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] [thread-next>] [day] [month] [year] [list]

Message-ID: <1673a119-838f-455e-88fb-528bbd72e1ea@amd.com>
Date: Fri, 16 Jan 2026 11:03:09 +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 1/16/26 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:
>>>    [<90000000047a8524>] show_stack+0x64/0x190
>>>    [<90000000047a0614>] dump_stack_lvl+0x6c/0x9c
>>>    [<90000000047cef34>] __warn+0xa4/0x1b0
>>>    [<90000000060a4884>] __report_bug+0xa4/0x1d0
>>>    [<90000000060a4a88>] report_bug+0x38/0xd0
>>>    [<90000000060df330>] do_bp+0x260/0x410
>>>    [<90000000047a6bc0>] handle_bp+0x120/0x1c0
>>>    [<ffff8000028bff40>] amdgpu_irq_put+0xb0/0x140 [amdgpu]
>>>    [<ffff8000027b1a8c>] amdgpu_fence_driver_hw_fini+0x12c/0x180 [amdgpu]
>>>    [<ffff800002f2c04c>] amdgpu_device_fini_hw+0xf0/0x3fc [amdgpu]
>>>    [<ffff80000279e2ac>] amdgpu_driver_load_kms+0x7c/0xa0 [amdgpu]
>>>    [<ffff800002791128>] amdgpu_pci_probe+0x298/0x810 [amdgpu]
>>>    [<90000000054d04a4>] local_pci_probe+0x44/0xc0
>>>    [<90000000047f4ab0>] work_for_cpu_fn+0x20/0x40
>>>    [<90000000047f93e0>] process_one_work+0x170/0x4e0
>>>    [<90000000047fa14c>] worker_thread+0x3ac/0x4e0
>>>    [<9000000004806824>] kthread+0x154/0x170
>>>    [<90000000060df5b4>] ret_from_kernel_thread+0x24/0xd0
>>>    [<90000000047a62a4>] ret_from_kernel_thread_asm+0xc/0x88
>>>
>>>    ---[ end trace 0000000000000000 ]---
>>>    amdgpu 0000:07:00.0: probe with driver amdgpu failed with error -110
>>>    amdgpu 0000:07:00.0: amdgpu: amdgpu: ttm finalized
>>>
>>> This is because amdgpu_irq_enabled() is false in amdgpu_irq_put(), then
>>> the condition of WARN_ON() is true.
>>>
>>> In order to avoid the unnecessary Call Traces, it can remove the check of
>>> amdgpu_irq_enabled() and only check atomic_read(&src->enabled_types[type]
>>> for three reasons:
>>>
>>> (1) The aim is to prevent refcount from being less than 0, it was added in
>>> commit 1fa8d710573f ("drm/amdgpu: Fix desktop freezed after gpu-reset").
>>> (2) There are already many useful failed log before the Call Trace, there
>>> is no need to WARN_ON().
>>
>> Well completely disagree. The call trace here is absolutely intentional.
> 
> If so, since the call trace is same, is it enough to use WARN_ON_ONCE()
> here?

I also don't see a justification for that.

>> That you get a lot of other backtraces because the driver doesn't initialize at all isn't a good rational to remove this one here.
>>
>> Regards,
>> Christian.
>>
>>> (3) The following checks in amdgpu_irq_put() are same with the checks in
>>> amdgpu_irq_enabled(), there is no need to do the redundant operations.
>>>
>>>     if (!adev->irq.installed)
>>>         return -ENOENT;
>>>
>>>     if (type >= src->num_types)
>>>         return -EINVAL;
>>>
>>>     if (!src->enabled_types || !src->funcs->set)
>>>         return -EINVAL;
> 
> Is this reasonable? Only check atomic_read(&src->enabled_types[type]?

No, absolutely not. That are two completely different things.

> That is to say, does it make sense to do the following change?

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.

Regards,
Christian.

> 
> ----->8-----
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c
> index 8112ffc85995..d10d6fcc525e 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c
> @@ -636,7 +636,7 @@ int amdgpu_irq_put(struct amdgpu_device *adev, struct amdgpu_irq_src *src,
>         if (!src->enabled_types || !src->funcs->set)
>                 return -EINVAL;
> 
> -       if (WARN_ON(!amdgpu_irq_enabled(adev, src, type)))
> +       if (WARN_ON_ONCE(!atomic_read(&src->enabled_types[type])))
>                 return -EINVAL;
> 
>         if (atomic_dec_and_test(&src->enabled_types[type]))
> 
> Thanks,
> Tiezhu
>