linux-kernel - Re: [RFC PATCH] drm/amdgpu: Avoid unnecessary Call Traces in amdgpu_irq

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] [day] [month] [year] [list]

Message-ID: <aeede61e-2c03-4711-8d77-efd551d17b29@amd.com>
Date: Thu, 22 Jan 2026 13:53:53 +0100
From: Christian König <christian.koenig@....com>
To: Tiezhu Yang <yangtiezhu@...ngson.cn>,
 Alex Deucher <alexander.deucher@....com>
Cc: Alan Liu <haoping.liu@....com>, amd-gfx@...ts.freedesktop.org,
 linux-kernel@...r.kernel.org
Subject: Re: [RFC PATCH] drm/amdgpu: Avoid unnecessary Call Traces in
 amdgpu_irq_put()

On 1/22/26 10:37, Tiezhu Yang wrote:
> On 2026/1/19 下午6:45, Christian König wrote:
>> On 1/19/26 09:53, Tiezhu Yang wrote:
>>> On 2026/1/16 下午6:03, Christian König wrote:
>>>> On 1/16/26 02:20, Tiezhu Yang wrote:
>>>>> On 2026/1/15 下午9:47, Christian König wrote:
>>>>>> On 1/15/26 02:28, Tiezhu Yang wrote:
>>>>>>> Currently, there are many Call Traces when booting kernel on LoongArch,
>>>>>>> here are the trimmed kernel log messages:
>>>>>>>
>>>>>>>      amdgpu 0000:07:00.0: amdgpu: hw_init of IP block <gfx_v6_0> failed -110
>>>>>>>      amdgpu 0000:07:00.0: amdgpu: amdgpu_device_ip_init failed
>>>>>>>      amdgpu 0000:07:00.0: amdgpu: Fatal error during GPU init
>>>>>>>      amdgpu 0000:07:00.0: amdgpu: amdgpu: finishing device.
>>>>>>>      ------------[ cut here ]------------
>>>>>>>      WARNING: drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c:639 at amdgpu_irq_put+0xb0/0x140 [amdgpu], CPU#0: kworker/0:0/9
>>>>>>>      ...
>>>>>>>      Call Trace:
>>>
>>> ...
>>>
>>>> The warning can basically only be triggered by two conditions:
>>>> 1. A fatal problem while loading the driver and the error handling is not 100% clean.
>>>> 2. A driver coding error.
>>>>
>>>> And we really need to catch all of those, so there is no real rational to limit the warning.
>>>>
>>>> I mean when you run into any of those they should potentially be fixed at some point.
>>>
>>> I did the following change and it can fix the problem, given that I am
>>> not familiar with amdgpu driver, could you please check it? If it is OK,
>>> I will send a formal patch later.
>>>
>>> ----->8-----
>>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c
>>> index 8112ffc85995..ac19565e7c45 100644
>>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c
>>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c
>>> @@ -138,6 +138,9 @@ void amdgpu_irq_disable_all(struct amdgpu_device *adev)
>>>                          if (!src || !src->funcs->set || !src->num_types)
>>>                                  continue;
>>>
>>> +                       kfree(src->enabled_types);
>>> +                       src->enabled_types = NULL;
>>> +
>>
>> Mhm, I need to double check that but that looks like not such a bad idea of hand.
> 
> Please let me know whether the above change is OK, if yes I will send a
> patch ASAP before the coming merge window.

Well, this has absolutely lowest priority. So don't expect anything in the near term.

Christian.

> 
> Thanks,
> Tiezhu
>