summaryrefslogtreecommitdiffstats
path: root/src/gallium/winsys/amdgpu
Commit message (Collapse)AuthorAgeFilesLines
* winsys/amdgpu: pad compute IBsMarek Olšák2018-03-081-1/+3
| | | | | | v2: pad with PKT2 NOPs on SI Reviewed-by: Alex Deucher <[email protected]>
* radeonsi: align command buffer starting address to fix some Raven hangsMarek Olšák2018-03-081-2/+3
| | | | | | Cc: 17.3 18.0 <[email protected]> Reviewed-by: Christian König <[email protected]> Reviewed-by: Alex Deucher <[email protected]>
* winsys/amdgpu: request high addressesChristian König2018-02-281-4/+12
| | | | | | | | | We now have hopefully fixed all bugs regarding high addresses on Vega10 and Raven. Start to use the high range to make room for SVM in the low range. Signed-off-by: Christian König <[email protected]> Reviewed-by: Marek Olšák <[email protected]>
* winsys/amdgpu:add uvd hevc enc support in amdgpu csJames Zhu2018-02-211-0/+6
| | | | | | | Support UVD HEVC encode in amdgpu cs Signed-off-by: James Zhu <[email protected]> Reviewed-by: Boyuan Zhang <[email protected]>
* winsys/amdgpu: enable 32-bit VM allocationsMarek Olšák2018-02-171-1/+2
| | | | Reviewed-by: Samuel Pitoiset <[email protected]>
* winsys/amdgpu: allow non page-aligned size bo creation from pointerMichal Navratil2018-02-061-4/+7
| | | | | | | | | Fix INVALID_OPERATION caused by BufferData with target EXTERNAL_VIRTUAL_MEMORY_BUFFER_AMD when the buffer size is not page aligned. Signed-off-by: Marek Olšák <[email protected]> Cc: 17.3 18.0 <[email protected]>
* winsys/amdgpu: add support for syncobj signaling v3Andres Rodriguez2018-01-302-1/+86
| | | | | | | | | | Add the ability to signal a syncobj when a cs completes execution. v2: corresponding changes for gallium fence->semaphore rename v3: s/semaphore/fence for pipe objects Signed-off-by: Andres Rodriguez <[email protected]> Reviewed-by: Marek Olšák <[email protected]>
* gallium/radeon: set number of pb_cache buckets = number of heapsMarek Olšák2018-01-272-17/+12
| | | | Reviewed-by: Samuel Pitoiset <[email protected]>
* pb_cache: let drivers choose the number of bucketsMarek Olšák2018-01-272-2/+2
| | | | Reviewed-by: Samuel Pitoiset <[email protected]>
* winsys/amdgpu: fix assertion failure with UVD and VCE ringsMarek Olšák2018-01-261-2/+1
| | | | Cc: 18.0 <[email protected]>
* radeonsi: Export signalled sync file instead of -1.Bas Nieuwenhuizen2018-01-261-0/+22
| | | | | | | | -1 is considered an error for EGL_ANDROID_native_fence_sync, so we need to actually create a sync file. Fixes: f536f45250 "radeonsi: implement sync_file import/export" Reviewed-by: Dave Airlie <[email protected]>
* autotools: include meson build files in tarballDylan Baker2018-01-191-0/+2
| | | | | | | | | | | | This adds the meson.build, meson_options.txt, and a few scripts that are used exclusively by the meson build. v2: - Remove accidentally included changes needed to test make dist with LLVM > 3.9 Signed-off-by: Dylan Baker <[email protected]> Acked-by: Eric Engestrom <[email protected]> Reviewed-by: Emil Velikov <[email protected]>
* winsys/amdgpu: disable local BOs again due to worse performanceMarek Olšák2017-12-111-2/+3
| | | | | Cc: 17.3 <[email protected]> Reviewed-by: Nicolai Hähnle <[email protected]>
* winsys/amdgpu: make IBs use read-only memoryMarek Olšák2017-12-061-0/+1
| | | | | Tested-by: Dieter Nützel <[email protected]> Reviewed-by: Nicolai Hähnle <[email protected]>
* radeonsi: always place sparse buffers in VRAMNicolai Hähnle2017-12-061-2/+3
| | | | | | | | | | | | Together with "radeonsi: fix the R600_RESOURCE_FLAG_UNMAPPABLE check", this ensures that sparse buffers are placed in VRAM. Noticed by an assertion that started triggering with commit d4fac1e1d7 ("gallium/radeon: enable suballocations for VRAM with no CPU access") Fixes KHR-GL45.sparse_buffer_tests.BufferStorageTest in debug builds. Reviewed-by: Marek Olšák <[email protected]> Tested-by: Dieter Nützel <[email protected]>
* winsys/amdgpu: add RADEON_FLAG_READ_ONLYMarek Olšák2017-12-051-1/+8
| | | | Reviewed-by: Nicolai Hähnle <[email protected]>
* gallium/radeon: move setting VRAM|GTT into winsysesMarek Olšák2017-12-051-1/+13
| | | | | | The combined VRAM|GTT heap will be removed. Reviewed-by: Nicolai Hähnle <[email protected]>
* amd: remove always-true BRAHMA_BUILD defineEric Engestrom2017-12-011-3/+1
| | | | | Signed-off-by: Eric Engestrom <[email protected]> Reviewed-by: Marek Olšák <[email protected]>
* r300,r600,radeonsi: replace RADEON_FLUSH_* with PIPE_FLUSH_*Marek Olšák2017-11-292-3/+3
| | | | | | and handle PIPE_FLUSH_HINT_FINISH in r300. Reviewed-by: Nicolai Hähnle <[email protected]>
* winsys/amdgpu: add vcn enc cs supportBoyuan Zhang2017-11-171-1/+6
| | | | | | | New cs support is needed for vcn encode Signed-off-by: Boyuan Zhang <[email protected]> Acked-by: Christian König <[email protected]>
* meson: don't use build_by_default for specific gallium driversDylan Baker2017-11-131-1/+0
| | | | | | | | | | | | | | | | | | | Using build_by_default : false is convenient for dependencies that can be pulled in by various diverse components of the build system, the gallium hardware/software drivers and state trackers do not fit that description. Instead, these should be guarded using the variable that tracks whether that driver should be enabled. This leaves a few helper libraries: trace, rbug, etc, and the generic winsys bits as `build_by_default : false` because there are a large number of gallium components that pull them in. v2: - remove build_by_default from winsys convenience libs as well. v3: - Always put drivers before winsys for consistency Signed-off-by: Dylan Baker <[email protected]> Tested-by: Lionel Landwerlin <[email protected]> (v1) Reviewed-by: Eric Anholt <[email protected]>
* winsys/amdgpu: handle cs_add_fence_dependency for deferred/unsubmitted fencesNicolai Hähnle2017-11-092-11/+19
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The idea is to fix the following interleaving of operations that can arise from deferred fences: Thread 1 / Context 1 Thread 2 / Context 2 -------------------- -------------------- f = deferred flush <------- application-side synchronization -------> fence_server_sync(f) ... flush() flush() We will now stall in fence_server_sync until the flush of context 1 has completed. This scenario was unlikely to occur previously, because applications seem to be doing Thread 1 / Context 1 Thread 2 / Context 2 -------------------- -------------------- f = glFenceSync() glFlush() <------- application-side synchronization -------> glWaitSync(f) ... and indeed they probably *have* to use this ordering to avoid deadlocks in the GLX model, where all GL operations conceptually go through a single connection to the X server. However, it's less clear whether applications have to do this with other WSI (i.e. EGL). Besides, even this sequence of GL commands can be translated into the Gallium-level sequence outlined above when Gallium threading and asynchronous flushes are used. So it makes sense to be more robust. As a side effect, we no longer busy-wait on submission_in_progress. We won't enable asynchronous flushes on radeon, but add a cs_add_fence_dependency stub anyway to document the potential issue. Reviewed-by: Marek Olšák <[email protected]>
* util: move os_time.[ch] to src/utilNicolai Hähnle2017-11-092-2/+3
| | | | Reviewed-by: Marek Olšák <[email protected]>
* amdgpu: use simple mtxTimothy Arceri2017-11-095-44/+45
| | | | | Reviewed-by: Marek Olšák <[email protected]> Reviewed-by: Nicolai Hähnle <[email protected]>
* amd/addrlib: update to latest versionMarek Olšák2017-11-081-1/+0
| | | | | | | | | | | | This uses C++11 initializer lists. I just overwrote all Mesa files with internal addrlib and discarded hunks that we should probably keep, but I might have missed something. The code depending on ADDR_AM_BUILD is removed. We can add it back next time if needed. Acked-by: Nicolai Hähnle <[email protected]>
* winsys/amdgpu: Add R600_DEBUG flag to reserve VMID per ctx.Andrey Grodzovsky2017-11-032-0/+13
| | | | | | | | | | Fixes reverted patch f03b7c9 by doing VMID reservation per process and not per context. Also updates required amdgpu libdrm version since the change involved interface updates in amdgpu libdrm. Signed-off-by: Andrey Grodzovsky <[email protected]> Signed-off-by: Marek Olšák <[email protected]>
* radeonsi: remove 'Authors:' commentsMarek Olšák2017-11-027-28/+0
| | | | | | | It's inaccurate. Instead, see the copyright and use "git log" and "git blame" to know the authorship. Acked-by: Nicolai Hähnle <[email protected]>
* Revert "winsys/amdgpu: Add R600_DEBUG flag to reserve VMID per ctx."Marek Olšák2017-11-014-13/+0
| | | | | | This reverts commit f03b7c9ad92c1656a221297819fbc6d065cc0af7. The libdrm interface is wrong.
* winsys/amdgpu: Add R600_DEBUG flag to reserve VMID per ctx.Andrey Grodzovsky2017-10-314-0/+13
| | | | | Signed-off-by: Andrey Grodzovsky <[email protected]> Signed-off-by: Marek Olšák <[email protected]>
* radeonsi: add GFX-IB-size query to the HUDMarek Olšák2017-10-173-0/+4
| | | | | | It shows the sum of all IBs per frame. Reviewed-by: Nicolai Hähnle <[email protected]>
* winsys/amdgpu: disable CPU caching for GFX & SDMA IBsMarek Olšák2017-10-171-4/+9
| | | | | | This should decrease IB fetch latency. Reviewed-by: Nicolai Hähnle <[email protected]>
* winsys/amdgpu: don't do read-modify-write on command buffersMarek Olšák2017-10-172-4/+16
| | | | | | i.e. don't use |= Reviewed-by: Nicolai Hähnle <[email protected]>
* meson: build radeonsiDylan Baker2017-10-161-0/+36
| | | | | | | | This builds the radeonsi (and radeon) window system bits and gallium driver bits. Signed-off-by: Dylan Baker <[email protected]> Reviewed-by: Eric Anholt <eric at anholt.net>
* winsys/amdgpu: implement sync_file import/exportMarek Olšák2017-10-122-7/+126
| | | | | | syncobj is used internally for interactions with command submission. Reviewed-by: Nicolai Hähnle <[email protected]>
* Revert "winsys/amdgpu: disable local BOs on Raven"Marek Olšák2017-09-121-2/+1
| | | | | | This reverts commit 1cda9a2fee05effd9c64bd773bc6005281593662. It works now.
* winsys/amdgpu: use the new raw CS APIMarek Olšák2017-09-112-77/+93
| | | | | | This also cleans things up. Reviewed-by: Nicolai Hähnle <[email protected]>
* radeonsi: implement pipe_context::fence_server_syncMarek Olšák2017-09-111-0/+16
| | | | | | This will be more useful once we have sync_file support. Reviewed-by: Nicolai Hähnle <[email protected]>
* winsys/amdgpu: factor out some fence dependency code into separate functionsMarek Olšák2017-09-111-21/+34
| | | | Reviewed-by: Nicolai Hähnle <[email protected]>
* winsys/amdgpu: rename fence_dependency functionsMarek Olšák2017-09-111-12/+12
| | | | Reviewed-by: Nicolai Hähnle <[email protected]>
* winsys/amdgpu: don't allow interprocess resource sharing for IBsMarek Olšák2017-09-111-1/+2
| | | | | | | Now we should get IB submissions with bo_list == NULL when DRI buffers aren't referenced. Reviewed-by: Nicolai Hähnle <[email protected]>
* winsys/amdgpu: disable local BOs on RavenMarek Olšák2017-09-071-1/+2
| | | | | | It hangs with a high degree of reproducibility. Acked-by: Nicolai Hähnle <[email protected]>
* winsys/amdgpu: set AMDGPU_GEM_CREATE_VM_ALWAYS_VALID if possible v2Christian König2017-08-313-5/+27
| | | | | | | | | | | When the kernel supports it set the local flag and stop adding those BOs to the BO list. Can probably be optimized much more. v2: rename new flag to AMDGPU_GEM_CREATE_VM_ALWAYS_VALID Reviewed-by: Marek Olšák <[email protected]>
* radeonsi: set a per-buffer flag that disables inter-process sharing (v4)Marek Olšák2017-08-311-13/+17
| | | | | | | | | | | For lower overhead in the CS ioctl. Winsys allocators are not used with interprocess-sharable resources. v2: It shouldn't crash anymore, but the kernel will reject the new flag. v3 (christian): Rename the flag, avoid sending those buffers in the BO list. v4 (christian): Remove setting the kernel flag for now Reviewed-by: Marek Olšák <[email protected]>
* winsys/amdgpu: add BO to the global list only when RADEON_ALL_BOS is setSamuel Pitoiset2017-08-304-11/+17
| | | | | | | Only useful when that debug option is enabled. Signed-off-by: Samuel Pitoiset <[email protected]> Reviewed-by: Marek Olšák <[email protected]>
* radeonsi: remove Constant Engine supportMarek Olšák2017-08-222-108/+2
| | | | | | | | We have come to the conclusion that it doesn't improve performance. Tested-by: Dieter Nützel <[email protected]> Reviewed-by: Samuel Pitoiset <[email protected]> Reviewed-by: Nicolai Hähnle <[email protected]>
* gallium/radeon: print all members of radeon_info with R600_DEBUG=infoMarek Olšák2017-08-071-1/+1
| | | | | | | also set max_alignment on amdgpu. Tested-by: Dieter Nützel <[email protected]> Reviewed-by: Nicolai Hähnle <[email protected]>
* winsys/amdgpu: enable computation of tile swizzleMarek Olšák2017-08-042-1/+13
| | | | | Reviewed-by: Dave Airlie <[email protected]> Reviewed-by: Nicolai Hähnle <[email protected]>
* ac/surface: increment surf_index only when tile swizzle is allowedMarek Olšák2017-08-041-0/+1
| | | | | Reviewed-by: Dave Airlie <[email protected]> Reviewed-by: Nicolai Hähnle <[email protected]>
* gallium: add pipe_screen_config to screen_create functionsNicolai Hähnle2017-08-022-4/+5
| | | | | | | This allows a more generic mechanism for passing user configurations into drivers by accessing the dri options directly. Reviewed-by: Marek Olšák <[email protected]>
* radeonsi: prevent a deadlock in util_queue_add_job with too many GL contextsMarek Olšák2017-07-171-1/+2
| | | | | | | | | | | | | | | | If the queue is full, util_queue_add_job will wait while bo_fence_lock is held. It pb_slab wants to reuse a buffer, it will lock the pb_slab mutex and try to check BO fence busyness, but it has to wait for bo_fence_lock to get released. Both bo_fence_lock and pb_slab mutex are locked now. When the CS thread unreferences and releases a suballocated buffer, it will try to lock the pb_slab mutex and has to wait. The CS thread can't finish its job in order to free a queue slot and unblock util_queue_add_job ==> deadlock. Reviewed-by: Nicolai Hähnle <[email protected]>
generated by cgit v1.2.3 (git 2.39.1) at 2024-12-20 00:41:05 +0000