All of lore.kernel.org
 help / color / mirror / Atom feed
From: David Rosca <david.rosca@amd.com>
To: Alex Deucher <alexander.deucher@amd.com>, amd-gfx@lists.freedesktop.org
Subject: Re: [PATCH] drm/amdgpu: handle pipeline sync without a VM fence
Date: Fri, 14 Aug 2026 19:14:25 +0200	[thread overview]
Message-ID: <3af940dd-adcf-43d2-a825-241ca54062b2@amd.com> (raw)
In-Reply-To: <20260814154212.2688253-1-alexander.deucher@amd.com>


On 8/14/26 17:42, Alex Deucher wrote:
> If we end up emitting a VM fence keep pipeline sync
> associated with that fence.  If not, emit them as
> part of the IB fence.
>
> Cc: David Rosca <david.rosca@amd.com>
> Fixes: cb1e657ccac8 ("drm/amdgpu: handle GDS and SPM without a VM fence")
> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
> ---
>   drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c | 6 +++++-
>   drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 6 ++++--
>   drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h | 2 +-
>   3 files changed, 10 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
> index da4dc489e80bd..360e6f00cb7c0 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
> @@ -222,7 +222,7 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned int num_ibs,
>   		vm_af = job->hw_vm_fence;
>   		/* VM sequence */
>   		vm_af->ib_wptr = ring->wptr;
> -		amdgpu_vm_flush(ring, job, need_pipe_sync, &emit_spm_needed,
> +		amdgpu_vm_flush(ring, job, &need_pipe_sync, &emit_spm_needed,
>   				&emit_gds_needed);
>   		vm_af->ib_dw_size =
>   			amdgpu_ring_get_dw_distance(ring, vm_af->ib_wptr, ring->wptr);
> @@ -235,6 +235,10 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned int num_ibs,
>   	if (ring->funcs->insert_start)
>   		ring->funcs->insert_start(ring);
>   
> +	/* this may have been handled by amdgpu_vm_flush */
> +	if (need_pipe_sync)
> +		amdgpu_ring_emit_pipeline_sync(ring);
> +
>   	if (emit_spm_needed)
>   		adev->gfx.rlc.funcs->update_spm_vmid(adev, ring->xcc_id, ring, job->vmid);
>   
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
> index 71050a86bcc3a..4bd9368b13d97 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
> @@ -772,7 +772,7 @@ bool amdgpu_vm_need_pipeline_sync(struct amdgpu_ring *ring,
>    * Emit a VM flush when it is necessary.
>    */
>   void amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job,
> -		     bool need_pipe_sync, bool *emit_spm_needed,
> +		     bool *need_pipe_sync, bool *emit_spm_needed,
>   		     bool *emit_gds_needed)
>   {
>   	struct amdgpu_device *adev = ring->adev;
need_pipe_sync should also be removed from this condition below:

     if (!vm_flush_needed && !gds_switch_needed && !need_pipe_sync &&
         !cleaner_shader_needed && !spm_update_needed)
         return;

With that, Reviewed-by: David Rosca <david.rosca@amd.com>

> @@ -847,8 +847,10 @@ void amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job,
>   		patch = amdgpu_ring_init_cond_exec(ring,
>   						   ring->cond_exe_gpu_addr);
>   
> -	if (need_pipe_sync)
> +	if (emit_fence && *need_pipe_sync) {
>   		amdgpu_ring_emit_pipeline_sync(ring);
> +		*need_pipe_sync = false;
> +	}
>   
>   	if (cleaner_shader_needed)
>   		ring->funcs->emit_cleaner_shader(ring);
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
> index 7f2ba728e3ed3..d32183cd9e0fc 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
> @@ -512,7 +512,7 @@ int amdgpu_vm_validate(struct amdgpu_device *adev, struct amdgpu_vm *vm,
>   		       int (*callback)(void *p, struct amdgpu_bo *bo),
>   		       void *param);
>   void amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job,
> -		     bool need_pipe_sync, bool *emit_spm_needed,
> +		     bool *need_pipe_sync, bool *emit_spm_needed,
>   		     bool *emit_gds_needed);
>   int amdgpu_vm_update_pdes(struct amdgpu_device *adev,
>   			  struct amdgpu_vm *vm, bool immediate);

  reply	other threads:[~2026-08-14 17:14 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-08-14 15:42 [PATCH] drm/amdgpu: handle pipeline sync without a VM fence Alex Deucher
2026-08-14 17:14 ` David Rosca [this message]
2026-08-14 17:28   ` Alex Deucher
  -- strict thread matches above, loose matches on Subject: below --
2026-08-14 17:29 Alex Deucher

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=3af940dd-adcf-43d2-a825-241ca54062b2@amd.com \
    --to=david.rosca@amd.com \
    --cc=alexander.deucher@amd.com \
    --cc=amd-gfx@lists.freedesktop.org \
    /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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.