Merge tag 'drm-fixes-2024-04-05' of https://gitlab.freedesktop.org/drm/kernel

Pull drm fixes from Dave Airlie:
 "Weekly fixes, mostly xe and i915, amdgpu on a week off, otherwise a
  nouveau fix for a crash with new vulkan cts tests, and a couple of
  cleanups and misc fixes.

  display:
   - fix typos in kerneldoc

  prime:
   - unbreak dma-buf export for virt-gpu

  nouveau:
   - uvmm: fix remap address calculation
   - minor cleanups

  panfrost:
   - fix power-transition timeouts

  xe:
   - Stop using system_unbound_wq for preempt fences
   - Fix saving unordered rebinding fences by attaching them as kernel
     feces to the vm's resv
   - Fix TLB invalidation fences completing out of order
   - Move rebind TLB invalidation to the ring ops to reduce the latency

  i915:
   - A few DisplayPort related fixes
   - eDP PSR fixes
   - Remove some VM space restrictions on older platforms
   - Disable automatic load CCS load balancing"

* tag 'drm-fixes-2024-04-05' of https://gitlab.freedesktop.org/drm/kernel: (22 commits)
  drm/xe: Use ordered wq for preempt fence waiting
  drm/xe: Move vma rebinding to the drm_exec locking loop
  drm/xe: Make TLB invalidation fences unordered
  drm/xe: Rework rebinding
  drm/xe: Use ring ops TLB invalidation for rebinds
  drm/i915/mst: Reject FEC+MST on ICL
  drm/i915/mst: Limit MST+DSC to TGL+
  drm/i915/dp: Fix the computation for compressed_bpp for DISPLAY < 13
  drm/i915/gt: Enable only one CCS for compute workload
  drm/i915/gt: Do not generate the command streamer for all the CCS
  drm/i915/gt: Disable HW load balancing for CCS
  drm/i915/gt: Limit the reserved VM space to only the platforms that need it
  drm/i915/psr: Fix intel_psr2_sel_fetch_et_alignment usage
  drm/i915/psr: Move writing early transport pipe src
  drm/i915/psr: Calculate PIPE_SRCSZ_ERLY_TPT value
  drm/i915/dp: Remove support for UHBR13.5
  drm/i915/dp: Fix DSC state HW readout for SST connectors
  drm/display: fix typo
  drm/prime: Unbreak virtgpu dma-buf export
  nouveau/uvmm: fix addr/range calcs for remap operations
  ...

35 files changed, 340 insertions, 197 deletions

diff --git a/drivers/gpu/drm/display/drm_dp_dual_mode_helper.c b/drivers/gpu/drm/display/drm_dp_dual_mode_helper.c
index bd61e20770a5..14a2a8473682 100644
--- a/drivers/gpu/drm/display/drm_dp_dual_mode_helper.c
+++ b/drivers/gpu/drm/display/drm_dp_dual_mode_helper.c
@@ -52,7 +52,7 @@
  * @adapter: I2C adapter for the DDC bus
  * @offset: register offset
  * @buffer: buffer for return data
- * @size: sizo of the buffer
+ * @size: size of the buffer
  *
  * Reads @size bytes from the DP dual mode adaptor registers
  * starting at @offset.
@@ -116,7 +116,7 @@ EXPORT_SYMBOL(drm_dp_dual_mode_read);
  * @adapter: I2C adapter for the DDC bus
  * @offset: register offset
  * @buffer: buffer for write data
- * @size: sizo of the buffer
+ * @size: size of the buffer
  *
  * Writes @size bytes to the DP dual mode adaptor registers
  * starting at @offset.
diff --git a/drivers/gpu/drm/drm_prime.c b/drivers/gpu/drm/drm_prime.c
index 7352bde299d5..03bd3c7bd0dc 100644
--- a/drivers/gpu/drm/drm_prime.c
+++ b/drivers/gpu/drm/drm_prime.c
@@ -582,7 +582,12 @@ int drm_gem_map_attach(struct dma_buf *dma_buf,
 {
 	struct drm_gem_object *obj = dma_buf->priv;
 
-	if (!obj->funcs->get_sg_table)
+	/*
+	 * drm_gem_map_dma_buf() requires obj->get_sg_table(), but drivers
+	 * that implement their own ->map_dma_buf() do not.
+	 */
+	if (dma_buf->ops->map_dma_buf == drm_gem_map_dma_buf &&
+	    !obj->funcs->get_sg_table)
 		return -ENOSYS;
 
 	return drm_gem_pin(obj);
diff --git a/drivers/gpu/drm/i915/Makefile b/drivers/gpu/drm/i915/Makefile
index 4c2f85632391..fba73c38e235 100644
--- a/drivers/gpu/drm/i915/Makefile
+++ b/drivers/gpu/drm/i915/Makefile
@@ -118,6 +118,7 @@ gt-y += \
 	gt/intel_ggtt_fencing.o \
 	gt/intel_gt.o \
 	gt/intel_gt_buffer_pool.o \
+	gt/intel_gt_ccs_mode.o \
 	gt/intel_gt_clock_utils.o \
 	gt/intel_gt_debugfs.o \
 	gt/intel_gt_engines_debugfs.o \
diff --git a/drivers/gpu/drm/i915/display/intel_display.c b/drivers/gpu/drm/i915/display/intel_display.c
index ab2f52d21bad..8af9e6128277 100644
--- a/drivers/gpu/drm/i915/display/intel_display.c
+++ b/drivers/gpu/drm/i915/display/intel_display.c
@@ -2709,15 +2709,6 @@ static void intel_set_pipe_src_size(const struct intel_crtc_state *crtc_state)
 	 */
 	intel_de_write(dev_priv, PIPESRC(pipe),
 		       PIPESRC_WIDTH(width - 1) | PIPESRC_HEIGHT(height - 1));
-
-	if (!crtc_state->enable_psr2_su_region_et)
-		return;
-
-	width = drm_rect_width(&crtc_state->psr2_su_area);
-	height = drm_rect_height(&crtc_state->psr2_su_area);
-
-	intel_de_write(dev_priv, PIPE_SRCSZ_ERLY_TPT(pipe),
-		       PIPESRC_WIDTH(width - 1) | PIPESRC_HEIGHT(height - 1));
 }
 
 static bool intel_pipe_is_interlaced(const struct intel_crtc_state *crtc_state)
diff --git a/drivers/gpu/drm/i915/display/intel_display_device.h b/drivers/gpu/drm/i915/display/intel_display_device.h
index fe4268813786..9b1bce2624b9 100644
--- a/drivers/gpu/drm/i915/display/intel_display_device.h
+++ b/drivers/gpu/drm/i915/display/intel_display_device.h
@@ -47,6 +47,7 @@ struct drm_printer;
 #define HAS_DPT(i915)			(DISPLAY_VER(i915) >= 13)
 #define HAS_DSB(i915)			(DISPLAY_INFO(i915)->has_dsb)
 #define HAS_DSC(__i915)			(DISPLAY_RUNTIME_INFO(__i915)->has_dsc)
+#define HAS_DSC_MST(__i915)		(DISPLAY_VER(__i915) >= 12 && HAS_DSC(__i915))
 #define HAS_FBC(i915)			(DISPLAY_RUNTIME_INFO(i915)->fbc_mask != 0)
 #define HAS_FPGA_DBG_UNCLAIMED(i915)	(DISPLAY_INFO(i915)->has_fpga_dbg)
 #define HAS_FW_BLC(i915)		(DISPLAY_VER(i915) >= 3)
diff --git a/drivers/gpu/drm/i915/display/intel_display_types.h b/drivers/gpu/drm/i915/display/intel_display_types.h
index 9104f18753b4..bf3f942e19c3 100644
--- a/drivers/gpu/drm/i915/display/intel_display_types.h
+++ b/drivers/gpu/drm/i915/display/intel_display_types.h
@@ -1423,6 +1423,8 @@ struct intel_crtc_state {
 
 	u32 psr2_man_track_ctl;
 
+	u32 pipe_srcsz_early_tpt;
+
 	struct drm_rect psr2_su_area;
 
 	/* Variable Refresh Rate state */
diff --git a/drivers/gpu/drm/i915/display/intel_dp.c b/drivers/gpu/drm/i915/display/intel_dp.c
index f98ef4b42a44..abd62bebc46d 100644
--- a/drivers/gpu/drm/i915/display/intel_dp.c
+++ b/drivers/gpu/drm/i915/display/intel_dp.c
@@ -499,7 +499,7 @@ intel_dp_set_source_rates(struct intel_dp *intel_dp)
 	/* The values must be in increasing order */
 	static const int mtl_rates[] = {
 		162000, 216000, 243000, 270000, 324000, 432000, 540000, 675000,
-		810000,	1000000, 1350000, 2000000,
+		810000,	1000000, 2000000,
 	};
 	static const int icl_rates[] = {
 		162000, 216000, 270000, 324000, 432000, 540000, 648000, 810000,
@@ -1422,7 +1422,8 @@ static bool intel_dp_source_supports_fec(struct intel_dp *intel_dp,
 	if (DISPLAY_VER(dev_priv) >= 12)
 		return true;
 
-	if (DISPLAY_VER(dev_priv) == 11 && encoder->port != PORT_A)
+	if (DISPLAY_VER(dev_priv) == 11 && encoder->port != PORT_A &&
+	    !intel_crtc_has_type(pipe_config, INTEL_OUTPUT_DP_MST))
 		return true;
 
 	return false;
@@ -1917,8 +1918,9 @@ icl_dsc_compute_link_config(struct intel_dp *intel_dp,
 	dsc_max_bpp = min(dsc_max_bpp, pipe_bpp - 1);
 
 	for (i = 0; i < ARRAY_SIZE(valid_dsc_bpp); i++) {
-		if (valid_dsc_bpp[i] < dsc_min_bpp ||
-		    valid_dsc_bpp[i] > dsc_max_bpp)
+		if (valid_dsc_bpp[i] < dsc_min_bpp)
+			continue;
+		if (valid_dsc_bpp[i] > dsc_max_bpp)
 			break;
 
 		ret = dsc_compute_link_config(intel_dp,
@@ -6557,6 +6559,7 @@ intel_dp_init_connector(struct intel_digital_port *dig_port,
 		intel_connector->get_hw_state = intel_ddi_connector_get_hw_state;
 	else
 		intel_connector->get_hw_state = intel_connector_get_hw_state;
+	intel_connector->sync_state = intel_dp_connector_sync_state;
 
 	if (!intel_edp_init_connector(intel_dp, intel_connector)) {
 		intel_dp_aux_fini(intel_dp);
diff --git a/drivers/gpu/drm/i915/display/intel_dp_mst.c b/drivers/gpu/drm/i915/display/intel_dp_mst.c
index 53aec023ce92..b651c990af85 100644
--- a/drivers/gpu/drm/i915/display/intel_dp_mst.c
+++ b/drivers/gpu/drm/i915/display/intel_dp_mst.c
@@ -1355,7 +1355,7 @@ intel_dp_mst_mode_valid_ctx(struct drm_connector *connector,
 		return 0;
 	}
 
-	if (DISPLAY_VER(dev_priv) >= 10 &&
+	if (HAS_DSC_MST(dev_priv) &&
 	    drm_dp_sink_supports_dsc(intel_connector->dp.dsc_dpcd)) {
 		/*
 		 * TBD pass the connector BPC,
diff --git a/drivers/gpu/drm/i915/display/intel_psr.c b/drivers/gpu/drm/i915/display/intel_psr.c
index 6927785fd6ff..b6e539f1342c 100644
--- a/drivers/gpu/drm/i915/display/intel_psr.c
+++ b/drivers/gpu/drm/i915/display/intel_psr.c
@@ -1994,6 +1994,7 @@ static void psr_force_hw_tracking_exit(struct intel_dp *intel_dp)
 
 void intel_psr2_program_trans_man_trk_ctl(const struct intel_crtc_state *crtc_state)
 {
+	struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc);
 	struct drm_i915_private *dev_priv = to_i915(crtc_state->uapi.crtc->dev);
 	enum transcoder cpu_transcoder = crtc_state->cpu_transcoder;
 	struct intel_encoder *encoder;
@@ -2013,6 +2014,12 @@ void intel_psr2_program_trans_man_trk_ctl(const struct intel_crtc_state *crtc_st
 
 	intel_de_write(dev_priv, PSR2_MAN_TRK_CTL(cpu_transcoder),
 		       crtc_state->psr2_man_track_ctl);
+
+	if (!crtc_state->enable_psr2_su_region_et)
+		return;
+
+	intel_de_write(dev_priv, PIPE_SRCSZ_ERLY_TPT(crtc->pipe),
+		       crtc_state->pipe_srcsz_early_tpt);
 }
 
 static void psr2_man_trk_ctl_calc(struct intel_crtc_state *crtc_state,
@@ -2051,6 +2058,20 @@ exit:
 	crtc_state->psr2_man_track_ctl = val;
 }
 
+static u32 psr2_pipe_srcsz_early_tpt_calc(struct intel_crtc_state *crtc_state,
+					  bool full_update)
+{
+	int width, height;
+
+	if (!crtc_state->enable_psr2_su_region_et || full_update)
+		return 0;
+
+	width = drm_rect_width(&crtc_state->psr2_su_area);
+	height = drm_rect_height(&crtc_state->psr2_su_area);
+
+	return PIPESRC_WIDTH(width - 1) | PIPESRC_HEIGHT(height - 1);
+}
+
 static void clip_area_update(struct drm_rect *overlap_damage_area,
 			     struct drm_rect *damage_area,
 			     struct drm_rect *pipe_src)
@@ -2095,21 +2116,36 @@ static void intel_psr2_sel_fetch_pipe_alignment(struct intel_crtc_state *crtc_st
  * cursor fully when cursor is in SU area.
  */
 static void
-intel_psr2_sel_fetch_et_alignment(struct intel_crtc_state *crtc_state,
-				  struct intel_plane_state *cursor_state)
+intel_psr2_sel_fetch_et_alignment(struct intel_atomic_state *state,
+				  struct intel_crtc *crtc)
 {
-	struct drm_rect inter;
+	struct intel_crtc_state *crtc_state = intel_atomic_get_new_crtc_state(state, crtc);
+	struct intel_plane_state *new_plane_state;
+	struct intel_plane *plane;
+	int i;
 
-	if (!crtc_state->enable_psr2_su_region_et ||
-	    !cursor_state->uapi.visible)
+	if (!crtc_state->enable_psr2_su_region_et)
 		return;
 
-	inter = crtc_state->psr2_su_area;
-	if (!drm_rect_intersect(&inter, &cursor_state->uapi.dst))
-		return;
+	for_each_new_intel_plane_in_state(state, plane, new_plane_state, i) {
+		struct drm_rect inter;
 
-	clip_area_update(&crtc_state->psr2_su_area, &cursor_state->uapi.dst,
-			 &crtc_state->pipe_src);
+		if (new_plane_state->uapi.crtc != crtc_state->uapi.crtc)
+			continue;
+
+		if (plane->id != PLANE_CURSOR)
+			continue;
+
+		if (!new_plane_state->uapi.visible)
+			continue;
+
+		inter = crtc_state->psr2_su_area;
+		if (!drm_rect_intersect(&inter, &new_plane_state->uapi.dst))
+			continue;
+
+		clip_area_update(&crtc_state->psr2_su_area, &new_plane_state->uapi.dst,
+				 &crtc_state->pipe_src);
+	}
 }
 
 /*
@@ -2152,8 +2188,7 @@ int intel_psr2_sel_fetch_update(struct intel_atomic_state *state,
 {
 	struct drm_i915_private *dev_priv = to_i915(state->base.dev);
 	struct intel_crtc_state *crtc_state = intel_atomic_get_new_crtc_state(state, crtc);
-	struct intel_plane_state *new_plane_state, *old_plane_state,
-		*cursor_plane_state = NULL;
+	struct intel_plane_state *new_plane_state, *old_plane_state;
 	struct intel_plane *plane;
 	bool full_update = false;
 	int i, ret;
@@ -2238,13 +2273,6 @@ int intel_psr2_sel_fetch_update(struct intel_atomic_state *state,
 		damaged_area.x2 += new_plane_state->uapi.dst.x1 - src.x1;
 
 		clip_area_update(&crtc_state->psr2_su_area, &damaged_area, &crtc_state->pipe_src);
-
-		/*
-		 * Cursor plane new state is stored to adjust su area to cover
-		 * cursor are fully.
-		 */
-		if (plane->id == PLANE_CURSOR)
-			cursor_plane_state = new_plane_state;
 	}
 
 	/*
@@ -2273,9 +2301,13 @@ int intel_psr2_sel_fetch_update(struct intel_atomic_state *state,
 	if (ret)
 		return ret;
 
-	/* Adjust su area to cover cursor fully as necessary */
-	if (cursor_plane_state)
-		intel_psr2_sel_fetch_et_alignment(crtc_state, cursor_plane_state);
+	/*
+	 * Adjust su area to cover cursor fully as necessary (early
+	 * transport). This needs to be done after
+	 * drm_atomic_add_affected_planes to ensure visible cursor is added into
+	 * affected planes even when cursor is not updated by itself.
+	 */
+	intel_psr2_sel_fetch_et_alignment(state, crtc);
 
 	intel_psr2_sel_fetch_pipe_alignment(crtc_state);
 
@@ -2338,6 +2370,8 @@ int intel_psr2_sel_fetch_update(struct intel_atomic_state *state,
 
 skip_sel_fetch_set_loop:
 	psr2_man_trk_ctl_calc(crtc_state, full_update);
+	crtc_state->pipe_srcsz_early_tpt =
+		psr2_pipe_srcsz_early_tpt_calc(crtc_state, full_update);
 	return 0;
 }
 
diff --git a/drivers/gpu/drm/i915/gt/gen8_ppgtt.c b/drivers/gpu/drm/i915/gt/gen8_ppgtt.c
index fa46d2308b0e..81bf2216371b 100644
--- a/drivers/gpu/drm/i915/gt/gen8_ppgtt.c
+++ b/drivers/gpu/drm/i915/gt/gen8_ppgtt.c
@@ -961,6 +961,9 @@ static int gen8_init_rsvd(struct i915_address_space *vm)
 	struct i915_vma *vma;
 	int ret;
 
+	if (!intel_gt_needs_wa_16018031267(vm->gt))
+		return 0;
+
 	/* The memory will be used only by GPU. */
 	obj = i915_gem_object_create_lmem(i915, PAGE_SIZE,
 					  I915_BO_ALLOC_VOLATILE |
diff --git a/drivers/gpu/drm/i915/gt/intel_engine_cs.c b/drivers/gpu/drm/i915/gt/intel_engine_cs.c
index 1ade568ffbfa..7a6dc371c384 100644
--- a/drivers/gpu/drm/i915/gt/intel_engine_cs.c
+++ b/drivers/gpu/drm/i915/gt/intel_engine_cs.c
@@ -908,6 +908,23 @@ static intel_engine_mask_t init_engine_mask(struct intel_gt *gt)
 		info->engine_mask &= ~BIT(GSC0);
 	}
 
+	/*
+	 * Do not create the command streamer for CCS slices beyond the first.
+	 * All the workload submitted to the first engine will be shared among
+	 * all the slices.
+	 *
+	 * Once the user will be allowed to customize the CCS mode, then this
+	 * check needs to be removed.
+	 */
+	if (IS_DG2(gt->i915)) {
+		u8 first_ccs = __ffs(CCS_MASK(gt));
+
+		/* Mask off all the CCS engine */
+		info->engine_mask &= ~GENMASK(CCS3, CCS0);
+		/* Put back in the first CCS engine */
+		info->engine_mask |= BIT(_CCS(first_ccs));
+	}
+
 	return info->engine_mask;
 }
 
diff --git a/drivers/gpu/drm/i915/gt/intel_gt.c b/drivers/gpu/drm/i915/gt/intel_gt.c
index a425db5ed3a2..6a2c2718bcc3 100644
--- a/drivers/gpu/drm/i915/gt/intel_gt.c
+++ b/drivers/gpu/drm/i915/gt/intel_gt.c
@@ -1024,6 +1024,12 @@ enum i915_map_type intel_gt_coherent_map_type(struct intel_gt *gt,
 		return I915_MAP_WC;
 }
 
+bool intel_gt_needs_wa_16018031267(struct intel_gt *gt)
+{
+	/* Wa_16018031267, Wa_16018063123 */
+	return IS_GFX_GT_IP_RANGE(gt, IP_VER(12, 55), IP_VER(12, 71));
+}
+
 bool intel_gt_needs_wa_22016122933(struct intel_gt *gt)
 {
 	return MEDIA_VER_FULL(gt->i915) == IP_VER(13, 0) && gt->type == GT_MEDIA;
diff --git a/drivers/gpu/drm/i915/gt/intel_gt.h b/drivers/gpu/drm/i915/gt/intel_gt.h
index 608f5c872928..003eb93b826f 100644
--- a/drivers/gpu/drm/i915/gt/intel_gt.h
+++ b/drivers/gpu/drm/i915/gt/intel_gt.h
@@ -82,17 +82,18 @@ struct drm_printer;
 		  ##__VA_ARGS__);					\
 } while (0)
 
-#define NEEDS_FASTCOLOR_BLT_WABB(engine) ( \
-	IS_GFX_GT_IP_RANGE(engine->gt, IP_VER(12, 55), IP_VER(12, 71)) && \
-	engine->class == COPY_ENGINE_CLASS && engine->instance == 0)
-
 static inline bool gt_is_root(struct intel_gt *gt)
 {
 	return !gt->info.id;
 }
 
+bool intel_gt_needs_wa_16018031267(struct intel_gt *gt);
 bool intel_gt_needs_wa_22016122933(struct intel_gt *gt);
 
+#define NEEDS_FASTCOLOR_BLT_WABB(engine) ( \
+	intel_gt_needs_wa_16018031267(engine->gt) && \
+	engine->class == COPY_ENGINE_CLASS && engine->instance == 0)
+
 static inline struct intel_gt *uc_to_gt(struct intel_uc *uc)
 {
 	return container_of(uc, struct intel_gt, uc);
diff --git a/drivers/gpu/drm/i915/gt/intel_gt_ccs_mode.c b/drivers/gpu/drm/i915/gt/intel_gt_ccs_mode.c
new file mode 100644
index 000000000000..044219c5960a
--- /dev/null
+++ b/drivers/gpu/drm/i915/gt/intel_gt_ccs_mode.c
@@ -0,0 +1,39 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright © 2024 Intel Corporation
+ */
+
+#include "i915_drv.h"
+#include "intel_gt.h"
+#include "intel_gt_ccs_mode.h"
+#include "intel_gt_regs.h"
+
+void intel_gt_apply_ccs_mode(struct intel_gt *gt)
+{
+	int cslice;
+	u32 mode = 0;
+	int first_ccs = __ffs(CCS_MASK(gt));
+
+	if (!IS_DG2(gt->i915))
+		return;
+
+	/* Build the value for the fixed CCS load balancing */
+	for (cslice = 0; cslice < I915_MAX_CCS; cslice++) {
+		if (CCS_MASK(gt) & BIT(cslice))
+			/*
+			 * If available, assign the cslice
+			 * to the first available engine...
+			 */
+			mode |= XEHP_CCS_MODE_CSLICE(cslice, first_ccs);
+
+		else
+			/*
+			 * ... otherwise, mark the cslice as
+			 * unavailable if no CCS dispatches here
+			 */
+			mode |= XEHP_CCS_MODE_CSLICE(cslice,
+						     XEHP_CCS_MODE_CSLICE_MASK);
+	}
+
+	intel_uncore_write(gt->uncore, XEHP_CCS_MODE, mode);
+}
diff --git a/drivers/gpu/drm/i915/gt/intel_gt_ccs_mode.h b/drivers/gpu/drm/i915/gt/intel_gt_ccs_mode.h
new file mode 100644
index 000000000000..9e5549caeb26
--- /dev/null
+++ b/drivers/gpu/drm/i915/gt/intel_gt_ccs_mode.h
@@ -0,0 +1,13 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright © 2024 Intel Corporation
+ */
+
+#ifndef __INTEL_GT_CCS_MODE_H__
+#define __INTEL_GT_CCS_MODE_H__
+
+struct intel_gt;
+
+void intel_gt_apply_ccs_mode(struct intel_gt *gt);
+
+#endif /* __INTEL_GT_CCS_MODE_H__ */
diff --git a/drivers/gpu/drm/i915/gt/intel_gt_regs.h b/drivers/gpu/drm/i915/gt/intel_gt_regs.h
index 50962cfd1353..743fe3566722 100644
--- a/drivers/gpu/drm/i915/gt/intel_gt_regs.h
+++ b/drivers/gpu/drm/i915/gt/intel_gt_regs.h
@@ -1477,8 +1477,14 @@
 #define   ECOBITS_PPGTT_CACHE4B			(0 << 8)
 
 #define GEN12_RCU_MODE				_MMIO(0x14800)
+#define   XEHP_RCU_MODE_FIXED_SLICE_CCS_MODE	REG_BIT(1)
 #define   GEN12_RCU_MODE_CCS_ENABLE		REG_BIT(0)
 
+#define XEHP_CCS_MODE				_MMIO(0x14804)
+#define   XEHP_CCS_MODE_CSLICE_MASK		REG_GENMASK(2, 0) /* CCS0-3 + rsvd */
+#define   XEHP_CCS_MODE_CSLICE_WIDTH		ilog2(XEHP_CCS_MODE_CSLICE_MASK + 1)
+#define   XEHP_CCS_MODE_CSLICE(cslice, ccs)	(ccs << (cslice * XEHP_CCS_MODE_CSLICE_WIDTH))
+
 #define CHV_FUSE_GT				_MMIO(VLV_GUNIT_BASE + 0x2168)
 #define   CHV_FGT_DISABLE_SS0			(1 << 10)
 #define   CHV_FGT_DISABLE_SS1			(1 << 11)
diff --git a/drivers/gpu/drm/i915/gt/intel_workarounds.c b/drivers/gpu/drm/i915/gt/intel_workarounds.c
index 25413809b9dc..6ec3582c9735 100644
--- a/drivers/gpu/drm/i915/gt/intel_workarounds.c
+++ b/drivers/gpu/drm/i915/gt/intel_workarounds.c
@@ -10,6 +10,7 @@
 #include "intel_engine_regs.h"
 #include "intel_gpu_commands.h"
 #include "intel_gt.h"
+#include "intel_gt_ccs_mode.h"
 #include "intel_gt_mcr.h"
 #include "intel_gt_print.h"
 #include "intel_gt_regs.h"
@@ -51,7 +52,8 @@
  *   registers belonging to BCS, VCS or VECS should be implemented in
  *   xcs_engine_wa_init(). Workarounds for registers not belonging to a specific
  *   engine's MMIO range but that are part of of the common RCS/CCS reset domain
- *   should be implemented in general_render_compute_wa_init().
+ *   should be implemented in general_render_compute_wa_init(). The settings
+ *   about the CCS load balancing should be added in ccs_engine_wa_mode().
  *
  * - GT workarounds: the list of these WAs is applied whenever these registers
  *   revert to their default values: on GPU reset, suspend/resume [1]_, etc.
@@ -2854,6 +2856,28 @@ add_render_compute_tuning_settings(struct intel_gt *gt,
 		wa_write_clr(wal, GEN8_GARBCNTL, GEN12_BUS_HASH_CTL_BIT_EXC);
 }
 
+static void ccs_engine_wa_mode(struct intel_engine_cs *engine, struct i915_wa_list *wal)
+{
+	struct intel_gt *gt = engine->gt;
+
+	if (!IS_DG2(gt->i915))
+		return;
+
+	/*
+	 * Wa_14019159160: This workaround, along with others, leads to
+	 * significant challenges in utilizing load balancing among the
+	 * CCS slices. Consequently, an architectural decision has been
+	 * made to completely disable automatic CCS load balancing.
+	 */
+	wa_masked_en(wal, GEN12_RCU_MODE, XEHP_RCU_MODE_FIXED_SLICE_CCS_MODE);
+
+	/*
+	 * After having disabled automatic load balancing we need to
+	 * assign all slices to a single CCS. We will call it CCS mode 1
+	 */
+	intel_gt_apply_ccs_mode(gt);
+}
+
 /*
  * The workarounds in this function apply to shared registers in
  * the general render reset domain that aren't tied to a
@@ -3004,8 +3028,10 @@ engine_init_workarounds(struct intel_engine_cs *engine, struct i915_wa_list *wal
 	 * to a single RCS/CCS engine's workaround list since
 	 * they're reset as part of the general render domain reset.
 	 */
-	if (engine->flags & I915_ENGINE_FIRST_RENDER_COMPUTE)
+	if (engine->flags & I915_ENGINE_FIRST_RENDER_COMPUTE) {
 		general_render_compute_wa_init(engine, wal);
+		ccs_engine_wa_mode(engine, wal);
+	}
 
 	if (engine->class == COMPUTE_CLASS)
 		ccs_engine_wa_init(engine, wal);
diff --git a/drivers/gpu/drm/nouveau/nouveau_uvmm.c b/drivers/gpu/drm/nouveau/nouveau_uvmm.c
index 0a0a11dc9ec0..ee02cd833c5e 100644
--- a/drivers/gpu/drm/nouveau/nouveau_uvmm.c
+++ b/drivers/gpu/drm/nouveau/nouveau_uvmm.c
@@ -812,15 +812,15 @@ op_remap(struct drm_gpuva_op_remap *r,
 	struct drm_gpuva_op_unmap *u = r->unmap;
 	struct nouveau_uvma *uvma = uvma_from_va(u->va);
 	u64 addr = uvma->va.va.addr;
-	u64 range = uvma->va.va.range;
+	u64 end = uvma->va.va.addr + uvma->va.va.range;
 
 	if (r->prev)
 		addr = r->prev->va.addr + r->prev->va.range;
 
 	if (r->next)
-		range = r->next->va.addr - addr;
+		end = r->next->va.addr;
 
-	op_unmap_range(u, addr, range);
+	op_unmap_range(u, addr, end - addr);
 }
 
 static int
diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf100.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf100.c
index 986e8d547c94..060c74a80eb1 100644
--- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf100.c
+++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf100.c
@@ -420,7 +420,7 @@ gf100_gr_chan_new(struct nvkm_gr *base, struct nvkm_chan *fifoch,
 			return ret;
 	} else {
 		ret = nvkm_memory_map(gr->attrib_cb, 0, chan->vmm, chan->attrib_cb,
-				      &args, sizeof(args));;
+				      &args, sizeof(args));
 		if (ret)
 			return ret;
 	}
diff --git a/drivers/gpu/drm/panfrost/panfrost_gpu.c b/drivers/gpu/drm/panfrost/panfrost_gpu.c
index 9063ce254642..fd8e44992184 100644
--- a/drivers/gpu/drm/panfrost/panfrost_gpu.c
+++ b/drivers/gpu/drm/panfrost/panfrost_gpu.c
@@ -441,19 +441,19 @@ void panfrost_gpu_power_off(struct panfrost_device *pfdev)
 
 	gpu_write(pfdev, SHADER_PWROFF_LO, pfdev->features.shader_present);
 	ret = readl_relaxed_poll_timeout(pfdev->iomem + SHADER_PWRTRANS_LO,
-					 val, !val, 1, 1000);
+					 val, !val, 1, 2000);
 	if (ret)
 		dev_err(pfdev->dev, "shader power transition timeout");
 
 	gpu_write(pfdev, TILER_PWROFF_LO, pfdev->features.tiler_present);
 	ret = readl_relaxed_poll_timeout(pfdev->iomem + TILER_PWRTRANS_LO,
-					 val, !val, 1, 1000);
+					 val, !val, 1, 2000);
 	if (ret)
 		dev_err(pfdev->dev, "tiler power transition timeout");
 
 	gpu_write(pfdev, L2_PWROFF_LO, pfdev->features.l2_present);
 	ret = readl_poll_timeout(pfdev->iomem + L2_PWRTRANS_LO,
-				 val, !val, 0, 1000);
+				 val, !val, 0, 2000);
 	if (ret)
 		dev_err(pfdev->dev, "l2 power transition timeout");
 }
diff --git a/drivers/gpu/drm/xe/xe_device.c b/drivers/gpu/drm/xe/xe_device.c
index ca85e81fdb44..d32ff3857e65 100644
--- a/drivers/gpu/drm/xe/xe_device.c
+++ b/drivers/gpu/drm/xe/xe_device.c
@@ -193,6 +193,9 @@ static void xe_device_destroy(struct drm_device *dev, void *dummy)
 {
 	struct xe_device *xe = to_xe_device(dev);
 
+	if (xe->preempt_fence_wq)
+		destroy_workqueue(xe->preempt_fence_wq);
+
 	if (xe->ordered_wq)
 		destroy_workqueue(xe->ordered_wq);
 
@@ -258,9 +261,15 @@ struct xe_device *xe_device_create(struct pci_dev *pdev,
 	INIT_LIST_HEAD(&xe->pinned.external_vram);
 	INIT_LIST_HEAD(&xe->pinned.evicted);
 
+	xe->preempt_fence_wq = alloc_ordered_workqueue("xe-preempt-fence-wq", 0);
 	xe->ordered_wq = alloc_ordered_workqueue("xe-ordered-wq", 0);
 	xe->unordered_wq = alloc_workqueue("xe-unordered-wq", 0, 0);
-	if (!xe->ordered_wq || !xe->unordered_wq) {
+	if (!xe->ordered_wq || !xe->unordered_wq ||
+	    !xe->preempt_fence_wq) {
+		/*
+		 * Cleanup done in xe_device_destroy via
+		 * drmm_add_action_or_reset register above
+		 */
 		drm_err(&xe->drm, "Failed to allocate xe workqueues\n");
 		err = -ENOMEM;
 		goto err;
diff --git a/drivers/gpu/drm/xe/xe_device_types.h b/drivers/gpu/drm/xe/xe_device_types.h
index 9785eef2e5a4..8e3a222b41cf 100644
--- a/drivers/gpu/drm/xe/xe_device_types.h
+++ b/drivers/gpu/drm/xe/xe_device_types.h
@@ -363,6 +363,9 @@ struct xe_device {
 	/** @ufence_wq: user fence wait queue */
 	wait_queue_head_t ufence_wq;
 
+	/** @preempt_fence_wq: used to serialize preempt fences */
+	struct workqueue_struct *preempt_fence_wq;
+
 	/** @ordered_wq: used to serialize compute mode resume */
 	struct workqueue_struct *ordered_wq;
 
diff --git a/drivers/gpu/drm/xe/xe_exec.c b/drivers/gpu/drm/xe/xe_exec.c
index 826c8b389672..cc5e0f75de3c 100644
--- a/drivers/gpu/drm/xe/xe_exec.c
+++ b/drivers/gpu/drm/xe/xe_exec.c
@@ -94,48 +94,16 @@
  *	Unlock all
  */
 
+/*
+ * Add validation and rebinding to the drm_exec locking loop, since both can
+ * trigger eviction which may require sleeping dma_resv locks.
+ */
 static int xe_exec_fn(struct drm_gpuvm_exec *vm_exec)
 {
 	struct xe_vm *vm = container_of(vm_exec->vm, struct xe_vm, gpuvm);
-	struct drm_gem_object *obj;
-	unsigned long index;
-	int num_fences;
-	int ret;
-
-	ret = drm_gpuvm_validate(vm_exec->vm, &vm_exec->exec);
-	if (ret)
-		return ret;
-
-	/*
-	 * 1 fence slot for the final submit, and 1 more for every per-tile for
-	 * GPU bind and 1 extra for CPU bind. Note that there are potentially
-	 * many vma per object/dma-resv, however the fence slot will just be
-	 * re-used, since they are largely the same timeline and the seqno
-	 * should be in order. In the case of CPU bind there is dummy fence used
-	 * for all CPU binds, so no need to have a per-tile slot for that.
-	 */
-	num_fences = 1 + 1 + vm->xe->info.tile_count;
 
-	/*
-	 * We don't know upfront exactly how many fence slots we will need at
-	 * the start of the exec, since the TTM bo_validate above can consume
-	 * numerous fence slots. Also due to how the dma_resv_reserve_fences()
-	 * works it only ensures that at least that many fence slots are
-	 * available i.e if there are already 10 slots available and we reserve
-	 * two more, it can just noop without reserving anything.  With this it
-	 * is quite possible that TTM steals some of the fence slots and then
-	 * when it comes time to do the vma binding and final exec stage we are
-	 * lacking enough fence slots, leading to some nasty BUG_ON() when
-	 * adding the fences. Hence just add our own fences here, after the
-	 * validate stage.
-	 */
-	drm_exec_for_each_locked_object(&vm_exec->exec, index, obj) {
-		ret = dma_resv_reserve_fences(obj->resv, num_fences);
-		if (ret)
-			return ret;
-	}
-
-	return 0;
+	/* The fence slot added here is intended for the exec sched job. */
+	return xe_vm_validate_rebind(vm, &vm_exec->exec, 1);
 }
 
 int xe_exec_ioctl(struct drm_device *dev, void *data, struct drm_file *file)
@@ -152,7 +120,6 @@ int xe_exec_ioctl(struct drm_device *dev, void *data, struct drm_file *file)
 	struct drm_exec *exec = &vm_exec.exec;
 	u32 i, num_syncs = 0, num_ufence = 0;
 	struct xe_sched_job *job;
-	struct dma_fence *rebind_fence;
 	struct xe_vm *vm;
 	bool write_locked, skip_retry = false;
 	ktime_t end = 0;
@@ -290,39 +257,7 @@ retry:
 		goto err_exec;
 	}
 
-	/*
-	 * Rebind any invalidated userptr or evicted BOs in the VM, non-compute
-	 * VM mode only.
-	 */
-	rebind_fence = xe_vm_rebind(vm, false);
-	if (IS_ERR(rebind_fence)) {
-		err = PTR_ERR(rebind_fence);
-		goto err_put_job;
-	}
-
-	/*
-	 * We store the rebind_fence in the VM so subsequent execs don't get
-	 * scheduled before the rebinds of userptrs / evicted BOs is complete.
-	 */
-	if (rebind_fence) {
-		dma_fence_put(vm->rebind_fence);
-		vm->rebind_fence = rebind_fence;
-	}
-	if (vm->rebind_fence) {
-		if (test_bit(DMA_FENCE_FLAG_SIGNALED_BIT,
-			     &vm->rebind_fence->flags)) {
-			dma_fence_put(vm->rebind_fence);
-			vm->rebind_fence = NULL;
-		} else {
-			dma_fence_get(vm->rebind_fence);
-			err = drm_sched_job_add_dependency(&job->drm,
-							   vm->rebind_fence);
-			if (err)
-				goto err_put_job;
-		}
-	}
-
-	/* Wait behind munmap style rebinds */
+	/* Wait behind rebinds */
 	if (!xe_vm_in_lr_mode(vm)) {
 		err = drm_sched_job_add_resv_dependencies(&job->drm,
 							  xe_vm_resv(vm),
diff --git a/drivers/gpu/drm/xe/xe_exec_queue_types.h b/drivers/gpu/drm/xe/xe_exec_queue_types.h
index 62b3d9d1d7cd..462b33195032 100644
--- a/drivers/gpu/drm/xe/xe_exec_queue_types.h
+++ b/drivers/gpu/drm/xe/xe_exec_queue_types.h
@@ -148,6 +148,11 @@ struct xe_exec_queue {
 	const struct xe_ring_ops *ring_ops;
 	/** @entity: DRM sched entity for this exec queue (1 to 1 relationship) */
 	struct drm_sched_entity *entity;
+	/**
+	 * @tlb_flush_seqno: The seqno of the last rebind tlb flush performed
+	 * Protected by @vm's resv. Unused if @vm == NULL.
+	 */
+	u64 tlb_flush_seqno;
 	/** @lrc: logical ring context for this exec queue */
 	struct xe_lrc lrc[];
 };
diff --git a/drivers/gpu/drm/xe/xe_gt_pagefault.c b/drivers/gpu/drm/xe/xe_gt_pagefault.c
index 241c294270d9..fa9e9853c53b 100644
--- a/drivers/gpu/drm/xe/xe_gt_pagefault.c
+++ b/drivers/gpu/drm/xe/xe_gt_pagefault.c
@@ -100,10 +100,9 @@ static int xe_pf_begin(struct drm_exec *exec, struct xe_vma *vma,
 {
 	struct xe_bo *bo = xe_vma_bo(vma);
 	struct xe_vm *vm = xe_vma_vm(vma);
-	unsigned int num_shared = 2; /* slots for bind + move */
 	int err;
 
-	err = xe_vm_prepare_vma(exec, vma, num_shared);
+	err = xe_vm_lock_vma(exec, vma);
 	if (err)
 		return err;
 
diff --git a/drivers/gpu/drm/xe/xe_gt_tlb_invalidation.c b/drivers/gpu/drm/xe/xe_gt_tlb_invalidation.c
index f03e077f81a0..e598a4363d01 100644
--- a/drivers/gpu/drm/xe/xe_gt_tlb_invalidation.c
+++ b/drivers/gpu/drm/xe/xe_gt_tlb_invalidation.c
@@ -61,7 +61,6 @@ int xe_gt_tlb_invalidation_init(struct xe_gt *gt)
 	INIT_LIST_HEAD(&gt->tlb_invalidation.pending_fences);
 	spin_lock_init(&gt->tlb_invalidation.pending_lock);
 	spin_lock_init(&gt->tlb_invalidation.lock);
-	gt->tlb_invalidation.fence_context = dma_fence_context_alloc(1);
 	INIT_DELAYED_WORK(&gt->tlb_invalidation.fence_tdr,
 			  xe_gt_tlb_fence_timeout);
 
diff --git a/drivers/gpu/drm/xe/xe_gt_types.h b/drivers/gpu/drm/xe/xe_gt_types.h
index 70c615dd1498..07b2f724ec45 100644
--- a/drivers/gpu/drm/xe/xe_gt_types.h
+++ b/drivers/gpu/drm/xe/xe_gt_types.h
@@ -177,13 +177,6 @@ struct xe_gt {
 		 * xe_gt_tlb_fence_timeout after the timeut interval is over.
 		 */
 		struct delayed_work fence_tdr;
-		/** @tlb_invalidation.fence_context: context for TLB invalidation fences */
-		u64 fence_context;
-		/**
-		 * @tlb_invalidation.fence_seqno: seqno to TLB invalidation fences, protected by
-		 * tlb_invalidation.lock
-		 */
-		u32 fence_seqno;
 		/** @tlb_invalidation.lock: protects TLB invalidation fences */
 		spinlock_t lock;
 	} tlb_invalidation;
diff --git a/drivers/gpu/drm/xe/xe_preempt_fence.c b/drivers/gpu/drm/xe/xe_preempt_fence.c
index 7bce2a332603..7d50c6e89d8e 100644
--- a/drivers/gpu/drm/xe/xe_preempt_fence.c
+++ b/drivers/gpu/drm/xe/xe_preempt_fence.c
@@ -49,7 +49,7 @@ static bool preempt_fence_enable_signaling(struct dma_fence *fence)
 	struct xe_exec_queue *q = pfence->q;
 
 	pfence->error = q->ops->suspend(q);
-	queue_work(system_unbound_wq, &pfence->preempt_work);
+	queue_work(q->vm->xe->preempt_fence_wq, &pfence->preempt_work);
 	return true;
 }
 
diff --git a/drivers/gpu/drm/xe/xe_pt.c b/drivers/gpu/drm/xe/xe_pt.c
index 7f54bc3e389d..4efc8c1a3d7a 100644
--- a/drivers/gpu/drm/xe/xe_pt.c
+++ b/drivers/gpu/drm/xe/xe_pt.c
@@ -1135,8 +1135,7 @@ static int invalidation_fence_init(struct xe_gt *gt,
 	spin_lock_irq(&gt->tlb_invalidation.lock);
 	dma_fence_init(&ifence->base.base, &invalidation_fence_ops,
 		       &gt->tlb_invalidation.lock,
-		       gt->tlb_invalidation.fence_context,
-		       ++gt->tlb_invalidation.fence_seqno);
+		       dma_fence_context_alloc(1), 1);
 	spin_unlock_irq(&gt->tlb_invalidation.lock);
 
 	INIT_LIST_HEAD(&ifence->base.link);
@@ -1236,6 +1235,13 @@ __xe_pt_bind_vma(struct xe_tile *tile, struct xe_vma *vma, struct xe_exec_queue
 	err = xe_pt_prepare_bind(tile, vma, entries, &num_entries);
 	if (err)
 		goto err;
+
+	err = dma_resv_reserve_fences(xe_vm_resv(vm), 1);
+	if (!err && !xe_vma_has_no_bo(vma) && !xe_vma_bo(vma)->vm)
+		err = dma_resv_reserve_fences(xe_vma_bo(vma)->ttm.base.resv, 1);
+	if (err)
+		goto err;
+
 	xe_tile_assert(tile, num_entries <= ARRAY_SIZE(entries));
 
 	xe_vm_dbg_print_entries(tile_to_xe(tile), entries, num_entries);
@@ -1254,11 +1260,13 @@ __xe_pt_bind_vma(struct xe_tile *tile, struct xe_vma *vma, struct xe_exec_queue
 	 * non-faulting LR, in particular on user-space batch buffer chaining,
 	 * it needs to be done here.
 	 */
-	if ((rebind && !xe_vm_in_lr_mode(vm) && !vm->batch_invalidate_tlb) ||
-	    (!rebind && xe_vm_has_scratch(vm) && xe_vm_in_preempt_fence_mode(vm))) {
+	if ((!rebind && xe_vm_has_scratch(vm) && xe_vm_in_preempt_fence_mode(vm))) {
 		ifence = kzalloc(sizeof(*ifence), GFP_KERNEL);
 		if (!ifence)
 			return ERR_PTR(-ENOMEM);
+	} else if (rebind && !xe_vm_in_lr_mode(vm)) {
+		/* We bump also if batch_invalidate_tlb is true */
+		vm->tlb_flush_seqno++;
 	}
 
 	rfence = kzalloc(sizeof(*rfence), GFP_KERNEL);
@@ -1297,7 +1305,7 @@ __xe_pt_bind_vma(struct xe_tile *tile, struct xe_vma *vma, struct xe_exec_queue
 		}
 
 		/* add shared fence now for pagetable delayed destroy */
-		dma_resv_add_fence(xe_vm_resv(vm), fence, !rebind &&
+		dma_resv_add_fence(xe_vm_resv(vm), fence, rebind ||
 				   last_munmap_rebind ?
 				   DMA_RESV_USAGE_KERNEL :
 				   DMA_RESV_USAGE_BOOKKEEP);
@@ -1576,6 +1584,7 @@ __xe_pt_unbind_vma(struct xe_tile *tile, struct xe_vma *vma, struct xe_exec_queu
 	struct dma_fence *fence = NULL;
 	struct invalidation_fence *ifence;
 	struct xe_range_fence *rfence;
+	int err;
 
 	LLIST_HEAD(deferred);
 
@@ -1593,6 +1602,12 @@ __xe_pt_unbind_vma(struct xe_tile *tile, struct xe_vma *vma, struct xe_exec_queu
 	xe_pt_calc_rfence_interval(vma, &unbind_pt_update, entries,
 				   num_entries);
 
+	err = dma_resv_reserve_fences(xe_vm_resv(vm), 1);
+	if (!err && !xe_vma_has_no_bo(vma) && !xe_vma_bo(vma)->vm)
+		err = dma_resv_reserve_fences(xe_vma_bo(vma)->ttm.base.resv, 1);
+	if (err)
+		return ERR_PTR(err);
+
 	ifence = kzalloc(sizeof(*ifence), GFP_KERNEL);
 	if (!ifence)
 		return ERR_PTR(-ENOMEM);
diff --git a/drivers/gpu/drm/xe/xe_ring_ops.c b/drivers/gpu/drm/xe/xe_ring_ops.c
index c4edffcd4a32..5b2b37b59813 100644
--- a/drivers/gpu/drm/xe/xe_ring_ops.c
+++ b/drivers/gpu/drm/xe/xe_ring_ops.c
@@ -219,10 +219,9 @@ static void __emit_job_gen12_simple(struct xe_sched_job *job, struct xe_lrc *lrc
 {
 	u32 dw[MAX_JOB_SIZE_DW], i = 0;
 	u32 ppgtt_flag = get_ppgtt_flag(job);
-	struct xe_vm *vm = job->q->vm;
 	struct xe_gt *gt = job->q->gt;
 
-	if (vm && vm->batch_invalidate_tlb) {
+	if (job->ring_ops_flush_tlb) {
 		dw[i++] = preparser_disable(true);
 		i = emit_flush_imm_ggtt(xe_lrc_start_seqno_ggtt_addr(lrc),
 					seqno, true, dw, i);
@@ -270,7 +269,6 @@ static void __emit_job_gen12_video(struct xe_sched_job *job, struct xe_lrc *lrc,
 	struct xe_gt *gt = job->q->gt;
 	struct xe_device *xe = gt_to_xe(gt);
 	bool decode = job->q->class == XE_ENGINE_CLASS_VIDEO_DECODE;
-	struct xe_vm *vm = job->q->vm;
 
 	dw[i++] = preparser_disable(true);
 
@@ -282,13 +280,13 @@ static void __emit_job_gen12_video(struct xe_sched_job *job, struct xe_lrc *lrc,
 			i = emit_aux_table_inv(gt, VE0_AUX_INV, dw, i);
 	}
 
-	if (vm && vm->batch_invalidate_tlb)
+	if (job->ring_ops_flush_tlb)
 		i = emit_flush_imm_ggtt(xe_lrc_start_seqno_ggtt_addr(lrc),
 					seqno, true, dw, i);
 
 	dw[i++] = preparser_disable(false);
 
-	if (!vm || !vm->batch_invalidate_tlb)
+	if (!job->ring_ops_flush_tlb)
 		i = emit_store_imm_ggtt(xe_lrc_start_seqno_ggtt_addr(lrc),
 					seqno, dw, i);
 
@@ -317,7 +315,6 @@ static void __emit_job_gen12_render_compute(struct xe_sched_job *job,
 	struct xe_gt *gt = job->q->gt;
 	struct xe_device *xe = gt_to_xe(gt);
 	bool lacks_render = !(gt->info.engine_mask & XE_HW_ENGINE_RCS_MASK);
-	struct xe_vm *vm = job->q->vm;
 	u32 mask_flags = 0;
 
 	dw[i++] = preparser_disable(true);
@@ -327,7 +324,7 @@ static void __emit_job_gen12_render_compute(struct xe_sched_job *job,
 		mask_flags = PIPE_CONTROL_3D_ENGINE_FLAGS;
 
 	/* See __xe_pt_bind_vma() for a discussion on TLB invalidations. */
-	i = emit_pipe_invalidate(mask_flags, vm && vm->batch_invalidate_tlb, dw, i);
+	i = emit_pipe_invalidate(mask_flags, job->ring_ops_flush_tlb, dw, i);
 
 	/* hsdes: 1809175790 */
 	if (has_aux_ccs(xe))
diff --git a/drivers/gpu/drm/xe/xe_sched_job.c b/drivers/gpu/drm/xe/xe_sched_job.c
index 8151ddafb940..b0c7fa4693cf 100644
--- a/drivers/gpu/drm/xe/xe_sched_job.c
+++ b/drivers/gpu/drm/xe/xe_sched_job.c
@@ -250,6 +250,16 @@ bool xe_sched_job_completed(struct xe_sched_job *job)
 
 void xe_sched_job_arm(struct xe_sched_job *job)
 {
+	struct xe_exec_queue *q = job->q;
+	struct xe_vm *vm = q->vm;
+
+	if (vm && !xe_sched_job_is_migration(q) && !xe_vm_in_lr_mode(vm) &&
+	    (vm->batch_invalidate_tlb || vm->tlb_flush_seqno != q->tlb_flush_seqno)) {
+		xe_vm_assert_held(vm);
+		q->tlb_flush_seqno = vm->tlb_flush_seqno;
+		job->ring_ops_flush_tlb = true;
+	}
+
 	drm_sched_job_arm(&job->drm);
 }
 
diff --git a/drivers/gpu/drm/xe/xe_sched_job_types.h b/drivers/gpu/drm/xe/xe_sched_job_types.h
index b1d83da50a53..5e12724219fd 100644
--- a/drivers/gpu/drm/xe/xe_sched_job_types.h
+++ b/drivers/gpu/drm/xe/xe_sched_job_types.h
@@ -39,6 +39,8 @@ struct xe_sched_job {
 	} user_fence;
 	/** @migrate_flush_flags: Additional flush flags for migration jobs */
 	u32 migrate_flush_flags;
+	/** @ring_ops_flush_tlb: The ring ops need to flush TLB before payload. */
+	bool ring_ops_flush_tlb;
 	/** @batch_addr: batch buffer address of job */
 	u64 batch_addr[];
 };
diff --git a/drivers/gpu/drm/xe/xe_vm.c b/drivers/gpu/drm/xe/xe_vm.c
index f88faef4142b..62d1ef8867a8 100644
--- a/drivers/gpu/drm/xe/xe_vm.c
+++ b/drivers/gpu/drm/xe/xe_vm.c
@@ -482,17 +482,53 @@ static int xe_gpuvm_validate(struct drm_gpuvm_bo *vm_bo, struct drm_exec *exec)
 	return 0;
 }
 
+/**
+ * xe_vm_validate_rebind() - Validate buffer objects and rebind vmas
+ * @vm: The vm for which we are rebinding.
+ * @exec: The struct drm_exec with the locked GEM objects.
+ * @num_fences: The number of fences to reserve for the operation, not
+ * including rebinds and validations.
+ *
+ * Validates all evicted gem objects and rebinds their vmas. Note that
+ * rebindings may cause evictions and hence the validation-rebind
+ * sequence is rerun until there are no more objects to validate.
+ *
+ * Return: 0 on success, negative error code on error. In particular,
+ * may return -EINTR or -ERESTARTSYS if interrupted, and -EDEADLK if
+ * the drm_exec transaction needs to be restarted.
+ */
+int xe_vm_validate_rebind(struct xe_vm *vm, struct drm_exec *exec,
+			  unsigned int num_fences)
+{
+	struct drm_gem_object *obj;
+	unsigned long index;
+	int ret;
+
+	do {
+		ret = drm_gpuvm_validate(&vm->gpuvm, exec);
+		if (ret)
+			return ret;
+
+		ret = xe_vm_rebind(vm, false);
+		if (ret)
+			return ret;
+	} while (!list_empty(&vm->gpuvm.evict.list));
+
+	drm_exec_for_each_locked_object(exec, index, obj) {
+		ret = dma_resv_reserve_fences(obj->resv, num_fences);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
 static int xe_preempt_work_begin(struct drm_exec *exec, struct xe_vm *vm,
 				 bool *done)
 {
 	int err;
 
-	/*
-	 * 1 fence for each preempt fence plus a fence for each tile from a
-	 * possible rebind
-	 */
-	err = drm_gpuvm_prepare_vm(&vm->gpuvm, exec, vm->preempt.num_exec_queues +
-				   vm->xe->info.tile_count);
+	err = drm_gpuvm_prepare_vm(&vm->gpuvm, exec, 0);
 	if (err)
 		return err;
 
@@ -507,7 +543,7 @@ static int xe_preempt_work_begin(struct drm_exec *exec, struct xe_vm *vm,
 		return 0;
 	}
 
-	err = drm_gpuvm_prepare_objects(&vm->gpuvm, exec, vm->preempt.num_exec_queues);
+	err = drm_gpuvm_prepare_objects(&vm->gpuvm, exec, 0);
 	if (err)
 		return err;
 
@@ -515,14 +551,19 @@ static int xe_preempt_work_begin(struct drm_exec *exec, struct xe_vm *vm,
 	if (err)
 		return err;
 
-	return drm_gpuvm_validate(&vm->gpuvm, exec);
+	/*
+	 * Add validation and rebinding to the locking loop since both can
+	 * cause evictions which may require blocing dma_resv locks.
+	 * The fence reservation here is intended for the new preempt fences
+	 * we attach at the end of the rebind work.
+	 */
+	return xe_vm_validate_rebind(vm, exec, vm->preempt.num_exec_queues);
 }
 
 static void preempt_rebind_work_func(struct work_struct *w)
 {
 	struct xe_vm *vm = container_of(w, struct xe_vm, preempt.rebind_work);
 	struct drm_exec exec;
-	struct dma_fence *rebind_fence;
 	unsigned int fence_count = 0;
 	LIST_HEAD(preempt_fences);
 	ktime_t end = 0;
@@ -568,18 +609,11 @@ retry:
 	if (err)
 		goto out_unlock;
 
-	rebind_fence = xe_vm_rebind(vm, true);
-	if (IS_ERR(rebind_fence)) {
-		err = PTR_ERR(rebind_fence);
+	err = xe_vm_rebind(vm, true);
+	if (err)
 		goto out_unlock;
-	}
 
-	if (rebind_fence) {
-		dma_fence_wait(rebind_fence, false);
-		dma_fence_put(rebind_fence);
-	}
-
-	/* Wait on munmap style VM unbinds */
+	/* Wait on rebinds and munmap style VM unbinds */
 	wait = dma_resv_wait_timeout(xe_vm_resv(vm),
 				     DMA_RESV_USAGE_KERNEL,
 				     false, MAX_SCHEDULE_TIMEOUT);
@@ -773,14 +807,14 @@ xe_vm_bind_vma(struct xe_vma *vma, struct xe_exec_queue *q,
 	       struct xe_sync_entry *syncs, u32 num_syncs,
 	       bool first_op, bool last_op);
 
-struct dma_fence *xe_vm_rebind(struct xe_vm *vm, bool rebind_worker)
+int xe_vm_rebind(struct xe_vm *vm, bool rebind_worker)
 {
-	struct dma_fence *fence = NULL;
+	struct dma_fence *fence;
 	struct xe_vma *vma, *next;
 
 	lockdep_assert_held(&vm->lock);
 	if (xe_vm_in_lr_mode(vm) && !rebind_worker)
-		return NULL;
+		return 0;
 
 	xe_vm_assert_held(vm);
 	list_for_each_entry_safe(vma, next, &vm->rebind_list,
@@ -788,17 +822,17 @@ struct dma_fence *xe_vm_rebind(struct xe_vm *vm, bool rebind_worker)
 		xe_assert(vm->xe, vma->tile_present);
 
 		list_del_init(&vma->combined_links.rebind);
-		dma_fence_put(fence);
 		if (rebind_worker)
 			trace_xe_vma_rebind_worker(vma);
 		else
 			trace_xe_vma_rebind_exec(vma);
 		fence = xe_vm_bind_vma(vma, NULL, NULL, 0, false, false);
 		if (IS_ERR(fence))
-			return fence;
+			return PTR_ERR(fence);
+		dma_fence_put(fence);
 	}
 
-	return fence;
+	return 0;
 }
 
 static void xe_vma_free(struct xe_vma *vma)
@@ -1004,35 +1038,26 @@ static void xe_vma_destroy(struct xe_vma *vma, struct dma_fence *fence)
 }
 
 /**
- * xe_vm_prepare_vma() - drm_exec utility to lock a vma
+ * xe_vm_lock_vma() - drm_exec utility to lock a vma
  * @exec: The drm_exec object we're currently locking for.
  * @vma: The vma for witch we want to lock the vm resv and any attached
  * object's resv.
- * @num_shared: The number of dma-fence slots to pre-allocate in the
- * objects' reservation objects.
  *
  * Return: 0 on success, negative error code on error. In particular
  * may return -EDEADLK on WW transaction contention and -EINTR if
  * an interruptible wait is terminated by a signal.
  */
-int xe_vm_prepare_vma(struct drm_exec *exec, struct xe_vma *vma,
-		      unsigned int num_shared)
+int xe_vm_lock_vma(struct drm_exec *exec, struct xe_vma *vma)
 {
 	struct xe_vm *vm = xe_vma_vm(vma);
 	struct xe_bo *bo = xe_vma_bo(vma);
 	int err;
 
 	XE_WARN_ON(!vm);
-	if (num_shared)
-		err = drm_exec_prepare_obj(exec, xe_vm_obj(vm), num_shared);
-	else
-		err = drm_exec_lock_obj(exec, xe_vm_obj(vm));
-	if (!err && bo && !bo->vm) {
-		if (num_shared)
-			err = drm_exec_prepare_obj(exec, &bo->ttm.base, num_shared);
-		else
-			err = drm_exec_lock_obj(exec, &bo->ttm.base);
-	}
+
+	err = drm_exec_lock_obj(exec, xe_vm_obj(vm));
+	if (!err && bo && !bo->vm)
+		err = drm_exec_lock_obj(exec, &bo->ttm.base);
 
 	return err;
 }
@@ -1044,7 +1069,7 @@ static void xe_vma_destroy_unlocked(struct xe_vma *vma)
 
 	drm_exec_init(&exec, 0, 0);
 	drm_exec_until_all_locked(&exec) {
-		err = xe_vm_prepare_vma(&exec, vma, 0);
+		err = xe_vm_lock_vma(&exec, vma);
 		drm_exec_retry_on_contention(&exec);
 		if (XE_WARN_ON(err))
 			break;
@@ -1589,7 +1614,6 @@ static void vm_destroy_work_func(struct work_struct *w)
 		XE_WARN_ON(vm->pt_root[id]);
 
 	trace_xe_vm_free(vm);
-	dma_fence_put(vm->rebind_fence);
 	kfree(vm);
 }
 
@@ -2512,7 +2536,7 @@ static int op_execute(struct drm_exec *exec, struct xe_vm *vm,
 
 	lockdep_assert_held_write(&vm->lock);
 
-	err = xe_vm_prepare_vma(exec, vma, 1);
+	err = xe_vm_lock_vma(exec, vma);
 	if (err)
 		return err;
 
diff --git a/drivers/gpu/drm/xe/xe_vm.h b/drivers/gpu/drm/xe/xe_vm.h
index 6df1f1c7f85d..306cd0934a19 100644
--- a/drivers/gpu/drm/xe/xe_vm.h
+++ b/drivers/gpu/drm/xe/xe_vm.h
@@ -207,7 +207,7 @@ int __xe_vm_userptr_needs_repin(struct xe_vm *vm);
 
 int xe_vm_userptr_check_repin(struct xe_vm *vm);
 
-struct dma_fence *xe_vm_rebind(struct xe_vm *vm, bool rebind_worker);
+int xe_vm_rebind(struct xe_vm *vm, bool rebind_worker);
 
 int xe_vm_invalidate_vma(struct xe_vma *vma);
 
@@ -242,8 +242,10 @@ bool xe_vm_validate_should_retry(struct drm_exec *exec, int err, ktime_t *end);
 
 int xe_analyze_vm(struct drm_printer *p, struct xe_vm *vm, int gt_id);
 
-int xe_vm_prepare_vma(struct drm_exec *exec, struct xe_vma *vma,
-		      unsigned int num_shared);
+int xe_vm_lock_vma(struct drm_exec *exec, struct xe_vma *vma);
+
+int xe_vm_validate_rebind(struct xe_vm *vm, struct drm_exec *exec,
+			  unsigned int num_fences);
 
 /**
  * xe_vm_resv() - Return's the vm's reservation object
diff --git a/drivers/gpu/drm/xe/xe_vm_types.h b/drivers/gpu/drm/xe/xe_vm_types.h
index ae5fb565f6bf..badf3945083d 100644
--- a/drivers/gpu/drm/xe/xe_vm_types.h
+++ b/drivers/gpu/drm/xe/xe_vm_types.h
@@ -177,9 +177,6 @@ struct xe_vm {
 	 */
 	struct list_head rebind_list;
 
-	/** @rebind_fence: rebind fence from execbuf */
-	struct dma_fence *rebind_fence;
-
 	/**
 	 * @destroy_work: worker to destroy VM, needed as a dma_fence signaling
 	 * from an irq context can be last put and the destroy needs to be able
@@ -264,6 +261,11 @@ struct xe_vm {
 		bool capture_once;
 	} error_capture;
 
+	/**
+	 * @tlb_flush_seqno: Required TLB flush seqno for the next exec.
+	 * protected by the vm resv.
+	 */
+	u64 tlb_flush_seqno;
 	/** @batch_invalidate_tlb: Always invalidate TLB before batch start */
 	bool batch_invalidate_tlb;
 	/** @xef: XE file handle for tracking this VM's drm client */