lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date:   Fri, 2 Dec 2022 09:46:48 +0100
From:   Thorsten Leemhuis <regressions@...mhuis.info>
To:     Andrew Morton <akpm@...ux-foundation.org>,
        Rik van Riel <riel@...riel.com>
Cc:     Yang Shi <shy828301@...il.com>,
        "Huang, Ying" <ying.huang@...el.com>,
        kernel test robot <yujie.liu@...el.com>, lkp@...ts.01.org,
        lkp@...el.com, Matthew Wilcox <willy@...radead.org>,
        linux-kernel@...r.kernel.org, linux-mm@...ck.org,
        feng.tang@...el.com, zhengjun.xing@...ux.intel.com,
        fengwei.yin@...el.com, Nathan Chancellor <nathan@...nel.org>
Subject: Re: [mm] f35b5d7d67: will-it-scale.per_process_ops -95.5% regression

On 01.12.22 22:22, Andrew Morton wrote:
> On Thu, 01 Dec 2022 15:29:41 -0500 Rik van Riel <riel@...riel.com> wrote:
>> On Thu, 2022-12-01 at 19:33 +0100, Thorsten Leemhuis wrote:
>>> On 28.11.22 07:40, Nathan Chancellor wrote:
>>> I wonder what we should do about below performance regression. Is
>>> reverting the culprit now and reapplying it later together with a fix
>>> a
>>> viable option? Or was anything done/is anybody doing something
>>> already
>>> to address the problem and I just missed it?
>>
>> The changeset in question speeds up kernel compiles with
>> GCC, as well as the runtime speed of other programs, due
>> to being able to use THPs more. However, it slows down kernel
>> compiles with clang, due to ... something clang does.
>>
>> I have not figured out what that something is yet.
>>
>> I don't know if I have the wrong version of clang here,
>> but I have not seen any smoking gun at all when tracing
>> clang system calls. I see predominantly small mmap and
>> unmap calls, and nothing that even triggers 2MB alignment.
> 
> 2.8% speedup for gcc is nice.  Massive slowdown in the malloc banchmark
> and in LLVM/clang is very bad - we don't know what other userspace will
> be so affected.
> 
> So I think we revert until this is fully understood.

Andrew, many thx for taking care of this. While at it let me please get
a small process issue of my chest:

What beverage of choice do I have to offer you to make you in the future
include 'Link:' tags linking to the report(s) when you add reverts like
the one for this issue?

My regression tracking bot heavily relies on them, that's why I care.
And our documentation (see [1]) also says that they should be used. But
the main reason why I ask in this particular case is different:

They in cases like afaics are especially helpful, as they make life a
whole lot easier for future code archeologists. And I guess that's not
something theoretical in this case, as I assume the patch that triggered
the issue will come back sooner or later -- and then those links will
help a lot to find this thread. Which is also why Linus really wants to
see them:
https://lore.kernel.org/all/CAHk-=wjMmSZzMJ3Xnskdg4+GGz=5p5p+GSYyFBTh0f-DgvdBWg@mail.gmail.com/
https://lore.kernel.org/all/CAHk-=wgs38ZrfPvy=nOwVkVzjpM3VFU1zobP37Fwd_h9iAD5JQ@mail.gmail.com/
https://lore.kernel.org/all/CAHk-=wjxzafG-=J8oT30s7upn4RhBs6TX-uVFZ5rME+L5_DoJA@mail.gmail.com/

Ciao, Thorsten

[1] see Documentation/process/submitting-patches.rst
(http://docs.kernel.org/process/submitting-patches.html) and
Documentation/process/5.Posting.rst
(https://docs.kernel.org/process/5.Posting.html)

Powered by blists - more mailing lists