[PATCH 1/3] drm/amdgpu: use HMM mirror callback to replace mmu notifier v6

[PATCH 1/3] drm/amdgpu: use HMM mirror callback to replace mmu notifier v6

[Yang, Philip]

Replace our MMU notifier with hmm_mirror_ops.sync_cpu_device_pagetables
callback. Enable CONFIG_HMM and CONFIG_HMM_MIRROR as a dependency in
DRM_AMDGPU_USERPTR Kconfig.

It supports both KFD userptr and gfx userptr paths.

The depdent HMM patchset from Jérôme Glisse are all merged into 4.20.0
kernel now.

Change-Id: Ie62c3c5e3c5b8521ab3b438d1eff2aa2a003835e
Signed-off-by: Philip Yang <Philip.Yang@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/Kconfig     |   6 +-
 drivers/gpu/drm/amd/amdgpu/Makefile    |   2 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c | 122 ++++++++++---------------
 drivers/gpu/drm/amd/amdgpu/amdgpu_mn.h |   2 +-
 4 files changed, 55 insertions(+), 77 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/Kconfig b/drivers/gpu/drm/amd/amdgpu/Kconfig
index 9221e5489069..960a63355705 100644
--- a/drivers/gpu/drm/amd/amdgpu/Kconfig
+++ b/drivers/gpu/drm/amd/amdgpu/Kconfig
@@ -26,10 +26,10 @@ config DRM_AMDGPU_CIK
 config DRM_AMDGPU_USERPTR
 	bool "Always enable userptr write support"
 	depends on DRM_AMDGPU
-	select MMU_NOTIFIER
+	select HMM_MIRROR
 	help
-	  This option selects CONFIG_MMU_NOTIFIER if it isn't already
-	  selected to enabled full userptr support.
+	  This option selects CONFIG_HMM and CONFIG_HMM_MIRROR if it
+	  isn't already selected to enabled full userptr support.
 
 config DRM_AMDGPU_GART_DEBUGFS
 	bool "Allow GART access through debugfs"
diff --git a/drivers/gpu/drm/amd/amdgpu/Makefile b/drivers/gpu/drm/amd/amdgpu/Makefile
index f76bcb9c45e4..675efc850ff4 100644
--- a/drivers/gpu/drm/amd/amdgpu/Makefile
+++ b/drivers/gpu/drm/amd/amdgpu/Makefile
@@ -172,7 +172,7 @@ endif
 amdgpu-$(CONFIG_COMPAT) += amdgpu_ioc32.o
 amdgpu-$(CONFIG_VGA_SWITCHEROO) += amdgpu_atpx_handler.o
 amdgpu-$(CONFIG_ACPI) += amdgpu_acpi.o
-amdgpu-$(CONFIG_MMU_NOTIFIER) += amdgpu_mn.o
+amdgpu-$(CONFIG_HMM_MIRROR) += amdgpu_mn.o
 
 include $(FULL_AMD_PATH)/powerplay/Makefile
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c
index e55508b39496..5d518d2bb9be 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c
@@ -45,7 +45,7 @@
 
 #include <linux/firmware.h>
 #include <linux/module.h>
-#include <linux/mmu_notifier.h>
+#include <linux/hmm.h>
 #include <linux/interval_tree.h>
 #include <drm/drmP.h>
 #include <drm/drm.h>
@@ -58,7 +58,6 @@
  *
  * @adev: amdgpu device pointer
  * @mm: process address space
- * @mn: MMU notifier structure
  * @type: type of MMU notifier
  * @work: destruction work item
  * @node: hash table node to find structure by adev and mn
@@ -66,6 +65,7 @@
  * @objects: interval tree containing amdgpu_mn_nodes
  * @read_lock: mutex for recursive locking of @lock
  * @recursion: depth of recursion
+ * @mirror: HMM mirror function support
  *
  * Data for each amdgpu device and process address space.
  */
@@ -73,7 +73,6 @@ struct amdgpu_mn {
 	/* constant after initialisation */
 	struct amdgpu_device	*adev;
 	struct mm_struct	*mm;
-	struct mmu_notifier	mn;
 	enum amdgpu_mn_type	type;
 
 	/* only used on destruction */
@@ -87,6 +86,9 @@ struct amdgpu_mn {
 	struct rb_root_cached	objects;
 	struct mutex		read_lock;
 	atomic_t		recursion;
+
+	/* HMM mirror */
+	struct hmm_mirror	mirror;
 };
 
 /**
@@ -103,7 +105,7 @@ struct amdgpu_mn_node {
 };
 
 /**
- * amdgpu_mn_destroy - destroy the MMU notifier
+ * amdgpu_mn_destroy - destroy the HMM mirror
  *
  * @work: previously sheduled work item
  *
@@ -129,28 +131,26 @@ static void amdgpu_mn_destroy(struct work_struct *work)
 	}
 	up_write(&amn->lock);
 	mutex_unlock(&adev->mn_lock);
-	mmu_notifier_unregister_no_release(&amn->mn, amn->mm);
+
+	hmm_mirror_unregister(&amn->mirror);
 	kfree(amn);
 }
 
 /**
- * amdgpu_mn_release - callback to notify about mm destruction
+ * amdgpu_hmm_mirror_release - callback to notify about mm destruction
  *
- * @mn: our notifier
- * @mm: the mm this callback is about
+ * @mirror: the HMM mirror (mm) this callback is about
  *
- * Shedule a work item to lazy destroy our notifier.
+ * Shedule a work item to lazy destroy HMM mirror.
  */
-static void amdgpu_mn_release(struct mmu_notifier *mn,
-			      struct mm_struct *mm)
+static void amdgpu_hmm_mirror_release(struct hmm_mirror *mirror)
 {
-	struct amdgpu_mn *amn = container_of(mn, struct amdgpu_mn, mn);
+	struct amdgpu_mn *amn = container_of(mirror, struct amdgpu_mn, mirror);
 
 	INIT_WORK(&amn->work, amdgpu_mn_destroy);
 	schedule_work(&amn->work);
 }
 
-
 /**
  * amdgpu_mn_lock - take the write side lock for this notifier
  *
@@ -237,21 +237,19 @@ static void amdgpu_mn_invalidate_node(struct amdgpu_mn_node *node,
 /**
  * amdgpu_mn_invalidate_range_start_gfx - callback to notify about mm change
  *
- * @mn: our notifier
- * @mm: the mm this callback is about
- * @start: start of updated range
- * @end: end of updated range
+ * @mirror: the hmm_mirror (mm) is about to update
+ * @update: the update start, end address
  *
  * Block for operations on BOs to finish and mark pages as accessed and
  * potentially dirty.
  */
-static int amdgpu_mn_invalidate_range_start_gfx(struct mmu_notifier *mn,
-						 struct mm_struct *mm,
-						 unsigned long start,
-						 unsigned long end,
-						 bool blockable)
+static int amdgpu_mn_sync_pagetables_gfx(struct hmm_mirror *mirror,
+			const struct hmm_update *update)
 {
-	struct amdgpu_mn *amn = container_of(mn, struct amdgpu_mn, mn);
+	struct amdgpu_mn *amn = container_of(mirror, struct amdgpu_mn, mirror);
+	unsigned long start = update->start;
+	unsigned long end = update->end;
+	bool blockable = update->blockable;
 	struct interval_tree_node *it;
 
 	/* notification is exclusive, but interval is inclusive */
@@ -278,28 +276,28 @@ static int amdgpu_mn_invalidate_range_start_gfx(struct mmu_notifier *mn,
 		amdgpu_mn_invalidate_node(node, start, end);
 	}
 
+	amdgpu_mn_read_unlock(amn);
+
 	return 0;
 }
 
 /**
  * amdgpu_mn_invalidate_range_start_hsa - callback to notify about mm change
  *
- * @mn: our notifier
- * @mm: the mm this callback is about
- * @start: start of updated range
- * @end: end of updated range
+ * @mirror: the hmm_mirror (mm) is about to update
+ * @update: the update start, end address
  *
  * We temporarily evict all BOs between start and end. This
  * necessitates evicting all user-mode queues of the process. The BOs
  * are restorted in amdgpu_mn_invalidate_range_end_hsa.
  */
-static int amdgpu_mn_invalidate_range_start_hsa(struct mmu_notifier *mn,
-						 struct mm_struct *mm,
-						 unsigned long start,
-						 unsigned long end,
-						 bool blockable)
+static int amdgpu_mn_sync_pagetables_hsa(struct hmm_mirror *mirror,
+			const struct hmm_update *update)
 {
-	struct amdgpu_mn *amn = container_of(mn, struct amdgpu_mn, mn);
+	struct amdgpu_mn *amn = container_of(mirror, struct amdgpu_mn, mirror);
+	unsigned long start = update->start;
+	unsigned long end = update->end;
+	bool blockable = update->blockable;
 	struct interval_tree_node *it;
 
 	/* notification is exclusive, but interval is inclusive */
@@ -326,59 +324,39 @@ static int amdgpu_mn_invalidate_range_start_hsa(struct mmu_notifier *mn,
 
 			if (amdgpu_ttm_tt_affect_userptr(bo->tbo.ttm,
 							 start, end))
-				amdgpu_amdkfd_evict_userptr(mem, mm);
+				amdgpu_amdkfd_evict_userptr(mem, amn->mm);
 		}
 	}
 
+	amdgpu_mn_read_unlock(amn);
+
 	return 0;
 }
 
-/**
- * amdgpu_mn_invalidate_range_end - callback to notify about mm change
- *
- * @mn: our notifier
- * @mm: the mm this callback is about
- * @start: start of updated range
- * @end: end of updated range
- *
- * Release the lock again to allow new command submissions.
+/* Low bits of any reasonable mm pointer will be unused due to struct
+ * alignment. Use these bits to make a unique key from the mm pointer
+ * and notifier type.
  */
-static void amdgpu_mn_invalidate_range_end(struct mmu_notifier *mn,
-					   struct mm_struct *mm,
-					   unsigned long start,
-					   unsigned long end)
-{
-	struct amdgpu_mn *amn = container_of(mn, struct amdgpu_mn, mn);
-
-	amdgpu_mn_read_unlock(amn);
-}
+#define AMDGPU_MN_KEY(mm, type) ((unsigned long)(mm) + (type))
 
-static const struct mmu_notifier_ops amdgpu_mn_ops[] = {
+static struct hmm_mirror_ops amdgpu_hmm_mirror_ops[] = {
 	[AMDGPU_MN_TYPE_GFX] = {
-		.release = amdgpu_mn_release,
-		.invalidate_range_start = amdgpu_mn_invalidate_range_start_gfx,
-		.invalidate_range_end = amdgpu_mn_invalidate_range_end,
+		.sync_cpu_device_pagetables = amdgpu_mn_sync_pagetables_gfx,
+		.release = amdgpu_hmm_mirror_release
 	},
 	[AMDGPU_MN_TYPE_HSA] = {
-		.release = amdgpu_mn_release,
-		.invalidate_range_start = amdgpu_mn_invalidate_range_start_hsa,
-		.invalidate_range_end = amdgpu_mn_invalidate_range_end,
+		.sync_cpu_device_pagetables = amdgpu_mn_sync_pagetables_hsa,
+		.release = amdgpu_hmm_mirror_release
 	},
 };
 
-/* Low bits of any reasonable mm pointer will be unused due to struct
- * alignment. Use these bits to make a unique key from the mm pointer
- * and notifier type.
- */
-#define AMDGPU_MN_KEY(mm, type) ((unsigned long)(mm) + (type))
-
 /**
- * amdgpu_mn_get - create notifier context
+ * amdgpu_mn_get - create HMM mirror context
  *
  * @adev: amdgpu device pointer
  * @type: type of MMU notifier context
  *
- * Creates a notifier context for current->mm.
+ * Creates a HMM mirror context for current->mm.
  */
 struct amdgpu_mn *amdgpu_mn_get(struct amdgpu_device *adev,
 				enum amdgpu_mn_type type)
@@ -408,12 +386,12 @@ struct amdgpu_mn *amdgpu_mn_get(struct amdgpu_device *adev,
 	amn->mm = mm;
 	init_rwsem(&amn->lock);
 	amn->type = type;
-	amn->mn.ops = &amdgpu_mn_ops[type];
 	amn->objects = RB_ROOT_CACHED;
 	mutex_init(&amn->read_lock);
 	atomic_set(&amn->recursion, 0);
 
-	r = __mmu_notifier_register(&amn->mn, mm);
+	amn->mirror.ops = &amdgpu_hmm_mirror_ops[type];
+	r = hmm_mirror_register(&amn->mirror, mm);
 	if (r)
 		goto free_amn;
 
@@ -439,7 +417,7 @@ struct amdgpu_mn *amdgpu_mn_get(struct amdgpu_device *adev,
  * @bo: amdgpu buffer object
  * @addr: userptr addr we should monitor
  *
- * Registers an MMU notifier for the given BO at the specified address.
+ * Registers an HMM mirror for the given BO at the specified address.
  * Returns 0 on success, -ERRNO if anything goes wrong.
  */
 int amdgpu_mn_register(struct amdgpu_bo *bo, unsigned long addr)
@@ -495,11 +473,11 @@ int amdgpu_mn_register(struct amdgpu_bo *bo, unsigned long addr)
 }
 
 /**
- * amdgpu_mn_unregister - unregister a BO for notifier updates
+ * amdgpu_mn_unregister - unregister a BO for HMM mirror updates
  *
  * @bo: amdgpu buffer object
  *
- * Remove any registration of MMU notifier updates from the buffer object.
+ * Remove any registration of HMM mirror updates from the buffer object.
  */
 void amdgpu_mn_unregister(struct amdgpu_bo *bo)
 {
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.h
index eb0f432f78fe..0a51fd00021c 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.h
@@ -34,7 +34,7 @@ enum amdgpu_mn_type {
 	AMDGPU_MN_TYPE_HSA,
 };
 
-#if defined(CONFIG_MMU_NOTIFIER)
+#if defined(CONFIG_HMM_MIRROR)
 void amdgpu_mn_lock(struct amdgpu_mn *mn);
 void amdgpu_mn_unlock(struct amdgpu_mn *mn);
 struct amdgpu_mn *amdgpu_mn_get(struct amdgpu_device *adev,
-- 
2.17.1

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

[PATCH 2/3] drm/amdkfd: avoid HMM change cause circular lock dependency v2

[Yang, Philip]

There is circular lock between gfx and kfd path with HMM change:
lock(dqm) -> bo::reserve -> amdgpu_mn_lock

To avoid this, move init/unint_mqd() out of lock(dqm), to remove nested
locking between mmap_sem and bo::reserve. The locking order
is: bo::reserve -> amdgpu_mn_lock(p->mn)

Change-Id: I2ec09a47571f6b4c8eaef93f22c0a600f5f70153
Signed-off-by: Philip Yang <Philip.Yang@amd.com>
---
 .../drm/amd/amdkfd/kfd_device_queue_manager.c | 32 ++++++++++---------
 1 file changed, 17 insertions(+), 15 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
index 8372556b52eb..efe0d3c0215b 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
@@ -1156,21 +1156,17 @@ static int create_queue_cpsch(struct device_queue_manager *dqm, struct queue *q,
 	int retval;
 	struct mqd_manager *mqd_mgr;
 
-	retval = 0;
-
-	dqm_lock(dqm);
-
 	if (dqm->total_queue_count >= max_num_of_queues_per_device) {
 		pr_warn("Can't create new usermode queue because %d queues were already created\n",
 				dqm->total_queue_count);
 		retval = -EPERM;
-		goto out_unlock;
+		goto out;
 	}
 
 	if (q->properties.type == KFD_QUEUE_TYPE_SDMA) {
 		retval = allocate_sdma_queue(dqm, &q->sdma_id);
 		if (retval)
-			goto out_unlock;
+			goto out;
 		q->properties.sdma_queue_id =
 			q->sdma_id / get_num_sdma_engines(dqm);
 		q->properties.sdma_engine_id =
@@ -1181,6 +1177,9 @@ static int create_queue_cpsch(struct device_queue_manager *dqm, struct queue *q,
 	if (retval)
 		goto out_deallocate_sdma_queue;
 
+	/* Do init_mqd before dqm_lock(dqm) to avoid circular locking order:
+	 * lock(dqm) -> bo::reserve
+	 */
 	mqd_mgr = dqm->ops.get_mqd_manager(dqm,
 			get_mqd_type_from_queue_type(q->properties.type));
 
@@ -1188,6 +1187,7 @@ static int create_queue_cpsch(struct device_queue_manager *dqm, struct queue *q,
 		retval = -ENOMEM;
 		goto out_deallocate_doorbell;
 	}
+
 	/*
 	 * Eviction state logic: we only mark active queues as evicted
 	 * to avoid the overhead of restoring inactive queues later
@@ -1196,9 +1196,7 @@ static int create_queue_cpsch(struct device_queue_manager *dqm, struct queue *q,
 		q->properties.is_evicted = (q->properties.queue_size > 0 &&
 					    q->properties.queue_percent > 0 &&
 					    q->properties.queue_address != 0);
-
 	dqm->asic_ops.init_sdma_vm(dqm, q, qpd);
-
 	q->properties.tba_addr = qpd->tba_addr;
 	q->properties.tma_addr = qpd->tma_addr;
 	retval = mqd_mgr->init_mqd(mqd_mgr, &q->mqd, &q->mqd_mem_obj,
@@ -1206,6 +1204,8 @@ static int create_queue_cpsch(struct device_queue_manager *dqm, struct queue *q,
 	if (retval)
 		goto out_deallocate_doorbell;
 
+	dqm_lock(dqm);
+
 	list_add(&q->list, &qpd->queues_list);
 	qpd->queue_count++;
 	if (q->properties.is_active) {
@@ -1233,9 +1233,7 @@ static int create_queue_cpsch(struct device_queue_manager *dqm, struct queue *q,
 out_deallocate_sdma_queue:
 	if (q->properties.type == KFD_QUEUE_TYPE_SDMA)
 		deallocate_sdma_queue(dqm, q->sdma_id);
-out_unlock:
-	dqm_unlock(dqm);
-
+out:
 	return retval;
 }
 
@@ -1398,8 +1396,6 @@ static int destroy_queue_cpsch(struct device_queue_manager *dqm,
 			qpd->reset_wavefronts = true;
 	}
 
-	mqd_mgr->uninit_mqd(mqd_mgr, q->mqd, q->mqd_mem_obj);
-
 	/*
 	 * Unconditionally decrement this counter, regardless of the queue's
 	 * type
@@ -1410,6 +1406,9 @@ static int destroy_queue_cpsch(struct device_queue_manager *dqm,
 
 	dqm_unlock(dqm);
 
+	/* Do uninit_mqd after dqm_unlock(dqm) to avoid circular locking */
+	mqd_mgr->uninit_mqd(mqd_mgr, q->mqd, q->mqd_mem_obj);
+
 	return retval;
 
 failed:
@@ -1631,7 +1630,11 @@ static int process_termination_cpsch(struct device_queue_manager *dqm,
 		qpd->reset_wavefronts = false;
 	}
 
-	/* lastly, free mqd resources */
+	dqm_unlock(dqm);
+
+	/* Lastly, free mqd resources.
+	 * Do uninit_mqd() after dqm_unlock to avoid circular locking.
+	 */
 	list_for_each_entry_safe(q, next, &qpd->queues_list, list) {
 		mqd_mgr = dqm->ops.get_mqd_manager(dqm,
 			get_mqd_type_from_queue_type(q->properties.type));
@@ -1645,7 +1648,6 @@ static int process_termination_cpsch(struct device_queue_manager *dqm,
 	}
 
 out:
-	dqm_unlock(dqm);
 	return retval;
 }
 
-- 
2.17.1

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

[PATCH 3/3] drm/amdgpu: replace get_user_pages with HMM address mirror helpers v6

[Yang, Philip]

Use HMM helper function hmm_vma_fault() to get physical pages backing
userptr and start CPU page table update track of those pages. Then use
hmm_vma_range_done() to check if those pages are updated before
amdgpu_cs_submit for gfx or before user queues are resumed for kfd.

If userptr pages are updated, for gfx, amdgpu_cs_ioctl will restart
from scratch, for kfd, restore worker is rescheduled to retry.

HMM simplify the CPU page table concurrent update check, so remove
guptasklock, mmu_invalidations, last_set_pages fields from
amdgpu_ttm_tt struct.

HMM does not pin the page (increase page ref count), so remove related
operations like release_pages(), put_page(), mark_page_dirty().

Change-Id: I2e8c0c6f0d8c21e5596a32d7fc91762778bc9e67
Signed-off-by: Philip Yang <Philip.Yang@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h    |   1 -
 .../gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c  |  95 +++-------
 drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.h   |   2 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c        | 158 +++++++---------
 drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c       |  14 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c        |  25 ++-
 drivers/gpu/drm/amd/amdgpu/amdgpu_mn.h        |   4 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c       | 173 ++++++++----------
 drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h       |   3 +-
 9 files changed, 198 insertions(+), 277 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
index 0b31a1859023..0e1711a75b68 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
@@ -61,7 +61,6 @@ struct kgd_mem {
 
 	atomic_t invalid;
 	struct amdkfd_process_info *process_info;
-	struct page **user_pages;
 
 	struct amdgpu_sync sync;
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
index d7b10d79f1de..ae2d838d31ea 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
@@ -582,28 +582,12 @@ static int init_user_pages(struct kgd_mem *mem, struct mm_struct *mm,
 		goto out;
 	}
 
-	/* If no restore worker is running concurrently, user_pages
-	 * should not be allocated
-	 */
-	WARN(mem->user_pages, "Leaking user_pages array");
-
-	mem->user_pages = kvmalloc_array(bo->tbo.ttm->num_pages,
-					   sizeof(struct page *),
-					   GFP_KERNEL | __GFP_ZERO);
-	if (!mem->user_pages) {
-		pr_err("%s: Failed to allocate pages array\n", __func__);
-		ret = -ENOMEM;
-		goto unregister_out;
-	}
-
-	ret = amdgpu_ttm_tt_get_user_pages(bo->tbo.ttm, mem->user_pages);
+	ret = amdgpu_ttm_tt_get_user_pages(bo->tbo.ttm, bo->tbo.ttm->pages);
 	if (ret) {
 		pr_err("%s: Failed to get user pages: %d\n", __func__, ret);
-		goto free_out;
+		goto unregister_out;
 	}
 
-	amdgpu_ttm_tt_set_user_pages(bo->tbo.ttm, mem->user_pages);
-
 	ret = amdgpu_bo_reserve(bo, true);
 	if (ret) {
 		pr_err("%s: Failed to reserve BO\n", __func__);
@@ -616,11 +600,7 @@ static int init_user_pages(struct kgd_mem *mem, struct mm_struct *mm,
 	amdgpu_bo_unreserve(bo);
 
 release_out:
-	if (ret)
-		release_pages(mem->user_pages, bo->tbo.ttm->num_pages);
-free_out:
-	kvfree(mem->user_pages);
-	mem->user_pages = NULL;
+	amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm);
 unregister_out:
 	if (ret)
 		amdgpu_mn_unregister(bo);
@@ -679,7 +659,6 @@ static int reserve_bo_and_vm(struct kgd_mem *mem,
 	ctx->kfd_bo.priority = 0;
 	ctx->kfd_bo.tv.bo = &bo->tbo;
 	ctx->kfd_bo.tv.num_shared = 1;
-	ctx->kfd_bo.user_pages = NULL;
 	list_add(&ctx->kfd_bo.tv.head, &ctx->list);
 
 	amdgpu_vm_get_pd_bo(vm, &ctx->list, &ctx->vm_pd[0]);
@@ -743,7 +722,6 @@ static int reserve_bo_and_cond_vms(struct kgd_mem *mem,
 	ctx->kfd_bo.priority = 0;
 	ctx->kfd_bo.tv.bo = &bo->tbo;
 	ctx->kfd_bo.tv.num_shared = 1;
-	ctx->kfd_bo.user_pages = NULL;
 	list_add(&ctx->kfd_bo.tv.head, &ctx->list);
 
 	i = 0;
@@ -1371,15 +1349,6 @@ int amdgpu_amdkfd_gpuvm_free_memory_of_gpu(
 	list_del(&bo_list_entry->head);
 	mutex_unlock(&process_info->lock);
 
-	/* Free user pages if necessary */
-	if (mem->user_pages) {
-		pr_debug("%s: Freeing user_pages array\n", __func__);
-		if (mem->user_pages[0])
-			release_pages(mem->user_pages,
-					mem->bo->tbo.ttm->num_pages);
-		kvfree(mem->user_pages);
-	}
-
 	ret = reserve_bo_and_cond_vms(mem, NULL, BO_VM_ALL, &ctx);
 	if (unlikely(ret))
 		return ret;
@@ -1855,25 +1824,11 @@ static int update_invalid_user_pages(struct amdkfd_process_info *process_info,
 
 		bo = mem->bo;
 
-		if (!mem->user_pages) {
-			mem->user_pages =
-				kvmalloc_array(bo->tbo.ttm->num_pages,
-						 sizeof(struct page *),
-						 GFP_KERNEL | __GFP_ZERO);
-			if (!mem->user_pages) {
-				pr_err("%s: Failed to allocate pages array\n",
-				       __func__);
-				return -ENOMEM;
-			}
-		} else if (mem->user_pages[0]) {
-			release_pages(mem->user_pages, bo->tbo.ttm->num_pages);
-		}
-
 		/* Get updated user pages */
 		ret = amdgpu_ttm_tt_get_user_pages(bo->tbo.ttm,
-						   mem->user_pages);
+						   bo->tbo.ttm->pages);
 		if (ret) {
-			mem->user_pages[0] = NULL;
+			bo->tbo.ttm->pages[0] = NULL;
 			pr_info("%s: Failed to get user pages: %d\n",
 				__func__, ret);
 			/* Pretend it succeeded. It will fail later
@@ -1882,12 +1837,6 @@ static int update_invalid_user_pages(struct amdkfd_process_info *process_info,
 			 * stalled user mode queues.
 			 */
 		}
-
-		/* Mark the BO as valid unless it was invalidated
-		 * again concurrently
-		 */
-		if (atomic_cmpxchg(&mem->invalid, invalid, 0) != invalid)
-			return -EAGAIN;
 	}
 
 	return 0;
@@ -1917,7 +1866,8 @@ static int validate_invalid_user_pages(struct amdkfd_process_info *process_info)
 				     GFP_KERNEL);
 	if (!pd_bo_list_entries) {
 		pr_err("%s: Failed to allocate PD BO list entries\n", __func__);
-		return -ENOMEM;
+		ret = -ENOMEM;
+		goto out_no_mem;
 	}
 
 	INIT_LIST_HEAD(&resv_list);
@@ -1941,7 +1891,7 @@ static int validate_invalid_user_pages(struct amdkfd_process_info *process_info)
 	ret = ttm_eu_reserve_buffers(&ticket, &resv_list, false, &duplicates);
 	WARN(!list_empty(&duplicates), "Duplicates should be empty");
 	if (ret)
-		goto out;
+		goto out_free;
 
 	amdgpu_sync_create(&sync);
 
@@ -1967,10 +1917,8 @@ static int validate_invalid_user_pages(struct amdkfd_process_info *process_info)
 
 		bo = mem->bo;
 
-		/* Copy pages array and validate the BO if we got user pages */
-		if (mem->user_pages[0]) {
-			amdgpu_ttm_tt_set_user_pages(bo->tbo.ttm,
-						     mem->user_pages);
+		/* Validate the BO if we got user pages */
+		if (bo->tbo.ttm->pages[0]) {
 			amdgpu_bo_placement_from_domain(bo, mem->domain);
 			ret = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
 			if (ret) {
@@ -1979,16 +1927,16 @@ static int validate_invalid_user_pages(struct amdkfd_process_info *process_info)
 			}
 		}
 
-		/* Validate succeeded, now the BO owns the pages, free
-		 * our copy of the pointer array. Put this BO back on
-		 * the userptr_valid_list. If we need to revalidate
-		 * it, we need to start from scratch.
-		 */
-		kvfree(mem->user_pages);
-		mem->user_pages = NULL;
 		list_move_tail(&mem->validate_list.head,
 			       &process_info->userptr_valid_list);
 
+		/* Stop HMM track the userptr update. We dont check the return
+		 * value for concurrent CPU page table update because we will
+		 * reschedule the restore worker if process_info->evicted_bos
+		 * is updated.
+		 */
+		amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm);
+
 		/* Update mapping. If the BO was not validated
 		 * (because we couldn't get user pages), this will
 		 * clear the page table entries, which will result in
@@ -2022,8 +1970,15 @@ static int validate_invalid_user_pages(struct amdkfd_process_info *process_info)
 	ttm_eu_backoff_reservation(&ticket, &resv_list);
 	amdgpu_sync_wait(&sync, false);
 	amdgpu_sync_free(&sync);
-out:
+out_free:
 	kfree(pd_bo_list_entries);
+out_no_mem:
+	list_for_each_entry_safe(mem, tmp_mem,
+				 &process_info->userptr_inval_list,
+				 validate_list.head) {
+		bo = mem->bo;
+		amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm);
+	}
 
 	return ret;
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.h
index 7c5f5d1601e6..a130e766cbdb 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.h
@@ -36,7 +36,7 @@ struct amdgpu_bo_list_entry {
 	struct amdgpu_bo_va		*bo_va;
 	uint32_t			priority;
 	struct page			**user_pages;
-	int				user_invalidated;
+	bool				user_invalidated;
 };
 
 struct amdgpu_bo_list {
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
index 1c49b8266d69..451a1fab17d6 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@ -52,7 +52,6 @@ static int amdgpu_cs_user_fence_chunk(struct amdgpu_cs_parser *p,
 	p->uf_entry.tv.bo = &bo->tbo;
 	/* One for TTM and one for the CS job */
 	p->uf_entry.tv.num_shared = 2;
-	p->uf_entry.user_pages = NULL;
 
 	drm_gem_object_put_unlocked(gobj);
 
@@ -539,14 +538,14 @@ static int amdgpu_cs_list_validate(struct amdgpu_cs_parser *p,
 		if (usermm && usermm != current->mm)
 			return -EPERM;
 
-		/* Check if we have user pages and nobody bound the BO already */
-		if (amdgpu_ttm_tt_userptr_needs_pages(bo->tbo.ttm) &&
-		    lobj->user_pages) {
+		if (amdgpu_ttm_tt_is_userptr(bo->tbo.ttm) &&
+		    lobj->user_invalidated && lobj->user_pages) {
 			amdgpu_bo_placement_from_domain(bo,
 							AMDGPU_GEM_DOMAIN_CPU);
 			r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
 			if (r)
 				return r;
+
 			amdgpu_ttm_tt_set_user_pages(bo->tbo.ttm,
 						     lobj->user_pages);
 			binding_userptr = true;
@@ -577,7 +576,6 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
 	struct amdgpu_bo *gds;
 	struct amdgpu_bo *gws;
 	struct amdgpu_bo *oa;
-	unsigned tries = 10;
 	int r;
 
 	INIT_LIST_HEAD(&p->validated);
@@ -613,79 +611,45 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
 	if (p->uf_entry.tv.bo && !ttm_to_amdgpu_bo(p->uf_entry.tv.bo)->parent)
 		list_add(&p->uf_entry.tv.head, &p->validated);
 
-	while (1) {
-		struct list_head need_pages;
-
-		r = ttm_eu_reserve_buffers(&p->ticket, &p->validated, true,
-					   &duplicates);
-		if (unlikely(r != 0)) {
-			if (r != -ERESTARTSYS)
-				DRM_ERROR("ttm_eu_reserve_buffers failed.\n");
-			goto error_free_pages;
-		}
-
-		INIT_LIST_HEAD(&need_pages);
-		amdgpu_bo_list_for_each_userptr_entry(e, p->bo_list) {
-			struct amdgpu_bo *bo = ttm_to_amdgpu_bo(e->tv.bo);
-
-			if (amdgpu_ttm_tt_userptr_invalidated(bo->tbo.ttm,
-				 &e->user_invalidated) && e->user_pages) {
-
-				/* We acquired a page array, but somebody
-				 * invalidated it. Free it and try again
-				 */
-				release_pages(e->user_pages,
-					      bo->tbo.ttm->num_pages);
-				kvfree(e->user_pages);
-				e->user_pages = NULL;
-			}
-
-			if (amdgpu_ttm_tt_userptr_needs_pages(bo->tbo.ttm) &&
-			    !e->user_pages) {
-				list_del(&e->tv.head);
-				list_add(&e->tv.head, &need_pages);
-
-				amdgpu_bo_unreserve(bo);
-			}
+	/* Get userptr backing pages. If pages are updated after registered
+	 * in amdgpu_gem_userptr_ioctl(), amdgpu_cs_list_validate() will do
+	 * amdgpu_ttm_backend_bind() to flush and invalidate new pages
+	 */
+	amdgpu_bo_list_for_each_userptr_entry(e, p->bo_list) {
+		struct amdgpu_bo *bo = ttm_to_amdgpu_bo(e->tv.bo);
+		bool userpage_invalidated = false;
+		int i;
+
+		e->user_pages = kvmalloc_array(bo->tbo.ttm->num_pages,
+					sizeof(struct page *),
+					GFP_KERNEL | __GFP_ZERO);
+		if (!e->user_pages) {
+			DRM_ERROR("calloc failure\n");
+			return -ENOMEM;
 		}
 
-		if (list_empty(&need_pages))
-			break;
-
-		/* Unreserve everything again. */
-		ttm_eu_backoff_reservation(&p->ticket, &p->validated);
-
-		/* We tried too many times, just abort */
-		if (!--tries) {
-			r = -EDEADLK;
-			DRM_ERROR("deadlock in %s\n", __func__);
-			goto error_free_pages;
+		r = amdgpu_ttm_tt_get_user_pages(bo->tbo.ttm, e->user_pages);
+		if (r) {
+			kvfree(e->user_pages);
+			e->user_pages = NULL;
+			return r;
 		}
 
-		/* Fill the page arrays for all userptrs. */
-		list_for_each_entry(e, &need_pages, tv.head) {
-			struct ttm_tt *ttm = e->tv.bo->ttm;
-
-			e->user_pages = kvmalloc_array(ttm->num_pages,
-							 sizeof(struct page*),
-							 GFP_KERNEL | __GFP_ZERO);
-			if (!e->user_pages) {
-				r = -ENOMEM;
-				DRM_ERROR("calloc failure in %s\n", __func__);
-				goto error_free_pages;
-			}
-
-			r = amdgpu_ttm_tt_get_user_pages(ttm, e->user_pages);
-			if (r) {
-				DRM_ERROR("amdgpu_ttm_tt_get_user_pages failed.\n");
-				kvfree(e->user_pages);
-				e->user_pages = NULL;
-				goto error_free_pages;
+		for (i = 0; i < bo->tbo.ttm->num_pages; i++) {
+			if (bo->tbo.ttm->pages[i] != e->user_pages[i]) {
+				userpage_invalidated = true;
+				break;
 			}
 		}
+		e->user_invalidated = userpage_invalidated;
+	}
 
-		/* And try again. */
-		list_splice(&need_pages, &p->validated);
+	r = ttm_eu_reserve_buffers(&p->ticket, &p->validated, true,
+				   &duplicates);
+	if (unlikely(r != 0)) {
+		if (r != -ERESTARTSYS)
+			DRM_ERROR("ttm_eu_reserve_buffers failed.\n");
+		goto out;
 	}
 
 	amdgpu_cs_get_threshold_for_moves(p->adev, &p->bytes_moved_threshold,
@@ -754,17 +718,7 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
 error_validate:
 	if (r)
 		ttm_eu_backoff_reservation(&p->ticket, &p->validated);
-
-error_free_pages:
-
-	amdgpu_bo_list_for_each_userptr_entry(e, p->bo_list) {
-		if (!e->user_pages)
-			continue;
-
-		release_pages(e->user_pages, e->tv.bo->ttm->num_pages);
-		kvfree(e->user_pages);
-	}
-
+out:
 	return r;
 }
 
@@ -1213,8 +1167,7 @@ static int amdgpu_cs_submit(struct amdgpu_cs_parser *p,
 	struct amdgpu_bo_list_entry *e;
 	struct amdgpu_job *job;
 	uint64_t seq;
-
-	int r;
+	int r = 0;
 
 	job = p->job;
 	p->job = NULL;
@@ -1223,15 +1176,21 @@ static int amdgpu_cs_submit(struct amdgpu_cs_parser *p,
 	if (r)
 		goto error_unlock;
 
-	/* No memory allocation is allowed while holding the mn lock */
+	/* No memory allocation is allowed while holding the mn lock.
+	 * p->mn is hold until amdgpu_cs_submit is finished and fence is added
+	 * to BOs.
+	 */
 	amdgpu_mn_lock(p->mn);
+
+	/* If userptr are updated after amdgpu_cs_parser_bos(), restart cs */
 	amdgpu_bo_list_for_each_userptr_entry(e, p->bo_list) {
 		struct amdgpu_bo *bo = ttm_to_amdgpu_bo(e->tv.bo);
 
-		if (amdgpu_ttm_tt_userptr_needs_pages(bo->tbo.ttm)) {
-			r = -ERESTARTSYS;
-			goto error_abort;
-		}
+		r |= !amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm);
+	}
+	if (r) {
+		r = -ERESTARTSYS;
+		goto error_abort;
 	}
 
 	job->owner = p->filp;
@@ -1278,14 +1237,20 @@ static int amdgpu_cs_submit(struct amdgpu_cs_parser *p,
 int amdgpu_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
 {
 	struct amdgpu_device *adev = dev->dev_private;
-	union drm_amdgpu_cs *cs = data;
-	struct amdgpu_cs_parser parser = {};
-	bool reserved_buffers = false;
+	union drm_amdgpu_cs *cs;
+	struct amdgpu_cs_parser parser;
+	bool reserved_buffers;
+	int tries = 10;
 	int i, r;
 
 	if (!adev->accel_working)
 		return -EBUSY;
 
+restart:
+	memset(&parser, 0, sizeof(parser));
+	cs = data;
+	reserved_buffers = false;
+
 	parser.adev = adev;
 	parser.filp = filp;
 
@@ -1327,6 +1292,15 @@ int amdgpu_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
 
 out:
 	amdgpu_cs_parser_fini(&parser, r, reserved_buffers);
+
+	if (r == -ERESTARTSYS) {
+		if (!--tries) {
+			DRM_ERROR("Possible deadlock? Retry too many times\n");
+			return -EDEADLK;
+		}
+		goto restart;
+	}
+
 	return r;
 }
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
index f4f00217546e..92025993bfb1 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
@@ -336,26 +336,24 @@ int amdgpu_gem_userptr_ioctl(struct drm_device *dev, void *data,
 
 		r = amdgpu_bo_reserve(bo, true);
 		if (r)
-			goto free_pages;
+			goto user_pages_done;
 
 		amdgpu_bo_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_GTT);
 		r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
 		amdgpu_bo_unreserve(bo);
 		if (r)
-			goto free_pages;
+			goto user_pages_done;
 	}
 
 	r = drm_gem_handle_create(filp, gobj, &handle);
-	/* drop reference from allocate - handle holds it now */
-	drm_gem_object_put_unlocked(gobj);
 	if (r)
-		return r;
+		goto user_pages_done;
 
 	args->handle = handle;
-	return 0;
 
-free_pages:
-	release_pages(bo->tbo.ttm->pages, bo->tbo.ttm->num_pages);
+user_pages_done:
+	if (args->flags & AMDGPU_GEM_USERPTR_VALIDATE)
+		amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm);
 
 release_object:
 	drm_gem_object_put_unlocked(gobj);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c
index 5d518d2bb9be..9d40f3001f3c 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c
@@ -229,8 +229,6 @@ static void amdgpu_mn_invalidate_node(struct amdgpu_mn_node *node,
 			true, false, MAX_SCHEDULE_TIMEOUT);
 		if (r <= 0)
 			DRM_ERROR("(%ld) failed to wait for user bo\n", r);
-
-		amdgpu_ttm_tt_mark_user_pages(bo->tbo.ttm);
 	}
 }
 
@@ -513,3 +511,26 @@ void amdgpu_mn_unregister(struct amdgpu_bo *bo)
 	mutex_unlock(&adev->mn_lock);
 }
 
+/* flags used by HMM internal, not related to CPU/GPU PTE flags */
+static const uint64_t hmm_range_flags[HMM_PFN_FLAG_MAX] = {
+		(1 << 0), /* HMM_PFN_VALID */
+		(1 << 1), /* HMM_PFN_WRITE */
+		0 /* HMM_PFN_DEVICE_PRIVATE */
+};
+
+static const uint64_t hmm_range_values[HMM_PFN_VALUE_MAX] = {
+		0xfffffffffffffffeUL, /* HMM_PFN_ERROR */
+		0, /* HMM_PFN_NONE */
+		0xfffffffffffffffcUL /* HMM_PFN_SPECIAL */
+};
+
+void amdgpu_hmm_init_range(struct hmm_range *range)
+{
+	if (range) {
+		range->flags = hmm_range_flags;
+		range->values = hmm_range_values;
+		range->pfn_shift = PAGE_SHIFT;
+		range->pfns = NULL;
+		INIT_LIST_HEAD(&range->list);
+	}
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.h
index 0a51fd00021c..4803e216e174 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.h
@@ -25,9 +25,10 @@
 #define __AMDGPU_MN_H__
 
 /*
- * MMU Notifier
+ * HMM mirror
  */
 struct amdgpu_mn;
+struct hmm_range;
 
 enum amdgpu_mn_type {
 	AMDGPU_MN_TYPE_GFX,
@@ -41,6 +42,7 @@ struct amdgpu_mn *amdgpu_mn_get(struct amdgpu_device *adev,
 				enum amdgpu_mn_type type);
 int amdgpu_mn_register(struct amdgpu_bo *bo, unsigned long addr);
 void amdgpu_mn_unregister(struct amdgpu_bo *bo);
+void amdgpu_hmm_init_range(struct hmm_range *range);
 #else
 static inline void amdgpu_mn_lock(struct amdgpu_mn *mn) {}
 static inline void amdgpu_mn_unlock(struct amdgpu_mn *mn) {}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
index c91ec3101d00..3d074f4b02f4 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
@@ -43,6 +43,7 @@
 #include <linux/pagemap.h>
 #include <linux/debugfs.h>
 #include <linux/iommu.h>
+#include <linux/hmm.h>
 #include "amdgpu.h"
 #include "amdgpu_object.h"
 #include "amdgpu_trace.h"
@@ -705,11 +706,6 @@ static unsigned long amdgpu_ttm_io_mem_pfn(struct ttm_buffer_object *bo,
 /*
  * TTM backend functions.
  */
-struct amdgpu_ttm_gup_task_list {
-	struct list_head	list;
-	struct task_struct	*task;
-};
-
 struct amdgpu_ttm_tt {
 	struct ttm_dma_tt	ttm;
 	u64			offset;
@@ -718,85 +714,96 @@ struct amdgpu_ttm_tt {
 	uint32_t		userflags;
 	spinlock_t              guptasklock;
 	struct list_head        guptasks;
-	atomic_t		mmu_invalidations;
-	uint32_t		last_set_pages;
+	struct hmm_range	range;
 };
 
 /**
- * amdgpu_ttm_tt_get_user_pages - Pin pages of memory pointed to by a USERPTR
- * pointer to memory
+ * amdgpu_ttm_tt_get_user_pages - get device accessible pages that back user
+ * memory and start HMM tracking CPU page table update
  *
- * Called by amdgpu_gem_userptr_ioctl() and amdgpu_cs_parser_bos().
- * This provides a wrapper around the get_user_pages() call to provide
- * device accessible pages that back user memory.
+ * Calling function must call amdgpu_ttm_tt_userptr_range_done() once and only
+ * once afterwards to stop HMM tracking
  */
 int amdgpu_ttm_tt_get_user_pages(struct ttm_tt *ttm, struct page **pages)
 {
 	struct amdgpu_ttm_tt *gtt = (void *)ttm;
 	struct mm_struct *mm = gtt->usertask->mm;
-	unsigned int flags = 0;
-	unsigned pinned = 0;
-	int r;
+	unsigned long end = gtt->userptr + ttm->num_pages * PAGE_SIZE;
+	struct hmm_range *range = &gtt->range;
+	int r = 0, i;
 
 	if (!mm) /* Happens during process shutdown */
 		return -ESRCH;
 
-	if (!(gtt->userflags & AMDGPU_GEM_USERPTR_READONLY))
-		flags |= FOLL_WRITE;
+	amdgpu_hmm_init_range(range);
 
 	down_read(&mm->mmap_sem);
 
-	if (gtt->userflags & AMDGPU_GEM_USERPTR_ANONONLY) {
-		/*
-		 * check that we only use anonymous memory to prevent problems
-		 * with writeback
-		 */
-		unsigned long end = gtt->userptr + ttm->num_pages * PAGE_SIZE;
-		struct vm_area_struct *vma;
+	range->vma = find_vma(mm, gtt->userptr);
+	if (!range_in_vma(range->vma, gtt->userptr, end))
+		r = -EFAULT;
+	else if ((gtt->userflags & AMDGPU_GEM_USERPTR_ANONONLY) &&
+		range->vma->vm_file)
+		r = -EPERM;
+	if (r)
+		goto out;
 
-		vma = find_vma(mm, gtt->userptr);
-		if (!vma || vma->vm_file || vma->vm_end < end) {
-			up_read(&mm->mmap_sem);
-			return -EPERM;
-		}
+	range->pfns = kvmalloc_array(ttm->num_pages, sizeof(uint64_t),
+				     GFP_KERNEL);
+	if (range->pfns == NULL) {
+		r = -ENOMEM;
+		goto out;
 	}
+	range->start = gtt->userptr;
+	range->end = end;
 
-	/* loop enough times using contiguous pages of memory */
-	do {
-		unsigned num_pages = ttm->num_pages - pinned;
-		uint64_t userptr = gtt->userptr + pinned * PAGE_SIZE;
-		struct page **p = pages + pinned;
-		struct amdgpu_ttm_gup_task_list guptask;
+	range->pfns[0] = range->flags[HMM_PFN_VALID];
+	range->pfns[0] |= amdgpu_ttm_tt_is_readonly(ttm) ?
+				0 : range->flags[HMM_PFN_WRITE];
+	for (i = 1; i < ttm->num_pages; i++)
+		range->pfns[i] = range->pfns[0];
 
-		guptask.task = current;
-		spin_lock(&gtt->guptasklock);
-		list_add(&guptask.list, &gtt->guptasks);
-		spin_unlock(&gtt->guptasklock);
+	/* This may trigger page table update */
+	r = hmm_vma_fault(range, true);
+	if (r)
+		goto out_free_pfns;
 
-		if (mm == current->mm)
-			r = get_user_pages(userptr, num_pages, flags, p, NULL);
-		else
-			r = get_user_pages_remote(gtt->usertask,
-					mm, userptr, num_pages,
-					flags, p, NULL, NULL);
+	up_read(&mm->mmap_sem);
 
-		spin_lock(&gtt->guptasklock);
-		list_del(&guptask.list);
-		spin_unlock(&gtt->guptasklock);
+	for (i = 0; i < ttm->num_pages; i++)
+		pages[i] = hmm_pfn_to_page(range, range->pfns[i]);
 
-		if (r < 0)
-			goto release_pages;
+	return 0;
 
-		pinned += r;
+out_free_pfns:
+	kvfree(range->pfns);
+	range->pfns = NULL;
+out:
+	up_read(&mm->mmap_sem);
+	return r;
+}
 
-	} while (pinned < ttm->num_pages);
+/**
+ * amdgpu_ttm_tt_userptr_range_done - stop HMM track the CPU page table change
+ * Check if the pages backing this ttm range have been invalidated
+ *
+ * Returns: true if pages are still valid
+ */
+bool amdgpu_ttm_tt_get_user_pages_done(struct ttm_tt *ttm)
+{
+	struct amdgpu_ttm_tt *gtt = (void *)ttm;
+	bool r = false;
 
-	up_read(&mm->mmap_sem);
-	return 0;
+	if (!gtt || !gtt->userptr)
+		return false;
+
+	WARN_ONCE(!gtt->range.pfns, "No user pages to check\n");
+	if (gtt->range.pfns) {
+		r = hmm_vma_range_done(&gtt->range);
+		kvfree(gtt->range.pfns);
+		gtt->range.pfns = NULL;
+	}
 
-release_pages:
-	release_pages(pages, pinned);
-	up_read(&mm->mmap_sem);
 	return r;
 }
 
@@ -809,16 +816,10 @@ int amdgpu_ttm_tt_get_user_pages(struct ttm_tt *ttm, struct page **pages)
  */
 void amdgpu_ttm_tt_set_user_pages(struct ttm_tt *ttm, struct page **pages)
 {
-	struct amdgpu_ttm_tt *gtt = (void *)ttm;
 	unsigned i;
 
-	gtt->last_set_pages = atomic_read(&gtt->mmu_invalidations);
-	for (i = 0; i < ttm->num_pages; ++i) {
-		if (ttm->pages[i])
-			put_page(ttm->pages[i]);
-
+	for (i = 0; i < ttm->num_pages; ++i)
 		ttm->pages[i] = pages ? pages[i] : NULL;
-	}
 }
 
 /**
@@ -903,10 +904,11 @@ static void amdgpu_ttm_tt_unpin_userptr(struct ttm_tt *ttm)
 	/* unmap the pages mapped to the device */
 	dma_unmap_sg(adev->dev, ttm->sg->sgl, ttm->sg->nents, direction);
 
-	/* mark the pages as dirty */
-	amdgpu_ttm_tt_mark_user_pages(ttm);
-
 	sg_free_table(ttm->sg);
+
+	if (gtt->range.pfns &&
+	    ttm->pages[0] == hmm_pfn_to_page(&gtt->range, gtt->range.pfns[0]))
+		WARN_ONCE(1, "Missing get_user_page_done\n");
 }
 
 int amdgpu_ttm_gart_bind(struct amdgpu_device *adev,
@@ -1258,8 +1260,6 @@ int amdgpu_ttm_tt_set_userptr(struct ttm_tt *ttm, uint64_t addr,
 
 	spin_lock_init(&gtt->guptasklock);
 	INIT_LIST_HEAD(&gtt->guptasks);
-	atomic_set(&gtt->mmu_invalidations, 0);
-	gtt->last_set_pages = 0;
 
 	return 0;
 }
@@ -1289,7 +1289,6 @@ bool amdgpu_ttm_tt_affect_userptr(struct ttm_tt *ttm, unsigned long start,
 				  unsigned long end)
 {
 	struct amdgpu_ttm_tt *gtt = (void *)ttm;
-	struct amdgpu_ttm_gup_task_list *entry;
 	unsigned long size;
 
 	if (gtt == NULL || !gtt->userptr)
@@ -1302,48 +1301,20 @@ bool amdgpu_ttm_tt_affect_userptr(struct ttm_tt *ttm, unsigned long start,
 	if (gtt->userptr > end || gtt->userptr + size <= start)
 		return false;
 
-	/* Search the lists of tasks that hold this mapping and see
-	 * if current is one of them.  If it is return false.
-	 */
-	spin_lock(&gtt->guptasklock);
-	list_for_each_entry(entry, &gtt->guptasks, list) {
-		if (entry->task == current) {
-			spin_unlock(&gtt->guptasklock);
-			return false;
-		}
-	}
-	spin_unlock(&gtt->guptasklock);
-
-	atomic_inc(&gtt->mmu_invalidations);
-
 	return true;
 }
 
 /**
- * amdgpu_ttm_tt_userptr_invalidated - Has the ttm_tt object been invalidated?
- */
-bool amdgpu_ttm_tt_userptr_invalidated(struct ttm_tt *ttm,
-				       int *last_invalidated)
-{
-	struct amdgpu_ttm_tt *gtt = (void *)ttm;
-	int prev_invalidated = *last_invalidated;
-
-	*last_invalidated = atomic_read(&gtt->mmu_invalidations);
-	return prev_invalidated != *last_invalidated;
-}
-
-/**
- * amdgpu_ttm_tt_userptr_needs_pages - Have the pages backing this ttm_tt object
- * been invalidated since the last time they've been set?
+ * amdgpu_ttm_tt_is_userptr - Have the pages backing by userptr?
  */
-bool amdgpu_ttm_tt_userptr_needs_pages(struct ttm_tt *ttm)
+bool amdgpu_ttm_tt_is_userptr(struct ttm_tt *ttm)
 {
 	struct amdgpu_ttm_tt *gtt = (void *)ttm;
 
 	if (gtt == NULL || !gtt->userptr)
 		return false;
 
-	return atomic_read(&gtt->mmu_invalidations) != gtt->last_set_pages;
+	return true;
 }
 
 /**
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h
index b5b2d101f7db..8988c87fff9d 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h
@@ -102,6 +102,7 @@ int amdgpu_ttm_alloc_gart(struct ttm_buffer_object *bo);
 int amdgpu_ttm_recover_gart(struct ttm_buffer_object *tbo);
 
 int amdgpu_ttm_tt_get_user_pages(struct ttm_tt *ttm, struct page **pages);
+bool amdgpu_ttm_tt_get_user_pages_done(struct ttm_tt *ttm);
 void amdgpu_ttm_tt_set_user_pages(struct ttm_tt *ttm, struct page **pages);
 void amdgpu_ttm_tt_mark_user_pages(struct ttm_tt *ttm);
 int amdgpu_ttm_tt_set_userptr(struct ttm_tt *ttm, uint64_t addr,
@@ -112,7 +113,7 @@ bool amdgpu_ttm_tt_affect_userptr(struct ttm_tt *ttm, unsigned long start,
 				  unsigned long end);
 bool amdgpu_ttm_tt_userptr_invalidated(struct ttm_tt *ttm,
 				       int *last_invalidated);
-bool amdgpu_ttm_tt_userptr_needs_pages(struct ttm_tt *ttm);
+bool amdgpu_ttm_tt_is_userptr(struct ttm_tt *ttm);
 bool amdgpu_ttm_tt_is_readonly(struct ttm_tt *ttm);
 uint64_t amdgpu_ttm_tt_pde_flags(struct ttm_tt *ttm, struct ttm_mem_reg *mem);
 uint64_t amdgpu_ttm_tt_pte_flags(struct amdgpu_device *adev, struct ttm_tt *ttm,
-- 
2.17.1

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

Re: [PATCH 3/3] drm/amdgpu: replace get_user_pages with HMM address mirror helpers v6

[Yang, Philip]

Ping Christian, any comments for the GEM and CS part changes?

Thanks. Philip

On 2019-01-10 12:02 p.m., Yang, Philip wrote:
> Use HMM helper function hmm_vma_fault() to get physical pages backing
> userptr and start CPU page table update track of those pages. Then use
> hmm_vma_range_done() to check if those pages are updated before
> amdgpu_cs_submit for gfx or before user queues are resumed for kfd.
> 
> If userptr pages are updated, for gfx, amdgpu_cs_ioctl will restart
> from scratch, for kfd, restore worker is rescheduled to retry.
> 
> HMM simplify the CPU page table concurrent update check, so remove
> guptasklock, mmu_invalidations, last_set_pages fields from
> amdgpu_ttm_tt struct.
> 
> HMM does not pin the page (increase page ref count), so remove related
> operations like release_pages(), put_page(), mark_page_dirty().
> 
> Change-Id: I2e8c0c6f0d8c21e5596a32d7fc91762778bc9e67
> Signed-off-by: Philip Yang <Philip.Yang@amd.com>
> ---
>   drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h    |   1 -
>   .../gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c  |  95 +++-------
>   drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.h   |   2 +-
>   drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c        | 158 +++++++---------
>   drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c       |  14 +-
>   drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c        |  25 ++-
>   drivers/gpu/drm/amd/amdgpu/amdgpu_mn.h        |   4 +-
>   drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c       | 173 ++++++++----------
>   drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h       |   3 +-
>   9 files changed, 198 insertions(+), 277 deletions(-)
> 
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
> index 0b31a1859023..0e1711a75b68 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
> @@ -61,7 +61,6 @@ struct kgd_mem {
>   
>   	atomic_t invalid;
>   	struct amdkfd_process_info *process_info;
> -	struct page **user_pages;
>   
>   	struct amdgpu_sync sync;
>   
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
> index d7b10d79f1de..ae2d838d31ea 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
> @@ -582,28 +582,12 @@ static int init_user_pages(struct kgd_mem *mem, struct mm_struct *mm,
>   		goto out;
>   	}
>   
> -	/* If no restore worker is running concurrently, user_pages
> -	 * should not be allocated
> -	 */
> -	WARN(mem->user_pages, "Leaking user_pages array");
> -
> -	mem->user_pages = kvmalloc_array(bo->tbo.ttm->num_pages,
> -					   sizeof(struct page *),
> -					   GFP_KERNEL | __GFP_ZERO);
> -	if (!mem->user_pages) {
> -		pr_err("%s: Failed to allocate pages array\n", __func__);
> -		ret = -ENOMEM;
> -		goto unregister_out;
> -	}
> -
> -	ret = amdgpu_ttm_tt_get_user_pages(bo->tbo.ttm, mem->user_pages);
> +	ret = amdgpu_ttm_tt_get_user_pages(bo->tbo.ttm, bo->tbo.ttm->pages);
>   	if (ret) {
>   		pr_err("%s: Failed to get user pages: %d\n", __func__, ret);
> -		goto free_out;
> +		goto unregister_out;
>   	}
>   
> -	amdgpu_ttm_tt_set_user_pages(bo->tbo.ttm, mem->user_pages);
> -
>   	ret = amdgpu_bo_reserve(bo, true);
>   	if (ret) {
>   		pr_err("%s: Failed to reserve BO\n", __func__);
> @@ -616,11 +600,7 @@ static int init_user_pages(struct kgd_mem *mem, struct mm_struct *mm,
>   	amdgpu_bo_unreserve(bo);
>   
>   release_out:
> -	if (ret)
> -		release_pages(mem->user_pages, bo->tbo.ttm->num_pages);
> -free_out:
> -	kvfree(mem->user_pages);
> -	mem->user_pages = NULL;
> +	amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm);
>   unregister_out:
>   	if (ret)
>   		amdgpu_mn_unregister(bo);
> @@ -679,7 +659,6 @@ static int reserve_bo_and_vm(struct kgd_mem *mem,
>   	ctx->kfd_bo.priority = 0;
>   	ctx->kfd_bo.tv.bo = &bo->tbo;
>   	ctx->kfd_bo.tv.num_shared = 1;
> -	ctx->kfd_bo.user_pages = NULL;
>   	list_add(&ctx->kfd_bo.tv.head, &ctx->list);
>   
>   	amdgpu_vm_get_pd_bo(vm, &ctx->list, &ctx->vm_pd[0]);
> @@ -743,7 +722,6 @@ static int reserve_bo_and_cond_vms(struct kgd_mem *mem,
>   	ctx->kfd_bo.priority = 0;
>   	ctx->kfd_bo.tv.bo = &bo->tbo;
>   	ctx->kfd_bo.tv.num_shared = 1;
> -	ctx->kfd_bo.user_pages = NULL;
>   	list_add(&ctx->kfd_bo.tv.head, &ctx->list);
>   
>   	i = 0;
> @@ -1371,15 +1349,6 @@ int amdgpu_amdkfd_gpuvm_free_memory_of_gpu(
>   	list_del(&bo_list_entry->head);
>   	mutex_unlock(&process_info->lock);
>   
> -	/* Free user pages if necessary */
> -	if (mem->user_pages) {
> -		pr_debug("%s: Freeing user_pages array\n", __func__);
> -		if (mem->user_pages[0])
> -			release_pages(mem->user_pages,
> -					mem->bo->tbo.ttm->num_pages);
> -		kvfree(mem->user_pages);
> -	}
> -
>   	ret = reserve_bo_and_cond_vms(mem, NULL, BO_VM_ALL, &ctx);
>   	if (unlikely(ret))
>   		return ret;
> @@ -1855,25 +1824,11 @@ static int update_invalid_user_pages(struct amdkfd_process_info *process_info,
>   
>   		bo = mem->bo;
>   
> -		if (!mem->user_pages) {
> -			mem->user_pages =
> -				kvmalloc_array(bo->tbo.ttm->num_pages,
> -						 sizeof(struct page *),
> -						 GFP_KERNEL | __GFP_ZERO);
> -			if (!mem->user_pages) {
> -				pr_err("%s: Failed to allocate pages array\n",
> -				       __func__);
> -				return -ENOMEM;
> -			}
> -		} else if (mem->user_pages[0]) {
> -			release_pages(mem->user_pages, bo->tbo.ttm->num_pages);
> -		}
> -
>   		/* Get updated user pages */
>   		ret = amdgpu_ttm_tt_get_user_pages(bo->tbo.ttm,
> -						   mem->user_pages);
> +						   bo->tbo.ttm->pages);
>   		if (ret) {
> -			mem->user_pages[0] = NULL;
> +			bo->tbo.ttm->pages[0] = NULL;
>   			pr_info("%s: Failed to get user pages: %d\n",
>   				__func__, ret);
>   			/* Pretend it succeeded. It will fail later
> @@ -1882,12 +1837,6 @@ static int update_invalid_user_pages(struct amdkfd_process_info *process_info,
>   			 * stalled user mode queues.
>   			 */
>   		}
> -
> -		/* Mark the BO as valid unless it was invalidated
> -		 * again concurrently
> -		 */
> -		if (atomic_cmpxchg(&mem->invalid, invalid, 0) != invalid)
> -			return -EAGAIN;
>   	}
>   
>   	return 0;
> @@ -1917,7 +1866,8 @@ static int validate_invalid_user_pages(struct amdkfd_process_info *process_info)
>   				     GFP_KERNEL);
>   	if (!pd_bo_list_entries) {
>   		pr_err("%s: Failed to allocate PD BO list entries\n", __func__);
> -		return -ENOMEM;
> +		ret = -ENOMEM;
> +		goto out_no_mem;
>   	}
>   
>   	INIT_LIST_HEAD(&resv_list);
> @@ -1941,7 +1891,7 @@ static int validate_invalid_user_pages(struct amdkfd_process_info *process_info)
>   	ret = ttm_eu_reserve_buffers(&ticket, &resv_list, false, &duplicates);
>   	WARN(!list_empty(&duplicates), "Duplicates should be empty");
>   	if (ret)
> -		goto out;
> +		goto out_free;
>   
>   	amdgpu_sync_create(&sync);
>   
> @@ -1967,10 +1917,8 @@ static int validate_invalid_user_pages(struct amdkfd_process_info *process_info)
>   
>   		bo = mem->bo;
>   
> -		/* Copy pages array and validate the BO if we got user pages */
> -		if (mem->user_pages[0]) {
> -			amdgpu_ttm_tt_set_user_pages(bo->tbo.ttm,
> -						     mem->user_pages);
> +		/* Validate the BO if we got user pages */
> +		if (bo->tbo.ttm->pages[0]) {
>   			amdgpu_bo_placement_from_domain(bo, mem->domain);
>   			ret = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
>   			if (ret) {
> @@ -1979,16 +1927,16 @@ static int validate_invalid_user_pages(struct amdkfd_process_info *process_info)
>   			}
>   		}
>   
> -		/* Validate succeeded, now the BO owns the pages, free
> -		 * our copy of the pointer array. Put this BO back on
> -		 * the userptr_valid_list. If we need to revalidate
> -		 * it, we need to start from scratch.
> -		 */
> -		kvfree(mem->user_pages);
> -		mem->user_pages = NULL;
>   		list_move_tail(&mem->validate_list.head,
>   			       &process_info->userptr_valid_list);
>   
> +		/* Stop HMM track the userptr update. We dont check the return
> +		 * value for concurrent CPU page table update because we will
> +		 * reschedule the restore worker if process_info->evicted_bos
> +		 * is updated.
> +		 */
> +		amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm);
> +
>   		/* Update mapping. If the BO was not validated
>   		 * (because we couldn't get user pages), this will
>   		 * clear the page table entries, which will result in
> @@ -2022,8 +1970,15 @@ static int validate_invalid_user_pages(struct amdkfd_process_info *process_info)
>   	ttm_eu_backoff_reservation(&ticket, &resv_list);
>   	amdgpu_sync_wait(&sync, false);
>   	amdgpu_sync_free(&sync);
> -out:
> +out_free:
>   	kfree(pd_bo_list_entries);
> +out_no_mem:
> +	list_for_each_entry_safe(mem, tmp_mem,
> +				 &process_info->userptr_inval_list,
> +				 validate_list.head) {
> +		bo = mem->bo;
> +		amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm);
> +	}
>   
>   	return ret;
>   }
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.h
> index 7c5f5d1601e6..a130e766cbdb 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.h
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.h
> @@ -36,7 +36,7 @@ struct amdgpu_bo_list_entry {
>   	struct amdgpu_bo_va		*bo_va;
>   	uint32_t			priority;
>   	struct page			**user_pages;
> -	int				user_invalidated;
> +	bool				user_invalidated;
>   };
>   
>   struct amdgpu_bo_list {
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
> index 1c49b8266d69..451a1fab17d6 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
> @@ -52,7 +52,6 @@ static int amdgpu_cs_user_fence_chunk(struct amdgpu_cs_parser *p,
>   	p->uf_entry.tv.bo = &bo->tbo;
>   	/* One for TTM and one for the CS job */
>   	p->uf_entry.tv.num_shared = 2;
> -	p->uf_entry.user_pages = NULL;
>   
>   	drm_gem_object_put_unlocked(gobj);
>   
> @@ -539,14 +538,14 @@ static int amdgpu_cs_list_validate(struct amdgpu_cs_parser *p,
>   		if (usermm && usermm != current->mm)
>   			return -EPERM;
>   
> -		/* Check if we have user pages and nobody bound the BO already */
> -		if (amdgpu_ttm_tt_userptr_needs_pages(bo->tbo.ttm) &&
> -		    lobj->user_pages) {
> +		if (amdgpu_ttm_tt_is_userptr(bo->tbo.ttm) &&
> +		    lobj->user_invalidated && lobj->user_pages) {
>   			amdgpu_bo_placement_from_domain(bo,
>   							AMDGPU_GEM_DOMAIN_CPU);
>   			r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
>   			if (r)
>   				return r;
> +
>   			amdgpu_ttm_tt_set_user_pages(bo->tbo.ttm,
>   						     lobj->user_pages);
>   			binding_userptr = true;
> @@ -577,7 +576,6 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
>   	struct amdgpu_bo *gds;
>   	struct amdgpu_bo *gws;
>   	struct amdgpu_bo *oa;
> -	unsigned tries = 10;
>   	int r;
>   
>   	INIT_LIST_HEAD(&p->validated);
> @@ -613,79 +611,45 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
>   	if (p->uf_entry.tv.bo && !ttm_to_amdgpu_bo(p->uf_entry.tv.bo)->parent)
>   		list_add(&p->uf_entry.tv.head, &p->validated);
>   
> -	while (1) {
> -		struct list_head need_pages;
> -
> -		r = ttm_eu_reserve_buffers(&p->ticket, &p->validated, true,
> -					   &duplicates);
> -		if (unlikely(r != 0)) {
> -			if (r != -ERESTARTSYS)
> -				DRM_ERROR("ttm_eu_reserve_buffers failed.\n");
> -			goto error_free_pages;
> -		}
> -
> -		INIT_LIST_HEAD(&need_pages);
> -		amdgpu_bo_list_for_each_userptr_entry(e, p->bo_list) {
> -			struct amdgpu_bo *bo = ttm_to_amdgpu_bo(e->tv.bo);
> -
> -			if (amdgpu_ttm_tt_userptr_invalidated(bo->tbo.ttm,
> -				 &e->user_invalidated) && e->user_pages) {
> -
> -				/* We acquired a page array, but somebody
> -				 * invalidated it. Free it and try again
> -				 */
> -				release_pages(e->user_pages,
> -					      bo->tbo.ttm->num_pages);
> -				kvfree(e->user_pages);
> -				e->user_pages = NULL;
> -			}
> -
> -			if (amdgpu_ttm_tt_userptr_needs_pages(bo->tbo.ttm) &&
> -			    !e->user_pages) {
> -				list_del(&e->tv.head);
> -				list_add(&e->tv.head, &need_pages);
> -
> -				amdgpu_bo_unreserve(bo);
> -			}
> +	/* Get userptr backing pages. If pages are updated after registered
> +	 * in amdgpu_gem_userptr_ioctl(), amdgpu_cs_list_validate() will do
> +	 * amdgpu_ttm_backend_bind() to flush and invalidate new pages
> +	 */
> +	amdgpu_bo_list_for_each_userptr_entry(e, p->bo_list) {
> +		struct amdgpu_bo *bo = ttm_to_amdgpu_bo(e->tv.bo);
> +		bool userpage_invalidated = false;
> +		int i;
> +
> +		e->user_pages = kvmalloc_array(bo->tbo.ttm->num_pages,
> +					sizeof(struct page *),
> +					GFP_KERNEL | __GFP_ZERO);
> +		if (!e->user_pages) {
> +			DRM_ERROR("calloc failure\n");
> +			return -ENOMEM;
>   		}
>   
> -		if (list_empty(&need_pages))
> -			break;
> -
> -		/* Unreserve everything again. */
> -		ttm_eu_backoff_reservation(&p->ticket, &p->validated);
> -
> -		/* We tried too many times, just abort */
> -		if (!--tries) {
> -			r = -EDEADLK;
> -			DRM_ERROR("deadlock in %s\n", __func__);
> -			goto error_free_pages;
> +		r = amdgpu_ttm_tt_get_user_pages(bo->tbo.ttm, e->user_pages);
> +		if (r) {
> +			kvfree(e->user_pages);
> +			e->user_pages = NULL;
> +			return r;
>   		}
>   
> -		/* Fill the page arrays for all userptrs. */
> -		list_for_each_entry(e, &need_pages, tv.head) {
> -			struct ttm_tt *ttm = e->tv.bo->ttm;
> -
> -			e->user_pages = kvmalloc_array(ttm->num_pages,
> -							 sizeof(struct page*),
> -							 GFP_KERNEL | __GFP_ZERO);
> -			if (!e->user_pages) {
> -				r = -ENOMEM;
> -				DRM_ERROR("calloc failure in %s\n", __func__);
> -				goto error_free_pages;
> -			}
> -
> -			r = amdgpu_ttm_tt_get_user_pages(ttm, e->user_pages);
> -			if (r) {
> -				DRM_ERROR("amdgpu_ttm_tt_get_user_pages failed.\n");
> -				kvfree(e->user_pages);
> -				e->user_pages = NULL;
> -				goto error_free_pages;
> +		for (i = 0; i < bo->tbo.ttm->num_pages; i++) {
> +			if (bo->tbo.ttm->pages[i] != e->user_pages[i]) {
> +				userpage_invalidated = true;
> +				break;
>   			}
>   		}
> +		e->user_invalidated = userpage_invalidated;
> +	}
>   
> -		/* And try again. */
> -		list_splice(&need_pages, &p->validated);
> +	r = ttm_eu_reserve_buffers(&p->ticket, &p->validated, true,
> +				   &duplicates);
> +	if (unlikely(r != 0)) {
> +		if (r != -ERESTARTSYS)
> +			DRM_ERROR("ttm_eu_reserve_buffers failed.\n");
> +		goto out;
>   	}
>   
>   	amdgpu_cs_get_threshold_for_moves(p->adev, &p->bytes_moved_threshold,
> @@ -754,17 +718,7 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
>   error_validate:
>   	if (r)
>   		ttm_eu_backoff_reservation(&p->ticket, &p->validated);
> -
> -error_free_pages:
> -
> -	amdgpu_bo_list_for_each_userptr_entry(e, p->bo_list) {
> -		if (!e->user_pages)
> -			continue;
> -
> -		release_pages(e->user_pages, e->tv.bo->ttm->num_pages);
> -		kvfree(e->user_pages);
> -	}
> -
> +out:
>   	return r;
>   }
>   
> @@ -1213,8 +1167,7 @@ static int amdgpu_cs_submit(struct amdgpu_cs_parser *p,
>   	struct amdgpu_bo_list_entry *e;
>   	struct amdgpu_job *job;
>   	uint64_t seq;
> -
> -	int r;
> +	int r = 0;
>   
>   	job = p->job;
>   	p->job = NULL;
> @@ -1223,15 +1176,21 @@ static int amdgpu_cs_submit(struct amdgpu_cs_parser *p,
>   	if (r)
>   		goto error_unlock;
>   
> -	/* No memory allocation is allowed while holding the mn lock */
> +	/* No memory allocation is allowed while holding the mn lock.
> +	 * p->mn is hold until amdgpu_cs_submit is finished and fence is added
> +	 * to BOs.
> +	 */
>   	amdgpu_mn_lock(p->mn);
> +
> +	/* If userptr are updated after amdgpu_cs_parser_bos(), restart cs */
>   	amdgpu_bo_list_for_each_userptr_entry(e, p->bo_list) {
>   		struct amdgpu_bo *bo = ttm_to_amdgpu_bo(e->tv.bo);
>   
> -		if (amdgpu_ttm_tt_userptr_needs_pages(bo->tbo.ttm)) {
> -			r = -ERESTARTSYS;
> -			goto error_abort;
> -		}
> +		r |= !amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm);
> +	}
> +	if (r) {
> +		r = -ERESTARTSYS;
> +		goto error_abort;
>   	}
>   
>   	job->owner = p->filp;
> @@ -1278,14 +1237,20 @@ static int amdgpu_cs_submit(struct amdgpu_cs_parser *p,
>   int amdgpu_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
>   {
>   	struct amdgpu_device *adev = dev->dev_private;
> -	union drm_amdgpu_cs *cs = data;
> -	struct amdgpu_cs_parser parser = {};
> -	bool reserved_buffers = false;
> +	union drm_amdgpu_cs *cs;
> +	struct amdgpu_cs_parser parser;
> +	bool reserved_buffers;
> +	int tries = 10;
>   	int i, r;
>   
>   	if (!adev->accel_working)
>   		return -EBUSY;
>   
> +restart:
> +	memset(&parser, 0, sizeof(parser));
> +	cs = data;
> +	reserved_buffers = false;
> +
>   	parser.adev = adev;
>   	parser.filp = filp;
>   
> @@ -1327,6 +1292,15 @@ int amdgpu_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
>   
>   out:
>   	amdgpu_cs_parser_fini(&parser, r, reserved_buffers);
> +
> +	if (r == -ERESTARTSYS) {
> +		if (!--tries) {
> +			DRM_ERROR("Possible deadlock? Retry too many times\n");
> +			return -EDEADLK;
> +		}
> +		goto restart;
> +	}
> +
>   	return r;
>   }
>   
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
> index f4f00217546e..92025993bfb1 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
> @@ -336,26 +336,24 @@ int amdgpu_gem_userptr_ioctl(struct drm_device *dev, void *data,
>   
>   		r = amdgpu_bo_reserve(bo, true);
>   		if (r)
> -			goto free_pages;
> +			goto user_pages_done;
>   
>   		amdgpu_bo_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_GTT);
>   		r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
>   		amdgpu_bo_unreserve(bo);
>   		if (r)
> -			goto free_pages;
> +			goto user_pages_done;
>   	}
>   
>   	r = drm_gem_handle_create(filp, gobj, &handle);
> -	/* drop reference from allocate - handle holds it now */
> -	drm_gem_object_put_unlocked(gobj);
>   	if (r)
> -		return r;
> +		goto user_pages_done;
>   
>   	args->handle = handle;
> -	return 0;
>   
> -free_pages:
> -	release_pages(bo->tbo.ttm->pages, bo->tbo.ttm->num_pages);
> +user_pages_done:
> +	if (args->flags & AMDGPU_GEM_USERPTR_VALIDATE)
> +		amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm);
>   
>   release_object:
>   	drm_gem_object_put_unlocked(gobj);
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c
> index 5d518d2bb9be..9d40f3001f3c 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c
> @@ -229,8 +229,6 @@ static void amdgpu_mn_invalidate_node(struct amdgpu_mn_node *node,
>   			true, false, MAX_SCHEDULE_TIMEOUT);
>   		if (r <= 0)
>   			DRM_ERROR("(%ld) failed to wait for user bo\n", r);
> -
> -		amdgpu_ttm_tt_mark_user_pages(bo->tbo.ttm);
>   	}
>   }
>   
> @@ -513,3 +511,26 @@ void amdgpu_mn_unregister(struct amdgpu_bo *bo)
>   	mutex_unlock(&adev->mn_lock);
>   }
>   
> +/* flags used by HMM internal, not related to CPU/GPU PTE flags */
> +static const uint64_t hmm_range_flags[HMM_PFN_FLAG_MAX] = {
> +		(1 << 0), /* HMM_PFN_VALID */
> +		(1 << 1), /* HMM_PFN_WRITE */
> +		0 /* HMM_PFN_DEVICE_PRIVATE */
> +};
> +
> +static const uint64_t hmm_range_values[HMM_PFN_VALUE_MAX] = {
> +		0xfffffffffffffffeUL, /* HMM_PFN_ERROR */
> +		0, /* HMM_PFN_NONE */
> +		0xfffffffffffffffcUL /* HMM_PFN_SPECIAL */
> +};
> +
> +void amdgpu_hmm_init_range(struct hmm_range *range)
> +{
> +	if (range) {
> +		range->flags = hmm_range_flags;
> +		range->values = hmm_range_values;
> +		range->pfn_shift = PAGE_SHIFT;
> +		range->pfns = NULL;
> +		INIT_LIST_HEAD(&range->list);
> +	}
> +}
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.h
> index 0a51fd00021c..4803e216e174 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.h
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.h
> @@ -25,9 +25,10 @@
>   #define __AMDGPU_MN_H__
>   
>   /*
> - * MMU Notifier
> + * HMM mirror
>    */
>   struct amdgpu_mn;
> +struct hmm_range;
>   
>   enum amdgpu_mn_type {
>   	AMDGPU_MN_TYPE_GFX,
> @@ -41,6 +42,7 @@ struct amdgpu_mn *amdgpu_mn_get(struct amdgpu_device *adev,
>   				enum amdgpu_mn_type type);
>   int amdgpu_mn_register(struct amdgpu_bo *bo, unsigned long addr);
>   void amdgpu_mn_unregister(struct amdgpu_bo *bo);
> +void amdgpu_hmm_init_range(struct hmm_range *range);
>   #else
>   static inline void amdgpu_mn_lock(struct amdgpu_mn *mn) {}
>   static inline void amdgpu_mn_unlock(struct amdgpu_mn *mn) {}
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
> index c91ec3101d00..3d074f4b02f4 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
> @@ -43,6 +43,7 @@
>   #include <linux/pagemap.h>
>   #include <linux/debugfs.h>
>   #include <linux/iommu.h>
> +#include <linux/hmm.h>
>   #include "amdgpu.h"
>   #include "amdgpu_object.h"
>   #include "amdgpu_trace.h"
> @@ -705,11 +706,6 @@ static unsigned long amdgpu_ttm_io_mem_pfn(struct ttm_buffer_object *bo,
>   /*
>    * TTM backend functions.
>    */
> -struct amdgpu_ttm_gup_task_list {
> -	struct list_head	list;
> -	struct task_struct	*task;
> -};
> -
>   struct amdgpu_ttm_tt {
>   	struct ttm_dma_tt	ttm;
>   	u64			offset;
> @@ -718,85 +714,96 @@ struct amdgpu_ttm_tt {
>   	uint32_t		userflags;
>   	spinlock_t              guptasklock;
>   	struct list_head        guptasks;
> -	atomic_t		mmu_invalidations;
> -	uint32_t		last_set_pages;
> +	struct hmm_range	range;
>   };
>   
>   /**
> - * amdgpu_ttm_tt_get_user_pages - Pin pages of memory pointed to by a USERPTR
> - * pointer to memory
> + * amdgpu_ttm_tt_get_user_pages - get device accessible pages that back user
> + * memory and start HMM tracking CPU page table update
>    *
> - * Called by amdgpu_gem_userptr_ioctl() and amdgpu_cs_parser_bos().
> - * This provides a wrapper around the get_user_pages() call to provide
> - * device accessible pages that back user memory.
> + * Calling function must call amdgpu_ttm_tt_userptr_range_done() once and only
> + * once afterwards to stop HMM tracking
>    */
>   int amdgpu_ttm_tt_get_user_pages(struct ttm_tt *ttm, struct page **pages)
>   {
>   	struct amdgpu_ttm_tt *gtt = (void *)ttm;
>   	struct mm_struct *mm = gtt->usertask->mm;
> -	unsigned int flags = 0;
> -	unsigned pinned = 0;
> -	int r;
> +	unsigned long end = gtt->userptr + ttm->num_pages * PAGE_SIZE;
> +	struct hmm_range *range = &gtt->range;
> +	int r = 0, i;
>   
>   	if (!mm) /* Happens during process shutdown */
>   		return -ESRCH;
>   
> -	if (!(gtt->userflags & AMDGPU_GEM_USERPTR_READONLY))
> -		flags |= FOLL_WRITE;
> +	amdgpu_hmm_init_range(range);
>   
>   	down_read(&mm->mmap_sem);
>   
> -	if (gtt->userflags & AMDGPU_GEM_USERPTR_ANONONLY) {
> -		/*
> -		 * check that we only use anonymous memory to prevent problems
> -		 * with writeback
> -		 */
> -		unsigned long end = gtt->userptr + ttm->num_pages * PAGE_SIZE;
> -		struct vm_area_struct *vma;
> +	range->vma = find_vma(mm, gtt->userptr);
> +	if (!range_in_vma(range->vma, gtt->userptr, end))
> +		r = -EFAULT;
> +	else if ((gtt->userflags & AMDGPU_GEM_USERPTR_ANONONLY) &&
> +		range->vma->vm_file)
> +		r = -EPERM;
> +	if (r)
> +		goto out;
>   
> -		vma = find_vma(mm, gtt->userptr);
> -		if (!vma || vma->vm_file || vma->vm_end < end) {
> -			up_read(&mm->mmap_sem);
> -			return -EPERM;
> -		}
> +	range->pfns = kvmalloc_array(ttm->num_pages, sizeof(uint64_t),
> +				     GFP_KERNEL);
> +	if (range->pfns == NULL) {
> +		r = -ENOMEM;
> +		goto out;
>   	}
> +	range->start = gtt->userptr;
> +	range->end = end;
>   
> -	/* loop enough times using contiguous pages of memory */
> -	do {
> -		unsigned num_pages = ttm->num_pages - pinned;
> -		uint64_t userptr = gtt->userptr + pinned * PAGE_SIZE;
> -		struct page **p = pages + pinned;
> -		struct amdgpu_ttm_gup_task_list guptask;
> +	range->pfns[0] = range->flags[HMM_PFN_VALID];
> +	range->pfns[0] |= amdgpu_ttm_tt_is_readonly(ttm) ?
> +				0 : range->flags[HMM_PFN_WRITE];
> +	for (i = 1; i < ttm->num_pages; i++)
> +		range->pfns[i] = range->pfns[0];
>   
> -		guptask.task = current;
> -		spin_lock(&gtt->guptasklock);
> -		list_add(&guptask.list, &gtt->guptasks);
> -		spin_unlock(&gtt->guptasklock);
> +	/* This may trigger page table update */
> +	r = hmm_vma_fault(range, true);
> +	if (r)
> +		goto out_free_pfns;
>   
> -		if (mm == current->mm)
> -			r = get_user_pages(userptr, num_pages, flags, p, NULL);
> -		else
> -			r = get_user_pages_remote(gtt->usertask,
> -					mm, userptr, num_pages,
> -					flags, p, NULL, NULL);
> +	up_read(&mm->mmap_sem);
>   
> -		spin_lock(&gtt->guptasklock);
> -		list_del(&guptask.list);
> -		spin_unlock(&gtt->guptasklock);
> +	for (i = 0; i < ttm->num_pages; i++)
> +		pages[i] = hmm_pfn_to_page(range, range->pfns[i]);
>   
> -		if (r < 0)
> -			goto release_pages;
> +	return 0;
>   
> -		pinned += r;
> +out_free_pfns:
> +	kvfree(range->pfns);
> +	range->pfns = NULL;
> +out:
> +	up_read(&mm->mmap_sem);
> +	return r;
> +}
>   
> -	} while (pinned < ttm->num_pages);
> +/**
> + * amdgpu_ttm_tt_userptr_range_done - stop HMM track the CPU page table change
> + * Check if the pages backing this ttm range have been invalidated
> + *
> + * Returns: true if pages are still valid
> + */
> +bool amdgpu_ttm_tt_get_user_pages_done(struct ttm_tt *ttm)
> +{
> +	struct amdgpu_ttm_tt *gtt = (void *)ttm;
> +	bool r = false;
>   
> -	up_read(&mm->mmap_sem);
> -	return 0;
> +	if (!gtt || !gtt->userptr)
> +		return false;
> +
> +	WARN_ONCE(!gtt->range.pfns, "No user pages to check\n");
> +	if (gtt->range.pfns) {
> +		r = hmm_vma_range_done(&gtt->range);
> +		kvfree(gtt->range.pfns);
> +		gtt->range.pfns = NULL;
> +	}
>   
> -release_pages:
> -	release_pages(pages, pinned);
> -	up_read(&mm->mmap_sem);
>   	return r;
>   }
>   
> @@ -809,16 +816,10 @@ int amdgpu_ttm_tt_get_user_pages(struct ttm_tt *ttm, struct page **pages)
>    */
>   void amdgpu_ttm_tt_set_user_pages(struct ttm_tt *ttm, struct page **pages)
>   {
> -	struct amdgpu_ttm_tt *gtt = (void *)ttm;
>   	unsigned i;
>   
> -	gtt->last_set_pages = atomic_read(&gtt->mmu_invalidations);
> -	for (i = 0; i < ttm->num_pages; ++i) {
> -		if (ttm->pages[i])
> -			put_page(ttm->pages[i]);
> -
> +	for (i = 0; i < ttm->num_pages; ++i)
>   		ttm->pages[i] = pages ? pages[i] : NULL;
> -	}
>   }
>   
>   /**
> @@ -903,10 +904,11 @@ static void amdgpu_ttm_tt_unpin_userptr(struct ttm_tt *ttm)
>   	/* unmap the pages mapped to the device */
>   	dma_unmap_sg(adev->dev, ttm->sg->sgl, ttm->sg->nents, direction);
>   
> -	/* mark the pages as dirty */
> -	amdgpu_ttm_tt_mark_user_pages(ttm);
> -
>   	sg_free_table(ttm->sg);
> +
> +	if (gtt->range.pfns &&
> +	    ttm->pages[0] == hmm_pfn_to_page(&gtt->range, gtt->range.pfns[0]))
> +		WARN_ONCE(1, "Missing get_user_page_done\n");
>   }
>   
>   int amdgpu_ttm_gart_bind(struct amdgpu_device *adev,
> @@ -1258,8 +1260,6 @@ int amdgpu_ttm_tt_set_userptr(struct ttm_tt *ttm, uint64_t addr,
>   
>   	spin_lock_init(&gtt->guptasklock);
>   	INIT_LIST_HEAD(&gtt->guptasks);
> -	atomic_set(&gtt->mmu_invalidations, 0);
> -	gtt->last_set_pages = 0;
>   
>   	return 0;
>   }
> @@ -1289,7 +1289,6 @@ bool amdgpu_ttm_tt_affect_userptr(struct ttm_tt *ttm, unsigned long start,
>   				  unsigned long end)
>   {
>   	struct amdgpu_ttm_tt *gtt = (void *)ttm;
> -	struct amdgpu_ttm_gup_task_list *entry;
>   	unsigned long size;
>   
>   	if (gtt == NULL || !gtt->userptr)
> @@ -1302,48 +1301,20 @@ bool amdgpu_ttm_tt_affect_userptr(struct ttm_tt *ttm, unsigned long start,
>   	if (gtt->userptr > end || gtt->userptr + size <= start)
>   		return false;
>   
> -	/* Search the lists of tasks that hold this mapping and see
> -	 * if current is one of them.  If it is return false.
> -	 */
> -	spin_lock(&gtt->guptasklock);
> -	list_for_each_entry(entry, &gtt->guptasks, list) {
> -		if (entry->task == current) {
> -			spin_unlock(&gtt->guptasklock);
> -			return false;
> -		}
> -	}
> -	spin_unlock(&gtt->guptasklock);
> -
> -	atomic_inc(&gtt->mmu_invalidations);
> -
>   	return true;
>   }
>   
>   /**
> - * amdgpu_ttm_tt_userptr_invalidated - Has the ttm_tt object been invalidated?
> - */
> -bool amdgpu_ttm_tt_userptr_invalidated(struct ttm_tt *ttm,
> -				       int *last_invalidated)
> -{
> -	struct amdgpu_ttm_tt *gtt = (void *)ttm;
> -	int prev_invalidated = *last_invalidated;
> -
> -	*last_invalidated = atomic_read(&gtt->mmu_invalidations);
> -	return prev_invalidated != *last_invalidated;
> -}
> -
> -/**
> - * amdgpu_ttm_tt_userptr_needs_pages - Have the pages backing this ttm_tt object
> - * been invalidated since the last time they've been set?
> + * amdgpu_ttm_tt_is_userptr - Have the pages backing by userptr?
>    */
> -bool amdgpu_ttm_tt_userptr_needs_pages(struct ttm_tt *ttm)
> +bool amdgpu_ttm_tt_is_userptr(struct ttm_tt *ttm)
>   {
>   	struct amdgpu_ttm_tt *gtt = (void *)ttm;
>   
>   	if (gtt == NULL || !gtt->userptr)
>   		return false;
>   
> -	return atomic_read(&gtt->mmu_invalidations) != gtt->last_set_pages;
> +	return true;
>   }
>   
>   /**
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h
> index b5b2d101f7db..8988c87fff9d 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h
> @@ -102,6 +102,7 @@ int amdgpu_ttm_alloc_gart(struct ttm_buffer_object *bo);
>   int amdgpu_ttm_recover_gart(struct ttm_buffer_object *tbo);
>   
>   int amdgpu_ttm_tt_get_user_pages(struct ttm_tt *ttm, struct page **pages);
> +bool amdgpu_ttm_tt_get_user_pages_done(struct ttm_tt *ttm);
>   void amdgpu_ttm_tt_set_user_pages(struct ttm_tt *ttm, struct page **pages);
>   void amdgpu_ttm_tt_mark_user_pages(struct ttm_tt *ttm);
>   int amdgpu_ttm_tt_set_userptr(struct ttm_tt *ttm, uint64_t addr,
> @@ -112,7 +113,7 @@ bool amdgpu_ttm_tt_affect_userptr(struct ttm_tt *ttm, unsigned long start,
>   				  unsigned long end);
>   bool amdgpu_ttm_tt_userptr_invalidated(struct ttm_tt *ttm,
>   				       int *last_invalidated);
> -bool amdgpu_ttm_tt_userptr_needs_pages(struct ttm_tt *ttm);
> +bool amdgpu_ttm_tt_is_userptr(struct ttm_tt *ttm);
>   bool amdgpu_ttm_tt_is_readonly(struct ttm_tt *ttm);
>   uint64_t amdgpu_ttm_tt_pde_flags(struct ttm_tt *ttm, struct ttm_mem_reg *mem);
>   uint64_t amdgpu_ttm_tt_pte_flags(struct amdgpu_device *adev, struct ttm_tt *ttm,
> 
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

[PATCH 0/3] Use HMM to replace get_user_pages

[Yang, Philip]

Hi Christian,

This patch is rebased to lastest HMM. Please review the GEM and CS part changes
in patch 3/3.

Regards,

Philip Yang (3):
  drm/amdgpu: use HMM mirror callback to replace mmu notifier v6
  drm/amdkfd: avoid HMM change cause circular lock dependency v2
  drm/amdgpu: replace get_user_pages with HMM address mirror helpers v6

 drivers/gpu/drm/amd/amdgpu/Kconfig            |   6 +-
 drivers/gpu/drm/amd/amdgpu/Makefile           |   2 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h    |   1 -
 .../gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c  |  95 +++-------
 drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.h   |   2 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c        | 158 +++++++---------
 drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c       |  14 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c        | 164 +++++++++--------
 drivers/gpu/drm/amd/amdgpu/amdgpu_mn.h        |   6 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c       | 173 ++++++++----------
 drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h       |   3 +-
 .../drm/amd/amdkfd/kfd_device_queue_manager.c |  32 ++--
 12 files changed, 282 insertions(+), 374 deletions(-)

-- 
2.17.1

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

[PATCH 2/3] drm/amdkfd: avoid HMM change cause circular lock dependency v2

[Yang, Philip]

There is circular lock between gfx and kfd path with HMM change:
lock(dqm) -> bo::reserve -> amdgpu_mn_lock

To avoid this, move init/unint_mqd() out of lock(dqm), to remove nested
locking between mmap_sem and bo::reserve. The locking order
is: bo::reserve -> amdgpu_mn_lock(p->mn)

Change-Id: I2ec09a47571f6b4c8eaef93f22c0a600f5f70153
Signed-off-by: Philip Yang <Philip.Yang@amd.com>
---
 .../drm/amd/amdkfd/kfd_device_queue_manager.c | 32 ++++++++++---------
 1 file changed, 17 insertions(+), 15 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
index 8372556b52eb..efe0d3c0215b 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
@@ -1156,21 +1156,17 @@ static int create_queue_cpsch(struct device_queue_manager *dqm, struct queue *q,
 	int retval;
 	struct mqd_manager *mqd_mgr;
 
-	retval = 0;
-
-	dqm_lock(dqm);
-
 	if (dqm->total_queue_count >= max_num_of_queues_per_device) {
 		pr_warn("Can't create new usermode queue because %d queues were already created\n",
 				dqm->total_queue_count);
 		retval = -EPERM;
-		goto out_unlock;
+		goto out;
 	}
 
 	if (q->properties.type == KFD_QUEUE_TYPE_SDMA) {
 		retval = allocate_sdma_queue(dqm, &q->sdma_id);
 		if (retval)
-			goto out_unlock;
+			goto out;
 		q->properties.sdma_queue_id =
 			q->sdma_id / get_num_sdma_engines(dqm);
 		q->properties.sdma_engine_id =
@@ -1181,6 +1177,9 @@ static int create_queue_cpsch(struct device_queue_manager *dqm, struct queue *q,
 	if (retval)
 		goto out_deallocate_sdma_queue;
 
+	/* Do init_mqd before dqm_lock(dqm) to avoid circular locking order:
+	 * lock(dqm) -> bo::reserve
+	 */
 	mqd_mgr = dqm->ops.get_mqd_manager(dqm,
 			get_mqd_type_from_queue_type(q->properties.type));
 
@@ -1188,6 +1187,7 @@ static int create_queue_cpsch(struct device_queue_manager *dqm, struct queue *q,
 		retval = -ENOMEM;
 		goto out_deallocate_doorbell;
 	}
+
 	/*
 	 * Eviction state logic: we only mark active queues as evicted
 	 * to avoid the overhead of restoring inactive queues later
@@ -1196,9 +1196,7 @@ static int create_queue_cpsch(struct device_queue_manager *dqm, struct queue *q,
 		q->properties.is_evicted = (q->properties.queue_size > 0 &&
 					    q->properties.queue_percent > 0 &&
 					    q->properties.queue_address != 0);
-
 	dqm->asic_ops.init_sdma_vm(dqm, q, qpd);
-
 	q->properties.tba_addr = qpd->tba_addr;
 	q->properties.tma_addr = qpd->tma_addr;
 	retval = mqd_mgr->init_mqd(mqd_mgr, &q->mqd, &q->mqd_mem_obj,
@@ -1206,6 +1204,8 @@ static int create_queue_cpsch(struct device_queue_manager *dqm, struct queue *q,
 	if (retval)
 		goto out_deallocate_doorbell;
 
+	dqm_lock(dqm);
+
 	list_add(&q->list, &qpd->queues_list);
 	qpd->queue_count++;
 	if (q->properties.is_active) {
@@ -1233,9 +1233,7 @@ static int create_queue_cpsch(struct device_queue_manager *dqm, struct queue *q,
 out_deallocate_sdma_queue:
 	if (q->properties.type == KFD_QUEUE_TYPE_SDMA)
 		deallocate_sdma_queue(dqm, q->sdma_id);
-out_unlock:
-	dqm_unlock(dqm);
-
+out:
 	return retval;
 }
 
@@ -1398,8 +1396,6 @@ static int destroy_queue_cpsch(struct device_queue_manager *dqm,
 			qpd->reset_wavefronts = true;
 	}
 
-	mqd_mgr->uninit_mqd(mqd_mgr, q->mqd, q->mqd_mem_obj);
-
 	/*
 	 * Unconditionally decrement this counter, regardless of the queue's
 	 * type
@@ -1410,6 +1406,9 @@ static int destroy_queue_cpsch(struct device_queue_manager *dqm,
 
 	dqm_unlock(dqm);
 
+	/* Do uninit_mqd after dqm_unlock(dqm) to avoid circular locking */
+	mqd_mgr->uninit_mqd(mqd_mgr, q->mqd, q->mqd_mem_obj);
+
 	return retval;
 
 failed:
@@ -1631,7 +1630,11 @@ static int process_termination_cpsch(struct device_queue_manager *dqm,
 		qpd->reset_wavefronts = false;
 	}
 
-	/* lastly, free mqd resources */
+	dqm_unlock(dqm);
+
+	/* Lastly, free mqd resources.
+	 * Do uninit_mqd() after dqm_unlock to avoid circular locking.
+	 */
 	list_for_each_entry_safe(q, next, &qpd->queues_list, list) {
 		mqd_mgr = dqm->ops.get_mqd_manager(dqm,
 			get_mqd_type_from_queue_type(q->properties.type));
@@ -1645,7 +1648,6 @@ static int process_termination_cpsch(struct device_queue_manager *dqm,
 	}
 
 out:
-	dqm_unlock(dqm);
 	return retval;
 }
 
-- 
2.17.1

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

[PATCH 0/3] Use HMM to replace get_user_pages

[Yang, Philip]

Hi Christian,

This patch is rebased to lastest HMM. Please review the GEM and CS part changes
in patch 3/3.

Thanks,

Philip Yang (3):
  drm/amdgpu: use HMM mirror callback to replace mmu notifier v7
  drm/amdkfd: avoid HMM change cause circular lock dependency v2
  drm/amdgpu: replace get_user_pages with HMM address mirror helpers v6

 drivers/gpu/drm/amd/amdgpu/Kconfig            |   6 +-
 drivers/gpu/drm/amd/amdgpu/Makefile           |   2 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h    |   1 -
 .../gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c  |  95 +++-------
 drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.h   |   2 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c        | 158 +++++++---------
 drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c       |  14 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c        | 179 +++++++++---------
 drivers/gpu/drm/amd/amdgpu/amdgpu_mn.h        |   6 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c       | 173 +++++++----------
 drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h       |   3 +-
 .../drm/amd/amdkfd/kfd_device_queue_manager.c |  32 ++--
 12 files changed, 285 insertions(+), 386 deletions(-)

-- 
2.17.1

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

[PATCH 2/3] drm/amdkfd: avoid HMM change cause circular lock dependency v2

[Yang, Philip]

There is circular lock between gfx and kfd path with HMM change:
lock(dqm) -> bo::reserve -> amdgpu_mn_lock

To avoid this, move init/unint_mqd() out of lock(dqm), to remove nested
locking between mmap_sem and bo::reserve. The locking order
is: bo::reserve -> amdgpu_mn_lock(p->mn)

Change-Id: I2ec09a47571f6b4c8eaef93f22c0a600f5f70153
Signed-off-by: Philip Yang <Philip.Yang@amd.com>
---
 .../drm/amd/amdkfd/kfd_device_queue_manager.c | 32 ++++++++++---------
 1 file changed, 17 insertions(+), 15 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
index 8372556b52eb..efe0d3c0215b 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
@@ -1156,21 +1156,17 @@ static int create_queue_cpsch(struct device_queue_manager *dqm, struct queue *q,
 	int retval;
 	struct mqd_manager *mqd_mgr;
 
-	retval = 0;
-
-	dqm_lock(dqm);
-
 	if (dqm->total_queue_count >= max_num_of_queues_per_device) {
 		pr_warn("Can't create new usermode queue because %d queues were already created\n",
 				dqm->total_queue_count);
 		retval = -EPERM;
-		goto out_unlock;
+		goto out;
 	}
 
 	if (q->properties.type == KFD_QUEUE_TYPE_SDMA) {
 		retval = allocate_sdma_queue(dqm, &q->sdma_id);
 		if (retval)
-			goto out_unlock;
+			goto out;
 		q->properties.sdma_queue_id =
 			q->sdma_id / get_num_sdma_engines(dqm);
 		q->properties.sdma_engine_id =
@@ -1181,6 +1177,9 @@ static int create_queue_cpsch(struct device_queue_manager *dqm, struct queue *q,
 	if (retval)
 		goto out_deallocate_sdma_queue;
 
+	/* Do init_mqd before dqm_lock(dqm) to avoid circular locking order:
+	 * lock(dqm) -> bo::reserve
+	 */
 	mqd_mgr = dqm->ops.get_mqd_manager(dqm,
 			get_mqd_type_from_queue_type(q->properties.type));
 
@@ -1188,6 +1187,7 @@ static int create_queue_cpsch(struct device_queue_manager *dqm, struct queue *q,
 		retval = -ENOMEM;
 		goto out_deallocate_doorbell;
 	}
+
 	/*
 	 * Eviction state logic: we only mark active queues as evicted
 	 * to avoid the overhead of restoring inactive queues later
@@ -1196,9 +1196,7 @@ static int create_queue_cpsch(struct device_queue_manager *dqm, struct queue *q,
 		q->properties.is_evicted = (q->properties.queue_size > 0 &&
 					    q->properties.queue_percent > 0 &&
 					    q->properties.queue_address != 0);
-
 	dqm->asic_ops.init_sdma_vm(dqm, q, qpd);
-
 	q->properties.tba_addr = qpd->tba_addr;
 	q->properties.tma_addr = qpd->tma_addr;
 	retval = mqd_mgr->init_mqd(mqd_mgr, &q->mqd, &q->mqd_mem_obj,
@@ -1206,6 +1204,8 @@ static int create_queue_cpsch(struct device_queue_manager *dqm, struct queue *q,
 	if (retval)
 		goto out_deallocate_doorbell;
 
+	dqm_lock(dqm);
+
 	list_add(&q->list, &qpd->queues_list);
 	qpd->queue_count++;
 	if (q->properties.is_active) {
@@ -1233,9 +1233,7 @@ static int create_queue_cpsch(struct device_queue_manager *dqm, struct queue *q,
 out_deallocate_sdma_queue:
 	if (q->properties.type == KFD_QUEUE_TYPE_SDMA)
 		deallocate_sdma_queue(dqm, q->sdma_id);
-out_unlock:
-	dqm_unlock(dqm);
-
+out:
 	return retval;
 }
 
@@ -1398,8 +1396,6 @@ static int destroy_queue_cpsch(struct device_queue_manager *dqm,
 			qpd->reset_wavefronts = true;
 	}
 
-	mqd_mgr->uninit_mqd(mqd_mgr, q->mqd, q->mqd_mem_obj);
-
 	/*
 	 * Unconditionally decrement this counter, regardless of the queue's
 	 * type
@@ -1410,6 +1406,9 @@ static int destroy_queue_cpsch(struct device_queue_manager *dqm,
 
 	dqm_unlock(dqm);
 
+	/* Do uninit_mqd after dqm_unlock(dqm) to avoid circular locking */
+	mqd_mgr->uninit_mqd(mqd_mgr, q->mqd, q->mqd_mem_obj);
+
 	return retval;
 
 failed:
@@ -1631,7 +1630,11 @@ static int process_termination_cpsch(struct device_queue_manager *dqm,
 		qpd->reset_wavefronts = false;
 	}
 
-	/* lastly, free mqd resources */
+	dqm_unlock(dqm);
+
+	/* Lastly, free mqd resources.
+	 * Do uninit_mqd() after dqm_unlock to avoid circular locking.
+	 */
 	list_for_each_entry_safe(q, next, &qpd->queues_list, list) {
 		mqd_mgr = dqm->ops.get_mqd_manager(dqm,
 			get_mqd_type_from_queue_type(q->properties.type));
@@ -1645,7 +1648,6 @@ static int process_termination_cpsch(struct device_queue_manager *dqm,
 	}
 
 out:
-	dqm_unlock(dqm);
 	return retval;
 }
 
-- 
2.17.1

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

Re: [PATCH 2/3] drm/amdkfd: avoid HMM change cause circular lock dependency v2

[Christian König]

Am 04.02.19 um 19:23 schrieb Yang, Philip:
> There is circular lock between gfx and kfd path with HMM change:
> lock(dqm) -> bo::reserve -> amdgpu_mn_lock
>
> To avoid this, move init/unint_mqd() out of lock(dqm), to remove nested
> locking between mmap_sem and bo::reserve. The locking order
> is: bo::reserve -> amdgpu_mn_lock(p->mn)

In general this sounds correct to me, but apart from that I don't know 
the code well enough to fully judge.

>
> Change-Id: I2ec09a47571f6b4c8eaef93f22c0a600f5f70153
> Signed-off-by: Philip Yang <Philip.Yang@amd.com>

Acked-by: Christian König <christian.koenig@amd.com>

> ---
>   .../drm/amd/amdkfd/kfd_device_queue_manager.c | 32 ++++++++++---------
>   1 file changed, 17 insertions(+), 15 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
> index 8372556b52eb..efe0d3c0215b 100644
> --- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
> +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
> @@ -1156,21 +1156,17 @@ static int create_queue_cpsch(struct device_queue_manager *dqm, struct queue *q,
>   	int retval;
>   	struct mqd_manager *mqd_mgr;
>   
> -	retval = 0;
> -
> -	dqm_lock(dqm);
> -
>   	if (dqm->total_queue_count >= max_num_of_queues_per_device) {
>   		pr_warn("Can't create new usermode queue because %d queues were already created\n",
>   				dqm->total_queue_count);
>   		retval = -EPERM;
> -		goto out_unlock;
> +		goto out;
>   	}
>   
>   	if (q->properties.type == KFD_QUEUE_TYPE_SDMA) {
>   		retval = allocate_sdma_queue(dqm, &q->sdma_id);
>   		if (retval)
> -			goto out_unlock;
> +			goto out;
>   		q->properties.sdma_queue_id =
>   			q->sdma_id / get_num_sdma_engines(dqm);
>   		q->properties.sdma_engine_id =
> @@ -1181,6 +1177,9 @@ static int create_queue_cpsch(struct device_queue_manager *dqm, struct queue *q,
>   	if (retval)
>   		goto out_deallocate_sdma_queue;
>   
> +	/* Do init_mqd before dqm_lock(dqm) to avoid circular locking order:
> +	 * lock(dqm) -> bo::reserve
> +	 */
>   	mqd_mgr = dqm->ops.get_mqd_manager(dqm,
>   			get_mqd_type_from_queue_type(q->properties.type));
>   
> @@ -1188,6 +1187,7 @@ static int create_queue_cpsch(struct device_queue_manager *dqm, struct queue *q,
>   		retval = -ENOMEM;
>   		goto out_deallocate_doorbell;
>   	}
> +
>   	/*
>   	 * Eviction state logic: we only mark active queues as evicted
>   	 * to avoid the overhead of restoring inactive queues later
> @@ -1196,9 +1196,7 @@ static int create_queue_cpsch(struct device_queue_manager *dqm, struct queue *q,
>   		q->properties.is_evicted = (q->properties.queue_size > 0 &&
>   					    q->properties.queue_percent > 0 &&
>   					    q->properties.queue_address != 0);
> -
>   	dqm->asic_ops.init_sdma_vm(dqm, q, qpd);
> -
>   	q->properties.tba_addr = qpd->tba_addr;
>   	q->properties.tma_addr = qpd->tma_addr;
>   	retval = mqd_mgr->init_mqd(mqd_mgr, &q->mqd, &q->mqd_mem_obj,
> @@ -1206,6 +1204,8 @@ static int create_queue_cpsch(struct device_queue_manager *dqm, struct queue *q,
>   	if (retval)
>   		goto out_deallocate_doorbell;
>   
> +	dqm_lock(dqm);
> +
>   	list_add(&q->list, &qpd->queues_list);
>   	qpd->queue_count++;
>   	if (q->properties.is_active) {
> @@ -1233,9 +1233,7 @@ static int create_queue_cpsch(struct device_queue_manager *dqm, struct queue *q,
>   out_deallocate_sdma_queue:
>   	if (q->properties.type == KFD_QUEUE_TYPE_SDMA)
>   		deallocate_sdma_queue(dqm, q->sdma_id);
> -out_unlock:
> -	dqm_unlock(dqm);
> -
> +out:
>   	return retval;
>   }
>   
> @@ -1398,8 +1396,6 @@ static int destroy_queue_cpsch(struct device_queue_manager *dqm,
>   			qpd->reset_wavefronts = true;
>   	}
>   
> -	mqd_mgr->uninit_mqd(mqd_mgr, q->mqd, q->mqd_mem_obj);
> -
>   	/*
>   	 * Unconditionally decrement this counter, regardless of the queue's
>   	 * type
> @@ -1410,6 +1406,9 @@ static int destroy_queue_cpsch(struct device_queue_manager *dqm,
>   
>   	dqm_unlock(dqm);
>   
> +	/* Do uninit_mqd after dqm_unlock(dqm) to avoid circular locking */
> +	mqd_mgr->uninit_mqd(mqd_mgr, q->mqd, q->mqd_mem_obj);
> +
>   	return retval;
>   
>   failed:
> @@ -1631,7 +1630,11 @@ static int process_termination_cpsch(struct device_queue_manager *dqm,
>   		qpd->reset_wavefronts = false;
>   	}
>   
> -	/* lastly, free mqd resources */
> +	dqm_unlock(dqm);
> +
> +	/* Lastly, free mqd resources.
> +	 * Do uninit_mqd() after dqm_unlock to avoid circular locking.
> +	 */
>   	list_for_each_entry_safe(q, next, &qpd->queues_list, list) {
>   		mqd_mgr = dqm->ops.get_mqd_manager(dqm,
>   			get_mqd_type_from_queue_type(q->properties.type));
> @@ -1645,7 +1648,6 @@ static int process_termination_cpsch(struct device_queue_manager *dqm,
>   	}
>   
>   out:
> -	dqm_unlock(dqm);
>   	return retval;
>   }
>   

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

[PATCH 0/3] Use HMM to replace get_user_pages

[Yang, Philip]

Hi Christian,

Resend patch 1/3, 2/3, added Reviewed-by in comments.

Change in patch 3/3, amdgpu_cs_submit, amdgpu_cs_ioctl return -EAGAIN
to user space to retry cs_ioctl.

Regards,
Philip

Philip Yang (3):
  drm/amdgpu: use HMM mirror callback to replace mmu notifier v7
  drm/amdkfd: avoid HMM change cause circular lock dependency v2
  drm/amdgpu: replace get_user_pages with HMM address mirror helpers v8

 drivers/gpu/drm/amd/amdgpu/Kconfig            |   6 +-
 drivers/gpu/drm/amd/amdgpu/Makefile           |   2 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h    |   1 -
 .../gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c  |  95 +++-------
 drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.h   |   2 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c        | 138 +++++---------
 drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c       |  14 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c        | 179 +++++++++---------
 drivers/gpu/drm/amd/amdgpu/amdgpu_mn.h        |   6 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c       | 178 +++++++----------
 drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h       |   3 +-
 .../drm/amd/amdkfd/kfd_device_queue_manager.c |  32 ++--
 12 files changed, 269 insertions(+), 387 deletions(-)

-- 
2.17.1

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

[PATCH 2/3] drm/amdkfd: avoid HMM change cause circular lock dependency v2

[Yang, Philip]

There is circular lock between gfx and kfd path with HMM change:
lock(dqm) -> bo::reserve -> amdgpu_mn_lock

To avoid this, move init/unint_mqd() out of lock(dqm), to remove nested
locking between mmap_sem and bo::reserve. The locking order
is: bo::reserve -> amdgpu_mn_lock(p->mn)

Change-Id: I2ec09a47571f6b4c8eaef93f22c0a600f5f70153
Signed-off-by: Philip Yang <Philip.Yang@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
---
 .../drm/amd/amdkfd/kfd_device_queue_manager.c | 32 ++++++++++---------
 1 file changed, 17 insertions(+), 15 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
index 8372556b52eb..efe0d3c0215b 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
@@ -1156,21 +1156,17 @@ static int create_queue_cpsch(struct device_queue_manager *dqm, struct queue *q,
 	int retval;
 	struct mqd_manager *mqd_mgr;
 
-	retval = 0;
-
-	dqm_lock(dqm);
-
 	if (dqm->total_queue_count >= max_num_of_queues_per_device) {
 		pr_warn("Can't create new usermode queue because %d queues were already created\n",
 				dqm->total_queue_count);
 		retval = -EPERM;
-		goto out_unlock;
+		goto out;
 	}
 
 	if (q->properties.type == KFD_QUEUE_TYPE_SDMA) {
 		retval = allocate_sdma_queue(dqm, &q->sdma_id);
 		if (retval)
-			goto out_unlock;
+			goto out;
 		q->properties.sdma_queue_id =
 			q->sdma_id / get_num_sdma_engines(dqm);
 		q->properties.sdma_engine_id =
@@ -1181,6 +1177,9 @@ static int create_queue_cpsch(struct device_queue_manager *dqm, struct queue *q,
 	if (retval)
 		goto out_deallocate_sdma_queue;
 
+	/* Do init_mqd before dqm_lock(dqm) to avoid circular locking order:
+	 * lock(dqm) -> bo::reserve
+	 */
 	mqd_mgr = dqm->ops.get_mqd_manager(dqm,
 			get_mqd_type_from_queue_type(q->properties.type));
 
@@ -1188,6 +1187,7 @@ static int create_queue_cpsch(struct device_queue_manager *dqm, struct queue *q,
 		retval = -ENOMEM;
 		goto out_deallocate_doorbell;
 	}
+
 	/*
 	 * Eviction state logic: we only mark active queues as evicted
 	 * to avoid the overhead of restoring inactive queues later
@@ -1196,9 +1196,7 @@ static int create_queue_cpsch(struct device_queue_manager *dqm, struct queue *q,
 		q->properties.is_evicted = (q->properties.queue_size > 0 &&
 					    q->properties.queue_percent > 0 &&
 					    q->properties.queue_address != 0);
-
 	dqm->asic_ops.init_sdma_vm(dqm, q, qpd);
-
 	q->properties.tba_addr = qpd->tba_addr;
 	q->properties.tma_addr = qpd->tma_addr;
 	retval = mqd_mgr->init_mqd(mqd_mgr, &q->mqd, &q->mqd_mem_obj,
@@ -1206,6 +1204,8 @@ static int create_queue_cpsch(struct device_queue_manager *dqm, struct queue *q,
 	if (retval)
 		goto out_deallocate_doorbell;
 
+	dqm_lock(dqm);
+
 	list_add(&q->list, &qpd->queues_list);
 	qpd->queue_count++;
 	if (q->properties.is_active) {
@@ -1233,9 +1233,7 @@ static int create_queue_cpsch(struct device_queue_manager *dqm, struct queue *q,
 out_deallocate_sdma_queue:
 	if (q->properties.type == KFD_QUEUE_TYPE_SDMA)
 		deallocate_sdma_queue(dqm, q->sdma_id);
-out_unlock:
-	dqm_unlock(dqm);
-
+out:
 	return retval;
 }
 
@@ -1398,8 +1396,6 @@ static int destroy_queue_cpsch(struct device_queue_manager *dqm,
 			qpd->reset_wavefronts = true;
 	}
 
-	mqd_mgr->uninit_mqd(mqd_mgr, q->mqd, q->mqd_mem_obj);
-
 	/*
 	 * Unconditionally decrement this counter, regardless of the queue's
 	 * type
@@ -1410,6 +1406,9 @@ static int destroy_queue_cpsch(struct device_queue_manager *dqm,
 
 	dqm_unlock(dqm);
 
+	/* Do uninit_mqd after dqm_unlock(dqm) to avoid circular locking */
+	mqd_mgr->uninit_mqd(mqd_mgr, q->mqd, q->mqd_mem_obj);
+
 	return retval;
 
 failed:
@@ -1631,7 +1630,11 @@ static int process_termination_cpsch(struct device_queue_manager *dqm,
 		qpd->reset_wavefronts = false;
 	}
 
-	/* lastly, free mqd resources */
+	dqm_unlock(dqm);
+
+	/* Lastly, free mqd resources.
+	 * Do uninit_mqd() after dqm_unlock to avoid circular locking.
+	 */
 	list_for_each_entry_safe(q, next, &qpd->queues_list, list) {
 		mqd_mgr = dqm->ops.get_mqd_manager(dqm,
 			get_mqd_type_from_queue_type(q->properties.type));
@@ -1645,7 +1648,6 @@ static int process_termination_cpsch(struct device_queue_manager *dqm,
 	}
 
 out:
-	dqm_unlock(dqm);
 	return retval;
 }
 
-- 
2.17.1

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx