|
From: Tom T. <tr...@un...> - 2005-03-11 16:42:19
|
valgrind-2.4.0.rc2 is quite nice but it (and rc1) sometimes glitches
if I resize the window after it starts running:
valgrind: vg_syscalls.c:6106 (vgPlain_client_syscall): Assertion `tst->sig_mask.sig[0] == tst->tmp_sig_mask.sig[0]' failed.
==9054== at 0xB002E2C9: ??? (vg_mylibc.c:1166)
==9054== by 0xB002E2C8: assert_fail (vg_mylibc.c:1166)
==9054== by 0xB002E307: vgPlain_core_assert_fail (vg_mylibc.c:1177)
==9054== by 0xB004C4AD: vgPlain_client_syscall (vg_syscalls.c:6194)
==9054== by 0xB00159B0: handle_syscall (vg_scheduler.c:632)
==9054== by 0xB0015C5A: vgPlain_scheduler (vg_scheduler.c:733)
==9054== by 0xB006EB3A: vgArch_thread_wrapper (core_os.c:69)
sched status:
running_tid=1
Thread 1: status = VgTs_Runnable
==9054== at 0x42028D69: sigsuspend (in /lib/i686/libc-2.2.93.so)
==9054== by 0x1B91D107: __pthread_wait_for_restart_signal (in /lib/i686/libpthread-0.10.so)
==9054== by 0x1B91A04A: pthread_cond_wait (in /lib/i686/libpthread-0.10.so)
==9054== by 0x1BAABF59: bktWait (/sas/dev/mva/TKLNX/src/TKLNXbktWait.c:110)
==9054== by 0x1BAAAD9B: sktWait (/sas/dev/mva/tkp/src/sktWait.c:124)
==9054== by 0x815EEFE: tkWait (/sas/dev/mva/tkp/src/tkwait.c:57)
==9054== by 0x8062A67: main (/sas/dev/mva/dsup/src/hamain.c:94)
Thread 2: status = VgTs_WaitSys
==9054== at 0x420D224B: poll (in /lib/i686/libc-2.2.93.so)
==9054== by 0x1B91AD9D: __pthread_manager (in /lib/i686/libpthread-0.10.so)
==9054== by 0x420DA1C9: clone (in /lib/i686/libc-2.2.93.so)
Thread 3: status = VgTs_Runnable
==9054== at 0x420D3B2E: select (in /lib/i686/libc-2.2.93.so)
==9054== by 0x1BAAC6F7: bktHandleChildProcess (/sas/dev/mva/TKPOS/src/TKPOSbktchsrv.c:198)
==9054== by 0x1BAAA4EE: sktMain (/sas/dev/mva/tkp/src/sktMain.c:95)
==9054== by 0x1BAABDC8: bktMain (/sas/dev/mva/TKPOS/src/TKPOSbktMain.c:97)
==9054== by 0x1B91B940: pthread_start_thread (in /lib/i686/libpthread-0.10.so)
==9054== by 0x420DA1C9: clone (in /lib/i686/libc-2.2.93.so)
Thread 4: status = VgTs_Runnable
==9054== at 0x420CDA84: open (in /lib/i686/libc-2.2.93.so)
As a possible clue, I noticed that if one runs the following program
and then changes the window size, it remains pause()d when run standalone
but terminates when run under valgrind.
#include <unistd.h>
int
main()
{
pause();
return 0;
}
Tom Truscott
|
|
From: Jeremy F. <je...@go...> - 2005-03-11 17:02:39
|
Tom Truscott wrote:
>valgrind-2.4.0.rc2 is quite nice but it (and rc1) sometimes glitches
>if I resize the window after it starts running:
>
> valgrind: vg_syscalls.c:6106 (vgPlain_client_syscall): Assertion `tst->sig_mask.sig[0] == tst->tmp_sig_mask.sig[0]' failed.
> ==9054== at 0xB002E2C9: ??? (vg_mylibc.c:1166)
> ==9054== by 0xB002E2C8: assert_fail (vg_mylibc.c:1166)
> ==9054== by 0xB002E307: vgPlain_core_assert_fail (vg_mylibc.c:1177)
> ==9054== by 0xB004C4AD: vgPlain_client_syscall (vg_syscalls.c:6194)
> ==9054== by 0xB00159B0: handle_syscall (vg_scheduler.c:632)
> ==9054== by 0xB0015C5A: vgPlain_scheduler (vg_scheduler.c:733)
> ==9054== by 0xB006EB3A: vgArch_thread_wrapper (core_os.c:69)
>
> sched status:
> running_tid=1
>
> Thread 1: status = VgTs_Runnable
> ==9054== at 0x42028D69: sigsuspend (in /lib/i686/libc-2.2.93.so)
> ==9054== by 0x1B91D107: __pthread_wait_for_restart_signal (in /lib/i686/libpthread-0.10.so)
> ==9054== by 0x1B91A04A: pthread_cond_wait (in /lib/i686/libpthread-0.10.so)
> ==9054== by 0x1BAABF59: bktWait (/sas/dev/mva/TKLNX/src/TKLNXbktWait.c:110)
> ==9054== by 0x1BAAAD9B: sktWait (/sas/dev/mva/tkp/src/sktWait.c:124)
> ==9054== by 0x815EEFE: tkWait (/sas/dev/mva/tkp/src/tkwait.c:57)
> ==9054== by 0x8062A67: main (/sas/dev/mva/dsup/src/hamain.c:94)
>
>
> Thread 2: status = VgTs_WaitSys
> ==9054== at 0x420D224B: poll (in /lib/i686/libc-2.2.93.so)
> ==9054== by 0x1B91AD9D: __pthread_manager (in /lib/i686/libpthread-0.10.so)
> ==9054== by 0x420DA1C9: clone (in /lib/i686/libc-2.2.93.so)
>
> Thread 3: status = VgTs_Runnable
> ==9054== at 0x420D3B2E: select (in /lib/i686/libc-2.2.93.so)
> ==9054== by 0x1BAAC6F7: bktHandleChildProcess (/sas/dev/mva/TKPOS/src/TKPOSbktchsrv.c:198)
> ==9054== by 0x1BAAA4EE: sktMain (/sas/dev/mva/tkp/src/sktMain.c:95)
> ==9054== by 0x1BAABDC8: bktMain (/sas/dev/mva/TKPOS/src/TKPOSbktMain.c:97)
> ==9054== by 0x1B91B940: pthread_start_thread (in /lib/i686/libpthread-0.10.so)
> ==9054== by 0x420DA1C9: clone (in /lib/i686/libc-2.2.93.so)
>
>
> Thread 4: status = VgTs_Runnable
> ==9054== at 0x420CDA84: open (in /lib/i686/libc-2.2.93.so)
>
>As a possible clue, I noticed that if one runs the following program
>and then changes the window size, it remains pause()d when run standalone
>but terminates when run under valgrind.
>
That's possible. The symptom you're seeing isn't directly related to
that, but it could certainly be indirectly related. Could you file a
bug for this?
Thanks,
J
|
|
From: Tom T. <tr...@un...> - 2005-03-11 21:28:25
|
The Assertion is on an old Red Hat system with an old pthreads library, and does not happen on newer ones. So forget that. I did (try to) submit a bug report for the interrupted-pause() quirk, since that still happens on newer systems. Tom Truscott |
|
From: Tom H. <to...@co...> - 2005-03-11 23:05:42
|
In message <200...@lo...>
Tom Truscott <tr...@un...> wrote:
> I did (try to) submit a bug report for the interrupted-pause() quirk,
> since that still happens on newer systems.
The bug came through a few hours ago.
Tom
--
Tom Hughes (to...@co...)
http://www.compton.nu/
|
|
From: Jeremy F. <je...@go...> - 2005-03-12 00:40:20
|
Tom Truscott wrote:
>valgrind-2.4.0.rc2 is quite nice but it (and rc1) sometimes glitches
>if I resize the window after it starts running:
>
> valgrind: vg_syscalls.c:6106 (vgPlain_client_syscall): Assertion `tst->sig_mask.sig[0] == tst->tmp_sig_mask.sig[0]' failed.
> ==9054== at 0xB002E2C9: ??? (vg_mylibc.c:1166)
> ==9054== by 0xB002E2C8: assert_fail (vg_mylibc.c:1166)
> ==9054== by 0xB002E307: vgPlain_core_assert_fail (vg_mylibc.c:1177)
> ==9054== by 0xB004C4AD: vgPlain_client_syscall (vg_syscalls.c:6194)
> ==9054== by 0xB00159B0: handle_syscall (vg_scheduler.c:632)
> ==9054== by 0xB0015C5A: vgPlain_scheduler (vg_scheduler.c:733)
> ==9054== by 0xB006EB3A: vgArch_thread_wrapper (core_os.c:69)
>[...]
>As a possible clue, I noticed that if one runs the following program
>and then changes the window size, it remains pause()d when run standalone
>but terminates when run under valgrind.
>
>
I've fixed the unexpected interrupted syscall problem, but the assertion
failure is still a concern; it shouldn't have happened either way. Can
you reproduce it with CVS HEAD?
J
|