KonstantinSeurer
/
mesa
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

egl/wayland: use surface dma-buf feedback to allocate surface buffers 

As explained in "egl/wayland: add initial dma-buf feedback support", we
still don't use the per-surface dma-buf feedback. In this patch we start
to use it.

If per-surface dma-buf feedback is advertised, use it to allocate
surface buffers. Also, the dma-buf protocol states that the feedback is
resent only when the client is using a suboptimal format/modifier pair.
So listen for new per-surface feedback events and reallocate the surface
buffers based on them. We can't change the format of a buffer, but we
can pick a new modifier.

This patch is based on previous work of Scott Anderson (@ascent).

Signed-off-by: Scott Anderson <scott.anderson@collabora.com>
Signed-off-by: Leandro Ribeiro <leandro.ribeiro@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11248>
This commit is contained in:
Leandro Ribeiro 2021-07-26 17:53:24 -03:00 committed by Marge Bot
parent 5b56fc748a
commit b5848b2dac
2 changed files with 78 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
src/egl/drivers/dri2/egl_dri2.h
View File

@ -332,6 +332,7 @@ struct dri2_egl_surface
bool compositor_using_another_device;
int format;
bool resized;
bool received_dmabuf_feedback;
#endif
#ifdef HAVE_DRM_PLATFORM

81
src/egl/drivers/dri2/platform_wayland.c
View File

@ -558,6 +558,13 @@ surface_dmabuf_feedback_done(void *data,
{
struct dri2_egl_surface *dri2_surf = data;
/* The dma-buf feedback protocol states that surface dma-buf feedback should
* be sent by the compositor only if its buffers are using a suboptimal pair
* of format and modifier. We can't change the buffer format, but we can
* reallocate with another modifier. So we raise this flag in order to force
* buffer reallocation based on the dma-buf feedback sent. */
dri2_surf->received_dmabuf_feedback = true;
dmabuf_feedback_fini(&dri2_surf->dmabuf_feedback);
dri2_surf->dmabuf_feedback = dri2_surf->pending_dmabuf_feedback;
dmabuf_feedback_init(&dri2_surf->pending_dmabuf_feedback);
@ -878,6 +885,68 @@ create_dri_image_diff_gpu(struct dri2_egl_surface *dri2_surf,
&linear_mod, 1, NULL);
}
static void
create_dri_image_from_dmabuf_feedback(struct dri2_egl_surface *dri2_surf,
unsigned int dri_image_format, uint32_t use_flags)
{
struct dri2_egl_display *dri2_dpy =
dri2_egl_display(dri2_surf->base.Resource.Display);
int visual_idx;
uint64_t *modifiers;
unsigned int num_modifiers;
uint32_t flags;
/* We don't have valid dma-buf feedback, so return */
if (dri2_surf->dmabuf_feedback.main_device == 0)
return;
visual_idx = dri2_wl_visual_idx_from_fourcc(dri2_surf->format);
assert(visual_idx != -1);
/* Iterates through the dma-buf feedback to pick a new set of modifiers. The
* tranches are sent in descending order of preference by the compositor, so
* the first set that we can pick is the best one. For now we still can't
* specify the target device in order to make the render device try its best
* to allocate memory that can be directly scanned out by the KMS device. But
* in the future this may change (newer versions of
* createImageWithModifiers). Also, we are safe to pick modifiers from
* tranches whose target device differs from the main device, as compositors
* do not expose (in dma-buf feedback tranches) formats/modifiers that are
* incompatible with the main device. */
util_dynarray_foreach(&dri2_surf->dmabuf_feedback.tranches,
struct dmabuf_feedback_tranche, tranche) {
/* Ignore tranches that do not contain dri2_surf->format */
if (!BITSET_TEST(tranche->formats.formats_bitmap, visual_idx))
continue;
modifiers = util_dynarray_begin(&tranche->formats.modifiers[visual_idx]);
num_modifiers = util_dynarray_num_elements(&tranche->formats.modifiers[visual_idx],
uint64_t);
/* For the purposes of this function, an INVALID modifier on
* its own means the modifiers aren't supported. */
if (num_modifiers == 0 ||
(num_modifiers == 1 && modifiers[0] == DRM_FORMAT_MOD_INVALID)) {
num_modifiers = 0;
modifiers = NULL;
}
flags = use_flags;
if (tranche->flags & ZWP_LINUX_DMABUF_FEEDBACK_V1_TRANCHE_FLAGS_SCANOUT)
flags |= __DRI_IMAGE_USE_SCANOUT;
dri2_surf->back->dri_image =
loader_dri_create_image(dri2_dpy->dri_screen, dri2_dpy->image,
dri2_surf->base.Width,
dri2_surf->base.Height,
dri_image_format,
dri2_dpy->is_different_gpu ? 0 : flags,
modifiers, num_modifiers, NULL);
if (dri2_surf->back->dri_image)
return;
}
}
static void
create_dri_image(struct dri2_egl_surface *dri2_surf,
unsigned int dri_image_format, uint32_t use_flags)
@ -987,7 +1056,10 @@ get_back_bo(struct dri2_egl_surface *dri2_surf)
}
if (dri2_surf->back->dri_image == NULL) {
create_dri_image(dri2_surf, dri_image_format, use_flags);
if (dri2_surf->wl_dmabuf_feedback)
create_dri_image_from_dmabuf_feedback(dri2_surf, dri_image_format, use_flags);
if (dri2_surf->back->dri_image == NULL)
create_dri_image(dri2_surf, dri_image_format, use_flags);
dri2_surf->back->age = 0;
}
@ -1036,9 +1108,10 @@ update_buffers(struct dri2_egl_surface *dri2_surf)
dri2_surf->dy = dri2_surf->wl_win->dy;
}
if (dri2_surf->resized) {
dri2_wl_release_buffers(dri2_surf);
dri2_surf->resized = false;
if (dri2_surf->resized || dri2_surf->received_dmabuf_feedback) {
dri2_wl_release_buffers(dri2_surf);
dri2_surf->resized = false;
dri2_surf->received_dmabuf_feedback = false;
}
if (get_back_bo(dri2_surf) < 0) {