This is the mail archive of the libc-alpha@sourceware.org mailing list for the glibc project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH] Test case for BZ 19329


On 24/11/16 11:40, Florian Weimer wrote:
> On 11/24/2016 12:18 PM, Szabolcs Nagy wrote:
>> Test concurrent dlopen and pthread_create when the loaded
>> modules have TLS.  This triggers dl-tls assertion failures
>> more reliably than the tst-stack4 test.
>>
>> The dlopened module has 100 DT_NEEDED dependencies and
>> for me about 4000 concurrent thread creations are needed
>> to see failures on x86_64.
>>
>> 2016-11-24  Szabolcs Nagy  <szabolcs.nagy@arm.com>
>>
>>     [BZ #19329]
>>     * nptl/Makefile (tests): Add tst-tls7.
>>     (modules-names): Add tst-tls7mod, tst-tls7mod-dep.
>>     * nptl/tst-tls7.c: New file.
>>     * nptl/tst-tls7mod-dep.c: New file.
>>     * nptl/tst-tls7mod.c: New file.
>>
> 
> Please do not use assert in tests.  Use the xpthread_* functions in test-skeleton.c instead.
> 

i can do that but it looks ugly.

we are looking for an assertion failure anyway
so it would be better to make the test system
deal with that..

> This test will fail on large SMP machines due to timeouts because thread creation is rather expensive there.  A
> less resource-intense way of writing this test would be to iterate multiple times and try to hit the race each
> time, with a smaller number of threads created in parallel.
> 

i can try, but iterating is not easy: there
is dynamic linker internal state that cannot
be reset (generation counter, max dtv idx,
allocated slotininfo entries) so i'd need to
start a new process for every iteration.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]