Hi all,
Today's linux-next merge of the drm-misc tree got a conflict in:
drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
between commit:
eca13f3c67b6 ("drm/amdgpu: use the last IB as gang leader v2")
from the drm-misc-fixes tree and commit:
1728baa7e4e6 ("drm/amdgpu: use scheduler dependencies for CS")
from the drm-misc tree.
I fixed it up (see below) and can carry the fix as necessary. This
is now fixed as far as linux-next is concerned, but any non trivial
conflicts should be mentioned to your upstream maintainer when your tree
is submitted for merging. You may also want to consider cooperating
with the maintainer of the conflicting tree to minimise any particularly
complex conflicts.
--
Cheers,
Stephen Rothwell
diff --cc drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
index de5cb056c9ad,0528c2b1db6e..000000000000
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@@ -1197,10 -1201,7 +1203,10 @@@ static int amdgpu_cs_sync_rings(struct
}
for (i = 0; i < p->gang_size; ++i) {
+ if (p->jobs[i] == leader)
+ continue;
+
- r = amdgpu_sync_clone(&leader->sync, &p->jobs[i]->sync);
+ r = amdgpu_sync_push_to_job(&p->sync, p->jobs[i]);
if (r)
return r;
}
@@@ -1241,14 -1243,11 +1247,14 @@@ static int amdgpu_cs_submit(struct amdg
for (i = 0; i < p->gang_size; ++i)
drm_sched_job_arm(&p->jobs[i]->base);
- for (i = 0; i < (p->gang_size - 1); ++i) {
+ for (i = 0; i < p->gang_size; ++i) {
struct dma_fence *fence;
+ if (p->jobs[i] == leader)
+ continue;
+
fence = &p->jobs[i]->base.s_fence->scheduled;
- r = amdgpu_sync_fence(&leader->sync, fence);
+ r = drm_sched_job_add_dependency(&leader->base, fence);
if (r)
goto error_cleanup;
}
Hi all,
On Wed, 16 Nov 2022 10:47:52 +1100 Stephen Rothwell <[email protected]> wrote:
>
> Today's linux-next merge of the drm-misc tree got a conflict in:
>
> drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
>
> between commit:
>
> eca13f3c67b6 ("drm/amdgpu: use the last IB as gang leader v2")
>
> from the drm-misc-fixes tree and commit:
>
> 1728baa7e4e6 ("drm/amdgpu: use scheduler dependencies for CS")
>
> from the drm-misc tree.
>
> I fixed it up (see below) and can carry the fix as necessary. This
> is now fixed as far as linux-next is concerned, but any non trivial
> conflicts should be mentioned to your upstream maintainer when your tree
> is submitted for merging. You may also want to consider cooperating
> with the maintainer of the conflicting tree to minimise any particularly
> complex conflicts.
>
> --
> Cheers,
> Stephen Rothwell
>
> diff --cc drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
> index de5cb056c9ad,0528c2b1db6e..000000000000
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
> @@@ -1197,10 -1201,7 +1203,10 @@@ static int amdgpu_cs_sync_rings(struct
> }
>
> for (i = 0; i < p->gang_size; ++i) {
> + if (p->jobs[i] == leader)
> + continue;
> +
> - r = amdgpu_sync_clone(&leader->sync, &p->jobs[i]->sync);
> + r = amdgpu_sync_push_to_job(&p->sync, p->jobs[i]);
> if (r)
> return r;
> }
> @@@ -1241,14 -1243,11 +1247,14 @@@ static int amdgpu_cs_submit(struct amdg
> for (i = 0; i < p->gang_size; ++i)
> drm_sched_job_arm(&p->jobs[i]->base);
>
> - for (i = 0; i < (p->gang_size - 1); ++i) {
> + for (i = 0; i < p->gang_size; ++i) {
> struct dma_fence *fence;
>
> + if (p->jobs[i] == leader)
> + continue;
> +
> fence = &p->jobs[i]->base.s_fence->scheduled;
> - r = amdgpu_sync_fence(&leader->sync, fence);
> + r = drm_sched_job_add_dependency(&leader->base, fence);
> if (r)
> goto error_cleanup;
> }
Note that I had to keep the declaration of "leader" in amdgpu_cs_sync_rings().
--
Cheers,
Stephen Rothwell
Am 16.11.22 um 01:25 schrieb Stephen Rothwell:
> Hi all,
>
> On Wed, 16 Nov 2022 10:47:52 +1100 Stephen Rothwell <[email protected]> wrote:
>> Today's linux-next merge of the drm-misc tree got a conflict in:
>>
>> drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
>>
>> between commit:
>>
>> eca13f3c67b6 ("drm/amdgpu: use the last IB as gang leader v2")
>>
>> from the drm-misc-fixes tree and commit:
>>
>> 1728baa7e4e6 ("drm/amdgpu: use scheduler dependencies for CS")
>>
>> from the drm-misc tree.
>>
>> I fixed it up (see below) and can carry the fix as necessary. This
>> is now fixed as far as linux-next is concerned, but any non trivial
>> conflicts should be mentioned to your upstream maintainer when your tree
>> is submitted for merging. You may also want to consider cooperating
>> with the maintainer of the conflicting tree to minimise any particularly
>> complex conflicts.
>>
>> --
>> Cheers,
>> Stephen Rothwell
>>
>> diff --cc drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
>> index de5cb056c9ad,0528c2b1db6e..000000000000
>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
>> @@@ -1197,10 -1201,7 +1203,10 @@@ static int amdgpu_cs_sync_rings(struct
>> }
>>
>> for (i = 0; i < p->gang_size; ++i) {
>> + if (p->jobs[i] == leader)
>> + continue;
>> +
>> - r = amdgpu_sync_clone(&leader->sync, &p->jobs[i]->sync);
>> + r = amdgpu_sync_push_to_job(&p->sync, p->jobs[i]);
>> if (r)
>> return r;
>> }
>> @@@ -1241,14 -1243,11 +1247,14 @@@ static int amdgpu_cs_submit(struct amdg
>> for (i = 0; i < p->gang_size; ++i)
>> drm_sched_job_arm(&p->jobs[i]->base);
>>
>> - for (i = 0; i < (p->gang_size - 1); ++i) {
>> + for (i = 0; i < p->gang_size; ++i) {
>> struct dma_fence *fence;
>>
>> + if (p->jobs[i] == leader)
>> + continue;
>> +
>> fence = &p->jobs[i]->base.s_fence->scheduled;
>> - r = amdgpu_sync_fence(&leader->sync, fence);
>> + r = drm_sched_job_add_dependency(&leader->base, fence);
>> if (r)
>> goto error_cleanup;
>> }
> Note that I had to keep the declaration of "leader" in amdgpu_cs_sync_rings().
This and all your other merge resolutions look good to me.
And sorry for the noise, drm-tip somehow doesn't seem to work any more
and we had a lot of conflicting patches going in through -fixes and -next.
Regards,
Christian.
>