AMD-GFX Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: "Christian König" <ckoenig.leichtzumerken@gmail.com>
To: SRINIVASAN SHANMUGAM <srinivasan.shanmugam@amd.com>,
	amd-gfx@lists.freedesktop.org, alexander.deucher@amd.com
Subject: Re: [PATCH 1/8] drm/amdgpu: grab an additional reference on the gang fence v2
Date: Fri, 14 Mar 2025 15:20:03 +0100	[thread overview]
Message-ID: <ff0378be-671c-4c60-8086-03fb4c21bc0b@gmail.com> (raw)
In-Reply-To: <de0e6edc-c3a8-4c9f-b0f3-37cfd8776db5@amd.com>

Am 14.03.25 um 05:09 schrieb SRINIVASAN SHANMUGAM:
>
> On 3/7/2025 7:18 PM, Christian König wrote:
>> We keep the gang submission fence around in adev, make sure that it
>> stays alive.
>>
>> v2: fix memory leak on retry
>>
>> Signed-off-by: Christian König <christian.koenig@amd.com>
>> ---
>>   drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 10 +++++++++-
>>   1 file changed, 9 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
>> index 198d29faa754..337543ec615c 100644
>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
>> @@ -6889,18 +6889,26 @@ struct dma_fence *amdgpu_device_switch_gang(struct amdgpu_device *adev,
>>   {
>>       struct dma_fence *old = NULL;
>>   +    dma_fence_get(gang);
>>       do {
>>           dma_fence_put(old);
>>           old = amdgpu_device_get_gang(adev);
>>           if (old == gang)
>>               break;
>>   -        if (!dma_fence_is_signaled(old))
>> +        if (!dma_fence_is_signaled(old)) {
>
> Here, should we need to check ?

The gang is initialized to a dummy fence on bootup. So even when there is never any gang submission the old value is never NULL.

Regards,
Christian.

>
>     // Check if old fence isn't signaled
>     if (old && !dma_fence_is_signaled(old)) {
>
>> +            dma_fence_put(gang);
>>               return old;
>> +        }
>>         } while (cmpxchg((struct dma_fence __force **)&adev->gang_submit,
>>                old, gang) != old);
>>   +    /*
>> +     * Drop it once for the exchanged reference in adev and once for the
>> +     * thread local reference acquired in amdgpu_device_get_gang().
>> +     */
>> +    dma_fence_put(old);
>
> if (old)
>     dma_fence_put(old); // Ensure to release old reference  only if it is valid?
>
>
>>       dma_fence_put(old);
>>       return NULL;
>>   }


  reply	other threads:[~2025-03-14 14:20 UTC|newest]

Thread overview: 19+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-03-07 13:48 [PATCH 1/8] drm/amdgpu: grab an additional reference on the gang fence v2 Christian König
2025-03-07 13:48 ` [PATCH 2/8] drm/amdgpu: use GFP_NOWAIT for memory allocations Christian König
2025-03-07 13:48 ` [PATCH 3/8] drm/amdgpu: overwrite signaled fence in amdgpu_sync Christian König
2025-03-12 15:06   ` SRINIVASAN SHANMUGAM
2025-03-14 14:13     ` Christian König
2025-03-07 13:48 ` [PATCH 4/8] drm/amdgpu: rework how isolation is enforced v2 Christian König
2025-03-14  7:07   ` Christian König
2025-03-07 13:48 ` [PATCH 5/8] drm/amdgpu: rework how the cleaner shader is emitted v3 Christian König
2025-03-14  4:24   ` SRINIVASAN SHANMUGAM
2025-03-14 14:21     ` Christian König
2025-03-07 13:48 ` [PATCH 6/8] drm/amdgpu: stop reserving VMIDs to enforce isolation Christian König
2025-03-12 15:10   ` SRINIVASAN SHANMUGAM
2025-03-14 14:15     ` Christian König
2025-03-12 15:16   ` SRINIVASAN SHANMUGAM
2025-03-07 13:48 ` [PATCH 7/8] drm/amdgpu: add isolation trace point Christian König
2025-03-07 13:48 ` [PATCH 8/8] drm/amdgpu: add cleaner shader " Christian König
2025-03-14  4:09 ` [PATCH 1/8] drm/amdgpu: grab an additional reference on the gang fence v2 SRINIVASAN SHANMUGAM
2025-03-14 14:20   ` Christian König [this message]
2025-03-15  2:20     ` SRINIVASAN SHANMUGAM

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=ff0378be-671c-4c60-8086-03fb4c21bc0b@gmail.com \
    --to=ckoenig.leichtzumerken@gmail.com \
    --cc=alexander.deucher@amd.com \
    --cc=amd-gfx@lists.freedesktop.org \
    --cc=srinivasan.shanmugam@amd.com \
    /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