Bunch of fixes to enable passing hotplug tests i previosly added here[1] with latest code. Once accepted I will enable the tests on libdrm side.
[1] - https://gitlab.freedesktop.org/mesa/drm/-/merge_requests/172
v2: Dropping VCE patch since relevant function already fixed in latest code. Moving IOMMU hnadling to TTM layer.
v3: Move pinned list to ttm device and a few others.
Andrey Grodzovsky (4): drm/ttm: Create pinned list drm/ttm: Clear all DMA mappings on demand drm/amdgpu: drm/amdgpu: Handle IOMMU enabled case drm/amdgpu: Add a UAPI flag for hot plug/unplug
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 2 + drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c | 3 +- drivers/gpu/drm/ttm/ttm_bo.c | 18 ++++++-- drivers/gpu/drm/ttm/ttm_device.c | 48 ++++++++++++++++++++++ include/drm/ttm/ttm_device.h | 2 + 5 files changed, 68 insertions(+), 5 deletions(-)
This list will be used to capture all non VRAM BOs not on LRU so when device is hot unplugged we can iterate the list and unmap DMA mappings before device is removed.
v2: Reanme function to ttm_bo_move_to_pinned v3: Move the pinned list to ttm device
Signed-off-by: Andrey Grodzovsky andrey.grodzovsky@amd.com Suggested-by: Christian König christian.koenig@amd.com --- drivers/gpu/drm/ttm/ttm_bo.c | 18 ++++++++++++++---- drivers/gpu/drm/ttm/ttm_device.c | 1 + include/drm/ttm/ttm_device.h | 1 + 3 files changed, 16 insertions(+), 4 deletions(-)
diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c index 1b950b45cf4b..1fedd0eb67ba 100644 --- a/drivers/gpu/drm/ttm/ttm_bo.c +++ b/drivers/gpu/drm/ttm/ttm_bo.c @@ -69,7 +69,17 @@ static void ttm_bo_mem_space_debug(struct ttm_buffer_object *bo, } }
-static void ttm_bo_del_from_lru(struct ttm_buffer_object *bo) +static inline void ttm_bo_move_to_pinned(struct ttm_buffer_object *bo) +{ + struct ttm_device *bdev = bo->bdev; + + list_move_tail(&bo->lru, &bdev->pinned); + + if (bdev->funcs->del_from_lru_notify) + bdev->funcs->del_from_lru_notify(bo); +} + +static inline void ttm_bo_del_from_lru(struct ttm_buffer_object *bo) { struct ttm_device *bdev = bo->bdev;
@@ -98,7 +108,7 @@ void ttm_bo_move_to_lru_tail(struct ttm_buffer_object *bo, dma_resv_assert_held(bo->base.resv);
if (bo->pin_count) { - ttm_bo_del_from_lru(bo); + ttm_bo_move_to_pinned(bo); return; }
@@ -339,7 +349,7 @@ static int ttm_bo_cleanup_refs(struct ttm_buffer_object *bo, return ret; }
- ttm_bo_del_from_lru(bo); + ttm_bo_move_to_pinned(bo); list_del_init(&bo->ddestroy); spin_unlock(&bo->bdev->lru_lock); ttm_bo_cleanup_memtype_use(bo); @@ -1154,7 +1164,7 @@ int ttm_bo_swapout(struct ttm_buffer_object *bo, struct ttm_operation_ctx *ctx, return 0; }
- ttm_bo_del_from_lru(bo); + ttm_bo_move_to_pinned(bo); /* TODO: Cleanup the locking */ spin_unlock(&bo->bdev->lru_lock);
diff --git a/drivers/gpu/drm/ttm/ttm_device.c b/drivers/gpu/drm/ttm/ttm_device.c index 5f31acec3ad7..530a9c36be37 100644 --- a/drivers/gpu/drm/ttm/ttm_device.c +++ b/drivers/gpu/drm/ttm/ttm_device.c @@ -208,6 +208,7 @@ int ttm_device_init(struct ttm_device *bdev, struct ttm_device_funcs *funcs, INIT_DELAYED_WORK(&bdev->wq, ttm_device_delayed_workqueue); spin_lock_init(&bdev->lru_lock); INIT_LIST_HEAD(&bdev->ddestroy); + INIT_LIST_HEAD(&bdev->pinned); bdev->dev_mapping = mapping; mutex_lock(&ttm_global_mutex); list_add_tail(&bdev->device_list, &glob->device_list); diff --git a/include/drm/ttm/ttm_device.h b/include/drm/ttm/ttm_device.h index cd592f8e941b..03fb44d061e0 100644 --- a/include/drm/ttm/ttm_device.h +++ b/include/drm/ttm/ttm_device.h @@ -265,6 +265,7 @@ struct ttm_device { */ spinlock_t lru_lock; struct list_head ddestroy; + struct list_head pinned;
/* * Protected by load / firstopen / lastclose /unload sync.
Am 27.08.21 um 22:39 schrieb Andrey Grodzovsky:
This list will be used to capture all non VRAM BOs not on LRU so when device is hot unplugged we can iterate the list and unmap DMA mappings before device is removed.
v2: Reanme function to ttm_bo_move_to_pinned v3: Move the pinned list to ttm device
As far as I can see there is not list_del() remaining. So this won't work correctly.
I suggest to rather rebase on top of the stuff I'm working on for a while to move the LRU into the resource instead.
Just send out the latest patch set of this with you in CC.
Christian.
Signed-off-by: Andrey Grodzovsky andrey.grodzovsky@amd.com Suggested-by: Christian König christian.koenig@amd.com
drivers/gpu/drm/ttm/ttm_bo.c | 18 ++++++++++++++---- drivers/gpu/drm/ttm/ttm_device.c | 1 + include/drm/ttm/ttm_device.h | 1 + 3 files changed, 16 insertions(+), 4 deletions(-)
diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c index 1b950b45cf4b..1fedd0eb67ba 100644 --- a/drivers/gpu/drm/ttm/ttm_bo.c +++ b/drivers/gpu/drm/ttm/ttm_bo.c @@ -69,7 +69,17 @@ static void ttm_bo_mem_space_debug(struct ttm_buffer_object *bo, } }
-static void ttm_bo_del_from_lru(struct ttm_buffer_object *bo) +static inline void ttm_bo_move_to_pinned(struct ttm_buffer_object *bo) +{
- struct ttm_device *bdev = bo->bdev;
- list_move_tail(&bo->lru, &bdev->pinned);
- if (bdev->funcs->del_from_lru_notify)
bdev->funcs->del_from_lru_notify(bo);+}
+static inline void ttm_bo_del_from_lru(struct ttm_buffer_object *bo) { struct ttm_device *bdev = bo->bdev;
@@ -98,7 +108,7 @@ void ttm_bo_move_to_lru_tail(struct ttm_buffer_object *bo, dma_resv_assert_held(bo->base.resv);
if (bo->pin_count) {
ttm_bo_del_from_lru(bo);
ttm_bo_move_to_pinned(bo);@@ -339,7 +349,7 @@ static int ttm_bo_cleanup_refs(struct ttm_buffer_object *bo, return ret; }
- ttm_bo_del_from_lru(bo);
- ttm_bo_move_to_pinned(bo); list_del_init(&bo->ddestroy); spin_unlock(&bo->bdev->lru_lock); ttm_bo_cleanup_memtype_use(bo);
@@ -1154,7 +1164,7 @@ int ttm_bo_swapout(struct ttm_buffer_object *bo, struct ttm_operation_ctx *ctx, return 0; }
- ttm_bo_del_from_lru(bo);
- ttm_bo_move_to_pinned(bo); /* TODO: Cleanup the locking */ spin_unlock(&bo->bdev->lru_lock);
diff --git a/drivers/gpu/drm/ttm/ttm_device.c b/drivers/gpu/drm/ttm/ttm_device.c index 5f31acec3ad7..530a9c36be37 100644 --- a/drivers/gpu/drm/ttm/ttm_device.c +++ b/drivers/gpu/drm/ttm/ttm_device.c @@ -208,6 +208,7 @@ int ttm_device_init(struct ttm_device *bdev, struct ttm_device_funcs *funcs, INIT_DELAYED_WORK(&bdev->wq, ttm_device_delayed_workqueue); spin_lock_init(&bdev->lru_lock); INIT_LIST_HEAD(&bdev->ddestroy);
- INIT_LIST_HEAD(&bdev->pinned); bdev->dev_mapping = mapping; mutex_lock(&ttm_global_mutex); list_add_tail(&bdev->device_list, &glob->device_list);
diff --git a/include/drm/ttm/ttm_device.h b/include/drm/ttm/ttm_device.h index cd592f8e941b..03fb44d061e0 100644 --- a/include/drm/ttm/ttm_device.h +++ b/include/drm/ttm/ttm_device.h @@ -265,6 +265,7 @@ struct ttm_device { */ spinlock_t lru_lock; struct list_head ddestroy;
struct list_head pinned;
/*
- Protected by load / firstopen / lastclose /unload sync.
On 2021-08-30 4:58 a.m., Christian König wrote:
Am 27.08.21 um 22:39 schrieb Andrey Grodzovsky:
This list will be used to capture all non VRAM BOs not on LRU so when device is hot unplugged we can iterate the list and unmap DMA mappings before device is removed.
v2: Reanme function to ttm_bo_move_to_pinned v3: Move the pinned list to ttm device
As far as I can see there is not list_del() remaining. So this won't work correctly.
It's in ttm_bo_release, there was no code change there hence it's not captured in the patch.
Andrey
I suggest to rather rebase on top of the stuff I'm working on for a while to move the LRU into the resource instead.
Just send out the latest patch set of this with you in CC.
Christian.
Signed-off-by: Andrey Grodzovsky andrey.grodzovsky@amd.com Suggested-by: Christian König christian.koenig@amd.com
drivers/gpu/drm/ttm/ttm_bo.c | 18 ++++++++++++++---- drivers/gpu/drm/ttm/ttm_device.c | 1 + include/drm/ttm/ttm_device.h | 1 + 3 files changed, 16 insertions(+), 4 deletions(-)
diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c index 1b950b45cf4b..1fedd0eb67ba 100644 --- a/drivers/gpu/drm/ttm/ttm_bo.c +++ b/drivers/gpu/drm/ttm/ttm_bo.c @@ -69,7 +69,17 @@ static void ttm_bo_mem_space_debug(struct ttm_buffer_object *bo, } } -static void ttm_bo_del_from_lru(struct ttm_buffer_object *bo) +static inline void ttm_bo_move_to_pinned(struct ttm_buffer_object *bo) +{ + struct ttm_device *bdev = bo->bdev;
+ list_move_tail(&bo->lru, &bdev->pinned);
+ if (bdev->funcs->del_from_lru_notify) + bdev->funcs->del_from_lru_notify(bo); +}
+static inline void ttm_bo_del_from_lru(struct ttm_buffer_object *bo) { struct ttm_device *bdev = bo->bdev; @@ -98,7 +108,7 @@ void ttm_bo_move_to_lru_tail(struct ttm_buffer_object *bo, dma_resv_assert_held(bo->base.resv); if (bo->pin_count) { - ttm_bo_del_from_lru(bo); + ttm_bo_move_to_pinned(bo); return; } @@ -339,7 +349,7 @@ static int ttm_bo_cleanup_refs(struct ttm_buffer_object *bo, return ret; } - ttm_bo_del_from_lru(bo); + ttm_bo_move_to_pinned(bo); list_del_init(&bo->ddestroy); spin_unlock(&bo->bdev->lru_lock); ttm_bo_cleanup_memtype_use(bo); @@ -1154,7 +1164,7 @@ int ttm_bo_swapout(struct ttm_buffer_object *bo, struct ttm_operation_ctx *ctx, return 0; } - ttm_bo_del_from_lru(bo); + ttm_bo_move_to_pinned(bo); /* TODO: Cleanup the locking */ spin_unlock(&bo->bdev->lru_lock); diff --git a/drivers/gpu/drm/ttm/ttm_device.c b/drivers/gpu/drm/ttm/ttm_device.c index 5f31acec3ad7..530a9c36be37 100644 --- a/drivers/gpu/drm/ttm/ttm_device.c +++ b/drivers/gpu/drm/ttm/ttm_device.c @@ -208,6 +208,7 @@ int ttm_device_init(struct ttm_device *bdev, struct ttm_device_funcs *funcs, INIT_DELAYED_WORK(&bdev->wq, ttm_device_delayed_workqueue); spin_lock_init(&bdev->lru_lock); INIT_LIST_HEAD(&bdev->ddestroy); + INIT_LIST_HEAD(&bdev->pinned); bdev->dev_mapping = mapping; mutex_lock(&ttm_global_mutex); list_add_tail(&bdev->device_list, &glob->device_list); diff --git a/include/drm/ttm/ttm_device.h b/include/drm/ttm/ttm_device.h index cd592f8e941b..03fb44d061e0 100644 --- a/include/drm/ttm/ttm_device.h +++ b/include/drm/ttm/ttm_device.h @@ -265,6 +265,7 @@ struct ttm_device { */ spinlock_t lru_lock; struct list_head ddestroy; + struct list_head pinned; /* * Protected by load / firstopen / lastclose /unload sync.
Am 30.08.21 um 16:16 schrieb Andrey Grodzovsky:
On 2021-08-30 4:58 a.m., Christian König wrote:
Am 27.08.21 um 22:39 schrieb Andrey Grodzovsky:
This list will be used to capture all non VRAM BOs not on LRU so when device is hot unplugged we can iterate the list and unmap DMA mappings before device is removed.
v2: Reanme function to ttm_bo_move_to_pinned v3: Move the pinned list to ttm device
As far as I can see there is not list_del() remaining. So this won't work correctly.
It's in ttm_bo_release, there was no code change there hence it's not captured in the patch.
Ah! So you keep the logic as is there. Sorry totally missed that.
In this case the patch is Reviewed-by: Christian König christian.koenig@amd.com
Can you push this to drm-misc-next?
Thanks, Christian.
Andrey
I suggest to rather rebase on top of the stuff I'm working on for a while to move the LRU into the resource instead.
Just send out the latest patch set of this with you in CC.
Christian.
Signed-off-by: Andrey Grodzovsky andrey.grodzovsky@amd.com Suggested-by: Christian König christian.koenig@amd.com
drivers/gpu/drm/ttm/ttm_bo.c | 18 ++++++++++++++---- drivers/gpu/drm/ttm/ttm_device.c | 1 + include/drm/ttm/ttm_device.h | 1 + 3 files changed, 16 insertions(+), 4 deletions(-)
diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c index 1b950b45cf4b..1fedd0eb67ba 100644 --- a/drivers/gpu/drm/ttm/ttm_bo.c +++ b/drivers/gpu/drm/ttm/ttm_bo.c @@ -69,7 +69,17 @@ static void ttm_bo_mem_space_debug(struct ttm_buffer_object *bo, } } -static void ttm_bo_del_from_lru(struct ttm_buffer_object *bo) +static inline void ttm_bo_move_to_pinned(struct ttm_buffer_object *bo) +{ + struct ttm_device *bdev = bo->bdev;
+ list_move_tail(&bo->lru, &bdev->pinned);
+ if (bdev->funcs->del_from_lru_notify) + bdev->funcs->del_from_lru_notify(bo); +}
+static inline void ttm_bo_del_from_lru(struct ttm_buffer_object *bo) { struct ttm_device *bdev = bo->bdev; @@ -98,7 +108,7 @@ void ttm_bo_move_to_lru_tail(struct ttm_buffer_object *bo, dma_resv_assert_held(bo->base.resv); if (bo->pin_count) { - ttm_bo_del_from_lru(bo); + ttm_bo_move_to_pinned(bo); return; } @@ -339,7 +349,7 @@ static int ttm_bo_cleanup_refs(struct ttm_buffer_object *bo, return ret; } - ttm_bo_del_from_lru(bo); + ttm_bo_move_to_pinned(bo); list_del_init(&bo->ddestroy); spin_unlock(&bo->bdev->lru_lock); ttm_bo_cleanup_memtype_use(bo); @@ -1154,7 +1164,7 @@ int ttm_bo_swapout(struct ttm_buffer_object *bo, struct ttm_operation_ctx *ctx, return 0; } - ttm_bo_del_from_lru(bo); + ttm_bo_move_to_pinned(bo); /* TODO: Cleanup the locking */ spin_unlock(&bo->bdev->lru_lock); diff --git a/drivers/gpu/drm/ttm/ttm_device.c b/drivers/gpu/drm/ttm/ttm_device.c index 5f31acec3ad7..530a9c36be37 100644 --- a/drivers/gpu/drm/ttm/ttm_device.c +++ b/drivers/gpu/drm/ttm/ttm_device.c @@ -208,6 +208,7 @@ int ttm_device_init(struct ttm_device *bdev, struct ttm_device_funcs *funcs, INIT_DELAYED_WORK(&bdev->wq, ttm_device_delayed_workqueue); spin_lock_init(&bdev->lru_lock); INIT_LIST_HEAD(&bdev->ddestroy); + INIT_LIST_HEAD(&bdev->pinned); bdev->dev_mapping = mapping; mutex_lock(&ttm_global_mutex); list_add_tail(&bdev->device_list, &glob->device_list); diff --git a/include/drm/ttm/ttm_device.h b/include/drm/ttm/ttm_device.h index cd592f8e941b..03fb44d061e0 100644 --- a/include/drm/ttm/ttm_device.h +++ b/include/drm/ttm/ttm_device.h @@ -265,6 +265,7 @@ struct ttm_device { */ spinlock_t lru_lock; struct list_head ddestroy; + struct list_head pinned; /* * Protected by load / firstopen / lastclose /unload sync.
On 2021-08-30 12:51 p.m., Christian König wrote:
Am 30.08.21 um 16:16 schrieb Andrey Grodzovsky:
On 2021-08-30 4:58 a.m., Christian König wrote:
Am 27.08.21 um 22:39 schrieb Andrey Grodzovsky:
This list will be used to capture all non VRAM BOs not on LRU so when device is hot unplugged we can iterate the list and unmap DMA mappings before device is removed.
v2: Reanme function to ttm_bo_move_to_pinned v3: Move the pinned list to ttm device
As far as I can see there is not list_del() remaining. So this won't work correctly.
It's in ttm_bo_release, there was no code change there hence it's not captured in the patch.
Ah! So you keep the logic as is there. Sorry totally missed that.
In this case the patch is Reviewed-by: Christian König christian.koenig@amd.com
Can you push this to drm-misc-next?
Thanks, Christian.
I think It's supposed to go on top of your changes you mention here which are not pushed yet. I will need to apply all the patches on top of yours and retest (I was doing everything in amd-staging-drm-next) until now.
Andrey
Andrey
I suggest to rather rebase on top of the stuff I'm working on for a while to move the LRU into the resource instead.
Just send out the latest patch set of this with you in CC.
Christian.
Signed-off-by: Andrey Grodzovsky andrey.grodzovsky@amd.com Suggested-by: Christian König christian.koenig@amd.com
drivers/gpu/drm/ttm/ttm_bo.c | 18 ++++++++++++++---- drivers/gpu/drm/ttm/ttm_device.c | 1 + include/drm/ttm/ttm_device.h | 1 + 3 files changed, 16 insertions(+), 4 deletions(-)
diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c index 1b950b45cf4b..1fedd0eb67ba 100644 --- a/drivers/gpu/drm/ttm/ttm_bo.c +++ b/drivers/gpu/drm/ttm/ttm_bo.c @@ -69,7 +69,17 @@ static void ttm_bo_mem_space_debug(struct ttm_buffer_object *bo, } } -static void ttm_bo_del_from_lru(struct ttm_buffer_object *bo) +static inline void ttm_bo_move_to_pinned(struct ttm_buffer_object *bo) +{ + struct ttm_device *bdev = bo->bdev;
+ list_move_tail(&bo->lru, &bdev->pinned);
+ if (bdev->funcs->del_from_lru_notify) + bdev->funcs->del_from_lru_notify(bo); +}
+static inline void ttm_bo_del_from_lru(struct ttm_buffer_object *bo) { struct ttm_device *bdev = bo->bdev; @@ -98,7 +108,7 @@ void ttm_bo_move_to_lru_tail(struct ttm_buffer_object *bo, dma_resv_assert_held(bo->base.resv); if (bo->pin_count) { - ttm_bo_del_from_lru(bo); + ttm_bo_move_to_pinned(bo); return; } @@ -339,7 +349,7 @@ static int ttm_bo_cleanup_refs(struct ttm_buffer_object *bo, return ret; } - ttm_bo_del_from_lru(bo); + ttm_bo_move_to_pinned(bo); list_del_init(&bo->ddestroy); spin_unlock(&bo->bdev->lru_lock); ttm_bo_cleanup_memtype_use(bo); @@ -1154,7 +1164,7 @@ int ttm_bo_swapout(struct ttm_buffer_object *bo, struct ttm_operation_ctx *ctx, return 0; } - ttm_bo_del_from_lru(bo); + ttm_bo_move_to_pinned(bo); /* TODO: Cleanup the locking */ spin_unlock(&bo->bdev->lru_lock); diff --git a/drivers/gpu/drm/ttm/ttm_device.c b/drivers/gpu/drm/ttm/ttm_device.c index 5f31acec3ad7..530a9c36be37 100644 --- a/drivers/gpu/drm/ttm/ttm_device.c +++ b/drivers/gpu/drm/ttm/ttm_device.c @@ -208,6 +208,7 @@ int ttm_device_init(struct ttm_device *bdev, struct ttm_device_funcs *funcs, INIT_DELAYED_WORK(&bdev->wq, ttm_device_delayed_workqueue); spin_lock_init(&bdev->lru_lock); INIT_LIST_HEAD(&bdev->ddestroy); + INIT_LIST_HEAD(&bdev->pinned); bdev->dev_mapping = mapping; mutex_lock(&ttm_global_mutex); list_add_tail(&bdev->device_list, &glob->device_list); diff --git a/include/drm/ttm/ttm_device.h b/include/drm/ttm/ttm_device.h index cd592f8e941b..03fb44d061e0 100644 --- a/include/drm/ttm/ttm_device.h +++ b/include/drm/ttm/ttm_device.h @@ -265,6 +265,7 @@ struct ttm_device { */ spinlock_t lru_lock; struct list_head ddestroy; + struct list_head pinned; /* * Protected by load / firstopen / lastclose /unload sync.
Am 30.08.21 um 19:02 schrieb Andrey Grodzovsky:
On 2021-08-30 12:51 p.m., Christian König wrote:
Am 30.08.21 um 16:16 schrieb Andrey Grodzovsky:
On 2021-08-30 4:58 a.m., Christian König wrote:
Am 27.08.21 um 22:39 schrieb Andrey Grodzovsky:
This list will be used to capture all non VRAM BOs not on LRU so when device is hot unplugged we can iterate the list and unmap DMA mappings before device is removed.
v2: Reanme function to ttm_bo_move_to_pinned v3: Move the pinned list to ttm device
As far as I can see there is not list_del() remaining. So this won't work correctly.
It's in ttm_bo_release, there was no code change there hence it's not captured in the patch.
Ah! So you keep the logic as is there. Sorry totally missed that.
In this case the patch is Reviewed-by: Christian König christian.koenig@amd.com
Can you push this to drm-misc-next?
Thanks, Christian.
I think It's supposed to go on top of your changes you mention here which are not pushed yet. I will need to apply all the patches on top of yours and retest (I was doing everything in amd-staging-drm-next) until now.
Works for me as well. Alternatively you can just push this patch here to drm-misc-next so that I can rebase on top and merge the rest through amd-staging-drm-next.
The patch pushed to drm-misc-next should automatically fall out when Alex rebases his stuff on upstream the next time.
Christian.
Andrey
Andrey
I suggest to rather rebase on top of the stuff I'm working on for a while to move the LRU into the resource instead.
Just send out the latest patch set of this with you in CC.
Christian.
Signed-off-by: Andrey Grodzovsky andrey.grodzovsky@amd.com Suggested-by: Christian König christian.koenig@amd.com
drivers/gpu/drm/ttm/ttm_bo.c | 18 ++++++++++++++---- drivers/gpu/drm/ttm/ttm_device.c | 1 + include/drm/ttm/ttm_device.h | 1 + 3 files changed, 16 insertions(+), 4 deletions(-)
diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c index 1b950b45cf4b..1fedd0eb67ba 100644 --- a/drivers/gpu/drm/ttm/ttm_bo.c +++ b/drivers/gpu/drm/ttm/ttm_bo.c @@ -69,7 +69,17 @@ static void ttm_bo_mem_space_debug(struct ttm_buffer_object *bo, } } -static void ttm_bo_del_from_lru(struct ttm_buffer_object *bo) +static inline void ttm_bo_move_to_pinned(struct ttm_buffer_object *bo) +{ + struct ttm_device *bdev = bo->bdev;
+ list_move_tail(&bo->lru, &bdev->pinned);
+ if (bdev->funcs->del_from_lru_notify) + bdev->funcs->del_from_lru_notify(bo); +}
+static inline void ttm_bo_del_from_lru(struct ttm_buffer_object *bo) { struct ttm_device *bdev = bo->bdev; @@ -98,7 +108,7 @@ void ttm_bo_move_to_lru_tail(struct ttm_buffer_object *bo, dma_resv_assert_held(bo->base.resv); if (bo->pin_count) { - ttm_bo_del_from_lru(bo); + ttm_bo_move_to_pinned(bo); return; } @@ -339,7 +349,7 @@ static int ttm_bo_cleanup_refs(struct ttm_buffer_object *bo, return ret; } - ttm_bo_del_from_lru(bo); + ttm_bo_move_to_pinned(bo); list_del_init(&bo->ddestroy); spin_unlock(&bo->bdev->lru_lock); ttm_bo_cleanup_memtype_use(bo); @@ -1154,7 +1164,7 @@ int ttm_bo_swapout(struct ttm_buffer_object *bo, struct ttm_operation_ctx *ctx, return 0; } - ttm_bo_del_from_lru(bo); + ttm_bo_move_to_pinned(bo); /* TODO: Cleanup the locking */ spin_unlock(&bo->bdev->lru_lock); diff --git a/drivers/gpu/drm/ttm/ttm_device.c b/drivers/gpu/drm/ttm/ttm_device.c index 5f31acec3ad7..530a9c36be37 100644 --- a/drivers/gpu/drm/ttm/ttm_device.c +++ b/drivers/gpu/drm/ttm/ttm_device.c @@ -208,6 +208,7 @@ int ttm_device_init(struct ttm_device *bdev, struct ttm_device_funcs *funcs, INIT_DELAYED_WORK(&bdev->wq, ttm_device_delayed_workqueue); spin_lock_init(&bdev->lru_lock); INIT_LIST_HEAD(&bdev->ddestroy); + INIT_LIST_HEAD(&bdev->pinned); bdev->dev_mapping = mapping; mutex_lock(&ttm_global_mutex); list_add_tail(&bdev->device_list, &glob->device_list); diff --git a/include/drm/ttm/ttm_device.h b/include/drm/ttm/ttm_device.h index cd592f8e941b..03fb44d061e0 100644 --- a/include/drm/ttm/ttm_device.h +++ b/include/drm/ttm/ttm_device.h @@ -265,6 +265,7 @@ struct ttm_device { */ spinlock_t lru_lock; struct list_head ddestroy; + struct list_head pinned; /* * Protected by load / firstopen / lastclose /unload sync.
On 2021-08-30 1:05 p.m., Christian König wrote:
Am 30.08.21 um 19:02 schrieb Andrey Grodzovsky:
On 2021-08-30 12:51 p.m., Christian König wrote:
Am 30.08.21 um 16:16 schrieb Andrey Grodzovsky:
On 2021-08-30 4:58 a.m., Christian König wrote:
Am 27.08.21 um 22:39 schrieb Andrey Grodzovsky:
This list will be used to capture all non VRAM BOs not on LRU so when device is hot unplugged we can iterate the list and unmap DMA mappings before device is removed.
v2: Reanme function to ttm_bo_move_to_pinned v3: Move the pinned list to ttm device
As far as I can see there is not list_del() remaining. So this won't work correctly.
It's in ttm_bo_release, there was no code change there hence it's not captured in the patch.
Ah! So you keep the logic as is there. Sorry totally missed that.
In this case the patch is Reviewed-by: Christian König christian.koenig@amd.com
Can you push this to drm-misc-next?
Thanks, Christian.
I think It's supposed to go on top of your changes you mention here which are not pushed yet. I will need to apply all the patches on top of yours and retest (I was doing everything in amd-staging-drm-next) until now.
Works for me as well. Alternatively you can just push this patch here to drm-misc-next so that I can rebase on top and merge the rest through amd-staging-drm-next.
The patch pushed to drm-misc-next should automatically fall out when Alex rebases his stuff on upstream the next time.
Christian.
So i can both push this specific patch to drm-misc-next and also push the entire 4 patch series to amd-stagin-drm-next (after rest of the patches RBed)?
Andrey
Andrey
Andrey
I suggest to rather rebase on top of the stuff I'm working on for a while to move the LRU into the resource instead.
Just send out the latest patch set of this with you in CC.
Christian.
Signed-off-by: Andrey Grodzovsky andrey.grodzovsky@amd.com Suggested-by: Christian König christian.koenig@amd.com
drivers/gpu/drm/ttm/ttm_bo.c | 18 ++++++++++++++---- drivers/gpu/drm/ttm/ttm_device.c | 1 + include/drm/ttm/ttm_device.h | 1 + 3 files changed, 16 insertions(+), 4 deletions(-)
diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c index 1b950b45cf4b..1fedd0eb67ba 100644 --- a/drivers/gpu/drm/ttm/ttm_bo.c +++ b/drivers/gpu/drm/ttm/ttm_bo.c @@ -69,7 +69,17 @@ static void ttm_bo_mem_space_debug(struct ttm_buffer_object *bo, } } -static void ttm_bo_del_from_lru(struct ttm_buffer_object *bo) +static inline void ttm_bo_move_to_pinned(struct ttm_buffer_object *bo) +{ + struct ttm_device *bdev = bo->bdev;
+ list_move_tail(&bo->lru, &bdev->pinned);
+ if (bdev->funcs->del_from_lru_notify) + bdev->funcs->del_from_lru_notify(bo); +}
+static inline void ttm_bo_del_from_lru(struct ttm_buffer_object *bo) { struct ttm_device *bdev = bo->bdev; @@ -98,7 +108,7 @@ void ttm_bo_move_to_lru_tail(struct ttm_buffer_object *bo, dma_resv_assert_held(bo->base.resv); if (bo->pin_count) { - ttm_bo_del_from_lru(bo); + ttm_bo_move_to_pinned(bo); return; } @@ -339,7 +349,7 @@ static int ttm_bo_cleanup_refs(struct ttm_buffer_object *bo, return ret; } - ttm_bo_del_from_lru(bo); + ttm_bo_move_to_pinned(bo); list_del_init(&bo->ddestroy); spin_unlock(&bo->bdev->lru_lock); ttm_bo_cleanup_memtype_use(bo); @@ -1154,7 +1164,7 @@ int ttm_bo_swapout(struct ttm_buffer_object *bo, struct ttm_operation_ctx *ctx, return 0; } - ttm_bo_del_from_lru(bo); + ttm_bo_move_to_pinned(bo); /* TODO: Cleanup the locking */ spin_unlock(&bo->bdev->lru_lock); diff --git a/drivers/gpu/drm/ttm/ttm_device.c b/drivers/gpu/drm/ttm/ttm_device.c index 5f31acec3ad7..530a9c36be37 100644 --- a/drivers/gpu/drm/ttm/ttm_device.c +++ b/drivers/gpu/drm/ttm/ttm_device.c @@ -208,6 +208,7 @@ int ttm_device_init(struct ttm_device *bdev, struct ttm_device_funcs *funcs, INIT_DELAYED_WORK(&bdev->wq, ttm_device_delayed_workqueue); spin_lock_init(&bdev->lru_lock); INIT_LIST_HEAD(&bdev->ddestroy); + INIT_LIST_HEAD(&bdev->pinned); bdev->dev_mapping = mapping; mutex_lock(&ttm_global_mutex); list_add_tail(&bdev->device_list, &glob->device_list); diff --git a/include/drm/ttm/ttm_device.h b/include/drm/ttm/ttm_device.h index cd592f8e941b..03fb44d061e0 100644 --- a/include/drm/ttm/ttm_device.h +++ b/include/drm/ttm/ttm_device.h @@ -265,6 +265,7 @@ struct ttm_device { */ spinlock_t lru_lock; struct list_head ddestroy; + struct list_head pinned; /* * Protected by load / firstopen / lastclose /unload sync.
Used by drivers supporting hot unplug to handle all DMA IOMMU group related dependencies before the group is removed during device removal and we try to access it after free when last device pointer from user space is dropped.
v3: Switch to ttm_bo_get_unless_zerom Iterate bdev for pinned list Switch to ttm_tt_unpopulate
Signed-off-by: Andrey Grodzovsky andrey.grodzovsky@amd.com --- drivers/gpu/drm/ttm/ttm_device.c | 47 ++++++++++++++++++++++++++++++++ include/drm/ttm/ttm_device.h | 1 + 2 files changed, 48 insertions(+)
diff --git a/drivers/gpu/drm/ttm/ttm_device.c b/drivers/gpu/drm/ttm/ttm_device.c index 530a9c36be37..a691c89f5b20 100644 --- a/drivers/gpu/drm/ttm/ttm_device.c +++ b/drivers/gpu/drm/ttm/ttm_device.c @@ -246,3 +246,50 @@ void ttm_device_fini(struct ttm_device *bdev) ttm_global_release(); } EXPORT_SYMBOL(ttm_device_fini); + +void ttm_device_clear_dma_mappings(struct ttm_device *bdev) +{ + struct ttm_resource_manager *man; + struct ttm_buffer_object *bo; + unsigned int i, j; + + spin_lock(&bdev->lru_lock); + while (!list_empty(&bdev->pinned)) { + bo = list_first_entry(&bdev->pinned, struct ttm_buffer_object, lru); + /* Take ref against racing releases once lru_lock is unlocked */ + if (ttm_bo_get_unless_zero(bo)) { + list_del_init(&bo->lru); + spin_unlock(&bdev->lru_lock); + + if (bo->ttm) + ttm_tt_unpopulate(bo->bdev, bo->ttm); + + ttm_bo_put(bo); + spin_lock(&bdev->lru_lock); + } + } + + for (i = TTM_PL_SYSTEM; i < TTM_NUM_MEM_TYPES; ++i) { + man = ttm_manager_type(bdev, i); + if (!man || !man->use_tt) + continue; + + for (j = 0; j < TTM_MAX_BO_PRIORITY; ++j) { + while (!list_empty(&man->lru[j])) { + bo = list_first_entry(&man->lru[j], struct ttm_buffer_object, lru); + if (ttm_bo_get_unless_zero(bo)) { + list_del_init(&bo->lru); + spin_unlock(&bdev->lru_lock); + + if (bo->ttm) + ttm_tt_unpopulate(bo->bdev, bo->ttm); + + ttm_bo_put(bo); + spin_lock(&bdev->lru_lock); + } + } + } + } + spin_unlock(&bdev->lru_lock); +} +EXPORT_SYMBOL(ttm_device_clear_dma_mappings); diff --git a/include/drm/ttm/ttm_device.h b/include/drm/ttm/ttm_device.h index 03fb44d061e0..07d722950d5b 100644 --- a/include/drm/ttm/ttm_device.h +++ b/include/drm/ttm/ttm_device.h @@ -299,5 +299,6 @@ int ttm_device_init(struct ttm_device *bdev, struct ttm_device_funcs *funcs, struct drm_vma_offset_manager *vma_manager, bool use_dma_alloc, bool use_dma32); void ttm_device_fini(struct ttm_device *bdev); +void ttm_device_clear_dma_mappings(struct ttm_device *bdev);
#endif
Handle all DMA IOMMU group related dependencies before the group is removed and we try to access it after free.
v2: Move the actul handling function to TTM
Signed-off-by: Andrey Grodzovsky andrey.grodzovsky@amd.com --- drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 2 ++ 1 file changed, 2 insertions(+)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index 0b5764aa98a4..653bd8fdaa33 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -3860,6 +3860,8 @@ void amdgpu_device_fini_hw(struct amdgpu_device *adev)
amdgpu_device_ip_fini_early(adev);
+ ttm_device_clear_dma_mappings(&adev->mman.bdev); + amdgpu_gart_dummy_page_fini(adev);
amdgpu_device_unmap_mmio(adev);
To support libdrm tests.
Signed-off-by: Andrey Grodzovsky andrey.grodzovsky@amd.com --- drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c index 6400259a7c4b..c2fdf67ff551 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c @@ -96,9 +96,10 @@ * - 3.40.0 - Add AMDGPU_IDS_FLAGS_TMZ * - 3.41.0 - Add video codec query * - 3.42.0 - Add 16bpc fixed point display support + * - 3.43.0 - Add device hot plug/unplug support */ #define KMS_DRIVER_MAJOR 3 -#define KMS_DRIVER_MINOR 42 +#define KMS_DRIVER_MINOR 43 #define KMS_DRIVER_PATCHLEVEL 0
int amdgpu_vram_limit;
dri-devel@lists.freedesktop.org
-
Andrey Grodzovsky -
Christian König -
Christian König