Driver Changes:

- Increase global invalidation timeout to handle some workloads
    (Kenneth Graunke)
  - Fix NPD while evicting BOs in an array of VM binds (Matthew Brost)
  - Fix resizable BAR to account for possibly needing to move BARs other
    than the LMEMBAR (Lucas De Marchi)
  - Fix error handling in xe_migrate_init() (Thomas Hellström)
  - Fix atomic fault handling with mixed mappings or if the page is
    already in VRAM (Matthew Brost)
  - Enable media samplers power gating for platforms before Xe2 (Vinay
    Belgaumkar)
  - Fix de-registering exec queue from GuC when unbinding (Matthew Brost)
  - Ensure data migration to system if indicated by madvise with SVM
    (Thomas Hellström)
  - Fix kerneldoc for kunit change (Matt Roper)
  - Always account for cacheline alignment on migration (Matthew Auld)
  - Drop bogus assertion on eviction (Matthew Auld)
 -----BEGIN PGP SIGNATURE-----
 
 iQIzBAABCgAdFiEE6rM8lpABPHM5FqyDm6KlpjDL6lMFAmjxG4kACgkQm6KlpjDL
 6lMVSxAAi89fShCW8/H7TJUDfaQdX1qTg574x+4kmsxAe5IdZLKR17iKagGwhXYt
 pGPOHLL6s13trhSDT9RHxrQ/iUhlUMAf3HGZyeC0/X86QuKA0qGbrXoJTdexaA/V
 AyaXmyPCh4CsDP7o/QNfkmaH9Ze3tYniYPxKmQXIsbJbG6hK8jgREpE3UC0ilveX
 9rgA8t66W08CbPsHX8bLEgpQ6dchSZHOvHSaXvW3X1xDIi9P5kd2A3JPW9q+T15M
 84xtbxan6JDZx+xguIKimlUti6ihTSksxkAV6nKyg0I3n56iLarf0HN5MDM6ZExU
 8uS1ZmocaKqLji51LroIL+0X31H4VnQZlT/eehheBukW8SF6/jXEnq2PtxNy01Yi
 NJTCcwvvA0jMhK02tc9gcpHgJcmjp08lbymlZ0QdEp4gIQn5dpXubhcvdNeOmUK9
 NJMD8aE+9JnQ6iD8GFVjvdTSHKMpKtsNl2kUShOU3oK1KNHAqn/v3r4iM8VbGBff
 TaCxusNeVqFCcWkh4R58ppKdKiwLzitjc0xP9kjNFtGDVtPS11fluxQ+BhhrzFKk
 84wnhG8Lry7Ss5TpCAWjirxQOANx/q4Nef7uby6QAF9SLuon7Q2XU7ShLOlWIeTH
 AmtX57A8TxTrXa0Smn0rIP7/sYAdGfWDAdTDdJjAoJ36w8T2rgo=
 =BqxC
 -----END PGP SIGNATURE-----

Merge tag 'drm-xe-fixes-2025-10-16' of https://gitlab.freedesktop.org/drm/xe/kernel into drm-fixes

Driver Changes:
 - Increase global invalidation timeout to handle some workloads
   (Kenneth Graunke)
 - Fix NPD while evicting BOs in an array of VM binds (Matthew Brost)
 - Fix resizable BAR to account for possibly needing to move BARs other
   than the LMEMBAR (Lucas De Marchi)
 - Fix error handling in xe_migrate_init() (Thomas Hellström)
 - Fix atomic fault handling with mixed mappings or if the page is
   already in VRAM (Matthew Brost)
 - Enable media samplers power gating for platforms before Xe2 (Vinay
   Belgaumkar)
 - Fix de-registering exec queue from GuC when unbinding (Matthew Brost)
 - Ensure data migration to system if indicated by madvise with SVM
   (Thomas Hellström)
 - Fix kerneldoc for kunit change (Matt Roper)
 - Always account for cacheline alignment on migration (Matthew Auld)
 - Drop bogus assertion on eviction (Matthew Auld)

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

From: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/rch735eqkmprfyutk3ux2fsqa3e5ve4p77w7a5j66qdpgyquxr@ao3wzcqtpn6s

drivers/gpu/drm/xe/regs/xe_gt_regs.h

@@ -342,6 +342,7 @@
 #define POWERGATE_ENABLE			XE_REG(0xa210)
 #define   RENDER_POWERGATE_ENABLE		REG_BIT(0)
 #define   MEDIA_POWERGATE_ENABLE		REG_BIT(1)
+#define   MEDIA_SAMPLERS_POWERGATE_ENABLE	REG_BIT(2)
 #define   VDN_HCP_POWERGATE_ENABLE(n)		REG_BIT(3 + 2 * (n))
 #define   VDN_MFXVDENC_POWERGATE_ENABLE(n)	REG_BIT(4 + 2 * (n))
 

drivers/gpu/drm/xe/tests/xe_pci.c

@@ -66,6 +66,7 @@ KUNIT_ARRAY_PARAM(platform, cases, xe_pci_fake_data_desc);
 
 /**
  * xe_pci_fake_data_gen_params - Generate struct xe_pci_fake_data parameters
+ * @test: test context object
  * @prev: the pointer to the previous parameter to iterate from or NULL
  * @desc: output buffer with minimum size of KUNIT_PARAM_DESC_SIZE
  *
@@ -242,6 +243,7 @@ KUNIT_ARRAY_PARAM(pci_id, pciidlist, xe_pci_id_kunit_desc);
 
 /**
  * xe_pci_graphics_ip_gen_param - Generate graphics struct xe_ip parameters
+ * @test: test context object
  * @prev: the pointer to the previous parameter to iterate from or NULL
  * @desc: output buffer with minimum size of KUNIT_PARAM_DESC_SIZE
  *
@@ -266,6 +268,7 @@ EXPORT_SYMBOL_IF_KUNIT(xe_pci_graphics_ip_gen_param);
 
 /**
  * xe_pci_media_ip_gen_param - Generate media struct xe_ip parameters
+ * @test: test context object
  * @prev: the pointer to the previous parameter to iterate from or NULL
  * @desc: output buffer with minimum size of KUNIT_PARAM_DESC_SIZE
  *
@@ -290,6 +293,7 @@ EXPORT_SYMBOL_IF_KUNIT(xe_pci_media_ip_gen_param);
 
 /**
  * xe_pci_id_gen_param - Generate struct pci_device_id parameters
+ * @test: test context object
  * @prev: the pointer to the previous parameter to iterate from or NULL
  * @desc: output buffer with minimum size of KUNIT_PARAM_DESC_SIZE
  *
@@ -376,6 +380,7 @@ EXPORT_SYMBOL_IF_KUNIT(xe_pci_fake_device_init);
 
 /**
  * xe_pci_live_device_gen_param - Helper to iterate Xe devices as KUnit parameters
+ * @test: test context object
  * @prev: the previously returned value, or NULL for the first iteration
  * @desc: the buffer for a parameter name
  *

drivers/gpu/drm/xe/xe_bo_evict.c

@@ -182,7 +182,6 @@ int xe_bo_evict_all(struct xe_device *xe)
 
 static int xe_bo_restore_and_map_ggtt(struct xe_bo *bo)
 {
-	struct xe_device *xe = xe_bo_device(bo);
 	int ret;
 
 	ret = xe_bo_restore_pinned(bo);
@@ -201,13 +200,6 @@ static int xe_bo_restore_and_map_ggtt(struct xe_bo *bo)
 		}
 	}
 
-	/*
-	 * We expect validate to trigger a move VRAM and our move code
-	 * should setup the iosys map.
-	 */
-	xe_assert(xe, !(bo->flags & XE_BO_FLAG_PINNED_LATE_RESTORE) ||
-		  !iosys_map_is_null(&bo->vmap));
-
 	return 0;
 }
 

drivers/gpu/drm/xe/xe_device.c

@@ -1070,7 +1070,7 @@ void xe_device_l2_flush(struct xe_device *xe)
 	spin_lock(&gt->global_invl_lock);
 
 	xe_mmio_write32(&gt->mmio, XE2_GLOBAL_INVAL, 0x1);
-	if (xe_mmio_wait32(&gt->mmio, XE2_GLOBAL_INVAL, 0x1, 0x0, 500, NULL, true))
+	if (xe_mmio_wait32(&gt->mmio, XE2_GLOBAL_INVAL, 0x1, 0x0, 1000, NULL, true))
 		xe_gt_err_once(gt, "Global invalidation timeout\n");
 
 	spin_unlock(&gt->global_invl_lock);

drivers/gpu/drm/xe/xe_gt_idle.c

@@ -124,6 +124,9 @@ void xe_gt_idle_enable_pg(struct xe_gt *gt)
 	if (xe_gt_is_main_type(gt))
 		gtidle->powergate_enable |= RENDER_POWERGATE_ENABLE;
 
+	if (MEDIA_VERx100(xe) >= 1100 && MEDIA_VERx100(xe) < 1255)
+		gtidle->powergate_enable |= MEDIA_SAMPLERS_POWERGATE_ENABLE;
+
 	if (xe->info.platform != XE_DG1) {
 		for (i = XE_HW_ENGINE_VCS0, j = 0; i <= XE_HW_ENGINE_VCS7; ++i, ++j) {
 			if ((gt->info.engine_mask & BIT(i)))
@@ -246,6 +249,11 @@ int xe_gt_idle_pg_print(struct xe_gt *gt, struct drm_printer *p)
 				drm_printf(p, "Media Slice%d Power Gate Status: %s\n", n,
 					   str_up_down(pg_status & media_slices[n].status_bit));
 	}
+
+	if (MEDIA_VERx100(xe) >= 1100 && MEDIA_VERx100(xe) < 1255)
+		drm_printf(p, "Media Samplers Power Gating Enabled: %s\n",
+			   str_yes_no(pg_enabled & MEDIA_SAMPLERS_POWERGATE_ENABLE));
+
 	return 0;
 }
 

drivers/gpu/drm/xe/xe_guc_submit.c

@@ -44,6 +44,7 @@
 #include "xe_ring_ops_types.h"
 #include "xe_sched_job.h"
 #include "xe_trace.h"
+#include "xe_uc_fw.h"
 #include "xe_vm.h"
 
 static struct xe_guc *
@@ -1489,7 +1490,17 @@ static void __guc_exec_queue_process_msg_cleanup(struct xe_sched_msg *msg)
 	xe_gt_assert(guc_to_gt(guc), !(q->flags & EXEC_QUEUE_FLAG_PERMANENT));
 	trace_xe_exec_queue_cleanup_entity(q);
 
-	if (exec_queue_registered(q))
+	/*
+	 * Expected state transitions for cleanup:
+	 * - If the exec queue is registered and GuC firmware is running, we must first
+	 *   disable scheduling and deregister the queue to ensure proper teardown and
+	 *   resource release in the GuC, then destroy the exec queue on driver side.
+	 * - If the GuC is already stopped (e.g., during driver unload or GPU reset),
+	 *   we cannot expect a response for the deregister request. In this case,
+	 *   it is safe to directly destroy the exec queue on driver side, as the GuC
+	 *   will not process further requests and all resources must be cleaned up locally.
+	 */
+	if (exec_queue_registered(q) && xe_uc_fw_is_running(&guc->fw))
 		disable_scheduling_deregister(guc, q);
 	else
 		__guc_exec_queue_destroy(guc, q);

drivers/gpu/drm/xe/xe_migrate.c

@@ -434,7 +434,7 @@ int xe_migrate_init(struct xe_migrate *m)
 
 	err = xe_migrate_lock_prepare_vm(tile, m, vm);
 	if (err)
-		return err;
+		goto err_out;
 
 	if (xe->info.has_usm) {
 		struct xe_hw_engine *hwe = xe_gt_hw_engine(primary_gt,
@@ -2113,7 +2113,9 @@ int xe_migrate_access_memory(struct xe_migrate *m, struct xe_bo *bo,
 		if (current_bytes & ~PAGE_MASK) {
 			int pitch = 4;
 
-			current_bytes = min_t(int, current_bytes, S16_MAX * pitch);
+			current_bytes = min_t(int, current_bytes,
+					      round_down(S16_MAX * pitch,
+							 XE_CACHELINE_BYTES));
 		}
 
 		__fence = xe_migrate_vram(m, current_bytes,

drivers/gpu/drm/xe/xe_pci.c

@@ -867,6 +867,8 @@ static int xe_pci_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
 	if (err)
 		return err;
 
+	xe_vram_resize_bar(xe);
+
 	err = xe_device_probe_early(xe);
 	/*
 	 * In Boot Survivability mode, no drm card is exposed and driver

drivers/gpu/drm/xe/xe_svm.c

@@ -1034,6 +1034,9 @@ retry:
 	if (err)
 		return err;
 
+	dpagemap = xe_vma_resolve_pagemap(vma, tile);
+	if (!dpagemap && !ctx.devmem_only)
+		ctx.device_private_page_owner = NULL;
 	range = xe_svm_range_find_or_insert(vm, fault_addr, vma, &ctx);
 
 	if (IS_ERR(range))
@@ -1054,7 +1057,6 @@ retry:
 
 	range_debug(range, "PAGE FAULT");
 
-	dpagemap = xe_vma_resolve_pagemap(vma, tile);
 	if (--migrate_try_count >= 0 &&
 	    xe_svm_range_needs_migrate_to_vram(range, vma, !!dpagemap || ctx.devmem_only)) {
 		ktime_t migrate_start = xe_svm_stats_ktime_get();
@@ -1073,7 +1075,17 @@ retry:
 				drm_dbg(&vm->xe->drm,
 					"VRAM allocation failed, falling back to retrying fault, asid=%u, errno=%pe\n",
 					vm->usm.asid, ERR_PTR(err));
-				goto retry;
+
+				/*
+				 * In the devmem-only case, mixed mappings may
+				 * be found. The get_pages function will fix
+				 * these up to a single location, allowing the
+				 * page fault handler to make forward progress.
+				 */
+				if (ctx.devmem_only)
+					goto get_pages;
+				else
+					goto retry;
 			} else {
 				drm_err(&vm->xe->drm,
 					"VRAM allocation failed, retry count exceeded, asid=%u, errno=%pe\n",
@@ -1083,6 +1095,7 @@ retry:
 		}
 	}
 
+get_pages:
 	get_pages_start = xe_svm_stats_ktime_get();
 
 	range_debug(range, "GET PAGES");

drivers/gpu/drm/xe/xe_vm.c

@@ -2832,7 +2832,7 @@ static void vm_bind_ioctl_ops_unwind(struct xe_vm *vm,
 }
 
 static int vma_lock_and_validate(struct drm_exec *exec, struct xe_vma *vma,
-				 bool validate)
+				 bool res_evict, bool validate)
 {
 	struct xe_bo *bo = xe_vma_bo(vma);
 	struct xe_vm *vm = xe_vma_vm(vma);
@@ -2843,7 +2843,8 @@ static int vma_lock_and_validate(struct drm_exec *exec, struct xe_vma *vma,
 			err = drm_exec_lock_obj(exec, &bo->ttm.base);
 		if (!err && validate)
 			err = xe_bo_validate(bo, vm,
-					     !xe_vm_in_preempt_fence_mode(vm), exec);
+					     !xe_vm_in_preempt_fence_mode(vm) &&
+					     res_evict, exec);
 	}
 
 	return err;
@@ -2913,14 +2914,23 @@ static int prefetch_ranges(struct xe_vm *vm, struct xe_vma_op *op)
 }
 
 static int op_lock_and_prep(struct drm_exec *exec, struct xe_vm *vm,
-			    struct xe_vma_op *op)
+			    struct xe_vma_ops *vops, struct xe_vma_op *op)
 {
 	int err = 0;
+	bool res_evict;
+
+	/*
+	 * We only allow evicting a BO within the VM if it is not part of an
+	 * array of binds, as an array of binds can evict another BO within the
+	 * bind.
+	 */
+	res_evict = !(vops->flags & XE_VMA_OPS_ARRAY_OF_BINDS);
 
 	switch (op->base.op) {
 	case DRM_GPUVA_OP_MAP:
 		if (!op->map.invalidate_on_bind)
 			err = vma_lock_and_validate(exec, op->map.vma,
+						    res_evict,
 						    !xe_vm_in_fault_mode(vm) ||
 						    op->map.immediate);
 		break;
@@ -2931,11 +2941,13 @@ static int op_lock_and_prep(struct drm_exec *exec, struct xe_vm *vm,
 
 		err = vma_lock_and_validate(exec,
 					    gpuva_to_vma(op->base.remap.unmap->va),
-					    false);
+					    res_evict, false);
 		if (!err && op->remap.prev)
-			err = vma_lock_and_validate(exec, op->remap.prev, true);
+			err = vma_lock_and_validate(exec, op->remap.prev,
+						    res_evict, true);
 		if (!err && op->remap.next)
-			err = vma_lock_and_validate(exec, op->remap.next, true);
+			err = vma_lock_and_validate(exec, op->remap.next,
+						    res_evict, true);
 		break;
 	case DRM_GPUVA_OP_UNMAP:
 		err = check_ufence(gpuva_to_vma(op->base.unmap.va));
@@ -2944,7 +2956,7 @@ static int op_lock_and_prep(struct drm_exec *exec, struct xe_vm *vm,
 
 		err = vma_lock_and_validate(exec,
 					    gpuva_to_vma(op->base.unmap.va),
-					    false);
+					    res_evict, false);
 		break;
 	case DRM_GPUVA_OP_PREFETCH:
 	{
@@ -2959,7 +2971,7 @@ static int op_lock_and_prep(struct drm_exec *exec, struct xe_vm *vm,
 
 		err = vma_lock_and_validate(exec,
 					    gpuva_to_vma(op->base.prefetch.va),
-					    false);
+					    res_evict, false);
 		if (!err && !xe_vma_has_no_bo(vma))
 			err = xe_bo_migrate(xe_vma_bo(vma),
 					    region_to_mem_type[region],
@@ -3005,7 +3017,7 @@ static int vm_bind_ioctl_ops_lock_and_prep(struct drm_exec *exec,
 		return err;
 
 	list_for_each_entry(op, &vops->list, link) {
-		err = op_lock_and_prep(exec, vm, op);
+		err = op_lock_and_prep(exec, vm, vops, op);
 		if (err)
 			return err;
 	}
@@ -3638,6 +3650,8 @@ int xe_vm_bind_ioctl(struct drm_device *dev, void *data, struct drm_file *file)
 	}
 
 	xe_vma_ops_init(&vops, vm, q, syncs, num_syncs);
+	if (args->num_binds > 1)
+		vops.flags |= XE_VMA_OPS_ARRAY_OF_BINDS;
 	for (i = 0; i < args->num_binds; ++i) {
 		u64 range = bind_ops[i].range;
 		u64 addr = bind_ops[i].addr;

drivers/gpu/drm/xe/xe_vm_types.h

@@ -476,6 +476,7 @@ struct xe_vma_ops {
 	/** @flag: signify the properties within xe_vma_ops*/
 #define XE_VMA_OPS_FLAG_HAS_SVM_PREFETCH BIT(0)
 #define XE_VMA_OPS_FLAG_MADVISE          BIT(1)
+#define XE_VMA_OPS_ARRAY_OF_BINDS	 BIT(2)
 	u32 flags;
 #ifdef TEST_VM_OPS_ERROR
 	/** @inject_error: inject error to test error handling */

drivers/gpu/drm/xe/xe_vram.c

@@ -26,15 +26,35 @@
 
 #define BAR_SIZE_SHIFT 20
 
-static void
+/*
-_resize_bar(struct xe_device *xe, int resno, resource_size_t size)
+ * Release all the BARs that could influence/block LMEMBAR resizing, i.e.
+ * assigned IORESOURCE_MEM_64 BARs
+ */
+static void release_bars(struct pci_dev *pdev)
+{
+	struct resource *res;
+	int i;
+
+	pci_dev_for_each_resource(pdev, res, i) {
+		/* Resource already un-assigned, do not reset it */
+		if (!res->parent)
+			continue;
+
+		/* No need to release unrelated BARs */
+		if (!(res->flags & IORESOURCE_MEM_64))
+			continue;
+
+		pci_release_resource(pdev, i);
+	}
+}
+
+static void resize_bar(struct xe_device *xe, int resno, resource_size_t size)
 {
 	struct pci_dev *pdev = to_pci_dev(xe->drm.dev);
 	int bar_size = pci_rebar_bytes_to_size(size);
 	int ret;
 
-	if (pci_resource_len(pdev, resno))
+	release_bars(pdev);
-		pci_release_resource(pdev, resno);
 
 	ret = pci_resize_resource(pdev, resno, bar_size);
 	if (ret) {
@@ -50,7 +70,7 @@ _resize_bar(struct xe_device *xe, int resno, resource_size_t size)
  * if force_vram_bar_size is set, attempt to set to the requested size
  * else set to maximum possible size
  */
-static void resize_vram_bar(struct xe_device *xe)
+void xe_vram_resize_bar(struct xe_device *xe)
 {
 	int force_vram_bar_size = xe_modparam.force_vram_bar_size;
 	struct pci_dev *pdev = to_pci_dev(xe->drm.dev);
@@ -119,7 +139,7 @@ static void resize_vram_bar(struct xe_device *xe)
 	pci_read_config_dword(pdev, PCI_COMMAND, &pci_cmd);
 	pci_write_config_dword(pdev, PCI_COMMAND, pci_cmd & ~PCI_COMMAND_MEMORY);
 
-	_resize_bar(xe, LMEM_BAR, rebar_size);
+	resize_bar(xe, LMEM_BAR, rebar_size);
 
 	pci_assign_unassigned_bus_resources(pdev->bus);
 	pci_write_config_dword(pdev, PCI_COMMAND, pci_cmd);
@@ -148,8 +168,6 @@ static int determine_lmem_bar_size(struct xe_device *xe, struct xe_vram_region *
 		return -ENXIO;
 	}
 
-	resize_vram_bar(xe);
-
 	lmem_bar->io_start = pci_resource_start(pdev, LMEM_BAR);
 	lmem_bar->io_size = pci_resource_len(pdev, LMEM_BAR);
 	if (!lmem_bar->io_size)

drivers/gpu/drm/xe/xe_vram.h

@@ -11,6 +11,7 @@
 struct xe_device;
 struct xe_vram_region;
 
+void xe_vram_resize_bar(struct xe_device *xe);
 int xe_vram_probe(struct xe_device *xe);
 
 struct xe_vram_region *xe_vram_region_alloc(struct xe_device *xe, u8 id, u32 placement);