Merge tag 'drm-fixes-2020-05-01' of git://anongit.freedesktop.org/drm…

…/drm

Pull drm fixes from Dave Airlie:
 "Regular scheduled fixes for graphics. Nothing to extreme bunch of
  amdgpu fixes, i915 and qxl fixes, along with some misc ones.

  All seems to be progressing normally.

  core:
   - EDID off by one DTD fix
   - DP mst write return code fix

  dma-buf:
   - fix SET_NAME ioctl uapi
   - doc fixes

  amdgpu:
   - Fix a green screen on resume issue
   - PM fixes for SR-IOV SDMA fix for navi
   - Renoir display fixes
   - Cursor and pageflip stuttering fixes
   - Misc additional display fixes
   - (uapi) Add additional DCC tiling flags for navi1x

  i915:
   - Fix selftest refcnt leak (Xiyu)
   - Fix gem vma lock (Chris)
   - Fix gt's i915_request.timeline acquire by checking if cacheline is
     valid (Chris)
   - Fix IRQ postinistall fault masks (Matt)

  qxl:
   - use after gree fix
   - fix lost kunmap
   - release leak fix

  virtio:
   - context destruction fix"

* tag 'drm-fixes-2020-05-01' of git://anongit.freedesktop.org/drm/drm: (26 commits)
  dma-buf: fix documentation build warnings
  drm/qxl: qxl_release use after free
  drm/qxl: lost qxl_bo_kunmap_atomic_page in qxl_image_init_helper()
  drm/i915: Use proper fault mask in interrupt postinstall too
  drm/amd/display: Use cursor locking to prevent flip delays
  drm/amd/display: Update downspread percent to match spreadsheet for DCN2.1
  drm/amd/display: Defer cursor update around VUPDATE for all ASIC
  drm/amd/display: fix rn soc bb update
  drm/amd/display: check if REFCLK_CNTL register is present
  drm/amdgpu: bump version for invalidate L2 before SDMA IBs
  drm/amdgpu: invalidate L2 before SDMA IBs (v2)
  drm/amdgpu: add tiling flags from Mesa
  drm/amd/powerplay: avoid using pm_en before it is initialized revised
  Revert "drm/amd/powerplay: avoid using pm_en before it is initialized"
  drm/qxl: qxl_release leak in qxl_hw_surface_alloc()
  drm/qxl: qxl_release leak in qxl_draw_dirty_fb()
  drm/virtio: only destroy created contexts
  drm/dp_mst: Fix drm_dp_send_dpcd_write() return code
  drm/i915/gt: Check cacheline is valid before acquiring
  drm/i915/gem: Hold obj->vma.lock over for_each_ggtt_vma()
  ...

drivers/dma-buf/dma-buf.c

@@ -388,7 +388,8 @@ static long dma_buf_ioctl(struct file *file,
 
 		return ret;
 
-	case DMA_BUF_SET_NAME:
+	case DMA_BUF_SET_NAME_A:
+	case DMA_BUF_SET_NAME_B:
 		return dma_buf_set_name(dmabuf, (const char __user *)arg);
 
 	default:
@@ -655,8 +656,8 @@ EXPORT_SYMBOL_GPL(dma_buf_put);
  * calls attach() of dma_buf_ops to allow device-specific attach functionality
  * @dmabuf:		[in]	buffer to attach device to.
  * @dev:		[in]	device to be attached.
- * @importer_ops	[in]	importer operations for the attachment
- * @importer_priv	[in]	importer private pointer for the attachment
+ * @importer_ops:	[in]	importer operations for the attachment
+ * @importer_priv:	[in]	importer private pointer for the attachment
  *
  * Returns struct dma_buf_attachment pointer for this attachment. Attachments
  * must be cleaned up by calling dma_buf_detach().

drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c

@@ -85,9 +85,10 @@
  * - 3.34.0 - Non-DC can flip correctly between buffers with different pitches
  * - 3.35.0 - Add drm_amdgpu_info_device::tcc_disabled_mask
  * - 3.36.0 - Allow reading more status registers on si/cik
+ * - 3.37.0 - L2 is invalidated before SDMA IBs, needed for correctness
  */
 #define KMS_DRIVER_MAJOR	3
-#define KMS_DRIVER_MINOR	36
+#define KMS_DRIVER_MINOR	37
 #define KMS_DRIVER_PATCHLEVEL	0
 
 int amdgpu_vram_limit = 0;

drivers/gpu/drm/amd/amdgpu/navi10_sdma_pkt_open.h

@@ -73,6 +73,22 @@
 #define SDMA_OP_AQL_COPY  0
 #define SDMA_OP_AQL_BARRIER_OR  0
 
+#define SDMA_GCR_RANGE_IS_PA		(1 << 18)
+#define SDMA_GCR_SEQ(x)			(((x) & 0x3) << 16)
+#define SDMA_GCR_GL2_WB			(1 << 15)
+#define SDMA_GCR_GL2_INV		(1 << 14)
+#define SDMA_GCR_GL2_DISCARD		(1 << 13)
+#define SDMA_GCR_GL2_RANGE(x)		(((x) & 0x3) << 11)
+#define SDMA_GCR_GL2_US			(1 << 10)
+#define SDMA_GCR_GL1_INV		(1 << 9)
+#define SDMA_GCR_GLV_INV		(1 << 8)
+#define SDMA_GCR_GLK_INV		(1 << 7)
+#define SDMA_GCR_GLK_WB			(1 << 6)
+#define SDMA_GCR_GLM_INV		(1 << 5)
+#define SDMA_GCR_GLM_WB			(1 << 4)
+#define SDMA_GCR_GL1_RANGE(x)		(((x) & 0x3) << 2)
+#define SDMA_GCR_GLI_INV(x)		(((x) & 0x3) << 0)
+
 /*define for op field*/
 #define SDMA_PKT_HEADER_op_offset 0
 #define SDMA_PKT_HEADER_op_mask   0x000000FF

drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c

@@ -382,6 +382,18 @@ static void sdma_v5_0_ring_emit_ib(struct amdgpu_ring *ring,
 	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
 	uint64_t csa_mc_addr = amdgpu_sdma_get_csa_mc_addr(ring, vmid);
 
+	/* Invalidate L2, because if we don't do it, we might get stale cache
+	 * lines from previous IBs.
+	 */
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_GCR_REQ));
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, (SDMA_GCR_GL2_INV |
+				 SDMA_GCR_GL2_WB |
+				 SDMA_GCR_GLM_INV |
+				 SDMA_GCR_GLM_WB) << 16);
+	amdgpu_ring_write(ring, 0xffffff80);
+	amdgpu_ring_write(ring, 0xffff);
+
 	/* An IB packet must end on a 8 DW boundary--the next dword
 	 * must be on a 8-dword boundary. Our IB packet below is 6
 	 * dwords long, thus add x number of NOPs, such that, in
@@ -1595,7 +1607,7 @@ static const struct amdgpu_ring_funcs sdma_v5_0_ring_funcs = {
 		SOC15_FLUSH_GPU_TLB_NUM_WREG * 3 +
 		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 6 * 2 +
 		10 + 10 + 10, /* sdma_v5_0_ring_emit_fence x3 for user fence, vm fence */
-	.emit_ib_size = 7 + 6, /* sdma_v5_0_ring_emit_ib */
+	.emit_ib_size = 5 + 7 + 6, /* sdma_v5_0_ring_emit_ib */
 	.emit_ib = sdma_v5_0_ring_emit_ib,
 	.emit_fence = sdma_v5_0_ring_emit_fence,
 	.emit_pipeline_sync = sdma_v5_0_ring_emit_pipeline_sync,

drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c

@@ -3340,7 +3340,8 @@ fill_plane_dcc_attributes(struct amdgpu_device *adev,
 			  const union dc_tiling_info *tiling_info,
 			  const uint64_t info,
 			  struct dc_plane_dcc_param *dcc,
-			  struct dc_plane_address *address)
+			  struct dc_plane_address *address,
+			  bool force_disable_dcc)
 {
 	struct dc *dc = adev->dm.dc;
 	struct dc_dcc_surface_param input;
@@ -3352,6 +3353,9 @@ fill_plane_dcc_attributes(struct amdgpu_device *adev,
 	memset(&input, 0, sizeof(input));
 	memset(&output, 0, sizeof(output));
 
+	if (force_disable_dcc)
+		return 0;
+
 	if (!offset)
 		return 0;
 
@@ -3401,7 +3405,8 @@ fill_plane_buffer_attributes(struct amdgpu_device *adev,
 			     union dc_tiling_info *tiling_info,
 			     struct plane_size *plane_size,
 			     struct dc_plane_dcc_param *dcc,
-			     struct dc_plane_address *address)
+			     struct dc_plane_address *address,
+			     bool force_disable_dcc)
 {
 	const struct drm_framebuffer *fb = &afb->base;
 	int ret;
@@ -3507,7 +3512,8 @@ fill_plane_buffer_attributes(struct amdgpu_device *adev,
 
 		ret = fill_plane_dcc_attributes(adev, afb, format, rotation,
 						plane_size, tiling_info,
-						tiling_flags, dcc, address);
+						tiling_flags, dcc, address,
+						force_disable_dcc);
 		if (ret)
 			return ret;
 	}
@@ -3599,7 +3605,8 @@ fill_dc_plane_info_and_addr(struct amdgpu_device *adev,
 			    const struct drm_plane_state *plane_state,
 			    const uint64_t tiling_flags,
 			    struct dc_plane_info *plane_info,
-			    struct dc_plane_address *address)
+			    struct dc_plane_address *address,
+			    bool force_disable_dcc)
 {
 	const struct drm_framebuffer *fb = plane_state->fb;
 	const struct amdgpu_framebuffer *afb =
@@ -3681,7 +3688,8 @@ fill_dc_plane_info_and_addr(struct amdgpu_device *adev,
 					   plane_info->rotation, tiling_flags,
 					   &plane_info->tiling_info,
 					   &plane_info->plane_size,
-					   &plane_info->dcc, address);
+					   &plane_info->dcc, address,
+					   force_disable_dcc);
 	if (ret)
 		return ret;
 
@@ -3704,6 +3712,7 @@ static int fill_dc_plane_attributes(struct amdgpu_device *adev,
 	struct dc_plane_info plane_info;
 	uint64_t tiling_flags;
 	int ret;
+	bool force_disable_dcc = false;
 
 	ret = fill_dc_scaling_info(plane_state, &scaling_info);
 	if (ret)
@@ -3718,9 +3727,11 @@ static int fill_dc_plane_attributes(struct amdgpu_device *adev,
 	if (ret)
 		return ret;
 
+	force_disable_dcc = adev->asic_type == CHIP_RAVEN && adev->in_suspend;
 	ret = fill_dc_plane_info_and_addr(adev, plane_state, tiling_flags,
 					  &plane_info,
-					  &dc_plane_state->address);
+					  &dc_plane_state->address,
+					  force_disable_dcc);
 	if (ret)
 		return ret;
 
@@ -5342,6 +5353,7 @@ static int dm_plane_helper_prepare_fb(struct drm_plane *plane,
 	uint64_t tiling_flags;
 	uint32_t domain;
 	int r;
+	bool force_disable_dcc = false;
 
 	dm_plane_state_old = to_dm_plane_state(plane->state);
 	dm_plane_state_new = to_dm_plane_state(new_state);
@@ -5400,11 +5412,13 @@ static int dm_plane_helper_prepare_fb(struct drm_plane *plane,
 			dm_plane_state_old->dc_state != dm_plane_state_new->dc_state) {
 		struct dc_plane_state *plane_state = dm_plane_state_new->dc_state;
 
+		force_disable_dcc = adev->asic_type == CHIP_RAVEN && adev->in_suspend;
 		fill_plane_buffer_attributes(
 			adev, afb, plane_state->format, plane_state->rotation,
 			tiling_flags, &plane_state->tiling_info,
 			&plane_state->plane_size, &plane_state->dcc,
-			&plane_state->address);
+			&plane_state->address,
+			force_disable_dcc);
 	}
 
 	return 0;
@@ -6676,7 +6690,12 @@ static void amdgpu_dm_commit_planes(struct drm_atomic_state *state,
 		fill_dc_plane_info_and_addr(
 			dm->adev, new_plane_state, tiling_flags,
 			&bundle->plane_infos[planes_count],
-			&bundle->flip_addrs[planes_count].address);
+			&bundle->flip_addrs[planes_count].address,
+			false);
+
+		DRM_DEBUG_DRIVER("plane: id=%d dcc_en=%d\n",
+				 new_plane_state->plane->index,
+				 bundle->plane_infos[planes_count].dcc.enable);
 
 		bundle->surface_updates[planes_count].plane_info =
 			&bundle->plane_infos[planes_count];
@@ -8096,7 +8115,8 @@ dm_determine_update_type_for_commit(struct amdgpu_display_manager *dm,
 				ret = fill_dc_plane_info_and_addr(
 					dm->adev, new_plane_state, tiling_flags,
 					plane_info,
-					&flip_addr->address);
+					&flip_addr->address,
+					false);
 				if (ret)
 					goto cleanup;
 

drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c

@@ -2908,6 +2908,12 @@ bool dc_link_handle_hpd_rx_irq(struct dc_link *link, union hpd_irq_data *out_hpd
 					sizeof(hpd_irq_dpcd_data),
 					"Status: ");
 
+		for (i = 0; i < MAX_PIPES; i++) {
+			pipe_ctx = &link->dc->current_state->res_ctx.pipe_ctx[i];
+			if (pipe_ctx && pipe_ctx->stream && pipe_ctx->stream->link == link)
+				link->dc->hwss.blank_stream(pipe_ctx);
+		}
+
 		for (i = 0; i < MAX_PIPES; i++) {
 			pipe_ctx = &link->dc->current_state->res_ctx.pipe_ctx[i];
 			if (pipe_ctx && pipe_ctx->stream && pipe_ctx->stream->link == link)
@@ -2927,6 +2933,12 @@ bool dc_link_handle_hpd_rx_irq(struct dc_link *link, union hpd_irq_data *out_hpd
 		if (pipe_ctx->stream->signal == SIGNAL_TYPE_DISPLAY_PORT_MST)
 			dc_link_reallocate_mst_payload(link);
 
+		for (i = 0; i < MAX_PIPES; i++) {
+			pipe_ctx = &link->dc->current_state->res_ctx.pipe_ctx[i];
+			if (pipe_ctx && pipe_ctx->stream && pipe_ctx->stream->link == link)
+				link->dc->hwss.unblank_stream(pipe_ctx, &previous_link_settings);
+		}
+
 		status = false;
 		if (out_link_loss)
 			*out_link_loss = true;
@@ -4227,6 +4239,21 @@ void dp_set_fec_enable(struct dc_link *link, bool enable)
 void dpcd_set_source_specific_data(struct dc_link *link)
 {
 	const uint32_t post_oui_delay = 30; // 30ms
+	uint8_t dspc = 0;
+	enum dc_status ret = DC_ERROR_UNEXPECTED;
+
+	ret = core_link_read_dpcd(link, DP_DOWN_STREAM_PORT_COUNT, &dspc,
+				  sizeof(dspc));
+
+	if (ret != DC_OK) {
+		DC_LOG_ERROR("Error in DP aux read transaction,"
+			     " not writing source specific data\n");
+		return;
+	}
+
+	/* Return if OUI unsupported */
+	if (!(dspc & DP_OUI_SUPPORT))
+		return;
 
 	if (!link->dc->vendor_signature.is_valid) {
 		struct dpcd_amd_signature amd_signature;

drivers/gpu/drm/amd/display/dc/core/dc_stream.c

@@ -231,34 +231,6 @@ struct dc_stream_status *dc_stream_get_status(
 	return dc_stream_get_status_from_state(dc->current_state, stream);
 }
 
-static void delay_cursor_until_vupdate(struct pipe_ctx *pipe_ctx, struct dc *dc)
-{
-#if defined(CONFIG_DRM_AMD_DC_DCN)
-	unsigned int vupdate_line;
-	unsigned int lines_to_vupdate, us_to_vupdate, vpos, nvpos;
-	struct dc_stream_state *stream = pipe_ctx->stream;
-	unsigned int us_per_line;
-
-	if (stream->ctx->asic_id.chip_family == FAMILY_RV &&
-			ASICREV_IS_RAVEN(stream->ctx->asic_id.hw_internal_rev)) {
-
-		vupdate_line = dc->hwss.get_vupdate_offset_from_vsync(pipe_ctx);
-		if (!dc_stream_get_crtc_position(dc, &stream, 1, &vpos, &nvpos))
-			return;
-
-		if (vpos >= vupdate_line)
-			return;
-
-		us_per_line = stream->timing.h_total * 10000 / stream->timing.pix_clk_100hz;
-		lines_to_vupdate = vupdate_line - vpos;
-		us_to_vupdate = lines_to_vupdate * us_per_line;
-
-		/* 70 us is a conservative estimate of cursor update time*/
-		if (us_to_vupdate < 70)
-			udelay(us_to_vupdate);
-	}
-#endif
-}
 
 /**
  * dc_stream_set_cursor_attributes() - Update cursor attributes and set cursor surface address
@@ -298,9 +270,7 @@ bool dc_stream_set_cursor_attributes(
 
 		if (!pipe_to_program) {
 			pipe_to_program = pipe_ctx;
-
-			delay_cursor_until_vupdate(pipe_ctx, dc);
-			dc->hwss.pipe_control_lock(dc, pipe_to_program, true);
+			dc->hwss.cursor_lock(dc, pipe_to_program, true);
 		}
 
 		dc->hwss.set_cursor_attribute(pipe_ctx);
@@ -309,7 +279,7 @@ bool dc_stream_set_cursor_attributes(
 	}
 
 	if (pipe_to_program)
-		dc->hwss.pipe_control_lock(dc, pipe_to_program, false);
+		dc->hwss.cursor_lock(dc, pipe_to_program, false);
 
 	return true;
 }
@@ -349,16 +319,14 @@ bool dc_stream_set_cursor_position(
 
 		if (!pipe_to_program) {
 			pipe_to_program = pipe_ctx;
-
-			delay_cursor_until_vupdate(pipe_ctx, dc);
-			dc->hwss.pipe_control_lock(dc, pipe_to_program, true);
+			dc->hwss.cursor_lock(dc, pipe_to_program, true);
 		}
 
 		dc->hwss.set_cursor_position(pipe_ctx);
 	}
 
 	if (pipe_to_program)
-		dc->hwss.pipe_control_lock(dc, pipe_to_program, false);
+		dc->hwss.cursor_lock(dc, pipe_to_program, false);
 
 	return true;
 }

drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c

@@ -2757,6 +2757,7 @@ static const struct hw_sequencer_funcs dce110_funcs = {
 	.disable_plane = dce110_power_down_fe,
 	.pipe_control_lock = dce_pipe_control_lock,
 	.interdependent_update_lock = NULL,
+	.cursor_lock = dce_pipe_control_lock,
 	.prepare_bandwidth = dce110_prepare_bandwidth,
 	.optimize_bandwidth = dce110_optimize_bandwidth,
 	.set_drr = set_drr,

drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c

@@ -1625,6 +1625,16 @@ void dcn10_pipe_control_lock(
 		hws->funcs.verify_allow_pstate_change_high(dc);
 }
 
+void dcn10_cursor_lock(struct dc *dc, struct pipe_ctx *pipe, bool lock)
+{
+	/* cursor lock is per MPCC tree, so only need to lock one pipe per stream */
+	if (!pipe || pipe->top_pipe)
+		return;
+
+	dc->res_pool->mpc->funcs->cursor_lock(dc->res_pool->mpc,
+			pipe->stream_res.opp->inst, lock);
+}
+
 static bool wait_for_reset_trigger_to_occur(
 	struct dc_context *dc_ctx,
 	struct timing_generator *tg)

drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.h

@@ -49,6 +49,7 @@ void dcn10_pipe_control_lock(
 	struct dc *dc,
 	struct pipe_ctx *pipe,
 	bool lock);
+void dcn10_cursor_lock(struct dc *dc, struct pipe_ctx *pipe, bool lock);
 void dcn10_blank_pixel_data(
 		struct dc *dc,
 		struct pipe_ctx *pipe_ctx,

drivers/gpu/drm/amd/display/dc/dcn10/dcn10_init.c

@@ -50,6 +50,7 @@ static const struct hw_sequencer_funcs dcn10_funcs = {
 	.disable_audio_stream = dce110_disable_audio_stream,
 	.disable_plane = dcn10_disable_plane,
 	.pipe_control_lock = dcn10_pipe_control_lock,
+	.cursor_lock = dcn10_cursor_lock,
 	.interdependent_update_lock = dcn10_lock_all_pipes,
 	.prepare_bandwidth = dcn10_prepare_bandwidth,
 	.optimize_bandwidth = dcn10_optimize_bandwidth,