Merge tag 'drm-xe-next-fixes-2024-11-28' of https://gitlab.freedesktop.org/drm/xe/kernel into drm-next

Driver Changes:
- Update xe2 graphics name string (Matt Roper)
- Fix a couple of guc submit races (Matt Auld)
- Fix pat index usage in migrate (Matt Auld)
- Ensure non-cached migrate pagetable bo mappings (Matt Auld)
- Take a PM ref in the delayed snapshot capture worker (Matt Brost)

Signed-off-by: Dave Airlie <airlied@redhat.com>

From: Thomas Hellstrom <thomas.hellstrom@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/Z0iOjKwEGVo_DmgY@fedora

Author: airlied

drivers/gpu/drm/xe/xe_devcoredump.c

@@ -23,6 +23,7 @@
 #include "xe_guc_submit.h"
 #include "xe_hw_engine.h"
 #include "xe_module.h"
+#include "xe_pm.h"
 #include "xe_sched_job.h"
 #include "xe_vm.h"
 
@@ -158,8 +159,11 @@ static void xe_devcoredump_deferred_snap_work(struct work_struct *work)
 {
 	struct xe_devcoredump_snapshot *ss = container_of(work, typeof(*ss), work);
 	struct xe_devcoredump *coredump = container_of(ss, typeof(*coredump), snapshot);
+	struct xe_device *xe = coredump_to_xe(coredump);
 	unsigned int fw_ref;
 
+	xe_pm_runtime_get(xe);
+
 	/* keep going if fw fails as we still want to save the memory and SW data */
 	fw_ref = xe_force_wake_get(gt_to_fw(ss->gt), XE_FORCEWAKE_ALL);
 	if (!xe_force_wake_ref_has_domain(fw_ref, XE_FORCEWAKE_ALL))
@@ -168,6 +172,8 @@ static void xe_devcoredump_deferred_snap_work(struct work_struct *work)
 	xe_guc_exec_queue_snapshot_capture_delayed(ss->ge);
 	xe_force_wake_put(gt_to_fw(ss->gt), fw_ref);
 
+	xe_pm_runtime_put(xe);
+
 	/* Calculate devcoredump size */
 	ss->read.size = __xe_devcoredump_read(NULL, INT_MAX, coredump);
 

drivers/gpu/drm/xe/xe_guc_submit.c

@@ -769,17 +769,19 @@ static void disable_scheduling_deregister(struct xe_guc *guc,
 					  struct xe_exec_queue *q)
 {
 	MAKE_SCHED_CONTEXT_ACTION(q, DISABLE);
-	struct xe_device *xe = guc_to_xe(guc);
 	int ret;
 
 	set_min_preemption_timeout(guc, q);
 	smp_rmb();
-	ret = wait_event_timeout(guc->ct.wq, !exec_queue_pending_enable(q) ||
-				 xe_guc_read_stopped(guc), HZ * 5);
+	ret = wait_event_timeout(guc->ct.wq,
+				 (!exec_queue_pending_enable(q) &&
+				  !exec_queue_pending_disable(q)) ||
+					 xe_guc_read_stopped(guc),
+				 HZ * 5);
 	if (!ret) {
 		struct xe_gpu_scheduler *sched = &q->guc->sched;
 
-		drm_warn(&xe->drm, "Pending enable failed to respond");
+		xe_gt_warn(q->gt, "Pending enable/disable failed to respond\n");
 		xe_sched_submission_start(sched);
 		xe_gt_reset_async(q->gt);
 		xe_sched_tdr_queue_imm(sched);
@@ -1101,7 +1103,8 @@ guc_exec_queue_timedout_job(struct drm_sched_job *drm_job)
 			 * modifying state
 			 */
 			ret = wait_event_timeout(guc->ct.wq,
-						 !exec_queue_pending_enable(q) ||
+						 (!exec_queue_pending_enable(q) &&
+						  !exec_queue_pending_disable(q)) ||
 						 xe_guc_read_stopped(guc), HZ * 5);
 			if (!ret || xe_guc_read_stopped(guc))
 				goto trigger_reset;
@@ -1330,8 +1333,8 @@ static void __guc_exec_queue_process_msg_suspend(struct xe_sched_msg *msg)
 
 	if (guc_exec_queue_allowed_to_change_state(q) && !exec_queue_suspended(q) &&
 	    exec_queue_enabled(q)) {
-		wait_event(guc->ct.wq, q->guc->resume_time != RESUME_PENDING ||
-			   xe_guc_read_stopped(guc));
+		wait_event(guc->ct.wq, (q->guc->resume_time != RESUME_PENDING ||
+			   xe_guc_read_stopped(guc)) && !exec_queue_pending_disable(q));
 
 		if (!xe_guc_read_stopped(guc)) {
 			s64 since_resume_ms =
@@ -1868,16 +1871,29 @@ static void handle_sched_done(struct xe_guc *guc, struct xe_exec_queue *q,
 		xe_gt_assert(guc_to_gt(guc), runnable_state == 0);
 		xe_gt_assert(guc_to_gt(guc), exec_queue_pending_disable(q));
 
-		clear_exec_queue_pending_disable(q);
 		if (q->guc->suspend_pending) {
 			suspend_fence_signal(q);
+			clear_exec_queue_pending_disable(q);
 		} else {
 			if (exec_queue_banned(q) || check_timeout) {
 				smp_wmb();
 				wake_up_all(&guc->ct.wq);
 			}
-			if (!check_timeout)
+			if (!check_timeout && exec_queue_destroyed(q)) {
+				/*
+				 * Make sure to clear the pending_disable only
+				 * after sampling the destroyed state. We want
+				 * to ensure we don't trigger the unregister too
+				 * early with something intending to only
+				 * disable scheduling. The caller doing the
+				 * destroy must wait for an ongoing
+				 * pending_disable before marking as destroyed.
+				 */
+				clear_exec_queue_pending_disable(q);
 				deregister_exec_queue(guc, q);
+			} else {
+				clear_exec_queue_pending_disable(q);
+			}
 		}
 	}
 }

drivers/gpu/drm/xe/xe_migrate.c

@@ -209,7 +209,8 @@ static int xe_migrate_prepare_vm(struct xe_tile *tile, struct xe_migrate *m,
 				  num_entries * XE_PAGE_SIZE,
 				  ttm_bo_type_kernel,
 				  XE_BO_FLAG_VRAM_IF_DGFX(tile) |
-				  XE_BO_FLAG_PINNED);
+				  XE_BO_FLAG_PINNED |
+				  XE_BO_FLAG_PAGETABLE);
 	if (IS_ERR(bo))
 		return PTR_ERR(bo);
 
@@ -1350,6 +1351,7 @@ __xe_migrate_update_pgtables(struct xe_migrate *m,
 
 	/* For sysmem PTE's, need to map them in our hole.. */
 	if (!IS_DGFX(xe)) {
+		u16 pat_index = xe->pat.idx[XE_CACHE_WB];
 		u32 ptes, ofs;
 
 		ppgtt_ofs = NUM_KERNEL_PDE - 1;
@@ -1409,7 +1411,7 @@ __xe_migrate_update_pgtables(struct xe_migrate *m,
 						pt_bo->update_index = current_update;
 
 					addr = vm->pt_ops->pte_encode_bo(pt_bo, 0,
-									 XE_CACHE_WB, 0);
+									 pat_index, 0);
 					bb->cs[bb->len++] = lower_32_bits(addr);
 					bb->cs[bb->len++] = upper_32_bits(addr);
 				}

drivers/gpu/drm/xe/xe_pci.c

@@ -174,7 +174,7 @@ static const struct xe_graphics_desc graphics_xelpg = {
 		GENMASK(XE_HW_ENGINE_CCS3, XE_HW_ENGINE_CCS0)
 
 static const struct xe_graphics_desc graphics_xe2 = {
-	.name = "Xe2_LPG / Xe2_HPG",
+	.name = "Xe2_LPG / Xe2_HPG / Xe3_LPG",
 
 	XE2_GFX_FEATURES,
 };