signal: multithread program returns with wrong errno on receiving SIGSTOP
Hi,
I found a bug on signal subsystem. If there is some multithread program and one of the thread is blocking on the system call, it returns with wrong errno on receiving SIGSTOP and following SIGCONT.
Arch dependency
===============
succeed to reproduce: i386, ia64 Unknown: any other arch
# I suspect that this problem occur on any arch...
How to reproduce
================
This process needs 2 terminals, term A and term B.
1. issue the test program (attached on this mail) from term A.
$ ./mt-wrong-errno
2. send SIGSTOP to test program from term B.
$ ps a
...
9685 pts/3 Sl+ 0:00 ./mt-wrong-errno
9688 pts/4 R+ 0:00 ps a
$ kill -STOP 96853. send SIGCONT to test program from term B.
$ kill -CONT 9685
Expected Result
===============
`./mt-wrong-errno' restarts read syscall (and stop again by receiving SIGTTIN).
Actual Result
=============
`./mt-wrong-errno' returns with wrong errno 512 means ERESTARTSYS. The comment on include/linux/errno.h says "this errno should never be seen by user program." So it's definitely a bug.
output of term A
----------------
$ ./mt-wrong-errno
[1]+ Stopped ./mt-wrong-errno
$ errno = 512
output of term B
----------------
$ ps a
...
9685 pts/3 Sl+ 0:00 ./mt-wrong-errno
9688 pts/4 R+ 0:00 ps a
$ kill -STOP 9685
$ kill -CONT 9685This problem can always reproduce, so it's no timing issue. I suspect that group_stop_count counting mechanism has something wrong and signal handling is canceled erroneously.
Any hints or patch itself are welcome.
Thanks,
Satoru
-------------------------------------------------------------------------------
/*
* mt-wrong-errno.c
*
* Copyright 2007 Satoru Takeuchi <[email protected]> *
* This software may be used and distributed according to the terms
* of the GNU General Public License, incorporated herein by reference. *
*/
#include <sys/types.h>
#include <pthread.h>
#include <stdio.h>
#include <stdlib.h>
#include <err.h>
#include <unistd.h>
#include <stdio.h>
#include <errno.h>
void *thread_fn(void *arg)
{
char c;
int ret;
ret = read(STDIN_FILENO, &c, 1);
if (ret < 0)
printf("errno = %d\n", errno);
return NULL;
}int main(int argc, char *argv[])
{
pthread_t t;
if (pthread_create(&t, NULL, thread_fn, NULL))
err(EXIT_FAILURE, "pthread_create() failed\n");
if (pthread_join(t, NULL))
warn("pthread_join() failed\n");
}Quoted 110 lines · On Mon, 28 May 2007 11:32:00 +0900 Satoru Takeuchi…
Hi,
I found a bug on signal subsystem. If there is some multithread program and one of the thread is blocking on the system call, it returns with wrong errno on receiving SIGSTOP and following SIGCONT.
Arch dependency
===============succeed to reproduce: i386, ia64 Unknown: any other arch# I suspect that this problem occur on any arch...
How to reproduce
================This process needs 2 terminals, term A and term B.
1. issue the test program (attached on this mail) from term A.
$ ./mt-wrong-errno2. send SIGSTOP to test program from term B.
$ ps a ... 9685 pts/3 Sl+ 0:00 ./mt-wrong-errno 9688 pts/4 R+ 0:00 ps a $ kill -STOP 96853. send SIGCONT to test program from term B.
$ kill -CONT 9685Expected Result
===============`./mt-wrong-errno' restarts read syscall (and stop again by receiving SIGTTIN).
Actual Result
=============`./mt-wrong-errno' returns with wrong errno 512 means ERESTARTSYS. The comment on
include/linux/errno.hsays "this errno should never be seen by user program." So it's definitely a bug.output of term A
----------------$ ./mt-wrong-errno[1]+ Stopped ./mt-wrong-errno
$ errno = 512output of term B
----------------$ ps a ... 9685 pts/3 Sl+ 0:00 ./mt-wrong-errno 9688 pts/4 R+ 0:00 ps a $ kill -STOP 9685 $ kill -CONT 9685This problem can always reproduce, so it's no timing issue. I suspect that
group_stop_countcounting mechanism has something wrong and signal handling is canceled erroneously.Any hints or patch itself are welcome.
Thanks,
Satoru
-------------------------------------------------------------------------------
/*
*mt-wrong-errno.c
*
* Copyright 2007 Satoru Takeuchi <[email protected]> *
* This software may be used and distributed according to the terms
* of the GNU General Public License, incorporated herein by reference. *
*/#include <
sys/types.h>
#include <pthread.h>
#include <stdio.h>
#include <stdlib.h>
#include <err.h>
#include <unistd.h>
#include <stdio.h>
#include <errno.h>void *thread_fn(void *arg)
{ char c; int ret; ret = read(STDIN_FILENO, &c, 1); if (ret < 0) printf("errno = %d\n", errno); return NULL; }int
main(int argc, char *argv[])
Quoted 9 lines
{
pthread_t t;
if (pthread_create(&t, NULL, thread_fn, NULL))
err(EXIT_FAILURE, "pthread_create() failed\n");
if (pthread_join(t, NULL))
warn("pthread_join() failed\n");
}What kernel versions is this occurring on?
Off hand I think it can happen on all 2.6 kernels in theory.
Hi Andrew,
Quoted 1 line
What kernel versions is this occurring on?
2.6.22-rc3 is.
Satoru
At Tue, 29 May 2007 01:03:15 -0700,
Quoted 130 lines · Andrew Morton
Quoted 110 lines · On Mon, 28 May 2007 11:32:00 +0900 Satoru Takeuchi…
Hi,
I found a bug on signal subsystem. If there is some multithread program and one of the thread is blocking on the system call, it returns with wrong errno on receiving SIGSTOP and following SIGCONT.
Arch dependency
===============succeed to reproduce: i386, ia64 Unknown: any other arch# I suspect that this problem occur on any arch...
How to reproduce
================This process needs 2 terminals, term A and term B.
1. issue the test program (attached on this mail) from term A.
$ ./mt-wrong-errno2. send SIGSTOP to test program from term B.
$ ps a ... 9685 pts/3 Sl+ 0:00 ./mt-wrong-errno 9688 pts/4 R+ 0:00 ps a $ kill -STOP 96853. send SIGCONT to test program from term B.
$ kill -CONT 9685Expected Result
===============`./mt-wrong-errno' restarts read syscall (and stop again by receiving SIGTTIN).
Actual Result
=============`./mt-wrong-errno' returns with wrong errno 512 means ERESTARTSYS. The comment on
include/linux/errno.hsays "this errno should never be seen by user program." So it's definitely a bug.output of term A
----------------$ ./mt-wrong-errno[1]+ Stopped ./mt-wrong-errno
$ errno = 512output of term B
----------------$ ps a ... 9685 pts/3 Sl+ 0:00 ./mt-wrong-errno 9688 pts/4 R+ 0:00 ps a $ kill -STOP 9685 $ kill -CONT 9685This problem can always reproduce, so it's no timing issue. I suspect that
group_stop_countcounting mechanism has something wrong and signal handling is canceled erroneously.Any hints or patch itself are welcome.
Thanks,
Satoru
-------------------------------------------------------------------------------
/*
*mt-wrong-errno.c
*
* Copyright 2007 Satoru Takeuchi <[email protected]> *
* This software may be used and distributed according to the terms
* of the GNU General Public License, incorporated herein by reference. *
*/#include <
sys/types.h>
#include <pthread.h>
#include <stdio.h>
#include <stdlib.h>
#include <err.h>
#include <unistd.h>
#include <stdio.h>
#include <errno.h>void *thread_fn(void *arg)
{ char c; int ret; ret = read(STDIN_FILENO, &c, 1); if (ret < 0) printf("errno = %d\n", errno); return NULL; }int
main(int argc, char *argv[])Quoted 9 lines
{ pthread_t t; if (pthread_create(&t, NULL, thread_fn, NULL)) err(EXIT_FAILURE, "pthread_create() failed\n"); if (pthread_join(t, NULL)) warn("pthread_join() failed\n"); }What kernel versions is this occurring on?
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/