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>
|
|
|
|
|
|
2016-06-09 10:30:00 +05:30
|
|
|
#include <proc/devname.h>
|
2017-05-12 10:33:00 +05:30
|
|
|
#include <proc/numa.h>
|
2016-06-09 10:30:00 +05:30
|
|
|
#include <proc/readproc.h>
|
2015-09-30 10:30:00 +05:30
|
|
|
#include <proc/sysinfo.h>
|
|
|
|
|
#include <proc/uptime.h>
|
2016-06-09 10:30:00 +05:30
|
|
|
#include <proc/wchan.h>
|
|
|
|
|
|
|
|
|
|
#include <proc/procps-private.h>
|
|
|
|
|
#include <proc/pids.h>
|
2015-08-19 10:30:00 +05:30
|
|
|
|
|
|
|
|
|
2016-09-28 20:40:10 +05:30
|
|
|
//#define UNREF_RPTHASH // report hash details at uref() time
|
2015-08-19 10:30:00 +05:30
|
|
|
|
2016-09-28 20:40:10 +05:30
|
|
|
#define FILL_ID_MAX 255 // upper limit with select of pid/uid
|
|
|
|
|
#define STACKS_INCR 128 // amount reap stack allocations grow
|
|
|
|
|
#define NEWOLD_INCR 128 // amt by which hist allocations grow
|
2015-08-19 10:30:00 +05:30
|
|
|
|
2020-08-10 10:30:00 +05:30
|
|
|
/* ------------------------------------------------------------------------- +
|
|
|
|
|
this provision can be used to ensure that our Item_table was synchronized |
|
|
|
|
|
with those enumerators found in the associated header file. It's intended |
|
|
|
|
|
to only be used locally (& temporarily) at some point prior to a release! | */
|
|
|
|
|
// #define ITEMTABLE_DEBUG //----------------------------------------------- |
|
|
|
|
|
// ------------------------------------------------------------------------- +
|
|
|
|
|
|
2015-08-19 10:30:00 +05:30
|
|
|
|
2015-10-01 10:30:00 +05:30
|
|
|
struct stacks_extent {
|
2015-08-19 10:30:00 +05:30
|
|
|
int ext_numstacks;
|
|
|
|
|
struct stacks_extent *next;
|
2017-09-14 10:30:00 +05:30
|
|
|
struct pids_stack **stacks;
|
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
|
|
|
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
|
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
|
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
|
|
|
};
|
|
|
|
|
|
2016-07-21 10:30:00 +05:30
|
|
|
struct pids_info {
|
2015-08-19 10:30:00 +05:30
|
|
|
int refcount;
|
2016-04-25 10:30:00 +05:30
|
|
|
int maxitems; // includes 'logical_end' delimiter
|
2015-08-19 10:30:00 +05:30
|
|
|
int curitems; // includes 'logical_end' delimiter
|
2016-04-25 10:30:00 +05:30
|
|
|
enum pids_item *items; // includes 'logical_end' delimiter
|
2015-09-10 10:30:00 +05:30
|
|
|
struct stacks_extent *extents; // anchor for all resettable extents
|
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
|
2015-08-19 10:30:00 +05:30
|
|
|
int history_yes; // need historical data
|
|
|
|
|
struct history_info *hist; // pointer to historical support data
|
2015-08-26 10:30:00 +05:30
|
|
|
proc_t*(*read_something)(PROCTAB*, proc_t*); // readproc/readeither via which
|
2015-08-19 10:30:00 +05:30
|
|
|
unsigned pgs2k_shift; // to convert some proc vaules
|
2016-06-06 10:30:00 +05:30
|
|
|
unsigned oldflags; // the old library PROC_FILL flagss
|
2016-06-18 10:30:00 +05:30
|
|
|
PROCTAB *fetch_PT; // oldlib interface for 'select' & 'reap'
|
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
|
2016-06-18 10:30:00 +05:30
|
|
|
PROCTAB *get_PT; // oldlib interface for active 'get'
|
2019-05-02 10:30:00 +05:30
|
|
|
struct stacks_extent *get_ext; // for active 'get' (also within '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
|
|
|
enum pids_fetch_type get_type; // last known type of 'get' request
|
2017-11-18 10:30:00 +05:30
|
|
|
int seterr; // an ENOMEM encountered during assign
|
2015-08-19 10:30:00 +05:30
|
|
|
};
|
|
|
|
|
|
|
|
|
|
|
library: minimize the use of 'cleanup_stacks' routines
Some parts of our newlib implementation are the result
of functions which have been propagated from module to
module. In particular, those 'cleanup_stacks' routines
are all similar & likely originated in the <pids> api.
In that interface there was a need to free dynamically
acquired memory before the result structure was reused
to satisfy subsequent 'get', 'select' or 'reap' calls.
This, in turn, led to a concept of 'dirty' stacks with
the need to call one of two 'cleanup_stack' functions.
None of the remaining interfaces deal with such memory
yet they each had their own 'cleanup_stack' functions.
Those functions were responsible for resetting each of
the result unions to zero, excluding any 'noop' items.
The bottom line is that for all interfaces, repetitive
calls would require iterating through the stack(s) two
separate times: once to 'cleanup' another to 'assign'.
With this commit we will reduce iterations to just the
'assign' routine. A reset to zero will be accomplished
in the 'extra' item set routine (which is the only one
actually requiring any reset). All other items will be
reinitialized automatically by a new current set value
or upon reallocation when an items compliment changes.
In the <pids> interface, any freeing of dynamic memory
could have been accomplished by adding that 'freefunc'
check to the 'assign' function. However, that requires
an Item_table test with every item. Instead, we'll now
satisfy such needs as the very first step in those set
functions responsible for dynamically acquired memory.
[ the <pids> api retains 2 'cleanup_stack' functions ]
[ to accommodate stack(s) 'reset' & to serve 'unref' ]
Lastly, all the 'itemize_stack' functions were tweaked
by eliminating an unnecessary initialization of result
unions. That objective was already accomplished by the
calloc() in a 'stacks_alloc' function or the remaining
'cleanup_stack' routine found in the <pids> interface.
Signed-off-by: Jim Warner <james.warner@comcast.net>
2019-05-22 10:30:00 +05:30
|
|
|
// ___ Free Storage Support |||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|
|
|
|
|
|
|
|
#define freNAME(t) free_pids_ ## t
|
|
|
|
|
|
|
|
|
|
static void freNAME(str) (struct pids_result *R) {
|
|
|
|
|
if (R->result.str) free(R->result.str);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void freNAME(strv) (struct pids_result *R) {
|
|
|
|
|
if (R->result.strv && *R->result.strv) free(*R->result.strv);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
2015-08-19 10:30:00 +05:30
|
|
|
// ___ Results 'Set' Support ||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|
|
|
|
2016-08-21 10:30:00 +05:30
|
|
|
#define setNAME(e) set_pids_ ## e
|
2015-08-19 10:30:00 +05:30
|
|
|
#define setDECL(e) static void setNAME(e) \
|
2016-07-21 10:30:00 +05:30
|
|
|
(struct pids_info *I, struct pids_result *R, proc_t *P)
|
2015-08-19 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
|
|
|
/* convert pages to kib */
|
2015-08-19 10:30:00 +05:30
|
|
|
#define CVT_set(e,t,x) setDECL(e) { \
|
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 */
|
2015-08-19 10:30:00 +05:30
|
|
|
#define DUP_set(e,x) setDECL(e) { \
|
library: minimize the use of 'cleanup_stacks' routines
Some parts of our newlib implementation are the result
of functions which have been propagated from module to
module. In particular, those 'cleanup_stacks' routines
are all similar & likely originated in the <pids> api.
In that interface there was a need to free dynamically
acquired memory before the result structure was reused
to satisfy subsequent 'get', 'select' or 'reap' calls.
This, in turn, led to a concept of 'dirty' stacks with
the need to call one of two 'cleanup_stack' functions.
None of the remaining interfaces deal with such memory
yet they each had their own 'cleanup_stack' functions.
Those functions were responsible for resetting each of
the result unions to zero, excluding any 'noop' items.
The bottom line is that for all interfaces, repetitive
calls would require iterating through the stack(s) two
separate times: once to 'cleanup' another to 'assign'.
With this commit we will reduce iterations to just the
'assign' routine. A reset to zero will be accomplished
in the 'extra' item set routine (which is the only one
actually requiring any reset). All other items will be
reinitialized automatically by a new current set value
or upon reallocation when an items compliment changes.
In the <pids> interface, any freeing of dynamic memory
could have been accomplished by adding that 'freefunc'
check to the 'assign' function. However, that requires
an Item_table test with every item. Instead, we'll now
satisfy such needs as the very first step in those set
functions responsible for dynamically acquired memory.
[ the <pids> api retains 2 'cleanup_stack' functions ]
[ to accommodate stack(s) 'reset' & to serve 'unref' ]
Lastly, all the 'itemize_stack' functions were tweaked
by eliminating an unnecessary initialization of result
unions. That objective was already accomplished by the
calloc() in a 'stacks_alloc' function or the remaining
'cleanup_stack' routine found in the <pids> interface.
Signed-off-by: Jim Warner <james.warner@comcast.net>
2019-05-22 10:30:00 +05:30
|
|
|
freNAME(str)(R); \
|
2017-11-18 10:30:00 +05:30
|
|
|
if (!(R->result.str = strdup(P-> x))) I->seterr = 1; }
|
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 */
|
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 ... */
|
2015-08-19 10:30:00 +05:30
|
|
|
#define STR_set(e,x) setDECL(e) { \
|
library: minimize the use of 'cleanup_stacks' routines
Some parts of our newlib implementation are the result
of functions which have been propagated from module to
module. In particular, those 'cleanup_stacks' routines
are all similar & likely originated in the <pids> api.
In that interface there was a need to free dynamically
acquired memory before the result structure was reused
to satisfy subsequent 'get', 'select' or 'reap' calls.
This, in turn, led to a concept of 'dirty' stacks with
the need to call one of two 'cleanup_stack' functions.
None of the remaining interfaces deal with such memory
yet they each had their own 'cleanup_stack' functions.
Those functions were responsible for resetting each of
the result unions to zero, excluding any 'noop' items.
The bottom line is that for all interfaces, repetitive
calls would require iterating through the stack(s) two
separate times: once to 'cleanup' another to 'assign'.
With this commit we will reduce iterations to just the
'assign' routine. A reset to zero will be accomplished
in the 'extra' item set routine (which is the only one
actually requiring any reset). All other items will be
reinitialized automatically by a new current set value
or upon reallocation when an items compliment changes.
In the <pids> interface, any freeing of dynamic memory
could have been accomplished by adding that 'freefunc'
check to the 'assign' function. However, that requires
an Item_table test with every item. Instead, we'll now
satisfy such needs as the very first step in those set
functions responsible for dynamically acquired memory.
[ the <pids> api retains 2 'cleanup_stack' functions ]
[ to accommodate stack(s) 'reset' & to serve 'unref' ]
Lastly, all the 'itemize_stack' functions were tweaked
by eliminating an unnecessary initialization of result
unions. That objective was already accomplished by the
calloc() in a 'stacks_alloc' function or the remaining
'cleanup_stack' routine found in the <pids> interface.
Signed-off-by: Jim Warner <james.warner@comcast.net>
2019-05-22 10:30:00 +05:30
|
|
|
freNAME(str)(R); \
|
2017-11-18 10:30:00 +05:30
|
|
|
if (NULL != P-> x) { R->result.str = P-> x; P-> x = NULL; } \
|
|
|
|
|
else { R->result.str = strdup("[ duplicate " STRINGIFY(e) " ]"); \
|
|
|
|
|
if (!R->result.str) I->seterr = 1; } }
|
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) { \
|
library: minimize the use of 'cleanup_stacks' routines
Some parts of our newlib implementation are the result
of functions which have been propagated from module to
module. In particular, those 'cleanup_stacks' routines
are all similar & likely originated in the <pids> api.
In that interface there was a need to free dynamically
acquired memory before the result structure was reused
to satisfy subsequent 'get', 'select' or 'reap' calls.
This, in turn, led to a concept of 'dirty' stacks with
the need to call one of two 'cleanup_stack' functions.
None of the remaining interfaces deal with such memory
yet they each had their own 'cleanup_stack' functions.
Those functions were responsible for resetting each of
the result unions to zero, excluding any 'noop' items.
The bottom line is that for all interfaces, repetitive
calls would require iterating through the stack(s) two
separate times: once to 'cleanup' another to 'assign'.
With this commit we will reduce iterations to just the
'assign' routine. A reset to zero will be accomplished
in the 'extra' item set routine (which is the only one
actually requiring any reset). All other items will be
reinitialized automatically by a new current set value
or upon reallocation when an items compliment changes.
In the <pids> interface, any freeing of dynamic memory
could have been accomplished by adding that 'freefunc'
check to the 'assign' function. However, that requires
an Item_table test with every item. Instead, we'll now
satisfy such needs as the very first step in those set
functions responsible for dynamically acquired memory.
[ the <pids> api retains 2 'cleanup_stack' functions ]
[ to accommodate stack(s) 'reset' & to serve 'unref' ]
Lastly, all the 'itemize_stack' functions were tweaked
by eliminating an unnecessary initialization of result
unions. That objective was already accomplished by the
calloc() in a 'stacks_alloc' function or the remaining
'cleanup_stack' routine found in the <pids> interface.
Signed-off-by: Jim Warner <james.warner@comcast.net>
2019-05-22 10:30:00 +05:30
|
|
|
freNAME(strv)(R); \
|
2017-11-18 10:30:00 +05:30
|
|
|
if (NULL != P-> x) { R->result.strv = P-> x; P-> x = NULL; } \
|
2019-04-25 10:30:00 +05:30
|
|
|
else { R->result.strv = vectorize_this_str("[ duplicate " STRINGIFY(e) " ]"); \
|
2019-09-26 10:30:00 +05:30
|
|
|
if (!R->result.strv) I->seterr = 1; } }
|
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
|
|
|
|
2015-08-19 10:30:00 +05:30
|
|
|
|
library: minimize the use of 'cleanup_stacks' routines
Some parts of our newlib implementation are the result
of functions which have been propagated from module to
module. In particular, those 'cleanup_stacks' routines
are all similar & likely originated in the <pids> api.
In that interface there was a need to free dynamically
acquired memory before the result structure was reused
to satisfy subsequent 'get', 'select' or 'reap' calls.
This, in turn, led to a concept of 'dirty' stacks with
the need to call one of two 'cleanup_stack' functions.
None of the remaining interfaces deal with such memory
yet they each had their own 'cleanup_stack' functions.
Those functions were responsible for resetting each of
the result unions to zero, excluding any 'noop' items.
The bottom line is that for all interfaces, repetitive
calls would require iterating through the stack(s) two
separate times: once to 'cleanup' another to 'assign'.
With this commit we will reduce iterations to just the
'assign' routine. A reset to zero will be accomplished
in the 'extra' item set routine (which is the only one
actually requiring any reset). All other items will be
reinitialized automatically by a new current set value
or upon reallocation when an items compliment changes.
In the <pids> interface, any freeing of dynamic memory
could have been accomplished by adding that 'freefunc'
check to the 'assign' function. However, that requires
an Item_table test with every item. Instead, we'll now
satisfy such needs as the very first step in those set
functions responsible for dynamically acquired memory.
[ the <pids> api retains 2 'cleanup_stack' functions ]
[ to accommodate stack(s) 'reset' & to serve 'unref' ]
Lastly, all the 'itemize_stack' functions were tweaked
by eliminating an unnecessary initialization of result
unions. That objective was already accomplished by the
calloc() in a 'stacks_alloc' function or the remaining
'cleanup_stack' routine found in the <pids> interface.
Signed-off-by: Jim Warner <james.warner@comcast.net>
2019-05-22 10:30:00 +05:30
|
|
|
setDECL(noop) { (void)I; (void)R; (void)P; }
|
|
|
|
|
setDECL(extra) { (void)I; (void)P; R->result.ull_int = 0; }
|
|
|
|
|
|
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: 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)
|
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)
|
2018-06-22 10:30:00 +05:30
|
|
|
STR_set(EXE, exe)
|
2015-08-19 10:30:00 +05:30
|
|
|
REG_set(EXIT_SIGNAL, s_int, exit_signal)
|
|
|
|
|
REG_set(FLAGS, ul_int, flags)
|
2016-07-31 10:30:00 +05:30
|
|
|
REG_set(FLT_MAJ, ul_int, maj_flt)
|
library: refactored some header file items and origins
This commit is intended as a refinement of the patches
mentioned below, where origins/sources of newlib items
were added to the header files for user documentation.
However, if those additions are to be truly effective,
along with kernel documentation (where available), the
following prerequisites must also have been satisfied:
. our identifiers closely align with linux field names
. our derived items are documented or self-documenting
Satisfying those prerequisites prompted this patch and
for these changes, kernel sources were emphasized over
available documentation (shame on me, it should always
have been so). And, while some 'new' fields were found
to be conditional, they were included unconditionally.
These changes appear more extensive than they actually
need be since I have attempted to enforce some spacing
conventions. So, I've summarize the significant things
in the sections that follow. For a proper perspective,
use: 'git diff --ignore-space-change' (good as alias).
___________________________________________ <PIDS> api
This api is unique in that there exists many different
file/directory origins subordinate to /proc/<pid>. And
our item identifiers are sometimes coerced so as to be
able to group related or similar enumerators together.
So, users needed more help relating our identifiers to
an actual documented field. Thus, we will now also add
the field names as with 'stat: delayacct_blkio_ticks'.
Each item ending with a '_C' now consistently includes
both the parent's count/time plus waited for children.
That 'RTPRIO' guy was renamed/relocated as PRIORITY_RT
since its original name is an implementation artifact.
___________________________________________ <STAT> api
The only api change was to correct a typo ('dervied').
_________________________________________ <VMSTAT> api
Even ignoring white space, this interface received the
largest number of changes. Mostly, this was because of
deficiencies in the proc(5) documentation. Recall that
this documentation already sorely lacks any substance.
Usually, just kernel releases are noted, not contents.
When compared to kernel source, that proc(5) contained
many non-existent fields and also omitted many others.
________________________________________ <MEMINFO> api
Sadly, with this api many of the changes were simply a
correction of some earlier 'human error' where several
fields where hashed then tracked but never represented
with an item enumerator in this meminfo.h header file.
_______________________________________ <SLABINFO> api
The 'SLABS' (summary) & 'SLABNODE' items were reversed
since the former are derived from the separate caches.
More significantly, those 'SLABNODE' guys were renamed
to 'SLAB' since they concern individual caches and the
concept of 'nodes' is really an implementation detail.
Also, several enumerators were changed to more closely
agree with official slabinfo(5) documentation referred
to in what we're treating as a base document: proc(5).
Lastly, while those 'SLABS' items are solely a product
of our library and not represented in slabinfo(5), the
names attempt to parallel those found as 'SLAB' items.
______________________________________ <DISKSTATS> api
One enumeration identifier was changed so as to better
reflect its relationship to that actual documentation:
'Documentation/iostats.txt', as referenced in proc(5).
Reference(s):
. 12/2018, item origins added (and commit msg history)
commit 96d59cbf46b3ff687bd29fad4708074a0e1cea14
. 01/2019, <stat> origins tweaked
commit 201e816b26ddaccc923ec40977c92037cdd0c34e
Signed-off-by: Jim Warner <james.warner@comcast.net>
2019-03-12 11:30:00 +05:30
|
|
|
setDECL(FLT_MAJ_C) { (void)I; R->result.ul_int = P->maj_flt + P->cmaj_flt; }
|
2016-07-31 10:30:00 +05:30
|
|
|
REG_set(FLT_MAJ_DELTA, s_int, maj_delta)
|
|
|
|
|
REG_set(FLT_MIN, ul_int, min_flt)
|
library: refactored some header file items and origins
This commit is intended as a refinement of the patches
mentioned below, where origins/sources of newlib items
were added to the header files for user documentation.
However, if those additions are to be truly effective,
along with kernel documentation (where available), the
following prerequisites must also have been satisfied:
. our identifiers closely align with linux field names
. our derived items are documented or self-documenting
Satisfying those prerequisites prompted this patch and
for these changes, kernel sources were emphasized over
available documentation (shame on me, it should always
have been so). And, while some 'new' fields were found
to be conditional, they were included unconditionally.
These changes appear more extensive than they actually
need be since I have attempted to enforce some spacing
conventions. So, I've summarize the significant things
in the sections that follow. For a proper perspective,
use: 'git diff --ignore-space-change' (good as alias).
___________________________________________ <PIDS> api
This api is unique in that there exists many different
file/directory origins subordinate to /proc/<pid>. And
our item identifiers are sometimes coerced so as to be
able to group related or similar enumerators together.
So, users needed more help relating our identifiers to
an actual documented field. Thus, we will now also add
the field names as with 'stat: delayacct_blkio_ticks'.
Each item ending with a '_C' now consistently includes
both the parent's count/time plus waited for children.
That 'RTPRIO' guy was renamed/relocated as PRIORITY_RT
since its original name is an implementation artifact.
___________________________________________ <STAT> api
The only api change was to correct a typo ('dervied').
_________________________________________ <VMSTAT> api
Even ignoring white space, this interface received the
largest number of changes. Mostly, this was because of
deficiencies in the proc(5) documentation. Recall that
this documentation already sorely lacks any substance.
Usually, just kernel releases are noted, not contents.
When compared to kernel source, that proc(5) contained
many non-existent fields and also omitted many others.
________________________________________ <MEMINFO> api
Sadly, with this api many of the changes were simply a
correction of some earlier 'human error' where several
fields where hashed then tracked but never represented
with an item enumerator in this meminfo.h header file.
_______________________________________ <SLABINFO> api
The 'SLABS' (summary) & 'SLABNODE' items were reversed
since the former are derived from the separate caches.
More significantly, those 'SLABNODE' guys were renamed
to 'SLAB' since they concern individual caches and the
concept of 'nodes' is really an implementation detail.
Also, several enumerators were changed to more closely
agree with official slabinfo(5) documentation referred
to in what we're treating as a base document: proc(5).
Lastly, while those 'SLABS' items are solely a product
of our library and not represented in slabinfo(5), the
names attempt to parallel those found as 'SLAB' items.
______________________________________ <DISKSTATS> api
One enumeration identifier was changed so as to better
reflect its relationship to that actual documentation:
'Documentation/iostats.txt', as referenced in proc(5).
Reference(s):
. 12/2018, item origins added (and commit msg history)
commit 96d59cbf46b3ff687bd29fad4708074a0e1cea14
. 01/2019, <stat> origins tweaked
commit 201e816b26ddaccc923ec40977c92037cdd0c34e
Signed-off-by: Jim Warner <james.warner@comcast.net>
2019-03-12 11:30:00 +05:30
|
|
|
setDECL(FLT_MIN_C) { (void)I; R->result.ul_int = P->min_flt + P->cmin_flt; }
|
2016-07-31 10:30:00 +05:30
|
|
|
REG_set(FLT_MIN_DELTA, s_int, min_delta)
|
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)
|
2018-02-11 12:41:11 +05:30
|
|
|
REG_set(ID_LOGIN, s_int, luid)
|
2015-08-19 10:30:00 +05:30
|
|
|
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)
|
2018-12-01 11:30:00 +05:30
|
|
|
REG_set(ID_TID, s_int, tid)
|
2015-08-19 10:30:00 +05:30
|
|
|
REG_set(ID_TPGID, s_int, tpgid)
|
2016-07-01 10:30:00 +05:30
|
|
|
REG_set(LXCNAME, str, lxcname)
|
2016-07-31 10:30:00 +05:30
|
|
|
CVT_set(MEM_CODE, ul_int, trs)
|
|
|
|
|
REG_set(MEM_CODE_PGS, ul_int, trs)
|
|
|
|
|
CVT_set(MEM_DATA, ul_int, drs)
|
|
|
|
|
REG_set(MEM_DATA_PGS, ul_int, drs)
|
|
|
|
|
CVT_set(MEM_RES, ul_int, resident)
|
|
|
|
|
REG_set(MEM_RES_PGS, ul_int, resident)
|
|
|
|
|
CVT_set(MEM_SHR, ul_int, share)
|
|
|
|
|
REG_set(MEM_SHR_PGS, ul_int, share)
|
|
|
|
|
CVT_set(MEM_VIRT, ul_int, size)
|
|
|
|
|
REG_set(MEM_VIRT_PGS, ul_int, size)
|
|
|
|
|
REG_set(NICE, s_int, nice)
|
2015-08-19 10:30:00 +05:30
|
|
|
REG_set(NLWP, s_int, nlwp)
|
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])
|
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)
|
library: refactored some header file items and origins
This commit is intended as a refinement of the patches
mentioned below, where origins/sources of newlib items
were added to the header files for user documentation.
However, if those additions are to be truly effective,
along with kernel documentation (where available), the
following prerequisites must also have been satisfied:
. our identifiers closely align with linux field names
. our derived items are documented or self-documenting
Satisfying those prerequisites prompted this patch and
for these changes, kernel sources were emphasized over
available documentation (shame on me, it should always
have been so). And, while some 'new' fields were found
to be conditional, they were included unconditionally.
These changes appear more extensive than they actually
need be since I have attempted to enforce some spacing
conventions. So, I've summarize the significant things
in the sections that follow. For a proper perspective,
use: 'git diff --ignore-space-change' (good as alias).
___________________________________________ <PIDS> api
This api is unique in that there exists many different
file/directory origins subordinate to /proc/<pid>. And
our item identifiers are sometimes coerced so as to be
able to group related or similar enumerators together.
So, users needed more help relating our identifiers to
an actual documented field. Thus, we will now also add
the field names as with 'stat: delayacct_blkio_ticks'.
Each item ending with a '_C' now consistently includes
both the parent's count/time plus waited for children.
That 'RTPRIO' guy was renamed/relocated as PRIORITY_RT
since its original name is an implementation artifact.
___________________________________________ <STAT> api
The only api change was to correct a typo ('dervied').
_________________________________________ <VMSTAT> api
Even ignoring white space, this interface received the
largest number of changes. Mostly, this was because of
deficiencies in the proc(5) documentation. Recall that
this documentation already sorely lacks any substance.
Usually, just kernel releases are noted, not contents.
When compared to kernel source, that proc(5) contained
many non-existent fields and also omitted many others.
________________________________________ <MEMINFO> api
Sadly, with this api many of the changes were simply a
correction of some earlier 'human error' where several
fields where hashed then tracked but never represented
with an item enumerator in this meminfo.h header file.
_______________________________________ <SLABINFO> api
The 'SLABS' (summary) & 'SLABNODE' items were reversed
since the former are derived from the separate caches.
More significantly, those 'SLABNODE' guys were renamed
to 'SLAB' since they concern individual caches and the
concept of 'nodes' is really an implementation detail.
Also, several enumerators were changed to more closely
agree with official slabinfo(5) documentation referred
to in what we're treating as a base document: proc(5).
Lastly, while those 'SLABS' items are solely a product
of our library and not represented in slabinfo(5), the
names attempt to parallel those found as 'SLAB' items.
______________________________________ <DISKSTATS> api
One enumeration identifier was changed so as to better
reflect its relationship to that actual documentation:
'Documentation/iostats.txt', as referenced in proc(5).
Reference(s):
. 12/2018, item origins added (and commit msg history)
commit 96d59cbf46b3ff687bd29fad4708074a0e1cea14
. 01/2019, <stat> origins tweaked
commit 201e816b26ddaccc923ec40977c92037cdd0c34e
Signed-off-by: Jim Warner <james.warner@comcast.net>
2019-03-12 11:30:00 +05:30
|
|
|
REG_set(PRIORITY_RT, s_int, rtprio)
|
2015-08-19 10:30:00 +05:30
|
|
|
REG_set(PROCESSOR, u_int, processor)
|
2017-05-12 10:33:00 +05:30
|
|
|
setDECL(PROCESSOR_NODE) { (void)I; R->result.s_int = numa_node_of_cpu(P->processor); }
|
2016-07-31 10:30:00 +05:30
|
|
|
REG_set(RSS, ul_int, rss)
|
2015-08-19 10:30:00 +05:30
|
|
|
REG_set(RSS_RLIM, ul_int, rss_rlim)
|
2016-07-31 10:30:00 +05:30
|
|
|
REG_set(SCHED_CLASS, s_int, sched)
|
2015-08-19 10:30:00 +05:30
|
|
|
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)
|
2016-07-31 10:30:00 +05:30
|
|
|
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; }
|
|
|
|
|
REG_set(TICS_ALL_DELTA, s_int, pcpu)
|
|
|
|
|
REG_set(TICS_BLKIO, ull_int, blkio_tics)
|
|
|
|
|
REG_set(TICS_GUEST, ull_int, gtime)
|
|
|
|
|
setDECL(TICS_GUEST_C) { (void)I; R->result.ull_int = P->gtime + P->cgtime; }
|
2015-08-19 10:30:00 +05:30
|
|
|
REG_set(TICS_SYSTEM, ull_int, stime)
|
2016-07-31 10:30:00 +05:30
|
|
|
setDECL(TICS_SYSTEM_C) { (void)I; R->result.ull_int = P->stime + P->cstime; }
|
2015-08-19 10:30:00 +05:30
|
|
|
REG_set(TICS_USER, ull_int, utime)
|
2016-07-31 10:30:00 +05:30
|
|
|
setDECL(TICS_USER_C) { (void)I; R->result.ull_int = P->utime + P->cutime; }
|
|
|
|
|
setDECL(TIME_ALL) { R->result.ull_int = (P->utime + P->stime) / I->hertz; }
|
|
|
|
|
setDECL(TIME_ELAPSED) { unsigned long long t = P->start_time / I->hertz; R->result.ull_int = I->boot_seconds >= t ? (I->boot_seconds - t) : 0; }
|
2015-08-19 10:30:00 +05:30
|
|
|
REG_set(TIME_START, ull_int, start_time)
|
|
|
|
|
REG_set(TTY, s_int, tty)
|
library: minimize the use of 'cleanup_stacks' routines
Some parts of our newlib implementation are the result
of functions which have been propagated from module to
module. In particular, those 'cleanup_stacks' routines
are all similar & likely originated in the <pids> api.
In that interface there was a need to free dynamically
acquired memory before the result structure was reused
to satisfy subsequent 'get', 'select' or 'reap' calls.
This, in turn, led to a concept of 'dirty' stacks with
the need to call one of two 'cleanup_stack' functions.
None of the remaining interfaces deal with such memory
yet they each had their own 'cleanup_stack' functions.
Those functions were responsible for resetting each of
the result unions to zero, excluding any 'noop' items.
The bottom line is that for all interfaces, repetitive
calls would require iterating through the stack(s) two
separate times: once to 'cleanup' another to 'assign'.
With this commit we will reduce iterations to just the
'assign' routine. A reset to zero will be accomplished
in the 'extra' item set routine (which is the only one
actually requiring any reset). All other items will be
reinitialized automatically by a new current set value
or upon reallocation when an items compliment changes.
In the <pids> interface, any freeing of dynamic memory
could have been accomplished by adding that 'freefunc'
check to the 'assign' function. However, that requires
an Item_table test with every item. Instead, we'll now
satisfy such needs as the very first step in those set
functions responsible for dynamically acquired memory.
[ the <pids> api retains 2 'cleanup_stack' functions ]
[ to accommodate stack(s) 'reset' & to serve 'unref' ]
Lastly, all the 'itemize_stack' functions were tweaked
by eliminating an unnecessary initialization of result
unions. That objective was already accomplished by the
calloc() in a 'stacks_alloc' function or the remaining
'cleanup_stack' routine found in the <pids> interface.
Signed-off-by: Jim Warner <james.warner@comcast.net>
2019-05-22 10:30:00 +05:30
|
|
|
setDECL(TTY_NAME) { char buf[64]; freNAME(str)(R); dev_to_tty(buf, sizeof(buf), P->tty, P->tid, ABBREV_DEV); if (!(R->result.str = strdup(buf))) I->seterr = 1; }
|
|
|
|
|
setDECL(TTY_NUMBER) { char buf[64]; freNAME(str)(R); dev_to_tty(buf, sizeof(buf), P->tty, P->tid, ABBREV_DEV|ABBREV_TTY|ABBREV_PTS); if (!(R->result.str = strdup(buf))) I->seterr = 1; }
|
2016-07-31 10:30:00 +05:30
|
|
|
REG_set(VM_DATA, ul_int, vm_data)
|
|
|
|
|
REG_set(VM_EXE, ul_int, vm_exe)
|
|
|
|
|
REG_set(VM_LIB, ul_int, vm_lib)
|
|
|
|
|
REG_set(VM_RSS, ul_int, vm_rss)
|
|
|
|
|
REG_set(VM_RSS_ANON, ul_int, vm_rss_anon)
|
|
|
|
|
REG_set(VM_RSS_FILE, ul_int, vm_rss_file)
|
|
|
|
|
REG_set(VM_RSS_LOCKED, ul_int, vm_lock)
|
|
|
|
|
REG_set(VM_RSS_SHARED, ul_int, vm_rss_shared)
|
|
|
|
|
REG_set(VM_SIZE, ul_int, vm_size)
|
|
|
|
|
REG_set(VM_STACK, ul_int, vm_stack)
|
|
|
|
|
REG_set(VM_SWAP, ul_int, vm_swap)
|
|
|
|
|
setDECL(VM_USED) { (void)I; R->result.ul_int = P->vm_swap + P->vm_rss; }
|
2015-08-19 10:30:00 +05:30
|
|
|
REG_set(VSIZE_PGS, ul_int, vsize)
|
library: minimize the use of 'cleanup_stacks' routines
Some parts of our newlib implementation are the result
of functions which have been propagated from module to
module. In particular, those 'cleanup_stacks' routines
are all similar & likely originated in the <pids> api.
In that interface there was a need to free dynamically
acquired memory before the result structure was reused
to satisfy subsequent 'get', 'select' or 'reap' calls.
This, in turn, led to a concept of 'dirty' stacks with
the need to call one of two 'cleanup_stack' functions.
None of the remaining interfaces deal with such memory
yet they each had their own 'cleanup_stack' functions.
Those functions were responsible for resetting each of
the result unions to zero, excluding any 'noop' items.
The bottom line is that for all interfaces, repetitive
calls would require iterating through the stack(s) two
separate times: once to 'cleanup' another to 'assign'.
With this commit we will reduce iterations to just the
'assign' routine. A reset to zero will be accomplished
in the 'extra' item set routine (which is the only one
actually requiring any reset). All other items will be
reinitialized automatically by a new current set value
or upon reallocation when an items compliment changes.
In the <pids> interface, any freeing of dynamic memory
could have been accomplished by adding that 'freefunc'
check to the 'assign' function. However, that requires
an Item_table test with every item. Instead, we'll now
satisfy such needs as the very first step in those set
functions responsible for dynamically acquired memory.
[ the <pids> api retains 2 'cleanup_stack' functions ]
[ to accommodate stack(s) 'reset' & to serve 'unref' ]
Lastly, all the 'itemize_stack' functions were tweaked
by eliminating an unnecessary initialization of result
unions. That objective was already accomplished by the
calloc() in a 'stacks_alloc' function or the remaining
'cleanup_stack' routine found in the <pids> interface.
Signed-off-by: Jim Warner <james.warner@comcast.net>
2019-05-22 10:30:00 +05:30
|
|
|
setDECL(WCHAN_NAME) { freNAME(str)(R); if (!(R->result.str = strdup(lookup_wchan(P->tid)))) I->seterr = 1;; }
|
2015-08-19 10:30:00 +05:30
|
|
|
|
|
|
|
|
#undef setDECL
|
|
|
|
|
#undef CVT_set
|
|
|
|
|
#undef DUP_set
|
|
|
|
|
#undef REG_set
|
|
|
|
|
#undef STR_set
|
|
|
|
|
#undef VEC_set
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
// ___ Sorting Support ||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|
|
|
|
|
|
|
|
struct sort_parms {
|
|
|
|
|
int offset;
|
|
|
|
|
enum pids_sort_order order;
|
|
|
|
|
};
|
|
|
|
|
|
2016-08-21 10:30:00 +05:30
|
|
|
#define srtNAME(t) sort_pids_ ## t
|
2016-06-09 10:30:00 +05:30
|
|
|
#define srtDECL(t) static int srtNAME(t) \
|
|
|
|
|
(const struct pids_stack **A, const struct pids_stack **B, struct sort_parms *P)
|
2015-08-19 10:30:00 +05:30
|
|
|
|
2016-06-09 10:30:00 +05:30
|
|
|
#define NUM_srt(T) srtDECL(T) { \
|
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; \
|
2015-09-13 10:30:00 +05:30
|
|
|
return P->order * (a->result. T - b->result. T); }
|
2015-08-19 10:30:00 +05:30
|
|
|
|
2016-06-09 10:30:00 +05:30
|
|
|
#define REG_srt(T) srtDECL(T) { \
|
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; \
|
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; \
|
2015-08-19 10:30:00 +05:30
|
|
|
return 0; }
|
|
|
|
|
|
|
|
|
|
NUM_srt(s_ch)
|
|
|
|
|
NUM_srt(s_int)
|
|
|
|
|
|
|
|
|
|
REG_srt(u_int)
|
|
|
|
|
REG_srt(ul_int)
|
|
|
|
|
REG_srt(ull_int)
|
|
|
|
|
|
2016-06-09 10:30:00 +05:30
|
|
|
srtDECL(str) {
|
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;
|
2015-09-13 10:30:00 +05:30
|
|
|
return P->order * strcoll(a->result.str, b->result.str);
|
2015-08-19 10:30:00 +05:30
|
|
|
}
|
|
|
|
|
|
2016-06-09 10:30:00 +05:30
|
|
|
srtDECL(strv) {
|
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;
|
2015-09-13 10:30:00 +05:30
|
|
|
return P->order * strcoll((*a->result.strv), (*b->result.strv));
|
2015-08-26 10:30:00 +05:30
|
|
|
}
|
|
|
|
|
|
2016-06-09 10:30:00 +05:30
|
|
|
srtDECL(strvers) {
|
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;
|
2015-09-13 10:30:00 +05:30
|
|
|
return P->order * strverscmp(a->result.str, b->result.str);
|
2015-09-07 10:30:00 +05:30
|
|
|
}
|
|
|
|
|
|
2016-06-09 10:30:00 +05:30
|
|
|
srtDECL(noop) {
|
|
|
|
|
(void)A; (void)B; (void)P;
|
2015-08-19 10:30:00 +05:30
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
2016-06-09 10:30:00 +05:30
|
|
|
#undef srtDECL
|
2015-08-19 10:30:00 +05:30
|
|
|
#undef NUM_srt
|
|
|
|
|
#undef REG_srt
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
// ___ Controlling Table ||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|
|
|
|
2016-08-21 10:30:00 +05:30
|
|
|
#define f_either PROC_SPARE_1 // either status or stat (favor stat)
|
2018-06-22 10:30:00 +05:30
|
|
|
#define f_exe PROC_FILL_EXE
|
2015-08-19 10:30:00 +05:30
|
|
|
#define f_grp PROC_FILLGRP
|
2018-02-11 12:41:11 +05:30
|
|
|
#define f_login PROC_FILL_LUID
|
2015-08-19 10:30:00 +05:30
|
|
|
#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
|
2015-08-28 10:30:00 +05:30
|
|
|
// these next three will yield true verctorized strings
|
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
|
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
|
|
|
|
|
|
2016-07-21 10:30:00 +05:30
|
|
|
typedef void (*SET_t)(struct pids_info *, struct pids_result *, proc_t *);
|
2015-08-26 10:30:00 +05:30
|
|
|
typedef void (*FRE_t)(struct pids_result *);
|
2015-08-19 10:30:00 +05:30
|
|
|
typedef int (*QSR_t)(const void *, const void *, void *);
|
|
|
|
|
|
2020-08-10 10:30:00 +05:30
|
|
|
#ifdef ITEMTABLE_DEBUG
|
|
|
|
|
#define RS(e) (SET_t)setNAME(e), PIDS_ ## e, STRINGIFY(PIDS_ ## e)
|
|
|
|
|
#else
|
2015-08-19 10:30:00 +05:30
|
|
|
#define RS(e) (SET_t)setNAME(e)
|
2020-08-10 10:30:00 +05:30
|
|
|
#endif
|
2016-07-07 10:30:00 +05:30
|
|
|
#define FF(t) (FRE_t)freNAME(t)
|
2015-08-19 10:30:00 +05:30
|
|
|
#define QS(t) (QSR_t)srtNAME(t)
|
2016-08-05 10:30:00 +05:30
|
|
|
#define TS(t) STRINGIFY(t)
|
|
|
|
|
#define TS_noop ""
|
2015-08-19 10:30:00 +05:30
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Need it be said?
|
|
|
|
|
* This table must be kept in the exact same order as
|
|
|
|
|
* those 'enum pids_item' guys ! */
|
|
|
|
|
static struct {
|
2015-08-26 10:30:00 +05:30
|
|
|
SET_t setsfunc; // the actual result setting routine
|
2020-08-10 10:30:00 +05:30
|
|
|
#ifdef ITEMTABLE_DEBUG
|
|
|
|
|
int enumnumb; // enumerator (must match position!)
|
|
|
|
|
char *enum2str; // enumerator name as a char* string
|
|
|
|
|
#endif
|
2015-08-19 10:30:00 +05:30
|
|
|
unsigned oldflags; // PROC_FILLxxxx flags for this item
|
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
|
2015-08-20 10:30:00 +05:30
|
|
|
int needhist; // a result requires history support
|
2016-08-05 10:30:00 +05:30
|
|
|
char *type2str; // the result type as a string value
|
2015-08-19 10:30:00 +05:30
|
|
|
} Item_table[] = {
|
2016-08-05 10:30:00 +05:30
|
|
|
/* setsfunc oldflags freefunc sortfunc needhist type2str
|
|
|
|
|
--------------------- ---------- --------- ------------- -------- ----------- */
|
|
|
|
|
{ RS(noop), 0, NULL, QS(noop), 0, TS_noop }, // user only, never altered
|
|
|
|
|
{ RS(extra), 0, NULL, QS(ull_int), 0, TS_noop }, // user only, reset to zero
|
|
|
|
|
|
|
|
|
|
{ RS(ADDR_END_CODE), f_stat, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(ADDR_KSTK_EIP), f_stat, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(ADDR_KSTK_ESP), f_stat, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(ADDR_START_CODE), f_stat, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(ADDR_START_STACK), f_stat, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(CGNAME), x_cgroup, FF(str), QS(str), 0, TS(str) },
|
|
|
|
|
{ RS(CGROUP), x_cgroup, FF(str), QS(str), 0, TS(str) },
|
|
|
|
|
{ RS(CGROUP_V), v_cgroup, FF(strv), QS(strv), 0, TS(strv) },
|
|
|
|
|
{ RS(CMD), f_either, FF(str), QS(str), 0, TS(str) },
|
|
|
|
|
{ RS(CMDLINE), x_cmdline, FF(str), QS(str), 0, TS(str) },
|
|
|
|
|
{ RS(CMDLINE_V), v_arg, FF(strv), QS(strv), 0, TS(strv) },
|
|
|
|
|
{ RS(ENVIRON), x_environ, FF(str), QS(str), 0, TS(str) },
|
|
|
|
|
{ RS(ENVIRON_V), v_env, FF(strv), QS(strv), 0, TS(strv) },
|
2018-06-22 10:30:00 +05:30
|
|
|
{ RS(EXE), f_exe, FF(str), QS(str), 0, TS(str) },
|
2016-08-05 10:30:00 +05:30
|
|
|
{ RS(EXIT_SIGNAL), f_stat, NULL, QS(s_int), 0, TS(s_int) },
|
|
|
|
|
{ RS(FLAGS), f_stat, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(FLT_MAJ), f_stat, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(FLT_MAJ_C), f_stat, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(FLT_MAJ_DELTA), f_stat, NULL, QS(s_int), +1, TS(s_int) },
|
|
|
|
|
{ RS(FLT_MIN), f_stat, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(FLT_MIN_C), f_stat, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(FLT_MIN_DELTA), f_stat, NULL, QS(s_int), +1, TS(s_int) },
|
|
|
|
|
{ RS(ID_EGID), 0, NULL, QS(u_int), 0, TS(u_int) }, // oldflags: free w/ simple_read
|
|
|
|
|
{ RS(ID_EGROUP), f_grp, NULL, QS(str), 0, TS(str) },
|
|
|
|
|
{ RS(ID_EUID), 0, NULL, QS(u_int), 0, TS(u_int) }, // oldflags: free w/ simple_read
|
|
|
|
|
{ RS(ID_EUSER), f_usr, NULL, QS(str), 0, TS(str) }, // freefunc NULL w/ cached string
|
|
|
|
|
{ RS(ID_FGID), f_status, NULL, QS(u_int), 0, TS(u_int) },
|
|
|
|
|
{ RS(ID_FGROUP), x_ogroup, NULL, QS(str), 0, TS(str) },
|
|
|
|
|
{ RS(ID_FUID), f_status, NULL, QS(u_int), 0, TS(u_int) },
|
|
|
|
|
{ RS(ID_FUSER), x_ouser, NULL, QS(str), 0, TS(str) }, // freefunc NULL w/ cached string
|
2018-02-11 12:41:11 +05:30
|
|
|
{ RS(ID_LOGIN), f_login, NULL, QS(s_int), 0, TS(s_int) },
|
2016-08-05 10:30:00 +05:30
|
|
|
{ RS(ID_PGRP), f_stat, NULL, QS(s_int), 0, TS(s_int) },
|
|
|
|
|
{ RS(ID_PID), 0, NULL, QS(s_int), 0, TS(s_int) }, // oldflags: free w/ simple_nextpid
|
|
|
|
|
{ RS(ID_PPID), f_either, NULL, QS(s_int), 0, TS(s_int) },
|
|
|
|
|
{ RS(ID_RGID), f_status, NULL, QS(u_int), 0, TS(u_int) },
|
|
|
|
|
{ RS(ID_RGROUP), x_ogroup, NULL, QS(str), 0, TS(str) },
|
|
|
|
|
{ RS(ID_RUID), f_status, NULL, QS(u_int), 0, TS(u_int) },
|
|
|
|
|
{ RS(ID_RUSER), x_ouser, NULL, QS(str), 0, TS(str) }, // freefunc NULL w/ cached string
|
|
|
|
|
{ RS(ID_SESSION), f_stat, NULL, QS(s_int), 0, TS(s_int) },
|
|
|
|
|
{ RS(ID_SGID), f_status, NULL, QS(u_int), 0, TS(u_int) },
|
|
|
|
|
{ RS(ID_SGROUP), x_ogroup, NULL, QS(str), 0, TS(str) },
|
|
|
|
|
{ RS(ID_SUID), f_status, NULL, QS(u_int), 0, TS(u_int) },
|
|
|
|
|
{ RS(ID_SUSER), x_ouser, NULL, QS(str), 0, TS(str) }, // freefunc NULL w/ cached string
|
|
|
|
|
{ RS(ID_TGID), 0, NULL, QS(s_int), 0, TS(s_int) }, // oldflags: free w/ simple_nextpid
|
2018-12-01 11:30:00 +05:30
|
|
|
{ RS(ID_TID), 0, NULL, QS(s_int), 0, TS(s_int) }, // oldflags: free w/ simple_nexttid
|
2016-08-05 10:30:00 +05:30
|
|
|
{ RS(ID_TPGID), f_stat, NULL, QS(s_int), 0, TS(s_int) },
|
|
|
|
|
{ RS(LXCNAME), f_lxc, NULL, QS(str), 0, TS(str) }, // freefunc NULL w/ cached string
|
|
|
|
|
{ RS(MEM_CODE), f_statm, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(MEM_CODE_PGS), f_statm, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(MEM_DATA), f_statm, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(MEM_DATA_PGS), f_statm, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(MEM_RES), f_statm, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(MEM_RES_PGS), f_statm, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(MEM_SHR), f_statm, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(MEM_SHR_PGS), f_statm, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(MEM_VIRT), f_statm, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(MEM_VIRT_PGS), f_statm, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(NICE), f_stat, NULL, QS(s_int), 0, TS(s_int) },
|
|
|
|
|
{ RS(NLWP), f_either, NULL, QS(s_int), 0, TS(s_int) },
|
|
|
|
|
{ RS(NS_IPC), f_ns, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(NS_MNT), f_ns, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(NS_NET), f_ns, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(NS_PID), f_ns, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(NS_USER), f_ns, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(NS_UTS), f_ns, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(OOM_ADJ), f_oom, NULL, QS(s_int), 0, TS(s_int) },
|
|
|
|
|
{ RS(OOM_SCORE), f_oom, NULL, QS(s_int), 0, TS(s_int) },
|
|
|
|
|
{ RS(PRIORITY), f_stat, NULL, QS(s_int), 0, TS(s_int) },
|
library: refactored some header file items and origins
This commit is intended as a refinement of the patches
mentioned below, where origins/sources of newlib items
were added to the header files for user documentation.
However, if those additions are to be truly effective,
along with kernel documentation (where available), the
following prerequisites must also have been satisfied:
. our identifiers closely align with linux field names
. our derived items are documented or self-documenting
Satisfying those prerequisites prompted this patch and
for these changes, kernel sources were emphasized over
available documentation (shame on me, it should always
have been so). And, while some 'new' fields were found
to be conditional, they were included unconditionally.
These changes appear more extensive than they actually
need be since I have attempted to enforce some spacing
conventions. So, I've summarize the significant things
in the sections that follow. For a proper perspective,
use: 'git diff --ignore-space-change' (good as alias).
___________________________________________ <PIDS> api
This api is unique in that there exists many different
file/directory origins subordinate to /proc/<pid>. And
our item identifiers are sometimes coerced so as to be
able to group related or similar enumerators together.
So, users needed more help relating our identifiers to
an actual documented field. Thus, we will now also add
the field names as with 'stat: delayacct_blkio_ticks'.
Each item ending with a '_C' now consistently includes
both the parent's count/time plus waited for children.
That 'RTPRIO' guy was renamed/relocated as PRIORITY_RT
since its original name is an implementation artifact.
___________________________________________ <STAT> api
The only api change was to correct a typo ('dervied').
_________________________________________ <VMSTAT> api
Even ignoring white space, this interface received the
largest number of changes. Mostly, this was because of
deficiencies in the proc(5) documentation. Recall that
this documentation already sorely lacks any substance.
Usually, just kernel releases are noted, not contents.
When compared to kernel source, that proc(5) contained
many non-existent fields and also omitted many others.
________________________________________ <MEMINFO> api
Sadly, with this api many of the changes were simply a
correction of some earlier 'human error' where several
fields where hashed then tracked but never represented
with an item enumerator in this meminfo.h header file.
_______________________________________ <SLABINFO> api
The 'SLABS' (summary) & 'SLABNODE' items were reversed
since the former are derived from the separate caches.
More significantly, those 'SLABNODE' guys were renamed
to 'SLAB' since they concern individual caches and the
concept of 'nodes' is really an implementation detail.
Also, several enumerators were changed to more closely
agree with official slabinfo(5) documentation referred
to in what we're treating as a base document: proc(5).
Lastly, while those 'SLABS' items are solely a product
of our library and not represented in slabinfo(5), the
names attempt to parallel those found as 'SLAB' items.
______________________________________ <DISKSTATS> api
One enumeration identifier was changed so as to better
reflect its relationship to that actual documentation:
'Documentation/iostats.txt', as referenced in proc(5).
Reference(s):
. 12/2018, item origins added (and commit msg history)
commit 96d59cbf46b3ff687bd29fad4708074a0e1cea14
. 01/2019, <stat> origins tweaked
commit 201e816b26ddaccc923ec40977c92037cdd0c34e
Signed-off-by: Jim Warner <james.warner@comcast.net>
2019-03-12 11:30:00 +05:30
|
|
|
{ RS(PRIORITY_RT), f_stat, NULL, QS(s_int), 0, TS(s_int) },
|
2016-08-05 10:30:00 +05:30
|
|
|
{ RS(PROCESSOR), f_stat, NULL, QS(u_int), 0, TS(u_int) },
|
2017-05-12 10:33:00 +05:30
|
|
|
{ RS(PROCESSOR_NODE), f_stat, NULL, QS(s_int), 0, TS(s_int) },
|
2016-08-05 10:30:00 +05:30
|
|
|
{ RS(RSS), f_stat, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(RSS_RLIM), f_stat, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(SCHED_CLASS), f_stat, NULL, QS(s_int), 0, TS(s_int) },
|
|
|
|
|
{ RS(SD_MACH), f_systemd, FF(str), QS(str), 0, TS(str) },
|
|
|
|
|
{ RS(SD_OUID), f_systemd, FF(str), QS(str), 0, TS(str) },
|
|
|
|
|
{ RS(SD_SEAT), f_systemd, FF(str), QS(str), 0, TS(str) },
|
|
|
|
|
{ RS(SD_SESS), f_systemd, FF(str), QS(str), 0, TS(str) },
|
|
|
|
|
{ RS(SD_SLICE), f_systemd, FF(str), QS(str), 0, TS(str) },
|
|
|
|
|
{ RS(SD_UNIT), f_systemd, FF(str), QS(str), 0, TS(str) },
|
|
|
|
|
{ RS(SD_UUNIT), f_systemd, FF(str), QS(str), 0, TS(str) },
|
|
|
|
|
{ RS(SIGBLOCKED), f_status, FF(str), QS(str), 0, TS(str) },
|
|
|
|
|
{ RS(SIGCATCH), f_status, FF(str), QS(str), 0, TS(str) },
|
|
|
|
|
{ RS(SIGIGNORE), f_status, FF(str), QS(str), 0, TS(str) },
|
|
|
|
|
{ RS(SIGNALS), f_status, FF(str), QS(str), 0, TS(str) },
|
|
|
|
|
{ RS(SIGPENDING), f_status, FF(str), QS(str), 0, TS(str) },
|
|
|
|
|
{ RS(STATE), f_either, NULL, QS(s_ch), 0, TS(s_ch) },
|
|
|
|
|
{ RS(SUPGIDS), f_status, FF(str), QS(str), 0, TS(str) },
|
|
|
|
|
{ RS(SUPGROUPS), x_supgrp, FF(str), QS(str), 0, TS(str) },
|
|
|
|
|
{ RS(TICS_ALL), f_stat, NULL, QS(ull_int), 0, TS(ull_int) },
|
|
|
|
|
{ RS(TICS_ALL_C), f_stat, NULL, QS(ull_int), 0, TS(ull_int) },
|
|
|
|
|
{ RS(TICS_ALL_DELTA), f_stat, NULL, QS(s_int), +1, TS(s_int) },
|
|
|
|
|
{ RS(TICS_BLKIO), f_stat, NULL, QS(ull_int), 0, TS(ull_int) },
|
|
|
|
|
{ RS(TICS_GUEST), f_stat, NULL, QS(ull_int), 0, TS(ull_int) },
|
|
|
|
|
{ RS(TICS_GUEST_C), f_stat, NULL, QS(ull_int), 0, TS(ull_int) },
|
|
|
|
|
{ RS(TICS_SYSTEM), f_stat, NULL, QS(ull_int), 0, TS(ull_int) },
|
|
|
|
|
{ RS(TICS_SYSTEM_C), f_stat, NULL, QS(ull_int), 0, TS(ull_int) },
|
|
|
|
|
{ RS(TICS_USER), f_stat, NULL, QS(ull_int), 0, TS(ull_int) },
|
|
|
|
|
{ RS(TICS_USER_C), f_stat, NULL, QS(ull_int), 0, TS(ull_int) },
|
|
|
|
|
{ RS(TIME_ALL), f_stat, NULL, QS(ull_int), 0, TS(ull_int) },
|
|
|
|
|
{ RS(TIME_ELAPSED), f_stat, NULL, QS(ull_int), 0, TS(ull_int) },
|
|
|
|
|
{ RS(TIME_START), f_stat, NULL, QS(ull_int), 0, TS(ull_int) },
|
|
|
|
|
{ RS(TTY), f_stat, NULL, QS(s_int), 0, TS(s_int) },
|
|
|
|
|
{ RS(TTY_NAME), f_stat, FF(str), QS(strvers), 0, TS(str) },
|
|
|
|
|
{ RS(TTY_NUMBER), f_stat, FF(str), QS(strvers), 0, TS(str) },
|
|
|
|
|
{ RS(VM_DATA), f_status, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(VM_EXE), f_status, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(VM_LIB), f_status, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(VM_RSS), f_status, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(VM_RSS_ANON), f_status, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(VM_RSS_FILE), f_status, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(VM_RSS_LOCKED), f_status, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(VM_RSS_SHARED), f_status, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(VM_SIZE), f_status, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(VM_STACK), f_status, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(VM_SWAP), f_status, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(VM_USED), f_status, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(VSIZE_PGS), f_stat, NULL, QS(ul_int), 0, TS(ul_int) },
|
|
|
|
|
{ RS(WCHAN_NAME), 0, FF(str), QS(str), 0, TS(str) }, // oldflags: tid already free
|
2015-08-19 10:30:00 +05:30
|
|
|
};
|
|
|
|
|
|
2016-07-07 10:30:00 +05:30
|
|
|
/* please note,
|
|
|
|
|
* this enum MUST be 1 greater than the highest value of any enum */
|
2017-09-21 10:30:00 +05:30
|
|
|
enum pids_item PIDS_logical_end = PIDS_WCHAN_NAME + 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
|
|
|
|
2015-08-19 10:30:00 +05:30
|
|
|
#undef setNAME
|
2015-08-26 10:30:00 +05:30
|
|
|
#undef freNAME
|
|
|
|
|
#undef srtNAME
|
2016-06-06 10:30:00 +05:30
|
|
|
#undef RS
|
|
|
|
|
#undef FF
|
|
|
|
|
#undef QS
|
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
|
2015-08-26 10:30:00 +05:30
|
|
|
#undef v_arg
|
|
|
|
|
#undef v_cgroup
|
|
|
|
|
#undef v_env
|
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
|
2016-07-31 10:30:00 +05:30
|
|
|
unsigned long maj, min; // last frame's maj/min_flt counts
|
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)
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
|
2016-08-21 10:30:00 +05:30
|
|
|
static void pids_config_history (
|
2016-07-21 10:30:00 +05:30
|
|
|
struct pids_info *info)
|
2015-08-19 10:30:00 +05:30
|
|
|
{
|
|
|
|
|
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);
|
2016-08-21 10:30:00 +05:30
|
|
|
} // end: pids_config_history
|
2015-08-19 10:30:00 +05:30
|
|
|
|
|
|
|
|
|
2016-08-21 10:30:00 +05:30
|
|
|
static inline HST_t *pids_histget (
|
2016-07-21 10:30:00 +05:30
|
|
|
struct pids_info *info,
|
2015-08-19 10:30:00 +05:30
|
|
|
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]);
|
2016-07-07 10:30:00 +05:30
|
|
|
V = Hr(PHist_sav[V].lnk);
|
|
|
|
|
}
|
2015-08-19 10:30:00 +05:30
|
|
|
return NULL;
|
2016-08-21 10:30:00 +05:30
|
|
|
} // end: pids_histget
|
2015-08-19 10:30:00 +05:30
|
|
|
|
|
|
|
|
|
2016-08-21 10:30:00 +05:30
|
|
|
static inline void pids_histput (
|
2016-07-21 10:30:00 +05:30
|
|
|
struct pids_info *info,
|
2015-08-19 10:30:00 +05:30
|
|
|
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);
|
2016-08-21 10:30:00 +05:30
|
|
|
} // end: pids_histput
|
2015-08-19 10:30:00 +05:30
|
|
|
|
|
|
|
|
#undef _HASH_PID_
|
|
|
|
|
|
|
|
|
|
|
2016-08-21 10:30:00 +05:30
|
|
|
static inline int pids_make_hist (
|
2016-07-21 10:30:00 +05:30
|
|
|
struct pids_info *info,
|
2015-08-19 10:30:00 +05:30
|
|
|
proc_t *p)
|
|
|
|
|
{
|
2015-08-20 10:30:00 +05:30
|
|
|
#define nSLOT info->hist->num_tasks
|
2015-08-19 10:30:00 +05:30
|
|
|
TIC_t tics;
|
|
|
|
|
HST_t *h;
|
|
|
|
|
|
2015-08-20 10:30:00 +05:30
|
|
|
if (nSLOT + 1 >= Hr(HHist_siz)) {
|
2016-09-09 15:14:44 +05:30
|
|
|
Hr(HHist_siz) += NEWOLD_INCR;
|
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));
|
2015-08-28 10:30:00 +05:30
|
|
|
if (!Hr(PHist_sav) || !Hr(PHist_new))
|
2017-11-18 10:30:00 +05:30
|
|
|
return 0;
|
2015-08-19 10:30:00 +05:30
|
|
|
}
|
2015-08-20 10:30:00 +05:30
|
|
|
Hr(PHist_new[nSLOT].pid) = p->tid;
|
|
|
|
|
Hr(PHist_new[nSLOT].maj) = p->maj_flt;
|
|
|
|
|
Hr(PHist_new[nSLOT].min) = p->min_flt;
|
2016-07-31 10:30:00 +05:30
|
|
|
Hr(PHist_new[nSLOT].tics) = tics = (p->utime + p->stime);
|
2015-08-19 10:30:00 +05:30
|
|
|
|
2016-08-21 10:30:00 +05:30
|
|
|
pids_histput(info, nSLOT);
|
2015-08-19 10:30:00 +05:30
|
|
|
|
2016-08-21 10:30:00 +05:30
|
|
|
if ((h = pids_histget(info, p->tid))) {
|
2016-07-31 10:30:00 +05:30
|
|
|
p->pcpu = tics - h->tics;
|
2015-08-19 10:30:00 +05:30
|
|
|
p->maj_delta = p->maj_flt - h->maj;
|
|
|
|
|
p->min_delta = p->min_flt - h->min;
|
|
|
|
|
}
|
|
|
|
|
|
2015-08-20 10:30:00 +05:30
|
|
|
nSLOT++;
|
2017-11-18 10:30:00 +05:30
|
|
|
return 1;
|
2015-08-20 10:30:00 +05:30
|
|
|
#undef nSLOT
|
2016-08-21 10:30:00 +05:30
|
|
|
} // end: pids_make_hist
|
2015-08-19 10:30:00 +05:30
|
|
|
|
|
|
|
|
|
2016-08-21 10:30:00 +05:30
|
|
|
static inline void pids_toggle_history (
|
2016-07-21 10:30:00 +05:30
|
|
|
struct pids_info *info)
|
2015-08-19 10:30:00 +05:30
|
|
|
{
|
|
|
|
|
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;
|
2016-08-21 10:30:00 +05:30
|
|
|
} // end: pids_toggle_history
|
2015-08-19 10:30:00 +05:30
|
|
|
|
|
|
|
|
|
|
|
|
|
#ifdef UNREF_RPTHASH
|
2016-08-21 10:30:00 +05:30
|
|
|
static void pids_unref_rpthash (
|
2016-07-21 10:30:00 +05:30
|
|
|
struct pids_info *info)
|
2015-08-19 10:30:00 +05:30
|
|
|
{
|
|
|
|
|
int i, j, pop, total_occupied, maxdepth, maxdepth_sav, numdepth
|
2015-08-20 10:30:00 +05:30
|
|
|
, cross_foot, sz = HHASH_SIZE * (int)sizeof(int)
|
|
|
|
|
, hsz = (int)sizeof(HST_t) * Hr(HHist_siz);
|
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,
|
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."
|
2015-08-19 10:30:00 +05:30
|
|
|
"\n"
|
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:"
|
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"
|
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)
|
2015-08-19 10:30:00 +05:30
|
|
|
, info->hist->num_tasks
|
|
|
|
|
, total_occupied, (total_occupied * 100) / HHASH_SIZE
|
2015-08-20 10:30:00 +05:30
|
|
|
, maxdepth);
|
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"
|
2015-08-20 10:30:00 +05:30
|
|
|
, numdepth, (numdepth * 100) / total_occupied, maxdepth);
|
2015-08-19 10:30:00 +05:30
|
|
|
pop -= numdepth;
|
|
|
|
|
cross_foot += numdepth;
|
|
|
|
|
if (0 == pop && cross_foot == total_occupied) break;
|
|
|
|
|
}
|
|
|
|
|
if (pop) {
|
2015-08-20 10:30:00 +05:30
|
|
|
fprintf(stderr, "\t %5d (%3d%%) unchained entries (at depth 0)\n"
|
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);
|
|
|
|
|
|
2015-08-20 10:30:00 +05:30
|
|
|
if (maxdepth_sav > 1) {
|
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");
|
|
|
|
|
}
|
|
|
|
|
}
|
2016-08-21 10:30:00 +05:30
|
|
|
} // end: pids_unref_rpthash
|
2015-08-19 10:30:00 +05:30
|
|
|
#endif // UNREF_RPTHASH
|
|
|
|
|
|
2015-08-28 10:30:00 +05:30
|
|
|
#undef Hr
|
2015-08-19 10:30:00 +05:30
|
|
|
#undef HHASH_SIZE
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
// ___ Standard Private Functions |||||||||||||||||||||||||||||||||||||||||||||
|
|
|
|
|
|
2017-11-18 10:30:00 +05:30
|
|
|
static inline int pids_assign_results (
|
2016-07-21 10:30:00 +05:30
|
|
|
struct pids_info *info,
|
2015-08-19 10:30:00 +05:30
|
|
|
struct pids_stack *stack,
|
|
|
|
|
proc_t *p)
|
|
|
|
|
{
|
|
|
|
|
struct pids_result *this = stack->head;
|
|
|
|
|
|
2017-11-18 10:30:00 +05:30
|
|
|
info->seterr = 0;
|
2015-08-19 10:30:00 +05:30
|
|
|
for (;;) {
|
|
|
|
|
enum pids_item item = this->item;
|
library: removed all the 'PROCPS_' enumerator prefixes
Many of our item enumerator identifiers are very long,
especially in that <VMSTAT> module. Additionally, they
all contain the exact same universal 'PROCPS_' prefix.
The origins for this are likely found in the desire to
avoid name clashes with other potential include files.
But with procps-ng newlib, we've probably gone way too
far. Did 'PROCPS_PIDS_TICS_SYSTEM' actually offer more
protection against clash than 'PIDS_TICS_SYSTEM' does?
I don't think so. Besides, no matter how big that name
becomes, one can never guarantee they'll never be some
clash. And, conversely, extremely short names will not
always create conflict. Of course, in either case when
some clash occurs, one can always #undef that problem.
Thus, this commit will eliminate that 'PROCPS_' prefix
making all of those enum identifiers a little shorter.
And, we'll still be well above some ridiculously short
(criminally short) names found in some common headers:
- - - - - - - - - - <term.h>
- 'tab', 'TTY', etc
- - - - - - - - - - - - - - - - <search.h>
- 'ENTER', ENTRY', 'FIND', etc
------------------------------------------------------
Finally, with this as a last of the wholesale changes,
we will have established the naming conventions below:
. only functions will begin with that 'procps_' prefix
. exposed structures begin with the module/header name
. item enumerators begin like structs, but capitalized
. other enumerators work exactly like item enumerators
. macros and constants begin just like the enumerators
------------------------------------------------------
Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-21 10:30:00 +05:30
|
|
|
if (item >= PIDS_logical_end)
|
2015-08-19 10:30:00 +05:30
|
|
|
break;
|
2015-08-26 10:30:00 +05:30
|
|
|
Item_table[item].setsfunc(info, this, p);
|
2015-08-19 10:30:00 +05:30
|
|
|
++this;
|
|
|
|
|
}
|
2017-11-18 10:30:00 +05:30
|
|
|
return !info->seterr;
|
2016-08-21 10:30:00 +05:30
|
|
|
} // end: pids_assign_results
|
2015-08-19 10:30:00 +05:30
|
|
|
|
|
|
|
|
|
2016-08-21 10:30:00 +05:30
|
|
|
static inline void pids_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)
|
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;
|
library: removed all the 'PROCPS_' enumerator prefixes
Many of our item enumerator identifiers are very long,
especially in that <VMSTAT> module. Additionally, they
all contain the exact same universal 'PROCPS_' prefix.
The origins for this are likely found in the desire to
avoid name clashes with other potential include files.
But with procps-ng newlib, we've probably gone way too
far. Did 'PROCPS_PIDS_TICS_SYSTEM' actually offer more
protection against clash than 'PIDS_TICS_SYSTEM' does?
I don't think so. Besides, no matter how big that name
becomes, one can never guarantee they'll never be some
clash. And, conversely, extremely short names will not
always create conflict. Of course, in either case when
some clash occurs, one can always #undef that problem.
Thus, this commit will eliminate that 'PROCPS_' prefix
making all of those enum identifiers a little shorter.
And, we'll still be well above some ridiculously short
(criminally short) names found in some common headers:
- - - - - - - - - - <term.h>
- 'tab', 'TTY', etc
- - - - - - - - - - - - - - - - <search.h>
- 'ENTER', ENTRY', 'FIND', etc
------------------------------------------------------
Finally, with this as a last of the wholesale changes,
we will have established the naming conventions below:
. only functions will begin with that 'procps_' prefix
. exposed structures begin with the module/header name
. item enumerators begin like structs, but capitalized
. other enumerators work exactly like item enumerators
. macros and constants begin just like the enumerators
------------------------------------------------------
Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-21 10:30:00 +05:30
|
|
|
if (item >= PIDS_logical_end)
|
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);
|
library: minimize the use of 'cleanup_stacks' routines
Some parts of our newlib implementation are the result
of functions which have been propagated from module to
module. In particular, those 'cleanup_stacks' routines
are all similar & likely originated in the <pids> api.
In that interface there was a need to free dynamically
acquired memory before the result structure was reused
to satisfy subsequent 'get', 'select' or 'reap' calls.
This, in turn, led to a concept of 'dirty' stacks with
the need to call one of two 'cleanup_stack' functions.
None of the remaining interfaces deal with such memory
yet they each had their own 'cleanup_stack' functions.
Those functions were responsible for resetting each of
the result unions to zero, excluding any 'noop' items.
The bottom line is that for all interfaces, repetitive
calls would require iterating through the stack(s) two
separate times: once to 'cleanup' another to 'assign'.
With this commit we will reduce iterations to just the
'assign' routine. A reset to zero will be accomplished
in the 'extra' item set routine (which is the only one
actually requiring any reset). All other items will be
reinitialized automatically by a new current set value
or upon reallocation when an items compliment changes.
In the <pids> interface, any freeing of dynamic memory
could have been accomplished by adding that 'freefunc'
check to the 'assign' function. However, that requires
an Item_table test with every item. Instead, we'll now
satisfy such needs as the very first step in those set
functions responsible for dynamically acquired memory.
[ the <pids> api retains 2 'cleanup_stack' functions ]
[ to accommodate stack(s) 'reset' & to serve 'unref' ]
Lastly, all the 'itemize_stack' functions were tweaked
by eliminating an unnecessary initialization of result
unions. That objective was already accomplished by the
calloc() in a 'stacks_alloc' function or the remaining
'cleanup_stack' routine found in the <pids> interface.
Signed-off-by: Jim Warner <james.warner@comcast.net>
2019-05-22 10:30:00 +05:30
|
|
|
this->result.ull_int = 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
|
|
|
++this;
|
2015-08-19 10:30:00 +05:30
|
|
|
}
|
2016-08-21 10:30:00 +05:30
|
|
|
} // end: pids_cleanup_stack
|
2015-08-19 10:30:00 +05:30
|
|
|
|
|
|
|
|
|
2016-08-21 10:30:00 +05:30
|
|
|
static inline void pids_cleanup_stacks_all (
|
2016-07-21 10:30:00 +05:30
|
|
|
struct pids_info *info)
|
2015-08-19 10:30:00 +05:30
|
|
|
{
|
|
|
|
|
struct stacks_extent *ext = info->extents;
|
|
|
|
|
int i;
|
|
|
|
|
|
|
|
|
|
while (ext) {
|
|
|
|
|
for (i = 0; ext->stacks[i]; i++)
|
2016-08-21 10:30:00 +05:30
|
|
|
pids_cleanup_stack(ext->stacks[i]->head);
|
2015-08-19 10:30:00 +05:30
|
|
|
ext = ext->next;
|
|
|
|
|
};
|
2016-08-21 10:30:00 +05:30
|
|
|
} // end: pids_cleanup_stacks_all
|
2015-08-19 10:30:00 +05:30
|
|
|
|
|
|
|
|
|
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. */
|
2016-08-21 10:30:00 +05:30
|
|
|
static struct stacks_extent *pids_extent_cut (
|
2016-07-21 10:30:00 +05:30
|
|
|
struct pids_info *info,
|
2015-08-19 10:30:00 +05:30
|
|
|
struct stacks_extent *ext)
|
|
|
|
|
{
|
|
|
|
|
struct stacks_extent *p = info->extents;
|
|
|
|
|
|
|
|
|
|
if (ext) {
|
|
|
|
|
if (ext == p) {
|
|
|
|
|
info->extents = p->next;
|
2015-09-10 10:30:00 +05:30
|
|
|
return ext;
|
2015-08-19 10:30:00 +05:30
|
|
|
}
|
|
|
|
|
do {
|
|
|
|
|
if (ext == p->next) {
|
|
|
|
|
p->next = p->next->next;
|
2015-09-10 10:30:00 +05:30
|
|
|
return ext;
|
2015-08-19 10:30:00 +05:30
|
|
|
}
|
|
|
|
|
p = p->next;
|
|
|
|
|
} while (p);
|
|
|
|
|
}
|
2015-09-10 10:30:00 +05:30
|
|
|
return NULL;
|
2016-08-21 10:30:00 +05:30
|
|
|
} // end: pids_extent_cut
|
2015-09-10 10:30:00 +05:30
|
|
|
|
|
|
|
|
|
2016-08-21 10:30:00 +05:30
|
|
|
static inline struct pids_result *pids_itemize_stack (
|
2016-04-25 10:30:00 +05:30
|
|
|
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;
|
|
|
|
|
}
|
|
|
|
|
return p_sav;
|
2016-08-21 10:30:00 +05:30
|
|
|
} // end: pids_itemize_stack
|
2016-04-25 10:30:00 +05:30
|
|
|
|
|
|
|
|
|
2016-08-21 10:30:00 +05:30
|
|
|
static void pids_itemize_stacks_all (
|
2016-07-21 10:30:00 +05:30
|
|
|
struct pids_info *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
|
|
|
{
|
|
|
|
|
struct stacks_extent *ext = info->extents;
|
|
|
|
|
|
|
|
|
|
while (ext) {
|
|
|
|
|
int i;
|
|
|
|
|
for (i = 0; ext->stacks[i]; i++)
|
2016-08-21 10:30:00 +05:30
|
|
|
pids_itemize_stack(ext->stacks[i]->head, info->curitems, info->items);
|
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 = ext->next;
|
|
|
|
|
};
|
2016-08-21 10:30:00 +05:30
|
|
|
} // end: pids_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
|
|
|
|
|
|
|
|
|
2016-08-21 10:30:00 +05:30
|
|
|
static inline int pids_items_check_failed (
|
2016-06-15 10:30:00 +05:30
|
|
|
enum pids_item *items,
|
|
|
|
|
int numitems)
|
2015-08-19 10:30:00 +05:30
|
|
|
{
|
|
|
|
|
int i;
|
|
|
|
|
|
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: removed all the 'PROCPS_' enumerator prefixes
Many of our item enumerator identifiers are very long,
especially in that <VMSTAT> module. Additionally, they
all contain the exact same universal 'PROCPS_' prefix.
The origins for this are likely found in the desire to
avoid name clashes with other potential include files.
But with procps-ng newlib, we've probably gone way too
far. Did 'PROCPS_PIDS_TICS_SYSTEM' actually offer more
protection against clash than 'PIDS_TICS_SYSTEM' does?
I don't think so. Besides, no matter how big that name
becomes, one can never guarantee they'll never be some
clash. And, conversely, extremely short names will not
always create conflict. Of course, in either case when
some clash occurs, one can always #undef that problem.
Thus, this commit will eliminate that 'PROCPS_' prefix
making all of those enum identifiers a little shorter.
And, we'll still be well above some ridiculously short
(criminally short) names found in some common headers:
- - - - - - - - - - <term.h>
- 'tab', 'TTY', etc
- - - - - - - - - - - - - - - - <search.h>
- 'ENTER', ENTRY', 'FIND', etc
------------------------------------------------------
Finally, with this as a last of the wholesale changes,
we will have established the naming conventions below:
. only functions will begin with that 'procps_' prefix
. exposed structures begin with the module/header name
. item enumerators begin like structs, but capitalized
. other enumerators work exactly like item enumerators
. macros and constants begin just like the enumerators
------------------------------------------------------
Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-21 10:30:00 +05:30
|
|
|
* if (procps_pids_new(&info, PIDS_noop, 3) < 0)
|
2016-05-17 00:44:14 +05:30
|
|
|
* ^~~~~~~~~~~~~~~~
|
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
|
2015-09-01 10:30:00 +05:30
|
|
|
|| (void *)items < (void *)0x8000) // twice as big as our largest enum
|
2017-11-18 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++) {
|
2015-08-28 10:30:00 +05:30
|
|
|
// a pids_item is currently unsigned, but we'll protect our future
|
2015-08-19 10:30:00 +05:30
|
|
|
if (items[i] < 0)
|
2017-11-18 10:30:00 +05:30
|
|
|
return 1;
|
library: removed all the 'PROCPS_' enumerator prefixes
Many of our item enumerator identifiers are very long,
especially in that <VMSTAT> module. Additionally, they
all contain the exact same universal 'PROCPS_' prefix.
The origins for this are likely found in the desire to
avoid name clashes with other potential include files.
But with procps-ng newlib, we've probably gone way too
far. Did 'PROCPS_PIDS_TICS_SYSTEM' actually offer more
protection against clash than 'PIDS_TICS_SYSTEM' does?
I don't think so. Besides, no matter how big that name
becomes, one can never guarantee they'll never be some
clash. And, conversely, extremely short names will not
always create conflict. Of course, in either case when
some clash occurs, one can always #undef that problem.
Thus, this commit will eliminate that 'PROCPS_' prefix
making all of those enum identifiers a little shorter.
And, we'll still be well above some ridiculously short
(criminally short) names found in some common headers:
- - - - - - - - - - <term.h>
- 'tab', 'TTY', etc
- - - - - - - - - - - - - - - - <search.h>
- 'ENTER', ENTRY', 'FIND', etc
------------------------------------------------------
Finally, with this as a last of the wholesale changes,
we will have established the naming conventions below:
. only functions will begin with that 'procps_' prefix
. exposed structures begin with the module/header name
. item enumerators begin like structs, but capitalized
. other enumerators work exactly like item enumerators
. macros and constants begin just like the enumerators
------------------------------------------------------
Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-21 10:30:00 +05:30
|
|
|
if (items[i] >= PIDS_logical_end) {
|
2017-11-18 10:30:00 +05:30
|
|
|
return 1;
|
2015-08-19 10:30:00 +05:30
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
return 0;
|
2016-08-21 10:30:00 +05:30
|
|
|
} // end: pids_items_check_failed
|
2015-08-19 10:30:00 +05:30
|
|
|
|
|
|
|
|
|
2016-08-21 10:30:00 +05:30
|
|
|
static inline void pids_libflags_set (
|
2016-07-21 10:30:00 +05:30
|
|
|
struct pids_info *info)
|
2015-08-19 10:30:00 +05:30
|
|
|
{
|
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;
|
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;
|
2015-08-19 10:30:00 +05:30
|
|
|
for (i = 0; i < info->curitems; i++) {
|
library: removed all the 'PROCPS_' enumerator prefixes
Many of our item enumerator identifiers are very long,
especially in that <VMSTAT> module. Additionally, they
all contain the exact same universal 'PROCPS_' prefix.
The origins for this are likely found in the desire to
avoid name clashes with other potential include files.
But with procps-ng newlib, we've probably gone way too
far. Did 'PROCPS_PIDS_TICS_SYSTEM' actually offer more
protection against clash than 'PIDS_TICS_SYSTEM' does?
I don't think so. Besides, no matter how big that name
becomes, one can never guarantee they'll never be some
clash. And, conversely, extremely short names will not
always create conflict. Of course, in either case when
some clash occurs, one can always #undef that problem.
Thus, this commit will eliminate that 'PROCPS_' prefix
making all of those enum identifiers a little shorter.
And, we'll still be well above some ridiculously short
(criminally short) names found in some common headers:
- - - - - - - - - - <term.h>
- 'tab', 'TTY', etc
- - - - - - - - - - - - - - - - <search.h>
- 'ENTER', ENTRY', 'FIND', etc
------------------------------------------------------
Finally, with this as a last of the wholesale changes,
we will have established the naming conventions below:
. only functions will begin with that 'procps_' prefix
. exposed structures begin with the module/header name
. item enumerators begin like structs, but capitalized
. other enumerators work exactly like item enumerators
. macros and constants begin just like the enumerators
------------------------------------------------------
Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-21 10:30:00 +05:30
|
|
|
if (((e = info->items[i])) >= PIDS_logical_end)
|
2016-04-25 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
|
|
|
info->oldflags |= Item_table[e].oldflags;
|
2016-04-25 10:30:00 +05:30
|
|
|
info->history_yes |= Item_table[e].needhist;
|
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) {
|
2016-08-21 10:30:00 +05:30
|
|
|
if (!(info->oldflags & (f_stat | f_status)))
|
|
|
|
|
info->oldflags |= f_stat;
|
2015-08-19 10:30:00 +05:30
|
|
|
}
|
|
|
|
|
return;
|
2016-08-21 10:30:00 +05:30
|
|
|
} // end: pids_libflags_set
|
2015-08-19 10:30:00 +05:30
|
|
|
|
|
|
|
|
|
2016-08-21 10:30:00 +05:30
|
|
|
static inline void pids_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)
|
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) {
|
2017-11-18 10:30:00 +05:30
|
|
|
int errsav = errno;
|
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
|
|
|
closeproc(*this);
|
|
|
|
|
*this = NULL;
|
2017-11-18 10:30:00 +05:30
|
|
|
errno = errsav;
|
2015-08-19 10:30:00 +05:30
|
|
|
}
|
2016-08-21 10:30:00 +05:30
|
|
|
} // end: pids_oldproc_close
|
2015-08-19 10:30:00 +05:30
|
|
|
|
|
|
|
|
|
2016-08-21 10:30:00 +05:30
|
|
|
static inline int pids_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,
|
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;
|
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);
|
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);
|
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)))
|
2015-08-19 10:30:00 +05:30
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
return 1;
|
2016-08-21 10:30:00 +05:30
|
|
|
} // end: pids_oldproc_open
|
2015-08-19 10:30:00 +05:30
|
|
|
|
|
|
|
|
|
2016-08-21 10:30:00 +05:30
|
|
|
static inline int pids_proc_tally (
|
2016-07-21 10:30:00 +05:30
|
|
|
struct pids_info *info,
|
2015-08-19 10:30:00 +05:30
|
|
|
struct pids_counts *counts,
|
|
|
|
|
proc_t *p)
|
|
|
|
|
{
|
|
|
|
|
switch (p->state) {
|
|
|
|
|
case 'R':
|
|
|
|
|
++counts->running;
|
|
|
|
|
break;
|
2018-01-13 11:30:00 +05:30
|
|
|
case 'D': // 'D' (disk sleep)
|
2015-08-19 10:30:00 +05:30
|
|
|
case 'S':
|
|
|
|
|
++counts->sleeping;
|
|
|
|
|
break;
|
2018-01-13 11:30:00 +05:30
|
|
|
case 't': // 't' (tracing stop)
|
2015-08-19 10:30:00 +05:30
|
|
|
case 'T':
|
|
|
|
|
++counts->stopped;
|
|
|
|
|
break;
|
|
|
|
|
case 'Z':
|
|
|
|
|
++counts->zombied;
|
|
|
|
|
break;
|
2018-01-13 11:30:00 +05:30
|
|
|
default:
|
|
|
|
|
/* currently: 'I' (idle),
|
|
|
|
|
'P' (parked),
|
|
|
|
|
'X' (dead - actually 'dying' & probably never seen)
|
|
|
|
|
*/
|
|
|
|
|
++counts->other;
|
2015-08-19 10:30:00 +05:30
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
++counts->total;
|
|
|
|
|
|
|
|
|
|
if (info->history_yes)
|
2017-11-18 10:30:00 +05:30
|
|
|
return pids_make_hist(info, p);
|
2015-08-19 10:30:00 +05:30
|
|
|
return 1;
|
2016-08-21 10:30:00 +05:30
|
|
|
} // end: pids_proc_tally
|
2015-08-19 10:30:00 +05:30
|
|
|
|
2015-10-01 10:30:00 +05:30
|
|
|
|
|
|
|
|
/*
|
2016-08-21 10:30:00 +05:30
|
|
|
* pids_stacks_alloc():
|
2015-10-01 10:30:00 +05:30
|
|
|
*
|
|
|
|
|
* Allocate and initialize one or more stacks each of which is anchored in an
|
2016-08-21 10:30:00 +05:30
|
|
|
* associated context structure.
|
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.
|
|
|
|
|
*/
|
2016-08-21 10:30:00 +05:30
|
|
|
static struct stacks_extent *pids_stacks_alloc (
|
2016-07-21 10:30:00 +05:30
|
|
|
struct pids_info *info,
|
2015-10-01 10:30:00 +05:30
|
|
|
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;
|
|
|
|
|
|
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. | */
|
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 |
|
2015-10-01 10:30:00 +05:30
|
|
|
|
|
|
|
|
for (i = 0; i < maxstacks; i++) {
|
|
|
|
|
p_head = (struct pids_stack *)v_head;
|
2016-08-21 10:30:00 +05:30
|
|
|
p_head->head = pids_itemize_stack((struct pids_result *)v_list, info->curitems, info->items);
|
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;
|
2016-08-21 10:30:00 +05:30
|
|
|
} // end: pids_stacks_alloc
|
2015-10-01 10:30:00 +05:30
|
|
|
|
|
|
|
|
|
2016-08-21 10:30:00 +05:30
|
|
|
static int pids_stacks_fetch (
|
2016-07-21 10:30:00 +05:30
|
|
|
struct pids_info *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) {
|
2016-09-09 15:14:44 +05:30
|
|
|
if (!(info->fetch.anchor = calloc(sizeof(void *), STACKS_INCR)))
|
2017-11-18 10:30:00 +05:30
|
|
|
return -1;
|
2019-05-02 10:30:00 +05:30
|
|
|
if (!(ext = pids_stacks_alloc(info, STACKS_INCR)))
|
2017-11-18 10:30:00 +05:30
|
|
|
return -1; // here, errno was set to ENOMEM
|
2019-05-02 10:30:00 +05:30
|
|
|
memcpy(info->fetch.anchor, ext->stacks, sizeof(void *) * STACKS_INCR);
|
|
|
|
|
n_alloc = STACKS_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
|
|
|
}
|
2016-08-21 10:30:00 +05:30
|
|
|
pids_toggle_history(info);
|
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;
|
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)) {
|
2016-09-09 15:14:44 +05:30
|
|
|
n_alloc += STACKS_INCR;
|
2017-11-18 10:30:00 +05:30
|
|
|
if (!(info->fetch.anchor = realloc(info->fetch.anchor, sizeof(void *) * n_alloc))
|
2016-09-09 15:14:44 +05:30
|
|
|
|| (!(ext = pids_stacks_alloc(info, STACKS_INCR))))
|
2017-11-18 10:30:00 +05:30
|
|
|
return -1; // here, errno was set to ENOMEM
|
2016-09-09 15:14:44 +05:30
|
|
|
memcpy(info->fetch.anchor + n_inuse, ext->stacks, sizeof(void *) * STACKS_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
|
|
|
}
|
2016-08-21 10:30:00 +05:30
|
|
|
if (!pids_proc_tally(info, &info->fetch.counts, &task))
|
2017-11-18 10:30:00 +05:30
|
|
|
return -1; // here, errno was set to ENOMEM
|
|
|
|
|
if (!pids_assign_results(info, info->fetch.anchor[n_inuse++], &task))
|
|
|
|
|
return -1; // here, errno was set to 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
|
|
|
}
|
2020-07-24 20:54:57 +05:30
|
|
|
/* while the possibility is extremely remote, the readproc.c (read_something) |
|
|
|
|
|
simple_readproc and simple_readtask guys could have encountered this error |
|
|
|
|
|
in which case they would have returned a NULL, thus ending our while loop. | */
|
2017-11-18 10:30:00 +05:30
|
|
|
if (errno == ENOMEM)
|
|
|
|
|
return -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
|
|
|
|
|
|
|
|
// finalize stuff -------------------------------------
|
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 |
|
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;
|
2016-06-06 10:30:00 +05:30
|
|
|
if (!(info->fetch.results.stacks = realloc(info->fetch.results.stacks, sizeof(void *) * n_saved)))
|
2017-11-18 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
|
|
|
}
|
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;
|
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
|
2016-08-21 10:30:00 +05:30
|
|
|
} // end: pids_stacks_fetch
|
2015-10-01 10:30:00 +05:30
|
|
|
|
|
|
|
|
|
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 --------------------------------------------
|
2015-09-11 10:30:00 +05:30
|
|
|
|
2015-08-19 10:30:00 +05:30
|
|
|
/*
|
|
|
|
|
* procps_pids_new():
|
|
|
|
|
*
|
|
|
|
|
* @info: location of returned new structure
|
|
|
|
|
*
|
2016-07-13 10:30:00 +05:30
|
|
|
* Returns: < 0 on failure, 0 on success along with
|
|
|
|
|
* a pointer to a new context struct
|
2015-08-19 10:30:00 +05:30
|
|
|
*/
|
|
|
|
|
PROCPS_EXPORT int procps_pids_new (
|
2016-07-21 10:30:00 +05:30
|
|
|
struct pids_info **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)
|
2015-08-19 10:30:00 +05:30
|
|
|
{
|
2016-07-21 10:30:00 +05:30
|
|
|
struct pids_info *p;
|
2015-09-30 10:30:00 +05:30
|
|
|
double uptime_secs;
|
2015-08-19 10:30:00 +05:30
|
|
|
int pgsz;
|
|
|
|
|
|
2020-08-10 10:30:00 +05:30
|
|
|
#ifdef ITEMTABLE_DEBUG
|
|
|
|
|
int i, failed = 0;
|
|
|
|
|
for (i = 0; i < MAXTABLE(Item_table); i++) {
|
|
|
|
|
if (i != Item_table[i].enumnumb) {
|
|
|
|
|
fprintf(stderr, "%s: enum/table error: Item_table[%d] was %s, but its value is %d\n"
|
|
|
|
|
, __FILE__, i, Item_table[i].enum2str, Item_table[i].enumnumb);
|
|
|
|
|
failed = 1;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
if (i != PIDS_logical_end) {
|
|
|
|
|
fprintf(stderr, "%s: PIDS_logical_end is %d, expected %d\n"
|
|
|
|
|
, __FILE__, PIDS_logical_end, i);
|
|
|
|
|
failed = 1;
|
|
|
|
|
}
|
|
|
|
|
if (failed) _Exit(EXIT_FAILURE);
|
|
|
|
|
#endif
|
|
|
|
|
|
2015-08-19 10:30:00 +05:30
|
|
|
if (info == NULL || *info != NULL)
|
|
|
|
|
return -EINVAL;
|
2016-07-21 10:30:00 +05:30
|
|
|
if (!(p = calloc(1, sizeof(struct pids_info))))
|
2015-08-19 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
|
|
|
|
|
|
|
|
/* if we're without items or numitems, a later call to
|
|
|
|
|
procps_pids_reset() will become mandatory */
|
|
|
|
|
if (items && numitems) {
|
2016-08-21 10:30:00 +05:30
|
|
|
if (pids_items_check_failed(items, numitems)) {
|
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;
|
2016-05-17 00:44:14 +05:30
|
|
|
}
|
library: removed all the 'PROCPS_' enumerator prefixes
Many of our item enumerator identifiers are very long,
especially in that <VMSTAT> module. Additionally, they
all contain the exact same universal 'PROCPS_' prefix.
The origins for this are likely found in the desire to
avoid name clashes with other potential include files.
But with procps-ng newlib, we've probably gone way too
far. Did 'PROCPS_PIDS_TICS_SYSTEM' actually offer more
protection against clash than 'PIDS_TICS_SYSTEM' does?
I don't think so. Besides, no matter how big that name
becomes, one can never guarantee they'll never be some
clash. And, conversely, extremely short names will not
always create conflict. Of course, in either case when
some clash occurs, one can always #undef that problem.
Thus, this commit will eliminate that 'PROCPS_' prefix
making all of those enum identifiers a little shorter.
And, we'll still be well above some ridiculously short
(criminally short) names found in some common headers:
- - - - - - - - - - <term.h>
- 'tab', 'TTY', etc
- - - - - - - - - - - - - - - - <search.h>
- 'ENTER', ENTRY', 'FIND', etc
------------------------------------------------------
Finally, with this as a last of the wholesale changes,
we will have established the naming conventions below:
. only functions will begin with that 'procps_' prefix
. exposed structures begin with the module/header name
. item enumerators begin like structs, but capitalized
. other enumerators work exactly like item enumerators
. macros and constants begin just like the enumerators
------------------------------------------------------
Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-21 10:30:00 +05:30
|
|
|
// allow for our 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
|
|
|
p->maxitems = numitems + 1;
|
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;
|
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);
|
library: removed all the 'PROCPS_' enumerator prefixes
Many of our item enumerator identifiers are very long,
especially in that <VMSTAT> module. Additionally, they
all contain the exact same universal 'PROCPS_' prefix.
The origins for this are likely found in the desire to
avoid name clashes with other potential include files.
But with procps-ng newlib, we've probably gone way too
far. Did 'PROCPS_PIDS_TICS_SYSTEM' actually offer more
protection against clash than 'PIDS_TICS_SYSTEM' does?
I don't think so. Besides, no matter how big that name
becomes, one can never guarantee they'll never be some
clash. And, conversely, extremely short names will not
always create conflict. Of course, in either case when
some clash occurs, one can always #undef that problem.
Thus, this commit will eliminate that 'PROCPS_' prefix
making all of those enum identifiers a little shorter.
And, we'll still be well above some ridiculously short
(criminally short) names found in some common headers:
- - - - - - - - - - <term.h>
- 'tab', 'TTY', etc
- - - - - - - - - - - - - - - - <search.h>
- 'ENTER', ENTRY', 'FIND', etc
------------------------------------------------------
Finally, with this as a last of the wholesale changes,
we will have established the naming conventions below:
. only functions will begin with that 'procps_' prefix
. exposed structures begin with the module/header name
. item enumerators begin like structs, but capitalized
. other enumerators work exactly like item enumerators
. macros and constants begin just like the enumerators
------------------------------------------------------
Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-21 10:30:00 +05:30
|
|
|
p->items[numitems] = 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
|
|
|
p->curitems = p->maxitems;
|
2016-08-21 10:30:00 +05:30
|
|
|
pids_libflags_set(p);
|
2015-08-19 10:30:00 +05:30
|
|
|
}
|
|
|
|
|
|
2016-09-09 15:14:44 +05:30
|
|
|
if (!(p->hist = calloc(NEWOLD_INCR, sizeof(struct history_info)))) {
|
2016-05-17 00:44:14 +05:30
|
|
|
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;
|
2016-05-17 00:44:14 +05:30
|
|
|
}
|
2016-08-21 10:30:00 +05:30
|
|
|
pids_config_history(p);
|
2015-08-19 10:30:00 +05:30
|
|
|
|
|
|
|
|
pgsz = getpagesize();
|
|
|
|
|
while (pgsz > 1024) { pgsz >>= 1; p->pgs2k_shift++; }
|
|
|
|
|
|
2015-09-30 10:30:00 +05:30
|
|
|
p->hertz = procps_hertz_get();
|
|
|
|
|
procps_uptime(&uptime_secs, NULL);
|
2015-10-07 10:30:00 +05:30
|
|
|
p->boot_seconds = uptime_secs;
|
2015-09-30 10:30:00 +05:30
|
|
|
|
2017-05-12 10:33:00 +05:30
|
|
|
numa_init();
|
|
|
|
|
|
2016-06-23 10:30:00 +05:30
|
|
|
p->fetch.results.counts = &p->fetch.counts;
|
|
|
|
|
|
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 (
|
2016-07-21 10:30:00 +05:30
|
|
|
struct pids_info *info)
|
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
|
|
|
if (info == NULL)
|
|
|
|
|
return -EINVAL;
|
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 (
|
2016-07-21 10:30:00 +05:30
|
|
|
struct pids_info **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 || *info == NULL)
|
|
|
|
|
return -EINVAL;
|
|
|
|
|
|
|
|
|
|
(*info)->refcount--;
|
2016-11-22 20:40:10 +05:30
|
|
|
|
|
|
|
|
if ((*info)->refcount < 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
|
|
|
#ifdef UNREF_RPTHASH
|
2016-08-21 10:30:00 +05:30
|
|
|
pids_unref_rpthash(*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
|
|
|
#endif
|
|
|
|
|
if ((*info)->extents) {
|
2016-08-21 10:30:00 +05:30
|
|
|
pids_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
|
|
|
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;
|
2016-08-21 10:30:00 +05:30
|
|
|
pids_cleanup_stack(ext->stacks[0]->head);
|
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
|
|
|
free(ext);
|
|
|
|
|
ext = nextext;
|
|
|
|
|
};
|
|
|
|
|
}
|
|
|
|
|
if ((*info)->fetch.anchor)
|
|
|
|
|
free((*info)->fetch.anchor);
|
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);
|
|
|
|
|
}
|
|
|
|
|
|
2019-05-02 10:30:00 +05:30
|
|
|
if ((*info)->get_ext)
|
|
|
|
|
pids_oldproc_close(&(*info)->get_PT);
|
|
|
|
|
|
2017-05-12 10:33:00 +05:30
|
|
|
numa_uninit();
|
|
|
|
|
|
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
|
|
|
free(*info);
|
|
|
|
|
*info = NULL;
|
|
|
|
|
return 0;
|
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
|
|
|
|
|
|
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 -------------------------------------------
|
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 (
|
2016-07-21 10:30:00 +05:30
|
|
|
struct pids_info *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)
|
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;
|
2015-08-26 10:30:00 +05:30
|
|
|
|
2016-06-15 10:30:00 +05:30
|
|
|
/* this is very likely the *only* newlib function where the
|
2016-07-21 10:30:00 +05:30
|
|
|
context (pids_info) of NULL will ever be permitted */
|
2017-11-18 10:30:00 +05:30
|
|
|
if (!look_up_our_self(&self)
|
|
|
|
|
|| (!return_self))
|
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;
|
2015-08-26 10:30:00 +05:30
|
|
|
|
2017-11-18 10:30:00 +05:30
|
|
|
errno = EINVAL;
|
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;
|
2017-11-18 10:30:00 +05:30
|
|
|
errno = 0;
|
2016-06-15 10:30:00 +05:30
|
|
|
|
2016-08-21 10:30:00 +05:30
|
|
|
if (!(ext = pids_stacks_alloc(info, 1)))
|
2016-06-15 10:30:00 +05:30
|
|
|
return NULL;
|
2017-11-18 10:30:00 +05:30
|
|
|
if (!pids_extent_cut(info, ext)) {
|
|
|
|
|
errno = EADDRNOTAVAIL;
|
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;
|
2017-11-18 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;
|
2017-11-18 10:30:00 +05:30
|
|
|
if (!pids_assign_results(info, ext->stacks[0], &self))
|
|
|
|
|
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
|
|
|
|
|
|
|
|
return ext->stacks[0];
|
|
|
|
|
} // end: fatal_proc_unmounted
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
PROCPS_EXPORT struct pids_stack *procps_pids_get (
|
2016-07-21 10:30:00 +05:30
|
|
|
struct pids_info *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)
|
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)
|
|
|
|
|
|
2017-11-18 10:30:00 +05:30
|
|
|
errno = EINVAL;
|
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 NULL;
|
library: removed all the 'PROCPS_' enumerator prefixes
Many of our item enumerator identifiers are very long,
especially in that <VMSTAT> module. Additionally, they
all contain the exact same universal 'PROCPS_' prefix.
The origins for this are likely found in the desire to
avoid name clashes with other potential include files.
But with procps-ng newlib, we've probably gone way too
far. Did 'PROCPS_PIDS_TICS_SYSTEM' actually offer more
protection against clash than 'PIDS_TICS_SYSTEM' does?
I don't think so. Besides, no matter how big that name
becomes, one can never guarantee they'll never be some
clash. And, conversely, extremely short names will not
always create conflict. Of course, in either case when
some clash occurs, one can always #undef that problem.
Thus, this commit will eliminate that 'PROCPS_' prefix
making all of those enum identifiers a little shorter.
And, we'll still be well above some ridiculously short
(criminally short) names found in some common headers:
- - - - - - - - - - <term.h>
- 'tab', 'TTY', etc
- - - - - - - - - - - - - - - - <search.h>
- 'ENTER', ENTRY', 'FIND', etc
------------------------------------------------------
Finally, with this as a last of the wholesale changes,
we will have established the naming conventions below:
. only functions will begin with that 'procps_' prefix
. exposed structures begin with the module/header name
. item enumerators begin like structs, but capitalized
. other enumerators work exactly like item enumerators
. macros and constants begin just like the enumerators
------------------------------------------------------
Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-21 10:30:00 +05:30
|
|
|
if (which != PIDS_FETCH_TASKS_ONLY && which != PIDS_FETCH_THREADS_TOO)
|
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;
|
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
|
|
|
|
|
|
|
|
if (!info->get_ext) {
|
2016-08-21 10:30:00 +05:30
|
|
|
if (!(info->get_ext = pids_stacks_alloc(info, 1)))
|
2017-11-18 10:30:00 +05:30
|
|
|
return NULL; // here, errno was overridden with ENOMEM
|
2019-05-02 10:30:00 +05:30
|
|
|
fresh_start:
|
2016-08-21 10:30:00 +05:30
|
|
|
if (!pids_oldproc_open(&info->get_PT, info->oldflags))
|
2017-11-18 10:30:00 +05:30
|
|
|
return NULL; // here, errno was overridden with ENOMEM/others
|
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->get_type = which;
|
|
|
|
|
info->read_something = which ? readeither : readproc;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (info->get_type != which) {
|
2016-08-21 10:30:00 +05:30
|
|
|
pids_oldproc_close(&info->get_PT);
|
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
|
|
|
goto fresh_start;
|
|
|
|
|
}
|
2017-11-18 10:30:00 +05:30
|
|
|
errno = 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 (NULL == info->read_something(info->get_PT, &task))
|
|
|
|
|
return NULL;
|
2017-11-18 10:30:00 +05:30
|
|
|
if (!pids_assign_results(info, info->get_ext->stacks[0], &task))
|
|
|
|
|
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
|
|
|
return info->get_ext->stacks[0];
|
|
|
|
|
} // end: procps_pids_get
|
2015-08-26 10:30:00 +05:30
|
|
|
|
|
|
|
|
|
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.
|
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 (
|
2016-07-21 10:30:00 +05:30
|
|
|
struct pids_info *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)
|
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;
|
2015-08-19 10:30:00 +05:30
|
|
|
|
2017-11-18 10:30:00 +05:30
|
|
|
errno = EINVAL;
|
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)
|
2015-08-26 10:30:00 +05:30
|
|
|
return NULL;
|
library: removed all the 'PROCPS_' enumerator prefixes
Many of our item enumerator identifiers are very long,
especially in that <VMSTAT> module. Additionally, they
all contain the exact same universal 'PROCPS_' prefix.
The origins for this are likely found in the desire to
avoid name clashes with other potential include files.
But with procps-ng newlib, we've probably gone way too
far. Did 'PROCPS_PIDS_TICS_SYSTEM' actually offer more
protection against clash than 'PIDS_TICS_SYSTEM' does?
I don't think so. Besides, no matter how big that name
becomes, one can never guarantee they'll never be some
clash. And, conversely, extremely short names will not
always create conflict. Of course, in either case when
some clash occurs, one can always #undef that problem.
Thus, this commit will eliminate that 'PROCPS_' prefix
making all of those enum identifiers a little shorter.
And, we'll still be well above some ridiculously short
(criminally short) names found in some common headers:
- - - - - - - - - - <term.h>
- 'tab', 'TTY', etc
- - - - - - - - - - - - - - - - <search.h>
- 'ENTER', ENTRY', 'FIND', etc
------------------------------------------------------
Finally, with this as a last of the wholesale changes,
we will have established the naming conventions below:
. only functions will begin with that 'procps_' prefix
. exposed structures begin with the module/header name
. item enumerators begin like structs, but capitalized
. other enumerators work exactly like item enumerators
. macros and constants begin just like the enumerators
------------------------------------------------------
Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-21 10:30:00 +05:30
|
|
|
if (which != PIDS_FETCH_TASKS_ONLY && which != PIDS_FETCH_THREADS_TOO)
|
2015-08-26 10:30:00 +05:30
|
|
|
return NULL;
|
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;
|
2017-11-18 10:30:00 +05:30
|
|
|
errno = 0;
|
2015-08-19 10:30:00 +05:30
|
|
|
|
2016-08-21 10:30:00 +05:30
|
|
|
if (!pids_oldproc_open(&info->fetch_PT, info->oldflags))
|
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;
|
2015-08-19 10:30:00 +05:30
|
|
|
|
2016-08-21 10:30:00 +05:30
|
|
|
rc = pids_stacks_fetch(info);
|
2015-08-19 10:30:00 +05:30
|
|
|
|
2016-08-21 10:30:00 +05:30
|
|
|
pids_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
|
2016-06-06 10:30:00 +05:30
|
|
|
return (rc > 0) ? &info->fetch.results : NULL;
|
2015-08-19 10:30:00 +05:30
|
|
|
} // end: procps_pids_reap
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
PROCPS_EXPORT int procps_pids_reset (
|
2016-07-21 10:30:00 +05:30
|
|
|
struct pids_info *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)
|
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)
|
2015-08-19 10:30:00 +05:30
|
|
|
return -EINVAL;
|
2016-08-21 10:30:00 +05:30
|
|
|
if (pids_items_check_failed(newitems, newnumitems))
|
2015-08-19 10:30:00 +05:30
|
|
|
return -EINVAL;
|
|
|
|
|
|
library: minimize the use of 'cleanup_stacks' routines
Some parts of our newlib implementation are the result
of functions which have been propagated from module to
module. In particular, those 'cleanup_stacks' routines
are all similar & likely originated in the <pids> api.
In that interface there was a need to free dynamically
acquired memory before the result structure was reused
to satisfy subsequent 'get', 'select' or 'reap' calls.
This, in turn, led to a concept of 'dirty' stacks with
the need to call one of two 'cleanup_stack' functions.
None of the remaining interfaces deal with such memory
yet they each had their own 'cleanup_stack' functions.
Those functions were responsible for resetting each of
the result unions to zero, excluding any 'noop' items.
The bottom line is that for all interfaces, repetitive
calls would require iterating through the stack(s) two
separate times: once to 'cleanup' another to 'assign'.
With this commit we will reduce iterations to just the
'assign' routine. A reset to zero will be accomplished
in the 'extra' item set routine (which is the only one
actually requiring any reset). All other items will be
reinitialized automatically by a new current set value
or upon reallocation when an items compliment changes.
In the <pids> interface, any freeing of dynamic memory
could have been accomplished by adding that 'freefunc'
check to the 'assign' function. However, that requires
an Item_table test with every item. Instead, we'll now
satisfy such needs as the very first step in those set
functions responsible for dynamically acquired memory.
[ the <pids> api retains 2 'cleanup_stack' functions ]
[ to accommodate stack(s) 'reset' & to serve 'unref' ]
Lastly, all the 'itemize_stack' functions were tweaked
by eliminating an unnecessary initialization of result
unions. That objective was already accomplished by the
calloc() in a 'stacks_alloc' function or the remaining
'cleanup_stack' routine found in the <pids> interface.
Signed-off-by: Jim Warner <james.warner@comcast.net>
2019-05-22 10:30:00 +05:30
|
|
|
pids_cleanup_stacks_all(info);
|
2019-05-02 10:30:00 +05:30
|
|
|
|
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))
|
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) {
|
2016-08-21 10:30:00 +05:30
|
|
|
while (info->extents) {
|
|
|
|
|
struct stacks_extent *p = info->extents;
|
|
|
|
|
info->extents = p->next;
|
|
|
|
|
free(p);
|
|
|
|
|
};
|
2019-05-02 10:30:00 +05:30
|
|
|
if (info->get_ext) {
|
|
|
|
|
pids_oldproc_close(&info->get_PT);
|
|
|
|
|
info->get_ext = NULL;
|
|
|
|
|
}
|
|
|
|
|
if (info->fetch.anchor) {
|
|
|
|
|
free(info->fetch.anchor);
|
|
|
|
|
info->fetch.anchor = NULL;
|
|
|
|
|
}
|
library: removed all the 'PROCPS_' enumerator prefixes
Many of our item enumerator identifiers are very long,
especially in that <VMSTAT> module. Additionally, they
all contain the exact same universal 'PROCPS_' prefix.
The origins for this are likely found in the desire to
avoid name clashes with other potential include files.
But with procps-ng newlib, we've probably gone way too
far. Did 'PROCPS_PIDS_TICS_SYSTEM' actually offer more
protection against clash than 'PIDS_TICS_SYSTEM' does?
I don't think so. Besides, no matter how big that name
becomes, one can never guarantee they'll never be some
clash. And, conversely, extremely short names will not
always create conflict. Of course, in either case when
some clash occurs, one can always #undef that problem.
Thus, this commit will eliminate that 'PROCPS_' prefix
making all of those enum identifiers a little shorter.
And, we'll still be well above some ridiculously short
(criminally short) names found in some common headers:
- - - - - - - - - - <term.h>
- 'tab', 'TTY', etc
- - - - - - - - - - - - - - - - <search.h>
- 'ENTER', ENTRY', 'FIND', etc
------------------------------------------------------
Finally, with this as a last of the wholesale changes,
we will have established the naming conventions below:
. only functions will begin with that 'procps_' prefix
. exposed structures begin with the module/header name
. item enumerators begin like structs, but capitalized
. other enumerators work exactly like item enumerators
. macros and constants begin just like the enumerators
------------------------------------------------------
Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-21 10:30:00 +05:30
|
|
|
// allow for our 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->maxitems = newnumitems + 1;
|
|
|
|
|
if (!(info->items = realloc(info->items, sizeof(enum pids_item) * info->maxitems)))
|
|
|
|
|
return -ENOMEM;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
memcpy(info->items, newitems, sizeof(enum pids_item) * newnumitems);
|
library: removed all the 'PROCPS_' enumerator prefixes
Many of our item enumerator identifiers are very long,
especially in that <VMSTAT> module. Additionally, they
all contain the exact same universal 'PROCPS_' prefix.
The origins for this are likely found in the desire to
avoid name clashes with other potential include files.
But with procps-ng newlib, we've probably gone way too
far. Did 'PROCPS_PIDS_TICS_SYSTEM' actually offer more
protection against clash than 'PIDS_TICS_SYSTEM' does?
I don't think so. Besides, no matter how big that name
becomes, one can never guarantee they'll never be some
clash. And, conversely, extremely short names will not
always create conflict. Of course, in either case when
some clash occurs, one can always #undef that problem.
Thus, this commit will eliminate that 'PROCPS_' prefix
making all of those enum identifiers a little shorter.
And, we'll still be well above some ridiculously short
(criminally short) names found in some common headers:
- - - - - - - - - - <term.h>
- 'tab', 'TTY', etc
- - - - - - - - - - - - - - - - <search.h>
- 'ENTER', ENTRY', 'FIND', etc
------------------------------------------------------
Finally, with this as a last of the wholesale changes,
we will have established the naming conventions below:
. only functions will begin with that 'procps_' prefix
. exposed structures begin with the module/header name
. item enumerators begin like structs, but capitalized
. other enumerators work exactly like item enumerators
. macros and constants begin just like the enumerators
------------------------------------------------------
Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-21 10:30:00 +05:30
|
|
|
info->items[newnumitems] = PIDS_logical_end;
|
|
|
|
|
// account for above 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;
|
2015-08-19 10:30:00 +05:30
|
|
|
|
library: minimize the use of 'cleanup_stacks' routines
Some parts of our newlib implementation are the result
of functions which have been propagated from module to
module. In particular, those 'cleanup_stacks' routines
are all similar & likely originated in the <pids> api.
In that interface there was a need to free dynamically
acquired memory before the result structure was reused
to satisfy subsequent 'get', 'select' or 'reap' calls.
This, in turn, led to a concept of 'dirty' stacks with
the need to call one of two 'cleanup_stack' functions.
None of the remaining interfaces deal with such memory
yet they each had their own 'cleanup_stack' functions.
Those functions were responsible for resetting each of
the result unions to zero, excluding any 'noop' items.
The bottom line is that for all interfaces, repetitive
calls would require iterating through the stack(s) two
separate times: once to 'cleanup' another to 'assign'.
With this commit we will reduce iterations to just the
'assign' routine. A reset to zero will be accomplished
in the 'extra' item set routine (which is the only one
actually requiring any reset). All other items will be
reinitialized automatically by a new current set value
or upon reallocation when an items compliment changes.
In the <pids> interface, any freeing of dynamic memory
could have been accomplished by adding that 'freefunc'
check to the 'assign' function. However, that requires
an Item_table test with every item. Instead, we'll now
satisfy such needs as the very first step in those set
functions responsible for dynamically acquired memory.
[ the <pids> api retains 2 'cleanup_stack' functions ]
[ to accommodate stack(s) 'reset' & to serve 'unref' ]
Lastly, all the 'itemize_stack' functions were tweaked
by eliminating an unnecessary initialization of result
unions. That objective was already accomplished by the
calloc() in a 'stacks_alloc' function or the remaining
'cleanup_stack' routine found in the <pids> interface.
Signed-off-by: Jim Warner <james.warner@comcast.net>
2019-05-22 10:30:00 +05:30
|
|
|
// if extents freed above, pids_stacks_alloc() will itemize ...
|
2016-08-21 10:30:00 +05:30
|
|
|
pids_itemize_stacks_all(info);
|
|
|
|
|
pids_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
|
|
|
|
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 (
|
2016-07-21 10:30:00 +05:30
|
|
|
struct pids_info *info,
|
2015-10-01 10:30:00 +05:30
|
|
|
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,
|
2016-04-18 10:30:00 +05:30
|
|
|
enum pids_select_type which)
|
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;
|
2015-10-01 10:30:00 +05:30
|
|
|
|
2017-11-18 10:30:00 +05:30
|
|
|
errno = EINVAL;
|
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)
|
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)
|
2015-10-01 10:30:00 +05:30
|
|
|
return NULL;
|
library: removed all the 'PROCPS_' enumerator prefixes
Many of our item enumerator identifiers are very long,
especially in that <VMSTAT> module. Additionally, they
all contain the exact same universal 'PROCPS_' prefix.
The origins for this are likely found in the desire to
avoid name clashes with other potential include files.
But with procps-ng newlib, we've probably gone way too
far. Did 'PROCPS_PIDS_TICS_SYSTEM' actually offer more
protection against clash than 'PIDS_TICS_SYSTEM' does?
I don't think so. Besides, no matter how big that name
becomes, one can never guarantee they'll never be some
clash. And, conversely, extremely short names will not
always create conflict. Of course, in either case when
some clash occurs, one can always #undef that problem.
Thus, this commit will eliminate that 'PROCPS_' prefix
making all of those enum identifiers a little shorter.
And, we'll still be well above some ridiculously short
(criminally short) names found in some common headers:
- - - - - - - - - - <term.h>
- 'tab', 'TTY', etc
- - - - - - - - - - - - - - - - <search.h>
- 'ENTER', ENTRY', 'FIND', etc
------------------------------------------------------
Finally, with this as a last of the wholesale changes,
we will have established the naming conventions below:
. only functions will begin with that 'procps_' prefix
. exposed structures begin with the module/header name
. item enumerators begin like structs, but capitalized
. other enumerators work exactly like item enumerators
. macros and constants begin just like the enumerators
------------------------------------------------------
Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-21 10:30:00 +05:30
|
|
|
if (which != PIDS_SELECT_PID && which != PIDS_SELECT_UID)
|
2015-10-01 10:30:00 +05:30
|
|
|
return NULL;
|
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;
|
2017-11-18 10:30:00 +05:30
|
|
|
errno = 0;
|
2015-10-01 10:30:00 +05:30
|
|
|
|
library: removed all the 'PROCPS_' enumerator prefixes
Many of our item enumerator identifiers are very long,
especially in that <VMSTAT> module. Additionally, they
all contain the exact same universal 'PROCPS_' prefix.
The origins for this are likely found in the desire to
avoid name clashes with other potential include files.
But with procps-ng newlib, we've probably gone way too
far. Did 'PROCPS_PIDS_TICS_SYSTEM' actually offer more
protection against clash than 'PIDS_TICS_SYSTEM' does?
I don't think so. Besides, no matter how big that name
becomes, one can never guarantee they'll never be some
clash. And, conversely, extremely short names will not
always create conflict. Of course, in either case when
some clash occurs, one can always #undef that problem.
Thus, this commit will eliminate that 'PROCPS_' prefix
making all of those enum identifiers a little shorter.
And, we'll still be well above some ridiculously short
(criminally short) names found in some common headers:
- - - - - - - - - - <term.h>
- 'tab', 'TTY', etc
- - - - - - - - - - - - - - - - <search.h>
- 'ENTER', ENTRY', 'FIND', etc
------------------------------------------------------
Finally, with this as a last of the wholesale changes,
we will have established the naming conventions below:
. only functions will begin with that 'procps_' prefix
. exposed structures begin with the module/header name
. item enumerators begin like structs, but capitalized
. other enumerators work exactly like item enumerators
. macros and constants begin just like the enumerators
------------------------------------------------------
Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-21 10:30:00 +05:30
|
|
|
// this zero delimiter is really only needed with PIDS_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;
|
2015-10-01 10:30:00 +05:30
|
|
|
|
2016-08-21 10:30:00 +05:30
|
|
|
if (!pids_oldproc_open(&info->fetch_PT, (info->oldflags | which), ids, numthese))
|
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;
|
2015-10-01 10:30:00 +05:30
|
|
|
|
2016-08-21 10:30:00 +05:30
|
|
|
rc = pids_stacks_fetch(info);
|
2015-10-01 10:30:00 +05:30
|
|
|
|
2016-08-21 10:30:00 +05:30
|
|
|
pids_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
|
2017-11-18 10:30:00 +05:30
|
|
|
return (rc >= 0) ? &info->fetch.results : NULL;
|
2015-10-01 10:30:00 +05:30
|
|
|
} // end: procps_pids_select
|
|
|
|
|
|
|
|
|
|
|
2015-08-19 10:30:00 +05:30
|
|
|
/*
|
2015-10-11 10:30:00 +05:30
|
|
|
* procps_pids_sort():
|
2015-08-19 10:30:00 +05:30
|
|
|
*
|
2016-07-07 10:30:00 +05:30
|
|
|
* Sort stacks anchored in the passed stack pointers array
|
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).
|
|
|
|
|
*/
|
2015-10-11 10:30:00 +05:30
|
|
|
PROCPS_EXPORT struct pids_stack **procps_pids_sort (
|
2016-07-21 10:30:00 +05:30
|
|
|
struct pids_info *info,
|
2015-08-20 10:30:00 +05:30
|
|
|
struct pids_stack *stacks[],
|
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,
|
2015-08-19 10:30:00 +05:30
|
|
|
enum pids_sort_order order)
|
|
|
|
|
{
|
|
|
|
|
struct sort_parms parms;
|
|
|
|
|
struct pids_result *p;
|
|
|
|
|
int offset;
|
|
|
|
|
|
2017-11-18 10:30:00 +05:30
|
|
|
errno = EINVAL;
|
2015-08-19 10:30:00 +05:30
|
|
|
if (info == NULL || stacks == NULL)
|
|
|
|
|
return NULL;
|
2015-08-28 10:30:00 +05:30
|
|
|
// a pids_item is currently unsigned, but we'll protect our future
|
library: removed all the 'PROCPS_' enumerator prefixes
Many of our item enumerator identifiers are very long,
especially in that <VMSTAT> module. Additionally, they
all contain the exact same universal 'PROCPS_' prefix.
The origins for this are likely found in the desire to
avoid name clashes with other potential include files.
But with procps-ng newlib, we've probably gone way too
far. Did 'PROCPS_PIDS_TICS_SYSTEM' actually offer more
protection against clash than 'PIDS_TICS_SYSTEM' does?
I don't think so. Besides, no matter how big that name
becomes, one can never guarantee they'll never be some
clash. And, conversely, extremely short names will not
always create conflict. Of course, in either case when
some clash occurs, one can always #undef that problem.
Thus, this commit will eliminate that 'PROCPS_' prefix
making all of those enum identifiers a little shorter.
And, we'll still be well above some ridiculously short
(criminally short) names found in some common headers:
- - - - - - - - - - <term.h>
- 'tab', 'TTY', etc
- - - - - - - - - - - - - - - - <search.h>
- 'ENTER', ENTRY', 'FIND', etc
------------------------------------------------------
Finally, with this as a last of the wholesale changes,
we will have established the naming conventions below:
. only functions will begin with that 'procps_' prefix
. exposed structures begin with the module/header name
. item enumerators begin like structs, but capitalized
. other enumerators work exactly like item enumerators
. macros and constants begin just like the enumerators
------------------------------------------------------
Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-21 10:30:00 +05:30
|
|
|
if (sortitem < 0 || sortitem >= PIDS_logical_end)
|
2015-08-19 10:30:00 +05:30
|
|
|
return NULL;
|
library: removed all the 'PROCPS_' enumerator prefixes
Many of our item enumerator identifiers are very long,
especially in that <VMSTAT> module. Additionally, they
all contain the exact same universal 'PROCPS_' prefix.
The origins for this are likely found in the desire to
avoid name clashes with other potential include files.
But with procps-ng newlib, we've probably gone way too
far. Did 'PROCPS_PIDS_TICS_SYSTEM' actually offer more
protection against clash than 'PIDS_TICS_SYSTEM' does?
I don't think so. Besides, no matter how big that name
becomes, one can never guarantee they'll never be some
clash. And, conversely, extremely short names will not
always create conflict. Of course, in either case when
some clash occurs, one can always #undef that problem.
Thus, this commit will eliminate that 'PROCPS_' prefix
making all of those enum identifiers a little shorter.
And, we'll still be well above some ridiculously short
(criminally short) names found in some common headers:
- - - - - - - - - - <term.h>
- 'tab', 'TTY', etc
- - - - - - - - - - - - - - - - <search.h>
- 'ENTER', ENTRY', 'FIND', etc
------------------------------------------------------
Finally, with this as a last of the wholesale changes,
we will have established the naming conventions below:
. only functions will begin with that 'procps_' prefix
. exposed structures begin with the module/header name
. item enumerators begin like structs, but capitalized
. other enumerators work exactly like item enumerators
. macros and constants begin just like the enumerators
------------------------------------------------------
Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-21 10:30:00 +05:30
|
|
|
if (order != PIDS_SORT_ASCEND && order != PIDS_SORT_DESCEND)
|
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)
|
2015-08-19 10:30:00 +05:30
|
|
|
break;
|
|
|
|
|
++offset;
|
|
|
|
|
if (offset >= info->curitems)
|
|
|
|
|
return NULL;
|
library: removed all the 'PROCPS_' enumerator prefixes
Many of our item enumerator identifiers are very long,
especially in that <VMSTAT> module. Additionally, they
all contain the exact same universal 'PROCPS_' prefix.
The origins for this are likely found in the desire to
avoid name clashes with other potential include files.
But with procps-ng newlib, we've probably gone way too
far. Did 'PROCPS_PIDS_TICS_SYSTEM' actually offer more
protection against clash than 'PIDS_TICS_SYSTEM' does?
I don't think so. Besides, no matter how big that name
becomes, one can never guarantee they'll never be some
clash. And, conversely, extremely short names will not
always create conflict. Of course, in either case when
some clash occurs, one can always #undef that problem.
Thus, this commit will eliminate that 'PROCPS_' prefix
making all of those enum identifiers a little shorter.
And, we'll still be well above some ridiculously short
(criminally short) names found in some common headers:
- - - - - - - - - - <term.h>
- 'tab', 'TTY', etc
- - - - - - - - - - - - - - - - <search.h>
- 'ENTER', ENTRY', 'FIND', etc
------------------------------------------------------
Finally, with this as a last of the wholesale changes,
we will have established the naming conventions below:
. only functions will begin with that 'procps_' prefix
. exposed structures begin with the module/header name
. item enumerators begin like structs, but capitalized
. other enumerators work exactly like item enumerators
. macros and constants begin just like the enumerators
------------------------------------------------------
Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-21 10:30:00 +05:30
|
|
|
if (p->item >= PIDS_logical_end)
|
2015-08-19 10:30:00 +05:30
|
|
|
return NULL;
|
|
|
|
|
++p;
|
|
|
|
|
}
|
2017-11-18 10:30:00 +05:30
|
|
|
errno = 0;
|
|
|
|
|
|
2015-08-19 10:30:00 +05:30
|
|
|
parms.offset = offset;
|
|
|
|
|
parms.order = order;
|
|
|
|
|
|
2015-08-26 10:30:00 +05:30
|
|
|
qsort_r(stacks, numstacked, sizeof(void *), (QSR_t)Item_table[p->item].sortfunc, &parms);
|
2015-08-19 10:30:00 +05:30
|
|
|
return stacks;
|
2015-10-11 10:30:00 +05:30
|
|
|
} // end: procps_pids_sort
|
2016-08-05 10:30:00 +05:30
|
|
|
|
|
|
|
|
|
|
|
|
|
// --- special debugging function(s) ------------------------------------------
|
|
|
|
|
/*
|
|
|
|
|
* The following isn't part of the normal programming interface. Rather,
|
|
|
|
|
* it exists to validate result types referenced in application programs.
|
|
|
|
|
*
|
|
|
|
|
* It's used only when:
|
|
|
|
|
* 1) the 'XTRA_PROCPS_DEBUG' has been defined, or
|
2020-07-13 10:30:00 +05:30
|
|
|
* 2) an #include of 'xtra-procps-debug.h' is used
|
2016-08-05 10:30:00 +05:30
|
|
|
*/
|
|
|
|
|
|
2016-08-08 11:53:45 +05:30
|
|
|
PROCPS_EXPORT struct pids_result *xtra_pids_val (
|
2016-08-05 10:30:00 +05:30
|
|
|
int relative_enum,
|
|
|
|
|
const char *typestr,
|
|
|
|
|
const struct pids_stack *stack,
|
|
|
|
|
struct pids_info *info,
|
|
|
|
|
const char *file,
|
|
|
|
|
int lineno)
|
|
|
|
|
{
|
|
|
|
|
char *str;
|
2016-08-08 11:53:45 +05:30
|
|
|
int i;
|
2016-08-05 10:30:00 +05:30
|
|
|
|
2016-08-08 11:53:45 +05:30
|
|
|
for (i = 0; stack->head[i].item < PIDS_logical_end; i++)
|
|
|
|
|
;
|
|
|
|
|
if (relative_enum < 0 || relative_enum >= i) {
|
2020-07-13 10:30:00 +05:30
|
|
|
fprintf(stderr, "%s line %d: invalid relative_enum = %d, valid range = 0-%d\n"
|
|
|
|
|
, file, lineno, relative_enum, i-1);
|
2016-08-08 11:53:45 +05:30
|
|
|
return NULL;
|
2016-08-05 10:30:00 +05:30
|
|
|
}
|
2016-08-08 11:53:45 +05:30
|
|
|
str = Item_table[stack->head[relative_enum].item].type2str;
|
2016-08-05 10:30:00 +05:30
|
|
|
if (str[0]
|
2016-08-08 11:53:45 +05:30
|
|
|
&& (strcmp(typestr, str))) {
|
2016-08-05 10:30:00 +05:30
|
|
|
fprintf(stderr, "%s line %d: was %s, expected %s\n", file, lineno, typestr, str);
|
2016-08-08 11:53:45 +05:30
|
|
|
}
|
|
|
|
|
return &stack->head[relative_enum];
|
2018-06-03 10:30:00 +05:30
|
|
|
(void)info;
|
2016-08-05 10:30:00 +05:30
|
|
|
} // end: xtra_pids_val
|
