glibc 2.35 i686 failures in tst-memset-rtm (bug 28033)

[Carlos O'Donell]

On 1/14/22 16:16, Florian Weimer wrote:
> * Carlos O'Donell via Libc-alpha:
> 
>> On Fedora 35, at commit eb024343726ee3b2e644b72ae03df9023898cf01, I see
>> the following somewhat random failures:
>>
>> [carlos@fedora ~]$ cat ~/build/glibc-review-i686/string/tst-memset-rtm.out
>> error: ../sysdeps/x86/tst-string-rtm.h:63: TSX abort rate: 100.00% (3000 out of 3000)
>> error: 1 test failures
>> [carlos@fedora ~]$ cat ~/build/glibc-review-i686/string/tst-memset-rtm.test-result 
>> FAIL: string/tst-memset-rtm
>> original exit status 1
>> [carlos@fedora ~]$ 
>>
>> Is this the same root cause as this issue?
>> https://sourceware.org/pipermail/libc-alpha/2022-January/135323.html
> 
> Please see <https://sourceware.org/bugzilla/show_bug.cgi?id=27398#c3>
> and check if you can reproduce this phenomenon.
> 
> This is a kernel/firmware bug.  RTM is not correctly disabled after
> resume.

Confirmed. Pinning to cpu 0 reproduces the problem. CPUs 1-7 are fine.

-- 
Cheers,
Carlos.