aboutsummaryrefslogtreecommitdiff
path: root/drivers/gpu/drm/amd
AgeCommit message (Collapse)AuthorFilesLines
2017-12-04drm/amd/scheduler:introduce guilty pointer memberMonk Liu10-11/+13
this member will be used later, it will points to the real var inside of context and CS_SUBMIT & gpu schdduler can decide if skip a job depends on context->guilty or *entity->guilty Signed-off-by: Monk Liu <[email protected]> Reviewed-by: Chunming Zhou <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2017-12-04drm/amdgpu:add hang_limit for sched(v2)Monk Liu3-3/+8
since gpu_scheduler source domain cannot access amdgpu variable so need create the hang_limit membewr for sched, and it can refer it for the upcoming GPU RESET patches v2: make hang_limit a parameter of sched_init() Signed-off-by: Monk Liu <[email protected]> Reviewed-by: Chunming Zhou <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2017-12-04drm/amdgpu:cleanup force_completionMonk Liu3-27/+11
cleanups, now only operate on the given ring Signed-off-by: Monk Liu <[email protected]> Reviewed-by: Alex Deucher <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2017-12-04drm/amdgpu: Avoid accessing job->entity after the job is scheduled.Andrey Grodzovsky4-23/+18
Bug: amdgpu_job_free_cb was accessing s_job->s_entity when the allocated amdgpu_ctx (and the entity inside it) were already deallocated from amdgpu_cs_parser_fini. Fix: Save job's priority on it's creation instead of accessing it from s_entity later on. Signed-off-by: Andrey Grodzovsky <[email protected]> Reviewed-by: Andres Rodriguez <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2017-12-04Merge tag 'drm-misc-next-2017-11-30' of ↵Dave Airlie3-9/+1
git://anongit.freedesktop.org/drm/drm-misc into drm-next Cross-subsystem Changes: - device tree doc for the Mitsubishi AA070MC01 and Tianma TM070RVHG71 panels (Lukasz Majewski) and for a 2nd endpoint on stm32 (Philippe Cornu) Core Changes: The most important changes are: - Add drm_driver .last_close and .output_poll_changed helpers to reduce fbdev emulation footprint in drivers (Noralf) - Fix plane clipping in core and for vmwgfx (Ville) Then we have a bunch of of improvement for print and debug such as the addition of a framebuffer debugfs file. ELD connector, HDMI and improvements. And a bunch of misc improvements, clean ups and style changes and doc updates [airlied: drop eld bits from amdgpu_dm] Driver Changes: - sii8620: filter unsupported modes and add DVI mode support (Maciej Purski) - rockchip: analogix_dp: Remove unnecessary init code (Jeffy Chen) - virtio, cirrus: add fb create_handle support to enable screenshots(Lepton Wu) - virtio: replace reference/unreference with get/put (Aastha Gupta) - vc4, gma500: Convert timers to use timer_setup() (Kees Cook) - vc4: Reject HDMI modes with too high of clocks (Eric) - vc4: Add support for more pixel formats (Dave Stevenson) - stm: dsi: Rename driver name to "stm32-display-dsi" (Philippe Cornu) - stm: ltdc: add a 2nd endpoint (Philippe Cornu) - via: use monotonic time for VIA_WAIT_IRQ (Arnd Bergmann) * tag 'drm-misc-next-2017-11-30' of git://anongit.freedesktop.org/drm/drm-misc: (96 commits) drm/bridge: tc358767: add copyright lines MAINTAINERS: change maintainer for Rockchip drm drivers drm/vblank: Fix vblank timestamp debugs drm/via: use monotonic time for VIA_WAIT_IRQ dma-buf: Fix ifnullfree.cocci warnings drm/printer: Add drm_vprintf() drm/edid: Allow HDMI infoframe without VIC or S3D video/hdmi: Allow "empty" HDMI infoframes dma-buf/fence: Fix lock inversion within dma-fence-array drm/sti: Handle return value of platform_get_irq_byname drm/vc4: Add support for NV21 and NV61. drm/vc4: Use .pixel_order instead of custom .flip_cbcr drm/vc4: Add support for DRM_FORMAT_RGB888 and DRM_FORMAT_BGR888 drm: Move drm_plane_helper_check_state() into drm_atomic_helper.c drm: Check crtc_state->enable rather than crtc->enabled in drm_plane_helper_check_state() drm/vmwgfx: Try to fix plane clipping drm/vmwgfx: Use drm_plane_helper_check_state() drm/vmwgfx: Remove bogus crtc coords vs fb size check gpu: gma500: remove unneeded DRIVER_LICENSE #define drm: don't link DP aux i2c adapter to the hardware device node ...
2017-12-01Merge branch 'drm-fixes-4.15' of git://people.freedesktop.org/~agd5f/linux ↵Dave Airlie27-182/+394
into drm-fixes Fixes for 4.15. Highlights: - DC fixes for S3, gamma, audio, pageflipping, etc. - fix a regression in radeon from kfd removal - fix a ttm regression with swiotlb disabled - misc other fixes * 'drm-fixes-4.15' of git://people.freedesktop.org/~agd5f/linux: (36 commits) drm/radeon: remove init of CIK VMIDs 8-16 for amdkfd drm/ttm: fix populate_and_map() functions once more drm/amd/display: USB-C / thunderbolt dock specific workaround drm/amd/display: Switch to drm_atomic_helper_wait_for_flip_done drm/amd/display: fix gamma setting drm/amd/display: Do not put drm_atomic_state on resume drm/amd/display: Fix couple more inconsistent NULL checks in dc_resource drm/amd/display: Fix potential NULL and mem leak in create_links drm/amd/display: Fix hubp check in set_cursor_position drm/amd/display: Fix use before NULL check in validate_timing drm/amd/display: Bunch of smatch error and warning fixes in DC drm/amd/display: Fix amdgpu_dm bugs found by smatch drm/amd/display: try to find matching audio inst for enc inst first drm/amd/display: fix seq issue: turn on clock before programming afmt. drm/amd/display: fix memory leaks on error exit return drm/amd/display: check plane state before validating fbc drm/amd/display: Do DC mode-change check when adding CRTCs drm/amd/display: Revert noisy assert messages drm/amd/display: fix split viewport rounding error drm/amd/display: Check aux channel before MST resume ...
2017-12-01Merge tag 'drm-amdkfd-fixes-2017-11-26' of ↵Dave Airlie4-16/+56
git://people.freedesktop.org/~gabbayo/linux into drm-fixes This is amdkfd pull request for -rc2. It contains three small fixes to the CIK SDMA code, compilation error fix in kfd_ioctl.h and fix to accessing a pointer after it was released. * tag 'drm-amdkfd-fixes-2017-11-26' of git://people.freedesktop.org/~gabbayo/linux: uapi: fix linux/kfd_ioctl.h userspace compilation errors drm/amdkfd: fix amdkfd use-after-free GP fault drm/amdkfd: Fix SDMA oversubsription handling drm/amdkfd: Fix SDMA ring buffer size calculation drm/amdgpu: Fix SDMA load/unload sequence on HWS disabled mode
2017-11-28drm/amd/display: USB-C / thunderbolt dock specific workaroundHersen Wu1-45/+40
reading dpcd 0x600 cause link loss for a particular USB-C dock with thurderbolt. workaround by avoiding dcpd 0x600 read unless it's necessary. Signed-off-by: Hersen Wu <[email protected]> Reviewed-by: Tony Cheng <[email protected]> Acked-by: Harry Wentland <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2017-11-28drm/amd/display: Switch to drm_atomic_helper_wait_for_flip_doneAndrey Grodzovsky1-1/+1
This new helper function is advised to be used for drviers that use the nonblocking commit tracking support instead of drm_atomic_helper_wait_for_vblanks. Signed-off-by: Andrey Grodzovsky <[email protected]> Reviewed-by: Harry Wentland <[email protected]> Reviewed-and-Tested-by: Michel Dänzer <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2017-11-28drm/amd/display: fix gamma settingRoman Li1-1/+2
Adding gamma changed check as condition for affected plane. We ignored adding plane as affected if modeset was not required. But for color management change we still need it. Signed-off-by: Roman Li <[email protected]> Reviewed-by: Tony Cheng <[email protected]> Acked-by: Harry Wentland <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2017-11-28drm/amd/display: Do not put drm_atomic_state on resumeLeo (Sunpeng) Li1-1/+0
drm_atomic_helper_resume now puts it for us. See relevant patch here: https://lists.freedesktop.org/archives/dri-devel/2017-October/154268.html Signed-off-by: Leo (Sunpeng) Li <[email protected]> Reviewed-by: Harry Wentland <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2017-11-28drm/amd/display: Fix couple more inconsistent NULL checks in dc_resourceHarry Wentland1-2/+7
Found by smatch: drivers/gpu/drm/amd/amdgpu/../display/dc/core/dc_resource.c:1001 acquire_free_pipe_for_stream() error: we previously assumed 'head_pipe' could be null (see line 998) drivers/gpu/drm/amd/amdgpu/../display/dc/core/dc_resource.c:1808 dc_validate_global_state() error: we previously assumed 'new_ctx' could be null (see line 1778) Signed-off-by: Harry Wentland <[email protected]> Reviewed-by: Tony Cheng <[email protected]> Acked-by: Harry Wentland <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2017-11-28drm/amd/display: Fix potential NULL and mem leak in create_linksHarry Wentland1-4/+10
Found by smatch: drivers/gpu/drm/amd/amdgpu/../display/dc/core/dc.c:148 create_links() error: potential null dereference 'link->link_enc'. (kzalloc returns null) Signed-off-by: Harry Wentland <[email protected]> Reviewed-by: Tony Cheng <[email protected]> Acked-by: Harry Wentland <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2017-11-28drm/amd/display: Fix hubp check in set_cursor_positionHarry Wentland2-12/+4
Found by smatch: drivers/gpu/drm/amd/amdgpu/../display/dc/core/dc_stream.c:298 dc_stream_set_cursor_position() error: we previously assumed 'hubp' could be null (see line 294) Signed-off-by: Harry Wentland <[email protected]> Reviewed-by: Tony Cheng <[email protected]> Acked-by: Harry Wentland <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2017-11-28drm/amd/display: Fix use before NULL check in validate_timingHarry Wentland1-4/+4
Found by smatch: drivers/gpu/drm/amd/amdgpu/../display/dc/dce110/dce110_timing_generator.c:1124 dce110_timing_generator_validate_timing() warn: variable dereferenced before check 'timing' (see line 1116) Signed-off-by: Harry Wentland <[email protected]> Reviewed-by: Tony Cheng <[email protected]> Acked-by: Harry Wentland <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2017-11-28drm/amd/display: Bunch of smatch error and warning fixes in DCHarry Wentland4-9/+16
drivers/gpu/drm/amd/amdgpu/../display/dc/basics/log_helpers.c:79 dc_conn_log() error: buffer overflow 'signal_type_info_tbl' 10 <= 10 drivers/gpu/drm/amd/amdgpu/../display/dc/bios/bios_parser.c:266 bios_parser_get_dst_obj() error: uninitialized symbol 'id'. drivers/gpu/drm/amd/amdgpu/../display/dc/dce/dce_audio.c:357 dce_aud_az_enable() warn: inconsistent indenting drivers/gpu/drm/amd/amdgpu/../display/dc/dcn10/dcn10_resource.c:958 dcn10_acquire_idle_pipe_for_layer() error: we previously assumed 'head_pipe' could be null (see line 952) Signed-off-by: Harry Wentland <[email protected]> Reviewed-by: Sun peng Li <[email protected]> Acked-by: Harry Wentland <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2017-11-28drm/amd/display: Fix amdgpu_dm bugs found by smatchHarry Wentland1-6/+10
drivers/gpu/drm/amd/amdgpu/../display/amdgpu_dm/amdgpu_dm.c:2760 create_eml_sink() warn: variable dereferenced before check 'aconnector->base.edid_blob_ptr' (see line 2758) drivers/gpu/drm/amd/amdgpu/../display/amdgpu_dm/amdgpu_dm.c:4270 amdgpu_dm_atomic_commit_tail() warn: variable dereferenced before check 'dm_new_crtc_state->stream' (see line 4266) drivers/gpu/drm/amd/amdgpu/../display/amdgpu_dm/amdgpu_dm.c:4417 dm_restore_drm_connector_state() warn: variable dereferenced before check 'disconnected_acrtc' (see line 4415) Signed-off-by: Harry Wentland <[email protected]> Reviewed-by: Sun peng Li <[email protected]> Acked-by: Harry Wentland <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2017-11-28drm/amd/display: try to find matching audio inst for enc inst firstCharlene Liu2-5/+8
[Description] in eDP+ HDMI/DP clone or extended configuration, audio inst changed from inst 1 to inst0. No failure related this though, just playback device endpoint inst changed. Also remove one addition register read. Signed-off-by: Charlene Liu <[email protected]> Reviewed-by: Tony Cheng <[email protected]> Acked-by: Harry Wentland <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2017-11-28drm/amd/display: fix seq issue: turn on clock before programming afmt.Charlene Liu2-12/+15
Signed-off-by: Charlene Liu <[email protected]> Reviewed-by: Krunoslav Kovac <[email protected]> Acked-by: Harry Wentland <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2017-11-28drm/amd/display: fix memory leaks on error exit returnColin Ian King1-5/+7
Currently in the case where some of the allocations fail for dce110_tgv, dce110_xfmv, dce110_miv or dce110_oppv then the exit return path ends up leaking allocated objects. Fix this by kfree'ing them before returning. Also re-work the comparison of the null pointers to use the !ptr idiom. Detected by CoverityScan, CID#1460246, 1460325, 1460324, 1460392 ("Resource Leak") Fixes: c4562236b3bc ("drm/amd/dc: Add dc display driver (v2)") Signed-off-by: Colin Ian King <[email protected]> Reviewed-by: Harry Wentland <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2017-11-28drm/amd/display: check plane state before validating fbcShirish S1-0/+4
While validation fbc, array_mode of the pipe is accessed without checking plane_state exists for it. Causing to null pointer dereferencing followed by reboot when a crtc associated with external display(not connected) is page flipped. This patch adds a check for plane_state before using it to validate fbc. Signed-off-by: Shirish S <[email protected]> Reviewed-by: Roman Li <[email protected]> Reviewed-by: Harry Wentland <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2017-11-28drm/amd/display: Do DC mode-change check when adding CRTCsLeo (Sunpeng) Li1-1/+1
Within atomic check, dm_update_crtcs_state is called twice. First to remove from the dc_state, and subsequently to add to it. In both calls, a secondary mode-change check is done using dc-level states. We shouldn't be doing this while removing, since a new dc_stream_state has not been created to do the necessary comparison. Because of this, the mode_changed flag within the DRM state can be mistakenly set to false. Doing so only when adding prevents this. We are also guaranteed that a call to add will come after remove, or else the atomic check fails (and a commit will not happen). Signed-off-by: Leo (Sunpeng) Li <[email protected]> Reviewed-by: Tony Cheng <[email protected]> Acked-by: Harry Wentland <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2017-11-28drm/amd/display: Revert noisy assert messagesJordan Lazare1-4/+0
This partially reverts commit 4fb48bb66211 ("dc: fix split viewport rounding error"). Signed-off-by: Jordan Lazare <[email protected]> Reviewed-by: Dmytro Laktyushkin <[email protected]> Acked-by: Harry Wentland <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2017-11-28drm/amd/display: fix split viewport rounding errorDmytro Laktyushkin2-6/+8
Signed-off-by: Dmytro Laktyushkin <[email protected]> Reviewed-by: Tony Cheng <[email protected]> Acked-by: Harry Wentland <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2017-11-28drm/amd/display: Check aux channel before MST resumeJerry (Fangzhi) Zuo1-1/+2
It is to fix: MST display failed to resume from S3 At the beginning of resume from S3, need to check if mgr->aux is NULL. Fake MST encoder doesn't have real aux channel. Signed-off-by: Jerry (Fangzhi) Zuo <[email protected]> Reviewed-by: Roman Li <[email protected]> Acked-by: Harry Wentland <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2017-11-28drm/amd/display: fix split recout offsetDmytro Laktyushkin1-2/+2
Previous recout calculation fix changed recout size rounding and affected the offset when it should not have Signed-off-by: Dmytro Laktyushkin <[email protected]> Reviewed-by: Tony Cheng <[email protected]> Acked-by: Harry Wentland <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2017-11-28drm/amd/display: Don't reject 3D timingsAndrew Jiang1-3/+0
Signed-off-by: Andrew Jiang <[email protected]> Reviewed-by: Tony Cheng <[email protected]> Acked-by: Harry Wentland <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2017-11-28drm/amd/display: Handle as MST first and then DP dongle if sink support bothHersen Wu1-16/+16
Signed-off-by: Hersen Wu <[email protected]> Reviewed-by: Tony Cheng <[email protected]> Reviewed-by: Wenjing Liu <[email protected]> Acked-by: Harry Wentland <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2017-11-28drm/amd/display: fix split recout calculationDmytro Laktyushkin1-5/+3
Recout split rounding code was wrong Signed-off-by: Dmytro Laktyushkin <[email protected]> Reviewed-by: Tony Cheng <[email protected]> Acked-by: Harry Wentland <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2017-11-28drm/amd/display: Fix S3 topology changeRoman Li1-0/+4
Clean fake sink flag on resume if real sink connected. Fixing S3 topology change problem like this: 1) x desktop with 1 or > displays 2) unplug display 3) suspend 4) replug same display 5) resume without this change replugged display doesn't light up Signed-off-by: Roman Li <[email protected]> Reviewed-by: Sun peng Li <[email protected]> Acked-by: Harry Wentland <[email protected]> Reviewed-by: Andrey Grodzovsky <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2017-11-28drm/amd/display: Add timing validation against dongle capEric Yang4-3/+91
For DP active dongles, the dpcd dongle caps are read but not used to validate mode timing. This addresses this. In particular, this change fixes light up on the HDMI 4k TV connected through DP active dongle. Since the 4k TV defaults to YCbCr420, which the dongle don't support. This change does not address MST cases, a more generalized approach must be taken for that. Signed-off-by: Eric Yang <[email protected]> Reviewed-by: Tony Cheng <[email protected]> Acked-by: Harry Wentland <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2017-11-28drm/amd/display: Should disable when new stream is nullLeo (Sunpeng) Li1-2/+4
core_link_disable_stream should be called when the new stream is null (i.e. want to disable). Modify the if condition to reflect that. Signed-off-by: Leo (Sunpeng) Li <[email protected]> Reviewed-by: Harry Wentland <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2017-11-28drm/amd/display: Add null check for 24BPP (xfm and dpp)Bhawanpreet Lakha1-6/+12
Fixes Nullptr error when trying 24BPP Signed-off-by: Bhawanpreet Lakha <[email protected]> Reviewed-by: Tony Cheng <[email protected]> Acked-by: Harry Wentland <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2017-11-28drm/amdgpu: drop experimental flag for ravenAlex Deucher1-1/+1
Signed-off-by: Alex Deucher <[email protected]>
2017-11-28drm/amdgpu: don't try to move pinned BOsChristian König1-0/+4
Never try to move pinned BOs during CS. Signed-off-by: Christian König <[email protected]> Reviewed-by: Michel Dänzer <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2017-11-28drm/amdgpu: Use unsigned ring indices in amdgpu_queue_mgr_mapMichel Dänzer2-4/+4
This matches the corresponding UAPI fields. Treating the ring index as signed could result in accessing random unrelated memory if the MSB was set. Fixes: effd924d2f3b ("drm/amdgpu: untie user ring ids from kernel ring ids v6") Cc: [email protected] Reviewed-by: Alex Deucher <[email protected]> Reviewed-by: Christian König <[email protected]> Signed-off-by: Michel Dänzer <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2017-11-28drm/amdgpu: Set adev->vcn.irq.num_types for VCNMichel Dänzer1-1/+1
We were setting adev->uvd.irq.num_types instead. Fixes: 9b257116e784 ("drm/amdgpu: add vcn enc irq support") Reviewed-by: Alex Deucher <[email protected]> Reviewed-by: Christian König <[email protected]> Signed-off-by: Michel Dänzer <[email protected]> Signed-off-by: Alex Deucher <[email protected]> Cc: [email protected]
2017-11-28Revert "drm/amdgpu: fix rmmod KCQ disable failed error"Alex Deucher3-3/+5
This reverts commit 446947b44fb8cabc0213ff4efd706931e36b1963. this patch is incorrrect, amdgpu_ucode_bo_fini always called after gfx_hw_fini. Signed-off-by: Alex Deucher <[email protected]>
2017-11-28drm/amdgpu: used cached gca values for cik_read_registerAlex Deucher1-16/+95
Using the cached values has less latency for bare metal and prevents reading back bogus values if the engine is powergated. This was implemented for VI and SI, but somehow CIK got missed. Reviewed-by: Christian König <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2017-11-28drm/amdgpu/gfx7: cache raster_config valuesAlex Deucher1-0/+16
We did this for gfx6 and 8, but somehow missed gfx7. Reviewed-by: Christian König <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2017-11-28drm/amdgpu: move UVD/VCE and VCN structure out from unionLeo Liu1-12/+8
With the enablement of VCN Dec and Enc from user space, User space queries kernel for the IP information, if HW has UVD/VCE, the info comes from these IP blocks, but this could end up mis-interpret for VCN when they are in the union, the other way same when HW with VCN block. Signed-off-by: Leo Liu <[email protected]> Acked-by: Alex Deucher <[email protected]> Reviewed-by: Christian König <[email protected]> Signed-off-by: Alex Deucher <[email protected]> Fixes: 95d0906f8506 ("drm/amdgpu: add initial vcn support and decode tests") Cc: [email protected] Reviewed-and-Tested-by: Michel Dänzer <[email protected]>
2017-11-27drm/amdkfd: Simplify locking during process creationYong Zhao1-25/+21
Also fixes error handling if kfd_process_init_cwsr fails. Signed-off-by: Yong Zhao <[email protected]> Signed-off-by: Felix Kuehling <[email protected]> Reviewed-by: Oded Gabbay <[email protected]> Signed-off-by: Oded Gabbay <[email protected]>
2017-11-27drm/amdkfd: Factor PDD destruction out of kfd_process_wq_releaseFelix Kuehling1-16/+24
Signed-off-by: Felix Kuehling <[email protected]> Reviewed-by: Oded Gabbay <[email protected]> Signed-off-by: Oded Gabbay <[email protected]>
2017-11-27drm/amdkfd: Reduce nesting in kfd_create_process_device_dataFelix Kuehling1-11/+12
Signed-off-by: Felix Kuehling <[email protected]> Reviewed-by: Oded Gabbay <[email protected]> Signed-off-by: Oded Gabbay <[email protected]>
2017-11-27drm/amdkfd: Return NULL if kfd_lookup_process_by_pasid failsYong Zhao1-2/+3
If no matching process is found, return NULL instead of a pointer to the last process in the kfd_processes_table. Signed-off-by: Yong Zhao <[email protected]> Signed-off-by: Felix Kuehling <[email protected]> Reviewed-by: Oded Gabbay <[email protected]> Signed-off-by: Oded Gabbay <[email protected]>
2017-11-27drm/amdkfd: Use ref count to prevent kfd_process destructionFelix Kuehling3-10/+21
Use a reference counter instead of a lock to prevent process destruction while functions running out of process context are using the kfd_process structure. In many cases these functions don't need the structure to be locked. In the few cases that really do need the process lock, take it explicitly. This helps simplify lock dependencies between the process lock and other locks, particularly amdgpu and mm_struct locks. This will be important when amdgpu calls back to amdkfd for memory evictions. Signed-off-by: Felix Kuehling <[email protected]> Acked-by: Christian König <[email protected]> Reviewed-by: Oded Gabbay <[email protected]> Signed-off-by: Oded Gabbay <[email protected]>
2017-11-27drm/amdkfd: Make kfd_process reference countedFelix Kuehling2-36/+26
This will be used to elliminate the use of the process lock for preventing concurrent process destruction. This will simplify lock dependencies between KFD and KGD. This also simplifies the process destruction in a few ways: * Don't allocate work struct dynamically * Remove unnecessary hack that increments mm reference counter * Remove unnecessary process locking during destruction Signed-off-by: Felix Kuehling <[email protected]> Reviewed-by: Oded Gabbay <[email protected]> Signed-off-by: Oded Gabbay <[email protected]>
2017-11-27drm/amdkfd: Get reference to lead_thread task structFelix Kuehling1-0/+4
Increment the kfd_process.lead_thread's reference counter to make it safe to dereference. This is needed for getting a safe reference to the process' mm_struct. Signed-off-by: Felix Kuehling <[email protected]> Reviewed-by: Oded Gabbay <[email protected]> Signed-off-by: Oded Gabbay <[email protected]>
2017-11-27drm/amdkfd: Add debugfs support to KFDFelix Kuehling12-0/+399
This commit adds several debugfs entries for kfd: kfd/hqds: dumps all HQDs on all GPUs for KFD-controlled compute and SDMA RLC queues kfd/mqds: dumps all MQDs of all KFD processes on all GPUs kfd/rls: dumps HWS runlists on all GPUs Signed-off-by: Yong Zhao <[email protected]> Signed-off-by: Felix Kuehling <[email protected]> Reviewed-by: Oded Gabbay <[email protected]> Signed-off-by: Oded Gabbay <[email protected]>
2017-11-27drm/amdgpu: Add kfd2kgd APIs for dumping HQDsFelix Kuehling3-0/+165
This can be used by KFD for debugging features, such as dumping HQDs in debugfs. Signed-off-by: Felix Kuehling <[email protected]> Acked-by: Oded Gabbay <[email protected]> Signed-off-by: Oded Gabbay <[email protected]>