public inbox for intel-xe@lists.freedesktop.org
 help / color / mirror / Atom feed
From: Jani Nikula <jani.nikula@linux.intel.com>
To: Ville Syrjala <ville.syrjala@linux.intel.com>,
	intel-gfx@lists.freedesktop.org
Cc: intel-xe@lists.freedesktop.org
Subject: Re: [PATCH 06/16] drm/i915: Introduce i915_fb_pin_ggtt_(un)pin()
Date: Fri, 24 Apr 2026 12:59:32 +0300	[thread overview]
Message-ID: <e954c272bba4200dc4147dafd024eb1943c043fb@intel.com> (raw)
In-Reply-To: <20260423165346.20884-7-ville.syrjala@linux.intel.com>

On Thu, 23 Apr 2026, Ville Syrjala <ville.syrjala@linux.intel.com> wrote:
> From: Ville Syrjälä <ville.syrjala@linux.intel.com>
>
> Extract the inner DPT parts of intel_plane_(un)pin() into the
> i915_fb_pin_ggtt_(un)pin(). These will become part of the new
> fb_pin parent interface.
>
> Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>

Reviewed-by: Jani Nikula <jani.nikula@intel.com>

> ---
>  drivers/gpu/drm/i915/i915_fb_pin.c | 68 +++++++++++++++++++-----------
>  1 file changed, 44 insertions(+), 24 deletions(-)
>
> diff --git a/drivers/gpu/drm/i915/i915_fb_pin.c b/drivers/gpu/drm/i915/i915_fb_pin.c
> index 0019f570d843..e14d2c1fb81f 100644
> --- a/drivers/gpu/drm/i915/i915_fb_pin.c
> +++ b/drivers/gpu/drm/i915/i915_fb_pin.c
> @@ -257,6 +257,41 @@ intel_plane_fb_vtd_guard(const struct intel_plane_state *plane_state)
>  				       plane_state->hw.rotation);
>  }
>  
> +static int i915_fb_pin_ggtt_pin(struct drm_gem_object *obj,
> +				const struct intel_fb_pin_params *pin_params,
> +				struct i915_vma **out_ggtt_vma,
> +				u32 *out_offset,
> +				int *out_fence_id)
> +{
> +	struct i915_vma *ggtt_vma;
> +
> +	ggtt_vma = intel_fb_pin_to_ggtt(obj, pin_params, out_fence_id);
> +	if (IS_ERR(ggtt_vma))
> +		return PTR_ERR(ggtt_vma);
> +
> +	*out_ggtt_vma = ggtt_vma;
> +
> +	/*
> +	 * Pre-populate the dma address before we enter the vblank
> +	 * evade critical section as i915_gem_object_get_dma_address()
> +	 * will trigger might_sleep() even if it won't actually sleep,
> +	 * which is the case when the fb has already been pinned.
> +	 */
> +	if (pin_params->needs_physical)
> +		*out_offset = i915_gem_object_get_dma_address(to_intel_bo(obj), 0);
> +	else
> +		*out_offset = i915_ggtt_offset(ggtt_vma);
> +
> +	return 0;
> +}
> +
> +static void i915_fb_pin_ggtt_unpin(struct i915_vma *ggtt_vma,
> +				   int fence_id)
> +{
> +	if (ggtt_vma)
> +		intel_fb_unpin_vma(ggtt_vma, fence_id);
> +}
> +
>  static int i915_fb_pin_dpt_pin(struct drm_gem_object *obj, struct intel_dpt *dpt,
>  			       const struct intel_fb_pin_params *pin_params,
>  			       struct i915_vma **out_dpt_vma,
> @@ -325,24 +360,11 @@ int intel_plane_pin_fb(struct intel_plane_state *plane_state,
>  			.needs_fence = intel_plane_needs_fence(display),
>  		};
>  
> -		ggtt_vma = intel_fb_pin_to_ggtt(intel_fb_bo(&fb->base), &pin_params,
> -						intel_plane_uses_fence(plane_state) ? &fence_id : NULL);
> -		if (IS_ERR(ggtt_vma))
> -			return PTR_ERR(ggtt_vma);
> -
> -		/*
> -		 * Pre-populate the dma address before we enter the vblank
> -		 * evade critical section as i915_gem_object_get_dma_address()
> -		 * will trigger might_sleep() even if it won't actually sleep,
> -		 * which is the case when the fb has already been pinned.
> -		 */
> -		if (intel_plane_needs_physical(plane)) {
> -			struct drm_i915_gem_object *obj = to_intel_bo(intel_fb_bo(&fb->base));
> -
> -			offset = i915_gem_object_get_dma_address(obj, 0);
> -		} else {
> -			offset = i915_ggtt_offset(ggtt_vma);
> -		}
> +		ret = i915_fb_pin_ggtt_pin(intel_fb_bo(&fb->base),
> +					   &pin_params, &ggtt_vma, &offset,
> +					   intel_plane_uses_fence(plane_state) ? &fence_id : NULL);
> +		if (ret)
> +			return ret;
>  	} else {
>  		struct intel_fb_pin_params pin_params = {
>  			.view = &plane_state->view.gtt,
> @@ -371,13 +393,11 @@ void intel_plane_unpin_fb(struct intel_plane_state *old_plane_state)
>  		to_intel_framebuffer(old_plane_state->hw.fb);
>  
>  	if (!intel_fb_uses_dpt(&fb->base)) {
> -		struct i915_vma *vma;
> +		i915_fb_pin_ggtt_unpin(old_plane_state->ggtt_vma,
> +				       old_plane_state->fence_id);
>  
> -		vma = fetch_and_zero(&old_plane_state->ggtt_vma);
> -		if (vma) {
> -			intel_fb_unpin_vma(vma, old_plane_state->fence_id);
> -			old_plane_state->fence_id = -1;
> -		}
> +		old_plane_state->ggtt_vma = NULL;
> +		old_plane_state->fence_id = -1;
>  	} else {
>  		i915_fb_pin_dpt_unpin(fb->dpt,
>  				      old_plane_state->dpt_vma,

-- 
Jani Nikula, Intel

  reply	other threads:[~2026-04-24  9:59 UTC|newest]

Thread overview: 36+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-04-23 16:53 [PATCH 00/16] drm/i915: Introduce 'fb_pin' parent interface Ville Syrjala
2026-04-23 16:53 ` [PATCH 01/16] drm/i915: Introduce intel_parent_fb_pin_get_map() Ville Syrjala
2026-04-24  9:14   ` Jani Nikula
2026-04-23 16:53 ` [PATCH 02/16] drm/i915: Move intel_fb_pin_params to the parent interface Ville Syrjala
2026-04-24  9:19   ` Jani Nikula
2026-04-23 16:53 ` [PATCH 03/16] drm/i915: Move the i915_dpt_offset()==0 assert Ville Syrjala
2026-04-24  9:19   ` Jani Nikula
2026-04-23 16:53 ` [PATCH 04/16] drm/i915: Reorganize intel_plane_pin_fb() a bit Ville Syrjala
2026-04-24  9:55   ` Jani Nikula
2026-04-23 16:53 ` [PATCH 05/16] drm/i915: Introduce i915_fb_pin_dpt_(un)pin() Ville Syrjala
2026-04-24  9:56   ` Jani Nikula
2026-04-23 16:53 ` [PATCH 06/16] drm/i915: Introduce i915_fb_pin_ggtt_(un)pin() Ville Syrjala
2026-04-24  9:59   ` Jani Nikula [this message]
2026-04-23 16:53 ` [PATCH 07/16] drm/xe: Move the FORCE_WC assert into __xe_pin_fb_vma() Ville Syrjala
2026-04-24 10:02   ` Jani Nikula
2026-04-23 16:53 ` [PATCH 08/16] drm/xe: Kill the fbdev vma reuse hack Ville Syrjala
2026-04-24 10:06   ` Jani Nikula
2026-04-23 16:53 ` [PATCH 09/16] drm/xe: Reorganize intel_plane_pin_fb() a bit Ville Syrjala
2026-04-24 10:07   ` Jani Nikula
2026-04-23 16:53 ` [PATCH 10/16] drm/xe: Introduce xe_fb_pin_dpt_(un)pin() Ville Syrjala
2026-04-24 10:12   ` Jani Nikula
2026-04-23 16:53 ` [PATCH 11/16] drm/xe: Introduce xe_fb_pin_ggtt_(un)pin() Ville Syrjala
2026-04-24 10:13   ` Jani Nikula
2026-04-23 16:53 ` [PATCH 12/16] drm/xe: Restructure reuse_vma() Ville Syrjala
2026-04-24 10:30   ` Jani Nikula
2026-04-23 16:53 ` [PATCH 13/16] drm/i915: Introduce the main fb_pin parent interface Ville Syrjala
2026-04-24 10:48   ` Jani Nikula
2026-04-24 10:55     ` Ville Syrjälä
2026-04-23 16:53 ` [PATCH 14/16] drm/i915/fbdev: Use intel_parent_fb_pin_ggtt_(un)pin() Ville Syrjala
2026-04-24 12:54   ` Jani Nikula
2026-04-23 16:53 ` [PATCH 15/16] drm/xe: Use xe_fb_pin_ggtt_pin() for the initial FB pin Ville Syrjala
2026-04-24 12:58   ` Jani Nikula
2026-04-23 16:53 ` [PATCH 16/16] drm/i915: Consolidate the intel_plane_(un)pin_fb() implementations Ville Syrjala
2026-04-24 13:30   ` Jani Nikula
2026-04-23 17:01 ` ✗ CI.checkpatch: warning for drm/i915: Introduce 'fb_pin' parent interface Patchwork
2026-04-23 17:02 ` ✓ CI.KUnit: success " Patchwork

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=e954c272bba4200dc4147dafd024eb1943c043fb@intel.com \
    --to=jani.nikula@linux.intel.com \
    --cc=intel-gfx@lists.freedesktop.org \
    --cc=intel-xe@lists.freedesktop.org \
    --cc=ville.syrjala@linux.intel.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