Re: [PATCH v3] malloc: Remove bin scanning from memalign (bug 30723)

[Stefan Liebler <stli@linux.ibm.com>]

On 19.09.23 19:35, Florian Weimer wrote:
> * Stefan Liebler:
> 
>> On 11.08.23 17:48, Florian Weimer via Libc-alpha wrote:
>>> On the test workload (mpv --cache=yes with VP9 video decoding), the
>>> bin scanning has a very poor success rate (less than 2%).  The tcache
>>> scanning has about 50% success rate, so keep that.
>>>
>>> Update comments in malloc/tst-memalign-2 to indicate the purpose
>>> of the tests.  Even with the scanning removed, the additional
>>> merging opportunities since commit 542b1105852568c3ebc712225ae78b
>>> ("malloc: Enable merging of remainders in memalign (bug 30723)")
>>> are sufficient to pass the existing large bins test.
>> Hi,
>>
>> on s390x, I've observed these test-fails starting with this commit:
>> FAIL: malloc/tst-memalign-2
>> FAIL: malloc/tst-memalign-2-malloc-hugetlb1
>>
>> With this output:
>> error: tst-memalign-2.c:158: not true: count > LN / 2
>> error: 1 test failures
>>
>> Note, that I only see the fails if /proc/sys/kernel/randomize_va_space
>> is set to 0. Then I get count=3; LN=8;
>>
>> If randomize_va_space is set to 2, the test passes and I get count=5; LN=8;
>>
>> Is s390x the only architecture where we get this failure?
> 
> Does your kernel have a fix for the early sbrk failure?  I suspect what
> happens is that we fall over to mmap from sbrk, making the heap
> non-contiguous.
> 
> The sbrk failure should be visible under strace.
> 
Yes, strace shows mmap with randomize_va_space=0:
brk(NULL)                               = 0x3fff7fb2000
brk(0x3fff7fd3000)                      = 0x3fff7fd3000
brk(0x3fff7ff7000)                      = 0x3fff7fd3000
mmap(NULL, 1048576, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1,
0) = 0x3fff7b00000
write(1, "error: tst-memalign-2.c:158: not"..., 54error:
tst-memalign-2.c:158: not true: count > LN / 2
) = 54

Note that there is this previous mmap-event before set_tid_address-syscall:
mmap(NULL, 8192, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0)
= 0x3fff7ff7000

With with randomize_va_space=2, I see:
brk(NULL)                               = 0x2aa005e8000
brk(0x2aa00609000)                      = 0x2aa00609000
brk(0x2aa0062d000)                      = 0x2aa0062d000
brk(0x2aa00654000)                      = 0x2aa00654000
exit_group(0)                           = ?


I think you mean the kernel fixes as documented in the static PIE
confiugre-check:
https://sourceware.org/git/?p=glibc.git;a=blob;f=sysdeps/s390/s390-64/configure.ac;h=4657de0d3795f40388805d438178c63f9558e936;hb=HEAD#l85
Those commits went into linux 5.19 and I see the fails on kernels with 6.xyz



> Thanks,
> Florian
>