linux-kernel - Re: [PATCH] drm/amd/display: Reduce stack size for dml21

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]

Message-ID: <CAK8P3a243ruO-7H79yZ49_+5RVQk4V7kE_=Ffy7Gko-um_0VBA@mail.gmail.com>
Date:   Tue, 14 Sep 2021 23:56:11 +0200
From:   Arnd Bergmann <arnd@...nel.org>
To:     Harry Wentland <harry.wentland@....com>
Cc:     amd-gfx list <amd-gfx@...ts.freedesktop.org>,
        Nick Desaulniers <ndesaulniers@...gle.com>,
        Linus Torvalds <torvalds@...ux-foundation.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        Leo Li <sunpeng.li@....com>,
        Alex Deucher <alexander.deucher@....com>,
        Christian König <christian.koenig@....com>,
        xinhui pan <Xinhui.Pan@....com>,
        Nathan Chancellor <nathan@...nel.org>,
        Guenter Roeck <linux@...ck-us.net>, llvm@...ts.linux.dev
Subject: Re: [PATCH] drm/amd/display: Reduce stack size for dml21_ModeSupportAndSystemConfigurationFull

On Tue, Sep 14, 2021 at 11:05 PM Harry Wentland <harry.wentland@....com> wrote:
>
> [Why & How]
> With Werror enabled in the kernel we were failing the clang build since
> dml21_ModeSupportAndSystemConfigurationFull's stack frame is 1064 when
> building with clang, and exceeding the default 1024 stack frame limit.
>
> The culprit seems to be the Pipe struct, so pull the relevant block
> out into its own sub-function.

I suspect it's not the Pipe struct but rather the way that you call another
function with a crazy number of arguments here. After your change,
this likely gets inlined and you avoid the problem, so the patch ends
up doing the right thing.

If you do more patches like this, I would suggest mentioning the new
stack usage of the calling function and the new noinline function, to
make sure that the combined number isn't actually worse than the old
number.

You can get these numbers by recompiling the file with the frame
size warning set to a low value, e.g. adding -Wframe-larger-than=100
to the command line.

Acked-by: Arnd Bergmann <arnd@...db.de>