[Intel-gfx] [PATCH xf86-video-intel] sna: Forcibly relinquish the GPU bo cache of a SHM pixmap on flushing

[Intel-gfx] [PATCH xf86-video-intel] sna: Forcibly relinquish the GPU bo cache of a SHM pixmap on flushing

[Chris Wilson]

Before we indicate return control of the SHM Pixmap to the client (that
is prior to the next XReply), we ensure that the original SHM buffer is
uptodate with any changes made on the GPU. We must flush the GPU writes
back to the CPU and so not allow ourselves to keep the dirty cache of
the CPU bo.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Alexei Podtelezhnikov <apodtele@gmail.com>
---
 src/sna/sna_accel.c | 12 ++++++++++--
 1 file changed, 10 insertions(+), 2 deletions(-)

diff --git a/src/sna/sna_accel.c b/src/sna/sna_accel.c
index edbfe5a4a..6e14cf7b4 100644
--- a/src/sna/sna_accel.c
+++ b/src/sna/sna_accel.c
@@ -2119,6 +2119,11 @@ static inline bool operate_inplace(struct sna_pixmap *priv, unsigned flags)
 	if (!USE_INPLACE)
 		return false;
 
+	if (flags & __MOVE_FORCE) {
+		DBG(("%s: no, inplace operation denied by force\n", __FUNCTION__));
+		return false;
+	}
+
 	if ((flags & MOVE_INPLACE_HINT) == 0) {
 		DBG(("%s: no, inplace operation not suitable\n", __FUNCTION__));
 		return false;
@@ -2327,7 +2332,7 @@ skip_inplace_map:
 
 	sna_pixmap_unmap(pixmap, priv);
 
-	if (USE_INPLACE &&
+	if (USE_INPLACE && !(flags & __MOVE_FORCE) &&
 	    (flags & MOVE_WRITE ? (void *)priv->gpu_bo : (void *)priv->gpu_damage) && priv->cpu_damage == NULL &&
 	    priv->gpu_bo->tiling == I915_TILING_NONE &&
 	    (flags & MOVE_READ || kgem_bo_can_map__cpu(&sna->kgem, priv->gpu_bo, flags & MOVE_WRITE)) &&
@@ -17409,7 +17414,10 @@ void sna_accel_flush(struct sna *sna)
 			     priv->pixmap->refcnt));
 			assert(!priv->flush);
 			ret = sna_pixmap_move_to_cpu(priv->pixmap,
-						     MOVE_READ | MOVE_WRITE);
+						     MOVE_READ |
+						     MOVE_WRITE |
+						     MOVE_WHOLE_HINT |
+						     __MOVE_FORCE);
 			assert(!ret || priv->gpu_bo == NULL);
 			if (priv->pixmap->refcnt == 0) {
 				sna_damage_destroy(&priv->cpu_damage);
-- 
2.26.0

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

Re: [Intel-gfx] [PATCH xf86-video-intel] sna: Forcibly relinquish the GPU bo cache of a SHM pixmap on flushing

[Alexei Podtelezhnikov]

On Fri, Apr 17, 2020 at 10:05 AM Chris Wilson <chris@chris-wilson.co.uk> wrote:
>
> Before we indicate return control of the SHM Pixmap to the client (that
> is prior to the next XReply), we ensure that the original SHM buffer is
> uptodate with any changes made on the GPU. We must flush the GPU writes
> back to the CPU and so not allow ourselves to keep the dirty cache of
> the CPU bo.
>
> Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
> Cc: Alexei Podtelezhnikov <apodtele@gmail.com>

The failed assertion is gone.

Tested-by: Alexei Podtelezhnikov <apodtele@gmail.com>

> ---
>  src/sna/sna_accel.c | 12 ++++++++++--
>  1 file changed, 10 insertions(+), 2 deletions(-)
>
> diff --git a/src/sna/sna_accel.c b/src/sna/sna_accel.c
> index edbfe5a4a..6e14cf7b4 100644
> --- a/src/sna/sna_accel.c
> +++ b/src/sna/sna_accel.c
> @@ -2119,6 +2119,11 @@ static inline bool operate_inplace(struct sna_pixmap *priv, unsigned flags)
>         if (!USE_INPLACE)
>                 return false;
>
> +       if (flags & __MOVE_FORCE) {
> +               DBG(("%s: no, inplace operation denied by force\n", __FUNCTION__));
> +               return false;
> +       }
> +
>         if ((flags & MOVE_INPLACE_HINT) == 0) {
>                 DBG(("%s: no, inplace operation not suitable\n", __FUNCTION__));
>                 return false;
> @@ -2327,7 +2332,7 @@ skip_inplace_map:
>
>         sna_pixmap_unmap(pixmap, priv);
>
> -       if (USE_INPLACE &&
> +       if (USE_INPLACE && !(flags & __MOVE_FORCE) &&
>             (flags & MOVE_WRITE ? (void *)priv->gpu_bo : (void *)priv->gpu_damage) && priv->cpu_damage == NULL &&
>             priv->gpu_bo->tiling == I915_TILING_NONE &&
>             (flags & MOVE_READ || kgem_bo_can_map__cpu(&sna->kgem, priv->gpu_bo, flags & MOVE_WRITE)) &&
> @@ -17409,7 +17414,10 @@ void sna_accel_flush(struct sna *sna)
>                              priv->pixmap->refcnt));
>                         assert(!priv->flush);
>                         ret = sna_pixmap_move_to_cpu(priv->pixmap,
> -                                                    MOVE_READ | MOVE_WRITE);
> +                                                    MOVE_READ |
> +                                                    MOVE_WRITE |
> +                                                    MOVE_WHOLE_HINT |
> +                                                    __MOVE_FORCE);
>                         assert(!ret || priv->gpu_bo == NULL);
>                         if (priv->pixmap->refcnt == 0) {
>                                 sna_damage_destroy(&priv->cpu_damage);
> --
> 2.26.0
>
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

[Intel-gfx] ✗ Fi.CI.BUILD: failure for sna: Forcibly relinquish the GPU bo cache of a SHM pixmap on flushing

[Patchwork]

== Series Details ==

Series: sna: Forcibly relinquish the GPU bo cache of a SHM pixmap on flushing
URL   : https://patchwork.freedesktop.org/series/76099/
State : failure

== Summary ==

Applying: sna: Forcibly relinquish the GPU bo cache of a SHM pixmap on flushing
error: sha1 information is lacking or useless (src/sna/sna_accel.c).
error: could not build fake ancestor
hint: Use 'git am --show-current-patch=diff' to see the failed patch
Patch failed at 0001 sna: Forcibly relinquish the GPU bo cache of a SHM pixmap on flushing
When you have resolved this problem, run "git am --continue".
If you prefer to skip this patch, run "git am --skip" instead.
To restore the original branch and stop patching, run "git am --abort".

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx