:30:00 +05:30
int n_alloc_save; // last known results.stacks allocation
struct pids_fetch results; // counts + stacks for return to caller
library: more pids_fetch struct opaqueness, <PIDS> api With all our 3rd generation interfaces, we're now well positioned to preserve binary compatibility should new fields be added to any public structure (assuming that the 'result' union already contains its largest type). This remains true even for the <pids> interface, which unlike the others, has one structure embedded within a separate struct rather than declaring a pointer to it. The counts struct was positioned after the stacks ptrs array so as to preserve that ABI if ever new ints were added. Logically, however, the counts (especially that total) should precede the stacks array if we wished to properly place a horse (total) before a cart (stacks). So to enable relocating those counts we will no longer embed that structure, but provide a pointer to it. And this will make accessing syntax feel more natural too. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-23 10:30:00 +05:30
struct pids_counts counts; // actual counts pointed to by 'results'
library: correct a flawed approach for PROCPS_FILL_UID Gosh, just because nobody uses some newlib provision I guess, since it is being offered, it ought to actually be tested at some point. Well, that point just arrived and guess what? A surprise: some bugs were discovered. The procps_pids_select function established a for loop wherein readproc is called until the passed 'maxthese' limit. Unfortunately this was incorrect for 2 reasons: 1. For PROCPS_FILL_PID results are limited by what the oldlib finds, having established the pid list at open. Total found already cannot exceed a passed 'maxthese'; 2. With PROCPS_FILL_UID, returned results could exceed a 'maxthese' thus making the for loop incorrect again. [ plus yours truly neglected to forward the required ] [ UIDs total to our old library, another oops biggie ] In summary: the loop should have been forever, exiting only when all those identified procs had been located. So, while addressing those bugs, I've consolidated all the retrieval code (initialize, iterate, summarize) in a single helper function which will now serve both the procps_pids_reap and select functions. And as a result those guys were reduced to quite trivial housekeeping. This patch, hopefully, completes the normalization for reap/select (fill), which began with references shown. Reference(s): commit 0c953eccc5fe7240be9d272e1b6c0ce8769d8ed2 commit 747dfc5987e6e91ea3a8575de307e2892790c598 Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-08 10:30:00 +05:30
};
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
struct procps_pidsinfo {
int refcount;
library: fix boo-boo left by earlier patch, <PIDS> api After the commit referenced below the potential exists for a SEGV (resulting from an out-of-bounds Item_table reference if PROCPS_PIDS_physical_end is encountered). So this patch eliminates that PROCPS_PIDS_physical_end as no longer necessary and completes the task of using PROCPS_PIDS_logical_end as a sole necessary fencepost. Reference(s): commit e7585992d9c0743246247b3d6ee0041942fe07d5
2016-04-25 10:30:00 +05:30
int maxitems; // includes 'logical_end' delimiter
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
int curitems; // includes 'logical_end' delimiter
library: fix boo-boo left by earlier patch, <PIDS> api After the commit referenced below the potential exists for a SEGV (resulting from an out-of-bounds Item_table reference if PROCPS_PIDS_physical_end is encountered). So this patch eliminates that PROCPS_PIDS_physical_end as no longer necessary and completes the task of using PROCPS_PIDS_logical_end as a sole necessary fencepost. Reference(s): commit e7585992d9c0743246247b3d6ee0041942fe07d5
2016-04-25 10:30:00 +05:30
enum pids_item *items; // includes 'logical_end' delimiter
library: enable an 'invariant' stack for that pids API After peeking at possible conversion of the ps program it appeared that we may ultimately need the concept of a 'static' pids_stack in suport of look_up_our_self(). In other words, one not impacted by procps_pids_reset. This patch simply sets the stage for that possibility. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-09-10 10:30:00 +05:30
struct stacks_extent *extents; // anchor for all resettable extents
library: introduce a fatal 'proc not mounted' function Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-09-11 10:30:00 +05:30
struct stacks_extent *otherexts; // anchor for single stack invariant extents
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
struct fetch_support fetch; // support for procps_pids_reap & select
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
int history_yes; // need historical data
struct history_info *hist; // pointer to historical support data
int dirty_stacks; // extents need dynamic storage clean
library: add sequential access to that pids collection To ease the transition to the new interface, for other than that top program, individual read provisions have been added to the <proc/pids.h> API. This represents a refinement of a position stated in a post noted below. Reference(s): http://www.freelists.org/post/procps/newlib-ps-fix Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-26 10:30:00 +05:30
proc_t*(*read_something)(PROCTAB*, proc_t*); // readproc/readeither via which
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
unsigned pgs2k_shift; // to convert some proc vaules
library: misc tweaks for code and comments, <PIDS> api This commit simply tries to keep naming plus formating conventions on a par with the continuing climb up that learning curve. These changes were suggested following <slabinfo> sources upgrade from 2nd to 3rd generation. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-06 10:30:00 +05:30
unsigned oldflags; // the old library PROC_FILL flagss
library: a couple of miscellaneous changes, <PIDS> api This patch represents the following minor adjustments: . old PROCTAB used with select & reap more descriptive . reformatted PROCPS_PIDS_VAL macro like other 3rd gen Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-18 10:30:00 +05:30
PROCTAB *fetch_PT; // oldlib interface for 'select' & 'reap'
library: add <pids> TIME_ALL/TIME_ELAPSED needed by ps Shoot, here's yet another bow to ps needs. But it's ok because it makes a lot of sense. Rather than force all users into their own calculations do but it once here. As an aside this need arose during ps testing when the sorts were using TIME_START or TICS_ALL. That was just fine for almost every need except 'etime' plus 'time'. That 'etime' was sorting the opposite of what's wanted when using TIME_START (of course) while 'time' yielded some weird ordering because TICS_ALL was too granular. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-09-30 10:30:00 +05:30
unsigned long hertz; // for TIME_ALL & TIME_ELAPSED calculations
unsigned long long boot_seconds; // for TIME_ELAPSED calculation
library: a couple of miscellaneous changes, <PIDS> api This patch represents the following minor adjustments: . old PROCTAB used with select & reap more descriptive . reformatted PROCPS_PIDS_VAL macro like other 3rd gen Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-18 10:30:00 +05:30
PROCTAB *get_PT; // oldlib interface for active 'get'
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
struct stacks_extent *get_ext; // an extent used for active 'get'
enum pids_fetch_type get_type; // last known type of 'get' request
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
};
// ___ Results 'Set' Support ||||||||||||||||||||||||||||||||||||||||||||||||||
library: eliminated a final potential NULL, <PIDS> api In that reference below a specific systemd problem was fixed in the commit shown. However lurking deep within the <pids> interface was yet one final case where NULL could be returned, involving 'strv' and the following: . a user requested both a single string vector (always returned as a normal string) & the vectorized version, as with PROCPS_PIDS_CMDLINE and PROCPS_PIDS_CMDLINE_V. . a user simply duplicated some vectorized enum items. The root of that NULL problem is the fact those single string vectors shared the same proc_t field with their true vectorized version. So while multiple occurrences for most strings could be satisfied with strdup versus the normal ownership usurpation, those true vectorized fields could not be quite so easily copied/duplicated. Thus newlib chose to return a NULL result.strv pointer under either of the above scenarios (which perhaps was just a user boo-boo in the first place). In any event, the NULL was a potential for true string vectors only. Now, since newlib is the sole caller into the readproc module, separate fields have been created for what are just normal strings (never vectorized) and those which remain the true vectorized versions. And, former flags which only worked if combined, now act as stand alone. Thus, both PROCPS_PIDS_CMDLINE & PROCPS_PIDS_CMDLINE_V can be used simultaneously (as they should have been). Also with this patch, items which a user duplicates in the stack (beyond the first such item) will return the the string "[ duplicate ENUM_ID ]". This practice will apply to both single strings and true vectorized ones. In addition to informing users of their error, it will also mean potential NULLs need now never be a concern. Reference(s); http://www.freelists.org/post/procps/systemd-binary-vs-library commit 0580a7b4c67d0297629d37281b4f690894429626 Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-19 10:30:00 +05:30
static char** vectorize_this (const char* src) {
#define pSZ (sizeof(char*))
char *cpy, **vec;
int adj, tot;
tot = strlen(src) + 1; // prep for our vectors
adj = (pSZ-1) - ((tot + pSZ-1) & (pSZ-1)); // calc alignment bytes
cpy = calloc(1, tot + adj + (2 * pSZ)); // get new larger buffer
if (!cpy) return NULL; // we no longer use xcalloc
snprintf(cpy, tot, "%s", src); // duplicate their string
vec = (char**)(cpy + tot + adj); // prep pointer to pointers
*vec = cpy; // point 1st vector to string
*(vec+1) = NULL; // null ptr 'list' delimit
return vec; // ==> free(*vec) to dealloc
#undef pSZ
} // end: vectorize_this
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
#define setNAME(e) set_results_ ## e
#define setDECL(e) static void setNAME(e) \
(struct procps_pidsinfo *I, struct pids_result *R, proc_t *P)
library: eliminated a final potential NULL, <PIDS> api In that reference below a specific systemd problem was fixed in the commit shown. However lurking deep within the <pids> interface was yet one final case where NULL could be returned, involving 'strv' and the following: . a user requested both a single string vector (always returned as a normal string) & the vectorized version, as with PROCPS_PIDS_CMDLINE and PROCPS_PIDS_CMDLINE_V. . a user simply duplicated some vectorized enum items. The root of that NULL problem is the fact those single string vectors shared the same proc_t field with their true vectorized version. So while multiple occurrences for most strings could be satisfied with strdup versus the normal ownership usurpation, those true vectorized fields could not be quite so easily copied/duplicated. Thus newlib chose to return a NULL result.strv pointer under either of the above scenarios (which perhaps was just a user boo-boo in the first place). In any event, the NULL was a potential for true string vectors only. Now, since newlib is the sole caller into the readproc module, separate fields have been created for what are just normal strings (never vectorized) and those which remain the true vectorized versions. And, former flags which only worked if combined, now act as stand alone. Thus, both PROCPS_PIDS_CMDLINE & PROCPS_PIDS_CMDLINE_V can be used simultaneously (as they should have been). Also with this patch, items which a user duplicates in the stack (beyond the first such item) will return the the string "[ duplicate ENUM_ID ]". This practice will apply to both single strings and true vectorized ones. In addition to informing users of their error, it will also mean potential NULLs need now never be a concern. Reference(s); http://www.freelists.org/post/procps/systemd-binary-vs-library commit 0580a7b4c67d0297629d37281b4f690894429626 Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-19 10:30:00 +05:30
/* convert pages to kib */
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
#define CVT_set(e,t,x) setDECL(e) { \
library: try to minimize the results types, <PIDS> api This is just beginning an effort to minimize/normalize the sheer variety of results types in use for our API. In taking these first baby steps, a few anomalies were found. There'll no doubt be many more yet to discover. . the _FLT_ (fault) fields were already signed long in the proc_t (even though their sscanf format used %lu). . although strtoul will alway return an unsigned long, all of the _VM_ fields were made signed long just like other memory fields (& signed sorts more efficiently). Reference(s): http://www.freelists.org/post/procps/newlib-drip-drip,4 Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-15 10:30:00 +05:30
R->result. t = (long)(P-> x) << I -> pgs2k_shift; }
library: eliminated a final potential NULL, <PIDS> api In that reference below a specific systemd problem was fixed in the commit shown. However lurking deep within the <pids> interface was yet one final case where NULL could be returned, involving 'strv' and the following: . a user requested both a single string vector (always returned as a normal string) & the vectorized version, as with PROCPS_PIDS_CMDLINE and PROCPS_PIDS_CMDLINE_V. . a user simply duplicated some vectorized enum items. The root of that NULL problem is the fact those single string vectors shared the same proc_t field with their true vectorized version. So while multiple occurrences for most strings could be satisfied with strdup versus the normal ownership usurpation, those true vectorized fields could not be quite so easily copied/duplicated. Thus newlib chose to return a NULL result.strv pointer under either of the above scenarios (which perhaps was just a user boo-boo in the first place). In any event, the NULL was a potential for true string vectors only. Now, since newlib is the sole caller into the readproc module, separate fields have been created for what are just normal strings (never vectorized) and those which remain the true vectorized versions. And, former flags which only worked if combined, now act as stand alone. Thus, both PROCPS_PIDS_CMDLINE & PROCPS_PIDS_CMDLINE_V can be used simultaneously (as they should have been). Also with this patch, items which a user duplicates in the stack (beyond the first such item) will return the the string "[ duplicate ENUM_ID ]". This practice will apply to both single strings and true vectorized ones. In addition to informing users of their error, it will also mean potential NULLs need now never be a concern. Reference(s); http://www.freelists.org/post/procps/systemd-binary-vs-library commit 0580a7b4c67d0297629d37281b4f690894429626 Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-19 10:30:00 +05:30
/* strdup of a static char array */
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
#define DUP_set(e,x) setDECL(e) { \
(void)I; R->result.str = strdup(P-> x); }
library: eliminated a final potential NULL, <PIDS> api In that reference below a specific systemd problem was fixed in the commit shown. However lurking deep within the <pids> interface was yet one final case where NULL could be returned, involving 'strv' and the following: . a user requested both a single string vector (always returned as a normal string) & the vectorized version, as with PROCPS_PIDS_CMDLINE and PROCPS_PIDS_CMDLINE_V. . a user simply duplicated some vectorized enum items. The root of that NULL problem is the fact those single string vectors shared the same proc_t field with their true vectorized version. So while multiple occurrences for most strings could be satisfied with strdup versus the normal ownership usurpation, those true vectorized fields could not be quite so easily copied/duplicated. Thus newlib chose to return a NULL result.strv pointer under either of the above scenarios (which perhaps was just a user boo-boo in the first place). In any event, the NULL was a potential for true string vectors only. Now, since newlib is the sole caller into the readproc module, separate fields have been created for what are just normal strings (never vectorized) and those which remain the true vectorized versions. And, former flags which only worked if combined, now act as stand alone. Thus, both PROCPS_PIDS_CMDLINE & PROCPS_PIDS_CMDLINE_V can be used simultaneously (as they should have been). Also with this patch, items which a user duplicates in the stack (beyond the first such item) will return the the string "[ duplicate ENUM_ID ]". This practice will apply to both single strings and true vectorized ones. In addition to informing users of their error, it will also mean potential NULLs need now never be a concern. Reference(s); http://www.freelists.org/post/procps/systemd-binary-vs-library commit 0580a7b4c67d0297629d37281b4f690894429626 Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-19 10:30:00 +05:30
/* regular assignment copy */
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
#define REG_set(e,t,x) setDECL(e) { \
(void)I; R->result. t = P-> x; }
library: eliminated a final potential NULL, <PIDS> api In that reference below a specific systemd problem was fixed in the commit shown. However lurking deep within the <pids> interface was yet one final case where NULL could be returned, involving 'strv' and the following: . a user requested both a single string vector (always returned as a normal string) & the vectorized version, as with PROCPS_PIDS_CMDLINE and PROCPS_PIDS_CMDLINE_V. . a user simply duplicated some vectorized enum items. The root of that NULL problem is the fact those single string vectors shared the same proc_t field with their true vectorized version. So while multiple occurrences for most strings could be satisfied with strdup versus the normal ownership usurpation, those true vectorized fields could not be quite so easily copied/duplicated. Thus newlib chose to return a NULL result.strv pointer under either of the above scenarios (which perhaps was just a user boo-boo in the first place). In any event, the NULL was a potential for true string vectors only. Now, since newlib is the sole caller into the readproc module, separate fields have been created for what are just normal strings (never vectorized) and those which remain the true vectorized versions. And, former flags which only worked if combined, now act as stand alone. Thus, both PROCPS_PIDS_CMDLINE & PROCPS_PIDS_CMDLINE_V can be used simultaneously (as they should have been). Also with this patch, items which a user duplicates in the stack (beyond the first such item) will return the the string "[ duplicate ENUM_ID ]". This practice will apply to both single strings and true vectorized ones. In addition to informing users of their error, it will also mean potential NULLs need now never be a concern. Reference(s); http://www.freelists.org/post/procps/systemd-binary-vs-library commit 0580a7b4c67d0297629d37281b4f690894429626 Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-19 10:30:00 +05:30
/* take ownership of a normal single string if possible, else return
some sort of hint that they duplicated this char * item ... */
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
#define STR_set(e,x) setDECL(e) { \
library: eliminated a final potential NULL, <PIDS> api In that reference below a specific systemd problem was fixed in the commit shown. However lurking deep within the <pids> interface was yet one final case where NULL could be returned, involving 'strv' and the following: . a user requested both a single string vector (always returned as a normal string) & the vectorized version, as with PROCPS_PIDS_CMDLINE and PROCPS_PIDS_CMDLINE_V. . a user simply duplicated some vectorized enum items. The root of that NULL problem is the fact those single string vectors shared the same proc_t field with their true vectorized version. So while multiple occurrences for most strings could be satisfied with strdup versus the normal ownership usurpation, those true vectorized fields could not be quite so easily copied/duplicated. Thus newlib chose to return a NULL result.strv pointer under either of the above scenarios (which perhaps was just a user boo-boo in the first place). In any event, the NULL was a potential for true string vectors only. Now, since newlib is the sole caller into the readproc module, separate fields have been created for what are just normal strings (never vectorized) and those which remain the true vectorized versions. And, former flags which only worked if combined, now act as stand alone. Thus, both PROCPS_PIDS_CMDLINE & PROCPS_PIDS_CMDLINE_V can be used simultaneously (as they should have been). Also with this patch, items which a user duplicates in the stack (beyond the first such item) will return the the string "[ duplicate ENUM_ID ]". This practice will apply to both single strings and true vectorized ones. In addition to informing users of their error, it will also mean potential NULLs need now never be a concern. Reference(s); http://www.freelists.org/post/procps/systemd-binary-vs-library commit 0580a7b4c67d0297629d37281b4f690894429626 Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-19 10:30:00 +05:30
(void)I; if (NULL != P-> x) { R->result.str = P-> x; P-> x = NULL; } \
library: final tweaks to code and/or comments, 3rd gen With the dust now settling on all those 3rd generation upgrades, this patch tries to provide some consistency among the separate modules involved. Someday we should consider a 4th generation where all redundant code has been removed and isolated in a new shared source file. Following is a summary of significant changes (if any) to each of these now upgraded 3rd gen library modules. <meminfo> ............................................ . strictly formatting/comment changes, code unaffected <pids> ............................................... . replaced a local mkSTR macro with existing STRINGIFY . added fetch narrative explaining duplicate addresses <slabinfo> ........................................... . rearranged some free logic for procps_slabinfo_unref . added fetch narrative explaining duplicate addresses <stat> ............................................... . added #define ENFORCE_LOGICAL, just as in <slabinfo> . replaced a local mkSTR macro with existing STRINGIFY . alphabetized the function declarations in the header <vmstat> ............................................. . made one coverity concession with read_vmstat_failed [ several of these changes may reflect this author's ] [ continuing pursuit of an unreasonable goal -- that ] [ of a 'perfect' (plus 'pretty') C language program! ] Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-09 10:30:00 +05:30
else R->result.str = strdup("[ duplicate " STRINGIFY(e) " ]"); }
library: eliminated a final potential NULL, <PIDS> api In that reference below a specific systemd problem was fixed in the commit shown. However lurking deep within the <pids> interface was yet one final case where NULL could be returned, involving 'strv' and the following: . a user requested both a single string vector (always returned as a normal string) & the vectorized version, as with PROCPS_PIDS_CMDLINE and PROCPS_PIDS_CMDLINE_V. . a user simply duplicated some vectorized enum items. The root of that NULL problem is the fact those single string vectors shared the same proc_t field with their true vectorized version. So while multiple occurrences for most strings could be satisfied with strdup versus the normal ownership usurpation, those true vectorized fields could not be quite so easily copied/duplicated. Thus newlib chose to return a NULL result.strv pointer under either of the above scenarios (which perhaps was just a user boo-boo in the first place). In any event, the NULL was a potential for true string vectors only. Now, since newlib is the sole caller into the readproc module, separate fields have been created for what are just normal strings (never vectorized) and those which remain the true vectorized versions. And, former flags which only worked if combined, now act as stand alone. Thus, both PROCPS_PIDS_CMDLINE & PROCPS_PIDS_CMDLINE_V can be used simultaneously (as they should have been). Also with this patch, items which a user duplicates in the stack (beyond the first such item) will return the the string "[ duplicate ENUM_ID ]". This practice will apply to both single strings and true vectorized ones. In addition to informing users of their error, it will also mean potential NULLs need now never be a concern. Reference(s); http://www.freelists.org/post/procps/systemd-binary-vs-library commit 0580a7b4c67d0297629d37281b4f690894429626 Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-19 10:30:00 +05:30
/* take ownership of true vectorized strings if possible, else return
some sort of hint that they duplicated this char ** item ... */
#define VEC_set(e,x) setDECL(e) { \
(void)I; if (NULL != P-> x) { R->result.strv = P-> x; P-> x = NULL; } \
library: final tweaks to code and/or comments, 3rd gen With the dust now settling on all those 3rd generation upgrades, this patch tries to provide some consistency among the separate modules involved. Someday we should consider a 4th generation where all redundant code has been removed and isolated in a new shared source file. Following is a summary of significant changes (if any) to each of these now upgraded 3rd gen library modules. <meminfo> ............................................ . strictly formatting/comment changes, code unaffected <pids> ............................................... . replaced a local mkSTR macro with existing STRINGIFY . added fetch narrative explaining duplicate addresses <slabinfo> ........................................... . rearranged some free logic for procps_slabinfo_unref . added fetch narrative explaining duplicate addresses <stat> ............................................... . added #define ENFORCE_LOGICAL, just as in <slabinfo> . replaced a local mkSTR macro with existing STRINGIFY . alphabetized the function declarations in the header <vmstat> ............................................. . made one coverity concession with read_vmstat_failed [ several of these changes may reflect this author's ] [ continuing pursuit of an unreasonable goal -- that ] [ of a 'perfect' (plus 'pretty') C language program! ] Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-09 10:30:00 +05:30
else R->result.strv = vectorize_this("[ duplicate " STRINGIFY(e) " ]"); }
library: eliminated a final potential NULL, <PIDS> api In that reference below a specific systemd problem was fixed in the commit shown. However lurking deep within the <pids> interface was yet one final case where NULL could be returned, involving 'strv' and the following: . a user requested both a single string vector (always returned as a normal string) & the vectorized version, as with PROCPS_PIDS_CMDLINE and PROCPS_PIDS_CMDLINE_V. . a user simply duplicated some vectorized enum items. The root of that NULL problem is the fact those single string vectors shared the same proc_t field with their true vectorized version. So while multiple occurrences for most strings could be satisfied with strdup versus the normal ownership usurpation, those true vectorized fields could not be quite so easily copied/duplicated. Thus newlib chose to return a NULL result.strv pointer under either of the above scenarios (which perhaps was just a user boo-boo in the first place). In any event, the NULL was a potential for true string vectors only. Now, since newlib is the sole caller into the readproc module, separate fields have been created for what are just normal strings (never vectorized) and those which remain the true vectorized versions. And, former flags which only worked if combined, now act as stand alone. Thus, both PROCPS_PIDS_CMDLINE & PROCPS_PIDS_CMDLINE_V can be used simultaneously (as they should have been). Also with this patch, items which a user duplicates in the stack (beyond the first such item) will return the the string "[ duplicate ENUM_ID ]". This practice will apply to both single strings and true vectorized ones. In addition to informing users of their error, it will also mean potential NULLs need now never be a concern. Reference(s); http://www.freelists.org/post/procps/systemd-binary-vs-library commit 0580a7b4c67d0297629d37281b4f690894429626 Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-19 10:30:00 +05:30
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
library: protect against a future breakage, <pids> ABI By co-mingling both external/internal identifiers with actual implementation code, potential future additions to our API would have been considerably more difficult. So, this patch will now rely solely on internal/hidden identifiers serving as fenceposts in validation logic. And if the following convention is used for new fields we will maintain that enumerator alphabetic ordering a a little longer (even though 2 user fields now don't): . PROCPS_PIDS_XTRA_FOO_A, PROCPS_PIDS_XTRA_FOO_B, etc. Reference(s): http://www.freelists.org/post/procps/me-too-newlib,7 Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-04-18 10:30:00 +05:30
setDECL(noop) { (void)I; (void)R; (void)P; return; }
setDECL(extra) { (void)I; (void)R; (void)P; return; }
library: exchange <pids> 'addr' for that 'ul_int' type Work on converting ps has revealed the desirability of trading a void pointer for that ul_int type. There was much arithmetic employed against such values and casts would otherwise have been required. Even pmap needed a cast on occasions when comparing an internal variable. Besides, there is much to be said for reducing demands on (and the complexity of) the result structure union. [ we choose ul_int over ull_int since that former is ] [ the exact same size and capacity as a void pointer ] [ regardless of whether compiled as 32-bit or 64-bit ] Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-09-09 10:30:00 +05:30
REG_set(ADDR_END_CODE, ul_int, end_code)
REG_set(ADDR_KSTK_EIP, ul_int, kstk_eip)
REG_set(ADDR_KSTK_ESP, ul_int, kstk_esp)
REG_set(ADDR_START_CODE, ul_int, start_code)
REG_set(ADDR_START_STACK, ul_int, start_stack)
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
REG_set(ALARM, sl_int, alarm)
library: eliminated a final potential NULL, <PIDS> api In that reference below a specific systemd problem was fixed in the commit shown. However lurking deep within the <pids> interface was yet one final case where NULL could be returned, involving 'strv' and the following: . a user requested both a single string vector (always returned as a normal string) & the vectorized version, as with PROCPS_PIDS_CMDLINE and PROCPS_PIDS_CMDLINE_V. . a user simply duplicated some vectorized enum items. The root of that NULL problem is the fact those single string vectors shared the same proc_t field with their true vectorized version. So while multiple occurrences for most strings could be satisfied with strdup versus the normal ownership usurpation, those true vectorized fields could not be quite so easily copied/duplicated. Thus newlib chose to return a NULL result.strv pointer under either of the above scenarios (which perhaps was just a user boo-boo in the first place). In any event, the NULL was a potential for true string vectors only. Now, since newlib is the sole caller into the readproc module, separate fields have been created for what are just normal strings (never vectorized) and those which remain the true vectorized versions. And, former flags which only worked if combined, now act as stand alone. Thus, both PROCPS_PIDS_CMDLINE & PROCPS_PIDS_CMDLINE_V can be used simultaneously (as they should have been). Also with this patch, items which a user duplicates in the stack (beyond the first such item) will return the the string "[ duplicate ENUM_ID ]". This practice will apply to both single strings and true vectorized ones. In addition to informing users of their error, it will also mean potential NULLs need now never be a concern. Reference(s); http://www.freelists.org/post/procps/systemd-binary-vs-library commit 0580a7b4c67d0297629d37281b4f690894429626 Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-19 10:30:00 +05:30
STR_set(CGNAME, cgname)
STR_set(CGROUP, cgroup)
VEC_set(CGROUP_V, cgroup_v)
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
STR_set(CMD, cmd)
library: eliminated a final potential NULL, <PIDS> api In that reference below a specific systemd problem was fixed in the commit shown. However lurking deep within the <pids> interface was yet one final case where NULL could be returned, involving 'strv' and the following: . a user requested both a single string vector (always returned as a normal string) & the vectorized version, as with PROCPS_PIDS_CMDLINE and PROCPS_PIDS_CMDLINE_V. . a user simply duplicated some vectorized enum items. The root of that NULL problem is the fact those single string vectors shared the same proc_t field with their true vectorized version. So while multiple occurrences for most strings could be satisfied with strdup versus the normal ownership usurpation, those true vectorized fields could not be quite so easily copied/duplicated. Thus newlib chose to return a NULL result.strv pointer under either of the above scenarios (which perhaps was just a user boo-boo in the first place). In any event, the NULL was a potential for true string vectors only. Now, since newlib is the sole caller into the readproc module, separate fields have been created for what are just normal strings (never vectorized) and those which remain the true vectorized versions. And, former flags which only worked if combined, now act as stand alone. Thus, both PROCPS_PIDS_CMDLINE & PROCPS_PIDS_CMDLINE_V can be used simultaneously (as they should have been). Also with this patch, items which a user duplicates in the stack (beyond the first such item) will return the the string "[ duplicate ENUM_ID ]". This practice will apply to both single strings and true vectorized ones. In addition to informing users of their error, it will also mean potential NULLs need now never be a concern. Reference(s); http://www.freelists.org/post/procps/systemd-binary-vs-library commit 0580a7b4c67d0297629d37281b4f690894429626 Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-19 10:30:00 +05:30
STR_set(CMDLINE, cmdline)
VEC_set(CMDLINE_V, cmdline_v)
STR_set(ENVIRON, environ)
VEC_set(ENVIRON_V, environ_v)
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
REG_set(EXIT_SIGNAL, s_int, exit_signal)
REG_set(FLAGS, ul_int, flags)
library: try to minimize the results types, <PIDS> api This is just beginning an effort to minimize/normalize the sheer variety of results types in use for our API. In taking these first baby steps, a few anomalies were found. There'll no doubt be many more yet to discover. . the _FLT_ (fault) fields were already signed long in the proc_t (even though their sscanf format used %lu). . although strtoul will alway return an unsigned long, all of the _VM_ fields were made signed long just like other memory fields (& signed sorts more efficiently). Reference(s): http://www.freelists.org/post/procps/newlib-drip-drip,4 Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-15 10:30:00 +05:30
REG_set(FLT_MAJ, sl_int, maj_flt)
REG_set(FLT_MAJ_C, sl_int, cmaj_flt)
REG_set(FLT_MAJ_DELTA, sl_int, maj_delta)
REG_set(FLT_MIN, sl_int, min_flt)
REG_set(FLT_MIN_C, sl_int, cmin_flt)
REG_set(FLT_MIN_DELTA, sl_int, min_delta)
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
REG_set(ID_EGID, u_int, egid)
REG_set(ID_EGROUP, str, egroup)
REG_set(ID_EUID, u_int, euid)
REG_set(ID_EUSER, str, euser)
REG_set(ID_FGID, u_int, fgid)
REG_set(ID_FGROUP, str, fgroup)
REG_set(ID_FUID, u_int, fuid)
REG_set(ID_FUSER, str, fuser)
REG_set(ID_PGRP, s_int, pgrp)
REG_set(ID_PID, s_int, tid)
REG_set(ID_PPID, s_int, ppid)
REG_set(ID_RGID, u_int, rgid)
REG_set(ID_RGROUP, str, rgroup)
REG_set(ID_RUID, u_int, ruid)
REG_set(ID_RUSER, str, ruser)
REG_set(ID_SESSION, s_int, session)
REG_set(ID_SGID, u_int, sgid)
REG_set(ID_SGROUP, str, sgroup)
REG_set(ID_SUID, u_int, suid)
REG_set(ID_SUSER, str, suser)
REG_set(ID_TGID, s_int, tgid)
REG_set(ID_TPGID, s_int, tpgid)
library: stop treating lxcname differently, <PIDS> api Unlike other cached string fields (such as USER name), that lxcname field was defined as 'const char *'. This required a custom 'set' function using a cast in order to avoid a compiler warning regarding that assignment. So this commit brings lxcname into line with all those other cached string fields. And while we're at it, the reason some string fields have no 'freefunc' will also be explained with some new comments in the Item_table. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-01 10:30:00 +05:30
REG_set(LXCNAME, str, lxcname)
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
REG_set(MEM_CODE, sl_int, trs)
library: try to minimize the results types, <PIDS> api This is just beginning an effort to minimize/normalize the sheer variety of results types in use for our API. In taking these first baby steps, a few anomalies were found. There'll no doubt be many more yet to discover. . the _FLT_ (fault) fields were already signed long in the proc_t (even though their sscanf format used %lu). . although strtoul will alway return an unsigned long, all of the _VM_ fields were made signed long just like other memory fields (& signed sorts more efficiently). Reference(s): http://www.freelists.org/post/procps/newlib-drip-drip,4 Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-15 10:30:00 +05:30
CVT_set(MEM_CODE_KIB, sl_int, trs)
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
REG_set(MEM_DATA, sl_int, drs)
library: try to minimize the results types, <PIDS> api This is just beginning an effort to minimize/normalize the sheer variety of results types in use for our API. In taking these first baby steps, a few anomalies were found. There'll no doubt be many more yet to discover. . the _FLT_ (fault) fields were already signed long in the proc_t (even though their sscanf format used %lu). . although strtoul will alway return an unsigned long, all of the _VM_ fields were made signed long just like other memory fields (& signed sorts more efficiently). Reference(s): http://www.freelists.org/post/procps/newlib-drip-drip,4 Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-15 10:30:00 +05:30
CVT_set(MEM_DATA_KIB, sl_int, drs)
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
REG_set(MEM_DT, sl_int, dt)
REG_set(MEM_LRS, sl_int, lrs)
REG_set(MEM_RES, sl_int, resident)
library: try to minimize the results types, <PIDS> api This is just beginning an effort to minimize/normalize the sheer variety of results types in use for our API. In taking these first baby steps, a few anomalies were found. There'll no doubt be many more yet to discover. . the _FLT_ (fault) fields were already signed long in the proc_t (even though their sscanf format used %lu). . although strtoul will alway return an unsigned long, all of the _VM_ fields were made signed long just like other memory fields (& signed sorts more efficiently). Reference(s): http://www.freelists.org/post/procps/newlib-drip-drip,4 Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-15 10:30:00 +05:30
CVT_set(MEM_RES_KIB, sl_int, resident)
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
REG_set(MEM_SHR, sl_int, share)
CVT_set(MEM_SHR_KIB, ul_int, share)
REG_set(MEM_VIRT, sl_int, size)
library: try to minimize the results types, <PIDS> api This is just beginning an effort to minimize/normalize the sheer variety of results types in use for our API. In taking these first baby steps, a few anomalies were found. There'll no doubt be many more yet to discover. . the _FLT_ (fault) fields were already signed long in the proc_t (even though their sscanf format used %lu). . although strtoul will alway return an unsigned long, all of the _VM_ fields were made signed long just like other memory fields (& signed sorts more efficiently). Reference(s): http://www.freelists.org/post/procps/newlib-drip-drip,4 Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-15 10:30:00 +05:30
CVT_set(MEM_VIRT_KIB, sl_int, size)
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
REG_set(NICE, sl_int, nice)
REG_set(NLWP, s_int, nlwp)
library: rework namespace calls Functions related to namespaces were half-in half-out of the procps library and didn't fit the standard naming scheme. While struct { long ns[x]} is a bit clunky, its the only way to "lock in" x. The alternative is to use ns_* variables. This work was needed before pgrep could be converted.
2015-09-03 18:02:19 +05:30
REG_set(NS_IPC, ul_int, ns.ns[0])
REG_set(NS_MNT, ul_int, ns.ns[1])
REG_set(NS_NET, ul_int, ns.ns[2])
REG_set(NS_PID, ul_int, ns.ns[3])
REG_set(NS_USER, ul_int, ns.ns[4])
REG_set(NS_UTS, ul_int, ns.ns[5])
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
REG_set(OOM_ADJ, s_int, oom_adj)
REG_set(OOM_SCORE, s_int, oom_score)
REG_set(PRIORITY, s_int, priority)
REG_set(PROCESSOR, u_int, processor)
REG_set(RSS, sl_int, rss)
REG_set(RSS_RLIM, ul_int, rss_rlim)
REG_set(RTPRIO, ul_int, rtprio)
REG_set(SCHED_CLASS, ul_int, sched)
STR_set(SD_MACH, sd_mach)
STR_set(SD_OUID, sd_ouid)
STR_set(SD_SEAT, sd_seat)
STR_set(SD_SESS, sd_sess)
STR_set(SD_SLICE, sd_slice)
STR_set(SD_UNIT, sd_unit)
STR_set(SD_UUNIT, sd_uunit)
DUP_set(SIGBLOCKED, blocked)
DUP_set(SIGCATCH, sigcatch)
DUP_set(SIGIGNORE, sigignore)
DUP_set(SIGNALS, signal)
DUP_set(SIGPENDING, _sigpnd)
REG_set(STATE, s_ch, state)
STR_set(SUPGIDS, supgid)
STR_set(SUPGROUPS, supgrp)
setDECL(TICS_ALL) { (void)I; R->result.ull_int = P->utime + P->stime; }
setDECL(TICS_ALL_C) { (void)I; R->result.ull_int = P->utime + P->stime + P->cutime + P->cstime; }
library: try to minimize the results types, <PIDS> api This is just beginning an effort to minimize/normalize the sheer variety of results types in use for our API. In taking these first baby steps, a few anomalies were found. There'll no doubt be many more yet to discover. . the _FLT_ (fault) fields were already signed long in the proc_t (even though their sscanf format used %lu). . although strtoul will alway return an unsigned long, all of the _VM_ fields were made signed long just like other memory fields (& signed sorts more efficiently). Reference(s): http://www.freelists.org/post/procps/newlib-drip-drip,4 Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-15 10:30:00 +05:30
REG_set(TICS_DELTA, sl_int, pcpu)
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
REG_set(TICS_SYSTEM, ull_int, stime)
REG_set(TICS_SYSTEM_C, ull_int, cstime)
REG_set(TICS_USER, ull_int, utime)
REG_set(TICS_USER_C, ull_int, cutime)
library: add <pids> TIME_ALL/TIME_ELAPSED needed by ps Shoot, here's yet another bow to ps needs. But it's ok because it makes a lot of sense. Rather than force all users into their own calculations do but it once here. As an aside this need arose during ps testing when the sorts were using TIME_START or TICS_ALL. That was just fine for almost every need except 'etime' plus 'time'. That 'etime' was sorting the opposite of what's wanted when using TIME_START (of course) while 'time' yielded some weird ordering because TICS_ALL was too granular. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-09-30 10:30:00 +05:30
setDECL(TIME_ALL) { R->result.ull_int = (P->utime + P->stime) / I->hertz; }
setDECL(TIME_ELAPSED) { R->result.ull_int = (I->boot_seconds >= (P->start_time / I->hertz)) ? I->boot_seconds - (P->start_time / I->hertz) : 0; }
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
REG_set(TIME_START, ull_int, start_time)
REG_set(TTY, s_int, tty)
library: add PROCPS_PIDS_TTY_NAME to compliment number OK, ok, this was kind of a huge omission. So please do not select the TTY field for display in top quite yet, at least until a next patch has been pushed to GitLab. And to produce a correct sort order for this new field the GNU 'strverscmp' routine was a necessary addition. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-09-07 10:30:00 +05:30
setDECL(TTY_NAME) { char buf[64]; (void)I; dev_to_tty(buf, sizeof(buf), P->tty, P->tid, ABBREV_DEV); R->result.str = strdup(buf); }
library: add abbreviated TTY_NUMBER to that <pids> API While not documented in the man page, ps allows 'tty4' as a valid output specifier complimenting 'tty8' & its derivatives. So, in order to eliminate a dev_to_name() call in the ps program the library will now offer this abbreviated tty version (consisting of a number only). Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-09-30 10:30:00 +05:30
setDECL(TTY_NUMBER) { char buf[64]; (void)I; dev_to_tty(buf, sizeof(buf), P->tty, P->tid, ABBREV_DEV|ABBREV_TTY|ABBREV_PTS); R->result.str = strdup(buf); }
library: try to minimize the results types, <PIDS> api This is just beginning an effort to minimize/normalize the sheer variety of results types in use for our API. In taking these first baby steps, a few anomalies were found. There'll no doubt be many more yet to discover. . the _FLT_ (fault) fields were already signed long in the proc_t (even though their sscanf format used %lu). . although strtoul will alway return an unsigned long, all of the _VM_ fields were made signed long just like other memory fields (& signed sorts more efficiently). Reference(s): http://www.freelists.org/post/procps/newlib-drip-drip,4 Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-15 10:30:00 +05:30
REG_set(VM_DATA, sl_int, vm_data)
REG_set(VM_EXE, sl_int, vm_exe)
REG_set(VM_LIB, sl_int, vm_lib)
REG_set(VM_RSS, sl_int, vm_rss)
REG_set(VM_RSS_ANON, sl_int, vm_rss_anon)
REG_set(VM_RSS_FILE, sl_int, vm_rss_file)
REG_set(VM_RSS_LOCKED, sl_int, vm_lock)
REG_set(VM_RSS_SHARED, sl_int, vm_rss_shared)
REG_set(VM_SIZE, sl_int, vm_size)
REG_set(VM_STACK, sl_int, vm_stack)
REG_set(VM_SWAP, sl_int, vm_swap)
setDECL(VM_USED) { (void)I; R->result.sl_int = P->vm_swap + P->vm_rss; }
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
REG_set(VSIZE_PGS, ul_int, vsize)
library: exchange <pids> 'addr' for that 'ul_int' type Work on converting ps has revealed the desirability of trading a void pointer for that ul_int type. There was much arithmetic employed against such values and casts would otherwise have been required. Even pmap needed a cast on occasions when comparing an internal variable. Besides, there is much to be said for reducing demands on (and the complexity of) the result structure union. [ we choose ul_int over ull_int since that former is ] [ the exact same size and capacity as a void pointer ] [ regardless of whether compiled as 32-bit or 64-bit ] Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-09-09 10:30:00 +05:30
REG_set(WCHAN_ADDR, ul_int, wchan)
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
setDECL(WCHAN_NAME) { (void)I; R->result.str = strdup(lookup_wchan(P->tid)); }
#undef setDECL
#undef CVT_set
#undef DUP_set
#undef REG_set
#undef STR_set
#undef VEC_set
library: add real string vectors to the pids interface After experimenting with an adaptation of pidof to the new pids interface, it became apparent that vectorized versions of those command lines would be necessary. So this commit adds that option and the strv result type. And since the stage had been set, a vectorized version of PROCPS_PIDS_ENVIRON & PROCPS_PIDS_CGROUP was added. Lastly, any use of 'const' in the result structure was removed so callers need not be bothered with casts and compiler warnings. Hopefully, they'll respect a stack. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-26 10:30:00 +05:30
// ___ Free Storage Support |||||||||||||||||||||||||||||||||||||||||||||||||||
library: the overlooked twerks (oops, tweaks), 3rd gen Yes, all of these changes are strictly cosmetic. It is likely symptomatic of some deep-seated character flaw. [ or, it might be because of a certain pride in this ] [ new library and the desire to make it even better! ] Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-07 10:30:00 +05:30
#define freNAME(t) free_results_ ## t
library: add real string vectors to the pids interface After experimenting with an adaptation of pidof to the new pids interface, it became apparent that vectorized versions of those command lines would be necessary. So this commit adds that option and the strv result type. And since the stage had been set, a vectorized version of PROCPS_PIDS_ENVIRON & PROCPS_PIDS_CGROUP was added. Lastly, any use of 'const' in the result structure was removed so callers need not be bothered with casts and compiler warnings. Hopefully, they'll respect a stack. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-26 10:30:00 +05:30
static void freNAME(str) (struct pids_result *R) {
library: miscellaneous minor accumulated <pids> tweaks This patch contains the following minor modifications: . When assigning to boot_seconds from a double, a cast to 'unsigned long' was employed that in reality should have been 'unsigned long long'. But, on second thought it's probably best if the compiler makes the decision. . Results for ID_TGID do not require an 'f_status' flg since both tid and tgid will be available by virtue of the 'simple_nextpid' function normal operations. Thus, that flag has been removed from the <pids> Item_table. . When the pids_stacks structure was eliminated with a change to the alloc/dealloc functions in the reference below, several casts became redundant and are removed. Reference(s): commit 747dfc5987e6e91ea3a8575de307e2892790c598. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-07 10:30:00 +05:30
if (R->result.str) free(R->result.str);
library: add real string vectors to the pids interface After experimenting with an adaptation of pidof to the new pids interface, it became apparent that vectorized versions of those command lines would be necessary. So this commit adds that option and the strv result type. And since the stage had been set, a vectorized version of PROCPS_PIDS_ENVIRON & PROCPS_PIDS_CGROUP was added. Lastly, any use of 'const' in the result structure was removed so callers need not be bothered with casts and compiler warnings. Hopefully, they'll respect a stack. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-26 10:30:00 +05:30
}
static void freNAME(strv) (struct pids_result *R) {
library: correct one <pids> reference from str to strv While not changing generated code this commit corrects one free reference from 'str' to a more proper 'strv'. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-25 10:30:00 +05:30
if (R->result.strv && *R->result.strv) free(*R->result.strv);
library: add real string vectors to the pids interface After experimenting with an adaptation of pidof to the new pids interface, it became apparent that vectorized versions of those command lines would be necessary. So this commit adds that option and the strv result type. And since the stage had been set, a vectorized version of PROCPS_PIDS_ENVIRON & PROCPS_PIDS_CGROUP was added. Lastly, any use of 'const' in the result structure was removed so callers need not be bothered with casts and compiler warnings. Hopefully, they'll respect a stack. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-26 10:30:00 +05:30
}
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
// ___ Sorting Support ||||||||||||||||||||||||||||||||||||||||||||||||||||||||
struct sort_parms {
int offset;
enum pids_sort_order order;
};
library: adjust 'noop' callback definition, <PIDS> api Though all those callback's parameters are ignored and qsort treats them as pointers to void, it's wrong when 3rd parm is 'enum pids_item', not 'struct sort_parms'. So we will fix it in a way that lessens the likelihood of another such a mistake when some new type is added. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-09 10:30:00 +05:30
#define srtNAME(t) sort_results_ ## t
#define srtDECL(t) static int srtNAME(t) \
(const struct pids_stack **A, const struct pids_stack **B, struct sort_parms *P)
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
library: adjust 'noop' callback definition, <PIDS> api Though all those callback's parameters are ignored and qsort treats them as pointers to void, it's wrong when 3rd parm is 'enum pids_item', not 'struct sort_parms'. So we will fix it in a way that lessens the likelihood of another such a mistake when some new type is added. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-09 10:30:00 +05:30
#define NUM_srt(T) srtDECL(T) { \
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
const struct pids_result *a = (*A)->head + P->offset; \
const struct pids_result *b = (*B)->head + P->offset; \
library: make the pids_sort_order enums more intuitive The values of PROCPS_SORT_ASCEND & PROCPS_SORT_DESCEND were a tad unintuitive. This patch will just make them a more natural +1 for ascending and -1 for descending. [ plus it still allows that fast path multiplication ] [ instead of a comparison for signed numbers/strings ] Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-09-13 10:30:00 +05:30
return P->order * (a->result. T - b->result. T); }
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
library: adjust 'noop' callback definition, <PIDS> api Though all those callback's parameters are ignored and qsort treats them as pointers to void, it's wrong when 3rd parm is 'enum pids_item', not 'struct sort_parms'. So we will fix it in a way that lessens the likelihood of another such a mistake when some new type is added. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-09 10:30:00 +05:30
#define REG_srt(T) srtDECL(T) { \
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
const struct pids_result *a = (*A)->head + P->offset; \
const struct pids_result *b = (*B)->head + P->offset; \
library: make the pids_sort_order enums more intuitive The values of PROCPS_SORT_ASCEND & PROCPS_SORT_DESCEND were a tad unintuitive. This patch will just make them a more natural +1 for ascending and -1 for descending. [ plus it still allows that fast path multiplication ] [ instead of a comparison for signed numbers/strings ] Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-09-13 10:30:00 +05:30
if ( a->result. T > b->result. T ) return P->order > 0 ? 1 : -1; \
if ( a->result. T < b->result. T ) return P->order > 0 ? -1 : 1; \
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
return 0; }
NUM_srt(s_ch)
NUM_srt(s_int)
NUM_srt(sl_int)
REG_srt(u_int)
REG_srt(ul_int)
REG_srt(ull_int)
library: adjust 'noop' callback definition, <PIDS> api Though all those callback's parameters are ignored and qsort treats them as pointers to void, it's wrong when 3rd parm is 'enum pids_item', not 'struct sort_parms'. So we will fix it in a way that lessens the likelihood of another such a mistake when some new type is added. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-09 10:30:00 +05:30
srtDECL(str) {
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
const struct pids_result *a = (*A)->head + P->offset;
const struct pids_result *b = (*B)->head + P->offset;
library: make the pids_sort_order enums more intuitive The values of PROCPS_SORT_ASCEND & PROCPS_SORT_DESCEND were a tad unintuitive. This patch will just make them a more natural +1 for ascending and -1 for descending. [ plus it still allows that fast path multiplication ] [ instead of a comparison for signed numbers/strings ] Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-09-13 10:30:00 +05:30
return P->order * strcoll(a->result.str, b->result.str);
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
}
library: adjust 'noop' callback definition, <PIDS> api Though all those callback's parameters are ignored and qsort treats them as pointers to void, it's wrong when 3rd parm is 'enum pids_item', not 'struct sort_parms'. So we will fix it in a way that lessens the likelihood of another such a mistake when some new type is added. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-09 10:30:00 +05:30
srtDECL(strv) {
library: add real string vectors to the pids interface After experimenting with an adaptation of pidof to the new pids interface, it became apparent that vectorized versions of those command lines would be necessary. So this commit adds that option and the strv result type. And since the stage had been set, a vectorized version of PROCPS_PIDS_ENVIRON & PROCPS_PIDS_CGROUP was added. Lastly, any use of 'const' in the result structure was removed so callers need not be bothered with casts and compiler warnings. Hopefully, they'll respect a stack. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-26 10:30:00 +05:30
const struct pids_result *a = (*A)->head + P->offset;
const struct pids_result *b = (*B)->head + P->offset;
if (!a->result.strv || !b->result.strv) return 0;
library: make the pids_sort_order enums more intuitive The values of PROCPS_SORT_ASCEND & PROCPS_SORT_DESCEND were a tad unintuitive. This patch will just make them a more natural +1 for ascending and -1 for descending. [ plus it still allows that fast path multiplication ] [ instead of a comparison for signed numbers/strings ] Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-09-13 10:30:00 +05:30
return P->order * strcoll((*a->result.strv), (*b->result.strv));
library: add real string vectors to the pids interface After experimenting with an adaptation of pidof to the new pids interface, it became apparent that vectorized versions of those command lines would be necessary. So this commit adds that option and the strv result type. And since the stage had been set, a vectorized version of PROCPS_PIDS_ENVIRON & PROCPS_PIDS_CGROUP was added. Lastly, any use of 'const' in the result structure was removed so callers need not be bothered with casts and compiler warnings. Hopefully, they'll respect a stack. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-26 10:30:00 +05:30
}
library: adjust 'noop' callback definition, <PIDS> api Though all those callback's parameters are ignored and qsort treats them as pointers to void, it's wrong when 3rd parm is 'enum pids_item', not 'struct sort_parms'. So we will fix it in a way that lessens the likelihood of another such a mistake when some new type is added. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-09 10:30:00 +05:30
srtDECL(strvers) {
library: add PROCPS_PIDS_TTY_NAME to compliment number OK, ok, this was kind of a huge omission. So please do not select the TTY field for display in top quite yet, at least until a next patch has been pushed to GitLab. And to produce a correct sort order for this new field the GNU 'strverscmp' routine was a necessary addition. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-09-07 10:30:00 +05:30
const struct pids_result *a = (*A)->head + P->offset;
const struct pids_result *b = (*B)->head + P->offset;
library: make the pids_sort_order enums more intuitive The values of PROCPS_SORT_ASCEND & PROCPS_SORT_DESCEND were a tad unintuitive. This patch will just make them a more natural +1 for ascending and -1 for descending. [ plus it still allows that fast path multiplication ] [ instead of a comparison for signed numbers/strings ] Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-09-13 10:30:00 +05:30
return P->order * strverscmp(a->result.str, b->result.str);
library: add PROCPS_PIDS_TTY_NAME to compliment number OK, ok, this was kind of a huge omission. So please do not select the TTY field for display in top quite yet, at least until a next patch has been pushed to GitLab. And to produce a correct sort order for this new field the GNU 'strverscmp' routine was a necessary addition. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-09-07 10:30:00 +05:30
}
library: adjust 'noop' callback definition, <PIDS> api Though all those callback's parameters are ignored and qsort treats them as pointers to void, it's wrong when 3rd parm is 'enum pids_item', not 'struct sort_parms'. So we will fix it in a way that lessens the likelihood of another such a mistake when some new type is added. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-09 10:30:00 +05:30
srtDECL(noop) {
(void)A; (void)B; (void)P;
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
return 0;
}
library: adjust 'noop' callback definition, <PIDS> api Though all those callback's parameters are ignored and qsort treats them as pointers to void, it's wrong when 3rd parm is 'enum pids_item', not 'struct sort_parms'. So we will fix it in a way that lessens the likelihood of another such a mistake when some new type is added. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-09 10:30:00 +05:30
#undef srtDECL
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
#undef NUM_srt
#undef REG_srt
// ___ Controlling Table ||||||||||||||||||||||||||||||||||||||||||||||||||||||
// from either 'stat' or 'status' (preferred)
#define f_either PROC_SPARE_1
#define f_grp PROC_FILLGRP
#define f_lxc PROC_FILL_LXC
#define f_ns PROC_FILLNS
#define f_oom PROC_FILLOOM
#define f_stat PROC_FILLSTAT
#define f_statm PROC_FILLMEM
#define f_status PROC_FILLSTATUS
#define f_systemd PROC_FILLSYSTEMD
#define f_usr PROC_FILLUSR
library: eliminated a final potential NULL, <PIDS> api In that reference below a specific systemd problem was fixed in the commit shown. However lurking deep within the <pids> interface was yet one final case where NULL could be returned, involving 'strv' and the following: . a user requested both a single string vector (always returned as a normal string) & the vectorized version, as with PROCPS_PIDS_CMDLINE and PROCPS_PIDS_CMDLINE_V. . a user simply duplicated some vectorized enum items. The root of that NULL problem is the fact those single string vectors shared the same proc_t field with their true vectorized version. So while multiple occurrences for most strings could be satisfied with strdup versus the normal ownership usurpation, those true vectorized fields could not be quite so easily copied/duplicated. Thus newlib chose to return a NULL result.strv pointer under either of the above scenarios (which perhaps was just a user boo-boo in the first place). In any event, the NULL was a potential for true string vectors only. Now, since newlib is the sole caller into the readproc module, separate fields have been created for what are just normal strings (never vectorized) and those which remain the true vectorized versions. And, former flags which only worked if combined, now act as stand alone. Thus, both PROCPS_PIDS_CMDLINE & PROCPS_PIDS_CMDLINE_V can be used simultaneously (as they should have been). Also with this patch, items which a user duplicates in the stack (beyond the first such item) will return the the string "[ duplicate ENUM_ID ]". This practice will apply to both single strings and true vectorized ones. In addition to informing users of their error, it will also mean potential NULLs need now never be a concern. Reference(s); http://www.freelists.org/post/procps/systemd-binary-vs-library commit 0580a7b4c67d0297629d37281b4f690894429626 Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-19 10:30:00 +05:30
// these next three will yield a single string (never vectorized)
#define x_cgroup PROC_EDITCGRPCVT
#define x_cmdline PROC_EDITCMDLCVT
#define x_environ PROC_EDITENVRCVT
library: accumulated changes to <pids> code & comments A patch containing the following miscellaneous tweaks: . avoided distortions unique to PROCPS_PIDS_TICS_DELTA . addressed several smatch warnings and/or suggestions . ensured oldproc_close invoked should tally_proc fail . keeping that namespace clean, added a missing #undef . added 2 comments acknowledging pids_item as unsigned . added/clarified comments regarding proc flags & strv From smatch analysis: . these were indeed boo-boos pids.c:580 make_hist() warn: the 'Hr' macro might need parens pids.c:1058 procps_pids_reap() warn: variable dereferenced before check 'info' (see line 1056) . these were not errors (and we did double check) pids.c:1067 procps_pids_reap() warn: double check that we're allocating correct size: 8 vs 128 pids.c:1068 procps_pids_reap() warn: double check that we're allocating correct size: 8 vs 128 Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-28 10:30:00 +05:30
// these next three will yield true verctorized strings
library: add real string vectors to the pids interface After experimenting with an adaptation of pidof to the new pids interface, it became apparent that vectorized versions of those command lines would be necessary. So this commit adds that option and the strv result type. And since the stage had been set, a vectorized version of PROCPS_PIDS_ENVIRON & PROCPS_PIDS_CGROUP was added. Lastly, any use of 'const' in the result structure was removed so callers need not be bothered with casts and compiler warnings. Hopefully, they'll respect a stack. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-26 10:30:00 +05:30
#define v_arg PROC_FILLARG
#define v_cgroup PROC_FILLCGROUP
#define v_env PROC_FILLENV
library: eliminated a final potential NULL, <PIDS> api In that reference below a specific systemd problem was fixed in the commit shown. However lurking deep within the <pids> interface was yet one final case where NULL could be returned, involving 'strv' and the following: . a user requested both a single string vector (always returned as a normal string) & the vectorized version, as with PROCPS_PIDS_CMDLINE and PROCPS_PIDS_CMDLINE_V. . a user simply duplicated some vectorized enum items. The root of that NULL problem is the fact those single string vectors shared the same proc_t field with their true vectorized version. So while multiple occurrences for most strings could be satisfied with strdup versus the normal ownership usurpation, those true vectorized fields could not be quite so easily copied/duplicated. Thus newlib chose to return a NULL result.strv pointer under either of the above scenarios (which perhaps was just a user boo-boo in the first place). In any event, the NULL was a potential for true string vectors only. Now, since newlib is the sole caller into the readproc module, separate fields have been created for what are just normal strings (never vectorized) and those which remain the true vectorized versions. And, former flags which only worked if combined, now act as stand alone. Thus, both PROCPS_PIDS_CMDLINE & PROCPS_PIDS_CMDLINE_V can be used simultaneously (as they should have been). Also with this patch, items which a user duplicates in the stack (beyond the first such item) will return the the string "[ duplicate ENUM_ID ]". This practice will apply to both single strings and true vectorized ones. In addition to informing users of their error, it will also mean potential NULLs need now never be a concern. Reference(s); http://www.freelists.org/post/procps/systemd-binary-vs-library commit 0580a7b4c67d0297629d37281b4f690894429626 Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-19 10:30:00 +05:30
// remaining are compound flags
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
#define x_ogroup PROC_FILLSTATUS | PROC_FILLGRP
#define x_ouser PROC_FILLSTATUS | PROC_FILLUSR
#define x_supgrp PROC_FILLSTATUS | PROC_FILLSUPGRP
typedef void (*SET_t)(struct procps_pidsinfo *, struct pids_result *, proc_t *);
library: add real string vectors to the pids interface After experimenting with an adaptation of pidof to the new pids interface, it became apparent that vectorized versions of those command lines would be necessary. So this commit adds that option and the strv result type. And since the stage had been set, a vectorized version of PROCPS_PIDS_ENVIRON & PROCPS_PIDS_CGROUP was added. Lastly, any use of 'const' in the result structure was removed so callers need not be bothered with casts and compiler warnings. Hopefully, they'll respect a stack. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-26 10:30:00 +05:30
typedef void (*FRE_t)(struct pids_result *);
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
typedef int (*QSR_t)(const void *, const void *, void *);
#define RS(e) (SET_t)setNAME(e)
library: the overlooked twerks (oops, tweaks), 3rd gen Yes, all of these changes are strictly cosmetic. It is likely symptomatic of some deep-seated character flaw. [ or, it might be because of a certain pride in this ] [ new library and the desire to make it even better! ] Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-07 10:30:00 +05:30
#define FF(t) (FRE_t)freNAME(t)
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
#define QS(t) (QSR_t)srtNAME(t)
/*
* Need it be said?
* This table must be kept in the exact same order as
* those 'enum pids_item' guys ! */
static struct {
library: add real string vectors to the pids interface After experimenting with an adaptation of pidof to the new pids interface, it became apparent that vectorized versions of those command lines would be necessary. So this commit adds that option and the strv result type. And since the stage had been set, a vectorized version of PROCPS_PIDS_ENVIRON & PROCPS_PIDS_CGROUP was added. Lastly, any use of 'const' in the result structure was removed so callers need not be bothered with casts and compiler warnings. Hopefully, they'll respect a stack. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-26 10:30:00 +05:30
SET_t setsfunc; // the actual result setting routine
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
unsigned oldflags; // PROC_FILLxxxx flags for this item
library: add real string vectors to the pids interface After experimenting with an adaptation of pidof to the new pids interface, it became apparent that vectorized versions of those command lines would be necessary. So this commit adds that option and the strv result type. And since the stage had been set, a vectorized version of PROCPS_PIDS_ENVIRON & PROCPS_PIDS_CGROUP was added. Lastly, any use of 'const' in the result structure was removed so callers need not be bothered with casts and compiler warnings. Hopefully, they'll respect a stack. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-26 10:30:00 +05:30
FRE_t freefunc; // free function for strings storage
QSR_t sortfunc; // sort cmp func for a specific type
library: miscellaneous tweaks to pid code and comments . traded a complex misaligned memory allocation scheme in the make_hist function for a simple aligned scheme. plus memory allocation increases are globally defined. . changed 1 parameter for procps_pids_stacks_sort() to better reflect the 'array of pointers', not an address of a pointer as is used with guys such as 'new/unref'. . the pids_reap struct was changed slightly to make it more reflective of it's actual implementation details. . the Item_table member .mustfree is now .needfree and that .makehist was now made .needhist for consistency. . reduced the number of separate 'return NULL;' source statements in that primary procps_pids_reap() routine. . ensured consistent reference to sizeof(void *) & not occasional reference to sizeof(void*) without a space. . rather than enable/disable validate_stacks via a #if in the function body, it is now handled via a #define. . some comments in the procps_pids_reset function were adjusted to reflect this current implementation. shown originally, they reflected an aborted attempt to avoid a testing aberration not fully understood at the time. . added a summary of the memory overhead cost of HST_t processing to that UNREF_RPTHASH output at unref time. . a 'PIDs at max depth:' portion of that UNREF_RPTHASH enabled #define is now published only when the maximum depth of hash table entry chains exceed depths of one. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-20 10:30:00 +05:30
int needhist; // a result requires history support
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
} Item_table[] = {
library: eliminated a final potential NULL, <PIDS> api In that reference below a specific systemd problem was fixed in the commit shown. However lurking deep within the <pids> interface was yet one final case where NULL could be returned, involving 'strv' and the following: . a user requested both a single string vector (always returned as a normal string) & the vectorized version, as with PROCPS_PIDS_CMDLINE and PROCPS_PIDS_CMDLINE_V. . a user simply duplicated some vectorized enum items. The root of that NULL problem is the fact those single string vectors shared the same proc_t field with their true vectorized version. So while multiple occurrences for most strings could be satisfied with strdup versus the normal ownership usurpation, those true vectorized fields could not be quite so easily copied/duplicated. Thus newlib chose to return a NULL result.strv pointer under either of the above scenarios (which perhaps was just a user boo-boo in the first place). In any event, the NULL was a potential for true string vectors only. Now, since newlib is the sole caller into the readproc module, separate fields have been created for what are just normal strings (never vectorized) and those which remain the true vectorized versions. And, former flags which only worked if combined, now act as stand alone. Thus, both PROCPS_PIDS_CMDLINE & PROCPS_PIDS_CMDLINE_V can be used simultaneously (as they should have been). Also with this patch, items which a user duplicates in the stack (beyond the first such item) will return the the string "[ duplicate ENUM_ID ]". This practice will apply to both single strings and true vectorized ones. In addition to informing users of their error, it will also mean potential NULLs need now never be a concern. Reference(s); http://www.freelists.org/post/procps/systemd-binary-vs-library commit 0580a7b4c67d0297629d37281b4f690894429626 Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-19 10:30:00 +05:30
/* setsfunc oldflags freefunc sortfunc needhist
--------------------- ---------- --------- ------------- -------- */
{ RS(noop), 0, NULL, QS(noop), 0 }, // user only, never altered
{ RS(extra), 0, NULL, QS(ull_int), 0 }, // user only, reset to zero
library: final tweaks to code and/or comments, 3rd gen With the dust now settling on all those 3rd generation upgrades, this patch tries to provide some consistency among the separate modules involved. Someday we should consider a 4th generation where all redundant code has been removed and isolated in a new shared source file. Following is a summary of significant changes (if any) to each of these now upgraded 3rd gen library modules. <meminfo> ............................................ . strictly formatting/comment changes, code unaffected <pids> ............................................... . replaced a local mkSTR macro with existing STRINGIFY . added fetch narrative explaining duplicate addresses <slabinfo> ........................................... . rearranged some free logic for procps_slabinfo_unref . added fetch narrative explaining duplicate addresses <stat> ............................................... . added #define ENFORCE_LOGICAL, just as in <slabinfo> . replaced a local mkSTR macro with existing STRINGIFY . alphabetized the function declarations in the header <vmstat> ............................................. . made one coverity concession with read_vmstat_failed [ several of these changes may reflect this author's ] [ continuing pursuit of an unreasonable goal -- that ] [ of a 'perfect' (plus 'pretty') C language program! ] Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-09 10:30:00 +05:30
library: eliminated a final potential NULL, <PIDS> api In that reference below a specific systemd problem was fixed in the commit shown. However lurking deep within the <pids> interface was yet one final case where NULL could be returned, involving 'strv' and the following: . a user requested both a single string vector (always returned as a normal string) & the vectorized version, as with PROCPS_PIDS_CMDLINE and PROCPS_PIDS_CMDLINE_V. . a user simply duplicated some vectorized enum items. The root of that NULL problem is the fact those single string vectors shared the same proc_t field with their true vectorized version. So while multiple occurrences for most strings could be satisfied with strdup versus the normal ownership usurpation, those true vectorized fields could not be quite so easily copied/duplicated. Thus newlib chose to return a NULL result.strv pointer under either of the above scenarios (which perhaps was just a user boo-boo in the first place). In any event, the NULL was a potential for true string vectors only. Now, since newlib is the sole caller into the readproc module, separate fields have been created for what are just normal strings (never vectorized) and those which remain the true vectorized versions. And, former flags which only worked if combined, now act as stand alone. Thus, both PROCPS_PIDS_CMDLINE & PROCPS_PIDS_CMDLINE_V can be used simultaneously (as they should have been). Also with this patch, items which a user duplicates in the stack (beyond the first such item) will return the the string "[ duplicate ENUM_ID ]". This practice will apply to both single strings and true vectorized ones. In addition to informing users of their error, it will also mean potential NULLs need now never be a concern. Reference(s); http://www.freelists.org/post/procps/systemd-binary-vs-library commit 0580a7b4c67d0297629d37281b4f690894429626 Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-19 10:30:00 +05:30
{ RS(ADDR_END_CODE), f_stat, NULL, QS(ul_int), 0 },
{ RS(ADDR_KSTK_EIP), f_stat, NULL, QS(ul_int), 0 },
{ RS(ADDR_KSTK_ESP), f_stat, NULL, QS(ul_int), 0 },
{ RS(ADDR_START_CODE), f_stat, NULL, QS(ul_int), 0 },
{ RS(ADDR_START_STACK), f_stat, NULL, QS(ul_int), 0 },
{ RS(ALARM), f_stat, NULL, QS(sl_int), 0 },
{ RS(CGNAME), x_cgroup, FF(str), QS(str), 0 },
{ RS(CGROUP), x_cgroup, FF(str), QS(str), 0 },
{ RS(CGROUP_V), v_cgroup, FF(strv), QS(strv), 0 },
{ RS(CMD), f_either, FF(str), QS(str), 0 },
{ RS(CMDLINE), x_cmdline, FF(str), QS(str), 0 },
{ RS(CMDLINE_V), v_arg, FF(strv), QS(strv), 0 },
{ RS(ENVIRON), x_environ, FF(str), QS(str), 0 },
{ RS(ENVIRON_V), v_env, FF(strv), QS(strv), 0 },
{ RS(EXIT_SIGNAL), f_stat, NULL, QS(s_int), 0 },
{ RS(FLAGS), f_stat, NULL, QS(ul_int), 0 },
{ RS(FLT_MAJ), f_stat, NULL, QS(sl_int), 0 },
{ RS(FLT_MAJ_C), f_stat, NULL, QS(sl_int), 0 },
{ RS(FLT_MAJ_DELTA), f_stat, NULL, QS(sl_int), +1 },
{ RS(FLT_MIN), f_stat, NULL, QS(sl_int), 0 },
{ RS(FLT_MIN_C), f_stat, NULL, QS(sl_int), 0 },
{ RS(FLT_MIN_DELTA), f_stat, NULL, QS(sl_int), +1 },
library: stop treating lxcname differently, <PIDS> api Unlike other cached string fields (such as USER name), that lxcname field was defined as 'const char *'. This required a custom 'set' function using a cast in order to avoid a compiler warning regarding that assignment. So this commit brings lxcname into line with all those other cached string fields. And while we're at it, the reason some string fields have no 'freefunc' will also be explained with some new comments in the Item_table. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-01 10:30:00 +05:30
{ RS(ID_EGID), 0, NULL, QS(u_int), 0 }, // oldflags: free w/ simple_read
library: eliminated a final potential NULL, <PIDS> api In that reference below a specific systemd problem was fixed in the commit shown. However lurking deep within the <pids> interface was yet one final case where NULL could be returned, involving 'strv' and the following: . a user requested both a single string vector (always returned as a normal string) & the vectorized version, as with PROCPS_PIDS_CMDLINE and PROCPS_PIDS_CMDLINE_V. . a user simply duplicated some vectorized enum items. The root of that NULL problem is the fact those single string vectors shared the same proc_t field with their true vectorized version. So while multiple occurrences for most strings could be satisfied with strdup versus the normal ownership usurpation, those true vectorized fields could not be quite so easily copied/duplicated. Thus newlib chose to return a NULL result.strv pointer under either of the above scenarios (which perhaps was just a user boo-boo in the first place). In any event, the NULL was a potential for true string vectors only. Now, since newlib is the sole caller into the readproc module, separate fields have been created for what are just normal strings (never vectorized) and those which remain the true vectorized versions. And, former flags which only worked if combined, now act as stand alone. Thus, both PROCPS_PIDS_CMDLINE & PROCPS_PIDS_CMDLINE_V can be used simultaneously (as they should have been). Also with this patch, items which a user duplicates in the stack (beyond the first such item) will return the the string "[ duplicate ENUM_ID ]". This practice will apply to both single strings and true vectorized ones. In addition to informing users of their error, it will also mean potential NULLs need now never be a concern. Reference(s); http://www.freelists.org/post/procps/systemd-binary-vs-library commit 0580a7b4c67d0297629d37281b4f690894429626 Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-19 10:30:00 +05:30
{ RS(ID_EGROUP), f_grp, NULL, QS(str), 0 },
library: stop treating lxcname differently, <PIDS> api Unlike other cached string fields (such as USER name), that lxcname field was defined as 'const char *'. This required a custom 'set' function using a cast in order to avoid a compiler warning regarding that assignment. So this commit brings lxcname into line with all those other cached string fields. And while we're at it, the reason some string fields have no 'freefunc' will also be explained with some new comments in the Item_table. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-01 10:30:00 +05:30
{ RS(ID_EUID), 0, NULL, QS(u_int), 0 }, // oldflags: free w/ simple_read
{ RS(ID_EUSER), f_usr, NULL, QS(str), 0 }, // freefunc NULL w/ cached string
library: eliminated a final potential NULL, <PIDS> api In that reference below a specific systemd problem was fixed in the commit shown. However lurking deep within the <pids> interface was yet one final case where NULL could be returned, involving 'strv' and the following: . a user requested both a single string vector (always returned as a normal string) & the vectorized version, as with PROCPS_PIDS_CMDLINE and PROCPS_PIDS_CMDLINE_V. . a user simply duplicated some vectorized enum items. The root of that NULL problem is the fact those single string vectors shared the same proc_t field with their true vectorized version. So while multiple occurrences for most strings could be satisfied with strdup versus the normal ownership usurpation, those true vectorized fields could not be quite so easily copied/duplicated. Thus newlib chose to return a NULL result.strv pointer under either of the above scenarios (which perhaps was just a user boo-boo in the first place). In any event, the NULL was a potential for true string vectors only. Now, since newlib is the sole caller into the readproc module, separate fields have been created for what are just normal strings (never vectorized) and those which remain the true vectorized versions. And, former flags which only worked if combined, now act as stand alone. Thus, both PROCPS_PIDS_CMDLINE & PROCPS_PIDS_CMDLINE_V can be used simultaneously (as they should have been). Also with this patch, items which a user duplicates in the stack (beyond the first such item) will return the the string "[ duplicate ENUM_ID ]". This practice will apply to both single strings and true vectorized ones. In addition to informing users of their error, it will also mean potential NULLs need now never be a concern. Reference(s); http://www.freelists.org/post/procps/systemd-binary-vs-library commit 0580a7b4c67d0297629d37281b4f690894429626 Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-19 10:30:00 +05:30
{ RS(ID_FGID), f_status, NULL, QS(u_int), 0 },
{ RS(ID_FGROUP), x_ogroup, NULL, QS(str), 0 },
{ RS(ID_FUID), f_status, NULL, QS(u_int), 0 },
library: stop treating lxcname differently, <PIDS> api Unlike other cached string fields (such as USER name), that lxcname field was defined as 'const char *'. This required a custom 'set' function using a cast in order to avoid a compiler warning regarding that assignment. So this commit brings lxcname into line with all those other cached string fields. And while we're at it, the reason some string fields have no 'freefunc' will also be explained with some new comments in the Item_table. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-01 10:30:00 +05:30
{ RS(ID_FUSER), x_ouser, NULL, QS(str), 0 }, // freefunc NULL w/ cached string
library: eliminated a final potential NULL, <PIDS> api In that reference below a specific systemd problem was fixed in the commit shown. However lurking deep within the <pids> interface was yet one final case where NULL could be returned, involving 'strv' and the following: . a user requested both a single string vector (always returned as a normal string) & the vectorized version, as with PROCPS_PIDS_CMDLINE and PROCPS_PIDS_CMDLINE_V. . a user simply duplicated some vectorized enum items. The root of that NULL problem is the fact those single string vectors shared the same proc_t field with their true vectorized version. So while multiple occurrences for most strings could be satisfied with strdup versus the normal ownership usurpation, those true vectorized fields could not be quite so easily copied/duplicated. Thus newlib chose to return a NULL result.strv pointer under either of the above scenarios (which perhaps was just a user boo-boo in the first place). In any event, the NULL was a potential for true string vectors only. Now, since newlib is the sole caller into the readproc module, separate fields have been created for what are just normal strings (never vectorized) and those which remain the true vectorized versions. And, former flags which only worked if combined, now act as stand alone. Thus, both PROCPS_PIDS_CMDLINE & PROCPS_PIDS_CMDLINE_V can be used simultaneously (as they should have been). Also with this patch, items which a user duplicates in the stack (beyond the first such item) will return the the string "[ duplicate ENUM_ID ]". This practice will apply to both single strings and true vectorized ones. In addition to informing users of their error, it will also mean potential NULLs need now never be a concern. Reference(s); http://www.freelists.org/post/procps/systemd-binary-vs-library commit 0580a7b4c67d0297629d37281b4f690894429626 Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-19 10:30:00 +05:30
{ RS(ID_PGRP), f_stat, NULL, QS(s_int), 0 },
{ RS(ID_PID), 0, NULL, QS(s_int), 0 }, // oldflags: free w/ simple_nextpid
{ RS(ID_PPID), f_either, NULL, QS(s_int), 0 },
{ RS(ID_RGID), f_status, NULL, QS(u_int), 0 },
{ RS(ID_RGROUP), x_ogroup, NULL, QS(str), 0 },
{ RS(ID_RUID), f_status, NULL, QS(u_int), 0 },
library: stop treating lxcname differently, <PIDS> api Unlike other cached string fields (such as USER name), that lxcname field was defined as 'const char *'. This required a custom 'set' function using a cast in order to avoid a compiler warning regarding that assignment. So this commit brings lxcname into line with all those other cached string fields. And while we're at it, the reason some string fields have no 'freefunc' will also be explained with some new comments in the Item_table. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-01 10:30:00 +05:30
{ RS(ID_RUSER), x_ouser, NULL, QS(str), 0 }, // freefunc NULL w/ cached string
library: eliminated a final potential NULL, <PIDS> api In that reference below a specific systemd problem was fixed in the commit shown. However lurking deep within the <pids> interface was yet one final case where NULL could be returned, involving 'strv' and the following: . a user requested both a single string vector (always returned as a normal string) & the vectorized version, as with PROCPS_PIDS_CMDLINE and PROCPS_PIDS_CMDLINE_V. . a user simply duplicated some vectorized enum items. The root of that NULL problem is the fact those single string vectors shared the same proc_t field with their true vectorized version. So while multiple occurrences for most strings could be satisfied with strdup versus the normal ownership usurpation, those true vectorized fields could not be quite so easily copied/duplicated. Thus newlib chose to return a NULL result.strv pointer under either of the above scenarios (which perhaps was just a user boo-boo in the first place). In any event, the NULL was a potential for true string vectors only. Now, since newlib is the sole caller into the readproc module, separate fields have been created for what are just normal strings (never vectorized) and those which remain the true vectorized versions. And, former flags which only worked if combined, now act as stand alone. Thus, both PROCPS_PIDS_CMDLINE & PROCPS_PIDS_CMDLINE_V can be used simultaneously (as they should have been). Also with this patch, items which a user duplicates in the stack (beyond the first such item) will return the the string "[ duplicate ENUM_ID ]". This practice will apply to both single strings and true vectorized ones. In addition to informing users of their error, it will also mean potential NULLs need now never be a concern. Reference(s); http://www.freelists.org/post/procps/systemd-binary-vs-library commit 0580a7b4c67d0297629d37281b4f690894429626 Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-19 10:30:00 +05:30
{ RS(ID_SESSION), f_stat, NULL, QS(s_int), 0 },
{ RS(ID_SGID), f_status, NULL, QS(u_int), 0 },
{ RS(ID_SGROUP), x_ogroup, NULL, QS(str), 0 },
{ RS(ID_SUID), f_status, NULL, QS(u_int), 0 },
library: stop treating lxcname differently, <PIDS> api Unlike other cached string fields (such as USER name), that lxcname field was defined as 'const char *'. This required a custom 'set' function using a cast in order to avoid a compiler warning regarding that assignment. So this commit brings lxcname into line with all those other cached string fields. And while we're at it, the reason some string fields have no 'freefunc' will also be explained with some new comments in the Item_table. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-01 10:30:00 +05:30
{ RS(ID_SUSER), x_ouser, NULL, QS(str), 0 }, // freefunc NULL w/ cached string
library: eliminated a final potential NULL, <PIDS> api In that reference below a specific systemd problem was fixed in the commit shown. However lurking deep within the <pids> interface was yet one final case where NULL could be returned, involving 'strv' and the following: . a user requested both a single string vector (always returned as a normal string) & the vectorized version, as with PROCPS_PIDS_CMDLINE and PROCPS_PIDS_CMDLINE_V. . a user simply duplicated some vectorized enum items. The root of that NULL problem is the fact those single string vectors shared the same proc_t field with their true vectorized version. So while multiple occurrences for most strings could be satisfied with strdup versus the normal ownership usurpation, those true vectorized fields could not be quite so easily copied/duplicated. Thus newlib chose to return a NULL result.strv pointer under either of the above scenarios (which perhaps was just a user boo-boo in the first place). In any event, the NULL was a potential for true string vectors only. Now, since newlib is the sole caller into the readproc module, separate fields have been created for what are just normal strings (never vectorized) and those which remain the true vectorized versions. And, former flags which only worked if combined, now act as stand alone. Thus, both PROCPS_PIDS_CMDLINE & PROCPS_PIDS_CMDLINE_V can be used simultaneously (as they should have been). Also with this patch, items which a user duplicates in the stack (beyond the first such item) will return the the string "[ duplicate ENUM_ID ]". This practice will apply to both single strings and true vectorized ones. In addition to informing users of their error, it will also mean potential NULLs need now never be a concern. Reference(s); http://www.freelists.org/post/procps/systemd-binary-vs-library commit 0580a7b4c67d0297629d37281b4f690894429626 Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-19 10:30:00 +05:30
{ RS(ID_TGID), 0, NULL, QS(s_int), 0 }, // oldflags: free w/ simple_nextpid
{ RS(ID_TPGID), f_stat, NULL, QS(s_int), 0 },
library: stop treating lxcname differently, <PIDS> api Unlike other cached string fields (such as USER name), that lxcname field was defined as 'const char *'. This required a custom 'set' function using a cast in order to avoid a compiler warning regarding that assignment. So this commit brings lxcname into line with all those other cached string fields. And while we're at it, the reason some string fields have no 'freefunc' will also be explained with some new comments in the Item_table. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-01 10:30:00 +05:30
{ RS(LXCNAME), f_lxc, NULL, QS(str), 0 }, // freefunc NULL w/ cached string
library: eliminated a final potential NULL, <PIDS> api In that reference below a specific systemd problem was fixed in the commit shown. However lurking deep within the <pids> interface was yet one final case where NULL could be returned, involving 'strv' and the following: . a user requested both a single string vector (always returned as a normal string) & the vectorized version, as with PROCPS_PIDS_CMDLINE and PROCPS_PIDS_CMDLINE_V. . a user simply duplicated some vectorized enum items. The root of that NULL problem is the fact those single string vectors shared the same proc_t field with their true vectorized version. So while multiple occurrences for most strings could be satisfied with strdup versus the normal ownership usurpation, those true vectorized fields could not be quite so easily copied/duplicated. Thus newlib chose to return a NULL result.strv pointer under either of the above scenarios (which perhaps was just a user boo-boo in the first place). In any event, the NULL was a potential for true string vectors only. Now, since newlib is the sole caller into the readproc module, separate fields have been created for what are just normal strings (never vectorized) and those which remain the true vectorized versions. And, former flags which only worked if combined, now act as stand alone. Thus, both PROCPS_PIDS_CMDLINE & PROCPS_PIDS_CMDLINE_V can be used simultaneously (as they should have been). Also with this patch, items which a user duplicates in the stack (beyond the first such item) will return the the string "[ duplicate ENUM_ID ]". This practice will apply to both single strings and true vectorized ones. In addition to informing users of their error, it will also mean potential NULLs need now never be a concern. Reference(s); http://www.freelists.org/post/procps/systemd-binary-vs-library commit 0580a7b4c67d0297629d37281b4f690894429626 Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-19 10:30:00 +05:30
{ RS(MEM_CODE), f_statm, NULL, QS(sl_int), 0 },
{ RS(MEM_CODE_KIB), f_statm, NULL, QS(sl_int), 0 },
{ RS(MEM_DATA), f_statm, NULL, QS(sl_int), 0 },
{ RS(MEM_DATA_KIB), f_statm, NULL, QS(sl_int), 0 },
{ RS(MEM_DT), f_statm, NULL, QS(sl_int), 0 },
{ RS(MEM_LRS), f_statm, NULL, QS(sl_int), 0 },
{ RS(MEM_RES), f_statm, NULL, QS(sl_int), 0 },
{ RS(MEM_RES_KIB), f_statm, NULL, QS(sl_int), 0 },
{ RS(MEM_SHR), f_statm, NULL, QS(sl_int), 0 },
{ RS(MEM_SHR_KIB), f_statm, NULL, QS(sl_int), 0 },
{ RS(MEM_VIRT), f_statm, NULL, QS(sl_int), 0 },
{ RS(MEM_VIRT_KIB), f_statm, NULL, QS(sl_int), 0 },
{ RS(NICE), f_stat, NULL, QS(sl_int), 0 },
{ RS(NLWP), f_either, NULL, QS(s_int), 0 },
{ RS(NS_IPC), f_ns, NULL, QS(ul_int), 0 },
{ RS(NS_MNT), f_ns, NULL, QS(ul_int), 0 },
{ RS(NS_NET), f_ns, NULL, QS(ul_int), 0 },
{ RS(NS_PID), f_ns, NULL, QS(ul_int), 0 },
{ RS(NS_USER), f_ns, NULL, QS(ul_int), 0 },
{ RS(NS_UTS), f_ns, NULL, QS(ul_int), 0 },
{ RS(OOM_ADJ), f_oom, NULL, QS(s_int), 0 },
{ RS(OOM_SCORE), f_oom, NULL, QS(s_int), 0 },
{ RS(PRIORITY), f_stat, NULL, QS(s_int), 0 },
{ RS(PROCESSOR), f_stat, NULL, QS(u_int), 0 },
{ RS(RSS), f_stat, NULL, QS(sl_int), 0 },
{ RS(RSS_RLIM), f_stat, NULL, QS(ul_int), 0 },
{ RS(RTPRIO), f_stat, NULL, QS(ul_int), 0 },
{ RS(SCHED_CLASS), f_stat, NULL, QS(ul_int), 0 },
{ RS(SD_MACH), f_systemd, FF(str), QS(str), 0 },
{ RS(SD_OUID), f_systemd, FF(str), QS(str), 0 },
{ RS(SD_SEAT), f_systemd, FF(str), QS(str), 0 },
{ RS(SD_SESS), f_systemd, FF(str), QS(str), 0 },
{ RS(SD_SLICE), f_systemd, FF(str), QS(str), 0 },
{ RS(SD_UNIT), f_systemd, FF(str), QS(str), 0 },
{ RS(SD_UUNIT), f_systemd, FF(str), QS(str), 0 },
{ RS(SIGBLOCKED), f_status, FF(str), QS(str), 0 },
{ RS(SIGCATCH), f_status, FF(str), QS(str), 0 },
{ RS(SIGIGNORE), f_status, FF(str), QS(str), 0 },
{ RS(SIGNALS), f_status, FF(str), QS(str), 0 },
{ RS(SIGPENDING), f_status, FF(str), QS(str), 0 },
{ RS(STATE), f_either, NULL, QS(s_ch), 0 },
{ RS(SUPGIDS), f_status, FF(str), QS(str), 0 },
{ RS(SUPGROUPS), x_supgrp, FF(str), QS(str), 0 },
{ RS(TICS_ALL), f_stat, NULL, QS(ull_int), 0 },
{ RS(TICS_ALL_C), f_stat, NULL, QS(ull_int), 0 },
{ RS(TICS_DELTA), f_stat, NULL, QS(sl_int), +1 },
{ RS(TICS_SYSTEM), f_stat, NULL, QS(ull_int), 0 },
{ RS(TICS_SYSTEM_C), f_stat, NULL, QS(ull_int), 0 },
{ RS(TICS_USER), f_stat, NULL, QS(ull_int), 0 },
{ RS(TICS_USER_C), f_stat, NULL, QS(ull_int), 0 },
{ RS(TIME_ALL), f_stat, NULL, QS(ull_int), 0 },
{ RS(TIME_ELAPSED), f_stat, NULL, QS(ull_int), 0 },
{ RS(TIME_START), f_stat, NULL, QS(ull_int), 0 },
{ RS(TTY), f_stat, NULL, QS(s_int), 0 },
{ RS(TTY_NAME), f_stat, FF(str), QS(strvers), 0 },
{ RS(TTY_NUMBER), f_stat, FF(str), QS(strvers), 0 },
{ RS(VM_DATA), f_status, NULL, QS(sl_int), 0 },
{ RS(VM_EXE), f_status, NULL, QS(sl_int), 0 },
{ RS(VM_LIB), f_status, NULL, QS(sl_int), 0 },
{ RS(VM_RSS), f_status, NULL, QS(sl_int), 0 },
{ RS(VM_RSS_ANON), f_status, NULL, QS(sl_int), 0 },
{ RS(VM_RSS_FILE), f_status, NULL, QS(sl_int), 0 },
{ RS(VM_RSS_LOCKED), f_status, NULL, QS(sl_int), 0 },
{ RS(VM_RSS_SHARED), f_status, NULL, QS(sl_int), 0 },
{ RS(VM_SIZE), f_status, NULL, QS(sl_int), 0 },
{ RS(VM_STACK), f_status, NULL, QS(sl_int), 0 },
{ RS(VM_SWAP), f_status, NULL, QS(sl_int), 0 },
{ RS(VM_USED), f_status, NULL, QS(sl_int), 0 },
{ RS(VSIZE_PGS), f_stat, NULL, QS(ul_int), 0 },
{ RS(WCHAN_ADDR), f_stat, NULL, QS(ul_int), 0 },
{ RS(WCHAN_NAME), 0, FF(str), QS(str), 0 }, // oldflags: tid already free
library: final tweaks to code and/or comments, 3rd gen With the dust now settling on all those 3rd generation upgrades, this patch tries to provide some consistency among the separate modules involved. Someday we should consider a 4th generation where all redundant code has been removed and isolated in a new shared source file. Following is a summary of significant changes (if any) to each of these now upgraded 3rd gen library modules. <meminfo> ............................................ . strictly formatting/comment changes, code unaffected <pids> ............................................... . replaced a local mkSTR macro with existing STRINGIFY . added fetch narrative explaining duplicate addresses <slabinfo> ........................................... . rearranged some free logic for procps_slabinfo_unref . added fetch narrative explaining duplicate addresses <stat> ............................................... . added #define ENFORCE_LOGICAL, just as in <slabinfo> . replaced a local mkSTR macro with existing STRINGIFY . alphabetized the function declarations in the header <vmstat> ............................................. . made one coverity concession with read_vmstat_failed [ several of these changes may reflect this author's ] [ continuing pursuit of an unreasonable goal -- that ] [ of a 'perfect' (plus 'pretty') C language program! ] Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-09 10:30:00 +05:30
library: fix boo-boo left by earlier patch, <PIDS> api After the commit referenced below the potential exists for a SEGV (resulting from an out-of-bounds Item_table reference if PROCPS_PIDS_physical_end is encountered). So this patch eliminates that PROCPS_PIDS_physical_end as no longer necessary and completes the task of using PROCPS_PIDS_logical_end as a sole necessary fencepost. Reference(s): commit e7585992d9c0743246247b3d6ee0041942fe07d5
2016-04-25 10:30:00 +05:30
// dummy entry corresponding to PROCPS_PIDS_logical_end ...
library: eliminated a final potential NULL, <PIDS> api In that reference below a specific systemd problem was fixed in the commit shown. However lurking deep within the <pids> interface was yet one final case where NULL could be returned, involving 'strv' and the following: . a user requested both a single string vector (always returned as a normal string) & the vectorized version, as with PROCPS_PIDS_CMDLINE and PROCPS_PIDS_CMDLINE_V. . a user simply duplicated some vectorized enum items. The root of that NULL problem is the fact those single string vectors shared the same proc_t field with their true vectorized version. So while multiple occurrences for most strings could be satisfied with strdup versus the normal ownership usurpation, those true vectorized fields could not be quite so easily copied/duplicated. Thus newlib chose to return a NULL result.strv pointer under either of the above scenarios (which perhaps was just a user boo-boo in the first place). In any event, the NULL was a potential for true string vectors only. Now, since newlib is the sole caller into the readproc module, separate fields have been created for what are just normal strings (never vectorized) and those which remain the true vectorized versions. And, former flags which only worked if combined, now act as stand alone. Thus, both PROCPS_PIDS_CMDLINE & PROCPS_PIDS_CMDLINE_V can be used simultaneously (as they should have been). Also with this patch, items which a user duplicates in the stack (beyond the first such item) will return the the string "[ duplicate ENUM_ID ]". This practice will apply to both single strings and true vectorized ones. In addition to informing users of their error, it will also mean potential NULLs need now never be a concern. Reference(s); http://www.freelists.org/post/procps/systemd-binary-vs-library commit 0580a7b4c67d0297629d37281b4f690894429626 Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-19 10:30:00 +05:30
{ NULL, 0, NULL, NULL, 0 }
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
};
library: the overlooked twerks (oops, tweaks), 3rd gen Yes, all of these changes are strictly cosmetic. It is likely symptomatic of some deep-seated character flaw. [ or, it might be because of a certain pride in this ] [ new library and the desire to make it even better! ] Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-07 10:30:00 +05:30
/* please note,
* this enum MUST be 1 greater than the highest value of any enum */
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
enum pids_item PROCPS_PIDS_logical_end = PROCPS_PIDS_WCHAN_NAME + 1;
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
#undef setNAME
library: add real string vectors to the pids interface After experimenting with an adaptation of pidof to the new pids interface, it became apparent that vectorized versions of those command lines would be necessary. So this commit adds that option and the strv result type. And since the stage had been set, a vectorized version of PROCPS_PIDS_ENVIRON & PROCPS_PIDS_CGROUP was added. Lastly, any use of 'const' in the result structure was removed so callers need not be bothered with casts and compiler warnings. Hopefully, they'll respect a stack. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-26 10:30:00 +05:30
#undef freNAME
#undef srtNAME
library: misc tweaks for code and comments, <PIDS> api This commit simply tries to keep naming plus formating conventions on a par with the continuing climb up that learning curve. These changes were suggested following <slabinfo> sources upgrade from 2nd to 3rd generation. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-06 10:30:00 +05:30
#undef RS
#undef FF
#undef QS
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
//#undef f_either // needed later
#undef f_grp
#undef f_lxc
#undef f_ns
#undef f_oom
//#undef f_stat // needed later
#undef f_statm
//#undef f_status // needed later
#undef f_systemd
#undef f_usr
library: add real string vectors to the pids interface After experimenting with an adaptation of pidof to the new pids interface, it became apparent that vectorized versions of those command lines would be necessary. So this commit adds that option and the strv result type. And since the stage had been set, a vectorized version of PROCPS_PIDS_ENVIRON & PROCPS_PIDS_CGROUP was added. Lastly, any use of 'const' in the result structure was removed so callers need not be bothered with casts and compiler warnings. Hopefully, they'll respect a stack. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-26 10:30:00 +05:30
#undef v_arg
#undef v_cgroup
#undef v_env
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
#undef x_cgroup
#undef x_cmdline
#undef x_environ
#undef x_ogroup
#undef x_ouser
#undef x_supgrp
// ___ History Support Private Functions ||||||||||||||||||||||||||||||||||||||
// ( stolen from top when he wasn't looking ) -------------------------------
#define HHASH_SIZE 1024
#define _HASH_PID_(K) (K & (HHASH_SIZE - 1))
#define Hr(x) info->hist->x // 'hist ref', minimize stolen impact
typedef unsigned long long TIC_t;
typedef struct HST_t {
TIC_t tics; // last frame's tics count
library: try to minimize the results types, <PIDS> api This is just beginning an effort to minimize/normalize the sheer variety of results types in use for our API. In taking these first baby steps, a few anomalies were found. There'll no doubt be many more yet to discover. . the _FLT_ (fault) fields were already signed long in the proc_t (even though their sscanf format used %lu). . although strtoul will alway return an unsigned long, all of the _VM_ fields were made signed long just like other memory fields (& signed sorts more efficiently). Reference(s): http://www.freelists.org/post/procps/newlib-drip-drip,4 Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-15 10:30:00 +05:30
long maj, min; // last frame's maj/min_flt counts
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
int pid; // record 'key'
int lnk; // next on hash chain
} HST_t;
struct history_info {
int num_tasks; // used as index (tasks tallied)
int HHist_siz; // max number of HST_t structs
HST_t *PHist_sav; // alternating 'old/new' HST_t anchors
HST_t *PHist_new;
int HHash_one [HHASH_SIZE]; // the actual hash tables
int HHash_two [HHASH_SIZE]; // (accessed via PHash_sav/PHash_new)
int HHash_nul [HHASH_SIZE]; // an 'empty' hash table image
int *PHash_sav; // alternating 'old/new' hash tables
int *PHash_new; // (aka. the 'one/two' actual tables)
};
static void config_history (
struct procps_pidsinfo *info)
{
int i;
for (i = 0; i < HHASH_SIZE; i++) // make the 'empty' table image
Hr(HHash_nul[i]) = -1;
memcpy(Hr(HHash_one), Hr(HHash_nul), sizeof(Hr(HHash_nul)));
memcpy(Hr(HHash_two), Hr(HHash_nul), sizeof(Hr(HHash_nul)));
Hr(PHash_sav) = Hr(HHash_one); // alternating 'old/new' hash tables
Hr(PHash_new) = Hr(HHash_two);
} // end: config_history
static inline HST_t *histget (
struct procps_pidsinfo *info,
int pid)
{
int V = Hr(PHash_sav[_HASH_PID_(pid)]);
while (-1 < V) {
if (Hr(PHist_sav[V].pid) == pid)
return &Hr(PHist_sav[V]);
library: the overlooked twerks (oops, tweaks), 3rd gen Yes, all of these changes are strictly cosmetic. It is likely symptomatic of some deep-seated character flaw. [ or, it might be because of a certain pride in this ] [ new library and the desire to make it even better! ] Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-07 10:30:00 +05:30
V = Hr(PHist_sav[V].lnk);
}
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
return NULL;
} // end: histget
static inline void histput (
struct procps_pidsinfo *info,
unsigned this)
{
int V = _HASH_PID_(Hr(PHist_new[this].pid));
Hr(PHist_new[this].lnk) = Hr(PHash_new[V]);
Hr(PHash_new[V] = this);
} // end: histput
#undef _HASH_PID_
static int make_hist (
struct procps_pidsinfo *info,
proc_t *p)
{
library: miscellaneous tweaks to pid code and comments . traded a complex misaligned memory allocation scheme in the make_hist function for a simple aligned scheme. plus memory allocation increases are globally defined. . changed 1 parameter for procps_pids_stacks_sort() to better reflect the 'array of pointers', not an address of a pointer as is used with guys such as 'new/unref'. . the pids_reap struct was changed slightly to make it more reflective of it's actual implementation details. . the Item_table member .mustfree is now .needfree and that .makehist was now made .needhist for consistency. . reduced the number of separate 'return NULL;' source statements in that primary procps_pids_reap() routine. . ensured consistent reference to sizeof(void *) & not occasional reference to sizeof(void*) without a space. . rather than enable/disable validate_stacks via a #if in the function body, it is now handled via a #define. . some comments in the procps_pids_reset function were adjusted to reflect this current implementation. shown originally, they reflected an aborted attempt to avoid a testing aberration not fully understood at the time. . added a summary of the memory overhead cost of HST_t processing to that UNREF_RPTHASH output at unref time. . a 'PIDs at max depth:' portion of that UNREF_RPTHASH enabled #define is now published only when the maximum depth of hash table entry chains exceed depths of one. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-20 10:30:00 +05:30
#define nSLOT info->hist->num_tasks
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
TIC_t tics;
HST_t *h;
library: miscellaneous tweaks to pid code and comments . traded a complex misaligned memory allocation scheme in the make_hist function for a simple aligned scheme. plus memory allocation increases are globally defined. . changed 1 parameter for procps_pids_stacks_sort() to better reflect the 'array of pointers', not an address of a pointer as is used with guys such as 'new/unref'. . the pids_reap struct was changed slightly to make it more reflective of it's actual implementation details. . the Item_table member .mustfree is now .needfree and that .makehist was now made .needhist for consistency. . reduced the number of separate 'return NULL;' source statements in that primary procps_pids_reap() routine. . ensured consistent reference to sizeof(void *) & not occasional reference to sizeof(void*) without a space. . rather than enable/disable validate_stacks via a #if in the function body, it is now handled via a #define. . some comments in the procps_pids_reset function were adjusted to reflect this current implementation. shown originally, they reflected an aborted attempt to avoid a testing aberration not fully understood at the time. . added a summary of the memory overhead cost of HST_t processing to that UNREF_RPTHASH output at unref time. . a 'PIDs at max depth:' portion of that UNREF_RPTHASH enabled #define is now published only when the maximum depth of hash table entry chains exceed depths of one. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-20 10:30:00 +05:30
if (nSLOT + 1 >= Hr(HHist_siz)) {
Hr(HHist_siz) += MEMORY_INCR;
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
Hr(PHist_sav) = realloc(Hr(PHist_sav), sizeof(HST_t) * Hr(HHist_siz));
Hr(PHist_new) = realloc(Hr(PHist_new), sizeof(HST_t) * Hr(HHist_siz));
library: accumulated changes to <pids> code & comments A patch containing the following miscellaneous tweaks: . avoided distortions unique to PROCPS_PIDS_TICS_DELTA . addressed several smatch warnings and/or suggestions . ensured oldproc_close invoked should tally_proc fail . keeping that namespace clean, added a missing #undef . added 2 comments acknowledging pids_item as unsigned . added/clarified comments regarding proc flags & strv From smatch analysis: . these were indeed boo-boos pids.c:580 make_hist() warn: the 'Hr' macro might need parens pids.c:1058 procps_pids_reap() warn: variable dereferenced before check 'info' (see line 1056) . these were not errors (and we did double check) pids.c:1067 procps_pids_reap() warn: double check that we're allocating correct size: 8 vs 128 pids.c:1068 procps_pids_reap() warn: double check that we're allocating correct size: 8 vs 128 Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-28 10:30:00 +05:30
if (!Hr(PHist_sav) || !Hr(PHist_new))
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
return -ENOMEM;
}
library: miscellaneous tweaks to pid code and comments . traded a complex misaligned memory allocation scheme in the make_hist function for a simple aligned scheme. plus memory allocation increases are globally defined. . changed 1 parameter for procps_pids_stacks_sort() to better reflect the 'array of pointers', not an address of a pointer as is used with guys such as 'new/unref'. . the pids_reap struct was changed slightly to make it more reflective of it's actual implementation details. . the Item_table member .mustfree is now .needfree and that .makehist was now made .needhist for consistency. . reduced the number of separate 'return NULL;' source statements in that primary procps_pids_reap() routine. . ensured consistent reference to sizeof(void *) & not occasional reference to sizeof(void*) without a space. . rather than enable/disable validate_stacks via a #if in the function body, it is now handled via a #define. . some comments in the procps_pids_reset function were adjusted to reflect this current implementation. shown originally, they reflected an aborted attempt to avoid a testing aberration not fully understood at the time. . added a summary of the memory overhead cost of HST_t processing to that UNREF_RPTHASH output at unref time. . a 'PIDs at max depth:' portion of that UNREF_RPTHASH enabled #define is now published only when the maximum depth of hash table entry chains exceed depths of one. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-20 10:30:00 +05:30
Hr(PHist_new[nSLOT].pid) = p->tid;
Hr(PHist_new[nSLOT].tics) = tics = (p->utime + p->stime);
Hr(PHist_new[nSLOT].maj) = p->maj_flt;
Hr(PHist_new[nSLOT].min) = p->min_flt;
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
library: miscellaneous tweaks to pid code and comments . traded a complex misaligned memory allocation scheme in the make_hist function for a simple aligned scheme. plus memory allocation increases are globally defined. . changed 1 parameter for procps_pids_stacks_sort() to better reflect the 'array of pointers', not an address of a pointer as is used with guys such as 'new/unref'. . the pids_reap struct was changed slightly to make it more reflective of it's actual implementation details. . the Item_table member .mustfree is now .needfree and that .makehist was now made .needhist for consistency. . reduced the number of separate 'return NULL;' source statements in that primary procps_pids_reap() routine. . ensured consistent reference to sizeof(void *) & not occasional reference to sizeof(void*) without a space. . rather than enable/disable validate_stacks via a #if in the function body, it is now handled via a #define. . some comments in the procps_pids_reset function were adjusted to reflect this current implementation. shown originally, they reflected an aborted attempt to avoid a testing aberration not fully understood at the time. . added a summary of the memory overhead cost of HST_t processing to that UNREF_RPTHASH output at unref time. . a 'PIDs at max depth:' portion of that UNREF_RPTHASH enabled #define is now published only when the maximum depth of hash table entry chains exceed depths of one. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-20 10:30:00 +05:30
histput(info, nSLOT);
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
if ((h = histget(info, p->tid))) {
tics -= h->tics;
library: accumulated changes to <pids> code & comments A patch containing the following miscellaneous tweaks: . avoided distortions unique to PROCPS_PIDS_TICS_DELTA . addressed several smatch warnings and/or suggestions . ensured oldproc_close invoked should tally_proc fail . keeping that namespace clean, added a missing #undef . added 2 comments acknowledging pids_item as unsigned . added/clarified comments regarding proc flags & strv From smatch analysis: . these were indeed boo-boos pids.c:580 make_hist() warn: the 'Hr' macro might need parens pids.c:1058 procps_pids_reap() warn: variable dereferenced before check 'info' (see line 1056) . these were not errors (and we did double check) pids.c:1067 procps_pids_reap() warn: double check that we're allocating correct size: 8 vs 128 pids.c:1068 procps_pids_reap() warn: double check that we're allocating correct size: 8 vs 128 Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-28 10:30:00 +05:30
p->pcpu = tics;
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
p->maj_delta = p->maj_flt - h->maj;
p->min_delta = p->min_flt - h->min;
}
library: miscellaneous tweaks to pid code and comments . traded a complex misaligned memory allocation scheme in the make_hist function for a simple aligned scheme. plus memory allocation increases are globally defined. . changed 1 parameter for procps_pids_stacks_sort() to better reflect the 'array of pointers', not an address of a pointer as is used with guys such as 'new/unref'. . the pids_reap struct was changed slightly to make it more reflective of it's actual implementation details. . the Item_table member .mustfree is now .needfree and that .makehist was now made .needhist for consistency. . reduced the number of separate 'return NULL;' source statements in that primary procps_pids_reap() routine. . ensured consistent reference to sizeof(void *) & not occasional reference to sizeof(void*) without a space. . rather than enable/disable validate_stacks via a #if in the function body, it is now handled via a #define. . some comments in the procps_pids_reset function were adjusted to reflect this current implementation. shown originally, they reflected an aborted attempt to avoid a testing aberration not fully understood at the time. . added a summary of the memory overhead cost of HST_t processing to that UNREF_RPTHASH output at unref time. . a 'PIDs at max depth:' portion of that UNREF_RPTHASH enabled #define is now published only when the maximum depth of hash table entry chains exceed depths of one. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-20 10:30:00 +05:30
nSLOT++;
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
return 0;
library: miscellaneous tweaks to pid code and comments . traded a complex misaligned memory allocation scheme in the make_hist function for a simple aligned scheme. plus memory allocation increases are globally defined. . changed 1 parameter for procps_pids_stacks_sort() to better reflect the 'array of pointers', not an address of a pointer as is used with guys such as 'new/unref'. . the pids_reap struct was changed slightly to make it more reflective of it's actual implementation details. . the Item_table member .mustfree is now .needfree and that .makehist was now made .needhist for consistency. . reduced the number of separate 'return NULL;' source statements in that primary procps_pids_reap() routine. . ensured consistent reference to sizeof(void *) & not occasional reference to sizeof(void*) without a space. . rather than enable/disable validate_stacks via a #if in the function body, it is now handled via a #define. . some comments in the procps_pids_reset function were adjusted to reflect this current implementation. shown originally, they reflected an aborted attempt to avoid a testing aberration not fully understood at the time. . added a summary of the memory overhead cost of HST_t processing to that UNREF_RPTHASH output at unref time. . a 'PIDs at max depth:' portion of that UNREF_RPTHASH enabled #define is now published only when the maximum depth of hash table entry chains exceed depths of one. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-20 10:30:00 +05:30
#undef nSLOT
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
} // end: make_hist
static inline void toggle_history (
struct procps_pidsinfo *info)
{
void *v;
v = Hr(PHist_sav);
Hr(PHist_sav) = Hr(PHist_new);
Hr(PHist_new) = v;
v = Hr(PHash_sav);
Hr(PHash_sav) = Hr(PHash_new);
Hr(PHash_new) = v;
memcpy(Hr(PHash_new), Hr(HHash_nul), sizeof(Hr(HHash_nul)));
info->hist->num_tasks = 0;
} // end: toggle_history
#ifdef UNREF_RPTHASH
static void unref_rpthash (
struct procps_pidsinfo *info)
{
int i, j, pop, total_occupied, maxdepth, maxdepth_sav, numdepth
library: miscellaneous tweaks to pid code and comments . traded a complex misaligned memory allocation scheme in the make_hist function for a simple aligned scheme. plus memory allocation increases are globally defined. . changed 1 parameter for procps_pids_stacks_sort() to better reflect the 'array of pointers', not an address of a pointer as is used with guys such as 'new/unref'. . the pids_reap struct was changed slightly to make it more reflective of it's actual implementation details. . the Item_table member .mustfree is now .needfree and that .makehist was now made .needhist for consistency. . reduced the number of separate 'return NULL;' source statements in that primary procps_pids_reap() routine. . ensured consistent reference to sizeof(void *) & not occasional reference to sizeof(void*) without a space. . rather than enable/disable validate_stacks via a #if in the function body, it is now handled via a #define. . some comments in the procps_pids_reset function were adjusted to reflect this current implementation. shown originally, they reflected an aborted attempt to avoid a testing aberration not fully understood at the time. . added a summary of the memory overhead cost of HST_t processing to that UNREF_RPTHASH output at unref time. . a 'PIDs at max depth:' portion of that UNREF_RPTHASH enabled #define is now published only when the maximum depth of hash table entry chains exceed depths of one. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-20 10:30:00 +05:30
, cross_foot, sz = HHASH_SIZE * (int)sizeof(int)
, hsz = (int)sizeof(HST_t) * Hr(HHist_siz);
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
int depths[HHASH_SIZE];
for (i = 0, total_occupied = 0, maxdepth = 0; i < HHASH_SIZE; i++) {
int V = Hr(PHash_new[i]);
j = 0;
if (-1 < V) {
++total_occupied;
while (-1 < V) {
V = Hr(PHist_new[V].lnk);
if (-1 < V) j++;
}
}
depths[i] = j;
if (maxdepth < j) maxdepth = j;
}
maxdepth_sav = maxdepth;
fprintf(stderr,
library: miscellaneous tweaks to pid code and comments . traded a complex misaligned memory allocation scheme in the make_hist function for a simple aligned scheme. plus memory allocation increases are globally defined. . changed 1 parameter for procps_pids_stacks_sort() to better reflect the 'array of pointers', not an address of a pointer as is used with guys such as 'new/unref'. . the pids_reap struct was changed slightly to make it more reflective of it's actual implementation details. . the Item_table member .mustfree is now .needfree and that .makehist was now made .needhist for consistency. . reduced the number of separate 'return NULL;' source statements in that primary procps_pids_reap() routine. . ensured consistent reference to sizeof(void *) & not occasional reference to sizeof(void*) without a space. . rather than enable/disable validate_stacks via a #if in the function body, it is now handled via a #define. . some comments in the procps_pids_reset function were adjusted to reflect this current implementation. shown originally, they reflected an aborted attempt to avoid a testing aberration not fully understood at the time. . added a summary of the memory overhead cost of HST_t processing to that UNREF_RPTHASH output at unref time. . a 'PIDs at max depth:' portion of that UNREF_RPTHASH enabled #define is now published only when the maximum depth of hash table entry chains exceed depths of one. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-20 10:30:00 +05:30
"\n History Memory Costs:"
"\n\tHST_t size = %d, total allocated = %d,"
"\n\tthus PHist_new & PHist_sav consumed %dk (%d) total bytes."
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
"\n"
library: miscellaneous tweaks to pid code and comments . traded a complex misaligned memory allocation scheme in the make_hist function for a simple aligned scheme. plus memory allocation increases are globally defined. . changed 1 parameter for procps_pids_stacks_sort() to better reflect the 'array of pointers', not an address of a pointer as is used with guys such as 'new/unref'. . the pids_reap struct was changed slightly to make it more reflective of it's actual implementation details. . the Item_table member .mustfree is now .needfree and that .makehist was now made .needhist for consistency. . reduced the number of separate 'return NULL;' source statements in that primary procps_pids_reap() routine. . ensured consistent reference to sizeof(void *) & not occasional reference to sizeof(void*) without a space. . rather than enable/disable validate_stacks via a #if in the function body, it is now handled via a #define. . some comments in the procps_pids_reset function were adjusted to reflect this current implementation. shown originally, they reflected an aborted attempt to avoid a testing aberration not fully understood at the time. . added a summary of the memory overhead cost of HST_t processing to that UNREF_RPTHASH output at unref time. . a 'PIDs at max depth:' portion of that UNREF_RPTHASH enabled #define is now published only when the maximum depth of hash table entry chains exceed depths of one. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-20 10:30:00 +05:30
"\n\tTwo hash tables provide for %d entries each + 1 extra 'empty' image,"
"\n\tthus %dk (%d) bytes per table for %dk (%d) total bytes."
"\n"
"\n\tGrand total = %dk (%d) bytes."
"\n"
"\n Hash Results Report:"
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
"\n\tTotal hashed = %d"
"\n\tLevel-0 hash entries = %d (%d%% occupied)"
"\n\tMax Depth = %d"
"\n\n"
library: miscellaneous tweaks to pid code and comments . traded a complex misaligned memory allocation scheme in the make_hist function for a simple aligned scheme. plus memory allocation increases are globally defined. . changed 1 parameter for procps_pids_stacks_sort() to better reflect the 'array of pointers', not an address of a pointer as is used with guys such as 'new/unref'. . the pids_reap struct was changed slightly to make it more reflective of it's actual implementation details. . the Item_table member .mustfree is now .needfree and that .makehist was now made .needhist for consistency. . reduced the number of separate 'return NULL;' source statements in that primary procps_pids_reap() routine. . ensured consistent reference to sizeof(void *) & not occasional reference to sizeof(void*) without a space. . rather than enable/disable validate_stacks via a #if in the function body, it is now handled via a #define. . some comments in the procps_pids_reset function were adjusted to reflect this current implementation. shown originally, they reflected an aborted attempt to avoid a testing aberration not fully understood at the time. . added a summary of the memory overhead cost of HST_t processing to that UNREF_RPTHASH output at unref time. . a 'PIDs at max depth:' portion of that UNREF_RPTHASH enabled #define is now published only when the maximum depth of hash table entry chains exceed depths of one. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-20 10:30:00 +05:30
, (int)sizeof(HST_t), Hr(HHist_siz)
, hsz / 1024, hsz
, HHASH_SIZE
, sz / 1024, sz, (sz * 3) / 1024, sz * 3
, (hsz + (sz * 3)) / 1024, hsz + (sz * 3)
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
, info->hist->num_tasks
, total_occupied, (total_occupied * 100) / HHASH_SIZE
library: miscellaneous tweaks to pid code and comments . traded a complex misaligned memory allocation scheme in the make_hist function for a simple aligned scheme. plus memory allocation increases are globally defined. . changed 1 parameter for procps_pids_stacks_sort() to better reflect the 'array of pointers', not an address of a pointer as is used with guys such as 'new/unref'. . the pids_reap struct was changed slightly to make it more reflective of it's actual implementation details. . the Item_table member .mustfree is now .needfree and that .makehist was now made .needhist for consistency. . reduced the number of separate 'return NULL;' source statements in that primary procps_pids_reap() routine. . ensured consistent reference to sizeof(void *) & not occasional reference to sizeof(void*) without a space. . rather than enable/disable validate_stacks via a #if in the function body, it is now handled via a #define. . some comments in the procps_pids_reset function were adjusted to reflect this current implementation. shown originally, they reflected an aborted attempt to avoid a testing aberration not fully understood at the time. . added a summary of the memory overhead cost of HST_t processing to that UNREF_RPTHASH output at unref time. . a 'PIDs at max depth:' portion of that UNREF_RPTHASH enabled #define is now published only when the maximum depth of hash table entry chains exceed depths of one. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-20 10:30:00 +05:30
, maxdepth);
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
if (total_occupied) {
for (pop = total_occupied, cross_foot = 0; maxdepth; maxdepth--) {
for (i = 0, numdepth = 0; i < HHASH_SIZE; i++)
if (depths[i] == maxdepth) ++numdepth;
fprintf(stderr,
"\t %5d (%3d%%) hash table entries at depth %d\n"
library: miscellaneous tweaks to pid code and comments . traded a complex misaligned memory allocation scheme in the make_hist function for a simple aligned scheme. plus memory allocation increases are globally defined. . changed 1 parameter for procps_pids_stacks_sort() to better reflect the 'array of pointers', not an address of a pointer as is used with guys such as 'new/unref'. . the pids_reap struct was changed slightly to make it more reflective of it's actual implementation details. . the Item_table member .mustfree is now .needfree and that .makehist was now made .needhist for consistency. . reduced the number of separate 'return NULL;' source statements in that primary procps_pids_reap() routine. . ensured consistent reference to sizeof(void *) & not occasional reference to sizeof(void*) without a space. . rather than enable/disable validate_stacks via a #if in the function body, it is now handled via a #define. . some comments in the procps_pids_reset function were adjusted to reflect this current implementation. shown originally, they reflected an aborted attempt to avoid a testing aberration not fully understood at the time. . added a summary of the memory overhead cost of HST_t processing to that UNREF_RPTHASH output at unref time. . a 'PIDs at max depth:' portion of that UNREF_RPTHASH enabled #define is now published only when the maximum depth of hash table entry chains exceed depths of one. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-20 10:30:00 +05:30
, numdepth, (numdepth * 100) / total_occupied, maxdepth);
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
pop -= numdepth;
cross_foot += numdepth;
if (0 == pop && cross_foot == total_occupied) break;
}
if (pop) {
library: miscellaneous tweaks to pid code and comments . traded a complex misaligned memory allocation scheme in the make_hist function for a simple aligned scheme. plus memory allocation increases are globally defined. . changed 1 parameter for procps_pids_stacks_sort() to better reflect the 'array of pointers', not an address of a pointer as is used with guys such as 'new/unref'. . the pids_reap struct was changed slightly to make it more reflective of it's actual implementation details. . the Item_table member .mustfree is now .needfree and that .makehist was now made .needhist for consistency. . reduced the number of separate 'return NULL;' source statements in that primary procps_pids_reap() routine. . ensured consistent reference to sizeof(void *) & not occasional reference to sizeof(void*) without a space. . rather than enable/disable validate_stacks via a #if in the function body, it is now handled via a #define. . some comments in the procps_pids_reset function were adjusted to reflect this current implementation. shown originally, they reflected an aborted attempt to avoid a testing aberration not fully understood at the time. . added a summary of the memory overhead cost of HST_t processing to that UNREF_RPTHASH output at unref time. . a 'PIDs at max depth:' portion of that UNREF_RPTHASH enabled #define is now published only when the maximum depth of hash table entry chains exceed depths of one. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-20 10:30:00 +05:30
fprintf(stderr, "\t %5d (%3d%%) unchained entries (at depth 0)\n"
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
, pop, (pop * 100) / total_occupied);
cross_foot += pop;
}
fprintf(stderr,
"\t -----\n"
"\t %5d total entries occupied\n", cross_foot);
library: miscellaneous tweaks to pid code and comments . traded a complex misaligned memory allocation scheme in the make_hist function for a simple aligned scheme. plus memory allocation increases are globally defined. . changed 1 parameter for procps_pids_stacks_sort() to better reflect the 'array of pointers', not an address of a pointer as is used with guys such as 'new/unref'. . the pids_reap struct was changed slightly to make it more reflective of it's actual implementation details. . the Item_table member .mustfree is now .needfree and that .makehist was now made .needhist for consistency. . reduced the number of separate 'return NULL;' source statements in that primary procps_pids_reap() routine. . ensured consistent reference to sizeof(void *) & not occasional reference to sizeof(void*) without a space. . rather than enable/disable validate_stacks via a #if in the function body, it is now handled via a #define. . some comments in the procps_pids_reset function were adjusted to reflect this current implementation. shown originally, they reflected an aborted attempt to avoid a testing aberration not fully understood at the time. . added a summary of the memory overhead cost of HST_t processing to that UNREF_RPTHASH output at unref time. . a 'PIDs at max depth:' portion of that UNREF_RPTHASH enabled #define is now published only when the maximum depth of hash table entry chains exceed depths of one. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-20 10:30:00 +05:30
if (maxdepth_sav > 1) {
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
fprintf(stderr, "\n PIDs at max depth: ");
for (i = 0; i < HHASH_SIZE; i++)
if (depths[i] == maxdepth_sav) {
j = Hr(PHash_new[i]);
fprintf(stderr, "\n\tpos %4d: %05d", i, Hr(PHist_new[j].pid));
while (-1 < j) {
j = Hr(PHist_new[j].lnk);
if (-1 < j) fprintf(stderr, ", %05d", Hr(PHist_new[j].pid));
}
}
fprintf(stderr, "\n");
}
}
} // end: unref_rpthash
#endif // UNREF_RPTHASH
library: accumulated changes to <pids> code & comments A patch containing the following miscellaneous tweaks: . avoided distortions unique to PROCPS_PIDS_TICS_DELTA . addressed several smatch warnings and/or suggestions . ensured oldproc_close invoked should tally_proc fail . keeping that namespace clean, added a missing #undef . added 2 comments acknowledging pids_item as unsigned . added/clarified comments regarding proc flags & strv From smatch analysis: . these were indeed boo-boos pids.c:580 make_hist() warn: the 'Hr' macro might need parens pids.c:1058 procps_pids_reap() warn: variable dereferenced before check 'info' (see line 1056) . these were not errors (and we did double check) pids.c:1067 procps_pids_reap() warn: double check that we're allocating correct size: 8 vs 128 pids.c:1068 procps_pids_reap() warn: double check that we're allocating correct size: 8 vs 128 Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-28 10:30:00 +05:30
#undef Hr
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
#undef HHASH_SIZE
// ___ Standard Private Functions |||||||||||||||||||||||||||||||||||||||||||||
static inline void assign_results (
struct procps_pidsinfo *info,
struct pids_stack *stack,
proc_t *p)
{
struct pids_result *this = stack->head;
for (;;) {
enum pids_item item = this->item;
if (item >= PROCPS_PIDS_logical_end)
break;
library: add real string vectors to the pids interface After experimenting with an adaptation of pidof to the new pids interface, it became apparent that vectorized versions of those command lines would be necessary. So this commit adds that option and the strv result type. And since the stage had been set, a vectorized version of PROCPS_PIDS_ENVIRON & PROCPS_PIDS_CGROUP was added. Lastly, any use of 'const' in the result structure was removed so callers need not be bothered with casts and compiler warnings. Hopefully, they'll respect a stack. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-26 10:30:00 +05:30
Item_table[item].setsfunc(info, this, p);
info->dirty_stacks |= Item_table[item].freefunc ? 1 : 0;
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
++this;
}
return;
} // end: assign_results
static inline void cleanup_stack (
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
struct pids_result *this)
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
{
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
for (;;) {
enum pids_item item = this->item;
if (item >= PROCPS_PIDS_logical_end)
library: protect against a future breakage, <pids> ABI By co-mingling both external/internal identifiers with actual implementation code, potential future additions to our API would have been considerably more difficult. So, this patch will now rely solely on internal/hidden identifiers serving as fenceposts in validation logic. And if the following convention is used for new fields we will maintain that enumerator alphabetic ordering a a little longer (even though 2 user fields now don't): . PROCPS_PIDS_XTRA_FOO_A, PROCPS_PIDS_XTRA_FOO_B, etc. Reference(s): http://www.freelists.org/post/procps/me-too-newlib,7 Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-04-18 10:30:00 +05:30
break;
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
if (Item_table[item].freefunc)
Item_table[item].freefunc(this);
if (item > PROCPS_PIDS_noop)
this->result.ull_int = 0;
++this;
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
}
} // end: cleanup_stack
static inline void cleanup_stacks_all (
struct procps_pidsinfo *info)
{
struct stacks_extent *ext = info->extents;
int i;
while (ext) {
for (i = 0; ext->stacks[i]; i++)
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
cleanup_stack(ext->stacks[i]->head);
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
ext = ext->next;
};
info->dirty_stacks = 0;
} // end: cleanup_stacks_all
library: enable an 'invariant' stack for that pids API After peeking at possible conversion of the ps program it appeared that we may ultimately need the concept of a 'static' pids_stack in suport of look_up_our_self(). In other words, one not impacted by procps_pids_reset. This patch simply sets the stage for that possibility. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-09-10 10:30:00 +05:30
/*
* This routine exists in case we ever want to offer something like
* 'static' or 'invarient' results stacks. By unsplicing an extent
* from the info anchor it will be isolated from future reset/free. */
static struct stacks_extent *extent_cut (
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
struct procps_pidsinfo *info,
struct stacks_extent *ext)
{
struct stacks_extent *p = info->extents;
if (ext) {
if (ext == p) {
info->extents = p->next;
library: enable an 'invariant' stack for that pids API After peeking at possible conversion of the ps program it appeared that we may ultimately need the concept of a 'static' pids_stack in suport of look_up_our_self(). In other words, one not impacted by procps_pids_reset. This patch simply sets the stage for that possibility. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-09-10 10:30:00 +05:30
return ext;
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
}
do {
if (ext == p->next) {
p->next = p->next->next;
library: enable an 'invariant' stack for that pids API After peeking at possible conversion of the ps program it appeared that we may ultimately need the concept of a 'static' pids_stack in suport of look_up_our_self(). In other words, one not impacted by procps_pids_reset. This patch simply sets the stage for that possibility. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-09-10 10:30:00 +05:30
return ext;
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
}
p = p->next;
} while (p);
}
library: enable an 'invariant' stack for that pids API After peeking at possible conversion of the ps program it appeared that we may ultimately need the concept of a 'static' pids_stack in suport of look_up_our_self(). In other words, one not impacted by procps_pids_reset. This patch simply sets the stage for that possibility. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-09-10 10:30:00 +05:30
return NULL;
} // end: extent_cut
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
static void extents_free_all (
struct procps_pidsinfo *info)
library: enable an 'invariant' stack for that pids API After peeking at possible conversion of the ps program it appeared that we may ultimately need the concept of a 'static' pids_stack in suport of look_up_our_self(). In other words, one not impacted by procps_pids_reset. This patch simply sets the stage for that possibility. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-09-10 10:30:00 +05:30
{
library: standardize extents_free_all() logic, 3rd gen As those 3rd generation newlib APIs evolved so too did the extents_free_all() function. Most versions of this function required the callers to first verify that the extents anchor wasn't empty, which was poor etiquette. This simple function should have been much more robust and forgiving. With this commit, it fnally becomes so. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-01 10:30:00 +05:30
while (info->extents) {
struct stacks_extent *p = info->extents;
info->extents = info->extents->next;
free(p);
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
};
} // end: extents_free_all
library: reorder (alphabetize) some things, <PIDS> api Without breaking either ABI or API just rearrange some stuff to provide a little more consistent source file. [ these were prompted by some work on the <stat> API ] [ in preparation for him borrowing some of our code. ] Reference(s): Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-04-25 10:30:00 +05:30
static inline struct pids_result *itemize_stack (
struct pids_result *p,
int depth,
enum pids_item *items)
{
struct pids_result *p_sav = p;
int i;
for (i = 0; i < depth; i++) {
p->item = items[i];
p->result.ull_int = 0;
++p;
}
return p_sav;
} // end: itemize_stack
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
static void itemize_stacks_all (
struct procps_pidsinfo *info)
{
struct stacks_extent *ext = info->extents;
while (ext) {
int i;
for (i = 0; ext->stacks[i]; i++)
itemize_stack(ext->stacks[i]->head, info->curitems, info->items);
ext = ext->next;
};
info->dirty_stacks = 0;
library: the overlooked twerks (oops, tweaks), 3rd gen Yes, all of these changes are strictly cosmetic. It is likely symptomatic of some deep-seated character flaw. [ or, it might be because of a certain pride in this ] [ new library and the desire to make it even better! ] Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-07 10:30:00 +05:30
} // end: itemize_stacks_all
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
library: miscellaneous tweaks to pid code and comments . traded a complex misaligned memory allocation scheme in the make_hist function for a simple aligned scheme. plus memory allocation increases are globally defined. . changed 1 parameter for procps_pids_stacks_sort() to better reflect the 'array of pointers', not an address of a pointer as is used with guys such as 'new/unref'. . the pids_reap struct was changed slightly to make it more reflective of it's actual implementation details. . the Item_table member .mustfree is now .needfree and that .makehist was now made .needhist for consistency. . reduced the number of separate 'return NULL;' source statements in that primary procps_pids_reap() routine. . ensured consistent reference to sizeof(void *) & not occasional reference to sizeof(void*) without a space. . rather than enable/disable validate_stacks via a #if in the function body, it is now handled via a #define. . some comments in the procps_pids_reset function were adjusted to reflect this current implementation. shown originally, they reflected an aborted attempt to avoid a testing aberration not fully understood at the time. . added a summary of the memory overhead cost of HST_t processing to that UNREF_RPTHASH output at unref time. . a 'PIDs at max depth:' portion of that UNREF_RPTHASH enabled #define is now published only when the maximum depth of hash table entry chains exceed depths of one. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-20 10:30:00 +05:30
static inline int items_check_failed (
library: optional parms protection missing, <PIDS> api When those items were made dynamic at 'new' time, some other functions, previously assured of their presence, failed to verify a 'reset' had acually been requested. This commit just corrects that oversight and avoids an attempt to 'assign_results' when no items are present. Reference(s): . when items/numitems became optional commit 9ebadc1438a6665a98a9f0782523b0f9a2a6248f Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-15 10:30:00 +05:30
enum pids_item *items,
int numitems)
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
{
int i;
library: beef up 'enum pids_item' parameter validation I was surprised to find that ol' gcc silently converts a single (different) enum into an address where one or more enums were expected to be dereferenced. Of course this was just yet another way to generate an old SEGV. So this commit will strengthen those parameter checks. [ we will *not* blame Craig for a failure to consult ] [ the documentation, since it doesn't even exist yet ] Reference(s): http://www.freelists.org/post/procps/newlib-ps-fix,8 Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-30 10:30:00 +05:30
/* if an enum is passed instead of an address of one or more enums, ol' gcc
* will silently convert it to an address (possibly NULL). only clang will
* offer any sort of warning like the following:
*
* warning: incompatible integer to pointer conversion passing 'int' to parameter of type 'enum pids_item *'
library: respond to coverity (reluctantly), <PIDS> api Calls to free() have now been reintroduce in the new() function to quiet coverity warnings. Those free's were removed originally as that library 'new' was returning with a fatal error and a caller should end abnormally. Plus, it is virtually impossible to fail a malloc call under linux. And lastly, they required braces with the if statement making the code considerably less pretty. [ commit also includes 2 unrelated whitespace tweaks ] Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-17 00:44:14 +05:30
* if (procps_pids_new(&info, PROCPS_PIDS_noop, 3) < 0)
* ^~~~~~~~~~~~~~~~
library: beef up 'enum pids_item' parameter validation I was surprised to find that ol' gcc silently converts a single (different) enum into an address where one or more enums were expected to be dereferenced. Of course this was just yet another way to generate an old SEGV. So this commit will strengthen those parameter checks. [ we will *not* blame Craig for a failure to consult ] [ the documentation, since it doesn't even exist yet ] Reference(s): http://www.freelists.org/post/procps/newlib-ps-fix,8 Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-30 10:30:00 +05:30
*/
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
if (numitems < 1
library: more miscellaneous pids implementation tweaks A patch containing the following miscellaneous tweaks: . make a supposedly robust parameter test truly robust [ ensure the largest enum value used with validation ] . remove duplicate item test in cleanup_stack function [ is already subordinate to test of PROCPS_PIDS_noop ] Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-09-01 10:30:00 +05:30
|| (void *)items < (void *)0x8000) // twice as big as our largest enum
library: beef up 'enum pids_item' parameter validation I was surprised to find that ol' gcc silently converts a single (different) enum into an address where one or more enums were expected to be dereferenced. Of course this was just yet another way to generate an old SEGV. So this commit will strengthen those parameter checks. [ we will *not* blame Craig for a failure to consult ] [ the documentation, since it doesn't even exist yet ] Reference(s): http://www.freelists.org/post/procps/newlib-ps-fix,8 Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-30 10:30:00 +05:30
return -1;
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
for (i = 0; i < numitems; i++) {
library: accumulated changes to <pids> code & comments A patch containing the following miscellaneous tweaks: . avoided distortions unique to PROCPS_PIDS_TICS_DELTA . addressed several smatch warnings and/or suggestions . ensured oldproc_close invoked should tally_proc fail . keeping that namespace clean, added a missing #undef . added 2 comments acknowledging pids_item as unsigned . added/clarified comments regarding proc flags & strv From smatch analysis: . these were indeed boo-boos pids.c:580 make_hist() warn: the 'Hr' macro might need parens pids.c:1058 procps_pids_reap() warn: variable dereferenced before check 'info' (see line 1056) . these were not errors (and we did double check) pids.c:1067 procps_pids_reap() warn: double check that we're allocating correct size: 8 vs 128 pids.c:1068 procps_pids_reap() warn: double check that we're allocating correct size: 8 vs 128 Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-28 10:30:00 +05:30
// a pids_item is currently unsigned, but we'll protect our future
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
if (items[i] < 0)
return -1;
library: protect against a future breakage, <pids> ABI By co-mingling both external/internal identifiers with actual implementation code, potential future additions to our API would have been considerably more difficult. So, this patch will now rely solely on internal/hidden identifiers serving as fenceposts in validation logic. And if the following convention is used for new fields we will maintain that enumerator alphabetic ordering a a little longer (even though 2 user fields now don't): . PROCPS_PIDS_XTRA_FOO_A, PROCPS_PIDS_XTRA_FOO_B, etc. Reference(s): http://www.freelists.org/post/procps/me-too-newlib,7 Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-04-18 10:30:00 +05:30
if (items[i] >= PROCPS_PIDS_logical_end) {
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
return -1;
}
}
return 0;
} // end: items_check_failed
static inline void libflags_set (
struct procps_pidsinfo *info)
{
library: fix boo-boo left by earlier patch, <PIDS> api After the commit referenced below the potential exists for a SEGV (resulting from an out-of-bounds Item_table reference if PROCPS_PIDS_physical_end is encountered). So this patch eliminates that PROCPS_PIDS_physical_end as no longer necessary and completes the task of using PROCPS_PIDS_logical_end as a sole necessary fencepost. Reference(s): commit e7585992d9c0743246247b3d6ee0041942fe07d5
2016-04-25 10:30:00 +05:30
enum pids_item e;
library: eliminated a final potential NULL, <PIDS> api In that reference below a specific systemd problem was fixed in the commit shown. However lurking deep within the <pids> interface was yet one final case where NULL could be returned, involving 'strv' and the following: . a user requested both a single string vector (always returned as a normal string) & the vectorized version, as with PROCPS_PIDS_CMDLINE and PROCPS_PIDS_CMDLINE_V. . a user simply duplicated some vectorized enum items. The root of that NULL problem is the fact those single string vectors shared the same proc_t field with their true vectorized version. So while multiple occurrences for most strings could be satisfied with strdup versus the normal ownership usurpation, those true vectorized fields could not be quite so easily copied/duplicated. Thus newlib chose to return a NULL result.strv pointer under either of the above scenarios (which perhaps was just a user boo-boo in the first place). In any event, the NULL was a potential for true string vectors only. Now, since newlib is the sole caller into the readproc module, separate fields have been created for what are just normal strings (never vectorized) and those which remain the true vectorized versions. And, former flags which only worked if combined, now act as stand alone. Thus, both PROCPS_PIDS_CMDLINE & PROCPS_PIDS_CMDLINE_V can be used simultaneously (as they should have been). Also with this patch, items which a user duplicates in the stack (beyond the first such item) will return the the string "[ duplicate ENUM_ID ]". This practice will apply to both single strings and true vectorized ones. In addition to informing users of their error, it will also mean potential NULLs need now never be a concern. Reference(s); http://www.freelists.org/post/procps/systemd-binary-vs-library commit 0580a7b4c67d0297629d37281b4f690894429626 Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-19 10:30:00 +05:30
int i;
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
info->oldflags = info->history_yes = 0;
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
for (i = 0; i < info->curitems; i++) {
library: fix boo-boo left by earlier patch, <PIDS> api After the commit referenced below the potential exists for a SEGV (resulting from an out-of-bounds Item_table reference if PROCPS_PIDS_physical_end is encountered). So this patch eliminates that PROCPS_PIDS_physical_end as no longer necessary and completes the task of using PROCPS_PIDS_logical_end as a sole necessary fencepost. Reference(s): commit e7585992d9c0743246247b3d6ee0041942fe07d5
2016-04-25 10:30:00 +05:30
if (((e = info->items[i])) >= PROCPS_PIDS_logical_end)
break;
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
info->oldflags |= Item_table[e].oldflags;
library: fix boo-boo left by earlier patch, <PIDS> api After the commit referenced below the potential exists for a SEGV (resulting from an out-of-bounds Item_table reference if PROCPS_PIDS_physical_end is encountered). So this patch eliminates that PROCPS_PIDS_physical_end as no longer necessary and completes the task of using PROCPS_PIDS_logical_end as a sole necessary fencepost. Reference(s): commit e7585992d9c0743246247b3d6ee0041942fe07d5
2016-04-25 10:30:00 +05:30
info->history_yes |= Item_table[e].needhist;
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
}
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
if (info->oldflags & f_either) {
if (!(info->oldflags & f_stat))
info->oldflags |= f_status;
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
}
return;
} // end: libflags_set
static inline void oldproc_close (
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
PROCTAB **this)
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
{
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
if (*this != NULL) {
closeproc(*this);
*this = NULL;
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
}
} // end: oldproc_close
static inline int oldproc_open (
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
PROCTAB **this,
unsigned flags,
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
...)
{
library: correct a flawed approach for PROCPS_FILL_UID Gosh, just because nobody uses some newlib provision I guess, since it is being offered, it ought to actually be tested at some point. Well, that point just arrived and guess what? A surprise: some bugs were discovered. The procps_pids_select function established a for loop wherein readproc is called until the passed 'maxthese' limit. Unfortunately this was incorrect for 2 reasons: 1. For PROCPS_FILL_PID results are limited by what the oldlib finds, having established the pid list at open. Total found already cannot exceed a passed 'maxthese'; 2. With PROCPS_FILL_UID, returned results could exceed a 'maxthese' thus making the for loop incorrect again. [ plus yours truly neglected to forward the required ] [ UIDs total to our old library, another oops biggie ] In summary: the loop should have been forever, exiting only when all those identified procs had been located. So, while addressing those bugs, I've consolidated all the retrieval code (initialize, iterate, summarize) in a single helper function which will now serve both the procps_pids_reap and select functions. And as a result those guys were reduced to quite trivial housekeeping. This patch, hopefully, completes the normalization for reap/select (fill), which began with references shown. Reference(s): commit 0c953eccc5fe7240be9d272e1b6c0ce8769d8ed2 commit 747dfc5987e6e91ea3a8575de307e2892790c598 Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-08 10:30:00 +05:30
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
va_list vl;
int *ids;
library: correct a flawed approach for PROCPS_FILL_UID Gosh, just because nobody uses some newlib provision I guess, since it is being offered, it ought to actually be tested at some point. Well, that point just arrived and guess what? A surprise: some bugs were discovered. The procps_pids_select function established a for loop wherein readproc is called until the passed 'maxthese' limit. Unfortunately this was incorrect for 2 reasons: 1. For PROCPS_FILL_PID results are limited by what the oldlib finds, having established the pid list at open. Total found already cannot exceed a passed 'maxthese'; 2. With PROCPS_FILL_UID, returned results could exceed a 'maxthese' thus making the for loop incorrect again. [ plus yours truly neglected to forward the required ] [ UIDs total to our old library, another oops biggie ] In summary: the loop should have been forever, exiting only when all those identified procs had been located. So, while addressing those bugs, I've consolidated all the retrieval code (initialize, iterate, summarize) in a single helper function which will now serve both the procps_pids_reap and select functions. And as a result those guys were reduced to quite trivial housekeeping. This patch, hopefully, completes the normalization for reap/select (fill), which began with references shown. Reference(s): commit 0c953eccc5fe7240be9d272e1b6c0ce8769d8ed2 commit 747dfc5987e6e91ea3a8575de307e2892790c598 Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-08 10:30:00 +05:30
int num = 0;
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
if (*this == NULL) {
va_start(vl, flags);
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
ids = va_arg(vl, int*);
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
if (flags & PROC_UID) num = va_arg(vl, int);
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
va_end(vl);
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
if (NULL == (*this = openproc(flags, ids, num)))
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
return 0;
}
return 1;
} // end: oldproc_open
library: reorder (alphabetize) some things, <PIDS> api Without breaking either ABI or API just rearrange some stuff to provide a little more consistent source file. [ these were prompted by some work on the <stat> API ] [ in preparation for him borrowing some of our code. ] Reference(s): Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-04-25 10:30:00 +05:30
static inline int proc_tally (
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
struct procps_pidsinfo *info,
struct pids_counts *counts,
proc_t *p)
{
switch (p->state) {
case 'R':
++counts->running;
break;
case 'S':
case 'D':
++counts->sleeping;
break;
case 'T':
++counts->stopped;
break;
case 'Z':
++counts->zombied;
break;
default: // keep gcc happy
break;
}
++counts->total;
if (info->history_yes)
return !make_hist(info, p);
return 1;
library: reorder (alphabetize) some things, <PIDS> api Without breaking either ABI or API just rearrange some stuff to provide a little more consistent source file. [ these were prompted by some work on the <stat> API ] [ in preparation for him borrowing some of our code. ] Reference(s): Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-04-25 10:30:00 +05:30
} // end: proc_tally
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
library: privatize two with <pids> select/fill changes After simplifying that select/fill interface, there is no longer a need for public 'alloc' & 'dealloc' stacks functions. There is now only one instance of stacks as an input parameter found in procps_pids_stacks_sort(). But sorting 'empty' stacks serves no possible purpose. So this commit retains both functions, since they will still be needed, but designates them private (static). Additionally, with their demise we will eliminate that pids_stacks structure from the header file, internally using what always was the true 'stacks_extent' struct. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-01 10:30:00 +05:30
/*
library: reorder (alphabetize) some things, <PIDS> api Without breaking either ABI or API just rearrange some stuff to provide a little more consistent source file. [ these were prompted by some work on the <stat> API ] [ in preparation for him borrowing some of our code. ] Reference(s): Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-04-25 10:30:00 +05:30
* stacks_alloc():
library: privatize two with <pids> select/fill changes After simplifying that select/fill interface, there is no longer a need for public 'alloc' & 'dealloc' stacks functions. There is now only one instance of stacks as an input parameter found in procps_pids_stacks_sort(). But sorting 'empty' stacks serves no possible purpose. So this commit retains both functions, since they will still be needed, but designates them private (static). Additionally, with their demise we will eliminate that pids_stacks structure from the header file, internally using what always was the true 'stacks_extent' struct. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-01 10:30:00 +05:30
*
* Allocate and initialize one or more stacks each of which is anchored in an
library: protect against a future breakage, <pids> ABI By co-mingling both external/internal identifiers with actual implementation code, potential future additions to our API would have been considerably more difficult. So, this patch will now rely solely on internal/hidden identifiers serving as fenceposts in validation logic. And if the following convention is used for new fields we will maintain that enumerator alphabetic ordering a a little longer (even though 2 user fields now don't): . PROCPS_PIDS_XTRA_FOO_A, PROCPS_PIDS_XTRA_FOO_B, etc. Reference(s): http://www.freelists.org/post/procps/me-too-newlib,7 Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-04-18 10:30:00 +05:30
* associated pids_stack structure.
library: privatize two with <pids> select/fill changes After simplifying that select/fill interface, there is no longer a need for public 'alloc' & 'dealloc' stacks functions. There is now only one instance of stacks as an input parameter found in procps_pids_stacks_sort(). But sorting 'empty' stacks serves no possible purpose. So this commit retains both functions, since they will still be needed, but designates them private (static). Additionally, with their demise we will eliminate that pids_stacks structure from the header file, internally using what always was the true 'stacks_extent' struct. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-01 10:30:00 +05:30
*
* All such stacks will will have their result structures properly primed with
* 'items', while the result itself will be zeroed.
*
* Returns an array of pointers representing the 'heads' of each new stack.
*/
library: reorder (alphabetize) some things, <PIDS> api Without breaking either ABI or API just rearrange some stuff to provide a little more consistent source file. [ these were prompted by some work on the <stat> API ] [ in preparation for him borrowing some of our code. ] Reference(s): Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-04-25 10:30:00 +05:30
static struct stacks_extent *stacks_alloc (
library: privatize two with <pids> select/fill changes After simplifying that select/fill interface, there is no longer a need for public 'alloc' & 'dealloc' stacks functions. There is now only one instance of stacks as an input parameter found in procps_pids_stacks_sort(). But sorting 'empty' stacks serves no possible purpose. So this commit retains both functions, since they will still be needed, but designates them private (static). Additionally, with their demise we will eliminate that pids_stacks structure from the header file, internally using what always was the true 'stacks_extent' struct. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-01 10:30:00 +05:30
struct procps_pidsinfo *info,
int maxstacks)
{
struct stacks_extent *p_blob;
struct pids_stack **p_vect;
struct pids_stack *p_head;
size_t vect_size, head_size, list_size, blob_size;
void *v_head, *v_list;
int i;
if (info == NULL || info->items == NULL)
return NULL;
if (maxstacks < 1)
return NULL;
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
vect_size = sizeof(void *) * maxstacks; // size of the addr vectors |
vect_size += sizeof(void *); // plus NULL addr delimiter |
head_size = sizeof(struct pids_stack); // size of that head struct |
list_size = sizeof(struct pids_result) * info->maxitems; // any single results stack |
blob_size = sizeof(struct stacks_extent); // the extent anchor itself |
blob_size += vect_size; // plus room for addr vects |
blob_size += head_size * maxstacks; // plus room for head thing |
blob_size += list_size * maxstacks; // plus room for our stacks |
/* note: all of our memory is allocated in a single blob, facilitating a later free(). |
as a minimum, it is important that the result structures themselves always be |
contiguous for every stack since they are accessed through relative position. | */
library: privatize two with <pids> select/fill changes After simplifying that select/fill interface, there is no longer a need for public 'alloc' & 'dealloc' stacks functions. There is now only one instance of stacks as an input parameter found in procps_pids_stacks_sort(). But sorting 'empty' stacks serves no possible purpose. So this commit retains both functions, since they will still be needed, but designates them private (static). Additionally, with their demise we will eliminate that pids_stacks structure from the header file, internally using what always was the true 'stacks_extent' struct. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-01 10:30:00 +05:30
if (NULL == (p_blob = calloc(1, blob_size)))
return NULL;
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
p_blob->next = info->extents; // push this extent onto... |
info->extents = p_blob; // ...some existing extents |
p_vect = (void *)p_blob + sizeof(struct stacks_extent); // prime our vector pointer |
p_blob->stacks = p_vect; // set actual vectors start |
v_head = (void *)p_vect + vect_size; // prime head pointer start |
v_list = v_head + (head_size * maxstacks); // prime our stacks pointer |
library: privatize two with <pids> select/fill changes After simplifying that select/fill interface, there is no longer a need for public 'alloc' & 'dealloc' stacks functions. There is now only one instance of stacks as an input parameter found in procps_pids_stacks_sort(). But sorting 'empty' stacks serves no possible purpose. So this commit retains both functions, since they will still be needed, but designates them private (static). Additionally, with their demise we will eliminate that pids_stacks structure from the header file, internally using what always was the true 'stacks_extent' struct. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-01 10:30:00 +05:30
for (i = 0; i < maxstacks; i++) {
p_head = (struct pids_stack *)v_head;
library: reorder (alphabetize) some things, <PIDS> api Without breaking either ABI or API just rearrange some stuff to provide a little more consistent source file. [ these were prompted by some work on the <stat> API ] [ in preparation for him borrowing some of our code. ] Reference(s): Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-04-25 10:30:00 +05:30
p_head->head = itemize_stack((struct pids_result *)v_list, info->curitems, info->items);
library: privatize two with <pids> select/fill changes After simplifying that select/fill interface, there is no longer a need for public 'alloc' & 'dealloc' stacks functions. There is now only one instance of stacks as an input parameter found in procps_pids_stacks_sort(). But sorting 'empty' stacks serves no possible purpose. So this commit retains both functions, since they will still be needed, but designates them private (static). Additionally, with their demise we will eliminate that pids_stacks structure from the header file, internally using what always was the true 'stacks_extent' struct. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-01 10:30:00 +05:30
p_blob->stacks[i] = p_head;
v_list += list_size;
v_head += head_size;
}
p_blob->ext_numstacks = maxstacks;
return p_blob;
library: reorder (alphabetize) some things, <PIDS> api Without breaking either ABI or API just rearrange some stuff to provide a little more consistent source file. [ these were prompted by some work on the <stat> API ] [ in preparation for him borrowing some of our code. ] Reference(s): Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-04-25 10:30:00 +05:30
} // end: stacks_alloc
library: privatize two with <pids> select/fill changes After simplifying that select/fill interface, there is no longer a need for public 'alloc' & 'dealloc' stacks functions. There is now only one instance of stacks as an input parameter found in procps_pids_stacks_sort(). But sorting 'empty' stacks serves no possible purpose. So this commit retains both functions, since they will still be needed, but designates them private (static). Additionally, with their demise we will eliminate that pids_stacks structure from the header file, internally using what always was the true 'stacks_extent' struct. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-01 10:30:00 +05:30
library: reorder (alphabetize) some things, <PIDS> api Without breaking either ABI or API just rearrange some stuff to provide a little more consistent source file. [ these were prompted by some work on the <stat> API ] [ in preparation for him borrowing some of our code. ] Reference(s): Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-04-25 10:30:00 +05:30
static int stacks_fetch (
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
struct procps_pidsinfo *info)
library: correct a flawed approach for PROCPS_FILL_UID Gosh, just because nobody uses some newlib provision I guess, since it is being offered, it ought to actually be tested at some point. Well, that point just arrived and guess what? A surprise: some bugs were discovered. The procps_pids_select function established a for loop wherein readproc is called until the passed 'maxthese' limit. Unfortunately this was incorrect for 2 reasons: 1. For PROCPS_FILL_PID results are limited by what the oldlib finds, having established the pid list at open. Total found already cannot exceed a passed 'maxthese'; 2. With PROCPS_FILL_UID, returned results could exceed a 'maxthese' thus making the for loop incorrect again. [ plus yours truly neglected to forward the required ] [ UIDs total to our old library, another oops biggie ] In summary: the loop should have been forever, exiting only when all those identified procs had been located. So, while addressing those bugs, I've consolidated all the retrieval code (initialize, iterate, summarize) in a single helper function which will now serve both the procps_pids_reap and select functions. And as a result those guys were reduced to quite trivial housekeeping. This patch, hopefully, completes the normalization for reap/select (fill), which began with references shown. Reference(s): commit 0c953eccc5fe7240be9d272e1b6c0ce8769d8ed2 commit 747dfc5987e6e91ea3a8575de307e2892790c598 Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-08 10:30:00 +05:30
{
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
#define n_alloc info->fetch.n_alloc
#define n_inuse info->fetch.n_inuse
#define n_saved info->fetch.n_alloc_save
library: correct a flawed approach for PROCPS_FILL_UID Gosh, just because nobody uses some newlib provision I guess, since it is being offered, it ought to actually be tested at some point. Well, that point just arrived and guess what? A surprise: some bugs were discovered. The procps_pids_select function established a for loop wherein readproc is called until the passed 'maxthese' limit. Unfortunately this was incorrect for 2 reasons: 1. For PROCPS_FILL_PID results are limited by what the oldlib finds, having established the pid list at open. Total found already cannot exceed a passed 'maxthese'; 2. With PROCPS_FILL_UID, returned results could exceed a 'maxthese' thus making the for loop incorrect again. [ plus yours truly neglected to forward the required ] [ UIDs total to our old library, another oops biggie ] In summary: the loop should have been forever, exiting only when all those identified procs had been located. So, while addressing those bugs, I've consolidated all the retrieval code (initialize, iterate, summarize) in a single helper function which will now serve both the procps_pids_reap and select functions. And as a result those guys were reduced to quite trivial housekeeping. This patch, hopefully, completes the normalization for reap/select (fill), which began with references shown. Reference(s): commit 0c953eccc5fe7240be9d272e1b6c0ce8769d8ed2 commit 747dfc5987e6e91ea3a8575de307e2892790c598 Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-08 10:30:00 +05:30
static proc_t task; // static for initial zeroes + later dynamic free(s)
struct stacks_extent *ext;
// initialize stuff -----------------------------------
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
if (!info->fetch.anchor) {
if (!(info->fetch.anchor = calloc(sizeof(void *), MEMORY_INCR)))
return -ENOMEM;
library: correct a flawed approach for PROCPS_FILL_UID Gosh, just because nobody uses some newlib provision I guess, since it is being offered, it ought to actually be tested at some point. Well, that point just arrived and guess what? A surprise: some bugs were discovered. The procps_pids_select function established a for loop wherein readproc is called until the passed 'maxthese' limit. Unfortunately this was incorrect for 2 reasons: 1. For PROCPS_FILL_PID results are limited by what the oldlib finds, having established the pid list at open. Total found already cannot exceed a passed 'maxthese'; 2. With PROCPS_FILL_UID, returned results could exceed a 'maxthese' thus making the for loop incorrect again. [ plus yours truly neglected to forward the required ] [ UIDs total to our old library, another oops biggie ] In summary: the loop should have been forever, exiting only when all those identified procs had been located. So, while addressing those bugs, I've consolidated all the retrieval code (initialize, iterate, summarize) in a single helper function which will now serve both the procps_pids_reap and select functions. And as a result those guys were reduced to quite trivial housekeeping. This patch, hopefully, completes the normalization for reap/select (fill), which began with references shown. Reference(s): commit 0c953eccc5fe7240be9d272e1b6c0ce8769d8ed2 commit 747dfc5987e6e91ea3a8575de307e2892790c598 Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-08 10:30:00 +05:30
n_alloc = MEMORY_INCR;
}
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
if (!info->extents) {
if (!(ext = stacks_alloc(info, n_alloc)))
return -ENOMEM;
memset(info->fetch.anchor, 0, sizeof(void *) * n_alloc);
memcpy(info->fetch.anchor, ext->stacks, sizeof(void *) * n_alloc);
itemize_stacks_all(info);
}
cleanup_stacks_all(info);
library: correct a flawed approach for PROCPS_FILL_UID Gosh, just because nobody uses some newlib provision I guess, since it is being offered, it ought to actually be tested at some point. Well, that point just arrived and guess what? A surprise: some bugs were discovered. The procps_pids_select function established a for loop wherein readproc is called until the passed 'maxthese' limit. Unfortunately this was incorrect for 2 reasons: 1. For PROCPS_FILL_PID results are limited by what the oldlib finds, having established the pid list at open. Total found already cannot exceed a passed 'maxthese'; 2. With PROCPS_FILL_UID, returned results could exceed a 'maxthese' thus making the for loop incorrect again. [ plus yours truly neglected to forward the required ] [ UIDs total to our old library, another oops biggie ] In summary: the loop should have been forever, exiting only when all those identified procs had been located. So, while addressing those bugs, I've consolidated all the retrieval code (initialize, iterate, summarize) in a single helper function which will now serve both the procps_pids_reap and select functions. And as a result those guys were reduced to quite trivial housekeeping. This patch, hopefully, completes the normalization for reap/select (fill), which began with references shown. Reference(s): commit 0c953eccc5fe7240be9d272e1b6c0ce8769d8ed2 commit 747dfc5987e6e91ea3a8575de307e2892790c598 Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-08 10:30:00 +05:30
toggle_history(info);
library: more pids_fetch struct opaqueness, <PIDS> api With all our 3rd generation interfaces, we're now well positioned to preserve binary compatibility should new fields be added to any public structure (assuming that the 'result' union already contains its largest type). This remains true even for the <pids> interface, which unlike the others, has one structure embedded within a separate struct rather than declaring a pointer to it. The counts struct was positioned after the stacks ptrs array so as to preserve that ABI if ever new ints were added. Logically, however, the counts (especially that total) should precede the stacks array if we wished to properly place a horse (total) before a cart (stacks). So to enable relocating those counts we will no longer embed that structure, but provide a pointer to it. And this will make accessing syntax feel more natural too. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-23 10:30:00 +05:30
memset(&info->fetch.counts, 0, sizeof(struct pids_counts));
library: correct a flawed approach for PROCPS_FILL_UID Gosh, just because nobody uses some newlib provision I guess, since it is being offered, it ought to actually be tested at some point. Well, that point just arrived and guess what? A surprise: some bugs were discovered. The procps_pids_select function established a for loop wherein readproc is called until the passed 'maxthese' limit. Unfortunately this was incorrect for 2 reasons: 1. For PROCPS_FILL_PID results are limited by what the oldlib finds, having established the pid list at open. Total found already cannot exceed a passed 'maxthese'; 2. With PROCPS_FILL_UID, returned results could exceed a 'maxthese' thus making the for loop incorrect again. [ plus yours truly neglected to forward the required ] [ UIDs total to our old library, another oops biggie ] In summary: the loop should have been forever, exiting only when all those identified procs had been located. So, while addressing those bugs, I've consolidated all the retrieval code (initialize, iterate, summarize) in a single helper function which will now serve both the procps_pids_reap and select functions. And as a result those guys were reduced to quite trivial housekeeping. This patch, hopefully, completes the normalization for reap/select (fill), which began with references shown. Reference(s): commit 0c953eccc5fe7240be9d272e1b6c0ce8769d8ed2 commit 747dfc5987e6e91ea3a8575de307e2892790c598 Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-08 10:30:00 +05:30
// iterate stuff --------------------------------------
n_inuse = 0;
library: a couple of miscellaneous changes, <PIDS> api This patch represents the following minor adjustments: . old PROCTAB used with select & reap more descriptive . reformatted PROCPS_PIDS_VAL macro like other 3rd gen Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-18 10:30:00 +05:30
while (info->read_something(info->fetch_PT, &task)) {
library: correct a flawed approach for PROCPS_FILL_UID Gosh, just because nobody uses some newlib provision I guess, since it is being offered, it ought to actually be tested at some point. Well, that point just arrived and guess what? A surprise: some bugs were discovered. The procps_pids_select function established a for loop wherein readproc is called until the passed 'maxthese' limit. Unfortunately this was incorrect for 2 reasons: 1. For PROCPS_FILL_PID results are limited by what the oldlib finds, having established the pid list at open. Total found already cannot exceed a passed 'maxthese'; 2. With PROCPS_FILL_UID, returned results could exceed a 'maxthese' thus making the for loop incorrect again. [ plus yours truly neglected to forward the required ] [ UIDs total to our old library, another oops biggie ] In summary: the loop should have been forever, exiting only when all those identified procs had been located. So, while addressing those bugs, I've consolidated all the retrieval code (initialize, iterate, summarize) in a single helper function which will now serve both the procps_pids_reap and select functions. And as a result those guys were reduced to quite trivial housekeeping. This patch, hopefully, completes the normalization for reap/select (fill), which began with references shown. Reference(s): commit 0c953eccc5fe7240be9d272e1b6c0ce8769d8ed2 commit 747dfc5987e6e91ea3a8575de307e2892790c598 Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-08 10:30:00 +05:30
if (!(n_inuse < n_alloc)) {
n_alloc += MEMORY_INCR;
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
if ((!(info->fetch.anchor = realloc(info->fetch.anchor, sizeof(void *) * n_alloc)))
library: reorder (alphabetize) some things, <PIDS> api Without breaking either ABI or API just rearrange some stuff to provide a little more consistent source file. [ these were prompted by some work on the <stat> API ] [ in preparation for him borrowing some of our code. ] Reference(s): Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-04-25 10:30:00 +05:30
|| (!(ext = stacks_alloc(info, MEMORY_INCR))))
library: correct a flawed approach for PROCPS_FILL_UID Gosh, just because nobody uses some newlib provision I guess, since it is being offered, it ought to actually be tested at some point. Well, that point just arrived and guess what? A surprise: some bugs were discovered. The procps_pids_select function established a for loop wherein readproc is called until the passed 'maxthese' limit. Unfortunately this was incorrect for 2 reasons: 1. For PROCPS_FILL_PID results are limited by what the oldlib finds, having established the pid list at open. Total found already cannot exceed a passed 'maxthese'; 2. With PROCPS_FILL_UID, returned results could exceed a 'maxthese' thus making the for loop incorrect again. [ plus yours truly neglected to forward the required ] [ UIDs total to our old library, another oops biggie ] In summary: the loop should have been forever, exiting only when all those identified procs had been located. So, while addressing those bugs, I've consolidated all the retrieval code (initialize, iterate, summarize) in a single helper function which will now serve both the procps_pids_reap and select functions. And as a result those guys were reduced to quite trivial housekeeping. This patch, hopefully, completes the normalization for reap/select (fill), which began with references shown. Reference(s): commit 0c953eccc5fe7240be9d272e1b6c0ce8769d8ed2 commit 747dfc5987e6e91ea3a8575de307e2892790c598 Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-08 10:30:00 +05:30
return -1;
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
memcpy(info->fetch.anchor + n_inuse, ext->stacks, sizeof(void *) * MEMORY_INCR);
library: correct a flawed approach for PROCPS_FILL_UID Gosh, just because nobody uses some newlib provision I guess, since it is being offered, it ought to actually be tested at some point. Well, that point just arrived and guess what? A surprise: some bugs were discovered. The procps_pids_select function established a for loop wherein readproc is called until the passed 'maxthese' limit. Unfortunately this was incorrect for 2 reasons: 1. For PROCPS_FILL_PID results are limited by what the oldlib finds, having established the pid list at open. Total found already cannot exceed a passed 'maxthese'; 2. With PROCPS_FILL_UID, returned results could exceed a 'maxthese' thus making the for loop incorrect again. [ plus yours truly neglected to forward the required ] [ UIDs total to our old library, another oops biggie ] In summary: the loop should have been forever, exiting only when all those identified procs had been located. So, while addressing those bugs, I've consolidated all the retrieval code (initialize, iterate, summarize) in a single helper function which will now serve both the procps_pids_reap and select functions. And as a result those guys were reduced to quite trivial housekeeping. This patch, hopefully, completes the normalization for reap/select (fill), which began with references shown. Reference(s): commit 0c953eccc5fe7240be9d272e1b6c0ce8769d8ed2 commit 747dfc5987e6e91ea3a8575de307e2892790c598 Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-08 10:30:00 +05:30
}
library: more pids_fetch struct opaqueness, <PIDS> api With all our 3rd generation interfaces, we're now well positioned to preserve binary compatibility should new fields be added to any public structure (assuming that the 'result' union already contains its largest type). This remains true even for the <pids> interface, which unlike the others, has one structure embedded within a separate struct rather than declaring a pointer to it. The counts struct was positioned after the stacks ptrs array so as to preserve that ABI if ever new ints were added. Logically, however, the counts (especially that total) should precede the stacks array if we wished to properly place a horse (total) before a cart (stacks). So to enable relocating those counts we will no longer embed that structure, but provide a pointer to it. And this will make accessing syntax feel more natural too. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-23 10:30:00 +05:30
if (!proc_tally(info, &info->fetch.counts, &task))
library: correct a flawed approach for PROCPS_FILL_UID Gosh, just because nobody uses some newlib provision I guess, since it is being offered, it ought to actually be tested at some point. Well, that point just arrived and guess what? A surprise: some bugs were discovered. The procps_pids_select function established a for loop wherein readproc is called until the passed 'maxthese' limit. Unfortunately this was incorrect for 2 reasons: 1. For PROCPS_FILL_PID results are limited by what the oldlib finds, having established the pid list at open. Total found already cannot exceed a passed 'maxthese'; 2. With PROCPS_FILL_UID, returned results could exceed a 'maxthese' thus making the for loop incorrect again. [ plus yours truly neglected to forward the required ] [ UIDs total to our old library, another oops biggie ] In summary: the loop should have been forever, exiting only when all those identified procs had been located. So, while addressing those bugs, I've consolidated all the retrieval code (initialize, iterate, summarize) in a single helper function which will now serve both the procps_pids_reap and select functions. And as a result those guys were reduced to quite trivial housekeeping. This patch, hopefully, completes the normalization for reap/select (fill), which began with references shown. Reference(s): commit 0c953eccc5fe7240be9d272e1b6c0ce8769d8ed2 commit 747dfc5987e6e91ea3a8575de307e2892790c598 Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-08 10:30:00 +05:30
return -1;
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
assign_results(info, info->fetch.anchor[n_inuse++], &task);
library: correct a flawed approach for PROCPS_FILL_UID Gosh, just because nobody uses some newlib provision I guess, since it is being offered, it ought to actually be tested at some point. Well, that point just arrived and guess what? A surprise: some bugs were discovered. The procps_pids_select function established a for loop wherein readproc is called until the passed 'maxthese' limit. Unfortunately this was incorrect for 2 reasons: 1. For PROCPS_FILL_PID results are limited by what the oldlib finds, having established the pid list at open. Total found already cannot exceed a passed 'maxthese'; 2. With PROCPS_FILL_UID, returned results could exceed a 'maxthese' thus making the for loop incorrect again. [ plus yours truly neglected to forward the required ] [ UIDs total to our old library, another oops biggie ] In summary: the loop should have been forever, exiting only when all those identified procs had been located. So, while addressing those bugs, I've consolidated all the retrieval code (initialize, iterate, summarize) in a single helper function which will now serve both the procps_pids_reap and select functions. And as a result those guys were reduced to quite trivial housekeeping. This patch, hopefully, completes the normalization for reap/select (fill), which began with references shown. Reference(s): commit 0c953eccc5fe7240be9d272e1b6c0ce8769d8ed2 commit 747dfc5987e6e91ea3a8575de307e2892790c598 Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-08 10:30:00 +05:30
}
// finalize stuff -------------------------------------
library: final tweaks to code and/or comments, 3rd gen With the dust now settling on all those 3rd generation upgrades, this patch tries to provide some consistency among the separate modules involved. Someday we should consider a 4th generation where all redundant code has been removed and isolated in a new shared source file. Following is a summary of significant changes (if any) to each of these now upgraded 3rd gen library modules. <meminfo> ............................................ . strictly formatting/comment changes, code unaffected <pids> ............................................... . replaced a local mkSTR macro with existing STRINGIFY . added fetch narrative explaining duplicate addresses <slabinfo> ........................................... . rearranged some free logic for procps_slabinfo_unref . added fetch narrative explaining duplicate addresses <stat> ............................................... . added #define ENFORCE_LOGICAL, just as in <slabinfo> . replaced a local mkSTR macro with existing STRINGIFY . alphabetized the function declarations in the header <vmstat> ............................................. . made one coverity concession with read_vmstat_failed [ several of these changes may reflect this author's ] [ continuing pursuit of an unreasonable goal -- that ] [ of a 'perfect' (plus 'pretty') C language program! ] Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-09 10:30:00 +05:30
/* note: we go to this trouble of maintaining a duplicate of the consolidated |
extent stacks addresses represented as our 'anchor' since these ptrs |
library: refactor exposed pointers management, 3rd gen This commit brings all of those 'fetch' type functions (supporting some form of 'reap') into closer alignment with one another. The biggest impact is to be found in the <stat> module, which now provides for the separate copy of stack pointers which will be exposed to users. The reason such a copy was not employed initially with <stat>, unlike those for <pids> and <slabinfo>, is due to the fact that such stacks were never sortable. Thus the original raw consolidated extent pointers wouldn't have been disturbed. But that meant no NULL delimiter. So with this commit, all reap/fetch operations now use pointer copies when returning results to callers. And, all such arrays are now NULL delimited meaning callers can choose their own access fencepost: totals or NULL. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-01 10:30:00 +05:30
are exposed to a user (um, not that we don't trust 'em or anything). |
plus, we can NULL delimit these ptrs which we couldn't do otherwise. | */
if (n_saved < n_inuse + 1) {
n_saved = n_inuse + 1;
library: misc tweaks for code and comments, <PIDS> api This commit simply tries to keep naming plus formating conventions on a par with the continuing climb up that learning curve. These changes were suggested following <slabinfo> sources upgrade from 2nd to 3rd generation. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-06 10:30:00 +05:30
if (!(info->fetch.results.stacks = realloc(info->fetch.results.stacks, sizeof(void *) * n_saved)))
library: refactor exposed pointers management, 3rd gen This commit brings all of those 'fetch' type functions (supporting some form of 'reap') into closer alignment with one another. The biggest impact is to be found in the <stat> module, which now provides for the separate copy of stack pointers which will be exposed to users. The reason such a copy was not employed initially with <stat>, unlike those for <pids> and <slabinfo>, is due to the fact that such stacks were never sortable. Thus the original raw consolidated extent pointers wouldn't have been disturbed. But that meant no NULL delimiter. So with this commit, all reap/fetch operations now use pointer copies when returning results to callers. And, all such arrays are now NULL delimited meaning callers can choose their own access fencepost: totals or NULL. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-01 10:30:00 +05:30
return -ENOMEM;
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
}
library: misc tweaks for code and comments, <PIDS> api This commit simply tries to keep naming plus formating conventions on a par with the continuing climb up that learning curve. These changes were suggested following <slabinfo> sources upgrade from 2nd to 3rd generation. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-06 10:30:00 +05:30
memcpy(info->fetch.results.stacks, info->fetch.anchor, sizeof(void *) * n_inuse);
info->fetch.results.stacks[n_inuse] = NULL;
library: refactor exposed pointers management, 3rd gen This commit brings all of those 'fetch' type functions (supporting some form of 'reap') into closer alignment with one another. The biggest impact is to be found in the <stat> module, which now provides for the separate copy of stack pointers which will be exposed to users. The reason such a copy was not employed initially with <stat>, unlike those for <pids> and <slabinfo>, is due to the fact that such stacks were never sortable. Thus the original raw consolidated extent pointers wouldn't have been disturbed. But that meant no NULL delimiter. So with this commit, all reap/fetch operations now use pointer copies when returning results to callers. And, all such arrays are now NULL delimited meaning callers can choose their own access fencepost: totals or NULL. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-01 10:30:00 +05:30
library: correct a flawed approach for PROCPS_FILL_UID Gosh, just because nobody uses some newlib provision I guess, since it is being offered, it ought to actually be tested at some point. Well, that point just arrived and guess what? A surprise: some bugs were discovered. The procps_pids_select function established a for loop wherein readproc is called until the passed 'maxthese' limit. Unfortunately this was incorrect for 2 reasons: 1. For PROCPS_FILL_PID results are limited by what the oldlib finds, having established the pid list at open. Total found already cannot exceed a passed 'maxthese'; 2. With PROCPS_FILL_UID, returned results could exceed a 'maxthese' thus making the for loop incorrect again. [ plus yours truly neglected to forward the required ] [ UIDs total to our old library, another oops biggie ] In summary: the loop should have been forever, exiting only when all those identified procs had been located. So, while addressing those bugs, I've consolidated all the retrieval code (initialize, iterate, summarize) in a single helper function which will now serve both the procps_pids_reap and select functions. And as a result those guys were reduced to quite trivial housekeeping. This patch, hopefully, completes the normalization for reap/select (fill), which began with references shown. Reference(s): commit 0c953eccc5fe7240be9d272e1b6c0ce8769d8ed2 commit 747dfc5987e6e91ea3a8575de307e2892790c598 Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-08 10:30:00 +05:30
return n_inuse; // callers beware, this might be zero !
#undef n_alloc
#undef n_inuse
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
#undef n_saved
library: reorder (alphabetize) some things, <PIDS> api Without breaking either ABI or API just rearrange some stuff to provide a little more consistent source file. [ these were prompted by some work on the <stat> API ] [ in preparation for him borrowing some of our code. ] Reference(s): Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-04-25 10:30:00 +05:30
} // end: stacks_fetch
library: privatize two with <pids> select/fill changes After simplifying that select/fill interface, there is no longer a need for public 'alloc' & 'dealloc' stacks functions. There is now only one instance of stacks as an input parameter found in procps_pids_stacks_sort(). But sorting 'empty' stacks serves no possible purpose. So this commit retains both functions, since they will still be needed, but designates them private (static). Additionally, with their demise we will eliminate that pids_stacks structure from the header file, internally using what always was the true 'stacks_extent' struct. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-01 10:30:00 +05:30
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
// ___ Public Functions |||||||||||||||||||||||||||||||||||||||||||||||||||||||
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
// --- standard required functions --------------------------------------------
library: introduce a fatal 'proc not mounted' function Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-09-11 10:30:00 +05:30
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
/*
* procps_pids_new():
*
* @info: location of returned new structure
*
* Returns: 0 on success <0 on failure
*/
PROCPS_EXPORT int procps_pids_new (
struct procps_pidsinfo **info,
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
enum pids_item *items,
int numitems)
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
{
struct procps_pidsinfo *p;
library: add <pids> TIME_ALL/TIME_ELAPSED needed by ps Shoot, here's yet another bow to ps needs. But it's ok because it makes a lot of sense. Rather than force all users into their own calculations do but it once here. As an aside this need arose during ps testing when the sorts were using TIME_START or TICS_ALL. That was just fine for almost every need except 'etime' plus 'time'. That 'etime' was sorting the opposite of what's wanted when using TIME_START (of course) while 'time' yielded some weird ordering because TICS_ALL was too granular. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-09-30 10:30:00 +05:30
double uptime_secs;
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
int pgsz;
if (info == NULL || *info != NULL)
return -EINVAL;
if (!(p = calloc(1, sizeof(struct procps_pidsinfo))))
return -ENOMEM;
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
/* if we're without items or numitems, a later call to
procps_pids_reset() will become mandatory */
if (items && numitems) {
library: optional parms protection missing, <PIDS> api When those items were made dynamic at 'new' time, some other functions, previously assured of their presence, failed to verify a 'reset' had acually been requested. This commit just corrects that oversight and avoids an attempt to 'assign_results' when no items are present. Reference(s): . when items/numitems became optional commit 9ebadc1438a6665a98a9f0782523b0f9a2a6248f Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-15 10:30:00 +05:30
if (items_check_failed(items, numitems)) {
library: respond to coverity (reluctantly), <PIDS> api Calls to free() have now been reintroduce in the new() function to quiet coverity warnings. Those free's were removed originally as that library 'new' was returning with a fatal error and a caller should end abnormally. Plus, it is virtually impossible to fail a malloc call under linux. And lastly, they required braces with the if statement making the code considerably less pretty. [ commit also includes 2 unrelated whitespace tweaks ] Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-17 00:44:14 +05:30
free(p);
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
return -EINVAL;
library: respond to coverity (reluctantly), <PIDS> api Calls to free() have now been reintroduce in the new() function to quiet coverity warnings. Those free's were removed originally as that library 'new' was returning with a fatal error and a caller should end abnormally. Plus, it is virtually impossible to fail a malloc call under linux. And lastly, they required braces with the if statement making the code considerably less pretty. [ commit also includes 2 unrelated whitespace tweaks ] Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-17 00:44:14 +05:30
}
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
// allow for our PROCPS_PIDS_logical_end
p->maxitems = numitems + 1;
library: respond to coverity (reluctantly), <PIDS> api Calls to free() have now been reintroduce in the new() function to quiet coverity warnings. Those free's were removed originally as that library 'new' was returning with a fatal error and a caller should end abnormally. Plus, it is virtually impossible to fail a malloc call under linux. And lastly, they required braces with the if statement making the code considerably less pretty. [ commit also includes 2 unrelated whitespace tweaks ] Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-17 00:44:14 +05:30
if (!(p->items = calloc(p->maxitems, sizeof(enum pids_item)))) {
free(p);
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
return -ENOMEM;
library: respond to coverity (reluctantly), <PIDS> api Calls to free() have now been reintroduce in the new() function to quiet coverity warnings. Those free's were removed originally as that library 'new' was returning with a fatal error and a caller should end abnormally. Plus, it is virtually impossible to fail a malloc call under linux. And lastly, they required braces with the if statement making the code considerably less pretty. [ commit also includes 2 unrelated whitespace tweaks ] Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-17 00:44:14 +05:30
}
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
memcpy(p->items, items, sizeof(enum pids_item) * numitems);
p->items[numitems] = PROCPS_PIDS_logical_end;
p->curitems = p->maxitems;
libflags_set(p);
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
}
library: respond to coverity (reluctantly), <PIDS> api Calls to free() have now been reintroduce in the new() function to quiet coverity warnings. Those free's were removed originally as that library 'new' was returning with a fatal error and a caller should end abnormally. Plus, it is virtually impossible to fail a malloc call under linux. And lastly, they required braces with the if statement making the code considerably less pretty. [ commit also includes 2 unrelated whitespace tweaks ] Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-17 00:44:14 +05:30
if (!(p->hist = calloc(MEMORY_INCR, sizeof(struct history_info)))) {
free(p->items);
free(p);
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
return -ENOMEM;
library: respond to coverity (reluctantly), <PIDS> api Calls to free() have now been reintroduce in the new() function to quiet coverity warnings. Those free's were removed originally as that library 'new' was returning with a fatal error and a caller should end abnormally. Plus, it is virtually impossible to fail a malloc call under linux. And lastly, they required braces with the if statement making the code considerably less pretty. [ commit also includes 2 unrelated whitespace tweaks ] Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-17 00:44:14 +05:30
}
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
config_history(p);
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
pgsz = getpagesize();
while (pgsz > 1024) { pgsz >>= 1; p->pgs2k_shift++; }
library: add <pids> TIME_ALL/TIME_ELAPSED needed by ps Shoot, here's yet another bow to ps needs. But it's ok because it makes a lot of sense. Rather than force all users into their own calculations do but it once here. As an aside this need arose during ps testing when the sorts were using TIME_START or TICS_ALL. That was just fine for almost every need except 'etime' plus 'time'. That 'etime' was sorting the opposite of what's wanted when using TIME_START (of course) while 'time' yielded some weird ordering because TICS_ALL was too granular. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-09-30 10:30:00 +05:30
p->hertz = procps_hertz_get();
procps_uptime(&uptime_secs, NULL);
library: miscellaneous minor accumulated <pids> tweaks This patch contains the following minor modifications: . When assigning to boot_seconds from a double, a cast to 'unsigned long' was employed that in reality should have been 'unsigned long long'. But, on second thought it's probably best if the compiler makes the decision. . Results for ID_TGID do not require an 'f_status' flg since both tid and tgid will be available by virtue of the 'simple_nextpid' function normal operations. Thus, that flag has been removed from the <pids> Item_table. . When the pids_stacks structure was eliminated with a change to the alloc/dealloc functions in the reference below, several casts became redundant and are removed. Reference(s): commit 747dfc5987e6e91ea3a8575de307e2892790c598. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-07 10:30:00 +05:30
p->boot_seconds = uptime_secs;
library: add <pids> TIME_ALL/TIME_ELAPSED needed by ps Shoot, here's yet another bow to ps needs. But it's ok because it makes a lot of sense. Rather than force all users into their own calculations do but it once here. As an aside this need arose during ps testing when the sorts were using TIME_START or TICS_ALL. That was just fine for almost every need except 'etime' plus 'time'. That 'etime' was sorting the opposite of what's wanted when using TIME_START (of course) while 'time' yielded some weird ordering because TICS_ALL was too granular. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-09-30 10:30:00 +05:30
library: more pids_fetch struct opaqueness, <PIDS> api With all our 3rd generation interfaces, we're now well positioned to preserve binary compatibility should new fields be added to any public structure (assuming that the 'result' union already contains its largest type). This remains true even for the <pids> interface, which unlike the others, has one structure embedded within a separate struct rather than declaring a pointer to it. The counts struct was positioned after the stacks ptrs array so as to preserve that ABI if ever new ints were added. Logically, however, the counts (especially that total) should precede the stacks array if we wished to properly place a horse (total) before a cart (stacks). So to enable relocating those counts we will no longer embed that structure, but provide a pointer to it. And this will make accessing syntax feel more natural too. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-23 10:30:00 +05:30
p->fetch.results.counts = &p->fetch.counts;
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
p->refcount = 1;
*info = p;
return 0;
} // end: procps_pids_new
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
PROCPS_EXPORT int procps_pids_ref (
library: add sequential access to that pids collection To ease the transition to the new interface, for other than that top program, individual read provisions have been added to the <proc/pids.h> API. This represents a refinement of a position stated in a post noted below. Reference(s): http://www.freelists.org/post/procps/newlib-ps-fix Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-26 10:30:00 +05:30
struct procps_pidsinfo *info)
{
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
if (info == NULL)
return -EINVAL;
library: add sequential access to that pids collection To ease the transition to the new interface, for other than that top program, individual read provisions have been added to the <proc/pids.h> API. This represents a refinement of a position stated in a post noted below. Reference(s): http://www.freelists.org/post/procps/newlib-ps-fix Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-26 10:30:00 +05:30
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
info->refcount++;
return info->refcount;
} // end: procps_pids_ref
PROCPS_EXPORT int procps_pids_unref (
struct procps_pidsinfo **info)
{
if (info == NULL || *info == NULL)
return -EINVAL;
(*info)->refcount--;
if ((*info)->refcount == 0) {
#ifdef UNREF_RPTHASH
unref_rpthash(*info);
#endif
if ((*info)->extents) {
cleanup_stacks_all(*info);
do {
struct stacks_extent *p = (*info)->extents;
(*info)->extents = (*info)->extents->next;
free(p);
} while ((*info)->extents);
}
if ((*info)->otherexts) {
struct stacks_extent *nextext, *ext = (*info)->otherexts;
while (ext) {
nextext = ext->next;
cleanup_stack(ext->stacks[0]->head);
free(ext);
ext = nextext;
};
}
if ((*info)->fetch.anchor)
free((*info)->fetch.anchor);
library: misc tweaks for code and comments, <PIDS> api This commit simply tries to keep naming plus formating conventions on a par with the continuing climb up that learning curve. These changes were suggested following <slabinfo> sources upgrade from 2nd to 3rd generation. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-06 10:30:00 +05:30
if ((*info)->fetch.results.stacks)
free((*info)->fetch.results.stacks);
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
if ((*info)->items)
free((*info)->items);
if ((*info)->hist) {
free((*info)->hist->PHist_sav);
free((*info)->hist->PHist_new);
free((*info)->hist);
}
free(*info);
*info = NULL;
return 0;
library: add sequential access to that pids collection To ease the transition to the new interface, for other than that top program, individual read provisions have been added to the <proc/pids.h> API. This represents a refinement of a position stated in a post noted below. Reference(s): http://www.freelists.org/post/procps/newlib-ps-fix Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-26 10:30:00 +05:30
}
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
return (*info)->refcount;
} // end: procps_pids_unref
library: add sequential access to that pids collection To ease the transition to the new interface, for other than that top program, individual read provisions have been added to the <proc/pids.h> API. This represents a refinement of a position stated in a post noted below. Reference(s): http://www.freelists.org/post/procps/newlib-ps-fix Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-26 10:30:00 +05:30
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
// --- variable interface functions -------------------------------------------
library: add sequential access to that pids collection To ease the transition to the new interface, for other than that top program, individual read provisions have been added to the <proc/pids.h> API. This represents a refinement of a position stated in a post noted below. Reference(s): http://www.freelists.org/post/procps/newlib-ps-fix Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-26 10:30:00 +05:30
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
PROCPS_EXPORT struct pids_stack *fatal_proc_unmounted (
library: add sequential access to that pids collection To ease the transition to the new interface, for other than that top program, individual read provisions have been added to the <proc/pids.h> API. This represents a refinement of a position stated in a post noted below. Reference(s): http://www.freelists.org/post/procps/newlib-ps-fix Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-26 10:30:00 +05:30
struct procps_pidsinfo *info,
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
int return_self)
library: add sequential access to that pids collection To ease the transition to the new interface, for other than that top program, individual read provisions have been added to the <proc/pids.h> API. This represents a refinement of a position stated in a post noted below. Reference(s): http://www.freelists.org/post/procps/newlib-ps-fix Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-26 10:30:00 +05:30
{
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
static proc_t self;
struct stacks_extent *ext;
library: add sequential access to that pids collection To ease the transition to the new interface, for other than that top program, individual read provisions have been added to the <proc/pids.h> API. This represents a refinement of a position stated in a post noted below. Reference(s): http://www.freelists.org/post/procps/newlib-ps-fix Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-26 10:30:00 +05:30
library: optional parms protection missing, <PIDS> api When those items were made dynamic at 'new' time, some other functions, previously assured of their presence, failed to verify a 'reset' had acually been requested. This commit just corrects that oversight and avoids an attempt to 'assign_results' when no items are present. Reference(s): . when items/numitems became optional commit 9ebadc1438a6665a98a9f0782523b0f9a2a6248f Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-15 10:30:00 +05:30
/* this is very likely the *only* newlib function where the
context (procps_pidsinfo) of NULL will ever be permitted */
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
look_up_our_self(&self);
if (!return_self)
return NULL;
library: add sequential access to that pids collection To ease the transition to the new interface, for other than that top program, individual read provisions have been added to the <proc/pids.h> API. This represents a refinement of a position stated in a post noted below. Reference(s): http://www.freelists.org/post/procps/newlib-ps-fix Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-26 10:30:00 +05:30
library: optional parms protection missing, <PIDS> api When those items were made dynamic at 'new' time, some other functions, previously assured of their presence, failed to verify a 'reset' had acually been requested. This commit just corrects that oversight and avoids an attempt to 'assign_results' when no items are present. Reference(s): . when items/numitems became optional commit 9ebadc1438a6665a98a9f0782523b0f9a2a6248f Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-15 10:30:00 +05:30
if (info == NULL)
return NULL;
/* with items & numitems technically optional at 'new' time, it's
expected 'reset' will have been called -- but just in case ... */
if (!info->curitems)
return NULL;
if (!(ext = stacks_alloc(info, 1)))
return NULL;
if (!extent_cut(info, ext))
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
return NULL;
library: add sequential access to that pids collection To ease the transition to the new interface, for other than that top program, individual read provisions have been added to the <proc/pids.h> API. This represents a refinement of a position stated in a post noted below. Reference(s): http://www.freelists.org/post/procps/newlib-ps-fix Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-26 10:30:00 +05:30
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
ext->next = info->otherexts;
info->otherexts = ext;
assign_results(info, ext->stacks[0], &self);
return ext->stacks[0];
} // end: fatal_proc_unmounted
PROCPS_EXPORT struct pids_stack *procps_pids_get (
struct procps_pidsinfo *info,
enum pids_fetch_type which)
library: add sequential access to that pids collection To ease the transition to the new interface, for other than that top program, individual read provisions have been added to the <proc/pids.h> API. This represents a refinement of a position stated in a post noted below. Reference(s): http://www.freelists.org/post/procps/newlib-ps-fix Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-26 10:30:00 +05:30
{
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
static proc_t task; // static for initial zeroes + later dynamic free(s)
if (info == NULL)
return NULL;
if (!info->curitems)
return NULL;
if (which != PROCPS_FETCH_TASKS_ONLY && which != PROCPS_FETCH_THREADS_TOO)
return NULL;
library: optional parms protection missing, <PIDS> api When those items were made dynamic at 'new' time, some other functions, previously assured of their presence, failed to verify a 'reset' had acually been requested. This commit just corrects that oversight and avoids an attempt to 'assign_results' when no items are present. Reference(s): . when items/numitems became optional commit 9ebadc1438a6665a98a9f0782523b0f9a2a6248f Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-15 10:30:00 +05:30
/* with items & numitems technically optional at 'new' time, it's
expected 'reset' will have been called -- but just in case ... */
if (!info->curitems)
return NULL;
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
fresh_start:
if (!info->get_ext) {
if (!(info->get_ext = stacks_alloc(info, 1)))
return NULL;
if (!oldproc_open(&info->get_PT, info->oldflags))
return NULL;
info->get_type = which;
info->read_something = which ? readeither : readproc;
}
if (info->get_type != which) {
oldproc_close(&info->get_PT);
cleanup_stack(info->get_ext->stacks[0]->head);
if (extent_cut(info, info->get_ext))
free(info->get_ext);
info->get_ext = NULL;
goto fresh_start;
}
cleanup_stack(info->get_ext->stacks[0]->head);
if (NULL == info->read_something(info->get_PT, &task))
return NULL;
assign_results(info, info->get_ext->stacks[0], &task);
return info->get_ext->stacks[0];
} // end: procps_pids_get
library: add sequential access to that pids collection To ease the transition to the new interface, for other than that top program, individual read provisions have been added to the <proc/pids.h> API. This represents a refinement of a position stated in a post noted below. Reference(s): http://www.freelists.org/post/procps/newlib-ps-fix Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-26 10:30:00 +05:30
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
/* procps_pids_reap():
*
* Harvest all the available tasks/threads and provide the result
* stacks along with a summary of the information gathered.
*
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
* Returns: pointer to a pids_fetch struct on success, NULL on error.
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
*/
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
PROCPS_EXPORT struct pids_fetch *procps_pids_reap (
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
struct procps_pidsinfo *info,
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
enum pids_fetch_type which)
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
{
library: correct a flawed approach for PROCPS_FILL_UID Gosh, just because nobody uses some newlib provision I guess, since it is being offered, it ought to actually be tested at some point. Well, that point just arrived and guess what? A surprise: some bugs were discovered. The procps_pids_select function established a for loop wherein readproc is called until the passed 'maxthese' limit. Unfortunately this was incorrect for 2 reasons: 1. For PROCPS_FILL_PID results are limited by what the oldlib finds, having established the pid list at open. Total found already cannot exceed a passed 'maxthese'; 2. With PROCPS_FILL_UID, returned results could exceed a 'maxthese' thus making the for loop incorrect again. [ plus yours truly neglected to forward the required ] [ UIDs total to our old library, another oops biggie ] In summary: the loop should have been forever, exiting only when all those identified procs had been located. So, while addressing those bugs, I've consolidated all the retrieval code (initialize, iterate, summarize) in a single helper function which will now serve both the procps_pids_reap and select functions. And as a result those guys were reduced to quite trivial housekeeping. This patch, hopefully, completes the normalization for reap/select (fill), which began with references shown. Reference(s): commit 0c953eccc5fe7240be9d272e1b6c0ce8769d8ed2 commit 747dfc5987e6e91ea3a8575de307e2892790c598 Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-08 10:30:00 +05:30
int rc;
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
if (info == NULL)
library: fix major oops in procps_pids_reap() function In my zeal to finalize the initial pids implementation I omitted some quite important parameter checking from the above function. Thank goodness top was kind to us. Also, in anticipation of the additions of single stack read and supporting functions some items were renamed. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-26 10:30:00 +05:30
return NULL;
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
if (which != PROCPS_FETCH_TASKS_ONLY && which != PROCPS_FETCH_THREADS_TOO)
library: fix major oops in procps_pids_reap() function In my zeal to finalize the initial pids implementation I omitted some quite important parameter checking from the above function. Thank goodness top was kind to us. Also, in anticipation of the additions of single stack read and supporting functions some items were renamed. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-26 10:30:00 +05:30
return NULL;
library: optional parms protection missing, <PIDS> api When those items were made dynamic at 'new' time, some other functions, previously assured of their presence, failed to verify a 'reset' had acually been requested. This commit just corrects that oversight and avoids an attempt to 'assign_results' when no items are present. Reference(s): . when items/numitems became optional commit 9ebadc1438a6665a98a9f0782523b0f9a2a6248f Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-15 10:30:00 +05:30
/* with items & numitems technically optional at 'new' time, it's
expected 'reset' will have been called -- but just in case ... */
if (!info->curitems)
return NULL;
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
library: a couple of miscellaneous changes, <PIDS> api This patch represents the following minor adjustments: . old PROCTAB used with select & reap more descriptive . reformatted PROCPS_PIDS_VAL macro like other 3rd gen Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-18 10:30:00 +05:30
if (!oldproc_open(&info->fetch_PT, info->oldflags))
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
return NULL;
library: correct a flawed approach for PROCPS_FILL_UID Gosh, just because nobody uses some newlib provision I guess, since it is being offered, it ought to actually be tested at some point. Well, that point just arrived and guess what? A surprise: some bugs were discovered. The procps_pids_select function established a for loop wherein readproc is called until the passed 'maxthese' limit. Unfortunately this was incorrect for 2 reasons: 1. For PROCPS_FILL_PID results are limited by what the oldlib finds, having established the pid list at open. Total found already cannot exceed a passed 'maxthese'; 2. With PROCPS_FILL_UID, returned results could exceed a 'maxthese' thus making the for loop incorrect again. [ plus yours truly neglected to forward the required ] [ UIDs total to our old library, another oops biggie ] In summary: the loop should have been forever, exiting only when all those identified procs had been located. So, while addressing those bugs, I've consolidated all the retrieval code (initialize, iterate, summarize) in a single helper function which will now serve both the procps_pids_reap and select functions. And as a result those guys were reduced to quite trivial housekeeping. This patch, hopefully, completes the normalization for reap/select (fill), which began with references shown. Reference(s): commit 0c953eccc5fe7240be9d272e1b6c0ce8769d8ed2 commit 747dfc5987e6e91ea3a8575de307e2892790c598 Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-08 10:30:00 +05:30
info->read_something = which ? readeither : readproc;
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
rc = stacks_fetch(info);
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
library: a couple of miscellaneous changes, <PIDS> api This patch represents the following minor adjustments: . old PROCTAB used with select & reap more descriptive . reformatted PROCPS_PIDS_VAL macro like other 3rd gen Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-18 10:30:00 +05:30
oldproc_close(&info->fetch_PT);
library: correct a flawed approach for PROCPS_FILL_UID Gosh, just because nobody uses some newlib provision I guess, since it is being offered, it ought to actually be tested at some point. Well, that point just arrived and guess what? A surprise: some bugs were discovered. The procps_pids_select function established a for loop wherein readproc is called until the passed 'maxthese' limit. Unfortunately this was incorrect for 2 reasons: 1. For PROCPS_FILL_PID results are limited by what the oldlib finds, having established the pid list at open. Total found already cannot exceed a passed 'maxthese'; 2. With PROCPS_FILL_UID, returned results could exceed a 'maxthese' thus making the for loop incorrect again. [ plus yours truly neglected to forward the required ] [ UIDs total to our old library, another oops biggie ] In summary: the loop should have been forever, exiting only when all those identified procs had been located. So, while addressing those bugs, I've consolidated all the retrieval code (initialize, iterate, summarize) in a single helper function which will now serve both the procps_pids_reap and select functions. And as a result those guys were reduced to quite trivial housekeeping. This patch, hopefully, completes the normalization for reap/select (fill), which began with references shown. Reference(s): commit 0c953eccc5fe7240be9d272e1b6c0ce8769d8ed2 commit 747dfc5987e6e91ea3a8575de307e2892790c598 Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-08 10:30:00 +05:30
// we better have found at least 1 pid
library: misc tweaks for code and comments, <PIDS> api This commit simply tries to keep naming plus formating conventions on a par with the continuing climb up that learning curve. These changes were suggested following <slabinfo> sources upgrade from 2nd to 3rd generation. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-06 10:30:00 +05:30
return (rc > 0) ? &info->fetch.results : NULL;
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
} // end: procps_pids_reap
PROCPS_EXPORT int procps_pids_reset (
struct procps_pidsinfo *info,
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
enum pids_item *newitems,
int newnumitems)
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
{
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
if (info == NULL || newitems == NULL)
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
return -EINVAL;
library: optional parms protection missing, <PIDS> api When those items were made dynamic at 'new' time, some other functions, previously assured of their presence, failed to verify a 'reset' had acually been requested. This commit just corrects that oversight and avoids an attempt to 'assign_results' when no items are present. Reference(s): . when items/numitems became optional commit 9ebadc1438a6665a98a9f0782523b0f9a2a6248f Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-15 10:30:00 +05:30
if (items_check_failed(newitems, newnumitems))
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
return -EINVAL;
library: miscellaneous tweaks to pid code and comments . traded a complex misaligned memory allocation scheme in the make_hist function for a simple aligned scheme. plus memory allocation increases are globally defined. . changed 1 parameter for procps_pids_stacks_sort() to better reflect the 'array of pointers', not an address of a pointer as is used with guys such as 'new/unref'. . the pids_reap struct was changed slightly to make it more reflective of it's actual implementation details. . the Item_table member .mustfree is now .needfree and that .makehist was now made .needhist for consistency. . reduced the number of separate 'return NULL;' source statements in that primary procps_pids_reap() routine. . ensured consistent reference to sizeof(void *) & not occasional reference to sizeof(void*) without a space. . rather than enable/disable validate_stacks via a #if in the function body, it is now handled via a #define. . some comments in the procps_pids_reset function were adjusted to reflect this current implementation. shown originally, they reflected an aborted attempt to avoid a testing aberration not fully understood at the time. . added a summary of the memory overhead cost of HST_t processing to that UNREF_RPTHASH output at unref time. . a 'PIDs at max depth:' portion of that UNREF_RPTHASH enabled #define is now published only when the maximum depth of hash table entry chains exceed depths of one. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-20 10:30:00 +05:30
/* shame on this caller, they didn't change anything. and unless they have
altered the depth of the stacks we're not gonna change anything either! */
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
if (info->curitems == newnumitems + 1
&& !memcmp(info->items, newitems, sizeof(enum pids_item) * newnumitems))
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
return 0;
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
if (info->maxitems < newnumitems + 1) {
if (info->dirty_stacks)
cleanup_stacks_all(info);
// allow for our PROCPS_PIDS_logical_end
info->maxitems = newnumitems + 1;
if (!(info->items = realloc(info->items, sizeof(enum pids_item) * info->maxitems)))
return -ENOMEM;
extents_free_all(info);
}
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
if (info->dirty_stacks)
cleanup_stacks_all(info);
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
memcpy(info->items, newitems, sizeof(enum pids_item) * newnumitems);
info->items[newnumitems] = PROCPS_PIDS_logical_end;
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
// account for above PROCPS_PIDS_logical_end
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
info->curitems = newnumitems + 1;
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
itemize_stacks_all(info);
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
libflags_set(info);
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
return 0;
} // end: procps_pids_reset
library: correct a flawed approach for PROCPS_FILL_UID Gosh, just because nobody uses some newlib provision I guess, since it is being offered, it ought to actually be tested at some point. Well, that point just arrived and guess what? A surprise: some bugs were discovered. The procps_pids_select function established a for loop wherein readproc is called until the passed 'maxthese' limit. Unfortunately this was incorrect for 2 reasons: 1. For PROCPS_FILL_PID results are limited by what the oldlib finds, having established the pid list at open. Total found already cannot exceed a passed 'maxthese'; 2. With PROCPS_FILL_UID, returned results could exceed a 'maxthese' thus making the for loop incorrect again. [ plus yours truly neglected to forward the required ] [ UIDs total to our old library, another oops biggie ] In summary: the loop should have been forever, exiting only when all those identified procs had been located. So, while addressing those bugs, I've consolidated all the retrieval code (initialize, iterate, summarize) in a single helper function which will now serve both the procps_pids_reap and select functions. And as a result those guys were reduced to quite trivial housekeeping. This patch, hopefully, completes the normalization for reap/select (fill), which began with references shown. Reference(s): commit 0c953eccc5fe7240be9d272e1b6c0ce8769d8ed2 commit 747dfc5987e6e91ea3a8575de307e2892790c598 Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-08 10:30:00 +05:30
/* procps_pids_select():
*
* Harvest any processes matching the specified PID or UID and provide the
* result stacks along with a summary of the information gathered.
*
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
* Returns: pointer to a pids_fetch struct on success, NULL on error.
library: correct a flawed approach for PROCPS_FILL_UID Gosh, just because nobody uses some newlib provision I guess, since it is being offered, it ought to actually be tested at some point. Well, that point just arrived and guess what? A surprise: some bugs were discovered. The procps_pids_select function established a for loop wherein readproc is called until the passed 'maxthese' limit. Unfortunately this was incorrect for 2 reasons: 1. For PROCPS_FILL_PID results are limited by what the oldlib finds, having established the pid list at open. Total found already cannot exceed a passed 'maxthese'; 2. With PROCPS_FILL_UID, returned results could exceed a 'maxthese' thus making the for loop incorrect again. [ plus yours truly neglected to forward the required ] [ UIDs total to our old library, another oops biggie ] In summary: the loop should have been forever, exiting only when all those identified procs had been located. So, while addressing those bugs, I've consolidated all the retrieval code (initialize, iterate, summarize) in a single helper function which will now serve both the procps_pids_reap and select functions. And as a result those guys were reduced to quite trivial housekeeping. This patch, hopefully, completes the normalization for reap/select (fill), which began with references shown. Reference(s): commit 0c953eccc5fe7240be9d272e1b6c0ce8769d8ed2 commit 747dfc5987e6e91ea3a8575de307e2892790c598 Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-08 10:30:00 +05:30
*/
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
PROCPS_EXPORT struct pids_fetch *procps_pids_select (
library: normalize the <pids> fill/selection interface After wrestling with the conversion of both top and ps the differences between reap (all) & fill (select) has become increasingly inconvenient. So this patch simply normalizes that API making returned results identical. The former procps_pids_stacks_fill identifier will now be known as procps_pids_select which serves as logical counterpart to the existing procps_pids_reap function. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-01 10:30:00 +05:30
struct procps_pidsinfo *info,
unsigned *these,
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
int numthese,
library: changed use of 'fill' to 'select', <pids> API Remove a remnant of this new API's evolution. Here we no longer fill stacks, rather we reap and select them. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-04-18 10:30:00 +05:30
enum pids_select_type which)
library: normalize the <pids> fill/selection interface After wrestling with the conversion of both top and ps the differences between reap (all) & fill (select) has become increasingly inconvenient. So this patch simply normalizes that API making returned results identical. The former procps_pids_stacks_fill identifier will now be known as procps_pids_select which serves as logical counterpart to the existing procps_pids_reap function. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-01 10:30:00 +05:30
{
unsigned ids[FILL_ID_MAX + 1];
library: correct a flawed approach for PROCPS_FILL_UID Gosh, just because nobody uses some newlib provision I guess, since it is being offered, it ought to actually be tested at some point. Well, that point just arrived and guess what? A surprise: some bugs were discovered. The procps_pids_select function established a for loop wherein readproc is called until the passed 'maxthese' limit. Unfortunately this was incorrect for 2 reasons: 1. For PROCPS_FILL_PID results are limited by what the oldlib finds, having established the pid list at open. Total found already cannot exceed a passed 'maxthese'; 2. With PROCPS_FILL_UID, returned results could exceed a 'maxthese' thus making the for loop incorrect again. [ plus yours truly neglected to forward the required ] [ UIDs total to our old library, another oops biggie ] In summary: the loop should have been forever, exiting only when all those identified procs had been located. So, while addressing those bugs, I've consolidated all the retrieval code (initialize, iterate, summarize) in a single helper function which will now serve both the procps_pids_reap and select functions. And as a result those guys were reduced to quite trivial housekeeping. This patch, hopefully, completes the normalization for reap/select (fill), which began with references shown. Reference(s): commit 0c953eccc5fe7240be9d272e1b6c0ce8769d8ed2 commit 747dfc5987e6e91ea3a8575de307e2892790c598 Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-08 10:30:00 +05:30
int rc;
library: normalize the <pids> fill/selection interface After wrestling with the conversion of both top and ps the differences between reap (all) & fill (select) has become increasingly inconvenient. So this patch simply normalizes that API making returned results identical. The former procps_pids_stacks_fill identifier will now be known as procps_pids_select which serves as logical counterpart to the existing procps_pids_reap function. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-01 10:30:00 +05:30
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
if (info == NULL || these == NULL)
library: normalize the <pids> fill/selection interface After wrestling with the conversion of both top and ps the differences between reap (all) & fill (select) has become increasingly inconvenient. So this patch simply normalizes that API making returned results identical. The former procps_pids_stacks_fill identifier will now be known as procps_pids_select which serves as logical counterpart to the existing procps_pids_reap function. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-01 10:30:00 +05:30
return NULL;
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
if (numthese < 1 || numthese > FILL_ID_MAX)
library: normalize the <pids> fill/selection interface After wrestling with the conversion of both top and ps the differences between reap (all) & fill (select) has become increasingly inconvenient. So this patch simply normalizes that API making returned results identical. The former procps_pids_stacks_fill identifier will now be known as procps_pids_select which serves as logical counterpart to the existing procps_pids_reap function. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-01 10:30:00 +05:30
return NULL;
library: changed use of 'fill' to 'select', <pids> API Remove a remnant of this new API's evolution. Here we no longer fill stacks, rather we reap and select them. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-04-18 10:30:00 +05:30
if (which != PROCPS_SELECT_PID && which != PROCPS_SELECT_UID)
library: normalize the <pids> fill/selection interface After wrestling with the conversion of both top and ps the differences between reap (all) & fill (select) has become increasingly inconvenient. So this patch simply normalizes that API making returned results identical. The former procps_pids_stacks_fill identifier will now be known as procps_pids_select which serves as logical counterpart to the existing procps_pids_reap function. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-01 10:30:00 +05:30
return NULL;
library: optional parms protection missing, <PIDS> api When those items were made dynamic at 'new' time, some other functions, previously assured of their presence, failed to verify a 'reset' had acually been requested. This commit just corrects that oversight and avoids an attempt to 'assign_results' when no items are present. Reference(s): . when items/numitems became optional commit 9ebadc1438a6665a98a9f0782523b0f9a2a6248f Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-15 10:30:00 +05:30
/* with items & numitems technically optional at 'new' time, it's
expected 'reset' will have been called -- but just in case ... */
if (!info->curitems)
return NULL;
library: normalize the <pids> fill/selection interface After wrestling with the conversion of both top and ps the differences between reap (all) & fill (select) has become increasingly inconvenient. So this patch simply normalizes that API making returned results identical. The former procps_pids_stacks_fill identifier will now be known as procps_pids_select which serves as logical counterpart to the existing procps_pids_reap function. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-01 10:30:00 +05:30
library: changed use of 'fill' to 'select', <pids> API Remove a remnant of this new API's evolution. Here we no longer fill stacks, rather we reap and select them. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-04-18 10:30:00 +05:30
// this zero delimiter is really only needed with PROCPS_SELECT_PID
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
memcpy(ids, these, sizeof(unsigned) * numthese);
ids[numthese] = 0;
library: normalize the <pids> fill/selection interface After wrestling with the conversion of both top and ps the differences between reap (all) & fill (select) has become increasingly inconvenient. So this patch simply normalizes that API making returned results identical. The former procps_pids_stacks_fill identifier will now be known as procps_pids_select which serves as logical counterpart to the existing procps_pids_reap function. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-01 10:30:00 +05:30
library: a couple of miscellaneous changes, <PIDS> api This patch represents the following minor adjustments: . old PROCTAB used with select & reap more descriptive . reformatted PROCPS_PIDS_VAL macro like other 3rd gen Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-18 10:30:00 +05:30
if (!oldproc_open(&info->fetch_PT, (info->oldflags | which), ids, numthese))
library: normalize the <pids> fill/selection interface After wrestling with the conversion of both top and ps the differences between reap (all) & fill (select) has become increasingly inconvenient. So this patch simply normalizes that API making returned results identical. The former procps_pids_stacks_fill identifier will now be known as procps_pids_select which serves as logical counterpart to the existing procps_pids_reap function. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-01 10:30:00 +05:30
return NULL;
library: correct a flawed approach for PROCPS_FILL_UID Gosh, just because nobody uses some newlib provision I guess, since it is being offered, it ought to actually be tested at some point. Well, that point just arrived and guess what? A surprise: some bugs were discovered. The procps_pids_select function established a for loop wherein readproc is called until the passed 'maxthese' limit. Unfortunately this was incorrect for 2 reasons: 1. For PROCPS_FILL_PID results are limited by what the oldlib finds, having established the pid list at open. Total found already cannot exceed a passed 'maxthese'; 2. With PROCPS_FILL_UID, returned results could exceed a 'maxthese' thus making the for loop incorrect again. [ plus yours truly neglected to forward the required ] [ UIDs total to our old library, another oops biggie ] In summary: the loop should have been forever, exiting only when all those identified procs had been located. So, while addressing those bugs, I've consolidated all the retrieval code (initialize, iterate, summarize) in a single helper function which will now serve both the procps_pids_reap and select functions. And as a result those guys were reduced to quite trivial housekeeping. This patch, hopefully, completes the normalization for reap/select (fill), which began with references shown. Reference(s): commit 0c953eccc5fe7240be9d272e1b6c0ce8769d8ed2 commit 747dfc5987e6e91ea3a8575de307e2892790c598 Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-08 10:30:00 +05:30
info->read_something = readproc;
library: normalize the <pids> fill/selection interface After wrestling with the conversion of both top and ps the differences between reap (all) & fill (select) has become increasingly inconvenient. So this patch simply normalizes that API making returned results identical. The former procps_pids_stacks_fill identifier will now be known as procps_pids_select which serves as logical counterpart to the existing procps_pids_reap function. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-01 10:30:00 +05:30
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
rc = stacks_fetch(info);
library: normalize the <pids> fill/selection interface After wrestling with the conversion of both top and ps the differences between reap (all) & fill (select) has become increasingly inconvenient. So this patch simply normalizes that API making returned results identical. The former procps_pids_stacks_fill identifier will now be known as procps_pids_select which serves as logical counterpart to the existing procps_pids_reap function. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-01 10:30:00 +05:30
library: a couple of miscellaneous changes, <PIDS> api This patch represents the following minor adjustments: . old PROCTAB used with select & reap more descriptive . reformatted PROCPS_PIDS_VAL macro like other 3rd gen Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-18 10:30:00 +05:30
oldproc_close(&info->fetch_PT);
library: correct a flawed approach for PROCPS_FILL_UID Gosh, just because nobody uses some newlib provision I guess, since it is being offered, it ought to actually be tested at some point. Well, that point just arrived and guess what? A surprise: some bugs were discovered. The procps_pids_select function established a for loop wherein readproc is called until the passed 'maxthese' limit. Unfortunately this was incorrect for 2 reasons: 1. For PROCPS_FILL_PID results are limited by what the oldlib finds, having established the pid list at open. Total found already cannot exceed a passed 'maxthese'; 2. With PROCPS_FILL_UID, returned results could exceed a 'maxthese' thus making the for loop incorrect again. [ plus yours truly neglected to forward the required ] [ UIDs total to our old library, another oops biggie ] In summary: the loop should have been forever, exiting only when all those identified procs had been located. So, while addressing those bugs, I've consolidated all the retrieval code (initialize, iterate, summarize) in a single helper function which will now serve both the procps_pids_reap and select functions. And as a result those guys were reduced to quite trivial housekeeping. This patch, hopefully, completes the normalization for reap/select (fill), which began with references shown. Reference(s): commit 0c953eccc5fe7240be9d272e1b6c0ce8769d8ed2 commit 747dfc5987e6e91ea3a8575de307e2892790c598 Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-08 10:30:00 +05:30
// no guarantee any pids/uids were found
library: misc tweaks for code and comments, <PIDS> api This commit simply tries to keep naming plus formating conventions on a par with the continuing climb up that learning curve. These changes were suggested following <slabinfo> sources upgrade from 2nd to 3rd generation. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-06 10:30:00 +05:30
return (rc > -1) ? &info->fetch.results : NULL;
library: normalize the <pids> fill/selection interface After wrestling with the conversion of both top and ps the differences between reap (all) & fill (select) has become increasingly inconvenient. So this patch simply normalizes that API making returned results identical. The former procps_pids_stacks_fill identifier will now be known as procps_pids_select which serves as logical counterpart to the existing procps_pids_reap function. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-01 10:30:00 +05:30
} // end: procps_pids_select
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
/*
library: rename the 'procps_pids_stacks_sort' function The above function was the sole public function in the <pids> interface to use the word 'stacks' in its name. All of the others dealt exclusively with their duties, So this commit normalizes that outlier by renaming it. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-11 10:30:00 +05:30
* procps_pids_sort():
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
*
library: the overlooked twerks (oops, tweaks), 3rd gen Yes, all of these changes are strictly cosmetic. It is likely symptomatic of some deep-seated character flaw. [ or, it might be because of a certain pride in this ] [ new library and the desire to make it even better! ] Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-07 10:30:00 +05:30
* Sort stacks anchored in the passed stack pointers array
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
* based on the designated sort enumerator and specified order.
*
* Returns those same addresses sorted.
*
* Note: all of the stacks must be homogeneous (of equal length and content).
*/
library: rename the 'procps_pids_stacks_sort' function The above function was the sole public function in the <pids> interface to use the word 'stacks' in its name. All of the others dealt exclusively with their duties, So this commit normalizes that outlier by renaming it. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-11 10:30:00 +05:30
PROCPS_EXPORT struct pids_stack **procps_pids_sort (
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
struct procps_pidsinfo *info,
library: miscellaneous tweaks to pid code and comments . traded a complex misaligned memory allocation scheme in the make_hist function for a simple aligned scheme. plus memory allocation increases are globally defined. . changed 1 parameter for procps_pids_stacks_sort() to better reflect the 'array of pointers', not an address of a pointer as is used with guys such as 'new/unref'. . the pids_reap struct was changed slightly to make it more reflective of it's actual implementation details. . the Item_table member .mustfree is now .needfree and that .makehist was now made .needhist for consistency. . reduced the number of separate 'return NULL;' source statements in that primary procps_pids_reap() routine. . ensured consistent reference to sizeof(void *) & not occasional reference to sizeof(void*) without a space. . rather than enable/disable validate_stacks via a #if in the function body, it is now handled via a #define. . some comments in the procps_pids_reset function were adjusted to reflect this current implementation. shown originally, they reflected an aborted attempt to avoid a testing aberration not fully understood at the time. . added a summary of the memory overhead cost of HST_t processing to that UNREF_RPTHASH output at unref time. . a 'PIDs at max depth:' portion of that UNREF_RPTHASH enabled #define is now published only when the maximum depth of hash table entry chains exceed depths of one. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-20 10:30:00 +05:30
struct pids_stack *stacks[],
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
int numstacked,
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
enum pids_item sortitem,
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
enum pids_sort_order order)
{
struct sort_parms parms;
struct pids_result *p;
int offset;
if (info == NULL || stacks == NULL)
return NULL;
library: accumulated changes to <pids> code & comments A patch containing the following miscellaneous tweaks: . avoided distortions unique to PROCPS_PIDS_TICS_DELTA . addressed several smatch warnings and/or suggestions . ensured oldproc_close invoked should tally_proc fail . keeping that namespace clean, added a missing #undef . added 2 comments acknowledging pids_item as unsigned . added/clarified comments regarding proc flags & strv From smatch analysis: . these were indeed boo-boos pids.c:580 make_hist() warn: the 'Hr' macro might need parens pids.c:1058 procps_pids_reap() warn: variable dereferenced before check 'info' (see line 1056) . these were not errors (and we did double check) pids.c:1067 procps_pids_reap() warn: double check that we're allocating correct size: 8 vs 128 pids.c:1068 procps_pids_reap() warn: double check that we're allocating correct size: 8 vs 128 Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-28 10:30:00 +05:30
// a pids_item is currently unsigned, but we'll protect our future
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
if (sortitem < 0 || sortitem >= PROCPS_PIDS_logical_end)
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
return NULL;
library: misc tweaks for code and comments, <PIDS> api This commit simply tries to keep naming plus formating conventions on a par with the continuing climb up that learning curve. These changes were suggested following <slabinfo> sources upgrade from 2nd to 3rd generation. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-06 10:30:00 +05:30
if (order != PROCPS_PIDS_ASCEND && order != PROCPS_PIDS_DESCEND)
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
return NULL;
if (numstacked < 2)
return stacks;
offset = 0;
p = stacks[0]->head;
for (;;) {
library: standardize portions of interface, <PIDS> api This represents a rather major interface redesign. The following highlights most of the changes/enhancements. . The 'read' interface (employed by pgrep & pidof) saw the biggest change. The 'open', 'next' and 'shut' guys all went bye-bye, replaced by a single 'get' function. . The items specified at 'new' time no longer serve as the maximum. In fact, items & numitems are now treated as optional, should callers prefer to wait until later when the 'reset' function would then become mandatory. . Even at 'reset' time, the stacks are not tied to any sort of maximum. They will grow dynamically as needed. . The order of some parameters was changed to parallel that found in our other APIs. Specifically, when items & numitems are needed they're specified in that order. . A user will no longer be prevented from concurrently employing any accessor functions. In other words, that 'get' (old 'read') won't preclude 'reap' and 'select'. . A duplicate enumerator was found dealing with locked resident pages. So, the name VM_LOCK was eliminated in favor of VM_RSS_LOCKED, which is way more descriptive. . The struct address returned to callers following any reap() or select() is now more sharable as pids_fetch. . Some input parameter names were changed to make them more descriptive of the intended purpose/requirements. ------------------------------------------------------ Internally, there were numerous implementation changes made that did not directly impact any potential users. . That #define FPRINT_STACKS was eliminated along with the associated supporting function and its invocation. . Addresses returned following 'reap' or 'select' will now be NULL delimited, so one has the option of stacks access via the total count or this new NULL fencepost. . Input params were simplified and generalized in both oldproc_open() & close() to enable more than 1 PROCTAB to be open simultaneously, which was required for get. . The PROCPS_PIDS_logical_end enum was relocated after the Item_table making the need to keep it synchronized more apparent (if the table expands it's right there). . The 'Public function' section of the source file was subdivided into 1) the three basic required functions; and 2) functions that can sometimes vary between APIs. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-05-14 10:30:00 +05:30
if (p->item == sortitem)
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
break;
++offset;
if (offset >= info->curitems)
return NULL;
library: protect against a future breakage, <pids> ABI By co-mingling both external/internal identifiers with actual implementation code, potential future additions to our API would have been considerably more difficult. So, this patch will now rely solely on internal/hidden identifiers serving as fenceposts in validation logic. And if the following convention is used for new fields we will maintain that enumerator alphabetic ordering a a little longer (even though 2 user fields now don't): . PROCPS_PIDS_XTRA_FOO_A, PROCPS_PIDS_XTRA_FOO_B, etc. Reference(s): http://www.freelists.org/post/procps/me-too-newlib,7 Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-04-18 10:30:00 +05:30
if (p->item >= PROCPS_PIDS_logical_end)
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
return NULL;
++p;
}
parms.offset = offset;
parms.order = order;
library: add real string vectors to the pids interface After experimenting with an adaptation of pidof to the new pids interface, it became apparent that vectorized versions of those command lines would be necessary. So this commit adds that option and the strv result type. And since the stage had been set, a vectorized version of PROCPS_PIDS_ENVIRON & PROCPS_PIDS_CGROUP was added. Lastly, any use of 'const' in the result structure was removed so callers need not be bothered with casts and compiler warnings. Hopefully, they'll respect a stack. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-26 10:30:00 +05:30
qsort_r(stacks, numstacked, sizeof(void *), (QSR_t)Item_table[p->item].sortfunc, &parms);
library: implement task/thread support via the new api This commit is the culmination of efforts to modernize the library api. It should be treated as a first blush attempt, especially since I have absolutely no library design experience. But I did have a very strong desire to lessen the new library's impact on the top program. Under this new api, a 'stack' is the equivalent of the old proc_t. It can be seen as a variable length record whose contents & order is under complete user control. That initial stack/record configuration is established at procps_pids_new() time and will probably serve most program needs. But, a dynamic & demanding program like top will later change a stack via procps_pids_reset(). For programs like top & ps, procps_pids_reap() will be the function that will retrieve all tasks and threads. Any program that needs to filter / select only certain processes or users have available other functions that can be used: procps_pids_stacks_alloc, fill & dealloc. This implementation attempts to maximize that existing proven libprocps code base. As we gain more experience such actual code can be migrated into the pids.c file. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-08-19 10:30:00 +05:30
return stacks;
library: rename the 'procps_pids_stacks_sort' function The above function was the sole public function in the <pids> interface to use the word 'stacks' in its name. All of the others dealt exclusively with their duties, So this commit normalizes that outlier by renaming it. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-11 10:30:00 +05:30
} // end: procps_pids_sort
Reference in New Issue Copy Permalink