From 0abdba47dc1df708c365421d481734d3f7fecb01 Mon Sep 17 00:00:00 2001 From: Lucas Stach Date: Thu, 28 Feb 2019 07:23:29 +0100 Subject: [PATCH 01/34] drm/msm: don't allocate pages from the MOVABLE zone The pages backing the GEM objects are kept pinned in place as long as they are alive, so they must not be allocated from the MOVABLE zone. Blocking page migration for too long will cause the VM subsystem headaches and will outright break CMA, as a few pinned pages in CMA will lead to failure to find the required large contiguous regions. Signed-off-by: Lucas Stach Signed-off-by: Rob Clark Signed-off-by: Rob Clark --- drivers/gpu/drm/msm/msm_gem.c | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/drivers/gpu/drm/msm/msm_gem.c b/drivers/gpu/drm/msm/msm_gem.c index 18ca651ab942..76940a9da980 100644 --- a/drivers/gpu/drm/msm/msm_gem.c +++ b/drivers/gpu/drm/msm/msm_gem.c @@ -1026,6 +1026,13 @@ static struct drm_gem_object *_msm_gem_new(struct drm_device *dev, ret = drm_gem_object_init(dev, obj, size); if (ret) goto fail; + /* + * Our buffers are kept pinned, so allocating them from the + * MOVABLE zone is a really bad idea, and conflicts with CMA. + * See comments above new_inode() why this is required _and_ + * expected if you're going to pin these pages. + */ + mapping_set_gfp_mask(obj->filp->f_mapping, GFP_HIGHUSER); } return obj; From a6bb79ab48796a6293b9c60597b631bca6011041 Mon Sep 17 00:00:00 2001 From: Jordan Crouse Date: Tue, 19 Feb 2019 11:48:20 -0700 Subject: [PATCH 02/34] drm/msm: Remove pm_runtime calls from msm_iommu.c Currently the IOMMU code calls pm_runtime_get/put on the GPU or display device before doing a IOMMU operation. This was because usually the IOMMU driver didn't do power control of its own and since the hardware used the same clocks and power as the respective multimedia device it was a easy way to make sure that the power was available. Now two things have changed. First, the SMMU devices can do their own power control and more important bringing up the a6xx GPU isn't as easy as turning on some clocks. To bring the GPU up we need the GMU which itself needs the IOMMU so we have a chicken and egg problem. Luckily this is easily fixed by removing the pm_runtime calls from the functions and letting the device link to the IOMMU device handle the magic. Signed-off-by: Jordan Crouse Tested-by: Bjorn Andersson Signed-off-by: Rob Clark Signed-off-by: Rob Clark --- drivers/gpu/drm/msm/msm_iommu.c | 13 +------------ 1 file changed, 1 insertion(+), 12 deletions(-) diff --git a/drivers/gpu/drm/msm/msm_iommu.c b/drivers/gpu/drm/msm/msm_iommu.c index 4d62790cd425..12bb54cefd46 100644 --- a/drivers/gpu/drm/msm/msm_iommu.c +++ b/drivers/gpu/drm/msm/msm_iommu.c @@ -38,13 +38,8 @@ static int msm_iommu_attach(struct msm_mmu *mmu, const char * const *names, int cnt) { struct msm_iommu *iommu = to_msm_iommu(mmu); - int ret; - pm_runtime_get_sync(mmu->dev); - ret = iommu_attach_device(iommu->domain, mmu->dev); - pm_runtime_put_sync(mmu->dev); - - return ret; + return iommu_attach_device(iommu->domain, mmu->dev); } static void msm_iommu_detach(struct msm_mmu *mmu, const char * const *names, @@ -52,9 +47,7 @@ static void msm_iommu_detach(struct msm_mmu *mmu, const char * const *names, { struct msm_iommu *iommu = to_msm_iommu(mmu); - pm_runtime_get_sync(mmu->dev); iommu_detach_device(iommu->domain, mmu->dev); - pm_runtime_put_sync(mmu->dev); } static int msm_iommu_map(struct msm_mmu *mmu, uint64_t iova, @@ -63,9 +56,7 @@ static int msm_iommu_map(struct msm_mmu *mmu, uint64_t iova, struct msm_iommu *iommu = to_msm_iommu(mmu); size_t ret; -// pm_runtime_get_sync(mmu->dev); ret = iommu_map_sg(iommu->domain, iova, sgt->sgl, sgt->nents, prot); -// pm_runtime_put_sync(mmu->dev); WARN_ON(!ret); return (ret == len) ? 0 : -EINVAL; @@ -75,9 +66,7 @@ static int msm_iommu_unmap(struct msm_mmu *mmu, uint64_t iova, unsigned len) { struct msm_iommu *iommu = to_msm_iommu(mmu); - pm_runtime_get_sync(mmu->dev); iommu_unmap(iommu->domain, iova, len); - pm_runtime_put_sync(mmu->dev); return 0; } From 7603df38cc8c1e5d540b18ec9eb9d62d823197d0 Mon Sep 17 00:00:00 2001 From: Luca Weiss Date: Sat, 2 Mar 2019 13:35:29 +0100 Subject: [PATCH 03/34] drm/msm: Fix NULL pointer dereference [ 3.707412] Unable to handle kernel NULL pointer dereference at virtual address 0000009c [ 3.714511] pgd = (ptrval) [ 3.722742] [0000009c] *pgd=00000000 [ 3.725238] Internal error: Oops: 5 [#1] PREEMPT SMP ARM [ 3.728968] Modules linked in: [ 3.734265] CPU: 3 PID: 112 Comm: kworker/3:2 Tainted: G W 5.0.0-rc7-00183-g06a1c31df9eb #4 [ 3.737142] Hardware name: Generic DT based system [ 3.746778] Workqueue: events deferred_probe_work_func [ 3.751542] PC is at msm_gem_map_vma+0x3c/0xac [ 3.756669] LR is at msm_gem_get_and_pin_iova+0xd8/0x134 [ 3.761086] pc : [] lr : [] psr: 60000013 [ 3.766560] sp : ee297be8 ip : ed9ab1c0 fp : ed93b800 [ 3.772546] r10: ee35e180 r9 : 00000000 r8 : ee297c80 [ 3.777752] r7 : 00000000 r6 : 7c100000 r5 : 00000000 r4 : ee35e180 [ 3.782968] r3 : 00000001 r2 : 00000003 r1 : ee35e180 r0 : 00000000 [ 3.789562] Flags: nZCv IRQs on FIQs on Mode SVC_32 ISA ARM Segment none [ 3.796079] Control: 10c5787d Table: 2e3a806a DAC: 00000051 [ 3.803282] Process kworker/3:2 (pid: 112, stack limit = 0x(ptrval)) [ 3.809006] Stack: (0xee297be8 to 0xee298000) [ 3.815445] 7be0: 00000000 c1108c48 eda8c000 00000003 eda8c0fc c1108c48 [ 3.819715] 7c00: eda8c000 00000003 eda8c0fc c07d14f8 00000001 c07d1100 7c100000 00000000 [ 3.827873] 7c20: eda8c000 bb7ffb78 00000000 eda8c000 00000000 00000000 c0c8b1d4 ee3bfa00 [ 3.836037] 7c40: ee3b9800 c07d1684 00000000 c1108c48 ee0d7810 ee3b9800 c0c8b1d4 c07d222c [ 3.844193] 7c60: ee3bfd84 ee297c80 00000000 c0b1d5b0 ee3bfc40 c07dcfd8 ee3bfd84 ee297c80 [ 3.852357] 7c80: 0000006d ee3bfc40 ee0d7810 bb7ffb78 c0c8b1d4 00000000 ee3bfc40 c07ddb48 [ 3.860516] 7ca0: 00002004 c0eba384 ee3bfc40 c079eba0 ee3bd040 ee3b9800 00000001 ed93b800 [ 3.868673] 7cc0: ed9aa100 c07db7e8 ee3bf240 ed9a6500 00000001 ee3b9800 ee3bf2d4 c07a0a30 [ 3.876834] 7ce0: ed93b800 7d100000 c1108c48 ee0d7610 ee3b9800 ed93b800 c1108c48 00000000 [ 3.884991] 7d00: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 [ 3.893151] 7d20: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 bb7ffb78 [ 3.901310] 7d40: c12113c4 ed93b800 ee3b9800 c1108c48 ee9eec10 00000000 ed93b800 7d100000 [ 3.909472] 7d60: eff7b000 c07cf748 7d100000 00000000 c0e9a350 c0b1d5b0 c12113c4 c0961e40 [ 3.917633] 7d80: c12113c4 40000113 eeff4bec c0ebe004 00000019 c0b1d230 ee9eeda8 60000113 [ 3.925791] 7da0: ee35d300 ee9eeda8 c07ce260 bb7ffb78 c07ce260 ee35d2c0 00000028 00000002 [ 3.933950] 7dc0: eeb76280 c118f884 ee0be640 c11c6128 c07ce260 c07ea4ac 00000000 c0962b48 [ 3.942108] 7de0: c118f868 00000001 c0ebbc98 ee35d2c0 00000000 eeb76280 00000000 c118f87c [ 3.950270] 7e00: ee35d2c0 00000000 c11c63e0 c118f694 00000019 c07ea5d0 ee0d7810 00000000 [ 3.958430] 7e20: c118f694 00000000 00000000 c07f2b0c c120f55c ee0d7810 c120f560 00000000 [ 3.966590] 7e40: 00000000 c07f08c4 c07f0e8c ee0d7810 c11ba3d0 ee0d7810 c118f694 c07f0e8c [ 3.974748] 7e60: c1108c48 00000001 c0ebc3cc c11c63f8 c11ba3d0 c07f0c08 00000001 c07f2f8c [ 3.982908] 7e80: c118f694 00000000 ee297ed4 c07f0e8c c1108c48 00000001 c0ebc3cc c11c63f8 [ 3.991068] 7ea0: c11ba3d0 c07ee8a0 c11ba3d0 ee82686c ee0baf38 bb7ffb78 ee0d7810 ee0d7810 [ 3.999227] 7ec0: c1108c48 ee0d7844 c118faac c07f05b0 ee0d7810 ee0d7810 00000001 bb7ffb78 [ 4.007389] 7ee0: ee0d7810 ee0d7810 c118fd18 c118faac c11c63e0 c07ef7d0 ee0d7810 c118fa90 [ 4.015548] 7f00: c118fa90 c07efd68 c118fac8 ee27fe00 eefd9c80 eefdcd00 00000000 c118facc [ 4.023708] 7f20: 00000000 c033c038 eefd9c80 eefd9c80 00000008 ee27fe00 ee27fe14 eefd9c80 [ 4.031866] 7f40: 00000008 c1103d00 eefd9c98 ee296000 eefd9c80 c033ce54 ee907eac c0b1d230 [ 4.040026] 7f60: ee907eac eea24440 ee285000 00000000 ee296000 ee27fe00 c033ce24 eea2445c [ 4.048188] 7f80: ee907eac c0341db0 00000000 ee285000 c0341c8c 00000000 00000000 00000000 [ 4.056346] 7fa0: 00000000 00000000 00000000 c03010e8 00000000 00000000 00000000 00000000 [ 4.064505] 7fc0: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 [ 4.072665] 7fe0: 00000000 00000000 00000000 00000000 00000013 00000000 00000000 00000000 [ 4.080828] [] (msm_gem_map_vma) from [] (msm_gem_get_and_pin_iova+0xd8/0x134) [ 4.088983] [] (msm_gem_get_and_pin_iova) from [] (_msm_gem_kernel_new+0x38/0xac) [ 4.097839] [] (_msm_gem_kernel_new) from [] (msm_gem_kernel_new+0x24/0x2c) [ 4.107130] [] (msm_gem_kernel_new) from [] (dsi_tx_buf_alloc_6g+0x44/0x90) [ 4.115631] [] (dsi_tx_buf_alloc_6g) from [] (msm_dsi_host_modeset_init+0x80/0x104) [ 4.124313] [] (msm_dsi_host_modeset_init) from [] (msm_dsi_modeset_init+0x34/0x1c0) [ 4.133691] [] (msm_dsi_modeset_init) from [] (mdp5_kms_init+0x764/0x7e0) [ 4.143409] [] (mdp5_kms_init) from [] (msm_drm_bind+0x56c/0x740) [ 4.151824] [] (msm_drm_bind) from [] (try_to_bring_up_master+0x238/0x2b4) [ 4.159636] [] (try_to_bring_up_master) from [] (component_add+0xa8/0x170) [ 4.168146] [] (component_add) from [] (platform_drv_probe+0x48/0x9c) [ 4.176737] [] (platform_drv_probe) from [] (really_probe+0x278/0x404) [ 4.184981] [] (really_probe) from [] (driver_probe_device+0x78/0x1c0) [ 4.193147] [] (driver_probe_device) from [] (bus_for_each_drv+0x74/0xb8) [ 4.201389] [] (bus_for_each_drv) from [] (__device_attach+0xd0/0x164) [ 4.209984] [] (__device_attach) from [] (bus_probe_device+0x84/0x8c) [ 4.218143] [] (bus_probe_device) from [] (deferred_probe_work_func+0x48/0xc4) [ 4.226398] [] (deferred_probe_work_func) from [] (process_one_work+0x204/0x574) [ 4.235254] [] (process_one_work) from [] (worker_thread+0x30/0x560) [ 4.244534] [] (worker_thread) from [] (kthread+0x124/0x154) [ 4.252606] [] (kthread) from [] (ret_from_fork+0x14/0x2c) [ 4.259966] Exception stack(0xee297fb0 to 0xee297ff8) [ 4.266998] 7fa0: 00000000 00000000 00000000 00000000 [ 4.272143] 7fc0: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 [ 4.280297] 7fe0: 00000000 00000000 00000000 00000000 00000013 00000000 [ 4.288451] Code: e5813080 1a000013 e3a03001 e5c4307c (e590009c) [ 4.294933] ---[ end trace 18729cc2bca2b4b3 ]--- Signed-off-by: Luca Weiss Signed-off-by: Rob Clark Signed-off-by: Rob Clark --- drivers/gpu/drm/msm/msm_gem_vma.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/msm/msm_gem_vma.c b/drivers/gpu/drm/msm/msm_gem_vma.c index 49c04829cf34..fcf7a83f0e6f 100644 --- a/drivers/gpu/drm/msm/msm_gem_vma.c +++ b/drivers/gpu/drm/msm/msm_gem_vma.c @@ -85,7 +85,7 @@ msm_gem_map_vma(struct msm_gem_address_space *aspace, vma->mapped = true; - if (aspace->mmu) + if (aspace && aspace->mmu) ret = aspace->mmu->funcs->map(aspace->mmu, vma->iova, sgt, size, prot); From 6117f86202720cae2397dfcba7acf374cde3c6c9 Mon Sep 17 00:00:00 2001 From: Sean Paul Date: Mon, 28 Jan 2019 15:42:48 -0500 Subject: [PATCH 04/34] drm/msm: Use drm_mode_vrefresh instead of mode->vrefresh Use the drm_mode_vrefresh helper where we need refresh rate in case vrefresh is empty. Reviewed-by: Abhinav Kumar Signed-off-by: Sean Paul Link: https://patchwork.freedesktop.org/patch/msgid/20190128204306.95076-1-sean@poorly.run Signed-off-by: Rob Clark --- drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c | 6 +++--- drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys_cmd.c | 5 +++-- drivers/gpu/drm/msm/disp/dpu1/dpu_plane.c | 2 +- drivers/gpu/drm/msm/disp/mdp5/mdp5_cmd_encoder.c | 4 ++-- 4 files changed, 9 insertions(+), 8 deletions(-) diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c index 5aa3307f3f0c..3a714695d536 100644 --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c @@ -520,8 +520,8 @@ static void _dpu_encoder_adjust_mode(struct drm_connector *connector, list_for_each_entry(cur_mode, &connector->modes, head) { if (cur_mode->vdisplay == adj_mode->vdisplay && - cur_mode->hdisplay == adj_mode->hdisplay && - cur_mode->vrefresh == adj_mode->vrefresh) { + cur_mode->hdisplay == adj_mode->hdisplay && + drm_mode_vrefresh(cur_mode) == drm_mode_vrefresh(adj_mode)) { adj_mode->private = cur_mode->private; adj_mode->private_flags |= cur_mode->private_flags; } @@ -1802,7 +1802,7 @@ void dpu_encoder_kickoff(struct drm_encoder *drm_enc, bool async) atomic_set(&dpu_enc->frame_done_timeout, DPU_FRAME_DONE_TIMEOUT * 1000 / - drm_enc->crtc->state->adjusted_mode.vrefresh); + drm_mode_vrefresh(&drm_enc->crtc->state->adjusted_mode)); mod_timer(&dpu_enc->frame_done_timer, jiffies + ((atomic_read(&dpu_enc->frame_done_timeout) * HZ) / 1000)); diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys_cmd.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys_cmd.c index a399e1edd313..973737fb5c9f 100644 --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys_cmd.c +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys_cmd.c @@ -404,7 +404,8 @@ static void dpu_encoder_phys_cmd_tearcheck_config( return; } - tc_cfg.vsync_count = vsync_hz / (mode->vtotal * mode->vrefresh); + tc_cfg.vsync_count = vsync_hz / + (mode->vtotal * drm_mode_vrefresh(mode)); /* enable external TE after kickoff to avoid premature autorefresh */ tc_cfg.hw_vsync_mode = 0; @@ -424,7 +425,7 @@ static void dpu_encoder_phys_cmd_tearcheck_config( DPU_DEBUG_CMDENC(cmd_enc, "tc %d vsync_clk_speed_hz %u vtotal %u vrefresh %u\n", phys_enc->hw_pp->idx - PINGPONG_0, vsync_hz, - mode->vtotal, mode->vrefresh); + mode->vtotal, drm_mode_vrefresh(mode)); DPU_DEBUG_CMDENC(cmd_enc, "tc %d enable %u start_pos %u rd_ptr_irq %u\n", phys_enc->hw_pp->idx - PINGPONG_0, tc_enable, tc_cfg.start_pos, diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_plane.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_plane.c index b01183b309b9..da1f727d7495 100644 --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_plane.c +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_plane.c @@ -387,7 +387,7 @@ static void _dpu_plane_set_ot_limit(struct drm_plane *plane, ot_params.width = drm_rect_width(&pdpu->pipe_cfg.src_rect); ot_params.height = drm_rect_height(&pdpu->pipe_cfg.src_rect); ot_params.is_wfd = !pdpu->is_rt_pipe; - ot_params.frame_rate = crtc->mode.vrefresh; + ot_params.frame_rate = drm_mode_vrefresh(&crtc->mode); ot_params.vbif_idx = VBIF_RT; ot_params.clk_ctrl = pdpu->pipe_hw->cap->clk_ctrl; ot_params.rd = true; diff --git a/drivers/gpu/drm/msm/disp/mdp5/mdp5_cmd_encoder.c b/drivers/gpu/drm/msm/disp/mdp5/mdp5_cmd_encoder.c index 9bf9d6065c55..7b9edc21bc2c 100644 --- a/drivers/gpu/drm/msm/disp/mdp5/mdp5_cmd_encoder.c +++ b/drivers/gpu/drm/msm/disp/mdp5/mdp5_cmd_encoder.c @@ -59,10 +59,10 @@ static int pingpong_tearcheck_setup(struct drm_encoder *encoder, return -EINVAL; } - total_lines_x100 = mode->vtotal * mode->vrefresh; + total_lines_x100 = mode->vtotal * drm_mode_vrefresh(mode); if (!total_lines_x100) { DRM_DEV_ERROR(dev, "%s: vtotal(%d) or vrefresh(%d) is 0\n", - __func__, mode->vtotal, mode->vrefresh); + __func__, mode->vtotal, drm_mode_vrefresh(mode)); return -EINVAL; } From 2e0391865bd71e076139ca8059bf68469012c27a Mon Sep 17 00:00:00 2001 From: Sean Paul Date: Mon, 28 Jan 2019 15:42:49 -0500 Subject: [PATCH 05/34] drm/msm: dpu: Simplify frame_done watchdog timeout calculation Instead of setting the timeout and then immediately reading it back (along with the hand-rolled msecs_to_jiffies calculation), just calculate it once and set it in both places at the same time. Reviewed-by: Abhinav Kumar Signed-off-by: Sean Paul Link: https://patchwork.freedesktop.org/patch/msgid/20190128204306.95076-2-sean@poorly.run Signed-off-by: Rob Clark --- drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c | 12 +++++++----- 1 file changed, 7 insertions(+), 5 deletions(-) diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c index 3a714695d536..107abd4b99d2 100644 --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c @@ -1788,6 +1788,7 @@ void dpu_encoder_kickoff(struct drm_encoder *drm_enc, bool async) { struct dpu_encoder_virt *dpu_enc; struct dpu_encoder_phys *phys; + unsigned long timeout_ms; ktime_t wakeup_time; unsigned int i; @@ -1800,11 +1801,12 @@ void dpu_encoder_kickoff(struct drm_encoder *drm_enc, bool async) trace_dpu_enc_kickoff(DRMID(drm_enc)); - atomic_set(&dpu_enc->frame_done_timeout, - DPU_FRAME_DONE_TIMEOUT * 1000 / - drm_mode_vrefresh(&drm_enc->crtc->state->adjusted_mode)); - mod_timer(&dpu_enc->frame_done_timer, jiffies + - ((atomic_read(&dpu_enc->frame_done_timeout) * HZ) / 1000)); + timeout_ms = DPU_FRAME_DONE_TIMEOUT * 1000 / + drm_mode_vrefresh(&drm_enc->crtc->state->adjusted_mode); + + atomic_set(&dpu_enc->frame_done_timeout, timeout_ms); + mod_timer(&dpu_enc->frame_done_timer, + jiffies + msecs_to_jiffies(timeout_ms)); /* All phys encs are ready to go, trigger the kickoff */ _dpu_encoder_kickoff_phys(dpu_enc, async); From 70df9610de0fd3bbcb746f501ab7426a3052250f Mon Sep 17 00:00:00 2001 From: Sean Paul Date: Mon, 28 Jan 2019 15:42:50 -0500 Subject: [PATCH 06/34] drm/msm: dpu: Untangle frame_done timeout units There exists a bunch of confusion as to what the actual units of frame_done is: - The definition states it's in # of frames - CRTC treats it like it's ms - frame_done_timeout comment thinks it's Hz, but it stores ms - frame_done timer is setup such that it _should_ be in frames, but the timeout is super long So this patch tries to interpret what the driver really wants. I've de-centralized the #define since the consumers are expecting different units. For crtc, we just use 60ms since that's what it was doing before. Perhaps we could get fancy and scale with vrefresh, but that's for another time. For encoder, fix the comments and rename frame_done_timeout so it's obvious what the units are. In practice, frame_done_timeout is really just checked against 0 || !0, which I guess is why the units being wrong didn't matter. I've also dropped the timeout from the previous 60 frames to 5. That seems like more than enough time to give up on a frame, and my guess is that no one intended for the timeout to _actually_ be 60 frames. Reviewed-by: Fritz Koenig Signed-off-by: Sean Paul Link: https://patchwork.freedesktop.org/patch/msgid/20190128204306.95076-3-sean@poorly.run Signed-off-by: Rob Clark --- drivers/gpu/drm/msm/disp/dpu1/dpu_crtc.c | 5 ++++- drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c | 19 +++++++++++-------- drivers/gpu/drm/msm/disp/dpu1/dpu_kms.h | 3 --- 3 files changed, 15 insertions(+), 12 deletions(-) diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_crtc.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_crtc.c index b776fca571f3..366e88f6c5d1 100644 --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_crtc.c +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_crtc.c @@ -46,6 +46,9 @@ #define LEFT_MIXER 0 #define RIGHT_MIXER 1 +/* timeout in ms waiting for frame done */ +#define DPU_CRTC_FRAME_DONE_TIMEOUT_MS 60 + static struct dpu_kms *_dpu_crtc_get_kms(struct drm_crtc *crtc) { struct msm_drm_private *priv = crtc->dev->dev_private; @@ -683,7 +686,7 @@ static int _dpu_crtc_wait_for_frame_done(struct drm_crtc *crtc) DPU_ATRACE_BEGIN("frame done completion wait"); ret = wait_for_completion_timeout(&dpu_crtc->frame_done_comp, - msecs_to_jiffies(DPU_FRAME_DONE_TIMEOUT)); + msecs_to_jiffies(DPU_CRTC_FRAME_DONE_TIMEOUT_MS)); if (!ret) { DRM_ERROR("frame done wait timed out, ret:%d\n", ret); rc = -ETIMEDOUT; diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c index 107abd4b99d2..e8dfdb48016e 100644 --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c @@ -69,6 +69,9 @@ #define MAX_VDISPLAY_SPLIT 1080 +/* timeout in frames waiting for frame done */ +#define DPU_ENCODER_FRAME_DONE_TIMEOUT_FRAMES 5 + /** * enum dpu_enc_rc_events - events for resource control state machine * @DPU_ENC_RC_EVENT_KICKOFF: @@ -158,7 +161,7 @@ enum dpu_enc_rc_states { * Bit0 = phys_encs[0] etc. * @crtc_frame_event_cb: callback handler for frame event * @crtc_frame_event_cb_data: callback handler private data - * @frame_done_timeout: frame done timeout in Hz + * @frame_done_timeout_ms: frame done timeout in ms * @frame_done_timer: watchdog timer for frame done event * @vsync_event_timer: vsync timer * @disp_info: local copy of msm_display_info struct @@ -196,7 +199,7 @@ struct dpu_encoder_virt { void (*crtc_frame_event_cb)(void *, u32 event); void *crtc_frame_event_cb_data; - atomic_t frame_done_timeout; + atomic_t frame_done_timeout_ms; struct timer_list frame_done_timer; struct timer_list vsync_event_timer; @@ -1182,7 +1185,7 @@ static void dpu_encoder_virt_disable(struct drm_encoder *drm_enc) } /* after phys waits for frame-done, should be no more frames pending */ - if (atomic_xchg(&dpu_enc->frame_done_timeout, 0)) { + if (atomic_xchg(&dpu_enc->frame_done_timeout_ms, 0)) { DPU_ERROR("enc%d timeout pending\n", drm_enc->base.id); del_timer_sync(&dpu_enc->frame_done_timer); } @@ -1339,7 +1342,7 @@ static void dpu_encoder_frame_done_callback( } if (!dpu_enc->frame_busy_mask[0]) { - atomic_set(&dpu_enc->frame_done_timeout, 0); + atomic_set(&dpu_enc->frame_done_timeout_ms, 0); del_timer(&dpu_enc->frame_done_timer); dpu_encoder_resource_control(drm_enc, @@ -1801,10 +1804,10 @@ void dpu_encoder_kickoff(struct drm_encoder *drm_enc, bool async) trace_dpu_enc_kickoff(DRMID(drm_enc)); - timeout_ms = DPU_FRAME_DONE_TIMEOUT * 1000 / + timeout_ms = DPU_ENCODER_FRAME_DONE_TIMEOUT_FRAMES * 1000 / drm_mode_vrefresh(&drm_enc->crtc->state->adjusted_mode); - atomic_set(&dpu_enc->frame_done_timeout, timeout_ms); + atomic_set(&dpu_enc->frame_done_timeout_ms, timeout_ms); mod_timer(&dpu_enc->frame_done_timer, jiffies + msecs_to_jiffies(timeout_ms)); @@ -2126,7 +2129,7 @@ static void dpu_encoder_frame_done_timeout(struct timer_list *t) DRM_DEBUG_KMS("id:%u invalid timeout frame_busy_mask=%lu\n", DRMID(drm_enc), dpu_enc->frame_busy_mask[0]); return; - } else if (!atomic_xchg(&dpu_enc->frame_done_timeout, 0)) { + } else if (!atomic_xchg(&dpu_enc->frame_done_timeout_ms, 0)) { DRM_DEBUG_KMS("id:%u invalid timeout\n", DRMID(drm_enc)); return; } @@ -2172,7 +2175,7 @@ int dpu_encoder_setup(struct drm_device *dev, struct drm_encoder *enc, spin_lock_init(&dpu_enc->enc_spinlock); - atomic_set(&dpu_enc->frame_done_timeout, 0); + atomic_set(&dpu_enc->frame_done_timeout_ms, 0); timer_setup(&dpu_enc->frame_done_timer, dpu_encoder_frame_done_timeout, 0); diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.h b/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.h index ac75cfc267f4..31e9ef96ca5d 100644 --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.h +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.h @@ -73,9 +73,6 @@ #define DPU_NAME_SIZE 12 -/* timeout in frames waiting for frame done */ -#define DPU_FRAME_DONE_TIMEOUT 60 - /* * struct dpu_irq_callback - IRQ callback handlers * @list: list to callback From be77ef760cd351b27916a3b4e74e7fefb4809c4a Mon Sep 17 00:00:00 2001 From: Sean Paul Date: Mon, 28 Jan 2019 15:42:51 -0500 Subject: [PATCH 07/34] drm/msm: dpu: Don't queue the frame_done watchdog for cursor In the case of an async/cursor update, we don't wait for the frame_done event, which means handle_frame_done is never called, and the frame_done watchdog isn't canceled. Currently, this results in a frame_done timeout every time the cursor moves without a synchronous frame following it up before the timeout expires. Since we don't wait for frame_done, and don't handle it, we shouldn't modify the watchdog. Reviewed-by: Fritz Koenig Signed-off-by: Sean Paul Link: https://patchwork.freedesktop.org/patch/msgid/20190128204306.95076-4-sean@poorly.run Signed-off-by: Rob Clark --- drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c | 19 +++++++++++++------ 1 file changed, 13 insertions(+), 6 deletions(-) diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c index e8dfdb48016e..472cdb464458 100644 --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c @@ -1791,7 +1791,6 @@ void dpu_encoder_kickoff(struct drm_encoder *drm_enc, bool async) { struct dpu_encoder_virt *dpu_enc; struct dpu_encoder_phys *phys; - unsigned long timeout_ms; ktime_t wakeup_time; unsigned int i; @@ -1804,12 +1803,20 @@ void dpu_encoder_kickoff(struct drm_encoder *drm_enc, bool async) trace_dpu_enc_kickoff(DRMID(drm_enc)); - timeout_ms = DPU_ENCODER_FRAME_DONE_TIMEOUT_FRAMES * 1000 / - drm_mode_vrefresh(&drm_enc->crtc->state->adjusted_mode); + /* + * Asynchronous frames don't handle FRAME_DONE events. As such, they + * shouldn't enable the frame_done watchdog since it will always time + * out. + */ + if (!async) { + unsigned long timeout_ms; + timeout_ms = DPU_ENCODER_FRAME_DONE_TIMEOUT_FRAMES * 1000 / + drm_mode_vrefresh(&drm_enc->crtc->state->adjusted_mode); - atomic_set(&dpu_enc->frame_done_timeout_ms, timeout_ms); - mod_timer(&dpu_enc->frame_done_timer, - jiffies + msecs_to_jiffies(timeout_ms)); + atomic_set(&dpu_enc->frame_done_timeout_ms, timeout_ms); + mod_timer(&dpu_enc->frame_done_timer, + jiffies + msecs_to_jiffies(timeout_ms)); + } /* All phys encs are ready to go, trigger the kickoff */ _dpu_encoder_kickoff_phys(dpu_enc, async); From f98baa3109cea46083d2361ab14a0207d1b1bd16 Mon Sep 17 00:00:00 2001 From: Sean Paul Date: Wed, 30 Jan 2019 11:32:12 -0500 Subject: [PATCH 08/34] drm/msm: dpu: Don't set frame_busy_mask for async updates The frame_busy mask is used in frame_done event handling, which is not invoked for async commits. So an async commit will leave the frame_busy mask populated after it completes and future commits will start with the busy mask incorrect. This showed up on disable after cursor move. I was hitting the "this should not happen" comment in the frame event worker since frame_busy was set, we queued the event, but there were no frames pending (since async also doesn't set that). Reviewed-by: Fritz Koenig Signed-off-by: Sean Paul Link: https://patchwork.freedesktop.org/patch/msgid/20190130163220.138637-1-sean@poorly.run Signed-off-by: Rob Clark --- drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c index 472cdb464458..c51bc0fdd9c3 100644 --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c @@ -1550,8 +1550,14 @@ static void _dpu_encoder_kickoff_phys(struct dpu_encoder_virt *dpu_enc, if (!ctl) continue; - if (phys->split_role != ENC_ROLE_SLAVE) + /* + * This is cleared in frame_done worker, which isn't invoked + * for async commits. So don't set this for async, since it'll + * roll over to the next commit. + */ + if (!async && phys->split_role != ENC_ROLE_SLAVE) set_bit(i, dpu_enc->frame_busy_mask); + if (!phys->ops.needs_single_flush || !phys->ops.needs_single_flush(phys)) _dpu_encoder_trigger_flush(&dpu_enc->base, phys, 0x0, From b6057cda8f6cac5f50dbac7ff5ca825fdafdeb39 Mon Sep 17 00:00:00 2001 From: Jeykumar Sankaran Date: Wed, 13 Feb 2019 17:19:10 -0800 Subject: [PATCH 09/34] drm/msm/dpu: move hw_inf encoder baseclass Both video and command physical encoders will have a hw interface assigned to it. So there is really no need to track the hw block in specific encoder subclass. Signed-off-by: Jeykumar Sankaran Signed-off-by: Sean Paul Link: https://patchwork.freedesktop.org/patch/msgid/1550107156-17625-2-git-send-email-jsanka@codeaurora.org Signed-off-by: Rob Clark --- .../gpu/drm/msm/disp/dpu1/dpu_encoder_phys.h | 4 +- .../drm/msm/disp/dpu1/dpu_encoder_phys_vid.c | 125 +++++++----------- 2 files changed, 52 insertions(+), 77 deletions(-) diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys.h b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys.h index db94f3d3bea3..66a1fd4b014c 100644 --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys.h +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys.h @@ -200,6 +200,7 @@ struct dpu_encoder_irq { * @hw_mdptop: Hardware interface to the top registers * @hw_ctl: Hardware interface to the ctl registers * @hw_pp: Hardware interface to the ping pong registers + * @hw_intf: Hardware interface to the intf registers * @dpu_kms: Pointer to the dpu_kms top level * @cached_mode: DRM mode cached at mode_set time, acted on in enable * @enabled: Whether the encoder has enabled and running a mode @@ -228,6 +229,7 @@ struct dpu_encoder_phys { struct dpu_hw_mdp *hw_mdptop; struct dpu_hw_ctl *hw_ctl; struct dpu_hw_pingpong *hw_pp; + struct dpu_hw_intf *hw_intf; struct dpu_kms *dpu_kms; struct drm_display_mode cached_mode; enum dpu_enc_split_role split_role; @@ -254,12 +256,10 @@ static inline int dpu_encoder_phys_inc_pending(struct dpu_encoder_phys *phys) * struct dpu_encoder_phys_vid - sub-class of dpu_encoder_phys to handle video * mode specific operations * @base: Baseclass physical encoder structure - * @hw_intf: Hardware interface to the intf registers * @timing_params: Current timing parameter */ struct dpu_encoder_phys_vid { struct dpu_encoder_phys base; - struct dpu_hw_intf *hw_intf; struct intf_timing_params timing_params; }; diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys_vid.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys_vid.c index 3c4eb470a82c..1b706eada90f 100644 --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys_vid.c +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys_vid.c @@ -18,14 +18,14 @@ #include "dpu_trace.h" #define DPU_DEBUG_VIDENC(e, fmt, ...) DPU_DEBUG("enc%d intf%d " fmt, \ - (e) && (e)->base.parent ? \ - (e)->base.parent->base.id : -1, \ + (e) && (e)->parent ? \ + (e)->parent->base.id : -1, \ (e) && (e)->hw_intf ? \ (e)->hw_intf->idx - INTF_0 : -1, ##__VA_ARGS__) #define DPU_ERROR_VIDENC(e, fmt, ...) DPU_ERROR("enc%d intf%d " fmt, \ - (e) && (e)->base.parent ? \ - (e)->base.parent->base.id : -1, \ + (e) && (e)->parent ? \ + (e)->parent->base.id : -1, \ (e) && (e)->hw_intf ? \ (e)->hw_intf->idx - INTF_0 : -1, ##__VA_ARGS__) @@ -44,7 +44,7 @@ static bool dpu_encoder_phys_vid_is_master( } static void drm_mode_to_intf_timing_params( - const struct dpu_encoder_phys_vid *vid_enc, + const struct dpu_encoder_phys *phys_enc, const struct drm_display_mode *mode, struct intf_timing_params *timing) { @@ -92,7 +92,7 @@ static void drm_mode_to_intf_timing_params( timing->hsync_skew = mode->hskew; /* DSI controller cannot handle active-low sync signals. */ - if (vid_enc->hw_intf->cap->type == INTF_DSI) { + if (phys_enc->hw_intf->cap->type == INTF_DSI) { timing->hsync_polarity = 0; timing->vsync_polarity = 0; } @@ -143,11 +143,11 @@ static u32 get_vertical_total(const struct intf_timing_params *timing) * lines based on the chip worst case latencies. */ static u32 programmable_fetch_get_num_lines( - struct dpu_encoder_phys_vid *vid_enc, + struct dpu_encoder_phys *phys_enc, const struct intf_timing_params *timing) { u32 worst_case_needed_lines = - vid_enc->hw_intf->cap->prog_fetch_lines_worst_case; + phys_enc->hw_intf->cap->prog_fetch_lines_worst_case; u32 start_of_frame_lines = timing->v_back_porch + timing->vsync_pulse_width; u32 needed_vfp_lines = worst_case_needed_lines - start_of_frame_lines; @@ -155,26 +155,26 @@ static u32 programmable_fetch_get_num_lines( /* Fetch must be outside active lines, otherwise undefined. */ if (start_of_frame_lines >= worst_case_needed_lines) { - DPU_DEBUG_VIDENC(vid_enc, + DPU_DEBUG_VIDENC(phys_enc, "prog fetch is not needed, large vbp+vsw\n"); actual_vfp_lines = 0; } else if (timing->v_front_porch < needed_vfp_lines) { /* Warn fetch needed, but not enough porch in panel config */ pr_warn_once ("low vbp+vfp may lead to perf issues in some cases\n"); - DPU_DEBUG_VIDENC(vid_enc, + DPU_DEBUG_VIDENC(phys_enc, "less vfp than fetch req, using entire vfp\n"); actual_vfp_lines = timing->v_front_porch; } else { - DPU_DEBUG_VIDENC(vid_enc, "room in vfp for needed prefetch\n"); + DPU_DEBUG_VIDENC(phys_enc, "room in vfp for needed prefetch\n"); actual_vfp_lines = needed_vfp_lines; } - DPU_DEBUG_VIDENC(vid_enc, + DPU_DEBUG_VIDENC(phys_enc, "v_front_porch %u v_back_porch %u vsync_pulse_width %u\n", timing->v_front_porch, timing->v_back_porch, timing->vsync_pulse_width); - DPU_DEBUG_VIDENC(vid_enc, + DPU_DEBUG_VIDENC(phys_enc, "wc_lines %u needed_vfp_lines %u actual_vfp_lines %u\n", worst_case_needed_lines, needed_vfp_lines, actual_vfp_lines); @@ -194,8 +194,6 @@ static u32 programmable_fetch_get_num_lines( static void programmable_fetch_config(struct dpu_encoder_phys *phys_enc, const struct intf_timing_params *timing) { - struct dpu_encoder_phys_vid *vid_enc = - to_dpu_encoder_phys_vid(phys_enc); struct intf_prog_fetch f = { 0 }; u32 vfp_fetch_lines = 0; u32 horiz_total = 0; @@ -203,10 +201,10 @@ static void programmable_fetch_config(struct dpu_encoder_phys *phys_enc, u32 vfp_fetch_start_vsync_counter = 0; unsigned long lock_flags; - if (WARN_ON_ONCE(!vid_enc->hw_intf->ops.setup_prg_fetch)) + if (WARN_ON_ONCE(!phys_enc->hw_intf->ops.setup_prg_fetch)) return; - vfp_fetch_lines = programmable_fetch_get_num_lines(vid_enc, timing); + vfp_fetch_lines = programmable_fetch_get_num_lines(phys_enc, timing); if (vfp_fetch_lines) { vert_total = get_vertical_total(timing); horiz_total = get_horizontal_total(timing); @@ -216,12 +214,12 @@ static void programmable_fetch_config(struct dpu_encoder_phys *phys_enc, f.fetch_start = vfp_fetch_start_vsync_counter; } - DPU_DEBUG_VIDENC(vid_enc, + DPU_DEBUG_VIDENC(phys_enc, "vfp_fetch_lines %u vfp_fetch_start_vsync_counter %u\n", vfp_fetch_lines, vfp_fetch_start_vsync_counter); spin_lock_irqsave(phys_enc->enc_spinlock, lock_flags); - vid_enc->hw_intf->ops.setup_prg_fetch(vid_enc->hw_intf, &f); + phys_enc->hw_intf->ops.setup_prg_fetch(phys_enc->hw_intf, &f); spin_unlock_irqrestore(phys_enc->enc_spinlock, lock_flags); } @@ -231,7 +229,7 @@ static bool dpu_encoder_phys_vid_mode_fixup( struct drm_display_mode *adj_mode) { if (phys_enc) - DPU_DEBUG_VIDENC(to_dpu_encoder_phys_vid(phys_enc), "\n"); + DPU_DEBUG_VIDENC(phys_enc, "\n"); /* * Modifying mode has consequences when the mode comes back to us @@ -257,12 +255,12 @@ static void dpu_encoder_phys_vid_setup_timing_engine( mode = phys_enc->cached_mode; vid_enc = to_dpu_encoder_phys_vid(phys_enc); - if (!vid_enc->hw_intf->ops.setup_timing_gen) { + if (!phys_enc->hw_intf->ops.setup_timing_gen) { DPU_ERROR("timing engine setup is not supported\n"); return; } - DPU_DEBUG_VIDENC(vid_enc, "enabling mode:\n"); + DPU_DEBUG_VIDENC(phys_enc, "enabling mode:\n"); drm_mode_debug_printmodeline(&mode); if (phys_enc->split_role != ENC_ROLE_SOLO) { @@ -271,25 +269,25 @@ static void dpu_encoder_phys_vid_setup_timing_engine( mode.hsync_start >>= 1; mode.hsync_end >>= 1; - DPU_DEBUG_VIDENC(vid_enc, + DPU_DEBUG_VIDENC(phys_enc, "split_role %d, halve horizontal %d %d %d %d\n", phys_enc->split_role, mode.hdisplay, mode.htotal, mode.hsync_start, mode.hsync_end); } - drm_mode_to_intf_timing_params(vid_enc, &mode, &timing_params); + drm_mode_to_intf_timing_params(phys_enc, &mode, &timing_params); fmt = dpu_get_dpu_format(fmt_fourcc); - DPU_DEBUG_VIDENC(vid_enc, "fmt_fourcc 0x%X\n", fmt_fourcc); + DPU_DEBUG_VIDENC(phys_enc, "fmt_fourcc 0x%X\n", fmt_fourcc); - intf_cfg.intf = vid_enc->hw_intf->idx; + intf_cfg.intf = phys_enc->hw_intf->idx; intf_cfg.intf_mode_sel = DPU_CTL_MODE_SEL_VID; intf_cfg.stream_sel = 0; /* Don't care value for video mode */ intf_cfg.mode_3d = dpu_encoder_helper_get_3d_blend_mode(phys_enc); spin_lock_irqsave(phys_enc->enc_spinlock, lock_flags); - vid_enc->hw_intf->ops.setup_timing_gen(vid_enc->hw_intf, + phys_enc->hw_intf->ops.setup_timing_gen(phys_enc->hw_intf, &timing_params, fmt); phys_enc->hw_ctl->ops.setup_intf_cfg(phys_enc->hw_ctl, &intf_cfg); spin_unlock_irqrestore(phys_enc->enc_spinlock, lock_flags); @@ -396,19 +394,15 @@ static void dpu_encoder_phys_vid_mode_set( struct drm_display_mode *mode, struct drm_display_mode *adj_mode) { - struct dpu_encoder_phys_vid *vid_enc; - if (!phys_enc || !phys_enc->dpu_kms) { DPU_ERROR("invalid encoder/kms\n"); return; } - vid_enc = to_dpu_encoder_phys_vid(phys_enc); - if (adj_mode) { phys_enc->cached_mode = *adj_mode; drm_mode_debug_printmodeline(adj_mode); - DPU_DEBUG_VIDENC(vid_enc, "caching mode:\n"); + DPU_DEBUG_VIDENC(phys_enc, "caching mode:\n"); } _dpu_encoder_phys_vid_setup_irq_hw_idx(phys_enc); @@ -419,7 +413,6 @@ static int dpu_encoder_phys_vid_control_vblank_irq( bool enable) { int ret = 0; - struct dpu_encoder_phys_vid *vid_enc; int refcount; if (!phys_enc) { @@ -428,7 +421,6 @@ static int dpu_encoder_phys_vid_control_vblank_irq( } refcount = atomic_read(&phys_enc->vblank_refcount); - vid_enc = to_dpu_encoder_phys_vid(phys_enc); /* Slave encoders don't report vblank */ if (!dpu_encoder_phys_vid_is_master(phys_enc)) @@ -453,7 +445,7 @@ end: if (ret) { DRM_ERROR("failed: id:%u intf:%d ret:%d enable:%d refcnt:%d\n", DRMID(phys_enc->parent), - vid_enc->hw_intf->idx - INTF_0, ret, enable, + phys_enc->hw_intf->idx - INTF_0, ret, enable, refcount); } return ret; @@ -462,7 +454,6 @@ end: static void dpu_encoder_phys_vid_enable(struct dpu_encoder_phys *phys_enc) { struct msm_drm_private *priv; - struct dpu_encoder_phys_vid *vid_enc; struct dpu_rm_hw_iter iter; struct dpu_hw_ctl *ctl; u32 flush_mask = 0; @@ -474,7 +465,6 @@ static void dpu_encoder_phys_vid_enable(struct dpu_encoder_phys *phys_enc) } priv = phys_enc->parent->dev->dev_private; - vid_enc = to_dpu_encoder_phys_vid(phys_enc); ctl = phys_enc->hw_ctl; dpu_rm_init_hw_iter(&iter, phys_enc->parent->base.id, DPU_HW_BLK_INTF); @@ -482,22 +472,22 @@ static void dpu_encoder_phys_vid_enable(struct dpu_encoder_phys *phys_enc) struct dpu_hw_intf *hw_intf = (struct dpu_hw_intf *)iter.hw; if (hw_intf->idx == phys_enc->intf_idx) { - vid_enc->hw_intf = hw_intf; + phys_enc->hw_intf = hw_intf; break; } } - if (!vid_enc->hw_intf) { + if (!phys_enc->hw_intf) { DPU_ERROR("hw_intf not assigned\n"); return; } - DPU_DEBUG_VIDENC(vid_enc, "\n"); + DPU_DEBUG_VIDENC(phys_enc, "\n"); - if (WARN_ON(!vid_enc->hw_intf->ops.enable_timing)) + if (WARN_ON(!phys_enc->hw_intf->ops.enable_timing)) return; - dpu_encoder_helper_split_config(phys_enc, vid_enc->hw_intf->idx); + dpu_encoder_helper_split_config(phys_enc, phys_enc->hw_intf->idx); dpu_encoder_phys_vid_setup_timing_engine(phys_enc); @@ -510,12 +500,13 @@ static void dpu_encoder_phys_vid_enable(struct dpu_encoder_phys *phys_enc) !dpu_encoder_phys_vid_is_master(phys_enc)) goto skip_flush; - ctl->ops.get_bitmask_intf(ctl, &flush_mask, vid_enc->hw_intf->idx); + ctl->ops.get_bitmask_intf(ctl, &flush_mask, phys_enc->hw_intf->idx); ctl->ops.update_pending_flush(ctl, flush_mask); skip_flush: - DPU_DEBUG_VIDENC(vid_enc, "update pending flush ctl %d flush_mask %x\n", - ctl->idx - CTL_0, flush_mask); + DPU_DEBUG_VIDENC(phys_enc, + "update pending flush ctl %d flush_mask %x\n", + ctl->idx - CTL_0, flush_mask); /* ctl_flush & timing engine enable will be triggered by framework */ if (phys_enc->enable_state == DPU_ENC_DISABLED) @@ -532,7 +523,7 @@ static void dpu_encoder_phys_vid_destroy(struct dpu_encoder_phys *phys_enc) } vid_enc = to_dpu_encoder_phys_vid(phys_enc); - DPU_DEBUG_VIDENC(vid_enc, "\n"); + DPU_DEBUG_VIDENC(phys_enc, "\n"); kfree(vid_enc); } @@ -589,7 +580,6 @@ static int dpu_encoder_phys_vid_wait_for_vblank( static void dpu_encoder_phys_vid_prepare_for_kickoff( struct dpu_encoder_phys *phys_enc) { - struct dpu_encoder_phys_vid *vid_enc; struct dpu_hw_ctl *ctl; int rc; @@ -597,7 +587,6 @@ static void dpu_encoder_phys_vid_prepare_for_kickoff( DPU_ERROR("invalid encoder/parameters\n"); return; } - vid_enc = to_dpu_encoder_phys_vid(phys_enc); ctl = phys_enc->hw_ctl; if (!ctl || !ctl->ops.wait_reset_status) @@ -609,7 +598,7 @@ static void dpu_encoder_phys_vid_prepare_for_kickoff( */ rc = ctl->ops.wait_reset_status(ctl); if (rc) { - DPU_ERROR_VIDENC(vid_enc, "ctl %d reset failure: %d\n", + DPU_ERROR_VIDENC(phys_enc, "ctl %d reset failure: %d\n", ctl->idx, rc); dpu_encoder_helper_unregister_irq(phys_enc, INTR_IDX_VSYNC); } @@ -618,7 +607,6 @@ static void dpu_encoder_phys_vid_prepare_for_kickoff( static void dpu_encoder_phys_vid_disable(struct dpu_encoder_phys *phys_enc) { struct msm_drm_private *priv; - struct dpu_encoder_phys_vid *vid_enc; unsigned long lock_flags; int ret; @@ -629,16 +617,13 @@ static void dpu_encoder_phys_vid_disable(struct dpu_encoder_phys *phys_enc) } priv = phys_enc->parent->dev->dev_private; - vid_enc = to_dpu_encoder_phys_vid(phys_enc); - if (!vid_enc->hw_intf || !phys_enc->hw_ctl) { + if (!phys_enc->hw_intf || !phys_enc->hw_ctl) { DPU_ERROR("invalid hw_intf %d hw_ctl %d\n", - vid_enc->hw_intf != 0, phys_enc->hw_ctl != 0); + phys_enc->hw_intf != 0, phys_enc->hw_ctl != 0); return; } - DPU_DEBUG_VIDENC(vid_enc, "\n"); - - if (WARN_ON(!vid_enc->hw_intf->ops.enable_timing)) + if (WARN_ON(!phys_enc->hw_intf->ops.enable_timing)) return; if (phys_enc->enable_state == DPU_ENC_DISABLED) { @@ -647,7 +632,7 @@ static void dpu_encoder_phys_vid_disable(struct dpu_encoder_phys *phys_enc) } spin_lock_irqsave(phys_enc->enc_spinlock, lock_flags); - vid_enc->hw_intf->ops.enable_timing(vid_enc->hw_intf, 0); + phys_enc->hw_intf->ops.enable_timing(phys_enc->hw_intf, 0); if (dpu_encoder_phys_vid_is_master(phys_enc)) dpu_encoder_phys_inc_pending(phys_enc); spin_unlock_irqrestore(phys_enc->enc_spinlock, lock_flags); @@ -666,7 +651,7 @@ static void dpu_encoder_phys_vid_disable(struct dpu_encoder_phys *phys_enc) atomic_set(&phys_enc->pending_kickoff_cnt, 0); DRM_ERROR("wait disable failed: id:%u intf:%d ret:%d\n", DRMID(phys_enc->parent), - vid_enc->hw_intf->idx - INTF_0, ret); + phys_enc->hw_intf->idx - INTF_0, ret); } } @@ -677,25 +662,21 @@ static void dpu_encoder_phys_vid_handle_post_kickoff( struct dpu_encoder_phys *phys_enc) { unsigned long lock_flags; - struct dpu_encoder_phys_vid *vid_enc; if (!phys_enc) { DPU_ERROR("invalid encoder\n"); return; } - vid_enc = to_dpu_encoder_phys_vid(phys_enc); - DPU_DEBUG_VIDENC(vid_enc, "enable_state %d\n", phys_enc->enable_state); - /* * Video mode must flush CTL before enabling timing engine * Video encoders need to turn on their interfaces now */ if (phys_enc->enable_state == DPU_ENC_ENABLING) { trace_dpu_enc_phys_vid_post_kickoff(DRMID(phys_enc->parent), - vid_enc->hw_intf->idx - INTF_0); + phys_enc->hw_intf->idx - INTF_0); spin_lock_irqsave(phys_enc->enc_spinlock, lock_flags); - vid_enc->hw_intf->ops.enable_timing(vid_enc->hw_intf, 1); + phys_enc->hw_intf->ops.enable_timing(phys_enc->hw_intf, 1); spin_unlock_irqrestore(phys_enc->enc_spinlock, lock_flags); phys_enc->enable_state = DPU_ENC_ENABLED; } @@ -704,16 +685,13 @@ static void dpu_encoder_phys_vid_handle_post_kickoff( static void dpu_encoder_phys_vid_irq_control(struct dpu_encoder_phys *phys_enc, bool enable) { - struct dpu_encoder_phys_vid *vid_enc; int ret; if (!phys_enc) return; - vid_enc = to_dpu_encoder_phys_vid(phys_enc); - trace_dpu_enc_phys_vid_irq_ctrl(DRMID(phys_enc->parent), - vid_enc->hw_intf->idx - INTF_0, + phys_enc->hw_intf->idx - INTF_0, enable, atomic_read(&phys_enc->vblank_refcount)); @@ -732,19 +710,16 @@ static void dpu_encoder_phys_vid_irq_control(struct dpu_encoder_phys *phys_enc, static int dpu_encoder_phys_vid_get_line_count( struct dpu_encoder_phys *phys_enc) { - struct dpu_encoder_phys_vid *vid_enc; - if (!phys_enc) return -EINVAL; if (!dpu_encoder_phys_vid_is_master(phys_enc)) return -EINVAL; - vid_enc = to_dpu_encoder_phys_vid(phys_enc); - if (!vid_enc->hw_intf || !vid_enc->hw_intf->ops.get_line_count) + if (!phys_enc->hw_intf || !phys_enc->hw_intf->ops.get_line_count) return -EINVAL; - return vid_enc->hw_intf->ops.get_line_count(vid_enc->hw_intf); + return phys_enc->hw_intf->ops.get_line_count(phys_enc->hw_intf); } static void dpu_encoder_phys_vid_init_ops(struct dpu_encoder_phys_ops *ops) @@ -791,7 +766,7 @@ struct dpu_encoder_phys *dpu_encoder_phys_vid_init( phys_enc->hw_mdptop = p->dpu_kms->hw_mdp; phys_enc->intf_idx = p->intf_idx; - DPU_DEBUG_VIDENC(vid_enc, "\n"); + DPU_DEBUG_VIDENC(phys_enc, "\n"); dpu_encoder_phys_vid_init_ops(&phys_enc->ops); phys_enc->parent = p->parent; @@ -825,7 +800,7 @@ struct dpu_encoder_phys *dpu_encoder_phys_vid_init( init_waitqueue_head(&phys_enc->pending_kickoff_wq); phys_enc->enable_state = DPU_ENC_DISABLED; - DPU_DEBUG_VIDENC(vid_enc, "created intf idx:%d\n", p->intf_idx); + DPU_DEBUG_VIDENC(phys_enc, "created intf idx:%d\n", p->intf_idx); return phys_enc; From 366334a3127eb654410660987919e1de231f15f2 Mon Sep 17 00:00:00 2001 From: Jeykumar Sankaran Date: Wed, 13 Feb 2019 17:19:11 -0800 Subject: [PATCH 10/34] drm/msm/dpu: remove phys_vid subclass Not holding any video encoder specific data. Get rid of it. Signed-off-by: Jeykumar Sankaran Signed-off-by: Sean Paul Link: https://patchwork.freedesktop.org/patch/msgid/1550107156-17625-3-git-send-email-jsanka@codeaurora.org Signed-off-by: Rob Clark --- .../gpu/drm/msm/disp/dpu1/dpu_encoder_phys.h | 11 ----------- .../drm/msm/disp/dpu1/dpu_encoder_phys_vid.c | 18 ++++-------------- 2 files changed, 4 insertions(+), 25 deletions(-) diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys.h b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys.h index 66a1fd4b014c..97fb868a4ef6 100644 --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys.h +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys.h @@ -252,17 +252,6 @@ static inline int dpu_encoder_phys_inc_pending(struct dpu_encoder_phys *phys) return atomic_inc_return(&phys->pending_kickoff_cnt); } -/** - * struct dpu_encoder_phys_vid - sub-class of dpu_encoder_phys to handle video - * mode specific operations - * @base: Baseclass physical encoder structure - * @timing_params: Current timing parameter - */ -struct dpu_encoder_phys_vid { - struct dpu_encoder_phys base; - struct intf_timing_params timing_params; -}; - /** * struct dpu_encoder_phys_cmd - sub-class of dpu_encoder_phys to handle command * mode specific operations diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys_vid.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys_vid.c index 1b706eada90f..ae5449c73c5e 100644 --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys_vid.c +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys_vid.c @@ -240,7 +240,6 @@ static bool dpu_encoder_phys_vid_mode_fixup( static void dpu_encoder_phys_vid_setup_timing_engine( struct dpu_encoder_phys *phys_enc) { - struct dpu_encoder_phys_vid *vid_enc; struct drm_display_mode mode; struct intf_timing_params timing_params = { 0 }; const struct dpu_format *fmt = NULL; @@ -254,7 +253,6 @@ static void dpu_encoder_phys_vid_setup_timing_engine( } mode = phys_enc->cached_mode; - vid_enc = to_dpu_encoder_phys_vid(phys_enc); if (!phys_enc->hw_intf->ops.setup_timing_gen) { DPU_ERROR("timing engine setup is not supported\n"); return; @@ -293,8 +291,6 @@ static void dpu_encoder_phys_vid_setup_timing_engine( spin_unlock_irqrestore(phys_enc->enc_spinlock, lock_flags); programmable_fetch_config(phys_enc, &timing_params); - - vid_enc->timing_params = timing_params; } static void dpu_encoder_phys_vid_vblank_irq(void *arg, int irq_idx) @@ -515,16 +511,13 @@ skip_flush: static void dpu_encoder_phys_vid_destroy(struct dpu_encoder_phys *phys_enc) { - struct dpu_encoder_phys_vid *vid_enc; - if (!phys_enc) { DPU_ERROR("invalid encoder\n"); return; } - vid_enc = to_dpu_encoder_phys_vid(phys_enc); DPU_DEBUG_VIDENC(phys_enc, "\n"); - kfree(vid_enc); + kfree(phys_enc); } static void dpu_encoder_phys_vid_get_hw_resources( @@ -746,7 +739,6 @@ struct dpu_encoder_phys *dpu_encoder_phys_vid_init( struct dpu_enc_phys_init_params *p) { struct dpu_encoder_phys *phys_enc = NULL; - struct dpu_encoder_phys_vid *vid_enc = NULL; struct dpu_encoder_irq *irq; int i, ret = 0; @@ -755,14 +747,12 @@ struct dpu_encoder_phys *dpu_encoder_phys_vid_init( goto fail; } - vid_enc = kzalloc(sizeof(*vid_enc), GFP_KERNEL); - if (!vid_enc) { + phys_enc = kzalloc(sizeof(*phys_enc), GFP_KERNEL); + if (!phys_enc) { ret = -ENOMEM; goto fail; } - phys_enc = &vid_enc->base; - phys_enc->hw_mdptop = p->dpu_kms->hw_mdp; phys_enc->intf_idx = p->intf_idx; @@ -806,7 +796,7 @@ struct dpu_encoder_phys *dpu_encoder_phys_vid_init( fail: DPU_ERROR("failed to create encoder\n"); - if (vid_enc) + if (phys_enc) dpu_encoder_phys_vid_destroy(phys_enc); return ERR_PTR(ret); From a7fcc3237f31a4e206953bb73cf41bd429442f09 Mon Sep 17 00:00:00 2001 From: Jeykumar Sankaran Date: Wed, 13 Feb 2019 17:19:12 -0800 Subject: [PATCH 11/34] drm/msm/dpu: release resources on modeset failure release resources allocated in mode_set if any of the hw check fails. Most of these checks are not necessary and they will be removed in the follow up patches with state based resource allocations. Signed-off-by: Jeykumar Sankaran Signed-off-by: Sean Paul Link: https://patchwork.freedesktop.org/patch/msgid/1550107156-17625-4-git-send-email-jsanka@codeaurora.org Signed-off-by: Rob Clark --- drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c index c51bc0fdd9c3..7d7779d70145 100644 --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c @@ -1026,13 +1026,13 @@ static void dpu_encoder_virt_mode_set(struct drm_encoder *drm_enc, if (!dpu_enc->hw_pp[i]) { DPU_ERROR_ENC(dpu_enc, "no pp block assigned" "at idx: %d\n", i); - return; + goto error; } if (!hw_ctl[i]) { DPU_ERROR_ENC(dpu_enc, "no ctl block assigned" "at idx: %d\n", i); - return; + goto error; } phys->hw_pp = dpu_enc->hw_pp[i]; @@ -1045,6 +1045,9 @@ static void dpu_encoder_virt_mode_set(struct drm_encoder *drm_enc, } dpu_enc->mode_set_complete = true; + +error: + dpu_rm_release(&dpu_kms->rm, drm_enc); } static void _dpu_encoder_virt_enable_helper(struct drm_encoder *drm_enc) From c2ab55a68a33746591971f3e1d4ef3374f1d7e18 Mon Sep 17 00:00:00 2001 From: Jeykumar Sankaran Date: Wed, 13 Feb 2019 17:19:13 -0800 Subject: [PATCH 12/34] drm/msm/dpu: dont use encoder->crtc in atomic path encoder->crtc is not really meaningful for atomic path. Use crtc->encoder_mask to identify the crtc attached with an encoder. Signed-off-by: Jeykumar Sankaran Signed-off-by: Sean Paul Link: https://patchwork.freedesktop.org/patch/msgid/1550107156-17625-5-git-send-email-jsanka@codeaurora.org Signed-off-by: Rob Clark --- drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c index 7d7779d70145..99cbf3524348 100644 --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c @@ -962,6 +962,7 @@ static void dpu_encoder_virt_mode_set(struct drm_encoder *drm_enc, struct dpu_kms *dpu_kms; struct list_head *connector_list; struct drm_connector *conn = NULL, *conn_iter; + struct drm_crtc *drm_crtc; struct dpu_rm_hw_iter pp_iter, ctl_iter; struct msm_display_topology topology; struct dpu_hw_ctl *hw_ctl[MAX_CHANNELS_PER_ENC] = { NULL }; @@ -993,10 +994,14 @@ static void dpu_encoder_virt_mode_set(struct drm_encoder *drm_enc, return; } + drm_for_each_crtc(drm_crtc, drm_enc->dev) + if (drm_crtc->state->encoder_mask & drm_encoder_mask(drm_enc)) + break; + topology = dpu_encoder_get_topology(dpu_enc, dpu_kms, adj_mode); /* Reserve dynamic resources now. Indicating non-AtomicTest phase */ - ret = dpu_rm_reserve(&dpu_kms->rm, drm_enc, drm_enc->crtc->state, + ret = dpu_rm_reserve(&dpu_kms->rm, drm_enc, drm_crtc->state, topology, false); if (ret) { DPU_ERROR_ENC(dpu_enc, From b107603b4ad0f2624425bb2402701480bb59cafe Mon Sep 17 00:00:00 2001 From: Jeykumar Sankaran Date: Wed, 13 Feb 2019 17:19:14 -0800 Subject: [PATCH 13/34] drm/msm/dpu: map mixer/ctl hw blocks in encoder modeset After resource allocation, iterate and populate mixer/ctl hw blocks in encoder modeset thereby centralizing all the resource mapping to the CRTC. This change is made for easy switching to state based allocation using private objects later in this series. Signed-off-by: Jeykumar Sankaran Signed-off-by: Sean Paul Link: https://patchwork.freedesktop.org/patch/msgid/1550107156-17625-6-git-send-email-jsanka@codeaurora.org Signed-off-by: Rob Clark --- drivers/gpu/drm/msm/disp/dpu1/dpu_crtc.c | 64 +-------------------- drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c | 37 +++++++++--- 2 files changed, 31 insertions(+), 70 deletions(-) diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_crtc.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_crtc.c index 366e88f6c5d1..dfdfa766da8f 100644 --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_crtc.c +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_crtc.c @@ -428,65 +428,6 @@ void dpu_crtc_complete_commit(struct drm_crtc *crtc, trace_dpu_crtc_complete_commit(DRMID(crtc)); } -static void _dpu_crtc_setup_mixer_for_encoder( - struct drm_crtc *crtc, - struct drm_encoder *enc) -{ - struct dpu_crtc_state *cstate = to_dpu_crtc_state(crtc->state); - struct dpu_kms *dpu_kms = _dpu_crtc_get_kms(crtc); - struct dpu_rm *rm = &dpu_kms->rm; - struct dpu_crtc_mixer *mixer; - struct dpu_hw_ctl *last_valid_ctl = NULL; - int i; - struct dpu_rm_hw_iter lm_iter, ctl_iter; - - dpu_rm_init_hw_iter(&lm_iter, enc->base.id, DPU_HW_BLK_LM); - dpu_rm_init_hw_iter(&ctl_iter, enc->base.id, DPU_HW_BLK_CTL); - - /* Set up all the mixers and ctls reserved by this encoder */ - for (i = cstate->num_mixers; i < ARRAY_SIZE(cstate->mixers); i++) { - mixer = &cstate->mixers[i]; - - if (!dpu_rm_get_hw(rm, &lm_iter)) - break; - mixer->hw_lm = (struct dpu_hw_mixer *)lm_iter.hw; - - /* CTL may be <= LMs, if <, multiple LMs controlled by 1 CTL */ - if (!dpu_rm_get_hw(rm, &ctl_iter)) { - DPU_DEBUG("no ctl assigned to lm %d, using previous\n", - mixer->hw_lm->idx - LM_0); - mixer->lm_ctl = last_valid_ctl; - } else { - mixer->lm_ctl = (struct dpu_hw_ctl *)ctl_iter.hw; - last_valid_ctl = mixer->lm_ctl; - } - - /* Shouldn't happen, mixers are always >= ctls */ - if (!mixer->lm_ctl) { - DPU_ERROR("no valid ctls found for lm %d\n", - mixer->hw_lm->idx - LM_0); - return; - } - - cstate->num_mixers++; - DPU_DEBUG("setup mixer %d: lm %d\n", - i, mixer->hw_lm->idx - LM_0); - DPU_DEBUG("setup mixer %d: ctl %d\n", - i, mixer->lm_ctl->idx - CTL_0); - } -} - -static void _dpu_crtc_setup_mixers(struct drm_crtc *crtc) -{ - struct drm_encoder *enc; - - WARN_ON(!drm_modeset_is_locked(&crtc->mutex)); - - /* Check for mixers on all encoders attached to this crtc */ - drm_for_each_encoder_mask(enc, crtc->dev, crtc->state->encoder_mask) - _dpu_crtc_setup_mixer_for_encoder(crtc, enc); -} - static void _dpu_crtc_setup_lm_bounds(struct drm_crtc *crtc, struct drm_crtc_state *state) { @@ -536,10 +477,7 @@ static void dpu_crtc_atomic_begin(struct drm_crtc *crtc, dev = crtc->dev; smmu_state = &dpu_crtc->smmu_state; - if (!cstate->num_mixers) { - _dpu_crtc_setup_mixers(crtc); - _dpu_crtc_setup_lm_bounds(crtc, crtc->state); - } + _dpu_crtc_setup_lm_bounds(crtc, crtc->state); if (dpu_crtc->event) { WARN_ON(dpu_crtc->event); diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c index 99cbf3524348..c4e4e19b5ea1 100644 --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c @@ -963,9 +963,12 @@ static void dpu_encoder_virt_mode_set(struct drm_encoder *drm_enc, struct list_head *connector_list; struct drm_connector *conn = NULL, *conn_iter; struct drm_crtc *drm_crtc; - struct dpu_rm_hw_iter pp_iter, ctl_iter; + struct dpu_crtc_state *cstate; + struct dpu_rm_hw_iter hw_iter; struct msm_display_topology topology; struct dpu_hw_ctl *hw_ctl[MAX_CHANNELS_PER_ENC] = { NULL }; + struct dpu_hw_mixer *hw_lm[MAX_CHANNELS_PER_ENC] = { NULL }; + int num_lm = 0, num_ctl = 0; int i = 0, ret; if (!drm_enc) { @@ -1009,21 +1012,41 @@ static void dpu_encoder_virt_mode_set(struct drm_encoder *drm_enc, return; } - dpu_rm_init_hw_iter(&pp_iter, drm_enc->base.id, DPU_HW_BLK_PINGPONG); + dpu_rm_init_hw_iter(&hw_iter, drm_enc->base.id, DPU_HW_BLK_PINGPONG); for (i = 0; i < MAX_CHANNELS_PER_ENC; i++) { dpu_enc->hw_pp[i] = NULL; - if (!dpu_rm_get_hw(&dpu_kms->rm, &pp_iter)) + if (!dpu_rm_get_hw(&dpu_kms->rm, &hw_iter)) break; - dpu_enc->hw_pp[i] = (struct dpu_hw_pingpong *) pp_iter.hw; + dpu_enc->hw_pp[i] = (struct dpu_hw_pingpong *) hw_iter.hw; } - dpu_rm_init_hw_iter(&ctl_iter, drm_enc->base.id, DPU_HW_BLK_CTL); + dpu_rm_init_hw_iter(&hw_iter, drm_enc->base.id, DPU_HW_BLK_CTL); for (i = 0; i < MAX_CHANNELS_PER_ENC; i++) { - if (!dpu_rm_get_hw(&dpu_kms->rm, &ctl_iter)) + if (!dpu_rm_get_hw(&dpu_kms->rm, &hw_iter)) break; - hw_ctl[i] = (struct dpu_hw_ctl *)ctl_iter.hw; + hw_ctl[i] = (struct dpu_hw_ctl *)hw_iter.hw; + num_ctl++; } + dpu_rm_init_hw_iter(&hw_iter, drm_enc->base.id, DPU_HW_BLK_LM); + for (i = 0; i < MAX_CHANNELS_PER_ENC; i++) { + if (!dpu_rm_get_hw(&dpu_kms->rm, &hw_iter)) + break; + hw_lm[i] = (struct dpu_hw_mixer *)hw_iter.hw; + num_lm++; + } + + cstate = to_dpu_crtc_state(drm_crtc->state); + + for (i = 0; i < num_lm; i++) { + int ctl_idx = (i < num_ctl) ? i : (num_ctl-1); + + cstate->mixers[i].hw_lm = hw_lm[i]; + cstate->mixers[i].lm_ctl = hw_ctl[ctl_idx]; + } + + cstate->num_mixers = num_lm; + for (i = 0; i < dpu_enc->num_phys_encs; i++) { struct dpu_encoder_phys *phys = dpu_enc->phys_encs[i]; From 9d4c8fd1af4687eb477ae26b7ecc386b336cea72 Mon Sep 17 00:00:00 2001 From: Jeykumar Sankaran Date: Wed, 13 Feb 2019 17:19:15 -0800 Subject: [PATCH 14/34] drm/msm/dpu: assign intf to encoder in mode_set Iterate and assign HW intf block to physical encoders in encoder modeset. Moving all the HW block assignments to encoder modeset to allow easy switching to state based resource management. Signed-off-by: Jeykumar Sankaran Signed-off-by: Sean Paul Link: https://patchwork.freedesktop.org/patch/msgid/1550107156-17625-7-git-send-email-jsanka@codeaurora.org Signed-off-by: Rob Clark --- drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c | 22 ++++++++++++++++- .../drm/msm/disp/dpu1/dpu_encoder_phys_vid.c | 24 ------------------- 2 files changed, 21 insertions(+), 25 deletions(-) diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c index c4e4e19b5ea1..82bf16d61a45 100644 --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c @@ -969,7 +969,7 @@ static void dpu_encoder_virt_mode_set(struct drm_encoder *drm_enc, struct dpu_hw_ctl *hw_ctl[MAX_CHANNELS_PER_ENC] = { NULL }; struct dpu_hw_mixer *hw_lm[MAX_CHANNELS_PER_ENC] = { NULL }; int num_lm = 0, num_ctl = 0; - int i = 0, ret; + int i, j, ret; if (!drm_enc) { DPU_ERROR("invalid encoder\n"); @@ -1066,6 +1066,26 @@ static void dpu_encoder_virt_mode_set(struct drm_encoder *drm_enc, phys->hw_pp = dpu_enc->hw_pp[i]; phys->hw_ctl = hw_ctl[i]; + dpu_rm_init_hw_iter(&hw_iter, drm_enc->base.id, + DPU_HW_BLK_INTF); + for (j = 0; j < MAX_CHANNELS_PER_ENC; j++) { + struct dpu_hw_intf *hw_intf; + + if (!dpu_rm_get_hw(&dpu_kms->rm, &hw_iter)) + break; + + hw_intf = (struct dpu_hw_intf *)hw_iter.hw; + if (hw_intf->idx == phys->intf_idx) + phys->hw_intf = hw_intf; + } + + if (!phys->hw_intf) { + DPU_ERROR_ENC(dpu_enc, + "no intf block assigned at idx: %d\n", + i); + goto error; + } + phys->connector = conn->state->connector; if (phys->ops.mode_set) phys->ops.mode_set(phys, mode, adj_mode); diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys_vid.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys_vid.c index ae5449c73c5e..0adfdfc8775f 100644 --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys_vid.c +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys_vid.c @@ -449,35 +449,11 @@ end: static void dpu_encoder_phys_vid_enable(struct dpu_encoder_phys *phys_enc) { - struct msm_drm_private *priv; - struct dpu_rm_hw_iter iter; struct dpu_hw_ctl *ctl; u32 flush_mask = 0; - if (!phys_enc || !phys_enc->parent || !phys_enc->parent->dev || - !phys_enc->parent->dev->dev_private) { - DPU_ERROR("invalid encoder/device\n"); - return; - } - priv = phys_enc->parent->dev->dev_private; - ctl = phys_enc->hw_ctl; - dpu_rm_init_hw_iter(&iter, phys_enc->parent->base.id, DPU_HW_BLK_INTF); - while (dpu_rm_get_hw(&phys_enc->dpu_kms->rm, &iter)) { - struct dpu_hw_intf *hw_intf = (struct dpu_hw_intf *)iter.hw; - - if (hw_intf->idx == phys_enc->intf_idx) { - phys_enc->hw_intf = hw_intf; - break; - } - } - - if (!phys_enc->hw_intf) { - DPU_ERROR("hw_intf not assigned\n"); - return; - } - DPU_DEBUG_VIDENC(phys_enc, "\n"); if (WARN_ON(!phys_enc->hw_intf->ops.enable_timing)) From 4e99851742470a64326d5484a44e7940c0e6d056 Mon Sep 17 00:00:00 2001 From: Jeykumar Sankaran Date: Wed, 13 Feb 2019 17:19:16 -0800 Subject: [PATCH 15/34] drm/msm/dpu: check split role for single flush Removing unwanted access of crtc_state for finding this information. Use split role information to know whether we have slave ctl. Signed-off-by: Jeykumar Sankaran Signed-off-by: Sean Paul Link: https://patchwork.freedesktop.org/patch/msgid/1550107156-17625-8-git-send-email-jsanka@codeaurora.org Signed-off-by: Rob Clark --- .../gpu/drm/msm/disp/dpu1/dpu_encoder_phys_vid.c | 14 +------------- 1 file changed, 1 insertion(+), 13 deletions(-) diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys_vid.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys_vid.c index 0adfdfc8775f..1b7a335a6140 100644 --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys_vid.c +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys_vid.c @@ -347,22 +347,10 @@ static void dpu_encoder_phys_vid_underrun_irq(void *arg, int irq_idx) phys_enc); } -static bool _dpu_encoder_phys_is_dual_ctl(struct dpu_encoder_phys *phys_enc) -{ - struct dpu_crtc_state *dpu_cstate; - - if (!phys_enc) - return false; - - dpu_cstate = to_dpu_crtc_state(phys_enc->parent->crtc->state); - - return dpu_cstate->num_ctls > 1; -} - static bool dpu_encoder_phys_vid_needs_single_flush( struct dpu_encoder_phys *phys_enc) { - return (phys_enc && _dpu_encoder_phys_is_dual_ctl(phys_enc)); + return phys_enc->split_role != ENC_ROLE_SOLO; } static void _dpu_encoder_phys_vid_setup_irq_hw_idx( From e1505f627e2edaf9ba6cad4a352d897db1f99ef3 Mon Sep 17 00:00:00 2001 From: Douglas Anderson Date: Wed, 16 Jan 2019 10:46:22 -0800 Subject: [PATCH 16/34] drm/msm: Cleanup A6XX opp-level reading The patch ("OPP: Add support for parsing the 'opp-level' property") adds an API enabling a cleaner way to read the opp-level. Let's use the new API. Signed-off-by: Douglas Anderson Reviewed-by: Jordan Crouse Signed-off-by: Rob Clark Signed-off-by: Rob Clark --- drivers/gpu/drm/msm/adreno/a6xx_gmu.c | 17 ++++++----------- 1 file changed, 6 insertions(+), 11 deletions(-) diff --git a/drivers/gpu/drm/msm/adreno/a6xx_gmu.c b/drivers/gpu/drm/msm/adreno/a6xx_gmu.c index d1662a75c7ec..a12aa4664c8b 100644 --- a/drivers/gpu/drm/msm/adreno/a6xx_gmu.c +++ b/drivers/gpu/drm/msm/adreno/a6xx_gmu.c @@ -948,25 +948,20 @@ static int a6xx_gmu_memory_probe(struct a6xx_gmu *gmu) } /* Return the 'arc-level' for the given frequency */ -static u32 a6xx_gmu_get_arc_level(struct device *dev, unsigned long freq) +static unsigned int a6xx_gmu_get_arc_level(struct device *dev, + unsigned long freq) { struct dev_pm_opp *opp; - struct device_node *np; - u32 val = 0; + unsigned int val; if (!freq) return 0; - opp = dev_pm_opp_find_freq_exact(dev, freq, true); + opp = dev_pm_opp_find_freq_exact(dev, freq, true); if (IS_ERR(opp)) return 0; - np = dev_pm_opp_get_of_node(opp); - - if (np) { - of_property_read_u32(np, "opp-level", &val); - of_node_put(np); - } + val = dev_pm_opp_get_level(opp); dev_pm_opp_put(opp); @@ -1002,7 +997,7 @@ static int a6xx_gmu_rpmh_arc_votes_init(struct device *dev, u32 *votes, /* Construct a vote for each frequency */ for (i = 0; i < freqs_count; i++) { u8 pindex = 0, sindex = 0; - u32 level = a6xx_gmu_get_arc_level(dev, freqs[i]); + unsigned int level = a6xx_gmu_get_arc_level(dev, freqs[i]); /* Get the primary index that matches the arc level */ for (j = 0; j < pri_count; j++) { From 6cd5235c3135ea84b32469ea51b2aae384eda8af Mon Sep 17 00:00:00 2001 From: Wen Yang Date: Thu, 4 Apr 2019 00:04:11 +0800 Subject: [PATCH 17/34] drm/msm: a5xx: fix possible object reference leak The call to of_get_child_by_name returns a node pointer with refcount incremented thus it must be explicitly decremented after the last usage. Detected by coccinelle with the following warnings: drivers/gpu/drm/msm/adreno/a5xx_gpu.c:57:2-8: ERROR: missing of_node_put; acquired a node pointer with refcount incremented on line 47, but without a corresponding object release within this function. drivers/gpu/drm/msm/adreno/a5xx_gpu.c:66:2-8: ERROR: missing of_node_put; acquired a node pointer with refcount incremented on line 47, but without a corresponding object release within this function. drivers/gpu/drm/msm/adreno/a5xx_gpu.c:118:1-7: ERROR: missing of_node_put; acquired a node pointer with refcount incremented on line 47, but without a corresponding object release within this function. drivers/gpu/drm/msm/adreno/a5xx_gpu.c:57:2-8: ERROR: missing of_node_put; acquired a node pointer with refcount incremented on line 51, but without a corresponding object release within this function. drivers/gpu/drm/msm/adreno/a5xx_gpu.c:66:2-8: ERROR: missing of_node_put; acquired a node pointer with refcount incremented on line 51, but without a corresponding object release within this function. drivers/gpu/drm/msm/adreno/a5xx_gpu.c:118:1-7: ERROR: missing of_node_put; acquired a node pointer with refcount incremented on line 51, but without a corresponding object release within this function. Signed-off-by: Wen Yang Cc: Rob Clark Cc: Sean Paul Cc: David Airlie Cc: Daniel Vetter Cc: Jordan Crouse Cc: Mamta Shukla Cc: Thomas Zimmermann Cc: Sharat Masetty Cc: linux-arm-msm@vger.kernel.org Cc: dri-devel@lists.freedesktop.org Cc: freedreno@lists.freedesktop.org Cc: linux-kernel@vger.kernel.org (open list) Reviewed-by: Jordan Crouse Signed-off-by: Rob Clark Signed-off-by: Rob Clark --- drivers/gpu/drm/msm/adreno/a5xx_gpu.c | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-) diff --git a/drivers/gpu/drm/msm/adreno/a5xx_gpu.c b/drivers/gpu/drm/msm/adreno/a5xx_gpu.c index d5f5e56422f5..270da14cba67 100644 --- a/drivers/gpu/drm/msm/adreno/a5xx_gpu.c +++ b/drivers/gpu/drm/msm/adreno/a5xx_gpu.c @@ -34,7 +34,7 @@ static int zap_shader_load_mdt(struct msm_gpu *gpu, const char *fwname) { struct device *dev = &gpu->pdev->dev; const struct firmware *fw; - struct device_node *np; + struct device_node *np, *mem_np; struct resource r; phys_addr_t mem_phys; ssize_t mem_size; @@ -48,11 +48,13 @@ static int zap_shader_load_mdt(struct msm_gpu *gpu, const char *fwname) if (!np) return -ENODEV; - np = of_parse_phandle(np, "memory-region", 0); - if (!np) + mem_np = of_parse_phandle(np, "memory-region", 0); + of_node_put(np); + if (!mem_np) return -EINVAL; - ret = of_address_to_resource(np, 0, &r); + ret = of_address_to_resource(mem_np, 0, &r); + of_node_put(mem_np); if (ret) return ret; From d674c963af7470b6394b4b7f98cf2716b3a757d7 Mon Sep 17 00:00:00 2001 From: Rob Clark Date: Mon, 15 Apr 2019 16:55:45 -0700 Subject: [PATCH 18/34] drm/msm/gpu: add per-process pagetables param For now it always returns '0' (false), but once the iommu work is in place to enable per-process pagetables we can update the value returned. Userspace needs to know this to make an informed decision about exposing KHR_robustness. Signed-off-by: Rob Clark Reviewed-by: Jordan Crouse --- drivers/gpu/drm/msm/adreno/adreno_gpu.c | 3 +++ include/uapi/drm/msm_drm.h | 1 + 2 files changed, 4 insertions(+) diff --git a/drivers/gpu/drm/msm/adreno/adreno_gpu.c b/drivers/gpu/drm/msm/adreno/adreno_gpu.c index 27898475cdf4..7727917806ce 100644 --- a/drivers/gpu/drm/msm/adreno/adreno_gpu.c +++ b/drivers/gpu/drm/msm/adreno/adreno_gpu.c @@ -63,6 +63,9 @@ int adreno_get_param(struct msm_gpu *gpu, uint32_t param, uint64_t *value) case MSM_PARAM_NR_RINGS: *value = gpu->nr_rings; return 0; + case MSM_PARAM_PP_PGTABLE: + *value = 0; + return 0; default: DBG("%s: invalid param: %u", gpu->name, param); return -EINVAL; diff --git a/include/uapi/drm/msm_drm.h b/include/uapi/drm/msm_drm.h index 91a16b333c69..a9fdcf1689ce 100644 --- a/include/uapi/drm/msm_drm.h +++ b/include/uapi/drm/msm_drm.h @@ -74,6 +74,7 @@ struct drm_msm_timespec { #define MSM_PARAM_TIMESTAMP 0x05 #define MSM_PARAM_GMEM_BASE 0x06 #define MSM_PARAM_NR_RINGS 0x07 +#define MSM_PARAM_PP_PGTABLE 0x08 /* => 1 for per-process pagetables, else 0 */ struct drm_msm_param { __u32 pipe; /* in, MSM_PIPE_x */ From 48dc4241c9cd62bfbe33625e669e21d7081d35fa Mon Sep 17 00:00:00 2001 From: Rob Clark Date: Tue, 16 Apr 2019 16:13:28 -0700 Subject: [PATCH 19/34] drm/msm: add param to retrieve # of GPU faults (global) For KHR_robustness, userspace wants to know two things, the count of GPU faults globally, and the count of faults attributed to a given context. This patch providees the former, and the next patch provides the latter. Signed-off-by: Rob Clark Reviewed-by: Jordan Crouse --- drivers/gpu/drm/msm/adreno/adreno_gpu.c | 3 +++ drivers/gpu/drm/msm/msm_gpu.c | 3 +++ drivers/gpu/drm/msm/msm_gpu.h | 3 +++ include/uapi/drm/msm_drm.h | 1 + 4 files changed, 10 insertions(+) diff --git a/drivers/gpu/drm/msm/adreno/adreno_gpu.c b/drivers/gpu/drm/msm/adreno/adreno_gpu.c index 7727917806ce..6de08cfc38aa 100644 --- a/drivers/gpu/drm/msm/adreno/adreno_gpu.c +++ b/drivers/gpu/drm/msm/adreno/adreno_gpu.c @@ -66,6 +66,9 @@ int adreno_get_param(struct msm_gpu *gpu, uint32_t param, uint64_t *value) case MSM_PARAM_PP_PGTABLE: *value = 0; return 0; + case MSM_PARAM_FAULTS: + *value = gpu->global_faults; + return 0; default: DBG("%s: invalid param: %u", gpu->name, param); return -EINVAL; diff --git a/drivers/gpu/drm/msm/msm_gpu.c b/drivers/gpu/drm/msm/msm_gpu.c index 10babd18e286..194847a220b6 100644 --- a/drivers/gpu/drm/msm/msm_gpu.c +++ b/drivers/gpu/drm/msm/msm_gpu.c @@ -443,6 +443,9 @@ static void recover_worker(struct work_struct *work) if (submit) { struct task_struct *task; + /* Increment the fault count */ + gpu->global_faults++; + task = get_pid_task(submit->pid, PIDTYPE_PID); if (task) { comm = kstrdup(task->comm, GFP_KERNEL); diff --git a/drivers/gpu/drm/msm/msm_gpu.h b/drivers/gpu/drm/msm/msm_gpu.h index 6241986bab51..f2739cd97cea 100644 --- a/drivers/gpu/drm/msm/msm_gpu.h +++ b/drivers/gpu/drm/msm/msm_gpu.h @@ -104,6 +104,9 @@ struct msm_gpu { /* does gpu need hw_init? */ bool needs_hw_init; + /* number of GPU hangs (for all contexts) */ + int global_faults; + /* worker for handling active-list retiring: */ struct work_struct retire_work; diff --git a/include/uapi/drm/msm_drm.h b/include/uapi/drm/msm_drm.h index a9fdcf1689ce..178d7b407f3a 100644 --- a/include/uapi/drm/msm_drm.h +++ b/include/uapi/drm/msm_drm.h @@ -75,6 +75,7 @@ struct drm_msm_timespec { #define MSM_PARAM_GMEM_BASE 0x06 #define MSM_PARAM_NR_RINGS 0x07 #define MSM_PARAM_PP_PGTABLE 0x08 /* => 1 for per-process pagetables, else 0 */ +#define MSM_PARAM_FAULTS 0x09 struct drm_msm_param { __u32 pipe; /* in, MSM_PIPE_x */ From b0fb66043a0d471a5acfe743adf90a0c447aca6b Mon Sep 17 00:00:00 2001 From: Jordan Crouse Date: Fri, 22 Mar 2019 14:21:22 -0600 Subject: [PATCH 20/34] drm/msm/gpu: Add submit queue queries Add the capability to query information from a submit queue. The first available parameter is for querying the number of GPU faults (hangs) that can be attributed to the queue. This is useful for implementing context robustness. A user context can regularly query the number of faults to see if it is responsible for any and if so it can invalidate itself. This is also helpful for testing by confirming to the user driver if a particular command stream caused a fault (or not as the case may be). Signed-off-by: Jordan Crouse Signed-off-by: Rob Clark --- drivers/gpu/drm/msm/msm_drv.c | 9 +++++- drivers/gpu/drm/msm/msm_drv.h | 2 ++ drivers/gpu/drm/msm/msm_gpu.c | 3 +- drivers/gpu/drm/msm/msm_submitqueue.c | 41 +++++++++++++++++++++++++++ include/uapi/drm/msm_drm.h | 12 ++++++++ 5 files changed, 65 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/msm/msm_drv.c b/drivers/gpu/drm/msm/msm_drv.c index 0bdd93648761..37f18b55d11f 100644 --- a/drivers/gpu/drm/msm/msm_drv.c +++ b/drivers/gpu/drm/msm/msm_drv.c @@ -39,9 +39,10 @@ * MSM_GEM_INFO ioctl. * - 1.4.0 - softpin, MSM_RELOC_BO_DUMP, and GEM_INFO support to set/get * GEM object's debug name + * - 1.5.0 - Add SUBMITQUERY_QUERY ioctl */ #define MSM_VERSION_MAJOR 1 -#define MSM_VERSION_MINOR 4 +#define MSM_VERSION_MINOR 5 #define MSM_VERSION_PATCHLEVEL 0 static const struct drm_mode_config_funcs mode_config_funcs = { @@ -964,6 +965,11 @@ static int msm_ioctl_submitqueue_new(struct drm_device *dev, void *data, args->flags, &args->id); } +static int msm_ioctl_submitqueue_query(struct drm_device *dev, void *data, + struct drm_file *file) +{ + return msm_submitqueue_query(dev, file->driver_priv, data); +} static int msm_ioctl_submitqueue_close(struct drm_device *dev, void *data, struct drm_file *file) @@ -984,6 +990,7 @@ static const struct drm_ioctl_desc msm_ioctls[] = { DRM_IOCTL_DEF_DRV(MSM_GEM_MADVISE, msm_ioctl_gem_madvise, DRM_AUTH|DRM_RENDER_ALLOW), DRM_IOCTL_DEF_DRV(MSM_SUBMITQUEUE_NEW, msm_ioctl_submitqueue_new, DRM_AUTH|DRM_RENDER_ALLOW), DRM_IOCTL_DEF_DRV(MSM_SUBMITQUEUE_CLOSE, msm_ioctl_submitqueue_close, DRM_AUTH|DRM_RENDER_ALLOW), + DRM_IOCTL_DEF_DRV(MSM_SUBMITQUEUE_QUERY, msm_ioctl_submitqueue_query, DRM_AUTH|DRM_RENDER_ALLOW), }; static const struct vm_operations_struct vm_ops = { diff --git a/drivers/gpu/drm/msm/msm_drv.h b/drivers/gpu/drm/msm/msm_drv.h index c56dade2c1dc..f09d12ddb3ac 100644 --- a/drivers/gpu/drm/msm/msm_drv.h +++ b/drivers/gpu/drm/msm/msm_drv.h @@ -420,6 +420,8 @@ struct msm_gpu_submitqueue *msm_submitqueue_get(struct msm_file_private *ctx, u32 id); int msm_submitqueue_create(struct drm_device *drm, struct msm_file_private *ctx, u32 prio, u32 flags, u32 *id); +int msm_submitqueue_query(struct drm_device *drm, struct msm_file_private *ctx, + struct drm_msm_submitqueue_query *args); int msm_submitqueue_remove(struct msm_file_private *ctx, u32 id); void msm_submitqueue_close(struct msm_file_private *ctx); diff --git a/drivers/gpu/drm/msm/msm_gpu.c b/drivers/gpu/drm/msm/msm_gpu.c index 194847a220b6..2b76657badd5 100644 --- a/drivers/gpu/drm/msm/msm_gpu.c +++ b/drivers/gpu/drm/msm/msm_gpu.c @@ -443,8 +443,9 @@ static void recover_worker(struct work_struct *work) if (submit) { struct task_struct *task; - /* Increment the fault count */ + /* Increment the fault counts */ gpu->global_faults++; + submit->queue->faults++; task = get_pid_task(submit->pid, PIDTYPE_PID); if (task) { diff --git a/drivers/gpu/drm/msm/msm_submitqueue.c b/drivers/gpu/drm/msm/msm_submitqueue.c index 5115f75b5b7f..f160ec40a39b 100644 --- a/drivers/gpu/drm/msm/msm_submitqueue.c +++ b/drivers/gpu/drm/msm/msm_submitqueue.c @@ -120,6 +120,47 @@ int msm_submitqueue_init(struct drm_device *drm, struct msm_file_private *ctx) return msm_submitqueue_create(drm, ctx, default_prio, 0, NULL); } +static int msm_submitqueue_query_faults(struct msm_gpu_submitqueue *queue, + struct drm_msm_submitqueue_query *args) +{ + size_t size = min_t(size_t, args->len, sizeof(queue->faults)); + int ret; + + /* If a zero length was passed in, return the data size we expect */ + if (!args->len) { + args->len = sizeof(queue->faults); + return 0; + } + + /* Set the length to the actual size of the data */ + args->len = size; + + ret = copy_to_user(u64_to_user_ptr(args->data), &queue->faults, size); + + return ret ? -EFAULT : 0; +} + +int msm_submitqueue_query(struct drm_device *drm, struct msm_file_private *ctx, + struct drm_msm_submitqueue_query *args) +{ + struct msm_gpu_submitqueue *queue; + int ret = -EINVAL; + + if (args->pad) + return -EINVAL; + + queue = msm_submitqueue_get(ctx, args->id); + if (!queue) + return -ENOENT; + + if (args->param == MSM_SUBMITQUEUE_PARAM_FAULTS) + ret = msm_submitqueue_query_faults(queue, args); + + msm_submitqueue_put(queue); + + return ret; +} + int msm_submitqueue_remove(struct msm_file_private *ctx, u32 id) { struct msm_gpu_submitqueue *entry; diff --git a/include/uapi/drm/msm_drm.h b/include/uapi/drm/msm_drm.h index 178d7b407f3a..0b85ed6a3710 100644 --- a/include/uapi/drm/msm_drm.h +++ b/include/uapi/drm/msm_drm.h @@ -288,6 +288,16 @@ struct drm_msm_submitqueue { __u32 id; /* out, identifier */ }; +#define MSM_SUBMITQUEUE_PARAM_FAULTS 0 + +struct drm_msm_submitqueue_query { + __u64 data; + __u32 id; + __u32 param; + __u32 len; + __u32 pad; +}; + #define DRM_MSM_GET_PARAM 0x00 /* placeholder: #define DRM_MSM_SET_PARAM 0x01 @@ -304,6 +314,7 @@ struct drm_msm_submitqueue { */ #define DRM_MSM_SUBMITQUEUE_NEW 0x0A #define DRM_MSM_SUBMITQUEUE_CLOSE 0x0B +#define DRM_MSM_SUBMITQUEUE_QUERY 0x0C #define DRM_IOCTL_MSM_GET_PARAM DRM_IOWR(DRM_COMMAND_BASE + DRM_MSM_GET_PARAM, struct drm_msm_param) #define DRM_IOCTL_MSM_GEM_NEW DRM_IOWR(DRM_COMMAND_BASE + DRM_MSM_GEM_NEW, struct drm_msm_gem_new) @@ -315,6 +326,7 @@ struct drm_msm_submitqueue { #define DRM_IOCTL_MSM_GEM_MADVISE DRM_IOWR(DRM_COMMAND_BASE + DRM_MSM_GEM_MADVISE, struct drm_msm_gem_madvise) #define DRM_IOCTL_MSM_SUBMITQUEUE_NEW DRM_IOWR(DRM_COMMAND_BASE + DRM_MSM_SUBMITQUEUE_NEW, struct drm_msm_submitqueue) #define DRM_IOCTL_MSM_SUBMITQUEUE_CLOSE DRM_IOW (DRM_COMMAND_BASE + DRM_MSM_SUBMITQUEUE_CLOSE, __u32) +#define DRM_IOCTL_MSM_SUBMITQUEUE_QUERY DRM_IOW (DRM_COMMAND_BASE + DRM_MSM_SUBMITQUEUE_QUERY, struct drm_msm_submitqueue_query) #if defined(__cplusplus) } From b94a6e3737ac7b5c67570cbbd02858b9aec8fbf1 Mon Sep 17 00:00:00 2001 From: Jordan Crouse Date: Mon, 4 Feb 2019 09:15:39 -0700 Subject: [PATCH 21/34] drm/msm/a6xx: Remove unwanted regulator code The GMU code currently has some misguided code to try to work around a hardware quirk that requires the power domains on the GPU be collapsed in a certain order. Upcoming patches will do this the right way so get rid of the unused and unwanted regulator code. Signed-off-by: Jordan Crouse Signed-off-by: Rob Clark --- drivers/gpu/drm/msm/adreno/a6xx_gmu.c | 4 ---- drivers/gpu/drm/msm/adreno/a6xx_gmu.h | 2 -- 2 files changed, 6 deletions(-) diff --git a/drivers/gpu/drm/msm/adreno/a6xx_gmu.c b/drivers/gpu/drm/msm/adreno/a6xx_gmu.c index a12aa4664c8b..4618438886df 100644 --- a/drivers/gpu/drm/msm/adreno/a6xx_gmu.c +++ b/drivers/gpu/drm/msm/adreno/a6xx_gmu.c @@ -681,9 +681,6 @@ int a6xx_gmu_reset(struct a6xx_gpu *a6xx_gpu) gmu_poll_timeout(gmu, REG_A6XX_RSCC_TCS3_DRV0_STATUS, val, (val & 1), 100, 1000); - /* Force off the GX GSDC */ - regulator_force_disable(gmu->gx); - /* Disable the resources */ clk_bulk_disable_unprepare(gmu->nr_clocks, gmu->clocks); pm_runtime_put_sync(gmu->dev); @@ -1218,7 +1215,6 @@ int a6xx_gmu_probe(struct a6xx_gpu *a6xx_gpu, struct device_node *node) gmu->idle_level = GMU_IDLE_STATE_ACTIVE; pm_runtime_enable(gmu->dev); - gmu->gx = devm_regulator_get(gmu->dev, "vdd"); /* Get the list of clocks */ ret = a6xx_gmu_clocks_probe(gmu); diff --git a/drivers/gpu/drm/msm/adreno/a6xx_gmu.h b/drivers/gpu/drm/msm/adreno/a6xx_gmu.h index c721d9165d8e..8081083cd062 100644 --- a/drivers/gpu/drm/msm/adreno/a6xx_gmu.h +++ b/drivers/gpu/drm/msm/adreno/a6xx_gmu.h @@ -52,8 +52,6 @@ struct a6xx_gmu { int hfi_irq; int gmu_irq; - struct regulator *gx; - struct iommu_domain *domain; u64 uncached_iova_base; From c6c598535c7cce130cc99fc29a1578a844900cd7 Mon Sep 17 00:00:00 2001 From: Jordan Crouse Date: Mon, 4 Feb 2019 09:15:40 -0700 Subject: [PATCH 22/34] dt-bindings: drm/msm/a6xx: Add GX power-domain for GMU bindings The GMU should have two power domains defined: "cx" and "gx". "cx" is the actual power domain for the device and "gx" will be attached at runtime to manage reference counting on the GPU device in case of a GMU crash. Signed-off-by: Jordan Crouse Signed-off-by: Rob Clark --- Documentation/devicetree/bindings/display/msm/gmu.txt | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-) diff --git a/Documentation/devicetree/bindings/display/msm/gmu.txt b/Documentation/devicetree/bindings/display/msm/gmu.txt index 3439b38e60f2..90af5b0a56a9 100644 --- a/Documentation/devicetree/bindings/display/msm/gmu.txt +++ b/Documentation/devicetree/bindings/display/msm/gmu.txt @@ -24,7 +24,10 @@ Required properties: * "cxo" * "axi" * "mnoc" -- power-domains: should be <&clock_gpucc GPU_CX_GDSC> +- power-domains: should be: + <&clock_gpucc GPU_CX_GDSC> + <&clock_gpucc GPU_GX_GDSC> +- power-domain-names: Matching names for the power domains - iommus: phandle to the adreno iommu - operating-points-v2: phandle to the OPP operating points @@ -51,7 +54,10 @@ Example: <&gcc GCC_GPU_MEMNOC_GFX_CLK>; clock-names = "gmu", "cxo", "axi", "memnoc"; - power-domains = <&gpucc GPU_CX_GDSC>; + power-domains = <&gpucc GPU_CX_GDSC>, + <&gpucc GPU_GX_GDSC>; + power-domain-names = "cx", "gx"; + iommus = <&adreno_smmu 5>; operating-points-v2 = <&gmu_opp_table>; From 9325d4266afd4e54ff4aae0018d420d596964126 Mon Sep 17 00:00:00 2001 From: Jordan Crouse Date: Mon, 4 Feb 2019 09:15:41 -0700 Subject: [PATCH 23/34] drm/msm/gpu: Attach to the GPU GX power domain 99.999% of the time during normal operation the GMU is responsible for power and clock control on the GX domain and the CPU remains blissfully unaware. However, there is one situation where the CPU needs to get involved: The power sequencing rules dictate that the GX needs to be turned off before the CX so that the CX can be turned on before the GX during power up. During normal operation when the CPU is taking down the CX domain a stop command is sent to the GMU which turns off the GX domain and then the CPU handles the CX domain. But if the GMU happened to be unresponsive while the GX domain was left then the CPU will need to step in and turn off the GX domain before resetting the CX and rebooting the GMU. This unfortunately means that the CPU needs to be marginally aware of the GX domain even though it is expected to usually keep its hands off. To support this we create a semi-disabled GX power domain that does nothing to the hardware on power up but tries to shut it down normally on power down. In this method the reference counting is correct and we can step in with the pm_runtime_put() at the right time during the failure path. This patch sets up the connection to the GX power domain and does the magic to "enable" and disable it at the right points. Signed-off-by: Jordan Crouse Signed-off-by: Rob Clark --- drivers/gpu/drm/msm/adreno/a6xx_gmu.c | 41 ++++++++++++++++++++++++++- drivers/gpu/drm/msm/adreno/a6xx_gmu.h | 2 ++ 2 files changed, 42 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/msm/adreno/a6xx_gmu.c b/drivers/gpu/drm/msm/adreno/a6xx_gmu.c index 4618438886df..f804cfe7407c 100644 --- a/drivers/gpu/drm/msm/adreno/a6xx_gmu.c +++ b/drivers/gpu/drm/msm/adreno/a6xx_gmu.c @@ -3,6 +3,7 @@ #include #include +#include #include #include @@ -681,6 +682,16 @@ int a6xx_gmu_reset(struct a6xx_gpu *a6xx_gpu) gmu_poll_timeout(gmu, REG_A6XX_RSCC_TCS3_DRV0_STATUS, val, (val & 1), 100, 1000); + /* + * Depending on the state of the GMU at this point the GX domain might + * have been left on. Hardware sequencing rules state that the GX has to + * be turned off before the CX domain so this is that one time that + * that calling pm_runtime_put_sync() is expected to do something useful + * (turn off the headswitch) + */ + if (!IS_ERR(gmu->gxpd)) + pm_runtime_put_sync(gmu->gxpd); + /* Disable the resources */ clk_bulk_disable_unprepare(gmu->nr_clocks, gmu->clocks); pm_runtime_put_sync(gmu->dev); @@ -747,6 +758,14 @@ int a6xx_gmu_resume(struct a6xx_gpu *a6xx_gpu) /* Set the GPU to the highest power frequency */ __a6xx_gmu_set_freq(gmu, gmu->nr_gpu_freqs - 1); + /* + * "enable" the GX power domain which won't actually do anything but it + * will make sure that the refcounting is correct in case we need to + * bring down the GX after a GMU failure + */ + if (!IS_ERR(gmu->gxpd)) + pm_runtime_get(gmu->gxpd); + out: /* Make sure to turn off the boot OOB request on error */ if (ret) @@ -823,6 +842,14 @@ int a6xx_gmu_stop(struct a6xx_gpu *a6xx_gpu) /* Remove the bus vote */ icc_set_bw(gpu->icc_path, 0, 0); + /* + * Mark the GPU power domain as off. During the shutdown process the GMU + * should actually turn off the power so this is really just a + * houskeeping step + */ + if (!IS_ERR(gmu->gxpd)) + pm_runtime_put_sync(gmu->gxpd); + clk_bulk_disable_unprepare(gmu->nr_clocks, gmu->clocks); pm_runtime_put_sync(gmu->dev); @@ -1187,9 +1214,15 @@ void a6xx_gmu_remove(struct a6xx_gpu *a6xx_gpu) if (IS_ERR_OR_NULL(gmu->mmio)) return; - pm_runtime_disable(gmu->dev); a6xx_gmu_stop(a6xx_gpu); + pm_runtime_disable(gmu->dev); + + if (!IS_ERR(gmu->gxpd)) { + pm_runtime_disable(gmu->gxpd); + dev_pm_domain_detach(gmu->gxpd, false); + } + a6xx_gmu_irq_disable(gmu); a6xx_gmu_memory_free(gmu, gmu->hfi); @@ -1248,6 +1281,12 @@ int a6xx_gmu_probe(struct a6xx_gpu *a6xx_gpu, struct device_node *node) if (gmu->hfi_irq < 0 || gmu->gmu_irq < 0) goto err; + /* + * Get a link to the GX power domain to reset the GPU in case of GMU + * crash + */ + gmu->gxpd = dev_pm_domain_attach_by_name(gmu->dev, "gx"); + /* Get the power levels for the GMU and GPU */ a6xx_gmu_pwrlevels_probe(gmu); diff --git a/drivers/gpu/drm/msm/adreno/a6xx_gmu.h b/drivers/gpu/drm/msm/adreno/a6xx_gmu.h index 8081083cd062..078d418c8179 100644 --- a/drivers/gpu/drm/msm/adreno/a6xx_gmu.h +++ b/drivers/gpu/drm/msm/adreno/a6xx_gmu.h @@ -55,6 +55,8 @@ struct a6xx_gmu { struct iommu_domain *domain; u64 uncached_iova_base; + struct device *gxpd; + int idle_level; struct a6xx_gmu_bo *hfi; From e31fdb74c183a8cfe32dc31974f3d626b3af4393 Mon Sep 17 00:00:00 2001 From: Jordan Crouse Date: Mon, 4 Feb 2019 09:15:42 -0700 Subject: [PATCH 24/34] drm/msm/a6xx: Make GMU reset useful Now that the GX domain is sorted we can wire up a working GMU reset. IF a GMU hang was detected then try to forcefully shut down the GMU in the power down sequence which should ensure that it can recover normally on the next power up. Signed-off-by: Jordan Crouse Signed-off-by: Rob Clark --- drivers/gpu/drm/msm/adreno/a6xx_gmu.c | 112 ++++++++++++-------------- drivers/gpu/drm/msm/adreno/a6xx_gmu.h | 4 +- drivers/gpu/drm/msm/adreno/a6xx_gpu.c | 2 +- drivers/gpu/drm/msm/adreno/a6xx_gpu.h | 3 +- 4 files changed, 56 insertions(+), 65 deletions(-) diff --git a/drivers/gpu/drm/msm/adreno/a6xx_gmu.c b/drivers/gpu/drm/msm/adreno/a6xx_gmu.c index f804cfe7407c..b3bc2a5c28e6 100644 --- a/drivers/gpu/drm/msm/adreno/a6xx_gmu.c +++ b/drivers/gpu/drm/msm/adreno/a6xx_gmu.c @@ -10,6 +10,24 @@ #include "a6xx_gpu.h" #include "a6xx_gmu.xml.h" +static void a6xx_gmu_fault(struct a6xx_gmu *gmu) +{ + struct a6xx_gpu *a6xx_gpu = container_of(gmu, struct a6xx_gpu, gmu); + struct adreno_gpu *adreno_gpu = &a6xx_gpu->base; + struct msm_gpu *gpu = &adreno_gpu->base; + struct drm_device *dev = gpu->dev; + struct msm_drm_private *priv = dev->dev_private; + + /* FIXME: add a banner here */ + gmu->hung = true; + + /* Turn off the hangcheck timer while we are resetting */ + del_timer(&gpu->hangcheck_timer); + + /* Queue the GPU handler because we need to treat this as a recovery */ + queue_work(priv->wq, &gpu->recover_work); +} + static irqreturn_t a6xx_gmu_irq(int irq, void *data) { struct a6xx_gmu *gmu = data; @@ -21,8 +39,7 @@ static irqreturn_t a6xx_gmu_irq(int irq, void *data) if (status & A6XX_GMU_AO_HOST_INTERRUPT_STATUS_WDOG_BITE) { dev_err_ratelimited(gmu->dev, "GMU watchdog expired\n"); - /* Temporary until we can recover safely */ - BUG(); + a6xx_gmu_fault(gmu); } if (status & A6XX_GMU_AO_HOST_INTERRUPT_STATUS_HOST_AHB_BUS_ERROR) @@ -46,8 +63,7 @@ static irqreturn_t a6xx_hfi_irq(int irq, void *data) if (status & A6XX_GMU_GMU2HOST_INTR_INFO_CM3_FAULT) { dev_err_ratelimited(gmu->dev, "GMU firmware fault\n"); - /* Temporary until we can recover safely */ - BUG(); + a6xx_gmu_fault(gmu); } return IRQ_HANDLED; @@ -166,10 +182,8 @@ static bool a6xx_gmu_check_idle_level(struct a6xx_gmu *gmu) } /* Wait for the GMU to get to its most idle state */ -int a6xx_gmu_wait_for_idle(struct a6xx_gpu *a6xx_gpu) +int a6xx_gmu_wait_for_idle(struct a6xx_gmu *gmu) { - struct a6xx_gmu *gmu = &a6xx_gpu->gmu; - return spin_until(a6xx_gmu_check_idle_level(gmu)); } @@ -568,7 +582,7 @@ static int a6xx_gmu_fw_start(struct a6xx_gmu *gmu, unsigned int state) if (!rpmh_init) { a6xx_gmu_rpmh_init(gmu); rpmh_init = true; - } else if (state != GMU_RESET) { + } else { ret = a6xx_rpmh_start(gmu); if (ret) return ret; @@ -657,10 +671,9 @@ static void a6xx_gmu_irq_disable(struct a6xx_gmu *gmu) gmu_write(gmu, REG_A6XX_GMU_GMU2HOST_INTR_MASK, ~0); } -int a6xx_gmu_reset(struct a6xx_gpu *a6xx_gpu) +/* Force the GMU off in case it isn't responsive */ +static void a6xx_gmu_force_off(struct a6xx_gmu *gmu) { - struct a6xx_gmu *gmu = &a6xx_gpu->gmu; - int ret; u32 val; /* Flush all the queues */ @@ -681,44 +694,6 @@ int a6xx_gmu_reset(struct a6xx_gpu *a6xx_gpu) (val & 1), 100, 10000); gmu_poll_timeout(gmu, REG_A6XX_RSCC_TCS3_DRV0_STATUS, val, (val & 1), 100, 1000); - - /* - * Depending on the state of the GMU at this point the GX domain might - * have been left on. Hardware sequencing rules state that the GX has to - * be turned off before the CX domain so this is that one time that - * that calling pm_runtime_put_sync() is expected to do something useful - * (turn off the headswitch) - */ - if (!IS_ERR(gmu->gxpd)) - pm_runtime_put_sync(gmu->gxpd); - - /* Disable the resources */ - clk_bulk_disable_unprepare(gmu->nr_clocks, gmu->clocks); - pm_runtime_put_sync(gmu->dev); - - /* Re-enable the resources */ - pm_runtime_get_sync(gmu->dev); - - /* Use a known rate to bring up the GMU */ - clk_set_rate(gmu->core_clk, 200000000); - ret = clk_bulk_prepare_enable(gmu->nr_clocks, gmu->clocks); - if (ret) - goto out; - - a6xx_gmu_irq_enable(gmu); - - ret = a6xx_gmu_fw_start(gmu, GMU_RESET); - if (!ret) - ret = a6xx_hfi_start(gmu, GMU_COLD_BOOT); - - /* Set the GPU back to the highest power frequency */ - __a6xx_gmu_set_freq(gmu, gmu->nr_gpu_freqs - 1); - -out: - if (ret) - a6xx_gmu_clear_oob(gmu, GMU_OOB_BOOT_SLUMBER); - - return ret; } int a6xx_gmu_resume(struct a6xx_gpu *a6xx_gpu) @@ -731,6 +706,8 @@ int a6xx_gmu_resume(struct a6xx_gpu *a6xx_gpu) if (WARN(!gmu->mmio, "The GMU is not set up yet\n")) return 0; + gmu->hung = false; + /* Turn on the resources */ pm_runtime_get_sync(gmu->dev); @@ -789,11 +766,9 @@ bool a6xx_gmu_isidle(struct a6xx_gmu *gmu) return true; } -int a6xx_gmu_stop(struct a6xx_gpu *a6xx_gpu) +/* Gracefully try to shut down the GMU and by extension the GPU */ +static void a6xx_gmu_shutdown(struct a6xx_gmu *gmu) { - struct adreno_gpu *adreno_gpu = &a6xx_gpu->base; - struct msm_gpu *gpu = &adreno_gpu->base; - struct a6xx_gmu *gmu = &a6xx_gpu->gmu; u32 val; /* @@ -803,10 +778,13 @@ int a6xx_gmu_stop(struct a6xx_gpu *a6xx_gpu) val = gmu_read(gmu, REG_A6XX_GPU_GMU_CX_GMU_RPMH_POWER_STATE); if (val != 0xf) { - int ret = a6xx_gmu_wait_for_idle(a6xx_gpu); + int ret = a6xx_gmu_wait_for_idle(gmu); - /* Temporary until we can recover safely */ - BUG_ON(ret); + /* If the GMU isn't responding assume it is hung */ + if (ret) { + a6xx_gmu_force_off(gmu); + return; + } /* tell the GMU we want to slumber */ a6xx_gmu_notify_slumber(gmu); @@ -838,14 +816,30 @@ int a6xx_gmu_stop(struct a6xx_gpu *a6xx_gpu) /* Tell RPMh to power off the GPU */ a6xx_rpmh_stop(gmu); +} + + +int a6xx_gmu_stop(struct a6xx_gpu *a6xx_gpu) +{ + struct a6xx_gmu *gmu = &a6xx_gpu->gmu; + struct msm_gpu *gpu = &a6xx_gpu->base.base; + + /* + * Force the GMU off if we detected a hang, otherwise try to shut it + * down gracefully + */ + if (gmu->hung) + a6xx_gmu_force_off(gmu); + else + a6xx_gmu_shutdown(gmu); /* Remove the bus vote */ icc_set_bw(gpu->icc_path, 0, 0); /* - * Mark the GPU power domain as off. During the shutdown process the GMU - * should actually turn off the power so this is really just a - * houskeeping step + * Make sure the GX domain is off before turning off the GMU (CX) + * domain. Usually the GMU does this but only if the shutdown sequence + * was successful */ if (!IS_ERR(gmu->gxpd)) pm_runtime_put_sync(gmu->gxpd); diff --git a/drivers/gpu/drm/msm/adreno/a6xx_gmu.h b/drivers/gpu/drm/msm/adreno/a6xx_gmu.h index 078d418c8179..c5b1887f259f 100644 --- a/drivers/gpu/drm/msm/adreno/a6xx_gmu.h +++ b/drivers/gpu/drm/msm/adreno/a6xx_gmu.h @@ -27,9 +27,6 @@ struct a6xx_gmu_bo { /* the GMU is coming up for the first time or back from a power collapse */ #define GMU_COLD_BOOT 1 -/* The GMU is being soft reset after a fault */ -#define GMU_RESET 2 - /* * These define the level of control that the GMU has - the higher the number * the more things that the GMU hardware controls on its own. @@ -79,6 +76,7 @@ struct a6xx_gmu { struct a6xx_hfi_queue queues[2]; struct tasklet_struct hfi_tasklet; + bool hung; }; static inline u32 gmu_read(struct a6xx_gmu *gmu, u32 offset) diff --git a/drivers/gpu/drm/msm/adreno/a6xx_gpu.c b/drivers/gpu/drm/msm/adreno/a6xx_gpu.c index fefe773c989e..f76d8cd06f93 100644 --- a/drivers/gpu/drm/msm/adreno/a6xx_gpu.c +++ b/drivers/gpu/drm/msm/adreno/a6xx_gpu.c @@ -698,7 +698,7 @@ static int a6xx_pm_suspend(struct msm_gpu *gpu) * Make sure the GMU is idle before continuing (because some transitions * may use VBIF */ - a6xx_gmu_wait_for_idle(a6xx_gpu); + a6xx_gmu_wait_for_idle(&a6xx_gpu->gmu); /* Clear the VBIF pipe before shutting down */ /* FIXME: This accesses the GPU - do we need to make sure it is on? */ diff --git a/drivers/gpu/drm/msm/adreno/a6xx_gpu.h b/drivers/gpu/drm/msm/adreno/a6xx_gpu.h index 528a4cfe07cd..b46279eb18c5 100644 --- a/drivers/gpu/drm/msm/adreno/a6xx_gpu.h +++ b/drivers/gpu/drm/msm/adreno/a6xx_gpu.h @@ -46,9 +46,8 @@ struct a6xx_gpu { int a6xx_gmu_resume(struct a6xx_gpu *gpu); int a6xx_gmu_stop(struct a6xx_gpu *gpu); -int a6xx_gmu_wait_for_idle(struct a6xx_gpu *gpu); +int a6xx_gmu_wait_for_idle(struct a6xx_gmu *gmu); -int a6xx_gmu_reset(struct a6xx_gpu *a6xx_gpu); bool a6xx_gmu_isidle(struct a6xx_gmu *gmu); int a6xx_gmu_set_oob(struct a6xx_gmu *gmu, enum a6xx_gmu_oob_state state); From 41570b747cf3ae8fcd1cb0ac954cb1b3240af9bd Mon Sep 17 00:00:00 2001 From: Jordan Crouse Date: Mon, 4 Feb 2019 09:15:43 -0700 Subject: [PATCH 25/34] msm/drm/a6xx: Turn off the GMU if resume fails Currently if the GMU resume function fails all we try to do is clear the BOOT_SLUMBER oob which usually times out and ends up in a cycle of death. If the resume function fails at any point remove any RPMh votes that might have been added and try to shut down the GMU hardware cleanly. Signed-off-by: Jordan Crouse Signed-off-by: Rob Clark --- drivers/gpu/drm/msm/adreno/a6xx_gmu.c | 84 ++++++++++++++-------- drivers/gpu/drm/msm/adreno/a6xx_gpu.c | 20 ++---- drivers/gpu/drm/msm/adreno/adreno_device.c | 1 + 3 files changed, 59 insertions(+), 46 deletions(-) diff --git a/drivers/gpu/drm/msm/adreno/a6xx_gmu.c b/drivers/gpu/drm/msm/adreno/a6xx_gmu.c index b3bc2a5c28e6..9155dafae2a9 100644 --- a/drivers/gpu/drm/msm/adreno/a6xx_gmu.c +++ b/drivers/gpu/drm/msm/adreno/a6xx_gmu.c @@ -648,20 +648,6 @@ static int a6xx_gmu_fw_start(struct a6xx_gmu *gmu, unsigned int state) A6XX_GMU_AO_HOST_INTERRUPT_STATUS_HOST_AHB_BUS_ERROR | \ A6XX_GMU_AO_HOST_INTERRUPT_STATUS_FENCE_ERR) -static void a6xx_gmu_irq_enable(struct a6xx_gmu *gmu) -{ - gmu_write(gmu, REG_A6XX_GMU_AO_HOST_INTERRUPT_CLR, ~0); - gmu_write(gmu, REG_A6XX_GMU_GMU2HOST_INTR_CLR, ~0); - - gmu_write(gmu, REG_A6XX_GMU_AO_HOST_INTERRUPT_MASK, - ~A6XX_GMU_IRQ_MASK); - gmu_write(gmu, REG_A6XX_GMU_GMU2HOST_INTR_MASK, - ~A6XX_HFI_IRQ_MASK); - - enable_irq(gmu->gmu_irq); - enable_irq(gmu->hfi_irq); -} - static void a6xx_gmu_irq_disable(struct a6xx_gmu *gmu) { disable_irq(gmu->gmu_irq); @@ -671,20 +657,10 @@ static void a6xx_gmu_irq_disable(struct a6xx_gmu *gmu) gmu_write(gmu, REG_A6XX_GMU_GMU2HOST_INTR_MASK, ~0); } -/* Force the GMU off in case it isn't responsive */ -static void a6xx_gmu_force_off(struct a6xx_gmu *gmu) +static void a6xx_gmu_rpmh_off(struct a6xx_gmu *gmu) { u32 val; - /* Flush all the queues */ - a6xx_hfi_stop(gmu); - - /* Stop the interrupts */ - a6xx_gmu_irq_disable(gmu); - - /* Force off SPTP in case the GMU is managing it */ - a6xx_sptprac_disable(gmu); - /* Make sure there are no outstanding RPMh votes */ gmu_poll_timeout(gmu, REG_A6XX_RSCC_TCS0_DRV0_STATUS, val, (val & 1), 100, 10000); @@ -696,6 +672,22 @@ static void a6xx_gmu_force_off(struct a6xx_gmu *gmu) (val & 1), 100, 1000); } +/* Force the GMU off in case it isn't responsive */ +static void a6xx_gmu_force_off(struct a6xx_gmu *gmu) +{ + /* Flush all the queues */ + a6xx_hfi_stop(gmu); + + /* Stop the interrupts */ + a6xx_gmu_irq_disable(gmu); + + /* Force off SPTP in case the GMU is managing it */ + a6xx_sptprac_disable(gmu); + + /* Make sure there are no outstanding RPMh votes */ + a6xx_gmu_rpmh_off(gmu); +} + int a6xx_gmu_resume(struct a6xx_gpu *a6xx_gpu) { struct adreno_gpu *adreno_gpu = &a6xx_gpu->base; @@ -714,13 +706,18 @@ int a6xx_gmu_resume(struct a6xx_gpu *a6xx_gpu) /* Use a known rate to bring up the GMU */ clk_set_rate(gmu->core_clk, 200000000); ret = clk_bulk_prepare_enable(gmu->nr_clocks, gmu->clocks); - if (ret) - goto out; + if (ret) { + pm_runtime_put(gmu->dev); + return ret; + } /* Set the bus quota to a reasonable value for boot */ icc_set_bw(gpu->icc_path, 0, MBps_to_icc(3072)); - a6xx_gmu_irq_enable(gmu); + /* Enable the GMU interrupt */ + gmu_write(gmu, REG_A6XX_GMU_AO_HOST_INTERRUPT_CLR, ~0); + gmu_write(gmu, REG_A6XX_GMU_AO_HOST_INTERRUPT_MASK, ~A6XX_GMU_IRQ_MASK); + enable_irq(gmu->gmu_irq); /* Check to see if we are doing a cold or warm boot */ status = gmu_read(gmu, REG_A6XX_GMU_GENERAL_7) == 1 ? @@ -731,6 +728,16 @@ int a6xx_gmu_resume(struct a6xx_gpu *a6xx_gpu) goto out; ret = a6xx_hfi_start(gmu, status); + if (ret) + goto out; + + /* + * Turn on the GMU firmware fault interrupt after we know the boot + * sequence is successful + */ + gmu_write(gmu, REG_A6XX_GMU_GMU2HOST_INTR_CLR, ~0); + gmu_write(gmu, REG_A6XX_GMU_GMU2HOST_INTR_MASK, ~A6XX_HFI_IRQ_MASK); + enable_irq(gmu->hfi_irq); /* Set the GPU to the highest power frequency */ __a6xx_gmu_set_freq(gmu, gmu->nr_gpu_freqs - 1); @@ -744,9 +751,12 @@ int a6xx_gmu_resume(struct a6xx_gpu *a6xx_gpu) pm_runtime_get(gmu->gxpd); out: - /* Make sure to turn off the boot OOB request on error */ - if (ret) - a6xx_gmu_clear_oob(gmu, GMU_OOB_BOOT_SLUMBER); + /* On failure, shut down the GMU to leave it in a good state */ + if (ret) { + disable_irq(gmu->gmu_irq); + a6xx_rpmh_stop(gmu); + pm_runtime_put(gmu->dev); + } return ret; } @@ -769,6 +779,9 @@ bool a6xx_gmu_isidle(struct a6xx_gmu *gmu) /* Gracefully try to shut down the GMU and by extension the GPU */ static void a6xx_gmu_shutdown(struct a6xx_gmu *gmu) { + struct a6xx_gpu *a6xx_gpu = container_of(gmu, struct a6xx_gpu, gmu); + struct adreno_gpu *adreno_gpu = &a6xx_gpu->base; + struct msm_gpu *gpu = &adreno_gpu->base; u32 val; /* @@ -786,6 +799,12 @@ static void a6xx_gmu_shutdown(struct a6xx_gmu *gmu) return; } + /* Clear the VBIF pipe before shutting down */ + gpu_write(gpu, REG_A6XX_VBIF_XIN_HALT_CTRL0, 0xf); + spin_until((gpu_read(gpu, REG_A6XX_VBIF_XIN_HALT_CTRL1) & 0xf) + == 0xf); + gpu_write(gpu, REG_A6XX_VBIF_XIN_HALT_CTRL0, 0); + /* tell the GMU we want to slumber */ a6xx_gmu_notify_slumber(gmu); @@ -824,6 +843,9 @@ int a6xx_gmu_stop(struct a6xx_gpu *a6xx_gpu) struct a6xx_gmu *gmu = &a6xx_gpu->gmu; struct msm_gpu *gpu = &a6xx_gpu->base.base; + if (!pm_runtime_active(gmu->dev)) + return 0; + /* * Force the GMU off if we detected a hang, otherwise try to shut it * down gracefully diff --git a/drivers/gpu/drm/msm/adreno/a6xx_gpu.c b/drivers/gpu/drm/msm/adreno/a6xx_gpu.c index f76d8cd06f93..576559a3a523 100644 --- a/drivers/gpu/drm/msm/adreno/a6xx_gpu.c +++ b/drivers/gpu/drm/msm/adreno/a6xx_gpu.c @@ -678,13 +678,15 @@ static int a6xx_pm_resume(struct msm_gpu *gpu) struct a6xx_gpu *a6xx_gpu = to_a6xx_gpu(adreno_gpu); int ret; - ret = a6xx_gmu_resume(a6xx_gpu); - gpu->needs_hw_init = true; + ret = a6xx_gmu_resume(a6xx_gpu); + if (ret) + return ret; + msm_gpu_resume_devfreq(gpu); - return ret; + return 0; } static int a6xx_pm_suspend(struct msm_gpu *gpu) @@ -694,18 +696,6 @@ static int a6xx_pm_suspend(struct msm_gpu *gpu) devfreq_suspend_device(gpu->devfreq.devfreq); - /* - * Make sure the GMU is idle before continuing (because some transitions - * may use VBIF - */ - a6xx_gmu_wait_for_idle(&a6xx_gpu->gmu); - - /* Clear the VBIF pipe before shutting down */ - /* FIXME: This accesses the GPU - do we need to make sure it is on? */ - gpu_write(gpu, REG_A6XX_VBIF_XIN_HALT_CTRL0, 0xf); - spin_until((gpu_read(gpu, REG_A6XX_VBIF_XIN_HALT_CTRL1) & 0xf) == 0xf); - gpu_write(gpu, REG_A6XX_VBIF_XIN_HALT_CTRL0, 0); - return a6xx_gmu_stop(a6xx_gpu); } diff --git a/drivers/gpu/drm/msm/adreno/adreno_device.c b/drivers/gpu/drm/msm/adreno/adreno_device.c index 714ed6505e47..0d87db75b710 100644 --- a/drivers/gpu/drm/msm/adreno/adreno_device.c +++ b/drivers/gpu/drm/msm/adreno/adreno_device.c @@ -229,6 +229,7 @@ struct msm_gpu *adreno_load_gpu(struct drm_device *dev) ret = pm_runtime_get_sync(&pdev->dev); if (ret < 0) { + pm_runtime_put_sync(&pdev->dev); DRM_DEV_ERROR(dev->dev, "Couldn't power up the GPU: %d\n", ret); return NULL; } From fb076b15d50dee36c75eb6c447c1b1026ad12e2e Mon Sep 17 00:00:00 2001 From: Jordan Crouse Date: Mon, 4 Feb 2019 09:15:44 -0700 Subject: [PATCH 26/34] drm/msm/a6xx: Remove an unused struct member The HFI tasklet was removed in df0dff1 ("drm/msm/a6xx: Poll for HFI responses") but the tasklet_struct was accidentally left behind. Signed-off-by: Jordan Crouse Signed-off-by: Rob Clark --- drivers/gpu/drm/msm/adreno/a6xx_gmu.h | 1 - 1 file changed, 1 deletion(-) diff --git a/drivers/gpu/drm/msm/adreno/a6xx_gmu.h b/drivers/gpu/drm/msm/adreno/a6xx_gmu.h index c5b1887f259f..bedd8e6a63aa 100644 --- a/drivers/gpu/drm/msm/adreno/a6xx_gmu.h +++ b/drivers/gpu/drm/msm/adreno/a6xx_gmu.h @@ -75,7 +75,6 @@ struct a6xx_gmu { struct a6xx_hfi_queue queues[2]; - struct tasklet_struct hfi_tasklet; bool hung; }; From 48e7f18392c66f9b69ebac11c54f1a2e033ced54 Mon Sep 17 00:00:00 2001 From: "Kristian H. Kristensen" Date: Wed, 20 Mar 2019 10:09:08 -0700 Subject: [PATCH 27/34] drm/msm: Implement .gem_free_object_unlocked We use a llist and a worker to delay the object cleanup. This avoids taking mmap_sem and struct_mutex in the wrong order when calling drm_gem_object_put_unlocked() from drm_gem_mmap(). Fixes lockdep problem with copy_from_user() in msm_ioctl_gem_submit(). Signed-off-by: Kristian H. Kristensen Signed-off-by: Rob Clark --- drivers/gpu/drm/msm/msm_drv.c | 5 ++++- drivers/gpu/drm/msm/msm_drv.h | 5 +++++ drivers/gpu/drm/msm/msm_gem.c | 35 ++++++++++++++++++++++++++++++++++- drivers/gpu/drm/msm/msm_gem.h | 3 +++ 4 files changed, 46 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/msm/msm_drv.c b/drivers/gpu/drm/msm/msm_drv.c index 37f18b55d11f..154dab0775f3 100644 --- a/drivers/gpu/drm/msm/msm_drv.c +++ b/drivers/gpu/drm/msm/msm_drv.c @@ -458,6 +458,9 @@ static int msm_drm_init(struct device *dev, struct drm_driver *drv) priv->wq = alloc_ordered_workqueue("msm", 0); + INIT_WORK(&priv->free_work, msm_gem_free_work); + init_llist_head(&priv->free_list); + INIT_LIST_HEAD(&priv->inactive_list); drm_mode_config_init(ddev); @@ -1026,7 +1029,7 @@ static struct drm_driver msm_driver = { .irq_uninstall = msm_irq_uninstall, .enable_vblank = msm_enable_vblank, .disable_vblank = msm_disable_vblank, - .gem_free_object = msm_gem_free_object, + .gem_free_object_unlocked = msm_gem_free_object, .gem_vm_ops = &vm_ops, .dumb_create = msm_gem_dumb_create, .dumb_map_offset = msm_gem_dumb_map_offset, diff --git a/drivers/gpu/drm/msm/msm_drv.h b/drivers/gpu/drm/msm/msm_drv.h index f09d12ddb3ac..bbf739d85381 100644 --- a/drivers/gpu/drm/msm/msm_drv.h +++ b/drivers/gpu/drm/msm/msm_drv.h @@ -185,6 +185,10 @@ struct msm_drm_private { /* list of GEM objects: */ struct list_head inactive_list; + /* worker for delayed free of objects: */ + struct work_struct free_work; + struct llist_head free_list; + struct workqueue_struct *wq; unsigned int num_planes; @@ -325,6 +329,7 @@ void msm_gem_kernel_put(struct drm_gem_object *bo, struct msm_gem_address_space *aspace, bool locked); struct drm_gem_object *msm_gem_import(struct drm_device *dev, struct dma_buf *dmabuf, struct sg_table *sgt); +void msm_gem_free_work(struct work_struct *work); __printf(2, 3) void msm_gem_object_set_name(struct drm_gem_object *bo, const char *fmt, ...); diff --git a/drivers/gpu/drm/msm/msm_gem.c b/drivers/gpu/drm/msm/msm_gem.c index 76940a9da980..d4dae0f0238e 100644 --- a/drivers/gpu/drm/msm/msm_gem.c +++ b/drivers/gpu/drm/msm/msm_gem.c @@ -853,8 +853,18 @@ void msm_gem_describe_objects(struct list_head *list, struct seq_file *m) /* don't call directly! Use drm_gem_object_put() and friends */ void msm_gem_free_object(struct drm_gem_object *obj) { - struct drm_device *dev = obj->dev; struct msm_gem_object *msm_obj = to_msm_bo(obj); + struct drm_device *dev = obj->dev; + struct msm_drm_private *priv = dev->dev_private; + + if (llist_add(&msm_obj->freed, &priv->free_list)) + queue_work(priv->wq, &priv->free_work); +} + +static void free_object(struct msm_gem_object *msm_obj) +{ + struct drm_gem_object *obj = &msm_obj->base; + struct drm_device *dev = obj->dev; WARN_ON(!mutex_is_locked(&dev->struct_mutex)); @@ -892,6 +902,29 @@ void msm_gem_free_object(struct drm_gem_object *obj) kfree(msm_obj); } +void msm_gem_free_work(struct work_struct *work) +{ + struct msm_drm_private *priv = + container_of(work, struct msm_drm_private, free_work); + struct drm_device *dev = priv->dev; + struct llist_node *freed; + struct msm_gem_object *msm_obj, *next; + + while ((freed = llist_del_all(&priv->free_list))) { + + mutex_lock(&dev->struct_mutex); + + llist_for_each_entry_safe(msm_obj, next, + freed, freed) + free_object(msm_obj); + + mutex_unlock(&dev->struct_mutex); + + if (need_resched()) + break; + } +} + /* convenience method to construct a GEM buffer object, and userspace handle */ int msm_gem_new_handle(struct drm_device *dev, struct drm_file *file, uint32_t size, uint32_t flags, uint32_t *handle, diff --git a/drivers/gpu/drm/msm/msm_gem.h b/drivers/gpu/drm/msm/msm_gem.h index 2064fac871b8..0617c44d99b0 100644 --- a/drivers/gpu/drm/msm/msm_gem.h +++ b/drivers/gpu/drm/msm/msm_gem.h @@ -84,6 +84,8 @@ struct msm_gem_object { struct list_head vmas; /* list of msm_gem_vma */ + struct llist_node freed; + /* normally (resv == &_resv) except for imported bo's */ struct reservation_object *resv; struct reservation_object _resv; @@ -133,6 +135,7 @@ enum msm_gem_lock { void msm_gem_purge(struct drm_gem_object *obj, enum msm_gem_lock subclass); void msm_gem_vunmap(struct drm_gem_object *obj, enum msm_gem_lock subclass); +void msm_gem_free_work(struct work_struct *work); /* Created per submit-ioctl, to track bo's and cmdstream bufs, etc, * associated with the cmdstream submission for synchronization (and From 8ea274accda92d36bd510a3272adf745c6376d3f Mon Sep 17 00:00:00 2001 From: "Kristian H. Kristensen" Date: Wed, 20 Mar 2019 10:09:09 -0700 Subject: [PATCH 28/34] drm/msm: Stop dropping struct_mutex in recover_worker() Now that we don't have the mmap_sem lock inversion, we don't need to jump through this particular hoop anymore. Signed-off-by: Kristian H. Kristensen Signed-off-by: Rob Clark --- drivers/gpu/drm/msm/msm_gpu.c | 13 ------------- 1 file changed, 13 deletions(-) diff --git a/drivers/gpu/drm/msm/msm_gpu.c b/drivers/gpu/drm/msm/msm_gpu.c index 2b76657badd5..bf4ee2766431 100644 --- a/drivers/gpu/drm/msm/msm_gpu.c +++ b/drivers/gpu/drm/msm/msm_gpu.c @@ -450,21 +450,8 @@ static void recover_worker(struct work_struct *work) task = get_pid_task(submit->pid, PIDTYPE_PID); if (task) { comm = kstrdup(task->comm, GFP_KERNEL); - - /* - * So slightly annoying, in other paths like - * mmap'ing gem buffers, mmap_sem is acquired - * before struct_mutex, which means we can't - * hold struct_mutex across the call to - * get_cmdline(). But submits are retired - * from the same in-order workqueue, so we can - * safely drop the lock here without worrying - * about the submit going away. - */ - mutex_unlock(&dev->struct_mutex); cmd = kstrdup_quotable_cmdline(task, GFP_KERNEL); put_task_struct(task); - mutex_lock(&dev->struct_mutex); } if (comm && cmd) { From b673499ae738e7790b1b9f65e784999e5804abd3 Mon Sep 17 00:00:00 2001 From: "Kristian H. Kristensen" Date: Wed, 20 Mar 2019 10:09:10 -0700 Subject: [PATCH 29/34] drm/msm: Split submit_lookup_objects() into two loops First loop does copy_from_user() without the table lock held and just stores the handle. Second loop looks up buffer objects with the table_lock held without potentially blocking or faulting. This lets us clean up a bunch of custom, non-faulting copy_from_user() code. Signed-off-by: Kristian H. Kristensen Signed-off-by: Rob Clark --- drivers/gpu/drm/msm/msm_gem.h | 5 +++- drivers/gpu/drm/msm/msm_gem_submit.c | 44 +++++++++++----------------- 2 files changed, 21 insertions(+), 28 deletions(-) diff --git a/drivers/gpu/drm/msm/msm_gem.h b/drivers/gpu/drm/msm/msm_gem.h index 0617c44d99b0..c5ac781dffee 100644 --- a/drivers/gpu/drm/msm/msm_gem.h +++ b/drivers/gpu/drm/msm/msm_gem.h @@ -166,7 +166,10 @@ struct msm_gem_submit { } *cmd; /* array of size nr_cmds */ struct { uint32_t flags; - struct msm_gem_object *obj; + union { + struct msm_gem_object *obj; + uint32_t handle; + }; uint64_t iova; } bos[0]; }; diff --git a/drivers/gpu/drm/msm/msm_gem_submit.c b/drivers/gpu/drm/msm/msm_gem_submit.c index 12b983fc0b56..648e0c1b92e6 100644 --- a/drivers/gpu/drm/msm/msm_gem_submit.c +++ b/drivers/gpu/drm/msm/msm_gem_submit.c @@ -74,27 +74,14 @@ void msm_gem_submit_free(struct msm_gem_submit *submit) kfree(submit); } -static inline unsigned long __must_check -copy_from_user_inatomic(void *to, const void __user *from, unsigned long n) -{ - if (access_ok(from, n)) - return __copy_from_user_inatomic(to, from, n); - return -EFAULT; -} - static int submit_lookup_objects(struct msm_gem_submit *submit, struct drm_msm_gem_submit *args, struct drm_file *file) { unsigned i; int ret = 0; - spin_lock(&file->table_lock); - pagefault_disable(); - for (i = 0; i < args->nr_bos; i++) { struct drm_msm_gem_submit_bo submit_bo; - struct drm_gem_object *obj; - struct msm_gem_object *msm_obj; void __user *userptr = u64_to_user_ptr(args->bos + (i * sizeof(submit_bo))); @@ -103,15 +90,10 @@ static int submit_lookup_objects(struct msm_gem_submit *submit, */ submit->bos[i].flags = 0; - if (copy_from_user_inatomic(&submit_bo, userptr, sizeof(submit_bo))) { - pagefault_enable(); - spin_unlock(&file->table_lock); - if (copy_from_user(&submit_bo, userptr, sizeof(submit_bo))) { - ret = -EFAULT; - goto out; - } - spin_lock(&file->table_lock); - pagefault_disable(); + if (copy_from_user(&submit_bo, userptr, sizeof(submit_bo))) { + ret = -EFAULT; + i = 0; + goto out; } /* at least one of READ and/or WRITE flags should be set: */ @@ -121,19 +103,28 @@ static int submit_lookup_objects(struct msm_gem_submit *submit, !(submit_bo.flags & MANDATORY_FLAGS)) { DRM_ERROR("invalid flags: %x\n", submit_bo.flags); ret = -EINVAL; - goto out_unlock; + i = 0; + goto out; } + submit->bos[i].handle = submit_bo.handle; submit->bos[i].flags = submit_bo.flags; /* in validate_objects() we figure out if this is true: */ submit->bos[i].iova = submit_bo.presumed; + } + + spin_lock(&file->table_lock); + + for (i = 0; i < args->nr_bos; i++) { + struct drm_gem_object *obj; + struct msm_gem_object *msm_obj; /* normally use drm_gem_object_lookup(), but for bulk lookup * all under single table_lock just hit object_idr directly: */ - obj = idr_find(&file->object_idr, submit_bo.handle); + obj = idr_find(&file->object_idr, submit->bos[i].handle); if (!obj) { - DRM_ERROR("invalid handle %u at index %u\n", submit_bo.handle, i); + DRM_ERROR("invalid handle %u at index %u\n", submit->bos[i].handle, i); ret = -EINVAL; goto out_unlock; } @@ -142,7 +133,7 @@ static int submit_lookup_objects(struct msm_gem_submit *submit, if (!list_empty(&msm_obj->submit_entry)) { DRM_ERROR("handle %u at index %u already on submit list\n", - submit_bo.handle, i); + submit->bos[i].handle, i); ret = -EINVAL; goto out_unlock; } @@ -155,7 +146,6 @@ static int submit_lookup_objects(struct msm_gem_submit *submit, } out_unlock: - pagefault_enable(); spin_unlock(&file->table_lock); out: From 026ef6354caa661ddd4c399590048e72a9d420bf Mon Sep 17 00:00:00 2001 From: Jordan Crouse Date: Fri, 19 Apr 2019 13:56:27 -0600 Subject: [PATCH 30/34] dt-bindings: drm/msm/a6xx: Document interconnect properties for GPU Add documentation for the interconnect and interconnect-names bindings for the GPU node as detailed by bindings/interconnect/interconnect.txt. Signed-off-by: Jordan Crouse Reviewed-by: Douglas Anderson Reviewed-by: Rob Herring Acked-by: Georgi Djakov Signed-off-by: Rob Clark --- Documentation/devicetree/bindings/display/msm/gpu.txt | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/Documentation/devicetree/bindings/display/msm/gpu.txt b/Documentation/devicetree/bindings/display/msm/gpu.txt index aad1aef682f7..c04614c63ea9 100644 --- a/Documentation/devicetree/bindings/display/msm/gpu.txt +++ b/Documentation/devicetree/bindings/display/msm/gpu.txt @@ -22,6 +22,8 @@ Required properties: - qcom,adreno-630.2 - iommus: optional phandle to an adreno iommu instance - operating-points-v2: optional phandle to the OPP operating points +- interconnects: optional phandle to an interconnect provider. See + ../interconnect/interconnect.txt for details. - qcom,gmu: For GMU attached devices a phandle to the GMU device that will control the power for the GPU. Applicable targets: - qcom,adreno-630.2 @@ -70,6 +72,8 @@ Example a6xx (with GMU): operating-points-v2 = <&gpu_opp_table>; + interconnects = <&rsc_hlos MASTER_GFX3D &rsc_hlos SLAVE_EBI1>; + qcom,gmu = <&gmu>; }; }; From a9e2559c931d7c44ae7d677b12093956ffa864b9 Mon Sep 17 00:00:00 2001 From: Jordan Crouse Date: Fri, 19 Apr 2019 13:46:14 -0600 Subject: [PATCH 31/34] drm/msm/gpu: Move zap shader loading to adreno a5xx and a6xx both share (mostly) the same code to load the zap shader and bring the GPU out of secure mode. Move the formerly 5xx specific code to adreno to make it available for a6xx too. Signed-off-by: Jordan Crouse Signed-off-by: Rob Clark --- drivers/gpu/drm/msm/adreno/a5xx_gpu.c | 111 +------------------ drivers/gpu/drm/msm/adreno/adreno_gpu.c | 135 ++++++++++++++++++++++++ drivers/gpu/drm/msm/adreno/adreno_gpu.h | 6 ++ 3 files changed, 142 insertions(+), 110 deletions(-) diff --git a/drivers/gpu/drm/msm/adreno/a5xx_gpu.c b/drivers/gpu/drm/msm/adreno/a5xx_gpu.c index 270da14cba67..e5fcefa49f19 100644 --- a/drivers/gpu/drm/msm/adreno/a5xx_gpu.c +++ b/drivers/gpu/drm/msm/adreno/a5xx_gpu.c @@ -15,9 +15,6 @@ #include #include #include -#include -#include -#include #include #include #include @@ -30,96 +27,6 @@ static void a5xx_dump(struct msm_gpu *gpu); #define GPU_PAS_ID 13 -static int zap_shader_load_mdt(struct msm_gpu *gpu, const char *fwname) -{ - struct device *dev = &gpu->pdev->dev; - const struct firmware *fw; - struct device_node *np, *mem_np; - struct resource r; - phys_addr_t mem_phys; - ssize_t mem_size; - void *mem_region = NULL; - int ret; - - if (!IS_ENABLED(CONFIG_ARCH_QCOM)) - return -EINVAL; - - np = of_get_child_by_name(dev->of_node, "zap-shader"); - if (!np) - return -ENODEV; - - mem_np = of_parse_phandle(np, "memory-region", 0); - of_node_put(np); - if (!mem_np) - return -EINVAL; - - ret = of_address_to_resource(mem_np, 0, &r); - of_node_put(mem_np); - if (ret) - return ret; - - mem_phys = r.start; - mem_size = resource_size(&r); - - /* Request the MDT file for the firmware */ - fw = adreno_request_fw(to_adreno_gpu(gpu), fwname); - if (IS_ERR(fw)) { - DRM_DEV_ERROR(dev, "Unable to load %s\n", fwname); - return PTR_ERR(fw); - } - - /* Figure out how much memory we need */ - mem_size = qcom_mdt_get_size(fw); - if (mem_size < 0) { - ret = mem_size; - goto out; - } - - /* Allocate memory for the firmware image */ - mem_region = memremap(mem_phys, mem_size, MEMREMAP_WC); - if (!mem_region) { - ret = -ENOMEM; - goto out; - } - - /* - * Load the rest of the MDT - * - * Note that we could be dealing with two different paths, since - * with upstream linux-firmware it would be in a qcom/ subdir.. - * adreno_request_fw() handles this, but qcom_mdt_load() does - * not. But since we've already gotten thru adreno_request_fw() - * we know which of the two cases it is: - */ - if (to_adreno_gpu(gpu)->fwloc == FW_LOCATION_LEGACY) { - ret = qcom_mdt_load(dev, fw, fwname, GPU_PAS_ID, - mem_region, mem_phys, mem_size, NULL); - } else { - char *newname; - - newname = kasprintf(GFP_KERNEL, "qcom/%s", fwname); - - ret = qcom_mdt_load(dev, fw, newname, GPU_PAS_ID, - mem_region, mem_phys, mem_size, NULL); - kfree(newname); - } - if (ret) - goto out; - - /* Send the image to the secure world */ - ret = qcom_scm_pas_auth_and_reset(GPU_PAS_ID); - if (ret) - DRM_DEV_ERROR(dev, "Unable to authorize the image\n"); - -out: - if (mem_region) - memunmap(mem_region); - - release_firmware(fw); - - return ret; -} - static void a5xx_flush(struct msm_gpu *gpu, struct msm_ringbuffer *ring) { struct adreno_gpu *adreno_gpu = to_adreno_gpu(gpu); @@ -565,8 +472,6 @@ static int a5xx_zap_shader_resume(struct msm_gpu *gpu) static int a5xx_zap_shader_init(struct msm_gpu *gpu) { static bool loaded; - struct adreno_gpu *adreno_gpu = to_adreno_gpu(gpu); - struct platform_device *pdev = gpu->pdev; int ret; /* @@ -576,23 +481,9 @@ static int a5xx_zap_shader_init(struct msm_gpu *gpu) if (loaded) return a5xx_zap_shader_resume(gpu); - /* We need SCM to be able to load the firmware */ - if (!qcom_scm_is_available()) { - DRM_DEV_ERROR(&pdev->dev, "SCM is not available\n"); - return -EPROBE_DEFER; - } - - /* Each GPU has a target specific zap shader firmware name to use */ - if (!adreno_gpu->info->zapfw) { - DRM_DEV_ERROR(&pdev->dev, - "Zap shader firmware file not specified for this target\n"); - return -ENODEV; - } - - ret = zap_shader_load_mdt(gpu, adreno_gpu->info->zapfw); + ret = adreno_zap_shader_load(gpu, GPU_PAS_ID); loaded = !ret; - return ret; } diff --git a/drivers/gpu/drm/msm/adreno/adreno_gpu.c b/drivers/gpu/drm/msm/adreno/adreno_gpu.c index 6de08cfc38aa..6f7f4114afcf 100644 --- a/drivers/gpu/drm/msm/adreno/adreno_gpu.c +++ b/drivers/gpu/drm/msm/adreno/adreno_gpu.c @@ -19,13 +19,148 @@ #include #include +#include #include +#include #include #include +#include #include "adreno_gpu.h" #include "msm_gem.h" #include "msm_mmu.h" +static bool zap_available = true; + +static int zap_shader_load_mdt(struct msm_gpu *gpu, const char *fwname, + u32 pasid) +{ + struct device *dev = &gpu->pdev->dev; + const struct firmware *fw; + struct device_node *np, *mem_np; + struct resource r; + phys_addr_t mem_phys; + ssize_t mem_size; + void *mem_region = NULL; + int ret; + + if (!IS_ENABLED(CONFIG_ARCH_QCOM)) { + zap_available = false; + return -EINVAL; + } + + np = of_get_child_by_name(dev->of_node, "zap-shader"); + if (!np) { + zap_available = false; + return -ENODEV; + } + + mem_np = of_parse_phandle(np, "memory-region", 0); + of_node_put(np); + if (!mem_np) { + zap_available = false; + return -EINVAL; + } + + ret = of_address_to_resource(mem_np, 0, &r); + of_node_put(mem_np); + if (ret) + return ret; + + mem_phys = r.start; + mem_size = resource_size(&r); + + /* Request the MDT file for the firmware */ + fw = adreno_request_fw(to_adreno_gpu(gpu), fwname); + if (IS_ERR(fw)) { + DRM_DEV_ERROR(dev, "Unable to load %s\n", fwname); + return PTR_ERR(fw); + } + + /* Figure out how much memory we need */ + mem_size = qcom_mdt_get_size(fw); + if (mem_size < 0) { + ret = mem_size; + goto out; + } + + /* Allocate memory for the firmware image */ + mem_region = memremap(mem_phys, mem_size, MEMREMAP_WC); + if (!mem_region) { + ret = -ENOMEM; + goto out; + } + + /* + * Load the rest of the MDT + * + * Note that we could be dealing with two different paths, since + * with upstream linux-firmware it would be in a qcom/ subdir.. + * adreno_request_fw() handles this, but qcom_mdt_load() does + * not. But since we've already gotten through adreno_request_fw() + * we know which of the two cases it is: + */ + if (to_adreno_gpu(gpu)->fwloc == FW_LOCATION_LEGACY) { + ret = qcom_mdt_load(dev, fw, fwname, pasid, + mem_region, mem_phys, mem_size, NULL); + } else { + char *newname; + + newname = kasprintf(GFP_KERNEL, "qcom/%s", fwname); + + ret = qcom_mdt_load(dev, fw, newname, pasid, + mem_region, mem_phys, mem_size, NULL); + kfree(newname); + } + if (ret) + goto out; + + /* Send the image to the secure world */ + ret = qcom_scm_pas_auth_and_reset(pasid); + + /* + * If the scm call returns -EOPNOTSUPP we assume that this target + * doesn't need/support the zap shader so quietly fail + */ + if (ret == -EOPNOTSUPP) + zap_available = false; + else if (ret) + DRM_DEV_ERROR(dev, "Unable to authorize the image\n"); + +out: + if (mem_region) + memunmap(mem_region); + + release_firmware(fw); + + return ret; +} + +int adreno_zap_shader_load(struct msm_gpu *gpu, u32 pasid) +{ + struct adreno_gpu *adreno_gpu = to_adreno_gpu(gpu); + struct platform_device *pdev = gpu->pdev; + + /* Short cut if we determine the zap shader isn't available/needed */ + if (!zap_available) + return -ENODEV; + + /* We need SCM to be able to load the firmware */ + if (!qcom_scm_is_available()) { + DRM_DEV_ERROR(&pdev->dev, "SCM is not available\n"); + return -EPROBE_DEFER; + } + + /* Each GPU has a target specific zap shader firmware name to use */ + if (!adreno_gpu->info->zapfw) { + zap_available = false; + DRM_DEV_ERROR(&pdev->dev, + "Zap shader firmware file not specified for this target\n"); + return -ENODEV; + } + + return zap_shader_load_mdt(gpu, adreno_gpu->info->zapfw, pasid); +} + int adreno_get_param(struct msm_gpu *gpu, uint32_t param, uint64_t *value) { struct adreno_gpu *adreno_gpu = to_adreno_gpu(gpu); diff --git a/drivers/gpu/drm/msm/adreno/adreno_gpu.h b/drivers/gpu/drm/msm/adreno/adreno_gpu.h index 5db459bc28a7..0925606ec9b5 100644 --- a/drivers/gpu/drm/msm/adreno/adreno_gpu.h +++ b/drivers/gpu/drm/msm/adreno/adreno_gpu.h @@ -252,6 +252,12 @@ void adreno_gpu_state_destroy(struct msm_gpu_state *state); int adreno_gpu_state_get(struct msm_gpu *gpu, struct msm_gpu_state *state); int adreno_gpu_state_put(struct msm_gpu_state *state); +/* + * For a5xx and a6xx targets load the zap shader that is used to pull the GPU + * out of secure mode + */ +int adreno_zap_shader_load(struct msm_gpu *gpu, u32 pasid); + /* ringbuffer helpers (the parts that are adreno specific) */ static inline void From abccb9fe326743b72c8099d5cc039e9038bc50da Mon Sep 17 00:00:00 2001 From: Jordan Crouse Date: Fri, 19 Apr 2019 13:46:15 -0600 Subject: [PATCH 32/34] drm/msm/a6xx: Add zap shader load The a6xx GPU powers on in secure mode which restricts what memory it can write to. To get out of secure mode the GPU driver can write to REG_A6XX_RBBM_SECVID_TRUST_CNTL but on targets that are "secure" that register region is blocked and writes will cause the system to go down. For those targets we need to execute a special sequence that involves loadinga special shader that clears the GPU registers and use a PM4 sequence to pull the GPU out of secure. Add support for loading the zap shader and executing the secure sequence. For targets that do not support SCM or the specific SCM sequence this should fail and we would fall back to writing the register. Signed-off-by: Jordan Crouse Signed-off-by: Rob Clark --- drivers/gpu/drm/msm/adreno/a6xx_gpu.c | 38 +++++++++++++++++++++- drivers/gpu/drm/msm/adreno/adreno_device.c | 1 + 2 files changed, 38 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/msm/adreno/a6xx_gpu.c b/drivers/gpu/drm/msm/adreno/a6xx_gpu.c index 576559a3a523..7028bb784427 100644 --- a/drivers/gpu/drm/msm/adreno/a6xx_gpu.c +++ b/drivers/gpu/drm/msm/adreno/a6xx_gpu.c @@ -10,6 +10,8 @@ #include +#define GPU_PAS_ID 13 + static inline bool _a6xx_check_idle(struct msm_gpu *gpu) { struct adreno_gpu *adreno_gpu = to_adreno_gpu(gpu); @@ -343,6 +345,20 @@ static int a6xx_ucode_init(struct msm_gpu *gpu) return 0; } +static int a6xx_zap_shader_init(struct msm_gpu *gpu) +{ + static bool loaded; + int ret; + + if (loaded) + return 0; + + ret = adreno_zap_shader_load(gpu, GPU_PAS_ID); + + loaded = !ret; + return ret; +} + #define A6XX_INT_MASK (A6XX_RBBM_INT_0_MASK_CP_AHB_ERROR | \ A6XX_RBBM_INT_0_MASK_RBBM_ATB_ASYNCFIFO_OVERFLOW | \ A6XX_RBBM_INT_0_MASK_CP_HW_ERROR | \ @@ -491,7 +507,27 @@ static int a6xx_hw_init(struct msm_gpu *gpu) if (ret) goto out; - gpu_write(gpu, REG_A6XX_RBBM_SECVID_TRUST_CNTL, 0x0); + /* + * Try to load a zap shader into the secure world. If successful + * we can use the CP to switch out of secure mode. If not then we + * have no resource but to try to switch ourselves out manually. If we + * guessed wrong then access to the RBBM_SECVID_TRUST_CNTL register will + * be blocked and a permissions violation will soon follow. + */ + ret = a6xx_zap_shader_init(gpu); + if (!ret) { + OUT_PKT7(gpu->rb[0], CP_SET_SECURE_MODE, 1); + OUT_RING(gpu->rb[0], 0x00000000); + + a6xx_flush(gpu, gpu->rb[0]); + if (!a6xx_idle(gpu, gpu->rb[0])) + return -EINVAL; + } else { + /* Print a warning so if we die, we know why */ + dev_warn_once(gpu->dev->dev, + "Zap shader not enabled - using SECVID_TRUST_CNTL instead\n"); + gpu_write(gpu, REG_A6XX_RBBM_SECVID_TRUST_CNTL, 0x0); + } out: /* diff --git a/drivers/gpu/drm/msm/adreno/adreno_device.c b/drivers/gpu/drm/msm/adreno/adreno_device.c index 0d87db75b710..b907245d3d96 100644 --- a/drivers/gpu/drm/msm/adreno/adreno_device.c +++ b/drivers/gpu/drm/msm/adreno/adreno_device.c @@ -155,6 +155,7 @@ static const struct adreno_info gpulist[] = { .gmem = SZ_1M, .inactive_period = DRM_MSM_INACTIVE_PERIOD, .init = a6xx_gpu_init, + .zapfw = "a630_zap.mdt", }, }; From b55ee6b243bd2743b9cb20b6aa812091748c4304 Mon Sep 17 00:00:00 2001 From: Jordan Crouse Date: Fri, 19 Apr 2019 13:46:16 -0600 Subject: [PATCH 33/34] dt-bindings: drm/msm/gpu: Document a5xx / a6xx zap shader region Describe the zap-shader node that defines a reserved memory region to store the zap shader. Signed-off-by: Jordan Crouse Signed-off-by: Rob Clark --- Documentation/devicetree/bindings/display/msm/gpu.txt | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/Documentation/devicetree/bindings/display/msm/gpu.txt b/Documentation/devicetree/bindings/display/msm/gpu.txt index c04614c63ea9..2b8fd26c43b0 100644 --- a/Documentation/devicetree/bindings/display/msm/gpu.txt +++ b/Documentation/devicetree/bindings/display/msm/gpu.txt @@ -27,6 +27,9 @@ Required properties: - qcom,gmu: For GMU attached devices a phandle to the GMU device that will control the power for the GPU. Applicable targets: - qcom,adreno-630.2 +- zap-shader: For a5xx and a6xx devices this node contains a memory-region that + points to reserved memory to store the zap shader that can be used to help + bring the GPU out of secure mode. Example 3xx/4xx/a5xx: @@ -75,5 +78,9 @@ Example a6xx (with GMU): interconnects = <&rsc_hlos MASTER_GFX3D &rsc_hlos SLAVE_EBI1>; qcom,gmu = <&gmu>; + + zap-shader { + memory-region = <&zap_shader_region>; + }; }; }; From b02872df58aca66d0e7af3ec5065dbc6f0630dd1 Mon Sep 17 00:00:00 2001 From: Jordan Crouse Date: Wed, 10 Apr 2019 10:58:16 -0600 Subject: [PATCH 34/34] drm/msm/a6xx: Don't enable GPU state code if dependencies are missing Add CONFIG_DRM_MSM_GPU_STATE to conditionally compile Adreno GPU state code depending on the availability of the dependencies. Reported-by: Hulk Robot Reported-by: YueHaibing Fixes: 1707add81551 ("drm/msm/a6xx: Add a6xx gpu state") Signed-off-by: Jordan Crouse Signed-off-by: Rob Clark --- drivers/gpu/drm/msm/Kconfig | 5 +++++ drivers/gpu/drm/msm/Makefile | 3 ++- drivers/gpu/drm/msm/adreno/a6xx_gpu.c | 4 +++- drivers/gpu/drm/msm/msm_debugfs.c | 2 +- 4 files changed, 11 insertions(+), 3 deletions(-) diff --git a/drivers/gpu/drm/msm/Kconfig b/drivers/gpu/drm/msm/Kconfig index 78c9e5a5e793..9f2029eca39f 100644 --- a/drivers/gpu/drm/msm/Kconfig +++ b/drivers/gpu/drm/msm/Kconfig @@ -21,6 +21,11 @@ config DRM_MSM help DRM/KMS driver for MSM/snapdragon. +config DRM_MSM_GPU_STATE + bool + depends on DRM_MSM && (DEBUG_FS || DEV_COREDUMP) + default y + config DRM_MSM_REGISTER_LOGGING bool "MSM DRM register logging" depends on DRM_MSM diff --git a/drivers/gpu/drm/msm/Makefile b/drivers/gpu/drm/msm/Makefile index 56a70c74af4e..72d1bfcaab7a 100644 --- a/drivers/gpu/drm/msm/Makefile +++ b/drivers/gpu/drm/msm/Makefile @@ -15,7 +15,6 @@ msm-y := \ adreno/a6xx_gpu.o \ adreno/a6xx_gmu.o \ adreno/a6xx_hfi.o \ - adreno/a6xx_gpu_state.o \ hdmi/hdmi.o \ hdmi/hdmi_audio.o \ hdmi/hdmi_bridge.o \ @@ -96,6 +95,8 @@ msm-y := \ msm-$(CONFIG_DEBUG_FS) += adreno/a5xx_debugfs.o +msm-$(CONFIG_DRM_MSM_GPU_STATE) += adreno/a6xx_gpu_state.o + msm-$(CONFIG_DRM_FBDEV_EMULATION) += msm_fbdev.o msm-$(CONFIG_COMMON_CLK) += disp/mdp4/mdp4_lvds_pll.o msm-$(CONFIG_COMMON_CLK) += hdmi/hdmi_pll_8960.o diff --git a/drivers/gpu/drm/msm/adreno/a6xx_gpu.c b/drivers/gpu/drm/msm/adreno/a6xx_gpu.c index 7028bb784427..ec24508b9d68 100644 --- a/drivers/gpu/drm/msm/adreno/a6xx_gpu.c +++ b/drivers/gpu/drm/msm/adreno/a6xx_gpu.c @@ -807,14 +807,16 @@ static const struct adreno_gpu_funcs funcs = { .active_ring = a6xx_active_ring, .irq = a6xx_irq, .destroy = a6xx_destroy, -#if defined(CONFIG_DEBUG_FS) || defined(CONFIG_DEV_COREDUMP) +#if defined(CONFIG_DRM_MSM_GPU_STATE) .show = a6xx_show, #endif .gpu_busy = a6xx_gpu_busy, .gpu_get_freq = a6xx_gmu_get_freq, .gpu_set_freq = a6xx_gmu_set_freq, +#if defined(CONFIG_DRM_MSM_GPU_STATE) .gpu_state_get = a6xx_gpu_state_get, .gpu_state_put = a6xx_gpu_state_put, +#endif }, .get_timestamp = a6xx_get_timestamp, }; diff --git a/drivers/gpu/drm/msm/msm_debugfs.c b/drivers/gpu/drm/msm/msm_debugfs.c index fb423d309e91..67ef300559cf 100644 --- a/drivers/gpu/drm/msm/msm_debugfs.c +++ b/drivers/gpu/drm/msm/msm_debugfs.c @@ -75,7 +75,7 @@ static int msm_gpu_open(struct inode *inode, struct file *file) struct msm_gpu_show_priv *show_priv; int ret; - if (!gpu) + if (!gpu || !gpu->funcs->gpu_state_get) return -ENODEV; show_priv = kmalloc(sizeof(*show_priv), GFP_KERNEL);