Merge: RHEL 10.1: DRM Stable Backport (v6.15.5)

MR: https://gitlab.com/redhat/centos-stream/src/kernel/centos-stream-10/-/merge_requests/1162

## Overview

The DRM backport's goal is to backport all the changes in the DRM subsystem to the kernel target version, with the biggest value being that we get the upstream hardware enablement (and bug fixes) into RHEL.

After the main DRM Backport, this MR backport as many fixes from the linux-stable tree for the target version: v6.15.5.

## Dependencies:

This MR depends on the DRM Backport for 6.15:

Depends: https://gitlab.com/redhat/centos-stream/src/kernel/centos-stream-10/-/merge_requests/834

Depends: https://gitlab.com/redhat/centos-stream/src/kernel/centos-stream-10/-/merge_requests/1124

And in a ACPI fix required for AMD Kraken and AMD Strix:

Depends: https://gitlab.com/redhat/centos-stream/src/kernel/centos-stream-10/-/merge_requests/1179

## Other:

JIRA: https://issues.redhat.com/browse/RHEL-75958

Signed-off-by: José Expósito <jexposit@redhat.com>

Approved-by: Jan Stancek <jstancek@redhat.com>
Approved-by: Eric Chanudet <echanude@redhat.com>
Approved-by: Jocelyn Falempe <jfalempe@redhat.com>
Approved-by: David Airlie <airlied@redhat.com>
Approved-by: CKI KWF Bot <cki-ci-bot+kwf-gitlab-com@redhat.com>

Merged-by: CKI GitLab Kmaint Pipeline Bot <26919896-cki-kmaint-pipeline-bot@users.noreply.gitlab.com>

Author: CKI KWF Bot

Documentation/gpu/xe/index.rst

@@ -16,6 +16,7 @@ DG2, etc is provided to prototype the driver.
    xe_migrate
    xe_cs
    xe_pm
+   xe_gt_freq
    xe_pcode
    xe_gt_mcr
    xe_wa

Documentation/gpu/xe/xe_gt_freq.rst

@@ -0,0 +1,14 @@
+.. SPDX-License-Identifier: (GPL-2.0+ OR MIT)
+
+==========================
+Xe GT Frequency Management
+==========================
+
+.. kernel-doc:: drivers/gpu/drm/xe/xe_gt_freq.c
+   :doc: Xe GT Frequency Management
+
+Internal API
+============
+
+.. kernel-doc:: drivers/gpu/drm/xe/xe_gt_freq.c
+   :internal:

Makefile

@@ -9,7 +9,7 @@ NAME = Baby Opossum Posse
 #
 RHEL_DRM_VERSION = 6
 RHEL_DRM_PATCHLEVEL = 15
-RHEL_DRM_SUBLEVEL =
+RHEL_DRM_SUBLEVEL = 5
 
 # *DOCUMENTATION*
 # To see a list of typical targets execute "make help"

drivers/accel/ivpu/ivpu_fw.c

@@ -55,18 +55,18 @@ static struct {
 	int gen;
 	const char *name;
 } fw_names[] = {
-	{ IVPU_HW_IP_37XX, "vpu_37xx.bin" },
+	{ IVPU_HW_IP_37XX, "intel/vpu/vpu_37xx_v1.bin" },
 	{ IVPU_HW_IP_37XX, "intel/vpu/vpu_37xx_v0.0.bin" },
-	{ IVPU_HW_IP_40XX, "vpu_40xx.bin" },
+	{ IVPU_HW_IP_40XX, "intel/vpu/vpu_40xx_v1.bin" },
 	{ IVPU_HW_IP_40XX, "intel/vpu/vpu_40xx_v0.0.bin" },
-	{ IVPU_HW_IP_50XX, "vpu_50xx.bin" },
+	{ IVPU_HW_IP_50XX, "intel/vpu/vpu_50xx_v1.bin" },
 	{ IVPU_HW_IP_50XX, "intel/vpu/vpu_50xx_v0.0.bin" },
 };
 
 /* Production fw_names from the table above */
-MODULE_FIRMWARE("intel/vpu/vpu_37xx_v0.0.bin");
-MODULE_FIRMWARE("intel/vpu/vpu_40xx_v0.0.bin");
-MODULE_FIRMWARE("intel/vpu/vpu_50xx_v0.0.bin");
+MODULE_FIRMWARE("intel/vpu/vpu_37xx_v1.bin");
+MODULE_FIRMWARE("intel/vpu/vpu_40xx_v1.bin");
+MODULE_FIRMWARE("intel/vpu/vpu_50xx_v1.bin");
 
 static int ivpu_fw_request(struct ivpu_device *vdev)
 {

drivers/accel/ivpu/ivpu_gem.c

@@ -28,11 +28,21 @@ static inline void ivpu_dbg_bo(struct ivpu_device *vdev, struct ivpu_bo *bo, con
 {
 	ivpu_dbg(vdev, BO,
 		 "%6s: bo %8p vpu_addr %9llx size %8zu ctx %d has_pages %d dma_mapped %d mmu_mapped %d wc %d imported %d\n",
-		 action, bo, bo->vpu_addr, ivpu_bo_size(bo), bo->ctx ? bo->ctx->id : 0,
+		 action, bo, bo->vpu_addr, ivpu_bo_size(bo), bo->ctx_id,
 		 (bool)bo->base.pages, (bool)bo->base.sgt, bo->mmu_mapped, bo->base.map_wc,
 		 (bool)bo->base.base.import_attach);
 }
 
+static inline int ivpu_bo_lock(struct ivpu_bo *bo)
+{
+	return dma_resv_lock(bo->base.base.resv, NULL);
+}
+
+static inline void ivpu_bo_unlock(struct ivpu_bo *bo)
+{
+	dma_resv_unlock(bo->base.base.resv);
+}
+
 /*
  * ivpu_bo_pin() - pin the backing physical pages and map them to VPU.
  *
@@ -43,22 +53,22 @@ static inline void ivpu_dbg_bo(struct ivpu_device *vdev, struct ivpu_bo *bo, con
 int __must_check ivpu_bo_pin(struct ivpu_bo *bo)
 {
 	struct ivpu_device *vdev = ivpu_bo_to_vdev(bo);
+	struct sg_table *sgt;
 	int ret = 0;
 
-	mutex_lock(&bo->lock);
-
 	ivpu_dbg_bo(vdev, bo, "pin");
-	drm_WARN_ON(&vdev->drm, !bo->ctx);
 
-	if (!bo->mmu_mapped) {
-		struct sg_table *sgt = drm_gem_shmem_get_pages_sgt(&bo->base);
+	sgt = drm_gem_shmem_get_pages_sgt(&bo->base);
+	if (IS_ERR(sgt)) {
+		ret = PTR_ERR(sgt);
+		ivpu_err(vdev, "Failed to map BO in IOMMU: %d\n", ret);
+		return ret;
+	}
 
-		if (IS_ERR(sgt)) {
-			ret = PTR_ERR(sgt);
-			ivpu_err(vdev, "Failed to map BO in IOMMU: %d\n", ret);
-			goto unlock;
-		}
+	ivpu_bo_lock(bo);
 
+	if (!bo->mmu_mapped) {
+		drm_WARN_ON(&vdev->drm, !bo->ctx);
 		ret = ivpu_mmu_context_map_sgt(vdev, bo->ctx, bo->vpu_addr, sgt,
 					       ivpu_bo_is_snooped(bo));
 		if (ret) {
@@ -69,7 +79,7 @@ int __must_check ivpu_bo_pin(struct ivpu_bo *bo)
 	}
 
 unlock:
-	mutex_unlock(&bo->lock);
+	ivpu_bo_unlock(bo);
 
 	return ret;
 }
@@ -84,7 +94,7 @@ ivpu_bo_alloc_vpu_addr(struct ivpu_bo *bo, struct ivpu_mmu_context *ctx,
 	if (!drm_dev_enter(&vdev->drm, &idx))
 		return -ENODEV;
 
-	mutex_lock(&bo->lock);
+	ivpu_bo_lock(bo);
 
 	ret = ivpu_mmu_context_insert_node(ctx, range, ivpu_bo_size(bo), &bo->mm_node);
 	if (!ret) {
@@ -94,9 +104,7 @@ ivpu_bo_alloc_vpu_addr(struct ivpu_bo *bo, struct ivpu_mmu_context *ctx,
 		ivpu_err(vdev, "Failed to add BO to context %u: %d\n", ctx->id, ret);
 	}
 
-	ivpu_dbg_bo(vdev, bo, "alloc");
-
-	mutex_unlock(&bo->lock);
+	ivpu_bo_unlock(bo);
 
 	drm_dev_exit(idx);
 
@@ -107,7 +115,7 @@ static void ivpu_bo_unbind_locked(struct ivpu_bo *bo)
 {
 	struct ivpu_device *vdev = ivpu_bo_to_vdev(bo);
 
-	lockdep_assert(lockdep_is_held(&bo->lock) || !kref_read(&bo->base.base.refcount));
+	lockdep_assert(dma_resv_held(bo->base.base.resv) || !kref_read(&bo->base.base.refcount));
 
 	if (bo->mmu_mapped) {
 		drm_WARN_ON(&vdev->drm, !bo->ctx);
@@ -125,14 +133,12 @@ static void ivpu_bo_unbind_locked(struct ivpu_bo *bo)
 	if (bo->base.base.import_attach)
 		return;
 
-	dma_resv_lock(bo->base.base.resv, NULL);
 	if (bo->base.sgt) {
 		dma_unmap_sgtable(vdev->drm.dev, bo->base.sgt, DMA_BIDIRECTIONAL, 0);
 		sg_free_table(bo->base.sgt);
 		kfree(bo->base.sgt);
 		bo->base.sgt = NULL;
 	}
-	dma_resv_unlock(bo->base.base.resv);
 }
 
 void ivpu_bo_unbind_all_bos_from_context(struct ivpu_device *vdev, struct ivpu_mmu_context *ctx)
@@ -144,12 +150,12 @@ void ivpu_bo_unbind_all_bos_from_context(struct ivpu_device *vdev, struct ivpu_m
 
 	mutex_lock(&vdev->bo_list_lock);
 	list_for_each_entry(bo, &vdev->bo_list, bo_list_node) {
-		mutex_lock(&bo->lock);
+		ivpu_bo_lock(bo);
 		if (bo->ctx == ctx) {
 			ivpu_dbg_bo(vdev, bo, "unbind");
 			ivpu_bo_unbind_locked(bo);
 		}
-		mutex_unlock(&bo->lock);
+		ivpu_bo_unlock(bo);
 	}
 	mutex_unlock(&vdev->bo_list_lock);
 }
@@ -169,7 +175,6 @@ struct drm_gem_object *ivpu_gem_create_object(struct drm_device *dev, size_t siz
 	bo->base.pages_mark_dirty_on_put = true; /* VPU can dirty a BO anytime */
 
 	INIT_LIST_HEAD(&bo->bo_list_node);
-	mutex_init(&bo->lock);
 
 	return &bo->base.base;
 }
@@ -215,7 +220,7 @@ struct drm_gem_object *ivpu_gem_prime_import(struct drm_device *dev,
 	return ERR_PTR(ret);
 }
 
-static struct ivpu_bo *ivpu_bo_alloc(struct ivpu_device *vdev, u64 size, u32 flags)
+static struct ivpu_bo *ivpu_bo_alloc(struct ivpu_device *vdev, u64 size, u32 flags, u32 ctx_id)
 {
 	struct drm_gem_shmem_object *shmem;
 	struct ivpu_bo *bo;
@@ -233,13 +238,16 @@ static struct ivpu_bo *ivpu_bo_alloc(struct ivpu_device *vdev, u64 size, u32 fla
 		return ERR_CAST(shmem);
 
 	bo = to_ivpu_bo(&shmem->base);
+	bo->ctx_id = ctx_id;
 	bo->base.map_wc = flags & DRM_IVPU_BO_WC;
 	bo->flags = flags;
 
 	mutex_lock(&vdev->bo_list_lock);
 	list_add_tail(&bo->bo_list_node, &vdev->bo_list);
 	mutex_unlock(&vdev->bo_list_lock);
 
+	ivpu_dbg_bo(vdev, bo, "alloc");
+
 	return bo;
 }
 
@@ -277,10 +285,14 @@ static void ivpu_gem_bo_free(struct drm_gem_object *obj)
 	list_del(&bo->bo_list_node);
 	mutex_unlock(&vdev->bo_list_lock);
 
-	drm_WARN_ON(&vdev->drm, !dma_resv_test_signaled(obj->resv, DMA_RESV_USAGE_READ));
+	drm_WARN_ON(&vdev->drm, !drm_gem_is_imported(&bo->base.base) &&
+		    !dma_resv_test_signaled(obj->resv, DMA_RESV_USAGE_READ));
+	drm_WARN_ON(&vdev->drm, ivpu_bo_size(bo) == 0);
+	drm_WARN_ON(&vdev->drm, bo->base.vaddr);
 
 	ivpu_bo_unbind_locked(bo);
-	mutex_destroy(&bo->lock);
+	drm_WARN_ON(&vdev->drm, bo->mmu_mapped);
+	drm_WARN_ON(&vdev->drm, bo->ctx);
 
 	drm_WARN_ON(obj->dev, bo->base.pages_use_count > 1);
 	drm_gem_shmem_free(&bo->base);
@@ -314,15 +326,18 @@ int ivpu_bo_create_ioctl(struct drm_device *dev, void *data, struct drm_file *fi
 	if (size == 0)
 		return -EINVAL;
 
-	bo = ivpu_bo_alloc(vdev, size, args->flags);
+	bo = ivpu_bo_alloc(vdev, size, args->flags, file_priv->ctx.id);
 	if (IS_ERR(bo)) {
 		ivpu_err(vdev, "Failed to allocate BO: %pe (ctx %u size %llu flags 0x%x)",
 			 bo, file_priv->ctx.id, args->size, args->flags);
 		return PTR_ERR(bo);
 	}
 
 	ret = drm_gem_handle_create(file, &bo->base.base, &args->handle);
-	if (!ret)
+	if (ret)
+		ivpu_err(vdev, "Failed to create handle for BO: %pe (ctx %u size %llu flags 0x%x)",
+			 bo, file_priv->ctx.id, args->size, args->flags);
+	else
 		args->vpu_addr = bo->vpu_addr;
 
 	drm_gem_object_put(&bo->base.base);
@@ -345,7 +360,7 @@ ivpu_bo_create(struct ivpu_device *vdev, struct ivpu_mmu_context *ctx,
 	drm_WARN_ON(&vdev->drm, !PAGE_ALIGNED(range->end));
 	drm_WARN_ON(&vdev->drm, !PAGE_ALIGNED(size));
 
-	bo = ivpu_bo_alloc(vdev, size, flags);
+	bo = ivpu_bo_alloc(vdev, size, flags, IVPU_GLOBAL_CONTEXT_MMU_SSID);
 	if (IS_ERR(bo)) {
 		ivpu_err(vdev, "Failed to allocate BO: %pe (vpu_addr 0x%llx size %llu flags 0x%x)",
 			 bo, range->start, size, flags);
@@ -361,9 +376,9 @@ ivpu_bo_create(struct ivpu_device *vdev, struct ivpu_mmu_context *ctx,
 		goto err_put;
 
 	if (flags & DRM_IVPU_BO_MAPPABLE) {
-		dma_resv_lock(bo->base.base.resv, NULL);
+		ivpu_bo_lock(bo);
 		ret = drm_gem_shmem_vmap(&bo->base, &map);
-		dma_resv_unlock(bo->base.base.resv);
+		ivpu_bo_unlock(bo);
 
 		if (ret)
 			goto err_put;
@@ -386,9 +401,9 @@ void ivpu_bo_free(struct ivpu_bo *bo)
 	struct iosys_map map = IOSYS_MAP_INIT_VADDR(bo->base.vaddr);
 
 	if (bo->flags & DRM_IVPU_BO_MAPPABLE) {
-		dma_resv_lock(bo->base.base.resv, NULL);
+		ivpu_bo_lock(bo);
 		drm_gem_shmem_vunmap(&bo->base, &map);
-		dma_resv_unlock(bo->base.base.resv);
+		ivpu_bo_unlock(bo);
 	}
 
 	drm_gem_object_put(&bo->base.base);
@@ -407,12 +422,12 @@ int ivpu_bo_info_ioctl(struct drm_device *dev, void *data, struct drm_file *file
 
 	bo = to_ivpu_bo(obj);
 
-	mutex_lock(&bo->lock);
+	ivpu_bo_lock(bo);
 	args->flags = bo->flags;
 	args->mmap_offset = drm_vma_node_offset_addr(&obj->vma_node);
 	args->vpu_addr = bo->vpu_addr;
 	args->size = obj->size;
-	mutex_unlock(&bo->lock);
+	ivpu_bo_unlock(bo);
 
 	drm_gem_object_put(obj);
 	return ret;
@@ -449,10 +464,10 @@ int ivpu_bo_wait_ioctl(struct drm_device *dev, void *data, struct drm_file *file
 
 static void ivpu_bo_print_info(struct ivpu_bo *bo, struct drm_printer *p)
 {
-	mutex_lock(&bo->lock);
+	ivpu_bo_lock(bo);
 
 	drm_printf(p, "%-9p %-3u 0x%-12llx %-10lu 0x%-8x %-4u",
-		   bo, bo->ctx ? bo->ctx->id : 0, bo->vpu_addr, bo->base.base.size,
+		   bo, bo->ctx_id, bo->vpu_addr, bo->base.base.size,
 		   bo->flags, kref_read(&bo->base.base.refcount));
 
 	if (bo->base.pages)
@@ -466,7 +481,7 @@ static void ivpu_bo_print_info(struct ivpu_bo *bo, struct drm_printer *p)
 
 	drm_printf(p, "\n");
 
-	mutex_unlock(&bo->lock);
+	ivpu_bo_unlock(bo);
 }
 
 void ivpu_bo_list(struct drm_device *dev, struct drm_printer *p)

drivers/accel/ivpu/ivpu_gem.h

@@ -17,10 +17,10 @@ struct ivpu_bo {
 	struct list_head bo_list_node;
 	struct drm_mm_node mm_node;
 
-	struct mutex lock; /* Protects: ctx, mmu_mapped, vpu_addr */
 	u64 vpu_addr;
 	u32 flags;
 	u32 job_status; /* Valid only for command buffer */
+	u32 ctx_id;
 	bool mmu_mapped;
 };
 

drivers/accel/ivpu/ivpu_job.c

@@ -247,17 +247,17 @@ static int ivpu_cmdq_unregister(struct ivpu_file_priv *file_priv, struct ivpu_cm
 	if (!cmdq->db_id)
 		return 0;
 
+	ret = ivpu_jsm_unregister_db(vdev, cmdq->db_id);
+	if (!ret)
+		ivpu_dbg(vdev, JOB, "DB %d unregistered\n", cmdq->db_id);
+
 	if (vdev->fw->sched_mode == VPU_SCHEDULING_MODE_HW) {
 		ret = ivpu_jsm_hws_destroy_cmdq(vdev, file_priv->ctx.id, cmdq->id);
 		if (!ret)
 			ivpu_dbg(vdev, JOB, "Command queue %d destroyed, ctx %d\n",
 				 cmdq->id, file_priv->ctx.id);
 	}
 
-	ret = ivpu_jsm_unregister_db(vdev, cmdq->db_id);
-	if (!ret)
-		ivpu_dbg(vdev, JOB, "DB %d unregistered\n", cmdq->db_id);
-
 	xa_erase(&file_priv->vdev->db_xa, cmdq->db_id);
 	cmdq->db_id = 0;
 
@@ -986,7 +986,8 @@ void ivpu_context_abort_work_fn(struct work_struct *work)
 		return;
 
 	if (vdev->fw->sched_mode == VPU_SCHEDULING_MODE_HW)
-		ivpu_jsm_reset_engine(vdev, 0);
+		if (ivpu_jsm_reset_engine(vdev, 0))
+			return;
 
 	mutex_lock(&vdev->context_list_lock);
 	xa_for_each(&vdev->context_xa, ctx_id, file_priv) {
@@ -1009,7 +1010,8 @@ void ivpu_context_abort_work_fn(struct work_struct *work)
 	if (vdev->fw->sched_mode != VPU_SCHEDULING_MODE_HW)
 		goto runtime_put;
 
-	ivpu_jsm_hws_resume_engine(vdev, 0);
+	if (ivpu_jsm_hws_resume_engine(vdev, 0))
+		return;
 	/*
 	 * In hardware scheduling mode NPU already has stopped processing jobs
 	 * and won't send us any further notifications, thus we have to free job related resources