It fits there much better and is an other step to get the
compiler to common code.
Signed-off-by: Christian Gmeiner <christian.gmeiner@gmail.com>
Reviewed-by: Lucas Stach <l.stach@pengutronix.de>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17819>
Native rewrite in current form doesn't check
type of register and may use output as a
temp.
Helps with 218 deqp-gles2 tests.
Cc: mesa-stable
Reviewed-by: Pavel Ondračka <pavel.ondracka@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17736>
Updating PC_TESSFACTOR_ADDR could race with the next draw which uses it.
Fixes GL CTS tests running via Zink:
KHR-Single-GL46.enhanced_layouts.glsl_contant_values
KHR-GLES32.core.tessellation_shader.tessellation_shader_point_mode.point_rendering
KHR-GLES32.core.tessellation_shader.winding.quads_cw
KHR-GLES32.core.tessellation_shader.winding.triangles_cw
And probably more.
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/6815
Suggested-by: Connor Abbott <cwabbott0@gmail.com>
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17833>
Now, only one instance of virgl_screen exists for a device
(/dev/dri/cardX), and it is shared by different frontends (eg GLX,
GBM, etc.). There is a problem with this, as follows:
/* Init GLX */
...
glXCreateContext(...);
...
/* GBM */
gbm_fd = open("/dev/dri/card0", O_RDWR);
dev = gbm_create_device(gbm_fd);
bo = gbm_bo_create(dev, ...);
plane_handle = gbm_bo_get_handle_for_plane(bo, ...);
drmPrimeHandleToFD(gbm_fd, handle.u32, flags, &plane_fd);
The above drmPrimeHandleToFD() call will fail with ENOENT.
The reason is that GBM and GLX share the same virgl_screen (file
descriptor), and it is not gbm_fd that is used to create gbm_bo,
but other fd (opened during GLX initialization). Since the scope
of prime handle is limited to drm_file, the above plane_handle is
invalid under gbm_fd.
By canceling the sharing of virgl_screen between different drm_files,
GBM can use the correct fd to create resources, thereby avoiding the
problem of invalid prime handle.
Signed-off-by: Feng Jiang <jiangfeng@kylinos.cn>
Reviewed-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16738>
Currently, res->maybe_busy is false by default. If wait immediately
after the resource is created, virgl_drm_resource_wait() will return
directly without checking the actual state of the kernel, which will
cause synchronization problems, such as:
On Guest:
pipe_buffer_create [mesa]
virgl_drm_winsys_resource_create
virtio_gpu_resource_create_ioctl [kernel]
virtio_gpu_fence_alloc
virtio_gpu_object_create
virtio_gpu_cmd_resource_create_3d
VIRTIO_GPU_CMD_RESOURCE_CREATE_3D
virtio_gpu_object_attach
virtio_gpu_cmd_resource_attach_backing
VIRTIO_GPU_CMD_RESOURCE_ATTACH_BACKING
resource_wait [mesa]
virgl_drm_resource_wait /* return directly without fence waiting */
pipe_buffer_map [mesa]
virgl_drm_resource_map
virtio_gpu_map_ioctl [kernel]
os_mmap
memcpy /* <== here */
On Host (with QEMU):
VIRTIO_GPU_CMD_RESOURCE_CREATE_3D
virgl_cmd_create_resource_3d [qemu]
virgl_renderer_resource_create [virglrenderer]
VIRTIO_GPU_CMD_RESOURCE_ATTACH_BACKING
virgl_resource_attach_backing [qemu]
virtio_gpu_create_mapping_iov
virgl_renderer_resource_attach_iov [virglrenderer]
virgl_resource_attach_iov
vrend_pipe_resource_attach_iov
vrend_write_to_iovec /* <== here */
virtio_gpu_cleanup_mapping_iov [qemu]
In the example above, there is a race condition between memcpy and
vrend_write_to_iovec.
Signed-off-by: Jiang Feng <jiangfeng@kylinos.cn>
Reviewed-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: Chia-I Wu <olvaffe@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17592>
Undefs can happen even in real GLSL shaders so it's best to handle them.
Lowering to zero is a perfectly valid implementation. Also, run DCE
because some of the undefs may be dead after from_ssa and there's no
point in processing those in the back-end.
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16817>
This was disabled ages ago because it provoked bugs between us and
xserver about context creation attributes, hopefully those servers are
out of circulation by now, let's find out.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17756>
As it is needed to have V3D_DEBUG defined. For the v3d case, I did it
restoring v3d_process_debug_variable, as it is at v3d_debug.c that
DEBUG_GET_ONCE_FLAGS_OPTION is called.
Fixes: 106b33405e ("vc4/v3d: stop adding NORAST when SHADERDB debug option is used")
Reviewed-by: Juan A. Suarez <jasuarez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17812>
There is no point in trying to get the softfp to work here,
since these cards don't support tesselation shaders, they will
never get OpenGL 4.0.
v2: Fix formatting (Filip)
Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: Filip Gawin <filip@gawin.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17822>
Pre-EG hardware handles the FS inputs differently, so we
need to prepare a different code path.
v2: Make m_interolators_used private (Filip)
Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: Filip Gawin <filip@gawin.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17822>
This is needed because when we switch between GLES and GL on the host,
we have to lower atomics to ssbo, and with that the shaders can't be
pulled from the cache anymore. Likewise when we move the disk image with
a shader cache to a different host, other features might change that
will need lowering. To avoid using stale shaders in this case, merge the
caps into into the shader cache sha.
Fixes: d6db4d2e08
virgl: Add simple disk cache
Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17798>
When the compiler pads a data structure, the padded bytes will not be
initialized. Shader keys are compared with memcmp and unitialized
bytes within the structure breaks this mechanism.
Explicitly pad the structures with members, so the compiler is forced
to initialize them. Add a warning to indicate if a change to
alignment in any of the data structures requires additional padding.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17749>
nine doesn't use viewids, and this breaks drivers that support it
cc: mesa-stable
Reviewed-by: Axel Davy <davyaxel0@gmail.com>
Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17792>
this needs to check the env var in order to work correctly
cc: mesa-stable
Reviewed-by: Axel Davy <davyaxel0@gmail.com>
Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17792>
