Re: [Linaro-TCWG-CI] gdb-14-branchpoint-1426-gb960445a459: FAIL: 1 regressions on arm

[Thiago Jung Bauermann <thiago.bauermann@linaro.org>]

Hello,

This is recorded as GDB bug #31254¹. I've been investigating it and
yesterday I realised it's a memory corruption issue in a data structure
used by the Arm unwinder which uses exception tables.

I hope to have a fix for it soon™.

¹ https://sourceware.org/bugzilla/show_bug.cgi?id=31254


Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org> writes:

> Hi All,
>
> This is a false positive, obviously.  We do our best to filter out flaky tests, but in
> this case "gdb.threads/staticthreads.exp: up 10" PASSed twice in the previous run, and
> then FAILed twice in the next run.  Sneaky!
>
> Re. the FAIL, the testcase expects to be " in main .*" after "up 10", but ends up in
> pthread_join() instead:
> ===
> up 10
> #4  0x0001b864 in pthread_join ()
> (gdb) FAIL: gdb.threads/staticthreads.exp: up 10
> ===
> See [1] for details.
>
> [1]
> https://ci.linaro.org/job/tcwg_gdb_check--master-arm-build/752/artifact/artifacts/00-sumfiles/gdb.log.1.xz
> .
>
> Hi Thiago,
>
> Would you please investigate whether ending up in pthread_join() is expected/reasonable
> for 32-bit ARM?  In other words, whether we have a GDB bug exposed by staticthreads.exp or
> the testcase needs to be generalized a bit.
>
> Thank you,


-- 
Thiago