Merge remote-tracking branch 'stable/linux-6.11.y' into v6.11+

Makefile

@@ -1,7 +1,7 @@
 # SPDX-License-Identifier: GPL-2.0
 VERSION = 6
 PATCHLEVEL = 11
-SUBLEVEL = 8
+SUBLEVEL = 9
 EXTRAVERSION =
 NAME = Baby Opossum Posse
 

arch/loongarch/include/asm/loongarch.h

@@ -256,7 +256,7 @@
 #define  CSR_ESTAT_IS_WIDTH		14
 #define  CSR_ESTAT_IS			(_ULCAST_(0x3fff) << CSR_ESTAT_IS_SHIFT)
 
-#define LOONGARCH_CSR_ERA		0x6	/* ERA */
+#define LOONGARCH_CSR_ERA		0x6	/* Exception return address */
 
 #define LOONGARCH_CSR_BADV		0x7	/* Bad virtual address */
 

arch/loongarch/kvm/timer.c

@@ -161,10 +161,11 @@ static void _kvm_save_timer(struct kvm_vcpu *vcpu)
 	if (kvm_vcpu_is_blocking(vcpu)) {
 
 		/*
-		 * HRTIMER_MODE_PINNED is suggested since vcpu may run in
-		 * the same physical cpu in next time
+		 * HRTIMER_MODE_PINNED_HARD is suggested since vcpu may run in
+		 * the same physical cpu in next time, and the timer should run
+		 * in hardirq context even in the PREEMPT_RT case.
 		 */
-		hrtimer_start(&vcpu->arch.swtimer, expire, HRTIMER_MODE_ABS_PINNED);
+		hrtimer_start(&vcpu->arch.swtimer, expire, HRTIMER_MODE_ABS_PINNED_HARD);
 	}
 }
 

arch/loongarch/kvm/vcpu.c

@@ -1144,7 +1144,7 @@ int kvm_arch_vcpu_create(struct kvm_vcpu *vcpu)
 	vcpu->arch.vpid = 0;
 	vcpu->arch.flush_gpa = INVALID_GPA;
 
-	hrtimer_init(&vcpu->arch.swtimer, CLOCK_MONOTONIC, HRTIMER_MODE_ABS_PINNED);
+	hrtimer_init(&vcpu->arch.swtimer, CLOCK_MONOTONIC, HRTIMER_MODE_ABS_PINNED_HARD);
 	vcpu->arch.swtimer.function = kvm_swtimer_wakeup;
 
 	vcpu->arch.handle_exit = kvm_handle_exit;

arch/powerpc/platforms/powernv/opal-irqchip.c

@@ -282,6 +282,7 @@ int __init opal_event_init(void)
 				 name, NULL);
 		if (rc) {
 			pr_warn("Error %d requesting OPAL irq %d\n", rc, (int)r->start);
+			kfree(name);
 			continue;
 		}
 	}

arch/riscv/kvm/aia_imsic.c

@@ -55,7 +55,7 @@ struct imsic {
 	/* IMSIC SW-file */
 	struct imsic_mrif *swfile;
 	phys_addr_t swfile_pa;
-	spinlock_t swfile_extirq_lock;
+	raw_spinlock_t swfile_extirq_lock;
 };
 
 #define imsic_vs_csr_read(__c)			\
@@ -622,15 +622,15 @@ static void imsic_swfile_extirq_update(struct kvm_vcpu *vcpu)
 	 * interruptions between reading topei and updating pending status.
 	 */
 
-	spin_lock_irqsave(&imsic->swfile_extirq_lock, flags);
+	raw_spin_lock_irqsave(&imsic->swfile_extirq_lock, flags);
 
 	if (imsic_mrif_atomic_read(mrif, &mrif->eidelivery) &&
 	    imsic_mrif_topei(mrif, imsic->nr_eix, imsic->nr_msis))
 		kvm_riscv_vcpu_set_interrupt(vcpu, IRQ_VS_EXT);
 	else
 		kvm_riscv_vcpu_unset_interrupt(vcpu, IRQ_VS_EXT);
 
-	spin_unlock_irqrestore(&imsic->swfile_extirq_lock, flags);
+	raw_spin_unlock_irqrestore(&imsic->swfile_extirq_lock, flags);
 }
 
 static void imsic_swfile_read(struct kvm_vcpu *vcpu, bool clear,
@@ -1051,7 +1051,7 @@ int kvm_riscv_vcpu_aia_imsic_init(struct kvm_vcpu *vcpu)
 	}
 	imsic->swfile = page_to_virt(swfile_page);
 	imsic->swfile_pa = page_to_phys(swfile_page);
-	spin_lock_init(&imsic->swfile_extirq_lock);
+	raw_spin_lock_init(&imsic->swfile_extirq_lock);
 
 	/* Setup IO device */
 	kvm_iodevice_init(&imsic->iodev, &imsic_iodoev_ops);

block/elevator.c

@@ -550,7 +550,7 @@ EXPORT_SYMBOL_GPL(elv_unregister);
 static inline bool elv_support_iosched(struct request_queue *q)
 {
 	if (!queue_is_mq(q) ||
-	    (q->tag_set && (q->tag_set->flags & BLK_MQ_F_NO_SCHED)))
+	    (q->tag_set->flags & BLK_MQ_F_NO_SCHED))
 		return false;
 	return true;
 }
@@ -561,7 +561,7 @@ static inline bool elv_support_iosched(struct request_queue *q)
  */
 static struct elevator_type *elevator_get_default(struct request_queue *q)
 {
-	if (q->tag_set && q->tag_set->flags & BLK_MQ_F_NO_SCHED_BY_DEFAULT)
+	if (q->tag_set->flags & BLK_MQ_F_NO_SCHED_BY_DEFAULT)
 		return NULL;
 
 	if (q->nr_hw_queues != 1 &&

crypto/algapi.c

@@ -396,7 +396,7 @@ void crypto_alg_tested(const char *name, int err)
 	q->cra_flags |= CRYPTO_ALG_DEAD;
 	alg = test->adult;
 
-	if (list_empty(&alg->cra_list))
+	if (crypto_is_dead(alg))
 		goto complete;
 
 	if (err == -ECANCELED)

drivers/crypto/marvell/cesa/hash.c

@@ -947,7 +947,7 @@ struct ahash_alg mv_md5_alg = {
 		.base = {
 			.cra_name = "md5",
 			.cra_driver_name = "mv-md5",
-			.cra_priority = 300,
+			.cra_priority = 0,
 			.cra_flags = CRYPTO_ALG_ASYNC |
 				     CRYPTO_ALG_ALLOCATES_MEMORY |
 				     CRYPTO_ALG_KERN_DRIVER_ONLY,
@@ -1018,7 +1018,7 @@ struct ahash_alg mv_sha1_alg = {
 		.base = {
 			.cra_name = "sha1",
 			.cra_driver_name = "mv-sha1",
-			.cra_priority = 300,
+			.cra_priority = 0,
 			.cra_flags = CRYPTO_ALG_ASYNC |
 				     CRYPTO_ALG_ALLOCATES_MEMORY |
 				     CRYPTO_ALG_KERN_DRIVER_ONLY,
@@ -1092,7 +1092,7 @@ struct ahash_alg mv_sha256_alg = {
 		.base = {
 			.cra_name = "sha256",
 			.cra_driver_name = "mv-sha256",
-			.cra_priority = 300,
+			.cra_priority = 0,
 			.cra_flags = CRYPTO_ALG_ASYNC |
 				     CRYPTO_ALG_ALLOCATES_MEMORY |
 				     CRYPTO_ALG_KERN_DRIVER_ONLY,
@@ -1302,7 +1302,7 @@ struct ahash_alg mv_ahmac_md5_alg = {
 		.base = {
 			.cra_name = "hmac(md5)",
 			.cra_driver_name = "mv-hmac-md5",
-			.cra_priority = 300,
+			.cra_priority = 0,
 			.cra_flags = CRYPTO_ALG_ASYNC |
 				     CRYPTO_ALG_ALLOCATES_MEMORY |
 				     CRYPTO_ALG_KERN_DRIVER_ONLY,
@@ -1373,7 +1373,7 @@ struct ahash_alg mv_ahmac_sha1_alg = {
 		.base = {
 			.cra_name = "hmac(sha1)",
 			.cra_driver_name = "mv-hmac-sha1",
-			.cra_priority = 300,
+			.cra_priority = 0,
 			.cra_flags = CRYPTO_ALG_ASYNC |
 				     CRYPTO_ALG_ALLOCATES_MEMORY |
 				     CRYPTO_ALG_KERN_DRIVER_ONLY,
@@ -1444,7 +1444,7 @@ struct ahash_alg mv_ahmac_sha256_alg = {
 		.base = {
 			.cra_name = "hmac(sha256)",
 			.cra_driver_name = "mv-hmac-sha256",
-			.cra_priority = 300,
+			.cra_priority = 0,
 			.cra_flags = CRYPTO_ALG_ASYNC |
 				     CRYPTO_ALG_ALLOCATES_MEMORY |
 				     CRYPTO_ALG_KERN_DRIVER_ONLY,

drivers/gpu/drm/amd/amdkfd/kfd_chardev.c

@@ -1170,7 +1170,7 @@ static int kfd_ioctl_alloc_memory_of_gpu(struct file *filep,
 
 		if (flags & KFD_IOC_ALLOC_MEM_FLAGS_AQL_QUEUE_MEM)
 			size >>= 1;
-		WRITE_ONCE(pdd->vram_usage, pdd->vram_usage + PAGE_ALIGN(size));
+		atomic64_add(PAGE_ALIGN(size), &pdd->vram_usage);
 	}
 
 	mutex_unlock(&p->mutex);
@@ -1241,7 +1241,7 @@ static int kfd_ioctl_free_memory_of_gpu(struct file *filep,
 		kfd_process_device_remove_obj_handle(
 			pdd, GET_IDR_HANDLE(args->handle));
 
-	WRITE_ONCE(pdd->vram_usage, pdd->vram_usage - size);
+	atomic64_sub(size, &pdd->vram_usage);
 
 err_unlock:
 err_pdd:
@@ -2346,7 +2346,7 @@ static int criu_restore_memory_of_gpu(struct kfd_process_device *pdd,
 	} else if (bo_bucket->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_VRAM) {
 		bo_bucket->restored_offset = offset;
 		/* Update the VRAM usage count */
-		WRITE_ONCE(pdd->vram_usage, pdd->vram_usage + bo_bucket->size);
+		atomic64_add(bo_bucket->size, &pdd->vram_usage);
 	}
 	return 0;
 }

drivers/gpu/drm/amd/amdkfd/kfd_priv.h

@@ -766,7 +766,7 @@ struct kfd_process_device {
 	enum kfd_pdd_bound bound;
 
 	/* VRAM usage */
-	uint64_t vram_usage;
+	atomic64_t vram_usage;
 	struct attribute attr_vram;
 	char vram_filename[MAX_SYSFS_FILENAME_LEN];
 

drivers/gpu/drm/amd/amdkfd/kfd_process.c

@@ -306,7 +306,7 @@ static ssize_t kfd_procfs_show(struct kobject *kobj, struct attribute *attr,
 	} else if (strncmp(attr->name, "vram_", 5) == 0) {
 		struct kfd_process_device *pdd = container_of(attr, struct kfd_process_device,
 							      attr_vram);
-		return snprintf(buffer, PAGE_SIZE, "%llu\n", READ_ONCE(pdd->vram_usage));
+		return snprintf(buffer, PAGE_SIZE, "%llu\n", atomic64_read(&pdd->vram_usage));
 	} else if (strncmp(attr->name, "sdma_", 5) == 0) {
 		struct kfd_process_device *pdd = container_of(attr, struct kfd_process_device,
 							      attr_sdma);
@@ -1599,7 +1599,7 @@ struct kfd_process_device *kfd_create_process_device_data(struct kfd_node *dev,
 	pdd->bound = PDD_UNBOUND;
 	pdd->already_dequeued = false;
 	pdd->runtime_inuse = false;
-	pdd->vram_usage = 0;
+	atomic64_set(&pdd->vram_usage, 0);
 	pdd->sdma_past_activity_counter = 0;
 	pdd->user_gpu_id = dev->id;
 	atomic64_set(&pdd->evict_duration_counter, 0);

drivers/gpu/drm/amd/amdkfd/kfd_svm.c

@@ -404,6 +404,27 @@ static void svm_range_bo_release(struct kref *kref)
 		spin_lock(&svm_bo->list_lock);
 	}
 	spin_unlock(&svm_bo->list_lock);
+
+	if (mmget_not_zero(svm_bo->eviction_fence->mm)) {
+		struct kfd_process_device *pdd;
+		struct kfd_process *p;
+		struct mm_struct *mm;
+
+		mm = svm_bo->eviction_fence->mm;
+		/*
+		 * The forked child process takes svm_bo device pages ref, svm_bo could be
+		 * released after parent process is gone.
+		 */
+		p = kfd_lookup_process_by_mm(mm);
+		if (p) {
+			pdd = kfd_get_process_device_data(svm_bo->node, p);
+			if (pdd)
+				atomic64_sub(amdgpu_bo_size(svm_bo->bo), &pdd->vram_usage);
+			kfd_unref_process(p);
+		}
+		mmput(mm);
+	}
+
 	if (!dma_fence_is_signaled(&svm_bo->eviction_fence->base))
 		/* We're not in the eviction worker. Signal the fence. */
 		dma_fence_signal(&svm_bo->eviction_fence->base);
@@ -531,6 +552,7 @@ int
 svm_range_vram_node_new(struct kfd_node *node, struct svm_range *prange,
 			bool clear)
 {
+	struct kfd_process_device *pdd;
 	struct amdgpu_bo_param bp;
 	struct svm_range_bo *svm_bo;
 	struct amdgpu_bo_user *ubo;
@@ -622,6 +644,10 @@ svm_range_vram_node_new(struct kfd_node *node, struct svm_range *prange,
 	list_add(&prange->svm_bo_list, &svm_bo->range_list);
 	spin_unlock(&svm_bo->list_lock);
 
+	pdd = svm_range_get_pdd_by_node(prange, node);
+	if (pdd)
+		atomic64_add(amdgpu_bo_size(bo), &pdd->vram_usage);
+
 	return 0;
 
 reserve_bo_failed:

drivers/gpu/drm/vmwgfx/vmwgfx_drv.h

@@ -62,7 +62,7 @@
 #define VMWGFX_DRIVER_MINOR 20
 #define VMWGFX_DRIVER_PATCHLEVEL 0
 #define VMWGFX_FIFO_STATIC_SIZE (1024*1024)
-#define VMWGFX_MAX_DISPLAYS 16
+#define VMWGFX_NUM_DISPLAY_UNITS 8
 #define VMWGFX_CMD_BOUNCE_INIT_SIZE 32768
 
 #define VMWGFX_MIN_INITIAL_WIDTH 1280
@@ -82,7 +82,7 @@
 #define VMWGFX_NUM_GB_CONTEXT 256
 #define VMWGFX_NUM_GB_SHADER 20000
 #define VMWGFX_NUM_GB_SURFACE 32768
-#define VMWGFX_NUM_GB_SCREEN_TARGET VMWGFX_MAX_DISPLAYS
+#define VMWGFX_NUM_GB_SCREEN_TARGET VMWGFX_NUM_DISPLAY_UNITS
 #define VMWGFX_NUM_DXCONTEXT 256
 #define VMWGFX_NUM_DXQUERY 512
 #define VMWGFX_NUM_MOB (VMWGFX_NUM_GB_CONTEXT +\

drivers/gpu/drm/vmwgfx/vmwgfx_kms.c

@@ -2197,7 +2197,7 @@ int vmw_kms_update_layout_ioctl(struct drm_device *dev, void *data,
 	struct drm_mode_config *mode_config = &dev->mode_config;
 	struct drm_vmw_update_layout_arg *arg =
 		(struct drm_vmw_update_layout_arg *)data;
-	void __user *user_rects;
+	const void __user *user_rects;
 	struct drm_vmw_rect *rects;
 	struct drm_rect *drm_rects;
 	unsigned rects_size;
@@ -2209,6 +2209,8 @@ int vmw_kms_update_layout_ioctl(struct drm_device *dev, void *data,
 					    VMWGFX_MIN_INITIAL_HEIGHT};
 		vmw_du_update_layout(dev_priv, 1, &def_rect);
 		return 0;
+	} else if (arg->num_outputs > VMWGFX_NUM_DISPLAY_UNITS) {
+		return -E2BIG;
 	}
 
 	rects_size = arg->num_outputs * sizeof(struct drm_vmw_rect);

drivers/gpu/drm/vmwgfx/vmwgfx_kms.h

@@ -199,9 +199,6 @@ struct vmw_kms_dirty {
 	s32 unit_y2;
 };
 
-#define VMWGFX_NUM_DISPLAY_UNITS 8
-
-
 #define vmw_framebuffer_to_vfb(x) \
 	container_of(x, struct vmw_framebuffer, base)
 #define vmw_framebuffer_to_vfbs(x) \

drivers/gpu/drm/xe/xe_device.c

@@ -870,7 +870,7 @@ void xe_device_l2_flush(struct xe_device *xe)
 	spin_lock(&gt->global_invl_lock);
 	xe_mmio_write32(gt, XE2_GLOBAL_INVAL, 0x1);
 
-	if (xe_mmio_wait32(gt, XE2_GLOBAL_INVAL, 0x1, 0x0, 150, NULL, true))
+	if (xe_mmio_wait32(gt, XE2_GLOBAL_INVAL, 0x1, 0x0, 500, NULL, true))
 		xe_gt_err_once(gt, "Global invalidation timeout\n");
 	spin_unlock(&gt->global_invl_lock);
 

drivers/gpu/drm/xe/xe_force_wake.c

@@ -115,9 +115,15 @@ static int __domain_wait(struct xe_gt *gt, struct xe_force_wake_domain *domain,
 			     XE_FORCE_WAKE_ACK_TIMEOUT_MS * USEC_PER_MSEC,
 			     &value, true);
 	if (ret)
-		xe_gt_notice(gt, "Force wake domain %d failed to ack %s (%pe) reg[%#x] = %#x\n",
-			     domain->id, str_wake_sleep(wake), ERR_PTR(ret),
-			     domain->reg_ack.addr, value);
+		xe_gt_err(gt, "Force wake domain %d failed to ack %s (%pe) reg[%#x] = %#x\n",
+			  domain->id, str_wake_sleep(wake), ERR_PTR(ret),
+			  domain->reg_ack.addr, value);
+	if (value == ~0) {
+		xe_gt_err(gt,
+			  "Force wake domain %d: %s. MMIO unreliable (forcewake register returns 0xFFFFFFFF)!\n",
+			  domain->id, str_wake_sleep(wake));
+		ret = -EIO;
+	}
 
 	return ret;
 }

drivers/gpu/drm/xe/xe_guc_ct.c

@@ -897,6 +897,24 @@ static int guc_ct_send_recv(struct xe_guc_ct *ct, const u32 *action, u32 len,
 		}
 	}
 
+	/*
+	 * Occasionally it is seen that the G2H worker starts running after a delay of more than
+	 * a second even after being queued and activated by the Linux workqueue subsystem. This
+	 * leads to G2H timeout error. The root cause of issue lies with scheduling latency of
+	 * Lunarlake Hybrid CPU. Issue dissappears if we disable Lunarlake atom cores from BIOS
+	 * and this is beyond xe kmd.
+	 *
+	 * TODO: Drop this change once workqueue scheduling delay issue is fixed on LNL Hybrid CPU.
+	 */
+	if (!ret) {
+		flush_work(&ct->g2h_worker);
+		if (g2h_fence.done) {
+			xe_gt_warn(gt, "G2H fence %u, action %04x, done\n",
+				   g2h_fence.seqno, action[0]);
+			ret = 1;
+		}
+	}
+
 	/*
 	 * Ensure we serialize with completion side to prevent UAF with fence going out of scope on
 	 * the stack, since we have no clue if it will fire after the timeout before we can erase

drivers/gpu/drm/xe/xe_guc_submit.c

@@ -1771,8 +1771,13 @@ void xe_guc_submit_stop(struct xe_guc *guc)
 
 	mutex_lock(&guc->submission_state.lock);
 
-	xa_for_each(&guc->submission_state.exec_queue_lookup, index, q)
+	xa_for_each(&guc->submission_state.exec_queue_lookup, index, q) {
+		/* Prevent redundant attempts to stop parallel queues */
+		if (q->guc->id != index)
+			continue;
+
 		guc_exec_queue_stop(guc, q);
+	}
 
 	mutex_unlock(&guc->submission_state.lock);
 
@@ -1810,8 +1815,13 @@ int xe_guc_submit_start(struct xe_guc *guc)
 
 	mutex_lock(&guc->submission_state.lock);
 	atomic_dec(&guc->submission_state.stopped);
-	xa_for_each(&guc->submission_state.exec_queue_lookup, index, q)
+	xa_for_each(&guc->submission_state.exec_queue_lookup, index, q) {
+		/* Prevent redundant attempts to start parallel queues */
+		if (q->guc->id != index)
+			continue;
+
 		guc_exec_queue_start(q);
+	}
 	mutex_unlock(&guc->submission_state.lock);
 
 	wake_up_all(&guc->ct.wq);