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] Linux: Adjust gedents64 buffer size to int range [BZ #24740]


* Rafal Luzynski:

> 27.06.2019 15:09 Florian Weimer <fweimer@redhat.com> wrote:
>> [...]
>> Sure, make sense.  Re-tested and pushed with that change.
>> 
>> Thanks,
>> Florian
>
> Florian, the test for this patch always fails with timeout on
> my machine, Fedora 30 and I don't think it's the CPU too slow.
> I did not have time to investigate this thoroughly.  I am going
> to do it later but I'd like to give this feedback already.

Ah.  I see.  It's the test harness asking malloc to perform a memset:

#0  __memset_avx2_erms () at ../sysdeps/x86_64/multiarch/memset-vec-unaligned-erms.S:151
#1  0x00007ffff7a9c86c in alloc_perturb (n=<optimized out>, p=<optimized out>) at malloc.c:1869
#2  _int_malloc (av=av@entry=0x7ffff7dd0b80 <main_arena>, bytes=bytes@entry=4294967297) at malloc.c:4143
#3  0x00007ffff7a9dd04 in __GI___libc_malloc (bytes=4294967297) at malloc.c:3058

Presumably I could replace it with mmap to avoid this.

I think the last time something similar came up, we decided to retain
MALLOC_PERTURB_.

Thanks,
Florian


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