|
From: Christian B. <bor...@de...> - 2011-08-09 20:12:00
Attachments:
diffs.gz
|
Nightly build on fedora390 ( Fedora 13/14/15 mix with gcc 3.5.3 on z196 (s390x) ) Started at 2011-08-09 21:45:01 CEST Ended at 2011-08-09 22:12:09 CEST Results differ from 24 hours ago Checking out valgrind source tree ... done Configuring valgrind ... done Building valgrind ... done Running regression tests ... failed Regression test results follow == 495 tests, 17 stderr failures, 0 stdout failures, 0 stderrB failures, 0 stdoutB failures, 0 post failures == memcheck/tests/err_disable3 (stderr) memcheck/tests/err_disable4 (stderr) memcheck/tests/linux/timerfd-syscall (stderr) none/tests/s390x/op00 (stderr) helgrind/tests/locked_vs_unlocked1_fwd (stderr) helgrind/tests/locked_vs_unlocked1_rev (stderr) helgrind/tests/locked_vs_unlocked2 (stderr) helgrind/tests/locked_vs_unlocked3 (stderr) helgrind/tests/pth_barrier3 (stderr) helgrind/tests/tc06_two_races_xml (stderr) helgrind/tests/tc17_sembar (stderr) helgrind/tests/tc18_semabuse (stderr) helgrind/tests/tc20_verifywrap (stderr) helgrind/tests/tc23_bogus_condwait (stderr) drd/tests/tc04_free_lock (stderr) drd/tests/tc09_bad_unlock (stderr) drd/tests/tc23_bogus_condwait (stderr) ================================================= == Results from 24 hours ago == ================================================= Checking out valgrind source tree ... done Configuring valgrind ... done Building valgrind ... done Running regression tests ... failed Regression test results follow == 494 tests, 17 stderr failures, 0 stdout failures, 0 stderrB failures, 0 stdoutB failures, 0 post failures == memcheck/tests/err_disable3 (stderr) memcheck/tests/err_disable4 (stderr) memcheck/tests/linux/timerfd-syscall (stderr) none/tests/s390x/op_exception (stderr) helgrind/tests/locked_vs_unlocked1_fwd (stderr) helgrind/tests/locked_vs_unlocked1_rev (stderr) helgrind/tests/locked_vs_unlocked2 (stderr) helgrind/tests/locked_vs_unlocked3 (stderr) helgrind/tests/pth_barrier3 (stderr) helgrind/tests/tc06_two_races_xml (stderr) helgrind/tests/tc18_semabuse (stderr) helgrind/tests/tc20_verifywrap (stderr) helgrind/tests/tc23_bogus_condwait (stderr) drd/tests/hg05_race2 (stderr) drd/tests/tc04_free_lock (stderr) drd/tests/tc09_bad_unlock (stderr) drd/tests/tc23_bogus_condwait (stderr) ================================================= == Difference between 24 hours ago and now == ================================================= *** old.short Tue Aug 9 21:59:09 2011 --- new.short Tue Aug 9 22:12:09 2011 *************** *** 8,10 **** ! == 494 tests, 17 stderr failures, 0 stdout failures, 0 stderrB failures, 0 stdoutB failures, 0 post failures == memcheck/tests/err_disable3 (stderr) --- 8,10 ---- ! == 495 tests, 17 stderr failures, 0 stdout failures, 0 stderrB failures, 0 stdoutB failures, 0 post failures == memcheck/tests/err_disable3 (stderr) *************** *** 12,14 **** memcheck/tests/linux/timerfd-syscall (stderr) ! none/tests/s390x/op_exception (stderr) helgrind/tests/locked_vs_unlocked1_fwd (stderr) --- 12,14 ---- memcheck/tests/linux/timerfd-syscall (stderr) ! none/tests/s390x/op00 (stderr) helgrind/tests/locked_vs_unlocked1_fwd (stderr) *************** *** 19,20 **** --- 19,21 ---- helgrind/tests/tc06_two_races_xml (stderr) + helgrind/tests/tc17_sembar (stderr) helgrind/tests/tc18_semabuse (stderr) *************** *** 22,24 **** helgrind/tests/tc23_bogus_condwait (stderr) - drd/tests/hg05_race2 (stderr) drd/tests/tc04_free_lock (stderr) --- 23,24 ---- |
|
From: Bart V. A. <bva...@ac...> - 2011-08-09 20:19:32
|
On Tue, Aug 9, 2011 at 10:11 PM, Christian Borntraeger <bor...@de...> wrote: > Nightly build on fedora390 ( Fedora 13/14/15 mix with gcc 3.5.3 on z196 (s390x) ) It surprised me to find this in diffs.gz: +error 13 Permission denied +cannot create shared_mem file /tmp/vgdb-pipe-shared-mem-vgdb-13786 Bart. |
|
From: Christian B. <bor...@de...> - 2011-08-10 10:58:34
|
On 09/08/11 22:19, Bart Van Assche wrote: > On Tue, Aug 9, 2011 at 10:11 PM, Christian Borntraeger > <bor...@de...> wrote: >> Nightly build on fedora390 ( Fedora 13/14/15 mix with gcc 3.5.3 on z196 (s390x) ) > > It surprised me to find this in diffs.gz: > > +error 13 Permission denied > +cannot create shared_mem file /tmp/vgdb-pipe-shared-mem-vgdb-13786 Hmm, dont know. thanks for noticing. Maybe the shm file system is not mounted or this test requires root? I am away until September, will look into that then. Christian |
|
From: Philippe W. <phi...@sk...> - 2011-08-12 20:40:18
|
>> It surprised me to find this in diffs.gz: >> >> +error 13 Permission denied >> +cannot create shared_mem file /tmp/vgdb-pipe-shared-mem-vgdb-13786 > > Hmm, dont know. thanks for noticing. > Maybe the shm file system is not mounted or this test requires root? > I am away until September, will look into that then. My guess is the following: if Valgrind terminates abnormally (e.g. due to a fatal internal error or due to a kill -9 or ...), then the vgdb FIFO and shmem files cannot be cleaned up. If such an abnormal termination is done by a Valgrind running under user xxx with pid yyy, then sometime later, it could be that another Valgrind is launched by user zzz with a "re-cycled" pid yyy. The file will then not be writeable by user zzz, giving the error 13. After running the Valgrind regression tests, there should be no vgdb files remaining (the test none/tests/require-text-symbol-2.vgtest by design terminates abnormally, but the test cleanup removes the vgdb files). Maybe on S390X, there are some other tests terminating abrnomally, leaving vgdb files behind ? A possible solution to avoid such error 13 is to add the user id in the FIFO and shmem file names e.g. /tmp/vgdb-pipe-shared-mem-vgdb-<here the user id>-<pid>. Then there should be no clash anymore. Philippe |
|
From: Bart V. A. <bva...@ac...> - 2011-08-13 05:14:24
|
On Fri, Aug 12, 2011 at 10:40 PM, Philippe Waroquiers < phi...@sk...> wrote: > It surprised me to find this in diffs.gz: >>> >>> +error 13 Permission denied >>> +cannot create shared_mem file /tmp/vgdb-pipe-shared-mem-**vgdb-13786 >>> >> >> Hmm, dont know. thanks for noticing. >> Maybe the shm file system is not mounted or this test requires root? I am >> away until September, will look into that then. >> > > My guess is the following: if Valgrind terminates abnormally (e.g. due to a > fatal internal > error or due to a kill -9 or ...), then the vgdb FIFO and shmem files > cannot be cleaned up. > If such an abnormal termination is done by a Valgrind running under user > xxx with pid yyy, > then sometime later, it could be that another Valgrind is launched by user > zzz with a "re-cycled" pid yyy. > The file will then not be writeable by user zzz, giving the error 13. > After running the Valgrind regression tests, there should be no vgdb files > remaining > (the test none/tests/require-text-**symbol-2.vgtest by design terminates > abnormally, > but the test cleanup removes the vgdb files). > > Maybe on S390X, there are some other tests terminating abrnomally, leaving > vgdb > files behind ? > > A possible solution to avoid such error 13 is to add the user id in the > FIFO and shmem > file names e.g. /tmp/vgdb-pipe-shared-mem-**vgdb-<here the user id>-<pid>. > Then there should be no clash anymore. > Hello Philippe, Do you think it would be sufficient to invoke unlink() on vgdb pipe names before invoking open_fifo() in coregrind/vgdb.c ? Bart. |
|
From: Florian K. <br...@ac...> - 2011-08-24 17:12:46
|
On 08/12/2011 04:40 PM, Philippe Waroquiers wrote:
>>> It surprised me to find this in diffs.gz:
>>>
>>> +error 13 Permission denied
>>> +cannot create shared_mem file /tmp/vgdb-pipe-shared-mem-vgdb-13786
>>
>> Hmm, dont know. thanks for noticing.
>> Maybe the shm file system is not mounted or this test requires root?
>> I am away until September, will look into that then.
>
This issue persists. One of the s390x builds lately had this:
=================================================
./valgrind-new/cachegrind/tests/wrap5.stderr.diff
=================================================
--- wrap5.stderr.exp 2011-08-23 21:56:44.459662017 +0200
+++ wrap5.stderr.out 2011-08-23 22:01:09.589662018 +0200
@@ -15,3 +15,5 @@
LL refs:
LL misses:
LL miss rate:
+could not unlink /tmp/vgdb-pipe-from-vgdb-to-30026
+could not unlink /tmp/vgdb-pipe-to-vgdb-from-30026
I suspect it has the same cause.
> My guess is the following: if Valgrind terminates abnormally (e.g. due to a fatal internal
> error or due to a kill -9 or ...), then the vgdb FIFO and shmem files cannot be cleaned up.
> If such an abnormal termination is done by a Valgrind running under user xxx with pid yyy,
> then sometime later, it could be that another Valgrind is launched by user zzz with a
> "re-cycled" pid yyy.
The recycling of pids seems like a plausible explanation to me.
>
> A possible solution to avoid such error 13 is to add the user id in the FIFO and shmem
> file names e.g. /tmp/vgdb-pipe-shared-mem-vgdb-<here the user id>-<pid>.
> Then there should be no clash anymore.
>
I think adding the userid is the right approach and should resolve the
issue.
Florian
|