* [PATCH 2/2] drm/amdgpu: add per vm bo validation order
[not found] ` <20180326082955.11488-1-david1.zhou-5C7GfCeVMHo@public.gmane.org>
@ 2018-03-26 8:29 ` Chunming Zhou
[not found] ` <20180326082955.11488-2-david1.zhou-5C7GfCeVMHo@public.gmane.org>
0 siblings, 1 reply; 5+ messages in thread
From: Chunming Zhou @ 2018-03-26 8:29 UTC (permalink / raw)
To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW; +Cc: Chunming Zhou
Change-Id: I59f1da8a765efae8196c0b1599f47a8b8485c4d3
Signed-off-by: Chunming Zhou <david1.zhou@amd.com>
---
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 12 +++++++++++-
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h | 6 ++++++
2 files changed, 17 insertions(+), 1 deletion(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index 7db411a282ce..3543e394adac 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -205,12 +205,20 @@ int amdgpu_vm_validate_pt_bos(struct amdgpu_device *adev, struct amdgpu_vm *vm,
spin_lock(&vm->status_lock);
while (!list_empty(&vm->evicted)) {
- struct amdgpu_vm_bo_base *bo_base;
+ struct amdgpu_vm_bo_base *bo_base, *tmp;
struct amdgpu_bo *bo;
+ u64 bo_order;
bo_base = list_first_entry(&vm->evicted,
struct amdgpu_vm_bo_base,
vm_status);
+ bo_order = bo_base->order;
+ list_for_each_entry(tmp, &vm->evicted, vm_status) {
+ if (tmp->order < bo_order) {
+ bo_order = bo_base->order;
+ bo_base = tmp;
+ }
+ }
spin_unlock(&vm->status_lock);
bo = bo_base->bo;
@@ -1903,6 +1911,7 @@ struct amdgpu_bo_va *amdgpu_vm_bo_add(struct amdgpu_device *adev,
* */
spin_lock(&vm->status_lock);
list_move_tail(&bo_va->base.vm_status, &vm->evicted);
+ bo_va->base.order = (u64)atomic64_inc_return(&vm->bo_order);
spin_unlock(&vm->status_lock);
return bo_va;
@@ -2424,6 +2433,7 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
INIT_LIST_HEAD(&vm->relocated);
INIT_LIST_HEAD(&vm->moved);
INIT_LIST_HEAD(&vm->freed);
+ atomic64_set(&vm->bo_order, 0);
/* create scheduler entity for page table updates */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
index cf2c667ee538..bac5349d291e 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
@@ -152,6 +152,9 @@ struct amdgpu_vm_bo_base {
/* protected by the BO being reserved */
bool moved;
+
+ /* validate order */
+ u64 order;
};
struct amdgpu_vm_pt {
@@ -214,6 +217,9 @@ struct amdgpu_vm {
/* Limit non-retry fault storms */
unsigned int fault_credit;
+
+ /* per vm bo order in this vm */
+ atomic64_t bo_order;
};
struct amdgpu_vm_manager {
--
2.14.1
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 5+ messages in thread
* Re: [PATCH 2/2] drm/amdgpu: add per vm bo validation order
[not found] ` <20180326082955.11488-2-david1.zhou-5C7GfCeVMHo@public.gmane.org>
@ 2018-03-26 8:34 ` zhoucm1
0 siblings, 0 replies; 5+ messages in thread
From: zhoucm1 @ 2018-03-26 8:34 UTC (permalink / raw)
To: Chunming Zhou, amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW
pls ignore this one, will re-send it.
On 2018年03月26日 16:29, Chunming Zhou wrote:
> Change-Id: I59f1da8a765efae8196c0b1599f47a8b8485c4d3
> Signed-off-by: Chunming Zhou <david1.zhou@amd.com>
> ---
> drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 12 +++++++++++-
> drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h | 6 ++++++
> 2 files changed, 17 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
> index 7db411a282ce..3543e394adac 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
> @@ -205,12 +205,20 @@ int amdgpu_vm_validate_pt_bos(struct amdgpu_device *adev, struct amdgpu_vm *vm,
>
> spin_lock(&vm->status_lock);
> while (!list_empty(&vm->evicted)) {
> - struct amdgpu_vm_bo_base *bo_base;
> + struct amdgpu_vm_bo_base *bo_base, *tmp;
> struct amdgpu_bo *bo;
> + u64 bo_order;
>
> bo_base = list_first_entry(&vm->evicted,
> struct amdgpu_vm_bo_base,
> vm_status);
> + bo_order = bo_base->order;
> + list_for_each_entry(tmp, &vm->evicted, vm_status) {
> + if (tmp->order < bo_order) {
> + bo_order = bo_base->order;
> + bo_base = tmp;
> + }
> + }
> spin_unlock(&vm->status_lock);
>
> bo = bo_base->bo;
> @@ -1903,6 +1911,7 @@ struct amdgpu_bo_va *amdgpu_vm_bo_add(struct amdgpu_device *adev,
> * */
> spin_lock(&vm->status_lock);
> list_move_tail(&bo_va->base.vm_status, &vm->evicted);
> + bo_va->base.order = (u64)atomic64_inc_return(&vm->bo_order);
> spin_unlock(&vm->status_lock);
>
> return bo_va;
> @@ -2424,6 +2433,7 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
> INIT_LIST_HEAD(&vm->relocated);
> INIT_LIST_HEAD(&vm->moved);
> INIT_LIST_HEAD(&vm->freed);
> + atomic64_set(&vm->bo_order, 0);
>
> /* create scheduler entity for page table updates */
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
> index cf2c667ee538..bac5349d291e 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
> @@ -152,6 +152,9 @@ struct amdgpu_vm_bo_base {
>
> /* protected by the BO being reserved */
> bool moved;
> +
> + /* validate order */
> + u64 order;
> };
>
> struct amdgpu_vm_pt {
> @@ -214,6 +217,9 @@ struct amdgpu_vm {
>
> /* Limit non-retry fault storms */
> unsigned int fault_credit;
> +
> + /* per vm bo order in this vm */
> + atomic64_t bo_order;
> };
>
> struct amdgpu_vm_manager {
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply [flat|nested] 5+ messages in thread
* [PATCH 1/2] drm/amdgpu: re-validate per VM BOs if required
@ 2018-03-26 8:47 Chunming Zhou
[not found] ` <20180326084717.15465-1-david1.zhou-5C7GfCeVMHo@public.gmane.org>
0 siblings, 1 reply; 5+ messages in thread
From: Chunming Zhou @ 2018-03-26 8:47 UTC (permalink / raw)
To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW
Cc: Christian König, Christian König
From: Christian König <ckoenig.leichtzumerken@gmail.com>
If a per VM BO ends up in a allowed domain it never moves back into the
prefered domain.
Change-Id: Ifb3e561785d3b464da28c439b271c26825224c5e
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-and-Tested-by: Chunming Zhou <david1.zhou@amd.com>
---
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 15 +++++++++++++--
1 file changed, 13 insertions(+), 2 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index e9a41dd05345..7db411a282ce 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -1808,14 +1808,16 @@ int amdgpu_vm_handle_moved(struct amdgpu_device *adev,
spin_lock(&vm->status_lock);
while (!list_empty(&vm->moved)) {
- struct amdgpu_bo_va *bo_va;
struct reservation_object *resv;
+ struct amdgpu_bo_va *bo_va;
+ struct amdgpu_bo *bo;
bo_va = list_first_entry(&vm->moved,
struct amdgpu_bo_va, base.vm_status);
spin_unlock(&vm->status_lock);
- resv = bo_va->base.bo->tbo.resv;
+ bo = bo_va->base.bo;
+ resv = bo->tbo.resv;
/* Per VM BOs never need to bo cleared in the page tables */
if (resv == vm->root.base.bo->tbo.resv)
@@ -1835,6 +1837,15 @@ int amdgpu_vm_handle_moved(struct amdgpu_device *adev,
kcl_reservation_object_unlock(resv);
spin_lock(&vm->status_lock);
+
+ /* If the BO prefers to be in VRAM, but currently isn't add it
+ * back to the evicted list so that it gets validated again on
+ * the next command submission.
+ */
+ if (resv == vm->root.base.bo->tbo.resv &&
+ bo->preferred_domains == AMDGPU_GEM_DOMAIN_VRAM &&
+ bo->tbo.mem.mem_type != TTM_PL_VRAM)
+ list_add_tail(&bo_va->base.vm_status, &vm->evicted);
}
spin_unlock(&vm->status_lock);
--
2.14.1
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 5+ messages in thread
* [PATCH 2/2] drm/amdgpu: add per vm bo validation order
[not found] ` <20180326084717.15465-1-david1.zhou-5C7GfCeVMHo@public.gmane.org>
@ 2018-03-26 8:47 ` Chunming Zhou
2018-03-26 10:38 ` [PATCH 1/2] drm/amdgpu: re-validate per VM BOs if required Christian König
1 sibling, 0 replies; 5+ messages in thread
From: Chunming Zhou @ 2018-03-26 8:47 UTC (permalink / raw)
To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW; +Cc: Chunming Zhou
Change-Id: I59f1da8a765efae8196c0b1599f47a8b8485c4d3
Signed-off-by: Chunming Zhou <david1.zhou@amd.com>
---
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 14 +++++++++++++-
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h | 6 ++++++
2 files changed, 19 insertions(+), 1 deletion(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index 7db411a282ce..4dd79276c579 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -205,12 +205,22 @@ int amdgpu_vm_validate_pt_bos(struct amdgpu_device *adev, struct amdgpu_vm *vm,
spin_lock(&vm->status_lock);
while (!list_empty(&vm->evicted)) {
- struct amdgpu_vm_bo_base *bo_base;
+ struct amdgpu_vm_bo_base *bo_base, *tmp;
struct amdgpu_bo *bo;
+ u64 bo_order;
bo_base = list_first_entry(&vm->evicted,
struct amdgpu_vm_bo_base,
vm_status);
+ bo_order = bo_base->order;
+ if (bo_order > 0) {
+ list_for_each_entry(tmp, &vm->evicted, vm_status) {
+ if (tmp->order < bo_order) {
+ bo_order = tmp->order;
+ bo_base = tmp;
+ }
+ }
+ }
spin_unlock(&vm->status_lock);
bo = bo_base->bo;
@@ -1903,6 +1913,7 @@ struct amdgpu_bo_va *amdgpu_vm_bo_add(struct amdgpu_device *adev,
* */
spin_lock(&vm->status_lock);
list_move_tail(&bo_va->base.vm_status, &vm->evicted);
+ bo_va->base.order = (u64)atomic64_inc_return(&vm->bo_order);
spin_unlock(&vm->status_lock);
return bo_va;
@@ -2424,6 +2435,7 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
INIT_LIST_HEAD(&vm->relocated);
INIT_LIST_HEAD(&vm->moved);
INIT_LIST_HEAD(&vm->freed);
+ atomic64_set(&vm->bo_order, 0);
/* create scheduler entity for page table updates */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
index cf2c667ee538..bac5349d291e 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
@@ -152,6 +152,9 @@ struct amdgpu_vm_bo_base {
/* protected by the BO being reserved */
bool moved;
+
+ /* validate order */
+ u64 order;
};
struct amdgpu_vm_pt {
@@ -214,6 +217,9 @@ struct amdgpu_vm {
/* Limit non-retry fault storms */
unsigned int fault_credit;
+
+ /* per vm bo order in this vm */
+ atomic64_t bo_order;
};
struct amdgpu_vm_manager {
--
2.14.1
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 5+ messages in thread
* Re: [PATCH 1/2] drm/amdgpu: re-validate per VM BOs if required
[not found] ` <20180326084717.15465-1-david1.zhou-5C7GfCeVMHo@public.gmane.org>
2018-03-26 8:47 ` [PATCH 2/2] drm/amdgpu: add per vm bo validation order Chunming Zhou
@ 2018-03-26 10:38 ` Christian König
1 sibling, 0 replies; 5+ messages in thread
From: Christian König @ 2018-03-26 10:38 UTC (permalink / raw)
To: Chunming Zhou, amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW
Cc: Christian König
Am 26.03.2018 um 10:47 schrieb Chunming Zhou:
> From: Christian König <ckoenig.leichtzumerken@gmail.com>
>
> If a per VM BO ends up in a allowed domain it never moves back into the
> prefered domain.
>
> Change-Id: Ifb3e561785d3b464da28c439b271c26825224c5e
> Signed-off-by: Christian König <christian.koenig@amd.com>
> Reviewed-and-Tested-by: Chunming Zhou <david1.zhou@amd.com>
> ---
> drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 15 +++++++++++++--
> 1 file changed, 13 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
> index e9a41dd05345..7db411a282ce 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
> @@ -1808,14 +1808,16 @@ int amdgpu_vm_handle_moved(struct amdgpu_device *adev,
>
> spin_lock(&vm->status_lock);
> while (!list_empty(&vm->moved)) {
> - struct amdgpu_bo_va *bo_va;
> struct reservation_object *resv;
> + struct amdgpu_bo_va *bo_va;
> + struct amdgpu_bo *bo;
>
> bo_va = list_first_entry(&vm->moved,
> struct amdgpu_bo_va, base.vm_status);
> spin_unlock(&vm->status_lock);
>
> - resv = bo_va->base.bo->tbo.resv;
> + bo = bo_va->base.bo;
> + resv = bo->tbo.resv;
>
> /* Per VM BOs never need to bo cleared in the page tables */
> if (resv == vm->root.base.bo->tbo.resv)
> @@ -1835,6 +1837,15 @@ int amdgpu_vm_handle_moved(struct amdgpu_device *adev,
> kcl_reservation_object_unlock(resv);
>
> spin_lock(&vm->status_lock);
> +
> + /* If the BO prefers to be in VRAM, but currently isn't add it
> + * back to the evicted list so that it gets validated again on
> + * the next command submission.
> + */
> + if (resv == vm->root.base.bo->tbo.resv &&
> + bo->preferred_domains == AMDGPU_GEM_DOMAIN_VRAM &&
> + bo->tbo.mem.mem_type != TTM_PL_VRAM)
> + list_add_tail(&bo_va->base.vm_status, &vm->evicted);
Thinking about this more, might be even better to put that into
amdgpu_vm_bo_update() instead of the
"list_del_init(&bo_va->base.vm_status);".
I'm on easter vacation this week, so checking mails is delayed. But I'm
going to look at patch #2 this evening or tomorrow.
Regards,
Christian.
> }
> spin_unlock(&vm->status_lock);
>
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply [flat|nested] 5+ messages in thread
end of thread, other threads:[~2018-03-26 10:38 UTC | newest]
Thread overview: 5+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2018-03-26 8:47 [PATCH 1/2] drm/amdgpu: re-validate per VM BOs if required Chunming Zhou
[not found] ` <20180326084717.15465-1-david1.zhou-5C7GfCeVMHo@public.gmane.org>
2018-03-26 8:47 ` [PATCH 2/2] drm/amdgpu: add per vm bo validation order Chunming Zhou
2018-03-26 10:38 ` [PATCH 1/2] drm/amdgpu: re-validate per VM BOs if required Christian König
-- strict thread matches above, loose matches on Subject: below --
2018-03-26 8:29 Chunming Zhou
[not found] ` <20180326082955.11488-1-david1.zhou-5C7GfCeVMHo@public.gmane.org>
2018-03-26 8:29 ` [PATCH 2/2] drm/amdgpu: add per vm bo validation order Chunming Zhou
[not found] ` <20180326082955.11488-2-david1.zhou-5C7GfCeVMHo@public.gmane.org>
2018-03-26 8:34 ` zhoucm1
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox