Merge tag 'drm-next-2019-07-16' of git://anongit.freedesktop.org/drm/drm

Pull drm updates from Dave Airlie:
 "The biggest thing in this is the AMD Navi GPU support, this again
  contains a bunch of header files that are large. These are the new AMD
  RX5700 GPUs that just recently became available.

  New drivers:
   - ST-Ericsson MCDE driver
   - Ingenic JZ47xx SoC

  UAPI change:
   - HDR source metadata property

  Core:
   - HDR inforframes and EDID parsing
   - drm hdmi infoframe unpacking
   - remove prime sg_table caching into dma-buf
   - New gem vram helpers to reduce driver code
   - Lots of drmP.h removal
   - reservation fencing fix
   - documentation updates
   - drm_fb_helper_connector removed
   - mode name command handler rewrite

  fbcon:
   - Remove the fbcon notifiers

  ttm:
   - forward progress fixes

  dma-buf:
   - make mmap call optional
   - debugfs refcount fixes
   - dma-fence free with pending signals fix
   - each dma-buf gets an inode

  Panels:
   - Lots of additional panel bindings

  amdgpu:
   - initial navi10 support
   - avoid hw reset
   - HDR metadata support
   - new thermal sensors for vega asics
   - RAS fixes
   - use HMM rather than MMU notifier
   - xgmi topology via kfd
   - SR-IOV fixes
   - driver reload fixes
   - DC use a core bpc attribute
   - Aux fixes for DC
   - Bandwidth calc updates for DC
   - Clock handling refactor
   - kfd VEGAM support

  vmwgfx:
   - Coherent memory support changes

  i915:
   - HDR Support
   - HDMI i2c link
   - Icelake multi-segmented gamma support
   - GuC firmware update
   - Mule Creek Canyon PCH support for EHL
   - EHL platform updtes
   - move i915.alpha_support to i915.force_probe
   - runtime PM refactoring
   - VBT parsing refactoring
   - DSI fixes
   - struct mutex dependency reduction
   - GEM code reorg

  mali-dp:
   - Komeda driver features

  msm:
   - dsi vs EPROBE_DEFER fixes
   - msm8998 snapdragon 835 support
   - a540 gpu support
   - mdp5 and dpu interconnect support

  exynos:
   - drmP.h removal

  tegra:
   - misc fixes

  tda998x:
   - audio support improvements
   - pixel repeated mode support
   - quantisation range handling corrections
   - HDMI vendor info fix

  armada:
   - interlace support fix
   - overlay/video plane register handling refactor
   - add gamma support

  rockchip:
   - RX3328 support

  panfrost:
   - expose perf counters via hidden ioctls

  vkms:
   - enumerate CRC sources list

  ast:
   - rework BO handling

  mgag200:
   - rework BO handling

  dw-hdmi:
   - suspend/resume support

  rcar-du:
   - R8A774A1 Soc Support
   - LVDS dual-link mode support
   - Additional formats
   - Misc fixes

  omapdrm:
   - DSI command mode display support

  stm
   - fb modifier support
   - runtime PM support

  sun4i:
   - use vmap ops

  vc4:
   - binner bo binding rework

  v3d:
   - compute shader support
   - resync/sync fixes
   - job management refactoring

  lima:
   - NULL pointer in irq handler fix
   - scheduler default timeout

  virtio:
   - fence seqno support
   - trace events

  bochs:
   - misc fixes

  tc458767:
   - IRQ/HDP handling

  sii902x:
   - HDMI audio support

  atmel-hlcdc:
   - misc fixes

  meson:
   - zpos support"

* tag 'drm-next-2019-07-16' of git://anongit.freedesktop.org/drm/drm: (1815 commits)
  Revert "Merge branch 'vmwgfx-next' of git://people.freedesktop.org/~thomash/linux into drm-next"
  Revert "mm: adjust apply_to_pfn_range interface for dropped token."
  mm: adjust apply_to_pfn_range interface for dropped token.
  drm/amdgpu/navi10: add uclk activity sensor
  drm/amdgpu: properly guard the generic discovery code
  drm/amdgpu: add missing documentation on new module parameters
  drm/amdgpu: don't invalidate caches in RELEASE_MEM, only do the writeback
  drm/amd/display: avoid 64-bit division
  drm/amdgpu/psp11: simplify the ucode register logic
  drm/amdgpu: properly guard DC support in navi code
  drm/amd/powerplay: vega20: fix uninitialized variable use
  drm/amd/display: dcn20: include linux/delay.h
  amdgpu: make pmu support optional
  drm/amd/powerplay: Zero initialize current_rpm in vega20_get_fan_speed_percent
  drm/amd/powerplay: Zero initialize freq in smu_v11_0_get_current_clk_freq
  drm/amd/powerplay: Use memset to initialize metrics structs
  drm/amdgpu/mes10.1: Fix header guard
  drm/amd/powerplay: add temperature sensor support for navi10
  drm/amdgpu: fix scheduler timeout calc
  drm/amdgpu: Prepare for hmm_range_register API change (v2)
  ...

1 files changed, 66 insertions, 40 deletions

diff --git a/drivers/gpu/drm/v3d/v3d_drv.h b/drivers/gpu/drm/v3d/v3d_drv.h
index e9d4a2fdcf44..9aad9da1eb11 100644
--- a/drivers/gpu/drm/v3d/v3d_drv.h
+++ b/drivers/gpu/drm/v3d/v3d_drv.h
@@ -16,9 +16,11 @@ enum v3d_queue {
 	V3D_BIN,
 	V3D_RENDER,
 	V3D_TFU,
+	V3D_CSD,
+	V3D_CACHE_CLEAN,
 };
 
-#define V3D_MAX_QUEUES (V3D_TFU + 1)
+#define V3D_MAX_QUEUES (V3D_CACHE_CLEAN + 1)
 
 struct v3d_queue_state {
 	struct drm_gpu_scheduler sched;
@@ -55,6 +57,8 @@ struct v3d_dev {
 	 */
 	void *mmu_scratch;
 	dma_addr_t mmu_scratch_paddr;
+	/* virtual address bits from V3D to the MMU. */
+	int va_width;
 
 	/* Number of V3D cores. */
 	u32 cores;
@@ -67,9 +71,10 @@ struct v3d_dev {
 
 	struct work_struct overflow_mem_work;
 
-	struct v3d_exec_info *bin_job;
-	struct v3d_exec_info *render_job;
+	struct v3d_bin_job *bin_job;
+	struct v3d_render_job *render_job;
 	struct v3d_tfu_job *tfu_job;
+	struct v3d_csd_job *csd_job;
 
 	struct v3d_queue_state queue[V3D_MAX_QUEUES];
 
@@ -92,6 +97,12 @@ struct v3d_dev {
 	 */
 	struct mutex sched_lock;
 
+	/* Lock taken during a cache clean and when initiating an L2
+	 * flush, to keep L2 flushes from interfering with the
+	 * synchronous L2 cleans.
+	 */
+	struct mutex cache_clean_lock;
+
 	struct {
 		u32 num_allocated;
 		u32 pages_allocated;
@@ -104,6 +115,12 @@ to_v3d_dev(struct drm_device *dev)
 	return (struct v3d_dev *)dev->dev_private;
 }
 
+static inline bool
+v3d_has_csd(struct v3d_dev *v3d)
+{
+	return v3d->ver >= 41;
+}
+
 /* The per-fd struct, which tracks the MMU mappings. */
 struct v3d_file_priv {
 	struct v3d_dev *v3d;
@@ -117,7 +134,7 @@ struct v3d_bo {
 	struct drm_mm_node node;
 
 	/* List entry for the BO's position in
-	 * v3d_exec_info->unref_list
+	 * v3d_render_job->unref_list
 	 */
 	struct list_head unref_head;
 };
@@ -157,67 +174,74 @@ to_v3d_fence(struct dma_fence *fence)
 struct v3d_job {
 	struct drm_sched_job base;
 
-	struct v3d_exec_info *exec;
+	struct kref refcount;
 
-	/* An optional fence userspace can pass in for the job to depend on. */
-	struct dma_fence *in_fence;
+	struct v3d_dev *v3d;
+
+	/* This is the array of BOs that were looked up at the start
+	 * of submission.
+	 */
+	struct drm_gem_object **bo;
+	u32 bo_count;
+
+	/* Array of struct dma_fence * to block on before submitting this job.
+	 */
+	struct xarray deps;
+	unsigned long last_dep;
 
 	/* v3d fence to be signaled by IRQ handler when the job is complete. */
 	struct dma_fence *irq_fence;
 
+	/* scheduler fence for when the job is considered complete and
+	 * the BO reservations can be released.
+	 */
+	struct dma_fence *done_fence;
+
+	/* Callback for the freeing of the job on refcount going to 0. */
+	void (*free)(struct kref *ref);
+};
+
+struct v3d_bin_job {
+	struct v3d_job base;
+
 	/* GPU virtual addresses of the start/end of the CL job. */
 	u32 start, end;
 
 	u32 timedout_ctca, timedout_ctra;
-};
 
-struct v3d_exec_info {
-	struct v3d_dev *v3d;
+	/* Corresponding render job, for attaching our overflow memory. */
+	struct v3d_render_job *render;
 
-	struct v3d_job bin, render;
-
-	/* Fence for when the scheduler considers the binner to be
-	 * done, for render to depend on.
-	 */
-	struct dma_fence *bin_done_fence;
+	/* Submitted tile memory allocation start/size, tile state. */
+	u32 qma, qms, qts;
+};
 
-	/* Fence for when the scheduler considers the render to be
-	 * done, for when the BOs reservations should be complete.
-	 */
-	struct dma_fence *render_done_fence;
+struct v3d_render_job {
+	struct v3d_job base;
 
-	struct kref refcount;
+	/* GPU virtual addresses of the start/end of the CL job. */
+	u32 start, end;
 
-	/* This is the array of BOs that were looked up at the start of exec. */
-	struct v3d_bo **bo;
-	u32 bo_count;
+	u32 timedout_ctca, timedout_ctra;
 
 	/* List of overflow BOs used in the job that need to be
 	 * released once the job is complete.
 	 */
 	struct list_head unref_list;
-
-	/* Submitted tile memory allocation start/size, tile state. */
-	u32 qma, qms, qts;
 };
 
 struct v3d_tfu_job {
-	struct drm_sched_job base;
+	struct v3d_job base;
 
 	struct drm_v3d_submit_tfu args;
+};
 
-	/* An optional fence userspace can pass in for the job to depend on. */
-	struct dma_fence *in_fence;
-
-	/* v3d fence to be signaled by IRQ handler when the job is complete. */
-	struct dma_fence *irq_fence;
-
-	struct v3d_dev *v3d;
+struct v3d_csd_job {
+	struct v3d_job base;
 
-	struct kref refcount;
+	u32 timedout_batches;
 
-	/* This is the array of BOs that were looked up at the start of exec. */
-	struct v3d_bo *bo[4];
+	struct drm_v3d_submit_csd args;
 };
 
 /**
@@ -281,12 +305,14 @@ int v3d_submit_cl_ioctl(struct drm_device *dev, void *data,
 			struct drm_file *file_priv);
 int v3d_submit_tfu_ioctl(struct drm_device *dev, void *data,
 			 struct drm_file *file_priv);
+int v3d_submit_csd_ioctl(struct drm_device *dev, void *data,
+			 struct drm_file *file_priv);
 int v3d_wait_bo_ioctl(struct drm_device *dev, void *data,
 		      struct drm_file *file_priv);
-void v3d_exec_put(struct v3d_exec_info *exec);
-void v3d_tfu_job_put(struct v3d_tfu_job *exec);
+void v3d_job_put(struct v3d_job *job);
 void v3d_reset(struct v3d_dev *v3d);
 void v3d_invalidate_caches(struct v3d_dev *v3d);
+void v3d_clean_caches(struct v3d_dev *v3d);
 
 /* v3d_irq.c */
 int v3d_irq_init(struct v3d_dev *v3d);