[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <55ca17a3-8ea6-450e-8ec6-9bda97808164@oss.qualcomm.com>
Date: Mon, 28 Oct 2024 11:39:16 +0100
From: Konrad Dybcio <konrad.dybcio@....qualcomm.com>
To: Akhil P Oommen <quic_akhilpo@...cinc.com>,
Arnd Bergmann
<arnd@...nel.org>, Rob Clark <robdclark@...il.com>,
Sean Paul <sean@...rly.run>, Konrad Dybcio <konradybcio@...nel.org>,
Abhinav Kumar <quic_abhinavk@...cinc.com>,
Dmitry Baryshkov <dmitry.baryshkov@...aro.org>,
Marijn Suijten <marijn.suijten@...ainline.org>,
Dave Airlie <airlied@...il.com>, Simona Vetter <simona@...ll.ch>,
Nathan Chancellor <nathan@...nel.org>,
Nick Desaulniers <ndesaulniers@...gle.com>,
Bill Wendling
<morbo@...gle.com>,
Justin Stitt <justinstitt@...gle.com>
Cc: linux-arm-msm@...r.kernel.org, dri-devel@...ts.freedesktop.org,
freedreno@...ts.freedesktop.org, linux-kernel@...r.kernel.org,
llvm@...ts.linux.dev
Subject: Re: [PATCH] drm/msm/a6xx: Fix excessive stack usage
On 28.10.2024 10:52 AM, Akhil P Oommen wrote:
> On 10/28/2024 12:13 AM, Arnd Bergmann wrote:
>> On Sun, Oct 27, 2024, at 18:05, Akhil P Oommen wrote:
>>> Clang-19 and above sometimes end up with multiple copies of the large
>>> a6xx_hfi_msg_bw_table structure on the stack. The problem is that
>>> a6xx_hfi_send_bw_table() calls a number of device specific functions to
>>> fill the structure, but these create another copy of the structure on
>>> the stack which gets copied to the first.
>>>
>>> If the functions get inlined, that busts the warning limit:
>>>
>>> drivers/gpu/drm/msm/adreno/a6xx_hfi.c:631:12: error: stack frame size
>>> (1032) exceeds limit (1024) in 'a6xx_hfi_send_bw_table'
>>> [-Werror,-Wframe-larger-than]
>>>
>>> Fix this by kmalloc-ating struct a6xx_hfi_msg_bw_table instead of using
>>> the stack. Also, use this opportunity to skip re-initializing this table
>>> to optimize gpu wake up latency.
>>>
>>> Cc: Arnd Bergmann <arnd@...nel.org>
>>
>> Please change this to "Reported-by:"
>
> Sure.
>
>>
>> The patch looks correct to me, just one idea for improvement.
>>
>>> b/drivers/gpu/drm/msm/adreno/a6xx_gmu.h
>>> index 94b6c5cab6f4..b4a79f88ccf4 100644
>>> --- a/drivers/gpu/drm/msm/adreno/a6xx_gmu.h
>>> +++ b/drivers/gpu/drm/msm/adreno/a6xx_gmu.h
>>> @@ -99,6 +99,7 @@ struct a6xx_gmu {
>>> struct completion pd_gate;
>>>
>>> struct qmp *qmp;
>>> + struct a6xx_hfi_msg_bw_table *bw_table;
>>> };
>>
>> I think the bw_table is better just embedded
>> in here rather than referenced as a pointer:
>>
> There are some low tier chipsets with relatively lower RAM size that
> doesn't require this table. So, dynamically allocating this here helps
> to save 640 bytes (minus the overhead of tracking).
I'd second this, said chipsets often ship with 1-2 GiB of RAM (which
is still a lot in comparison, but you know.. every little bit counts)
Konrad
Powered by blists - more mailing lists