weston: update bbappend 9.0.0 -> 10.0.2
PATCH 1/1 weston: update bbappend 9.0.0 -> 10.0.2
To match upstream oe-core/master.
---
...equire-GL_EXT_unpack_subimage-commit.patch | 91 +++++++++++++++++++
...ct-plane-based-on-current-attached-C.patch | 52 +++++------
..._9.0.0.bbappend => weston_10.0.2.bbappend} | 3 +-
3 files changed, 115 insertions(+), 31 deletions(-)
create mode 100644 meta-arago-distro/recipes-graphics/wayland/weston/0001-Revert-require-GL_EXT_unpack_subimage-commit.patch
rename meta-arago-distro/recipes-graphics/wayland/{weston_9.0.0.bbappend => weston_10.0.2.bbappend} (78%)meta-arago-distro/recipes-graphics/wayland/weston/0001-Revert-require-GL_EXT_unpack_subimage-commit.patch new file +91 -0
@@ -0,0 +1,91 @@ +From 6558a9153cc81199146132041c61023a2c2e1b2e Mon Sep 17 00:00:00 2001 +From: Andrew Davis <[email protected]> +Date: Thu, 6 Oct 2022 15:49:31 -0500 +Subject: [PATCH] Revert require GL_EXT_unpack_subimage commit + +This reverts commit 593d5af43a8e2c2a3371088fa7ae430d0517c82d. + +That commit removed support for GPU drivers without GL_EXT_unpack_subimage +which SGX does not support. Add back support for GPUs without this +extension. + +Signed-off-by: Andrew Davis <[email protected]> +--- + libweston/renderer-gl/gl-renderer-internal.h | 2 ++ + libweston/renderer-gl/gl-renderer.c | 29 ++++++++++++++++---- + 2 files changed, 26 insertions(+), 5 deletions(-) + +diff --git a/libweston/renderer-gl/gl-renderer-internal.h b/libweston/renderer-gl/gl-renderer-internal.h +index 72101b47..7a6e2f48 100644 +--- a/libweston/renderer-gl/gl-renderer-internal.h ++++ b/libweston/renderer-gl/gl-renderer-internal.h +@@ -133,6 +133,8 @@ struct gl_renderer { + PFNEGLCREATEPLATFORMWINDOWSURFACEEXTPROC create_platform_window; + bool has_platform_base; + ++ bool has_unpack_subimage; ++ + PFNEGLBINDWAYLANDDISPLAYWL bind_display; + PFNEGLUNBINDWAYLANDDISPLAYWL unbind_display; + PFNEGLQUERYWAYLANDBUFFERWL query_buffer; +diff --git a/libweston/renderer-gl/gl-renderer.c b/libweston/renderer-gl/gl-renderer.c +index a5f5eae4..410ba85c 100644 +--- a/libweston/renderer-gl/gl-renderer.c ++++ b/libweston/renderer-gl/gl-renderer.c +@@ -1835,6 +1835,7 @@ gl_renderer_flush_damage(struct weston_surface *surface) + { + const struct weston_testsuite_quirks *quirks = + &surface->compositor->test_data.test_quirks; ++ struct gl_renderer *gr = get_renderer(surface->compositor); + struct gl_surface_state *gs = get_surface_state(surface); + struct weston_buffer *buffer = gs->buffer_ref.buffer; + struct weston_view *view; +@@ -1872,6 +1873,24 @@ gl_renderer_flush_damage(struct weston_surface *surface) + + glActiveTexture(GL_TEXTURE0); + ++ if (!gr->has_unpack_subimage) { ++ wl_shm_buffer_begin_access(buffer->shm_buffer); ++ for (j = 0; j < gs->num_textures; j++) { ++ glBindTexture(GL_TEXTURE_2D, gs->textures[j]); ++ glTexImage2D(GL_TEXTURE_2D, 0, ++ gs->gl_format[j], ++ gs->pitch / gs->hsub[j], ++ buffer->height / gs->vsub[j], ++ 0, ++ gl_format_from_internal(gs->gl_format[j]), ++ gs->gl_pixel_type, ++ data + gs->offset[j]); ++ } ++ wl_shm_buffer_end_access(buffer->shm_buffer); ++ ++ goto done; ++ } ++ + if (gs->needs_full_upload || quirks->gl_force_full_upload) { + glPixelStorei(GL_UNPACK_SKIP_PIXELS_EXT, 0); + glPixelStorei(GL_UNPACK_SKIP_ROWS_EXT, 0); +@@ -3932,11 +3951,9 @@ gl_renderer_setup(struct weston_compositor *ec, EGLSurface egl_surface) + else + ec->read_format = PIXMAN_a8b8g8r8; + +- if (gr->gl_version < gr_gl_version(3, 0) && +- !weston_check_egl_extension(extensions, "GL_EXT_unpack_subimage")) { +- weston_log("GL_EXT_unpack_subimage not available.\n"); +- return -1; +- } ++ if (gr->gl_version >= gr_gl_version(3, 0) || ++ weston_check_egl_extension(extensions, "GL_EXT_unpack_subimage")) ++ gr->has_unpack_subimage = true; + + if (gr->gl_version >= gr_gl_version(3, 0) || + weston_check_egl_extension(extensions, "GL_EXT_texture_type_2_10_10_10_REV")) +@@ -3977,6 +3994,8 @@ gl_renderer_setup(struct weston_compositor *ec, EGLSurface egl_surface) + gr_gl_version_minor(gr->gl_version)); + weston_log_continue(STAMP_SPACE "read-back format: %s\n", + ec->read_format == PIXMAN_a8r8g8b8 ? "BGRA" : "RGBA"); ++ weston_log_continue(STAMP_SPACE "wl_shm sub-image to texture: %s\n", ++ gr->has_unpack_subimage ? "yes" : "no"); + weston_log_continue(STAMP_SPACE "EGL Wayland extension: %s\n", + gr->has_bind_display ? "yes" : "no"); +
meta-arago-distro/recipes-graphics/wayland/weston/0001-backend-drm-Select-plane-based-on-current-attached-C.patch +22 -30
84 lines not on this page · raw message
meta-arago-distro/recipes-graphics/wayland/weston_10.0.2.bbappend renamed from meta-arago-distro/recipes-graphics/wayland/weston_9.0.0.bbappend +2 -1
@@ -1,10 +1,11 @@ FILESEXTRAPATHS:prepend := "${THISDIR}/${PN}:" -PR:append = ".arago3" +PR:append = ".arago1" # file://0002-weston-Allow-visual_id-to-be-0.patch SRC_URI += " \ file://0003-weston-Fix-virtual-keyboard-display-issue-for-QT5-ap.patch \ file://0004-weston-Fix-touch-screen-crash-issue.patch \ file://0001-backend-drm-Select-plane-based-on-current-attached-C.patch \ + file://0001-Revert-require-GL_EXT_unpack_subimage-commit.patch \ "
Signature · 1 line
2.38.1
Quoted 17 lines · Andrew Davis via lists.yoctoproject.org
To match upstream oe-core/master.
Signed-off-by Andrew Davis <[email protected]>--- ...equire-GL_EXT_unpack_subimage-commit.patch | 91 +++++++++++++++++++ ...ct-plane-based-on-current-attached-C.patch | 52 +++++------ ..._9.0.0.bbappend => weston_10.0.2.bbappend} | 3 +- 3 files changed, 115 insertions(+), 31 deletions(-) create mode 100644 meta-arago-distro/recipes-graphics/wayland/weston/0001-Revert-require-GL_EXT_unpack_subimage-commit.patch rename meta-arago-distro/recipes-graphics/wayland/{weston_9.0.0.bbappend => weston_10.0.2.bbappend} (78%)meta-arago-distro/recipes-graphics/wayland/weston/0001-Revert-require-GL_EXT_unpack_subimage-commit.patch new file +0
-0
meta-arago-distro/recipes-graphics/wayland/weston/0001-Revert-require-GL_EXT_unpack_subimage-commit.patch line 10
Quoted 10 lines
meta-arago-distro/recipes-graphics/wayland/weston/0001-Revert-require-GL_EXT_unpack_subimage-commit.patch new file +10
-0@@ -0,0 +1,91 @@ +From 6558a9153cc81199146132041c61023a2c2e1b2e Mon Sep 17 00:00:00 2001 +From: Andrew Davis <[email protected]> +Date: Thu, 6 Oct 2022 15:49:31 -0500 +Subject: [PATCH] Revert require GL_EXT_unpack_subimage commit + +This reverts commit 593d5af43a8e2c2a3371088fa7ae430d0517c82d. + +That commit removed support for GPU drivers without GL_EXT_unpack_subimage +which SGX does not support. Add back support for GPUs without this +extension.
The obvious question - how safe is it to assume GL_EXT_unpack_subimage is not being used some place else outside of this commit and not guarded by:
if (!gr->has_unpack_subimage)
Is this extention really in the driver, or just in a Mesa API translation layer? Could it be better fixed in the DDK?
Signature · 1 line
Denys
Quoted 193 lines
meta-arago-distro/recipes-graphics/wayland/weston/0001-Revert-require-GL_EXT_unpack_subimage-commit.patch new file +80
-0+Signed-off-by: Andrew Davis <[email protected]> +--- + libweston/renderer-gl/gl-renderer-internal.h | 2 ++ + libweston/renderer-gl/gl-renderer.c | 29 ++++++++++++++++---- + 2 files changed, 26 insertions(+), 5 deletions(-) + +diff --git a/libweston/renderer-gl/gl-renderer-internal.h b/libweston/renderer-gl/gl-renderer-internal.h +index 72101b47..7a6e2f48 100644 +--- a/libweston/renderer-gl/gl-renderer-internal.h ++++ b/libweston/renderer-gl/gl-renderer-internal.h +@@ -133,6 +133,8 @@ struct gl_renderer { + PFNEGLCREATEPLATFORMWINDOWSURFACEEXTPROC create_platform_window; + bool has_platform_base; + ++ bool has_unpack_subimage; ++ + PFNEGLBINDWAYLANDDISPLAYWL bind_display; + PFNEGLUNBINDWAYLANDDISPLAYWL unbind_display; + PFNEGLQUERYWAYLANDBUFFERWL query_buffer; +diff --git a/libweston/renderer-gl/gl-renderer.c b/libweston/renderer-gl/gl-renderer.c +index a5f5eae4..410ba85c 100644 +--- a/libweston/renderer-gl/gl-renderer.c ++++ b/libweston/renderer-gl/gl-renderer.c +@@ -1835,6 +1835,7 @@ gl_renderer_flush_damage(struct weston_surface *surface) + { + const struct weston_testsuite_quirks *quirks = + &surface->compositor->test_data.test_quirks; ++ struct gl_renderer *gr = get_renderer(surface->compositor); + struct gl_surface_state *gs = get_surface_state(surface); + struct weston_buffer *buffer = gs->buffer_ref.buffer; + struct weston_view *view; +@@ -1872,6 +1873,24 @@ gl_renderer_flush_damage(struct weston_surface *surface) + + glActiveTexture(GL_TEXTURE0); + ++ if (!gr->has_unpack_subimage) { ++ wl_shm_buffer_begin_access(buffer->shm_buffer); ++ for (j = 0; j < gs->num_textures; j++) { ++ glBindTexture(GL_TEXTURE_2D, gs->textures[j]); ++ glTexImage2D(GL_TEXTURE_2D, 0, ++ gs->gl_format[j], ++ gs->pitch / gs->hsub[j], ++ buffer->height / gs->vsub[j], ++ 0, ++ gl_format_from_internal(gs->gl_format[j]), ++ gs->gl_pixel_type, ++ data + gs->offset[j]); ++ } ++ wl_shm_buffer_end_access(buffer->shm_buffer); ++ ++ goto done; ++ } ++ + if (gs->needs_full_upload || quirks->gl_force_full_upload) { + glPixelStorei(GL_UNPACK_SKIP_PIXELS_EXT, 0); + glPixelStorei(GL_UNPACK_SKIP_ROWS_EXT, 0); +@@ -3932,11 +3951,9 @@ gl_renderer_setup(struct weston_compositor *ec, EGLSurface egl_surface) + else + ec->read_format = PIXMAN_a8b8g8r8; + +- if (gr->gl_version < gr_gl_version(3, 0) && +- !weston_check_egl_extension(extensions, "GL_EXT_unpack_subimage")) { +- weston_log("GL_EXT_unpack_subimage not available.\n"); +- return -1; +- } ++ if (gr->gl_version >= gr_gl_version(3, 0) || ++ weston_check_egl_extension(extensions, "GL_EXT_unpack_subimage")) ++ gr->has_unpack_subimage = true; + + if (gr->gl_version >= gr_gl_version(3, 0) || + weston_check_egl_extension(extensions, "GL_EXT_texture_type_2_10_10_10_REV")) +@@ -3977,6 +3994,8 @@ gl_renderer_setup(struct weston_compositor *ec, EGLSurface egl_surface) + gr_gl_version_minor(gr->gl_version)); + weston_log_continue(STAMP_SPACE "read-back format: %s\n", + ec->read_format == PIXMAN_a8r8g8b8 ? "BGRA" : "RGBA"); ++ weston_log_continue(STAMP_SPACE "wl_shm sub-image to texture: %s\n", ++ gr->has_unpack_subimage ? "yes" : "no"); + weston_log_continue(STAMP_SPACE "EGL Wayland extension: %s\n", + gr->has_bind_display ? "yes" : "no"); +meta-arago-distro/recipes-graphics/wayland/weston/0001-backend-drm-Select-plane-based-on-current-attached-C.patch +22
-30@@ -1,5 +1,5 @@ -From 943323d177ef33494d9d063aeb7f0e9785d9b3c1 Mon Sep 17 00:00:00 2001 -From: "Andrew F. Davis" <[email protected]> +From 8c4445d4dbd4c2d5de5a19370fcc5d8777a7e6d4 Mon Sep 17 00:00:00 2001 +From: Andrew Davis <[email protected]> Date: Fri, 6 Mar 2020 13:06:55 -0500 Subject: [PATCH] backend-drm: Select plane based on current attached CRTC@@ -11,45 +11,37 @@ This prevents changing a CRTC's primary plane when it is active which is not allowed by the DRM framework. Based-on-patch-by: Eric Ruei <[email protected]> -Signed-off-by: Andrew F. Davis <[email protected]> +Signed-off-by: Andrew Davis <[email protected]> --- - libweston/backend-drm/drm-internal.h | 1 + - libweston/backend-drm/drm.c | 10 ++++++++++ - 2 files changed, 11 insertions(+) + libweston/backend-drm/drm-internal.h | 1 + + libweston/backend-drm/drm.c | 9 +++++++++ + 2 files changed, 10 insertions(+) diff --git a/libweston/backend-drm/drm-internal.h b/libweston/backend-drm/drm-internal.h -index 6f5a9880..65b7486d 100644 +index 48600880..b381ca31 100644 --- a/libweston/backend-drm/drm-internal.h +++ b/libweston/backend-drm/drm-internal.h -@@ -398,6 +398,7 @@ struct drm_plane { - +@@ -468,6 +468,7 @@ struct drm_plane { uint32_t possible_crtcs; uint32_t plane_id; + uint32_t plane_idx; + uint32_t crtc_id; - uint32_t count_formats; struct drm_property_info props[WDRM_PLANE__COUNT]; + diff --git a/libweston/backend-drm/drm.c b/libweston/backend-drm/drm.c -index 041a15a5..544b1aff 100644 +index 42787702..b5b7c4ba 100644 --- a/libweston/backend-drm/drm.c +++ b/libweston/backend-drm/drm.c -@@ -819,6 +819,7 @@ drm_plane_create(struct drm_backend *b, const drmModePlane *kplane, - if (kplane) { - plane->possible_crtcs = kplane->possible_crtcs; - plane->plane_id = kplane->plane_id; -+ plane->crtc_id = kplane->crtc_id; +@@ -776,6 +776,7 @@ drm_plane_create(struct drm_backend *b, const drmModePlane *kplane) + plane->state_cur->complete = true; + plane->possible_crtcs = kplane->possible_crtcs; + plane->plane_id = kplane->plane_id; ++ plane->crtc_id = kplane->crtc_id; + + weston_drm_format_array_init(&plane->formats); - props = drmModeObjectGetProperties(b->drm.fd, kplane->plane_id, - DRM_MODE_OBJECT_PLANE); -@@ -843,6 +844,7 @@ drm_plane_create(struct drm_backend *b, const drmModePlane *kplane, - else { - plane->possible_crtcs = (1 << output->pipe); - plane->plane_id = 0; -+ plane->crtc_id = 0; - plane->count_formats = 1; - plane->formats[0].format = format; - plane->type = type; -@@ -950,6 +952,14 @@ drm_output_find_special_plane(struct drm_backend *b, struct drm_output *output, +@@ -866,6 +867,14 @@ drm_output_find_special_plane(struct drm_backend *b, struct drm_output *output, if (found_elsewhere) continue;@@ -58,12 +50,12 @@ index 041a15a5..544b1aff 100644 + * switch away a plane from a CTRC when active. */ + if ((type == WDRM_PLANE_TYPE_PRIMARY) && + (plane->crtc_id != 0) && -+ (plane->crtc_id != output->crtc_id)) ++ (plane->crtc_id != output->crtc->crtc_id)) + continue; + - plane->possible_crtcs = (1 << output->pipe); + plane->possible_crtcs = (1 << output->crtc->pipe); return plane; } -- -2.17.1 +2.38.1meta-arago-distro/recipes-graphics/wayland/weston_10.0.2.bbappend renamed from meta-arago-distro/recipes-graphics/wayland/weston_9.0.0.bbappend +2
-1@@ -1,10 +1,11 @@ FILESEXTRAPATHS:prepend := "${THISDIR}/${PN}:" -PR:append = ".arago3" +PR:append = ".arago1" # file://0002-weston-Allow-visual_id-to-be-0.patch SRC_URI += " \ file://0003-weston-Fix-virtual-keyboard-display-issue-for-QT5-ap.patch \ file://0004-weston-Fix-touch-screen-crash-issue.patch \ file://0001-backend-drm-Select-plane-based-on-current-attached-C.patch \ + file://0001-Revert-require-GL_EXT_unpack_subimage-commit.patch \ "Signature · 1 line
2.38.1
Quoted 32 lines · Denys Dmytriyenko
Quoted 17 lines · Andrew Davis via lists.yoctoproject.org
To match upstream oe-core/master.
Signed-off-by Andrew Davis <[email protected]>--- ...equire-GL_EXT_unpack_subimage-commit.patch | 91 +++++++++++++++++++ ...ct-plane-based-on-current-attached-C.patch | 52 +++++------ ..._9.0.0.bbappend => weston_10.0.2.bbappend} | 3 +- 3 files changed, 115 insertions(+), 31 deletions(-)create mode 100644
meta-arago-distro/recipes-graphics/wayland/weston/0001-Revert-require-GL_EXT_unpack_subimage-commit.patchrename meta-arago-distro/recipes-graphics/wayland/{weston_9.0.0.bbappend => weston_10.0.2.bbappend} (78%)meta-arago-distro/recipes-graphics/wayland/weston/0001-Revert-require-GL_EXT_unpack_subimage-commit.patch new file +0
-0meta-arago-distro/recipes-graphics/wayland/weston/0001-Revert-require-GL_EXT_unpack_subimage-commit.patch line 10
Quoted 10 lines
meta-arago-distro/recipes-graphics/wayland/weston/0001-Revert-require-GL_EXT_unpack_subimage-commit.patch new file +10
-0@@ -0,0 +1,91 @@ +From 6558a9153cc81199146132041c61023a2c2e1b2e Mon Sep 17 00:00:00 2001 +From: Andrew Davis <[email protected]> +Date: Thu, 6 Oct 2022 15:49:31 -0500 +Subject: [PATCH] Revert require GL_EXT_unpack_subimage commit + +This reverts commit 593d5af43a8e2c2a3371088fa7ae430d0517c82d. + +That commit removed support for GPU drivers without GL_EXT_unpack_subimage +which SGX does not support. Add back support for GPUs without this +extension.The obvious question - how safe is it to assume
GL_EXT_unpack_subimageis not being used some place else outside of this commit and not guarded by:
Quoted 6 lines
if (!gr->has_unpack_subimage)
Is this extention really in the driver, or just in a Mesa API translation layer? Could it be better fixed in the DDK?
You are right, it is not a very safe assumption, this patch is almost certainly not complete for all cases. And even if it was, it only fixes Weston, others like wlroots have the same issues[0], so other compositors are going to be broken too.
Long term this extension will need added to the driver. Unfortunately all work on keeping this driver alive is unofficial, and I just don't have the free time to add and test new extensions right now :(
Andrew
[0] https://github.com/tmlind/wlroots/commit/04491889240bbadb693aa37036dc55835118b2cc
Quoted 44 lines · Andrew Davis via lists.yoctoproject.org
Quoted 32 lines · Denys Dmytriyenko
Quoted 17 lines · Andrew Davis via lists.yoctoproject.org
To match upstream oe-core/master.
Signed-off-by Andrew Davis <[email protected]>--- ...equire-GL_EXT_unpack_subimage-commit.patch | 91 +++++++++++++++++++ ...ct-plane-based-on-current-attached-C.patch | 52 +++++------ ..._9.0.0.bbappend => weston_10.0.2.bbappend} | 3 +- 3 files changed, 115 insertions(+), 31 deletions(-) create mode 100644 meta-arago-distro/recipes-graphics/wayland/weston/0001-Revert-require-GL_EXT_unpack_subimage-commit.patch rename meta-arago-distro/recipes-graphics/wayland/{weston_9.0.0.bbappend => weston_10.0.2.bbappend} (78%)meta-arago-distro/recipes-graphics/wayland/weston/0001-Revert-require-GL_EXT_unpack_subimage-commit.patch new file +0
-0meta-arago-distro/recipes-graphics/wayland/weston/0001-Revert-require-GL_EXT_unpack_subimage-commit.patch line 10
Quoted 10 lines
meta-arago-distro/recipes-graphics/wayland/weston/0001-Revert-require-GL_EXT_unpack_subimage-commit.patch new file +10
-0@@ -0,0 +1,91 @@ +From 6558a9153cc81199146132041c61023a2c2e1b2e Mon Sep 17 00:00:00 2001 +From: Andrew Davis <[email protected]> +Date: Thu, 6 Oct 2022 15:49:31 -0500 +Subject: [PATCH] Revert require GL_EXT_unpack_subimage commit + +This reverts commit 593d5af43a8e2c2a3371088fa7ae430d0517c82d. + +That commit removed support for GPU drivers without GL_EXT_unpack_subimage +which SGX does not support. Add back support for GPUs without this +extension.The obvious question - how safe is it to assume
GL_EXT_unpack_subimageis not being used some place else outside of this commit and not guarded by:Quoted 6 lines
if (!gr->has_unpack_subimage)
Is this extention really in the driver, or just in a Mesa API translation layer? Could it be better fixed in the DDK?
You are right, it is not a very safe assumption, this patch is almost certainly not complete for all cases. And even if it was, it only fixes Weston, others like wlroots have the same issues[0], so other compositors are going to be broken too.
Quoted 7 lines
Long term this extension will need added to the driver. Unfortunately all work on keeping this driver alive is unofficial, and I just don't have the free time to add and test new extensions right now :(
Andrew
[0] https://github.com/tmlind/wlroots/commit/04491889240bbadb693aa37036dc55835118b2cc
Ryan,
I see you are pulling this patch to the WIP branches - what's the consensus here? As discussed, this is a very partial solution. Moreover, it only "fixes" meta-arago, so won't help people using just meta-ti with other distros...
Signature · 1 line
Denys
Quoted 51 lines · Denys Dmytriyenko
Quoted 50 lines · Andrew Davis via lists.yoctoproject.org
Quoted 32 lines · Denys Dmytriyenko
Quoted 17 lines · Andrew Davis via lists.yoctoproject.org
To match upstream oe-core/master.
Signed-off-by Andrew Davis <[email protected]>--- ...equire-GL_EXT_unpack_subimage-commit.patch | 91 +++++++++++++++++++ ...ct-plane-based-on-current-attached-C.patch | 52 +++++------ ..._9.0.0.bbappend => weston_10.0.2.bbappend} | 3 +- 3 files changed, 115 insertions(+), 31 deletions(-)create mode 100644
meta-arago-distro/recipes-graphics/wayland/weston/0001-Revert-require-GL_EXT_unpack_subimage-commit.patchrename meta-arago-distro/recipes-graphics/wayland/{weston_9.0.0.bbappend => weston_10.0.2.bbappend} (78%)meta-arago-distro/recipes-graphics/wayland/weston/0001-Revert-require-GL_EXT_unpack_subimage-commit.patch new file +0
-0meta-arago-distro/recipes-graphics/wayland/weston/0001-Revert-require-GL_EXT_unpack_subimage-commit.patch line 10
Quoted 10 lines
meta-arago-distro/recipes-graphics/wayland/weston/0001-Revert-require-GL_EXT_unpack_subimage-commit.patch new file +10
-0@@ -0,0 +1,91 @@ +From 6558a9153cc81199146132041c61023a2c2e1b2e Mon Sep 17 00:00:00 2001 +From: Andrew Davis <[email protected]> +Date: Thu, 6 Oct 2022 15:49:31 -0500 +Subject: [PATCH] Revert require GL_EXT_unpack_subimage commit + +This reverts commit 593d5af43a8e2c2a3371088fa7ae430d0517c82d. + +That commit removed support for GPU drivers without GL_EXT_unpack_subimage +which SGX does not support. Add back support for GPUs without this +extension.The obvious question - how safe is it to assume
GL_EXT_unpack_subimageis not being used some place else outside of this commit and not guarded by:Quoted 6 lines
if (!gr->has_unpack_subimage)
Is this extention really in the driver, or just in a Mesa API translation layer? Could it be better fixed in the DDK?
You are right, it is not a very safe assumption, this patch is almost certainly not complete for all cases. And even if it was, it only fixes Weston, others like wlroots have the same issues[0], so other compositors are going to be broken too.
Long term this extension will need added to the driver. Unfortunately all work on keeping this driver alive is unofficial, and I just don't have the free time to add and test new extensions right now :(
Andrew
Quoted 8 lines
Ryan,
I see you are pulling this patch to the WIP branches - what's the consensus here? As discussed, this is a very partial solution. Moreover, it only "fixes" meta-arago, so won't help people using just meta-ti with other distros...
Partial solution is better than no solution, right? Today even *with* meta-arago your images will have broken graphics, at least with this we can start building and testing some part of the graphics stack kirkstone.
(I don't like half-measures either, but it's all I have today)
Andrew
Quoted 61 lines · Andrew Davis via lists.yoctoproject.org
Quoted 51 lines · Denys Dmytriyenko
Quoted 50 lines · Andrew Davis via lists.yoctoproject.org
Quoted 32 lines · Denys Dmytriyenko
Quoted 17 lines · Andrew Davis via lists.yoctoproject.org
To match upstream oe-core/master.
Signed-off-by Andrew Davis <[email protected]>--- ...equire-GL_EXT_unpack_subimage-commit.patch | 91 +++++++++++++++++++ ...ct-plane-based-on-current-attached-C.patch | 52 +++++------ ..._9.0.0.bbappend => weston_10.0.2.bbappend} | 3 +- 3 files changed, 115 insertions(+), 31 deletions(-) create mode 100644 meta-arago-distro/recipes-graphics/wayland/weston/0001-Revert-require-GL_EXT_unpack_subimage-commit.patch rename meta-arago-distro/recipes-graphics/wayland/{weston_9.0.0.bbappend => weston_10.0.2.bbappend} (78%)meta-arago-distro/recipes-graphics/wayland/weston/0001-Revert-require-GL_EXT_unpack_subimage-commit.patch new file +0
-0meta-arago-distro/recipes-graphics/wayland/weston/0001-Revert-require-GL_EXT_unpack_subimage-commit.patch line 10
Quoted 10 lines
meta-arago-distro/recipes-graphics/wayland/weston/0001-Revert-require-GL_EXT_unpack_subimage-commit.patch new file +10
-0@@ -0,0 +1,91 @@ +From 6558a9153cc81199146132041c61023a2c2e1b2e Mon Sep 17 00:00:00 2001 +From: Andrew Davis <[email protected]> +Date: Thu, 6 Oct 2022 15:49:31 -0500 +Subject: [PATCH] Revert require GL_EXT_unpack_subimage commit + +This reverts commit 593d5af43a8e2c2a3371088fa7ae430d0517c82d. + +That commit removed support for GPU drivers without GL_EXT_unpack_subimage +which SGX does not support. Add back support for GPUs without this +extension.The obvious question - how safe is it to assume
GL_EXT_unpack_subimageis not being used some place else outside of this commit and not guarded by:Quoted 6 lines
if (!gr->has_unpack_subimage)
Is this extention really in the driver, or just in a Mesa API translation layer? Could it be better fixed in the DDK?
You are right, it is not a very safe assumption, this patch is almost certainly not complete for all cases. And even if it was, it only fixes Weston, others like wlroots have the same issues[0], so other compositors are going to be broken too.
Long term this extension will need added to the driver. Unfortunately all work on keeping this driver alive is unofficial, and I just don't have the free time to add and test new extensions right now :(
Andrew
Quoted 8 lines
Ryan,
I see you are pulling this patch to the WIP branches - what's the consensus here? As discussed, this is a very partial solution. Moreover, it only "fixes" meta-arago, so won't help people using just meta-ti with other distros...
Quoted 5 lines
Partial solution is better than no solution, right? Today even *with* meta-arago your images will have broken graphics, at least with this we can start building and testing some part of the graphics stack kirkstone.
(I don't like half-measures either, but it's all I have today)
Ok, sounds fair... Next question - was it tested and is it supposed to build/work? Or is it again a maintainers' problem? :)
Signature · 1 line
Denys