Kill CONFIG_FEATURE_USE_DEVPS_PATCH and the devps patch. I'm not

maintaining it anymore, and it is now terribly out of date.
 -Erik
This commit is contained in:
Eric Andersen 2002-12-05 07:24:08 +00:00
parent 6f9b45b9ef
commit deca106b6d
7 changed files with 3 additions and 1859 deletions

View File

@ -1,104 +0,0 @@
I have been asked several times whether the devps patch will go into the
mainline Linux kernel. The following emails from Alan Cox and Linus Torvalds
make it clear that it is not going to happen. This does not mean this patch
had no value -- it does. It just means that those that like it get to apply it
themselves...
-Erik
-------------------------------
From alan@lxorguk.ukuu.org.uk Thu Apr 13 08:07:22 2000
Return-Path: <alan@lxorguk.ukuu.org.uk>
Delivered-To: andersen@dillweed.dsl.xmission.com
Received: from localhost (dillweed.dsl.xmission.com [10.0.0.1])
by dillweed.dsl.xmission.com (Postfix) with ESMTP id 1D57A11A4F5
for <andersen@dillweed.dsl.xmission.com>; Thu, 13 Apr 2000 08:07:22 -0600 (MDT)
Envelope-to: andersen@xmission.com
Received: from mail.xmission.com
by localhost with IMAP (fetchmail-5.3.3)
for andersen@dillweed.dsl.xmission.com (single-drop); Thu, 13 Apr 2000 08:07:22 -0600 (MDT)
Received: from [194.168.151.1] (helo=the-village.bc.nu)
by mail.xmission.com with esmtp (Exim 3.03 #3)
id 12fhQk-0002OZ-00
for andersen@xmission.com; Thu, 13 Apr 2000 05:05:03 -0600
Received: from alan by the-village.bc.nu with local (Exim 2.12 #1)
id 12fhQ9-0002nD-00
for andersen@xmission.com; Thu, 13 Apr 2000 12:04:25 +0100
Subject: Re: kernel ps drivers [Was: vm locking question]
To: andersen@xmission.com
Date: Thu, 13 Apr 2000 12:04:23 +0100 (BST)
In-Reply-To: <20000412224130.A2748@xmission.com> from "Erik Andersen" at Apr 12, 2000 10:41:30 PM
X-Mailer: ELM [version 2.5 PL1]
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Message-Id: <E12fhQ9-0002nD-00@the-village.bc.nu>
From: Alan Cox <alan@lxorguk.ukuu.org.uk>
Status: RO
X-Status: A
Content-Length: 242
Lines: 6
> On the subject of ps, would you be willing to accept my /dev/ps
> patch into the kernel? If no, any suggestions on what should
> be done differently (if anything) to make it worthy of inclusion?
For 2.2.x no, for 2.3.x ask Linus not me
From torvalds@transmeta.com Thu Apr 13 09:18:16 2000
Return-Path: <torvalds@transmeta.com>
Delivered-To: andersen@dillweed.dsl.xmission.com
Received: from localhost (dillweed.dsl.xmission.com [10.0.0.1])
by dillweed.dsl.xmission.com (Postfix) with ESMTP id 3776411A3DF
for <andersen@dillweed.dsl.xmission.com>; Thu, 13 Apr 2000 09:18:16 -0600 (MDT)
Envelope-to: andersen@xmission.com
Received: from mail.xmission.com
by localhost with IMAP (fetchmail-5.3.3)
for andersen@dillweed.dsl.xmission.com (single-drop); Thu, 13 Apr 2000 09:18:16 -0600 (MDT)
Received: from [209.10.217.66] (helo=neon-gw.transmeta.com)
by mail.xmission.com with esmtp (Exim 3.03 #3)
id 12flK2-0004dd-00
for andersen@xmission.com; Thu, 13 Apr 2000 09:14:22 -0600
Received: (from root@localhost)
by neon-gw.transmeta.com (8.9.3/8.9.3) id IAA18635;
Thu, 13 Apr 2000 08:10:51 -0700
Received: from mailhost.transmeta.com(10.1.1.15) by neon-gw.transmeta.com via smap (V2.1)
id xma018629; Thu, 13 Apr 00 08:10:25 -0700
Received: from penguin.transmeta.com (root@penguin.transmeta.com [10.1.2.202])
by deepthought.transmeta.com (8.9.3/8.9.3) with ESMTP id IAA12264;
Thu, 13 Apr 2000 08:13:53 -0700 (PDT)
Received: from localhost (torvalds@localhost) by penguin.transmeta.com (8.9.3/8.7.3) with ESMTP id IAA02051; Thu, 13 Apr 2000 08:13:53 -0700
X-Authentication-Warning: penguin.transmeta.com: torvalds owned process doing -bs
Date: Thu, 13 Apr 2000 08:13:53 -0700 (PDT)
From: Linus Torvalds <torvalds@transmeta.com>
To: Erik Andersen <andersen@xmission.com>
Cc: Alan Cox <alan@redhat.com>
Subject: Re: kernel ps drivers [Was: vm locking question]
In-Reply-To: <20000413083127.A976@xmission.com>
Message-ID: <Pine.LNX.4.10.10004130812170.2000-100000@penguin.transmeta.com>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Status: RO
Content-Length: 659
Lines: 16
On Thu, 13 Apr 2000, Erik Andersen wrote:
>
> For 2.3.x would you be willing to accept my /dev/ps driver into the kernel?
> (Assuming I remove the /dev/modules driver (since it was pointed out that there
> is a perfectly good syscall providing that interface). If no, is there anything
> that could be done differently (if anything) to make it worthy of inclusion?
I do dislike /dev/ps mightily. If the problem is that /proc is too large,
then the right solution is to just clean up /proc. Which is getting done.
And yes, /proc will be larger than /dev/ps, but I still find that
preferable to having two incompatible ways to do the same thing.
Linus

File diff suppressed because it is too large Load Diff

View File

@ -27,88 +27,6 @@
#define READ_BUF_SIZE 50
/* For Erik's nifty devps device driver */
#ifdef CONFIG_FEATURE_USE_DEVPS_PATCH
#include <linux/devps.h>
/* find_pid_by_name()
*
* This finds the pid of the specified process,
* by using the /dev/ps device driver.
*
* Returns a list of all matching PIDs
*/
extern long* find_pid_by_name( const char* pidName)
{
int fd, i, j;
char device[] = "/dev/ps";
pid_t num_pids;
pid_t* pid_array = NULL;
long* pidList=NULL;
/* open device */
fd = open(device, O_RDONLY);
if (fd < 0)
perror_msg_and_die("open failed for `%s'", device);
/* Find out how many processes there are */
if (ioctl (fd, DEVPS_GET_NUM_PIDS, &num_pids)<0)
perror_msg_and_die("\nDEVPS_GET_PID_LIST");
/* Allocate some memory -- grab a few extras just in case
* some new processes start up while we wait. The kernel will
* just ignore any extras if we give it too many, and will trunc.
* the list if we give it too few. */
pid_array = (pid_t*) xcalloc( num_pids+10, sizeof(pid_t));
pid_array[0] = num_pids+10;
/* Now grab the pid list */
if (ioctl (fd, DEVPS_GET_PID_LIST, pid_array)<0)
perror_msg_and_die("\nDEVPS_GET_PID_LIST");
/* Now search for a match */
for (i=1, j=0; i<pid_array[0] ; i++) {
char* p;
struct pid_info info;
info.pid = pid_array[i];
if (ioctl (fd, DEVPS_GET_PID_INFO, &info)<0)
perror_msg_and_die("\nDEVPS_GET_PID_INFO");
/* Make sure we only match on the process name */
p=info.command_line+1;
while ((*p != 0) && !isspace(*(p)) && (*(p-1) != '\\')) {
(p)++;
}
if (isspace(*(p)))
*p='\0';
if ((strstr(info.command_line, pidName) != NULL)
&& (strlen(pidName) == strlen(info.command_line))) {
pidList=xrealloc( pidList, sizeof(long) * (j+2));
pidList[j++]=info.pid;
}
}
if (pidList) {
pidList[j]=0;
} else {
pidList=xrealloc( pidList, sizeof(long));
pidList[0]=-1;
}
/* Free memory */
free( pid_array);
/* close device */
if (close (fd) != 0)
perror_msg_and_die("close failed for `%s'", device);
return pidList;
}
#else /* CONFIG_FEATURE_USE_DEVPS_PATCH */
/* find_pid_by_name()
*
* Modified by Vladimir Oleynik for use with libbb/procps.c
@ -135,7 +53,6 @@ extern long* find_pid_by_name( const char* pidName)
pidList[i] = i==0 ? -1 : 0;
return pidList;
}
#endif /* CONFIG_FEATURE_USE_DEVPS_PATCH */
/* END CODE */
/*

View File

@ -23,17 +23,13 @@
#include "libbb.h"
/* Busybox mount uses either /proc/mounts or /dev/mtab to
/* Busybox mount uses either /proc/mounts or /etc/mtab to
* get the list of currently mounted filesystems */
#if defined CONFIG_FEATURE_MTAB_SUPPORT
const char mtab_file[] = "/etc/mtab";
#else
# if defined CONFIG_FEATURE_USE_DEVPS_PATCH
const char mtab_file[] = "/dev/mtab";
#else
const char mtab_file[] = "/proc/mounts";
#endif
#endif
/* END CODE */

View File

@ -8,7 +8,6 @@
*
*/
#if ! defined CONFIG_FEATURE_USE_DEVPS_PATCH
#include <dirent.h>
#include <string.h>
#include <stdlib.h>
@ -131,8 +130,6 @@ extern procps_status_t * procps_scan(int save_user_arg0)
}
}
#endif /* CONFIG_FEATURE_USE_DEVPS_PATCH. Else this file is empty */
/* END CODE */
/*
Local Variables:

View File

@ -20,12 +20,6 @@
* Place, Suite 330, Boston, MA 02111-1307 USA
*/
/*
* This contains _two_ implementations of ps for Linux. One uses the
* traditional /proc virtual filesystem, and the other use the devps kernel
* driver (written by Erik Andersen to avoid using /proc thereby saving 100k+).
*/
#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
@ -42,8 +36,6 @@ static const int TERMINAL_WIDTH = 79; /* not 80 in case terminal has linefo
#if ! defined CONFIG_FEATURE_USE_DEVPS_PATCH
extern int ps_main(int argc, char **argv)
{
procps_status_t * p;
@ -91,99 +83,3 @@ extern int ps_main(int argc, char **argv)
return EXIT_SUCCESS;
}
#else /* CONFIG_FEATURE_USE_DEVPS_PATCH */
/* The following is the second ps implementation --
* this one uses the nifty new devps kernel device.
*/
#include <linux/devps.h> /* For Erik's nifty devps device driver */
extern int ps_main(int argc, char **argv)
{
char device[] = "/dev/ps";
int i, j, len, fd;
pid_t num_pids;
pid_t* pid_array = NULL;
struct pid_info info;
char uidName[9];
#ifdef CONFIG_FEATURE_AUTOWIDTH
struct winsize win = { 0, 0, 0, 0 };
int terminal_width = TERMINAL_WIDTH;
#else
#define terminal_width TERMINAL_WIDTH
#endif
if (argc > 1 && **(argv + 1) == '-')
show_usage();
/* open device */
fd = open(device, O_RDONLY);
if (fd < 0)
perror_msg_and_die( "open failed for `%s'", device);
/* Find out how many processes there are */
if (ioctl (fd, DEVPS_GET_NUM_PIDS, &num_pids)<0)
perror_msg_and_die( "\nDEVPS_GET_PID_LIST");
/* Allocate some memory -- grab a few extras just in case
* some new processes start up while we wait. The kernel will
* just ignore any extras if we give it too many, and will trunc.
* the list if we give it too few. */
pid_array = (pid_t*) xcalloc( num_pids+10, sizeof(pid_t));
pid_array[0] = num_pids+10;
/* Now grab the pid list */
if (ioctl (fd, DEVPS_GET_PID_LIST, pid_array)<0)
perror_msg_and_die("\nDEVPS_GET_PID_LIST");
#ifdef CONFIG_FEATURE_AUTOWIDTH
ioctl(fileno(stdout), TIOCGWINSZ, &win);
if (win.ws_col > 0)
terminal_width = win.ws_col - 1;
#endif
/* Print up a ps listing */
printf(" PID Uid Stat Command\n");
for (i=1; i<pid_array[0] ; i++) {
info.pid = pid_array[i];
if (ioctl (fd, DEVPS_GET_PID_INFO, &info)<0)
perror_msg_and_die("\nDEVPS_GET_PID_INFO");
/* Make some adjustments as needed */
my_getpwuid(uidName, info.euid);
if(p.vmsize == 0)
len = printf("%5d %-8s %c ", p.pid, uidName, p.state);
else
len = printf("%5d %-8s %6d %c ", p.pid, uidName, p.vmsize, p.state);
if (strlen(info.command_line) > 1) {
for( j=0; j<(sizeof(info.command_line)-1) && j < (terminal_width-len); j++) {
if (*(info.command_line+j) == '\0' && *(info.command_line+j+1) != '\0') {
*(info.command_line+j) = ' ';
}
}
*(info.command_line+j) = '\0';
puts(info.command_line);
} else {
printf("[%s]\n", info.name);
}
}
/* Free memory */
free( pid_array);
/* close device */
if (close (fd) != 0)
perror_msg_and_die("close failed for `%s'", device);
exit (0);
}
#endif /* CONFIG_FEATURE_USE_DEVPS_PATCH */

View File

@ -53,9 +53,6 @@
#include <mntent.h>
#include <ctype.h>
#include "busybox.h"
#if defined CONFIG_FEATURE_USE_DEVPS_PATCH
# include <linux/devmtab.h> /* For Erik's nifty devmtab device driver */
#endif
enum {
MS_MGC_VAL = 0xc0ed0000, /* Magic number indicatng "new" flags */
@ -247,39 +244,6 @@ static int mount_one(char *blockDevice, char *directory, char *filesystemType,
int mount_all)
{
int status = 0;
#if defined CONFIG_FEATURE_USE_DEVPS_PATCH
if (strcmp(filesystemType, "auto") == 0) {
static const char *noauto_array[] =
{ "tmpfs", "shm", "proc", "ramfs", "devpts", "devfs", "usbdevfs",
0 };
const char **noauto_fstype;
const int num_of_filesystems = sysfs(3, 0, 0);
char buf[255];
int i = 0;
filesystemType = buf;
while (i < num_of_filesystems) {
sysfs(2, i++, filesystemType);
for (noauto_fstype = noauto_array; *noauto_fstype;
noauto_fstype++) {
if (!strcmp(filesystemType, *noauto_fstype)) {
break;
}
}
if (!*noauto_fstype) {
status =
do_mount(blockDevice, directory, filesystemType,
flags | MS_MGC_VAL, string_flags, useMtab,
fakeIt, mtab_opts, mount_all);
if (status) {
break;
}
}
}
}
#else
if (strcmp(filesystemType, "auto") == 0) {
char buf[255];
FILE *f;
@ -346,9 +310,7 @@ static int mount_one(char *blockDevice, char *directory, char *filesystemType,
}
}
fclose(f);
}
#endif
else {
} else {
status =
do_mount(blockDevice, directory, filesystemType,
flags | MS_MGC_VAL, string_flags, useMtab, fakeIt,
@ -366,47 +328,6 @@ static int mount_one(char *blockDevice, char *directory, char *filesystemType,
static void show_mounts(char *onlytype)
{
#if defined CONFIG_FEATURE_USE_DEVPS_PATCH
int fd, i, numfilesystems;
char device[] = "/dev/mtab";
struct k_mntent *mntentlist;
/* open device */
fd = open(device, O_RDONLY);
if (fd < 0) {
perror_msg_and_die("open failed for `%s'", device);
}
/* How many mounted filesystems? We need to know to
* allocate enough space for later... */
numfilesystems = ioctl(fd, DEVMTAB_COUNT_MOUNTS);
if (numfilesystems < 0) {
perror_msg_and_die("\nDEVMTAB_COUNT_MOUNTS");
}
mntentlist =
(struct k_mntent *) xcalloc(numfilesystems, sizeof(struct k_mntent));
/* Grab the list of mounted filesystems */
if (ioctl(fd, DEVMTAB_GET_MOUNTS, mntentlist) < 0) {
perror_msg_and_die("\nDEVMTAB_GET_MOUNTS");
}
for (i = 0; i < numfilesystems; i++) {
if (!onlytype || (strcmp(mntentlist[i].mnt_type, onlytype) == 0)) {
printf("%s %s %s %s %d %d\n", mntentlist[i].mnt_fsname,
mntentlist[i].mnt_dir, mntentlist[i].mnt_type,
mntentlist[i].mnt_opts, mntentlist[i].mnt_freq,
mntentlist[i].mnt_passno);
}
}
#ifdef CONFIG_FEATURE_CLEAN_UP
/* Don't bother to close files or free memory. Exit
* does that automagically, so we can save a few bytes */
free(mntentlist);
close(fd);
#endif
exit(EXIT_SUCCESS);
#else
FILE *mountTable = setmntent(mtab_file, "r");
if (mountTable) {
@ -433,7 +354,6 @@ static void show_mounts(char *onlytype)
perror_msg_and_die("%s", mtab_file);
}
exit(EXIT_SUCCESS);
#endif
}
extern int mount_main(int argc, char **argv)