ARC testsuite regressions (was Re: [PATCH v7.2 07/13] ARC: Linux Syscall Interface)

Florian Weimer fweimer at redhat.com
Fri Jul 10 16:32:14 EDT 2020


* Vineet Gupta:

> On 7/10/20 10:02 AM, Florian Weimer via Libc-alpha wrote:
>> * Vineet Gupta via Libc-alpha:
>> 
>>> From scratch meaning glibc alone or the whole toolchain. I used
>>> buildroot and glibc-dirclean to nuke entire glibc but gcc was not
>>> rebuilt. I can try that too.
>> 
>> No, rebuilding glibc from scratch should be fine.
>
> Right, I clean built the entire toolchain and that didn't help - but turned out to
> be sane thing anyways as installed headers seems were not rebuilt/installed and I
> needed the signum.h -> signum-arch.h fixup in vfork.S
>
>
>>> Some of the failed tests have prints about static TLS block ... so I'm
>>> wondering if they could be related ?
>>>
>>> | $ cat dlfcn/tststatic.out
>>> | .../build/libc.so.6: cannot allocate memory in static TLS block
>> 
>> This suggests to me that the static initialization code does not produce
>> sufficient alignment for the TCB, given the 32-byte alignment required
>> by the rseq area.
>> 
>> You could try and see what happens if you change
>> sysdeps/arc/nptl/pthreaddef.h to this:
>> 
>> /* Alignment requirement for TCB.  */
>> #define TCB_ALIGNMENT           32
>> 
>> If that helps, we have more of a generic issue here. 8-/
>
> Bingo, that fixes all the dlfcn/* failures (full testsuite still running)
>
> PASS: dlfcn/tststatic
> PASS: dlfcn/tststatic2
> PASS: dlfcn/tststatic3
> PASS: dlfcn/tststatic4
> PASS: dlfcn/tststatic5
>
> So is that the fix or more is needed ?

I have just posted an architecture-independent change here:

  <https://sourceware.org/pipermail/libc-alpha/2020-July/116143.html>

I had forgotten about the similar code for static dlopen when I made the
adjustment in elf/dl-tls.c.  I think we need to clean this up
eventually, it's fairly broken unfortunately.  But the patch I posted
should paper over this problem as far libc.so is concerned (like the
previous patch for dynamic TLS).

Would you mind reverting the TCB_ALIGNMENT change and test this patch?

Thanks,
Florian




More information about the linux-snps-arc mailing list