Changes in msm_gpu.h

d7a5ac67 - drm/msm: Extend gpu devcore dumps with pgtbl info

Tue, 10 Dec 2024 16:51:22 +0000

drm/msm: Extend gpu devcore dumps with pgtbl infoIn the case of iova fault triggered devcore dumps, include additionaldebug information based on what we think is the current page tables,including the TTBR0 value (which should match what we have inadreno_smmu_fault_info unless things have gone horribly wrong), andthe pagetable entries traversed in the process of resolving thefaulting iova.Signed-off-by: Rob Clark Patchwork: https://patchwork.freedesktop.org/patch/628117/ List of files: /linux-6.15/drivers/gpu/drm/msm/msm_gpu.h

3241504e - drm/msm/a6xx: Track current_ctx_seqno per ring

Thu, 03 Oct 2024 16:12:51 +0000

drm/msm/a6xx: Track current_ctx_seqno per ringWith preemption it is not enough to track the current_ctx_seqno globallyas execution might switch between rings.This is especially problematic when current_ctx_seqno is used todetermine whether a page table switch is necessary as it might lead tosecurity bugs.Track current context per ring.Tested-by: Rob Clark Tested-by: Neil Armstrong # on SM8650-QRDTested-by: Neil Armstrong # on SM8550-QRDTested-by: Neil Armstrong # on SM8450-HDKSigned-off-by: Antonino Maniscalco Patchwork: https://patchwork.freedesktop.org/patch/618012/Signed-off-by: Rob Clark List of files: /linux-6.15/drivers/gpu/drm/msm/msm_gpu.h

f2608b70 - drm/msm: Add obj flags to gpu devcoredump

Mon, 13 May 2024 15:51:47 +0000

drm/msm: Add obj flags to gpu devcoredumpWhen debugging faults, it is useful to know how the BO is mapped (cachedvs WC, gpu readonly, etc).Signed-off-by: Rob Clark Reviewed-by: Akhil P Oommen Acked-by: Konrad Dybcio Reviewed-by: Dmitry Baryshkov Patchwork: https://patchwork.freedesktop.org/patch/593854/ List of files: /linux-6.15/drivers/gpu/drm/msm/msm_gpu.h

204226de - drm/msm: Add GPU memory traces

Fri, 01 Mar 2024 18:53:45 +0000

drm/msm: Add GPU memory tracesPerfetto can use these traces to track global and per-process GPU memoryusage.Signed-off-by: Rob Clark Patchwork: https://patchwork.freedesktop.org/patch/580854/ List of files: /linux-6.15/drivers/gpu/drm/msm/msm_gpu.h

0efadfb0 - drm/msm: Drop msm_read/writel

Wed, 10 Apr 2024 21:52:52 +0000

drm/msm: Drop msm_read/writelTotally useless.Signed-off-by: Konrad Dybcio Reviewed-by: Andrew Halaney Patchwork: https://patchwork.freedesktop.org/patch/588804/Link: https://lore.kernel.org/r/20240410-topic-msm_rw-v1-1-e1fede9ffaba@linaro.orgSigned-off-by: Dmitry Baryshkov List of files: /linux-6.15/drivers/gpu/drm/msm/msm_gpu.h

38f922a5 - drm/sched: Reverse run-queue priority enumeration

Thu, 23 Nov 2023 03:08:48 +0000

drm/sched: Reverse run-queue priority enumerationReverse run-queue priority enumeration such that the higest priority is now 0,and for each consecutive integer the prioirty diminishes.Run-queues correspond to priorities. To an external observer a schedulercreated with a single run-queue, and another created withDRM_SCHED_PRIORITY_COUNT number of run-queues, should always schedulesched->sched_rq[0] with the same "priority", as that index run-queue exists inboth schedulers, i.e. a scheduler with one run-queue or many. This patch makesit so.In other words, the "priority" of sched->sched_rq[n], n >= 0, is the same forany scheduler created with any allowable number of run-queues (priorities), 0to DRM_SCHED_PRIORITY_COUNT.Cc: Rob Clark Cc: Abhinav Kumar Cc: Dmitry Baryshkov Cc: Danilo Krummrich Cc: Alex Deucher Cc: Christian König Cc: linux-arm-msm@vger.kernel.orgCc: freedreno@lists.freedesktop.orgCc: dri-devel@lists.freedesktop.orgSigned-off-by: Luben Tuikov Reviewed-by: Christian König Link: https://patchwork.freedesktop.org/patch/msgid/20231124052752.6915-6-ltuikov89@gmail.com List of files: /linux-6.15/drivers/gpu/drm/msm/msm_gpu.h

fe375c74 - drm/sched: Rename priority MIN to LOW

Wed, 15 Nov 2023 01:40:58 +0000

drm/sched: Rename priority MIN to LOWRename DRM_SCHED_PRIORITY_MIN to DRM_SCHED_PRIORITY_LOW.This mirrors DRM_SCHED_PRIORITY_HIGH, for a list of DRM scheduler prioritiesin ascending order, DRM_SCHED_PRIORITY_LOW, DRM_SCHED_PRIORITY_NORMAL, DRM_SCHED_PRIORITY_HIGH, DRM_SCHED_PRIORITY_KERNEL.Cc: Rob Clark Cc: Abhinav Kumar Cc: Dmitry Baryshkov Cc: Danilo Krummrich Cc: Alex Deucher Cc: Christian König Cc: linux-arm-msm@vger.kernel.orgCc: freedreno@lists.freedesktop.orgCc: dri-devel@lists.freedesktop.orgSigned-off-by: Luben Tuikov Reviewed-by: Christian König Link: https://patchwork.freedesktop.org/patch/msgid/20231124052752.6915-5-ltuikov89@gmail.com List of files: /linux-6.15/drivers/gpu/drm/msm/msm_gpu.h

42854f8d - drm/msm: Disallow relocs on a6xx+

Thu, 03 Aug 2023 18:32:45 +0000

drm/msm: Disallow relocs on a6xx+Mesa stopped using these pretty early in a6xx bringup[1]. Take advantageof this to disallow some legacy UABI.[1] https://gitlab.freedesktop.org/mesa/mesa/-/commit/7ef722861b691ce99be3827ed05f8c0ddf2cd66eSigned-off-by: Rob Clark Acked-by: Daniel Vetter Patchwork: https://patchwork.freedesktop.org/patch/551175/ List of files: /linux-6.15/drivers/gpu/drm/msm/msm_gpu.h

8ead9678 - drm/msm/gpu: Move BO allocation out of hw_init

Mon, 20 Mar 2023 14:43:35 +0000

drm/msm/gpu: Move BO allocation out of hw_initThese allocations are only done the first (successful) time throughhw_init() so they won't actually happen in the job_run() path. Butlockdep doesn't know this. So dis-entangle them from the hw_init()path.Signed-off-by: Rob Clark Patchwork: https://patchwork.freedesktop.org/patch/527852/Link: https://lore.kernel.org/r/20230320144356.803762-14-robdclark@gmail.com List of files: /linux-6.15/drivers/gpu/drm/msm/msm_gpu.h

e4f020c6 - drm/msm: Switch idr_lock to spinlock

Mon, 20 Mar 2023 14:43:32 +0000

drm/msm: Switch idr_lock to spinlockNeeded to idr_preload() which returns with preemption disabled.Signed-off-by: Rob Clark Patchwork: https://patchwork.freedesktop.org/patch/527846/Link: https://lore.kernel.org/r/20230320144356.803762-11-robdclark@gmail.com List of files: /linux-6.15/drivers/gpu/drm/msm/msm_gpu.h

d4843012 - drm/msm/a6xx: Remove cx gdsc polling using 'reset'

Mon, 02 Jan 2023 10:48:30 +0000

drm/msm/a6xx: Remove cx gdsc polling using 'reset'Remove the unused 'reset' interface which was supposed to help to ensurethat cx gdsc has collapsed during gpu recovery. This is was not enabledso far due to missing gpucc driver support. Similar functionality usinggenpd framework will be implemented in the upcoming patch.This effectively reverts commit 1f6cca404918("drm/msm/a6xx: Ensure CX collapse during gpu recovery").Signed-off-by: Akhil P Oommen Reviewed-by: Ulf Hansson Reviewed-by: Philipp Zabel Patchwork: https://patchwork.freedesktop.org/patch/516470/Link: https://lore.kernel.org/r/20230102161757.v5.4.I96e0bf9eaf96dd866111c1eec8a4c9b70fd7cbcb@changeidSigned-off-by: Rob Clark List of files: /linux-6.15/drivers/gpu/drm/msm/msm_gpu.h

fadcc3ab - drm/msm/gpu: Bypass PM QoS constraint for idle clamp

Tue, 10 Jan 2023 23:14:43 +0000

drm/msm/gpu: Bypass PM QoS constraint for idle clampChange idle freq clamping back to the direct method, bypassing PM QoSrequests. The problem with using PM QoS requests is they call(indirectly) the governors ->get_target_freq() which goes thru aget_dev_status() cycle. The problem comes when the GPU becomes activeagain and we remove the idle-clamp request, we go through anotherget_dev_status() cycle for the period that the GPU has been idle, whichtriggers the governor to lower the target freq excessively.This partially reverts commit 7c0ffcd40b16 ("drm/msm/gpu: Respect PM QoSconstraints"), but preserves the use of boost QoS request, so that itwill continue to play nicely with other QoS requests such as a coolingdevice. This also mostly undoes commit 78f815c1cf8f ("drm/msm: return theaverage load over the polling period")Signed-off-by: Rob Clark Patchwork: https://patchwork.freedesktop.org/patch/517785/Link: https://lore.kernel.org/r/20230110231447.1939101-3-robdclark@gmail.comReviewed-by: Chia-I Wu List of files: /linux-6.15/drivers/gpu/drm/msm/msm_gpu.h

6563f60f - drm/msm/gpu: Add devfreq tuning debugfs

Tue, 10 Jan 2023 23:14:42 +0000

drm/msm/gpu: Add devfreq tuning debugfsMake the handful of tuning knobs available visible via debugfs.v2: select DEVFREQ_GOV_SIMPLE_ONDEMAND because for some reason struct devfreq_simple_ondemand_data depends on thisSigned-off-by: Rob Clark Patchwork: https://patchwork.freedesktop.org/patch/517784/Link: https://lore.kernel.org/r/20230110231447.1939101-2-robdclark@gmail.comReviewed-by: Chia-I Wu List of files: /linux-6.15/drivers/gpu/drm/msm/msm_gpu.h

a66f1efc - drm/msm/gpu: Fix potential double-free

Tue, 10 Jan 2023 21:28:59 +0000

drm/msm/gpu: Fix potential double-freeIf userspace was calling the MSM_SET_PARAM ioctl on multiple threads toset the COMM or CMDLINE param, it could trigger a race causing theprevious value to be kfree'd multiple times. Fix this by serializing onthe gpu lock.Signed-off-by: Rob Clark Fixes: d4726d770068 ("drm/msm: Add a way to override processes comm/cmdline")Patchwork: https://patchwork.freedesktop.org/patch/517778/Link: https://lore.kernel.org/r/20230110212903.1925878-1-robdclark@gmail.com List of files: /linux-6.15/drivers/gpu/drm/msm/msm_gpu.h

d73b1d02 - drm/msm: Hangcheck progress detection

Mon, 14 Nov 2022 19:30:41 +0000

drm/msm: Hangcheck progress detectionIf the hangcheck timer expires, check if the fw's position in thecmdstream has advanced (changed) since last timer expiration, andallow it up to three additional "extensions" to it's alotted time.The intention is to continue to catch "shader stuck in a loop" typehangs quickly, but allow more time for things that are actuallymaking forward progress.Because we need to sample the CP state twice to detect if there hasnot been progress, this also cuts the the timer's duration in half.v2: Fix typo (REG_A6XX_CP_CSQ_IB2_STAT), add commentv3: Only halve hangcheck timer duration for generations which support progress detection (hdanton); removed unused a5xx progress (without knowing how to adjust for data buffered in ROQ it is too likely to report a false negative)v4: Comment updates to better describe the total hangcheck duration when progress detection is appliedReviewed-by: Chia-I Wu Tested-by: Chia-I Wu # dEQP-GLES2.functional.flush_finish.waitSigned-off-by: Rob Clark Reviewed-by: Akhil P Oommen Patchwork: https://patchwork.freedesktop.org/patch/511584/Link: https://lore.kernel.org/r/20221114193049.1533391-3-robdclark@gmail.com List of files: /linux-6.15/drivers/gpu/drm/msm/msm_gpu.h

cade05b2 - drm/msm/adreno: Simplify read64/write64 helpers

Mon, 14 Nov 2022 19:30:40 +0000

drm/msm/adreno: Simplify read64/write64 helpersThe _HI reg is always following the _LO reg, so no need to pass theseoffsets seprately.Signed-off-by: Rob Clark Reviewed-by: Dmitry Baryshkov Reviewed-by: Akhil P Oommen Patchwork: https://patchwork.freedesktop.org/patch/511581/Link: https://lore.kernel.org/r/20221114193049.1533391-2-robdclark@gmail.com List of files: /linux-6.15/drivers/gpu/drm/msm/msm_gpu.h

76efc245 - drm/msm/gpu: Fix crash during system suspend after unbind

Wed, 28 Sep 2022 07:19:00 +0000

drm/msm/gpu: Fix crash during system suspend after unbindIn adreno_unbind, we should clean up gpu device's drvdata to avoidaccessing a stale pointer during system suspend. Also, check for NULLptr in both system suspend/resume callbacks.Signed-off-by: Akhil P Oommen Patchwork: https://patchwork.freedesktop.org/patch/505075/Link: https://lore.kernel.org/r/20220928124830.2.I5ee0ac073ccdeb81961e5ec0cce5f741a7207a71@changeidSigned-off-by: Rob Clark List of files: /linux-6.15/drivers/gpu/drm/msm/msm_gpu.h

1f6cca40 - drm/msm/a6xx: Ensure CX collapse during gpu recovery

Thu, 18 Aug 2022 20:22:13 +0000

drm/msm/a6xx: Ensure CX collapse during gpu recoveryBecause there could be transient votes from other drivers/tz/hyp whichmay keep the cx gdsc enabled, we should poll until cx gdsc collapses.We can use the reset framework to poll for cx gdsc collapse from gpuccclk driver.This feature requires support from the platform's gpucc driver.Signed-off-by: Akhil P Oommen Reviewed-by: Dmitry Baryshkov Reviewed-by: Philipp Zabel Patchwork: https://patchwork.freedesktop.org/patch/498397/Link: https://lore.kernel.org/r/20220819015030.v5.5.I176567525af2b9439a7e485d0ca130528666a55c@changeidSigned-off-by: Rob Clark List of files: /linux-6.15/drivers/gpu/drm/msm/msm_gpu.h

b352ba54 - drm/msm/gem: Convert to using drm_gem_lru

Tue, 02 Aug 2022 15:51:43 +0000

drm/msm/gem: Convert to using drm_gem_lruThis converts over to use the shared GEM LRU/shrinker helpers. Notethat it means we are no longer tracking purgeable or willneed buffersthat are active separately. But the most recently pinned buffers shouldbe at the tail of the various LRUs, and the shrinker is already preparedto encounter objects which are still active.Signed-off-by: Rob Clark Patchwork: https://patchwork.freedesktop.org/patch/496131/Link: https://lore.kernel.org/r/20220802155152.1727594-11-robdclark@gmail.com List of files: /linux-6.15/drivers/gpu/drm/msm/msm_gpu.h

05ba44b3 - drm/msm: Split out idr_lock

Tue, 02 Aug 2022 15:51:36 +0000

drm/msm: Split out idr_lockOtherwise if we hit reclaim pinning objects in the submit path, we'll beblocking retire_worker trying to free a submit.Signed-off-by: Rob Clark Patchwork: https://patchwork.freedesktop.org/patch/496116/Link: https://lore.kernel.org/r/20220802155152.1727594-4-robdclark@gmail.com List of files: /linux-6.15/drivers/gpu/drm/msm/msm_gpu.h