From: "Christian König" <christian.koenig@amd.com>
To: "Kuehling, Felix" <felix.kuehling@amd.com>,
phasta@kernel.org, alexdeucher@gmail.com, simona.vetter@ffwll.ch,
tursulin@ursulin.net, airlied@gmail.com, matthew.brost@intel.com
Cc: dri-devel@lists.freedesktop.org, amd-gfx@lists.freedesktop.org
Subject: Re: [PATCH 11/20] drm/amdgpu: fix KFD eviction fence enable_signaling path
Date: Thu, 6 Nov 2025 17:46:39 +0100 [thread overview]
Message-ID: <9484fec0-ca36-4968-9edc-ca517825fe68@amd.com> (raw)
In-Reply-To: <887de3dc-f6a8-46c7-baef-862fe1acaf02@amd.com>
On 11/6/25 17:37, Kuehling, Felix wrote:
>
> On 2025-11-06 08:43, Christian König wrote:
>> On 11/4/25 17:28, Philipp Stanner wrote:
>>> On Fri, 2025-10-31 at 14:16 +0100, Christian König wrote:
>>>> Calling dma_fence_is_signaled() here is illegal!
>>> The series was sent as a v2. But is this still an RFC?
>> I think when Matthew came up with the XE patches we pretty much agreed that this is the way to go.
>>
>>> If not, more detailed commit messages are a desirable thing.
>> Good point, how about:
>>
>> The enable_signaling callback is called with the same irqsave spinlock held than dma_fence_is_signaled() tries to grab. That will 100% reliable deadlock if that happens.
>
> I guess we could use dma_fence_is_signaled_locked instead. That said, it only tries to take the lock (in dma_fence_signal) if fence->ops->signal is set, which isn't the case for these fences. That's why this has never caused a problem up till now.
But when fence->ops->signal isn't set then why are we calling this?
Regards,
Christian.
>
> Regards,
> Felix
>
>
>>
>> Thanks,
>> Christian.
>>
>>>
>>> P.
>>>
>>>> Signed-off-by: Christian König <christian.koenig@amd.com>
>>>> ---
>>>> drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_fence.c | 6 ------
>>>> 1 file changed, 6 deletions(-)
>>>>
>>>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_fence.c
>>>> b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_fence.c
>>>> index 1ef758ac5076..09c919f72b6c 100644
>>>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_fence.c
>>>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_fence.c
>>>> @@ -120,12 +120,6 @@ static bool amdkfd_fence_enable_signaling(struct
>>>> dma_fence *f)
>>>> {
>>>> struct amdgpu_amdkfd_fence *fence =
>>>> to_amdgpu_amdkfd_fence(f);
>>>> - if (!fence)
>>>> - return false;
>>>> -
>>>> - if (dma_fence_is_signaled(f))
>>>> - return true;
>>>> -
>>>> if (!fence->svm_bo) {
>>>> if
>>>> (!kgd2kfd_schedule_evict_and_restore_process(fence->mm, f))
>>>> return true;
next prev parent reply other threads:[~2025-11-06 16:46 UTC|newest]
Thread overview: 55+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-10-31 13:16 Independence for dma_fences! v2 Christian König
2025-10-31 13:16 ` [PATCH 01/20] dma-buf: cleanup dma_fence_describe v2 Christian König
2025-10-31 14:04 ` Tvrtko Ursulin
2025-10-31 13:16 ` [PATCH 02/20] dma-buf: rework stub fence initialisation v2 Christian König
2025-10-31 14:05 ` Tvrtko Ursulin
2025-11-04 15:01 ` Tvrtko Ursulin
2025-11-06 13:16 ` Christian König
2025-10-31 13:16 ` [PATCH 03/20] dma-buf: protected fence ops by RCU v2 Christian König
2025-10-31 14:29 ` Tvrtko Ursulin
2025-11-06 13:14 ` Christian König
2025-11-07 11:09 ` Tvrtko Ursulin
2025-10-31 13:16 ` [PATCH 04/20] dma-buf: detach fence ops on signal Christian König
2025-11-07 11:04 ` Philipp Stanner
2025-10-31 13:16 ` [PATCH 05/20] dma-buf: inline spinlock for fence protection Christian König
2025-11-07 11:59 ` Philipp Stanner
2025-10-31 13:16 ` [PATCH 06/20] dma-buf: use inline lock for the stub fence Christian König
2025-11-04 15:05 ` Tvrtko Ursulin
2025-10-31 13:16 ` [PATCH 07/20] dma-buf: use inline lock for the dma-fence-array Christian König
2025-11-05 8:50 ` Tvrtko Ursulin
2025-11-07 12:04 ` Philipp Stanner
2025-11-12 13:53 ` Christian König
2025-11-12 14:00 ` Philipp Stanner
2025-10-31 13:16 ` [PATCH 08/20] dma-buf: use inline lock for the dma-fence-chain Christian König
2025-11-04 15:08 ` Tvrtko Ursulin
2025-10-31 13:16 ` [PATCH 09/20] drm/sched: use inline locks for the drm-sched-fence Christian König
2025-11-04 15:12 ` Tvrtko Ursulin
2025-11-06 13:23 ` Christian König
2025-11-06 13:45 ` Tvrtko Ursulin
2025-11-07 8:33 ` Philipp Stanner
2025-11-12 13:58 ` Christian König
2025-10-31 13:16 ` [PATCH 10/20] drm/amdgpu: clean up and unify hw fence handling Christian König
2025-11-04 15:14 ` Tvrtko Ursulin
2025-10-31 13:16 ` [PATCH 11/20] drm/amdgpu: fix KFD eviction fence enable_signaling path Christian König
2025-11-04 16:28 ` Philipp Stanner
2025-11-06 13:43 ` Christian König
2025-11-06 16:37 ` Kuehling, Felix
2025-11-06 16:46 ` Christian König [this message]
2025-11-06 17:07 ` Kuehling, Felix
2025-11-06 17:09 ` Christian König
2025-11-06 17:25 ` Kuehling, Felix
2025-11-13 14:37 ` Christian König
2025-11-13 17:46 ` Kuehling, Felix
2025-10-31 13:16 ` [PATCH 12/20] drm/amdgpu: independence for the amdgpu_fence! Christian König
2025-10-31 13:16 ` [PATCH 13/20] drm/amdgpu: independence for the amdgpu_eviction_fence! Christian König
2025-11-04 15:45 ` Tvrtko Ursulin
2025-10-31 13:16 ` [PATCH 14/20] drm/amdgpu: independence for the amdgpu_vm_tlb_fence! Christian König
2025-11-04 15:45 ` Tvrtko Ursulin
2025-10-31 13:16 ` [PATCH 15/20] drm/amdgpu: independence for the amdkfd_fence! Christian König
2025-10-31 14:34 ` Kuehling, Felix
2025-10-31 13:16 ` [PATCH 16/20] drm/amdgpu: independence for the amdgpu_userq__fence! Christian König
2025-11-04 15:59 ` Tvrtko Ursulin
2025-10-31 13:16 ` [PATCH 17/20] drm/xe: Disconnect the low hanging fences from Xe module Christian König
2025-10-31 13:16 ` [PATCH 18/20] drm/xe: Drop HW fence slab Christian König
2025-10-31 13:16 ` [PATCH 19/20] drm/xe: Promote xe_hw_fence_irq to an ref counted object Christian König
2025-10-31 13:16 ` [PATCH 20/20] drm/xe: Finish disconnect HW fences from module Christian König
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=9484fec0-ca36-4968-9edc-ca517825fe68@amd.com \
--to=christian.koenig@amd.com \
--cc=airlied@gmail.com \
--cc=alexdeucher@gmail.com \
--cc=amd-gfx@lists.freedesktop.org \
--cc=dri-devel@lists.freedesktop.org \
--cc=felix.kuehling@amd.com \
--cc=matthew.brost@intel.com \
--cc=phasta@kernel.org \
--cc=simona.vetter@ffwll.ch \
--cc=tursulin@ursulin.net \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox