[valgrind] [Bug 361615] Inconsistent termination when an instrumented multithreaded process is terminated by signal

[via KDE Bugzilla]

https://bugs.kde.org/show_bug.cgi?id=361615

--- Comment #13 from earl_c...@yahoo.com ---
Created attachment 101149
  --> https://bugs.kde.org/attachment.cgi?id=101149=edit
Standalone test case

I could not find the previously mentioned standalone test case, so I've
uploaded a reconstructed version that illustrates the same behaviour.

Note also that I have included a candidate patch with this ticket:
https://bugs.kde.org/attachment.cgi?id=98326

When run outside valgrind, the test program prints:
---
$ gcc -o test test.c -lpthread && ./test ; echo $?
Signal 15
0
---

When run under valgrind, the result is different:
---
$ gcc -o test test.c -lpthread && valgrind ./test ; echo $?
==3351== Memcheck, a memory error detector
==3351== Copyright (C) 2002-2015, and GNU GPL'd, by Julian Seward et al.
==3351== Using Valgrind-3.11.0 and LibVEX; rerun with -h for copyright info
==3351== Command: ./test
==3351== 
==3352== 
==3352== Process terminating with default action of signal 15 (SIGTERM)
==3352==at 0x405DF3F: pthread_sigmask (pthread_sigmask.c:48)
==3352==by 0x804875E: slavethread (in /home/earl/test)
==3352==by 0x4058F71: start_thread (pthread_create.c:312)
==3352==by 0x415AF8D: clone (clone.S:129)
==3352== 
==3352== HEAP SUMMARY:
==3352== in use at exit: 136 bytes in 1 blocks
==3352==   total heap usage: 1 allocs, 0 frees, 136 bytes allocated
==3352== 
==3352== LEAK SUMMARY:
==3352==definitely lost: 0 bytes in 0 blocks
==3352==indirectly lost: 0 bytes in 0 blocks
==3352==  possibly lost: 136 bytes in 1 blocks
==3352==still reachable: 0 bytes in 0 blocks
==3352== suppressed: 0 bytes in 0 blocks
==3352== Rerun with --leak-check=full to see details of leaked memory
==3352== 
==3352== For counts of detected and suppressed errors, rerun with: -v
==3352== ERROR SUMMARY: 0 errors from 0 contexts (suppressed: 0 from 0)
Signal 9
test: test.c:88: main: Assertion `(((__extension__ (((union { __typeof(status)
__in; int __i; }) { .__in = (status) }).__i))) & 0x7f) == 15' failed.
==3351== 
==3351== Process terminating with default action of signal 6 (SIGABRT)
==3351==at 0x409D687: raise (raise.c:56)
==3351==by 0x40A0AB2: abort (abort.c:89)
==3351==by 0x40967C6: __assert_fail_base (assert.c:92)
==3351==by 0x4096876: __assert_fail (assert.c:101)
==3351==by 0x80489E5: main (in /home/earl/test)
==3351== 
==3351== HEAP SUMMARY:
==3351== in use at exit: 0 bytes in 0 blocks
==3351==   total heap usage: 3 allocs, 3 frees, 550 bytes allocated
==3351== 
==3351== All heap blocks were freed -- no leaks are possible
==3351== 
==3351== For counts of detected and suppressed errors, rerun with: -v
==3351== ERROR SUMMARY: 0 errors from 0 contexts (suppressed: 0 from 0)
Aborted
134
---

-- 
You are receiving this mail because:
You are watching all bug changes.

[valgrind] [Bug 361615] Inconsistent termination when an instrumented multithreaded process is terminated by signal

[Ben Cooksley via KDE Bugzilla]

https://bugs.kde.org/show_bug.cgi?id=361615

--- Comment #12 from Ben Cooksley  ---
The content of attachment 100145 has been deleted for the following reason:

Removing non-relevant content

-- 
You are receiving this mail because:
You are watching all bug changes.

[valgrind] [Bug 361615] Inconsistent termination when an instrumented multithreaded process is terminated by signal

[via KDE Bugzilla]

https://bugs.kde.org/show_bug.cgi?id=361615

--- Comment #11 from earl_c...@yahoo.com ---
(In reply to Ivo Raisr from comment #10)
> As Philippe pointed out, there is a problem with the existing attachment.
> Please have a look what is there inside and re-attach.

Yes, that's obviously wrong. Let me resurrect the intended content.

-- 
You are receiving this mail because:
You are watching all bug changes.

[valgrind] [Bug 361615] Inconsistent termination when an instrumented multithreaded process is terminated by signal

[Ivo Raisr via KDE Bugzilla]

https://bugs.kde.org/show_bug.cgi?id=361615

--- Comment #10 from Ivo Raisr  ---
As Philippe pointed out, there is a problem with the existing attachment.
Please have a look what is there inside and re-attach.

-- 
You are receiving this mail because:
You are watching all bug changes.

[valgrind] [Bug 361615] Inconsistent termination when an instrumented multithreaded process is terminated by signal

[via KDE Bugzilla]

https://bugs.kde.org/show_bug.cgi?id=361615

--- Comment #9 from earl_c...@yahoo.com ---
Comment on attachment 100145
  --> https://bugs.kde.org/attachment.cgi?id=100145
Self contained test case

.

-- 
You are receiving this mail because:
You are watching all bug changes.

[valgrind] [Bug 361615] Inconsistent termination when an instrumented multithreaded process is terminated by signal

[Philippe Waroquiers via KDE Bugzilla]

https://bugs.kde.org/show_bug.cgi?id=361615

--- Comment #8 from Philippe Waroquiers  ---
(In reply to earl_chew from comment #4)
> Created attachment 100145 [details]
> Self contained test case

It seems there was an attachment problem.
Could you re-attach the self contained test case ?
Thanks

-- 
You are receiving this mail because:
You are watching all bug changes.

[valgrind] [Bug 361615] Inconsistent termination when an instrumented multithreaded process is terminated by signal

[Philippe Waroquiers via KDE Bugzilla]

https://bugs.kde.org/show_bug.cgi?id=361615

--- Comment #7 from Philippe Waroquiers  ---
(In reply to Julian Seward from comment #6)
> Philippe, didn't you fix something like this recently?
Not recently.
Related to thread termination, I did something some years ago, to fix bug
324227.
I can take a look at this bug and see if I find something.

-- 
You are receiving this mail because:
You are watching all bug changes.

[valgrind] [Bug 361615] Inconsistent termination when an instrumented multithreaded process is terminated by signal

[Julian Seward via KDE Bugzilla]

https://bugs.kde.org/show_bug.cgi?id=361615

Julian Seward  changed:

   What|Removed |Added

 CC||philippe.waroquiers@skynet.
   ||be

--- Comment #6 from Julian Seward  ---
Philippe, didn't you fix something like this recently?

-- 
You are receiving this mail because:
You are watching all bug changes.

[valgrind] [Bug 361615] Inconsistent termination when an instrumented multithreaded process is terminated by signal

[via KDE Bugzilla]

https://bugs.kde.org/show_bug.cgi?id=361615

--- Comment #5 from earl_c...@yahoo.com ---
I compiled and ran the submitted test case both with and without valgrind. This
shows that under valgrind, the child process terminates with SIGKILL rather
than the expected SIGTERM.

$ gcc -o test test.c -lpthread && ./test
Signalled with 15

$ gcc -o test test.c -lpthread && valgrind ./test
==14625== Memcheck, a memory error detector
==14625== Copyright (C) 2002-2015, and GNU GPL'd, by Julian Seward et al.
==14625== Using Valgrind-3.11.0 and LibVEX; rerun with -h for copyright info
==14625== Command: ./test
==14625== 
==14626== 
==14626== Process terminating with default action of signal 15 (SIGTERM)
==14626==at 0x4124B96: ??? (syscall-template.S:81)
==14626==by 0x412493C: sleep (sleep.c:137)
==14626==by 0x804868E: slavethread (in /tmp/test)
==14626==by 0x4058F71: start_thread (pthread_create.c:312)
==14626==by 0x415AF8D: clone (clone.S:129)
==14626== 
==14626== HEAP SUMMARY:
==14626== in use at exit: 136 bytes in 1 blocks
==14626==   total heap usage: 1 allocs, 0 frees, 136 bytes allocated
==14626== 
==14626== LEAK SUMMARY:
==14626==definitely lost: 0 bytes in 0 blocks
==14626==indirectly lost: 0 bytes in 0 blocks
==14626==  possibly lost: 136 bytes in 1 blocks
==14626==still reachable: 0 bytes in 0 blocks
==14626== suppressed: 0 bytes in 0 blocks
==14626== Rerun with --leak-check=full to see details of leaked memory
==14626== 
==14626== For counts of detected and suppressed errors, rerun with: -v
==14626== ERROR SUMMARY: 0 errors from 0 contexts (suppressed: 0 from 0)
Signalled with 9
test: test.c:52: main: Assertion `15 == (((__extension__ (((union {
__typeof(status) __in; int __i; }) { .__in = (status) }).__i))) & 0x7f)'
failed.
==14625== 
==14625== Process terminating with default action of signal 6 (SIGABRT)
==14625==at 0x409D687: raise (raise.c:56)
==14625==by 0x40A0AB2: abort (abort.c:89)
==14625==by 0x40967C6: __assert_fail_base (assert.c:92)
==14625==by 0x4096876: __assert_fail (assert.c:101)
==14625==by 0x8048816: main (in /tmp/test)
==14625== 
==14625== HEAP SUMMARY:
==14625== in use at exit: 0 bytes in 0 blocks
==14625==   total heap usage: 3 allocs, 3 frees, 550 bytes allocated
==14625== 
==14625== All heap blocks were freed -- no leaks are possible
==14625== 
==14625== For counts of detected and suppressed errors, rerun with: -v
==14625== ERROR SUMMARY: 0 errors from 0 contexts (suppressed: 0 from 0)
Aborted

-- 
You are receiving this mail because:
You are watching all bug changes.

[valgrind] [Bug 361615] Inconsistent termination when an instrumented multithreaded process is terminated by signal

[via KDE Bugzilla]

https://bugs.kde.org/show_bug.cgi?id=361615

--- Comment #4 from earl_c...@yahoo.com ---
Created attachment 100145
  --> https://bugs.kde.org/attachment.cgi?id=100145=edit
Self contained test case

-- 
You are receiving this mail because:
You are watching all bug changes.

[valgrind] [Bug 361615] Inconsistent termination when an instrumented multithreaded process is terminated by signal

[via KDE Bugzilla]

https://bugs.kde.org/show_bug.cgi?id=361615

--- Comment #3 from earl_c...@yahoo.com ---
(In reply to Julian Seward from comment #2)
> Do you have a small test case which demonstrates that Valgrind's behaviour at
> present, differs from when the program is run "natively" ?

In the bug description, I pasted a small C program and accompanying shell
script that demonstrates the difference between a natively run program, and the
same running under valgrind.

Are you looking for something different ?

-- 
You are receiving this mail because:
You are watching all bug changes.

[valgrind] [Bug 361615] Inconsistent termination when an instrumented multithreaded process is terminated by signal

[Julian Seward via KDE Bugzilla]

https://bugs.kde.org/show_bug.cgi?id=361615

--- Comment #2 from Julian Seward  ---
Do you have a small test case which demonstrates that Valgrind's behaviour at
present, differs from when the program is run "natively" ?  I'd be happier
about
this if you have something directly that demonstrates that V's behaviour
differs
from that of the kernel, in this respect.

-- 
You are receiving this mail because:
You are watching all bug changes.

[valgrind] [Bug 361615] Inconsistent termination when an instrumented multithreaded process is terminated by signal

[Ivo Raisr via KDE Bugzilla]

https://bugs.kde.org/show_bug.cgi?id=361615

Ivo Raisr  changed:

   What|Removed |Added

 CC||iv...@ivosh.net

-- 
You are receiving this mail because:
You are watching all bug changes.

[valgrind] [Bug 361615] Inconsistent termination when an instrumented multithreaded process is terminated by signal

[via KDE Bugzilla]

https://bugs.kde.org/show_bug.cgi?id=361615

--- Comment #1 from earl_c...@yahoo.com ---
Created attachment 98326
  --> https://bugs.kde.org/attachment.cgi?id=98326=edit
Candidate patch to propagate termination signal

Here is an attempt to propagate the signal that terminates a thread into the
entire program.

-- 
You are receiving this mail because:
You are watching all bug changes.