- 18 Mar, 2022 3 commits
-
-
Specify the `parallel` variable at job level to be able to fine-tune it per job. Signed-off-by:
Corentin Noël <corentin.noel@collabora.com> Part-of: <virgl/virglrenderer!743>
-
Backtracks on one change from commit 5304e7d3 ("vrend: add vrend_resource_needs_redblue_swizzle() to fix arcvm deqp"), that incorrectly inverted the behavior for setting swizzle parameters for a sampler of eglimage-backed bgra resources. Fixes: 5304e7d3 ("vrend: add vrend_resource_needs_redblue_swizzle() to fix arcvm deqp") Signed-off-by:
Ryan Neph <ryanneph@google.com> Part-of: <virgl/virglrenderer!755>
-
Signed-off-by:
Yonggang Luo <luoyonggang@gmail.com> Part-of: <virgl/virglrenderer!713>
-
- 17 Mar, 2022 6 commits
-
-
Part-of: <virgl/virglrenderer!735>
-
In addition don't base the artifact filename on the CI_JOB_NAME, because of the "parallel" specifier the name contains a '/' which breaks the naming. Part-of: <virgl/virglrenderer!735>
-
Signed-off-by:
Gert Wollny <gert.wollny@collabora.com> Part-of: <virgl/virglrenderer!735>
-
On GLES hosts it is not supported at all, on GL the support is shaky, and there is no real interest to actually fix this anytime soon. Part-of: <virgl/virglrenderer!735>
-
The ctx variable is set to 0 on the line before so ctx.prog_type is always 0 which is TGSI_PROCESSOR_FRAGMENT. Signed-off-by:
-
These are leftover variables from e853d88d . Signed-off-by:
-
- 16 Mar, 2022 4 commits
-
-
Corentin Noël authored
The SAMPLEMASK is always returning an integer, we previously always had another instruction that would convert it to a float but this is no longer the case so always convert it to a floating number. Signed-off-by:
-
Allows the extension passthrough and implements the extension calls. Signed-off-by:
Igor Torrente <igor.torrente@collabora.com>
Part-of: <virgl/virglrenderer!747> -
Update venus-protocol files to move vkGetCalibratedTimestamps function from vn_protocol_driver_transport.h to vn_protocol_driver_device.h. Signed-off-by:
Igor Torrente <igor.torrente@collabora.com>
Part-of: <virgl/virglrenderer!747> -
Get rid of the ex_ prefix and use the common per stage prefix plus the _c or _bc suffix. With that we can also cover the compatibility contexts. v2: remove ducplicate assignments (John) Part-of: <virgl/virglrenderer!742>
-
- 15 Mar, 2022 7 commits
-
-
I remembered wrong; needs is the good new one, and dependencies is the bad old one. Flip all the jobs over to using needs instead. Signed-off-by:
Daniel Stone <daniels@collabora.com>
Part-of: <virgl/virglrenderer!748> -
We mark the build job as never running post-merge, because we just don't want a pipeline running. GitLab seems to respond to this by just trying to run the job anyway, so mark it as never. In all other cases, it will be run on_success when the upstream jobs it depends on have completed. Signed-off-by:
Daniel Stone <daniels@collabora.com>
Part-of: <virgl/virglrenderer!748> -
Mesa has VIRGL_FORMAT_R8G8_SRGB since commit e33cb6a763532279fc6fa6c3c1583ce3dd530827. Define it also in virglrenderer for the header compatibility with Mesa and to avoid acceidental collision of the definition. Signed-off-by:
Akihiko Odaki <akihiko.odaki@gmail.com> Part-of: <virgl/virglrenderer!738>
-
VirGL just runs way too many jobs, with each pipeline consuming the entire committed x86-64 build capacity we have available, so we need to restrict it anyway. This is made worse by the fact it will create duplicate pipelines for merge requests. Signed-off-by:
Daniel Stone <daniels@collabora.com>
Part-of: <virgl/virglrenderer!744> -
It is possible for two variable to have the same sid and array_id as long as they have a different name. Make sure that we are not considering them as two components. Also fix the already_found_one variable define in the loop when we need it to persist between iterations. Signed-off-by:
-
Drop /ci/.gitlab-ci.yml* files since the project CI/CD settings have been updated to make use of /.gitlab-ci.yml. Signed-off-by:
Cristian Ciocaltea <cristian.ciocaltea@collabora.com>
Part-of: <!746> -
The new definition is copied from commit 2116a4f5cc3a570a30a1378d03dbe5bd9c1ae50c of Mesa. Signed-off-by:
-
- 14 Mar, 2022 8 commits
-
-
Yiwei Zhang authored
The renderer doesn't advertise this ext to the driver. Signed-off-by:
Yiwei Zhang <zzyiwei@chromium.org> Part-of: <virgl/virglrenderer!745>
-
Yiwei Zhang authored
Commands for VK_KHR_external_fence_fd are disabled between the driver and renderer, but we need this extension for renderer proc table. Signed-off-by:
-
Replaces all the direct vk* calls with the proc-table to avoid using the loader trampoline. Signed-off-by:
Igor Torrente <igor.torrente@collabora.com>
Part-of: <!719> -
From a performance standpoint, is preferable to use the proc-table and avoid the loader trampoline. Signed-off-by:
Igor Torrente <igor.torrente@collabora.com>
Part-of: <virgl/virglrenderer!719> -
Removes the old method and update all the affected functions. Co-authored-by:
Igor Torrente <igor.torrente@collabora.com>
Co-authored-by: Chia-I Wu <olvaffe@gmail.com> Part-of: <virgl/virglrenderer!719>
-
Adds the code to make use of the new proc table functions in the [physical_]device initialization. Co-authored-by:
Igor Torrente <igor.torrente@collabora.com>
Co-authored-by: -
Add the automatic generated code to the proc tables. Co-authored-by:
Igor Torrente <igor.torrente@collabora.com>
Co-authored-by: -
Ryan Neph authored
Fixes dEQP-EGL.functional.wide_color#window_888_colorspace_srgb on ARCVM, which uses a 24 bit-per-pixel R8G8B8_UNORM surface format that slipped through the cracks during the latest BGRA rework in virglrenderer!725. Fixes: 187e6513 ("vrend: avoid srgb views on eglimage-backed surfaces") Signed-off-by:
-
- 11 Mar, 2022 1 commit
-
-
Corentin Noël authored
Part-of: <virgl/virglrenderer!739>
-
- 09 Mar, 2022 1 commit
-
-
小田喜陽彦 authored
Signed-off-by:
-
- 08 Mar, 2022 1 commit
-
-
Converting from valid pointer to void to intptr_t and uintptr_t, and then converting back to pointer to void will result in a pointer which compares equal to the original pointer. Using another integer type as intermediate would result in an implementation-defined behavior. Especially using unsigned long for the purpose is problematic for LLP64 model like Windows. Signed-off-by:
-
- 07 Mar, 2022 9 commits
-
-
Add VK_EXT_shader_demote_to_helper_invocation VK_EXT_conservative_rasterization VK_EXT_depth_clip_enable VK_EXT_robustness2 VK_EXT_shader_stencil_export VK_EXT_vertex_attribute_divisor Part-of: <virgl/virglrenderer!728>
-
For VK_EXT_conservative_rasterization VK_EXT_depth_clip_enable VK_EXT_robustness2 VK_EXT_shader_stencil_export VK_EXT_vertex_attribute_divisor Part-of: <virgl/virglrenderer!728>
-
Now passing for GLES hosts with the previous commit. Signed-off-by:
-
On GLES hosts, pixel data for BGRA resources are byte-reordered and passed to the host driver for compatibility reasons. The clear color used in glClearTexSubImage[EXT] for such resources on GLES hosts must be reordered to match. Signed-off-by:
-
Signed-off-by:
-
After cleaning up many of the previous BGRA hacks, it is clear now that we must manually reorder all bgr* data en-route to/from the guest, but only for GLES hosts. GL hosts instead enjoy the use of the GL_BGRA pixel transfer format, prompting the host driver to the the reordering internally, and only when it deems it necessary. Trying to use GL_BGRA_EXT for GLES hosts is somewhat possible with the new GL_MESA_bgra extension, but it still struggles when multisampled textures are involved. It is more robust to use the GL_RGBA pixel transfer format for GLES hosts instead and handle reordering ourselves. Signed-off-by:
-
Now that the glTextureView() issues for eglimage-backed textures have been discovered, we can switch back to using the right gbm formats for externally allocated resources (this commit) and remove many of the bgra-specific emulation (following commits). Signed-off-by:
-
There's no reason to keep them separated. gles_bgra_formats already includes both UNORM and SRGB versions anyways. Signed-off-by:
-
Views on samplers of eglimage-backed rgb* textures cause unintended red/blue channel-swapping due to the lack of an internalformat that specifically for BGR* ordering. Since we have control over the sampler's swizzle parameters, we can still use the texture view to get automatic colorspace conversion, but need to compensate for the unintended swizzle with our own to swizzle back. If the view intended to swap the channels, we just leave the swizzle parameters alone. With this approach, no shader augmentation is necessary. Signed-off-by:
-
