public inbox for drm-ai-reviews@public-inbox.freedesktop.org
 help / color / mirror / Atom feed
From: Christian König <christian.koenig@amd.com>
To: Ulisses Paixao <ulissespaixao@usp.br>,
	alexander.deucher@amd.com, airlied@gmail.com, simona@ffwll.ch
Cc: Felipe Sousa <felipesousa@usp.br>,
	amd-gfx@lists.freedesktop.org, dri-devel@lists.freedesktop.org
Subject: Re: [PATCH] drm/amd/amdgpu: remove duplicated code in gfx_v11 and gfx_v12
Date: Thu, 30 Apr 2026 08:21:56 +0200	[thread overview]
Message-ID: <dda47c3b-1303-4f10-b375-1f3eab032ee5@amd.com> (raw)
In-Reply-To: <20260429173646.7653-1-ulissespaixao@usp.br>



On 4/29/26 19:36, Ulisses Paixao wrote:
> [Sie erhalten nicht häufig E-Mails von ulissespaixao@usp.br. Weitere Informationen, warum dies wichtig ist, finden Sie unter https://aka.ms/LearnAboutSenderIdentification ]
> 
> The functions gfx_v11_0_handle_priv_fault and gfx_v12_0_handle_priv_fault
> are identical. This patch replaces them with a single implementation in
> amdgpu_gfx, called amdgpu_gfx_handle_priv_fault, to reduce code
> duplication.
> 
> Signed-off-by: Ulisses Paixao <ulissespaixao@usp.br>
> Co-developed-by: Felipe Sousa <felipesousa@usp.br>
> Signed-off-by: Felipe Sousa <felipesousa@usp.br>
> ---
>  drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c | 46 +++++++++++++++++++++++++
>  drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h |  2 ++
>  drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c  | 43 ++---------------------
>  drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c  | 43 ++---------------------
>  4 files changed, 54 insertions(+), 80 deletions(-)
> 
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
> index b8ca87669..c8d769cb0 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
> @@ -830,6 +830,52 @@ int amdgpu_gfx_enable_kgq(struct amdgpu_device *adev, int xcc_id)
>         return r;
>  }
> 
> +/**
> + * amdgpu_gfx_handle_priv_fault - Handle privileged instruction fault
> + *
> + * @adev: amdgpu_device pointer
> + * @entry: interrupt vector entry from the hardware
> + *
> + * This function handles privileged instruction faults by identifying
> + * the faulty ring (gfx or compute) and triggering a scheduler fault.
> + */
> +void amdgpu_gfx_handle_priv_fault(struct amdgpu_device *adev,
> +                                       struct amdgpu_iv_entry *entry)
> +{
> +       u8 me_id, pipe_id, queue_id;
> +       struct amdgpu_ring *ring;
> +       int i;
> +
> +       me_id = (entry->ring_id & 0x0c) >> 2;
> +       pipe_id = (entry->ring_id & 0x03) >> 0;
> +       queue_id = (entry->ring_id & 0x70) >> 4;

Even when they are identical on gfx11 and gfx12 this decoding here is HW specific and doesn't belong here.

> +
> +       if (!adev->gfx.disable_kq) {

That check can probably be removed. Both num_gfx_rings and num_compute_rings should be zero in that case.

> +               switch (me_id) {
> +               case 0:
> +                       for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
> +                               ring = &adev->gfx.gfx_ring[i];
> +                               if (ring->me == me_id && ring->pipe == pipe_id &&
> +                                   ring->queue == queue_id)
> +                                       drm_sched_fault(&ring->sched);
> +                       }
> +                       break;
> +               case 1:
> +               case 2:
> +                       for (i = 0; i < adev->gfx.num_compute_rings; i++) {
> +                               ring = &adev->gfx.compute_ring[i];
> +                               if (ring->me == me_id && ring->pipe == pipe_id &&
> +                                   ring->queue == queue_id)
> +                                       drm_sched_fault(&ring->sched);
> +                       }
> +                       break;
> +               default:
> +                       BUG();
> +                       break;
> +               }
> +       }

This part can be moved into amdgpu_gfx. But I would remove the switch (me_id) part and just go over all gfx and compute rings to search for the matching me and pipe.

Regards,
Christian.

> +}
> +
>  static void amdgpu_gfx_do_off_ctrl(struct amdgpu_device *adev, bool enable,
>                                    bool no_delay)
>  {
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
> index a0cf0a3b4..5655af43d 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
> @@ -611,6 +611,8 @@ bool amdgpu_gfx_is_high_priority_graphics_queue(struct amdgpu_device *adev,
>                                                 struct amdgpu_ring *ring);
>  bool amdgpu_gfx_is_me_queue_enabled(struct amdgpu_device *adev, int me,
>                                     int pipe, int queue);
> +void amdgpu_gfx_handle_priv_fault(struct amdgpu_device *adev,
> +                                       struct amdgpu_iv_entry *entry);
>  void amdgpu_gfx_off_ctrl(struct amdgpu_device *adev, bool enable);
>  void amdgpu_gfx_off_ctrl_immediate(struct amdgpu_device *adev, bool enable);
>  int amdgpu_get_gfx_off_status(struct amdgpu_device *adev, uint32_t *value);
> diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c
> index 2c6f1e25c..da869f928 100644
> --- a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c
> +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c
> @@ -6684,49 +6684,12 @@ static int gfx_v11_0_set_priv_inst_fault_state(struct amdgpu_device *adev,
>         return 0;
>  }
> 
> -static void gfx_v11_0_handle_priv_fault(struct amdgpu_device *adev,
> -                                       struct amdgpu_iv_entry *entry)
> -{
> -       u8 me_id, pipe_id, queue_id;
> -       struct amdgpu_ring *ring;
> -       int i;
> -
> -       me_id = (entry->ring_id & 0x0c) >> 2;
> -       pipe_id = (entry->ring_id & 0x03) >> 0;
> -       queue_id = (entry->ring_id & 0x70) >> 4;
> -
> -       if (!adev->gfx.disable_kq) {
> -               switch (me_id) {
> -               case 0:
> -                       for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
> -                               ring = &adev->gfx.gfx_ring[i];
> -                               if (ring->me == me_id && ring->pipe == pipe_id &&
> -                                   ring->queue == queue_id)
> -                                       drm_sched_fault(&ring->sched);
> -                       }
> -                       break;
> -               case 1:
> -               case 2:
> -                       for (i = 0; i < adev->gfx.num_compute_rings; i++) {
> -                               ring = &adev->gfx.compute_ring[i];
> -                               if (ring->me == me_id && ring->pipe == pipe_id &&
> -                                   ring->queue == queue_id)
> -                                       drm_sched_fault(&ring->sched);
> -                       }
> -                       break;
> -               default:
> -                       BUG();
> -                       break;
> -               }
> -       }
> -}
> -
>  static int gfx_v11_0_priv_reg_irq(struct amdgpu_device *adev,
>                                   struct amdgpu_irq_src *source,
>                                   struct amdgpu_iv_entry *entry)
>  {
>         DRM_ERROR("Illegal register access in command stream\n");
> -       gfx_v11_0_handle_priv_fault(adev, entry);
> +       amdgpu_gfx_handle_priv_fault(adev, entry);
>         return 0;
>  }
> 
> @@ -6735,7 +6698,7 @@ static int gfx_v11_0_bad_op_irq(struct amdgpu_device *adev,
>                                 struct amdgpu_iv_entry *entry)
>  {
>         DRM_ERROR("Illegal opcode in command stream\n");
> -       gfx_v11_0_handle_priv_fault(adev, entry);
> +       amdgpu_gfx_handle_priv_fault(adev, entry);
>         return 0;
>  }
> 
> @@ -6744,7 +6707,7 @@ static int gfx_v11_0_priv_inst_irq(struct amdgpu_device *adev,
>                                    struct amdgpu_iv_entry *entry)
>  {
>         DRM_ERROR("Illegal instruction in command stream\n");
> -       gfx_v11_0_handle_priv_fault(adev, entry);
> +       amdgpu_gfx_handle_priv_fault(adev, entry);
>         return 0;
>  }
> 
> diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c
> index 6baac533a..883878e23 100644
> --- a/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c
> +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c
> @@ -5015,49 +5015,12 @@ static int gfx_v12_0_set_priv_inst_fault_state(struct amdgpu_device *adev,
>         return 0;
>  }
> 
> -static void gfx_v12_0_handle_priv_fault(struct amdgpu_device *adev,
> -                                       struct amdgpu_iv_entry *entry)
> -{
> -       u8 me_id, pipe_id, queue_id;
> -       struct amdgpu_ring *ring;
> -       int i;
> -
> -       me_id = (entry->ring_id & 0x0c) >> 2;
> -       pipe_id = (entry->ring_id & 0x03) >> 0;
> -       queue_id = (entry->ring_id & 0x70) >> 4;
> -
> -       if (!adev->gfx.disable_kq) {
> -               switch (me_id) {
> -               case 0:
> -                       for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
> -                               ring = &adev->gfx.gfx_ring[i];
> -                               if (ring->me == me_id && ring->pipe == pipe_id &&
> -                                   ring->queue == queue_id)
> -                                       drm_sched_fault(&ring->sched);
> -                       }
> -                       break;
> -               case 1:
> -               case 2:
> -                       for (i = 0; i < adev->gfx.num_compute_rings; i++) {
> -                               ring = &adev->gfx.compute_ring[i];
> -                               if (ring->me == me_id && ring->pipe == pipe_id &&
> -                                   ring->queue == queue_id)
> -                                       drm_sched_fault(&ring->sched);
> -                       }
> -                       break;
> -               default:
> -                       BUG();
> -                       break;
> -               }
> -       }
> -}
> -
>  static int gfx_v12_0_priv_reg_irq(struct amdgpu_device *adev,
>                                   struct amdgpu_irq_src *source,
>                                   struct amdgpu_iv_entry *entry)
>  {
>         DRM_ERROR("Illegal register access in command stream\n");
> -       gfx_v12_0_handle_priv_fault(adev, entry);
> +       amdgpu_gfx_handle_priv_fault(adev, entry);
>         return 0;
>  }
> 
> @@ -5066,7 +5029,7 @@ static int gfx_v12_0_bad_op_irq(struct amdgpu_device *adev,
>                                 struct amdgpu_iv_entry *entry)
>  {
>         DRM_ERROR("Illegal opcode in command stream\n");
> -       gfx_v12_0_handle_priv_fault(adev, entry);
> +       amdgpu_gfx_handle_priv_fault(adev, entry);
>         return 0;
>  }
> 
> @@ -5075,7 +5038,7 @@ static int gfx_v12_0_priv_inst_irq(struct amdgpu_device *adev,
>                                    struct amdgpu_iv_entry *entry)
>  {
>         DRM_ERROR("Illegal instruction in command stream\n");
> -       gfx_v12_0_handle_priv_fault(adev, entry);
> +       amdgpu_gfx_handle_priv_fault(adev, entry);
>         return 0;
>  }
> 
> --
> 2.34.1
> 


  reply	other threads:[~2026-04-30  6:22 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-04-29 17:36 [PATCH] drm/amd/amdgpu: remove duplicated code in gfx_v11 and gfx_v12 Ulisses Paixao
2026-04-30  6:21 ` Christian König [this message]
2026-05-05  0:53   ` Claude review: " Claude Code Review Bot
2026-05-05  0:53   ` Claude Code Review Bot

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=dda47c3b-1303-4f10-b375-1f3eab032ee5@amd.com \
    --to=christian.koenig@amd.com \
    --cc=airlied@gmail.com \
    --cc=alexander.deucher@amd.com \
    --cc=amd-gfx@lists.freedesktop.org \
    --cc=dri-devel@lists.freedesktop.org \
    --cc=felipesousa@usp.br \
    --cc=simona@ffwll.ch \
    --cc=ulissespaixao@usp.br \
    /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