| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
| |
We need to look at the Dst for image stores.
|
|
|
|
| |
to stop using the workaround in si_make_buffer_descriptor.
|
|
|
|
|
|
| |
no change in behavior.
Reviewed-by: Bas Nieuwenhuizen <[email protected]>
|
|
|
|
|
|
| |
Ported from virglrenderer.
Signed-off-by: Dave Airlie <[email protected]>
|
|
|
|
|
| |
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=108311
Cc: <[email protected]>
|
|
|
|
|
|
|
|
|
| |
Vulkan and Gallium don't use Mesa's gl_program data structure, so they
can't poke at 'prog'. But we can simply use the copy of the shader info
stored with the NIR shader, which is guaranteed to exist.
Reviewed-by: Jason Ekstrand <[email protected]>
Reviewed-by: Iago Toral Quiroga <[email protected]>
|
|
|
|
|
|
| |
This makes some of the code more clear.
Reviewed-by: Thomas Helland <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Otherwise, I get this error:
main/egldevice.h:54:13: error: ‘NULL’ undeclared (first use in this function)
dev = NULL;
^~~~
with this config:
./autogen.sh --enable-gles1 --enable-gles2 --with-platforms='surfaceless' --disable-glx
--with-dri-drivers="i965" --with-gallium-drivers="" --enable-gbm
v3: Use stddef.h (Matt)
v4: Modify commit message (Eric)
Reviewed-by: Matt Turner <[email protected]>
Reviewed-by: Eric Engestrom <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
| |
The common truncf(x + 0.5) fails for the floating-point value just less
than 0.5 (nextafterf(0.5, 0.0)). nextafterf(0.5, 0.0) + 0.5, after
rounding is 1.0, thus truncf does not produce the desired value.
The solution is to add nextafterf(0.5, 0.0) instead of 0.5 before
truncating. This works for all values.
Reviewed-by: Roland Scheidegger <[email protected]>
|
|
|
|
| |
Signed-off-by: Juan A. Suarez Romero <[email protected]>
|
|
|
|
|
| |
Signed-off-by: Juan A. Suarez Romero <[email protected]>
(cherry picked from commit cfd1f8b92cae9dde3e5bed42109b5142f50a2ee5)
|
|
|
|
|
| |
Signed-off-by: Juan A. Suarez Romero <[email protected]>
(cherry picked from commit 3e741344d79e3ae67b1ad645e7d56fe6c0fb2ae2)
|
|
|
|
|
| |
Fixes: ce74a7bb8de7 ("egl/wayland: plug memory leak in drm_handle_device()")
Fixes: c59d3aa4b9bc ("egl/wayland: bail out when drmGetMagic fails")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Introduce a new driver-private transfer flag RADEON_TRANSFER_TEMPORARY
that specifies whether the caller will use buffer_unmap or not. The
default behavior is set to permanent maps, because that's what drivers
do for Gallium buffer maps.
This should eliminate the need for hacks in libdrm. Assertions are added
to catch when the buffer_unmap calls don't match the (temporary)
buffer_map calls.
I did my best to update r600 for consistency (r300 needs no changes
because it never calls buffer_unmap), even though the radeon winsys
ignores the new flag.
As an added bonus, this should actually improve the performance of
the normal fast path, because we no longer call into libdrm at all
after the first map, and there's one less atomic in the winsys itself
(there are now no atomics left in the UNSYNCHRONIZED fast path).
Cc: Leo Liu <[email protected]>
v2:
- remove comment about visible VRAM (Marek)
- don't rely on amdgpu_bo_cpu_map doing an atomic write
Reviewed-by: Marek Olšák <[email protected]>
|
|
|
|
|
|
|
| |
We'll use it in the upcoming mapping change. Sparse buffers have always
had one.
Reviewed-by: Marek Olšák <[email protected]>
|
|
|
|
|
|
| |
Fixes: 59e58c348e6af16a5f2dd "vulkan/wsi: Only wait on semaphores on the first swapchain"
Signed-off-by: Eric Engestrom <[email protected]>
Reviewed-by: Jason Ekstrand <[email protected]>
|
|
|
|
|
|
|
|
|
|
| |
As we fail to open the node, we leak the node/device name.
v2: Log and then free() (Eric)
Cc: [email protected]
Signed-off-by: Emil Velikov <[email protected]>
Reviewed-by: Eric Engestrom <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Currently as the function fails, we pass uninitialized data to the
authentication function. Stop doing that and print an warning when
the function fails.
v2: Plug memory leak in error path (Eric)
Cc: [email protected]
Signed-off-by: Emil Velikov <[email protected]>
Reviewed-by: Tapani Pälli <[email protected]> (v1)
Reviewed-by: Eric Engestrom <[email protected]>
|
|
|
|
|
|
| |
Fixes: da997ebec92942193955 "vulkan: Add KHR_display extension using DRM [v10]"
Signed-off-by: Eric Engestrom <[email protected]>
Reviewed-by: Keith Packard <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Originally the driver reported GL_FRAMEBUFFER_UNSUPPORTED in all cases,
adding more specific error messages was not correct and broke many tests.
Mostly revert this and only report GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT
for MESA_FORMAT_R_SRGB8.
Fixes: ebcde3454552adc6d3fea8af2207aafaba857796
i965: be more specific about FBO completeness errors
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=108805
Signed-off-by: Gert Wollny <[email protected]>
Reviewed-by: Tapani Pälli <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
| |
The format is emulated by using ISL_FORMAT_L8_SRGB, therefore we need to
force swizzles for the GBA channels. However, doing this only based on the
data type GL_RED breaks other formats, therefore, test specifically for the
format.
Fixes: c5363869d4971780401b21bb75083ef2518c12be
i965: Force zero swizzles for unused components in GL_RED and GL_RG
Signed-off-by: Gert Wollny <[email protected]>
Reviewed-by: Tapani Pälli <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
vtest doesn't implement the according API and would segfault:
Program received signal SIGSEGV, Segmentation fault.
#0 0x0000000000000000 in ?? ()
#1 in virgl_fence_server_sync at
src/gallium/drivers/virgl/virgl_context.c:1049
#2 in st_server_wait_sync at
src/mesa/state_tracker/st_cb_syncobj.c:155
so just don't do the call when the function pointers are not set.
Fixes dEQP:
dEQP-GLES3.functional.fence_sync.wait_sync_smalldraw
dEQP-GLES3.functional.fence_sync.wait_sync_largedraw
Fixes: d1a1c21e7621b5177febf191fcd3d3b8ef69dc96
virgl: native fence fd support
Signed-off-by: Gert Wollny <[email protected]>
Reviewed-by: Dave Airlie <[email protected]>
Reviewed-by: Robert Foss <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Avoids:
Conditional jump or move depends on uninitialised value(s)
at 0x9E2B39F: virgl_vtest_winsys_resource_cache_create (virgl_vtest_winsys.c:379)
by 0x9E2725F: virgl_buffer_create (virgl_buffer.c:169)
by 0x9E246D5: virgl_resource_create (virgl_resource.c:60)
by 0xA0C1B9F: bufferobj_data (st_cb_bufferobjects.c:344)
by 0xA0C1B9F: st_bufferobj_data (st_cb_bufferobjects.c:390)
by 0x9F4ACE3: vbo_use_buffer_objects (vbo_exec_api.c:1136)
by 0xA0C68C3: st_create_context_priv (st_context.c:416)
by 0xA0C707A: st_create_context (st_context.c:598)
by 0x9F81C6B: st_api_create_context (st_manager.c:918)
by 0x9BBE591: dri_create_context (dri_context.c:161)
by 0x9BB6931: driCreateContextAttribs (dri_util.c:473)
by 0x4E97A44: drisw_create_context_attribs (drisw_glx.c:630)
by 0x4E7C591: glXCreateContextAttribsARB (create_context.c:78)
Uninitialised value was created by a stack allocation
at 0x9E2B249: virgl_vtest_winsys_resource_cache_create (virgl_vtest_winsys.c:342)
Signed-off-by: Gert Wollny <[email protected]>
Reviewed-by: Dave Airlie <[email protected]>
Reviewed-by: Robert Foss <[email protected]>
|
|
|
|
|
|
| |
This wasn't handling 64-bit cases properly. Found by inspection.
Reviewed-by: Ian Romanick <[email protected]>
|
|
|
|
| |
Reviewed-by: Tapani Pälli <[email protected]>
|
|
|
|
|
|
|
|
| |
We normally call with stderr which is unbuffered, so this won't affect
that, but it does let me call nir_print_shader(nir, fopen("log", "w+"))
from gdb and actually get the whole shader in my file.
Reviewed-by: Tapani Pälli <[email protected]>
|
|
|
|
|
|
| |
I've been using this with the kmsro series to test v3d on VKMS without my
old KMS hack in the v3d kernel driver. KMSRO still needs some cleanup,
but v3d RO support seems reasonable.
|
|
|
|
|
| |
Fixes: 0d17b685b1ff ("gallium/u_tests: add a compute shader test that clears an image")
Reviewed-by: Marek Olšák <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Improves performance in Talos by about 15% (and significant improvements
in RotR and possibly other but did not bench with final patch) on
kernel 4.19 and earlier.
On 4.20+ a similar effect comes from
433ca054949a "drm/amdgpu: try allocating VRAM as power of two"
v2: Do not impact the alignment of the physical memory.
Reviewed-by: Dave Airlie <[email protected]>
Reviewed-by: Samuel Pitoiset <[email protected]>
CC: <[email protected]>
|
|
|
|
|
|
|
|
|
|
| |
Since 1285f71d3e landed, it needs to provide apps with proper sample
position for MSAA.
Currently no way to query this to hw, these are taken from blob driver.
Fixes: dEQP-GLES31.functional.texture.multisample.samples_#.sample_position
Signed-off-by: Rob Clark <[email protected]>
|
|
|
|
|
|
|
|
|
|
| |
We set equiv bit in SP_FS_CTRL_REG0. Somehow the hw doesn't hang with
this mismatched config, but does run slower. It is faster with either
neither bit set, or both bits set, but both is the fastest of the three
configurations. Worth a bit over 10% gain in glmark2.
Spotted-by: Jonathan Marek <[email protected]>
Signed-off-by: Rob Clark <[email protected]>
|
|
|
|
| |
Signed-off-by: Jonathan Marek <[email protected]>
|
|
|
|
|
|
|
| |
blip_fp uses GENERIC as input, so blit_vp should match for linking
Signed-off-by: Jonathan Marek <[email protected]>
Signed-off-by: Rob Clark <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
| |
Adds all missing texture related logic. For everything to work it also
needs changes to ir2/fd2_program, which are part of the ir2 update patch.
Note: it needs rnndb update
Signed-off-by: Jonathan Marek <[email protected]>
[remove stray patch]
Signed-off-by: Rob Clark <[email protected]>
|
|
|
|
|
|
|
|
| |
Note: it needs rnndb update
Signed-off-by: Marek Vasut <[email protected]>
Signed-off-by: Jonathan Marek <[email protected]>
Signed-off-by: Rob Clark <[email protected]>
|
|
|
|
|
| |
Signed-off-by: Jonathan Marek <[email protected]>
Signed-off-by: Rob Clark <[email protected]>
|
|
|
|
|
|
|
|
| |
200: 256KiB GMEM A200 (imx53)
201: 128KiB GMEM A200 (imx51)
Signed-off-by: Jonathan Marek <[email protected]>
Signed-off-by: Rob Clark <[email protected]>
|
|
|
|
|
|
|
| |
As it stands, it overflows to zero.
Signed-off-by: Jonathan Marek <[email protected]>
Signed-off-by: Rob Clark <[email protected]>
|
|
|
|
|
| |
Signed-off-by: Jonathan Marek <[email protected]>
Signed-off-by: Rob Clark <[email protected]>
|
|
|
|
|
|
|
|
|
| |
lowers ceil(x) as -floor(-x)
Signed-off-by: Jonathan Marek <[email protected]>
Reviewed-by: Jason Ekstrand <[email protected]>
Reviewed-by: Christian Gmeiner <[email protected]>
Signed-off-by: Rob Clark <[email protected]>
|
|
|
|
| |
Signed-off-by: Rob Clark <[email protected]>
|
|
|
|
|
|
|
|
|
| |
On older gens, the CLIP_ADJ bitfields were actually 3.6 fixed point.
Which might make more sense. Although this formula comes up with values
pretty close to what blob does for various viewport sizes (for at least
a5xx and a6xx), and seems to work.
Signed-off-by: Rob Clark <[email protected]>
|
|
|
|
|
|
|
|
|
| |
f6131d4ec7a had the side effect of enabling LRZ w/ 32b depth buffers.
But there are some bugs with this, which aren't fully understood yet,
so for now just skip LRZ w/ z32..
Fixes: f6131d4ec7a freedreno/a6xx: Clear z32 and separate stencil with blitter
Signed-off-by: Rob Clark <[email protected]>
|
|
|
|
|
|
|
|
|
| |
We generate an IB to clear the gmem at flush time and jump to it
before rendering each tile. This lets us get rid of the command stream
patching for gmem offsets.
Signed-off-by: Kristian H. Kristensen <[email protected]>
Signed-off-by: Rob Clark <[email protected]>
|
|
|
|
|
| |
Signed-off-by: Kristian H. Kristensen <[email protected]>
Signed-off-by: Rob Clark <[email protected]>
|
|
|
|
|
| |
Signed-off-by: Kristian H. Kristensen <[email protected]>
Signed-off-by: Rob Clark <[email protected]>
|
|
|
|
|
|
|
|
|
| |
For RGB surfaces (for example) we don't really care that the colormask
is 0x7 instead of 0xf. This should not trigger clear_with_quad()
slowpath.
Signed-off-by: Rob Clark <[email protected]>
Reviewed-by: Marek Olšák <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
| |
If we can't clear all the buffers with pctx->clear() (say, for example,
because of ColorMask), push the buffers we *can* clear with pctx->clear()
first. Tilers want to see clears coming before draws to enable fast-
paths, and clearing one of the attachments with a quad-draw first
confuses that logic.
Signed-off-by: Rob Clark <[email protected]>
Reviewed-by: Eric Anholt <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Move (most of) the ir3 compiler to src/freedreno/ir3 so that it can be
re-used by some future vulkan driver. The parts that are gallium
specific have been refactored out and remain in the gallium driver.
Getting the move done now so that it can happen before further
refactoring to support a6xx specific instructions.
NOTE also removes ir3_cmdline compiler tool from autotools build since
that was easier than fixing it and I normally use meson build. Waiting
patiently for the day that we can remove *everything* from the autotools
build.
Signed-off-by: Rob Clark <[email protected]>
|
|
|
|
| |
Signed-off-by: Rob Clark <[email protected]>
|