Skip to contentSkip to conversation

drm/amdgpu: fix sdma doorbell comments typo

PATCH NAK Frank.Min 6 participants 8 messages Oct 11, 2018 – Oct 12, 2018 7 messages not in the archive

Subsystems (2)
DRM drivers Radeon and amdgpu DRM drivers
lore
Share & tools
lore

Frank.Min

author
#0 lore raw

PATCH 1/2 drm/amdgpu: fix sdma doorbell comments typo

Change-Id I0a3dff9f01a90717e0c32b7fa81a5e891bd1d52d
Signed-off-by Frank.Min <[email protected]>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/amd/amdgpu/amdgpu.h +1 -1
index c21d9b9..6317f35 100644
@@ -432,7 +432,7 @@ typedef enum _AMDGPU_DOORBELL64_ASSIGNMENT
 	 * default non-graphics QWORD index is 0xe0 - 0xFF inclusive
 	 */
 
-	/* sDMA engines  reserved from 0xe0 -oxef  */
+	/* sDMA engines  reserved from 0xe0 -0xef  */
 	AMDGPU_DOORBELL64_sDMA_ENGINE0            = 0xE0,
 	AMDGPU_DOORBELL64_sDMA_HI_PRI_ENGINE0     = 0xE1,
 	AMDGPU_DOORBELL64_sDMA_ENGINE1            = 0xE8,
Signature · 6 lines
2.7.4

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

A message that is not in the archive (sent off-list or not yet received): 1539259752-15172-1-git-send-email-Frank.Min-5C7GfCeVMHo@public.gmane.org

author
#2 lore raw

PATCH 2/2 drm/amdgpu: skip vm entries checking while in sriov vf

vm page table would be allocated while in amdgpu_vm_init by amdgpu_allocate_static_csa for sriov, so the checking here would be skipped.

Change-Id Id30b86ad15ae509aeed9ed8ab60c259c88af3df5
Signed-off-by Frank.Min <[email protected]>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +5 -4
index 6904d79..6066f87 100644
@@ -3100,11 +3100,12 @@ int amdgpu_vm_make_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm, uns
 		return r;
 
 	/* Sanity checks */
-	if (!RB_EMPTY_ROOT(&vm->va.rb_root) || vm->root.entries) {
-		r = -EINVAL;
-		goto unreserve_bo;
+	if (!amdgpu_sriov_vf(adev)) {
+		if (!RB_EMPTY_ROOT(&vm->va.rb_root) || vm->root.entries) {
+			r = -EINVAL;
+			goto unreserve_bo;
+		}
 	}
-
 	if (pasid) {
 		unsigned long flags;
Signature · 6 lines
2.7.4

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

A message that is not in the archive (sent off-list or not yet received): 1539259752-15172-2-git-send-email-Frank.Min-5C7GfCeVMHo@public.gmane.org

NAK
#4 lore raw

Re: [PATCH 2/2] drm/amdgpu: skip vm entries checking while in sriov vf

Am 11.10.2018 um 14:09 schrieb Frank.Min:

Quoted 3 lines

vm page table would be allocated while in amdgpu_vm_init by amdgpu_allocate_static_csa for sriov, so the checking here would be skipped.

NAK, that checking is actually correct and points out that this doesn't work correctly.

Fix the function to be able to handle the case when there are already mappings in the VM.

Christian.

Quoted 18 lines
Change-Id Id30b86ad15ae509aeed9ed8ab60c259c88af3df5
Signed-off-by Frank.Min <[email protected]>
---
  drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 9 +++++----
  1 file changed, 5 insertions(+), 4 deletions(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +0 -3
index 6904d79..6066f87 100644
@@ -3100,11 +3100,12 @@ int amdgpu_vm_make_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm, uns
  		return r;
 
  	/* Sanity checks */
-	if (!RB_EMPTY_ROOT(&vm->va.rb_root) || vm->root.entries) {
-		r = -EINVAL;
-		goto unreserve_bo;

drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c int amdgpu_vm_make_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm, uns · line 3111

Quoted 10 lines
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +5 -1
+	if (!amdgpu_sriov_vf(adev)) {
+		if (!RB_EMPTY_ROOT(&vm->va.rb_root) || vm->root.entries) {
+			r = -EINVAL;
+			goto unreserve_bo;
+		}
  	}
-
  	if (pasid) {
  		unsigned long flags;

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

A message that is not in the archive (sent off-list or not yet received): b776d7bb-dd47-4130-fd45-9b4a432b92d5-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org

#6 lore raw

Re: [PATCH 2/2] drm/amdgpu: skip vm entries checking while in sriov vf

Quoted 10 lines · Christian König

Am 11.10.2018 um 14:09 schrieb Frank.Min:

Quoted 3 lines

vm page table would be allocated while in amdgpu_vm_init by amdgpu_allocate_static_csa for sriov, so the checking here would be skipped.

NAK, that checking is actually correct and points out that this doesn't work correctly.

Fix the function to be able to handle the case when there are already mappings in the VM.

KFD and the Thunk assume that they own the entire address space. If something is already mapped, KFD and the Thunk would not be aware of it.

amdgpu_allocate_static_csa runs during driver initialization, before there are any VMs. Looks like amdgpu_map_static_csa is the more important function here, which currently runs in amdgpu_driver_open_kms.

Could amdgpu_map_static_csa be delayed to the first time user mode maps memory into the VM through GEM APIs? That way it would allow turning a VM into a compute VM that hasn't been used for any GEM memory mappings.

Regards,
  Felix

Quoted 29 lines

Christian.

Quoted 26 lines
Change-Id Id30b86ad15ae509aeed9ed8ab60c259c88af3df5
Signed-off-by Frank.Min <[email protected]>

---
  drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 9 +++++----

  1 file changed, 5 insertions(+), 4 deletions(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +0 -0
index 6904d79..6066f87 100644
@@ -3100,11 +3100,12 @@ int amdgpu_vm_make_compute(struct

amdgpu_device *adev, struct amdgpu_vm *vm, uns
          return r;
        /* Sanity checks */
-    if (!RB_EMPTY_ROOT(&vm->va.rb_root) || vm->root.entries) { -        r = -EINVAL;
-        goto unreserve_bo;
+    if (!amdgpu_sriov_vf(adev)) {
+        if (!RB_EMPTY_ROOT(&vm->va.rb_root) || vm->root.entries) { +            r = -EINVAL;
+            goto unreserve_bo;
+        }
      }
-

Quoted 8 lines
Quoted 3 lines

      if (pasid) {
          unsigned long flags;

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

A message that is not in the archive (sent off-list or not yet received): 9dcce91a-2176-feea-9a89-19fa1b55959a-5C7GfCeVMHo@public.gmane.org

#8 lore raw

RE: [PATCH 2/2] drm/amdgpu: skip vm entries checking while in sriov vf

Quoted 24 lines

-----Original Message-----
From: amd-gfx [mailto:[email protected]] On Behalf Of Felix Kuehling
Sent: Friday, October 12, 2018 3:46 AM
To: Koenig, Christian <[email protected]>; Min, Frank <[email protected]>; [email protected]
Subject: Re: [PATCH 2/2] drm/amdgpu: skip vm entries checking while in sriov vf

Quoted 10 lines · Christian König

Am 11.10.2018 um 14:09 schrieb Frank.Min:

Quoted 3 lines

vm page table would be allocated while in amdgpu_vm_init by amdgpu_allocate_static_csa for sriov, so the checking here would be skipped.

NAK, that checking is actually correct and points out that this doesn't work correctly.

Fix the function to be able to handle the case when there are already mappings in the VM.

KFD and the Thunk assume that they own the entire address space. If something is already mapped, KFD and the Thunk would not be aware of it.

Quoted 9 lines

amdgpu_allocate_static_csa runs during driver initialization, before there are any VMs. Looks like amdgpu_map_static_csa is the more important function here, which currently runs in amdgpu_driver_open_kms.

Could amdgpu_map_static_csa be delayed to the first time user mode maps memory into the VM through GEM APIs? That way it would allow turning a VM into a compute VM that hasn't been used for any GEM memory mappings.

Yes. Frank and I found, while we do acquire_process_vm(), the vm->va RB trees are always not empty under SRIOV, because amdgpu_map_static_csa() will be called while vm inits under SRIOV, that will actually map the CSA BO to the VM and insert the node into vm->va before it acquires process VM in KFD.

Thanks,
Ray

Quoted 41 lines

Regards,
  Felix

Quoted 29 lines

Christian.

Quoted 26 lines
Change-Id Id30b86ad15ae509aeed9ed8ab60c259c88af3df5
Signed-off-by Frank.Min <[email protected]>

---
  drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 9 +++++----

  1 file changed, 5 insertions(+), 4 deletions(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +0 -0
index 6904d79..6066f87 100644
@@ -3100,11 +3100,12 @@ int amdgpu_vm_make_compute(struct

amdgpu_device *adev, struct amdgpu_vm *vm, uns
          return r;
        /* Sanity checks */
-    if (!RB_EMPTY_ROOT(&vm->va.rb_root) || vm->root.entries) { -        r = -EINVAL;
-        goto unreserve_bo;
+    if (!amdgpu_sriov_vf(adev)) {
+        if (!RB_EMPTY_ROOT(&vm->va.rb_root) || vm->root.entries) { +            r = -EINVAL;
+            goto unreserve_bo;
+        }
      }
-

Quoted 8 lines
Quoted 3 lines

      if (pasid) {
          unsigned long flags;

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

Quoted 4 lines

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

A message that is not in the archive (sent off-list or not yet received): BY2PR12MB00408D294FB316A26BD59330ECE20-K//h7OWB4q7Zvl48JdS6+wdYzm3356FpvxpqHgZTriW3zl9H0oFU5g@public.gmane.org

#10 lore raw

Re: [PATCH 2/2] drm/amdgpu: skip vm entries checking while in sriov vf

Am 12.10.2018 um 06:05 schrieb Huang, Ray:

Quoted 26 lines
Quoted 26 lines

-----Original Message-----
From: amd-gfx [mailto:[email protected]] On Behalf Of Felix Kuehling
Sent: Friday, October 12, 2018 3:46 AM
To: Koenig, Christian <[email protected]>; Min, Frank <[email protected]>; [email protected]
Subject: Re: [PATCH 2/2] drm/amdgpu: skip vm entries checking while in sriov vf

Quoted 9 lines · Christian König

Am 11.10.2018 um 14:09 schrieb Frank.Min:

Quoted 3 lines

vm page table would be allocated while in amdgpu_vm_init by amdgpu_allocate_static_csa for sriov, so the checking here would be skipped.

NAK, that checking is actually correct and points out that this doesn't work correctly.

Fix the function to be able to handle the case when there are already mappings in the VM.

KFD and the Thunk assume that they own the entire address space. If something is already mapped, KFD and the Thunk would not be aware of it.

amdgpu_allocate_static_csa runs during driver initialization, before there are any VMs. Looks like amdgpu_map_static_csa is the more important function here, which currently runs in amdgpu_driver_open_kms.

Quoted 6 lines
Quoted 5 lines

Could amdgpu_map_static_csa be delayed to the first time user mode maps memory into the VM through GEM APIs? That way it would allow turning a VM into a compute VM that hasn't been used for any GEM memory mappings.

Yes. Frank and I found, while we do acquire_process_vm(), the vm->va RB trees are always not empty under SRIOV, because amdgpu_map_static_csa() will be called while vm inits under SRIOV, that will actually map the CSA BO to the VM and insert the node into vm->va before it acquires process VM in KFD.

Yeah and exactly that will clash with the KFD.

Could be that it still works initially because the CSA is not mapped at the beginning of the address space IIRC, but we need to clean that up in the long term.

We should either delay adding the CSA BO to the VM (probably best done during context creation).

Or we start to make nails with heads and move the CSA BO into the context altogether. We are going to need this anyway for MCBP, so that would be a good first step.

Ray/Frank can anybody of you take care of this?

Thanks,
Christian.

Quoted 45 lines

Thanks,
Ray

Quoted 41 lines

Regards,
  Felix

Quoted 26 lines

Christian.

Quoted 24 lines
Change-Id Id30b86ad15ae509aeed9ed8ab60c259c88af3df5
Signed-off-by Frank.Min <[email protected]>
---
   drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 9 +++++----
   1 file changed, 5 insertions(+), 4 deletions(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +0 -0
index 6904d79..6066f87 100644
@@ -3100,11 +3100,12 @@ int amdgpu_vm_make_compute(struct

amdgpu_device *adev, struct amdgpu_vm *vm, uns
          return r;
        /* Sanity checks */
-    if (!RB_EMPTY_ROOT(&vm->va.rb_root) || vm->root.entries) { -        r = -EINVAL;
-        goto unreserve_bo;
+    if (!amdgpu_sriov_vf(adev)) {
+        if (!RB_EMPTY_ROOT(&vm->va.rb_root) || vm->root.entries) { +            r = -EINVAL;
+            goto unreserve_bo;
+        }
      }

Quoted 8 lines
Quoted 4 lines

-
      if (pasid) {
          unsigned long flags;

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

A message that is not in the archive (sent off-list or not yet received): 27b55866-bacf-422e-b357-79459cf8a911-5C7GfCeVMHo@public.gmane.org

#12 lore raw

RE: [PATCH 2/2] drm/amdgpu: skip vm entries checking while in sriov vf

Quoted 51 lines

-----Original Message-----
From: amd-gfx <[email protected]> On Behalf Of Koenig, Christian
Sent: Friday, October 12, 2018 3:25 PM
To: Huang, Ray <[email protected]>; Kuehling, Felix <[email protected]>; Min, Frank <[email protected]>; amd- [email protected]
Subject: Re: [PATCH 2/2] drm/amdgpu: skip vm entries checking while in sriov vf

Am 12.10.2018 um 06:05 schrieb Huang, Ray:

Quoted 14 lines
Quoted 14 lines

-----Original Message-----
From: amd-gfx [mailto:[email protected]] On Behalf Of Felix Kuehling
Sent: Friday, October 12, 2018 3:46 AM
To: Koenig, Christian <[email protected]>; Min, Frank <[email protected]>; [email protected]
Subject: Re: [PATCH 2/2] drm/amdgpu: skip vm entries checking while in sriov vf

Quoted 4 lines · Christian König

Am 11.10.2018 um 14:09 schrieb Frank.Min:

Quoted 3 lines

vm page table would be allocated while in amdgpu_vm_init by amdgpu_allocate_static_csa for sriov, so the checking here would be skipped.

Quoted 7 lines
Quoted 7 lines
Quoted 5 lines

NAK, that checking is actually correct and points out that this doesn't work correctly.

Fix the function to be able to handle the case when there are already mappings in the VM.

KFD and the Thunk assume that they own the entire address space. If something is already mapped, KFD and the Thunk would not be aware of

it.

Quoted 4 lines
Quoted 4 lines

amdgpu_allocate_static_csa runs during driver initialization, before there are any VMs. Looks like amdgpu_map_static_csa is the more important function here, which currently runs in

amdgpu_driver_open_kms.

Quoted 7 lines
Quoted 6 lines

Could amdgpu_map_static_csa be delayed to the first time user mode maps memory into the VM through GEM APIs? That way it would allow turning a VM into a compute VM that hasn't been used for any GEM memory mappings.

Yes. Frank and I found, while we do acquire_process_vm(), the vm->va RB

trees are always not empty under SRIOV, because amdgpu_map_static_csa() will be called while vm inits under SRIOV, that will actually map the CSA BO to the VM and insert the node into vm->va before it acquires process VM in KFD.

Quoted 8 lines

Yeah and exactly that will clash with the KFD.

Could be that it still works initially because the CSA is not mapped at the beginning of the address space IIRC, but we need to clean that up in the long term.

We should either delay adding the CSA BO to the VM (probably best done during context creation).

Yes, we planed to create sdma csa buffer when context creation.

Quoted 1 line

Or we start to make nails with heads and move the CSA BO into the context

So we can add the sdma_csa bo/static_csa in the struct amdgpu_ctx?

Best Regards
Rex

Quoted 54 lines

altogether. We are going to need this anyway for MCBP, so that would be a good first step.

Ray/Frank can anybody of you take care of this?

Thanks,
Christian.

Quoted 45 lines

Thanks,
Ray

Quoted 41 lines

Regards,
  Felix

Quoted 26 lines

Christian.

Quoted 24 lines
Change-Id Id30b86ad15ae509aeed9ed8ab60c259c88af3df5
Signed-off-by Frank.Min <[email protected]>
---
   drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 9 +++++----
   1 file changed, 5 insertions(+), 4 deletions(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +0 -0
index 6904d79..6066f87 100644
@@ -3100,11 +3100,12 @@ int amdgpu_vm_make_compute(struct

amdgpu_device *adev, struct amdgpu_vm *vm, uns
          return r;
        /* Sanity checks */
-    if (!RB_EMPTY_ROOT(&vm->va.rb_root) || vm->root.entries) { -        r = -EINVAL;
-        goto unreserve_bo;
+    if (!amdgpu_sriov_vf(adev)) {
+        if (!RB_EMPTY_ROOT(&vm->va.rb_root) || vm->root.entries) { +            r = -EINVAL;
+            goto unreserve_bo;
+        }
      }

Quoted 8 lines
Quoted 4 lines

-
      if (pasid) {
          unsigned long flags;

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

Quoted 4 lines

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

A message that is not in the archive (sent off-list or not yet received): BYAPR12MB2775CBBCA321F3C4A0317BBFFBE20-ZGDeBxoHBPmJeBUhB162ZQdYzm3356FpvxpqHgZTriW3zl9H0oFU5g@public.gmane.org

#14 lore raw

Re: [PATCH 2/2] drm/amdgpu: skip vm entries checking while in sriov vf

Am 12.10.2018 um 09:31 schrieb Zhu, Rex:

Quoted 54 lines
Quoted 54 lines

-----Original Message-----
From: amd-gfx <[email protected]> On Behalf Of Koenig, Christian
Sent: Friday, October 12, 2018 3:25 PM
To: Huang, Ray <[email protected]>; Kuehling, Felix <[email protected]>; Min, Frank <[email protected]>; amd- [email protected]
Subject: Re: [PATCH 2/2] drm/amdgpu: skip vm entries checking while in sriov vf

Am 12.10.2018 um 06:05 schrieb Huang, Ray:

Quoted 14 lines
Quoted 14 lines

-----Original Message-----
From: amd-gfx [mailto:[email protected]] On Behalf Of Felix Kuehling
Sent: Friday, October 12, 2018 3:46 AM
To: Koenig, Christian <[email protected]>; Min, Frank <[email protected]>; [email protected]
Subject: Re: [PATCH 2/2] drm/amdgpu: skip vm entries checking while in sriov vf

Quoted 4 lines · Christian König

Am 11.10.2018 um 14:09 schrieb Frank.Min:

Quoted 3 lines

vm page table would be allocated while in amdgpu_vm_init by amdgpu_allocate_static_csa for sriov, so the checking here would be skipped.

Quoted 7 lines
Quoted 7 lines
Quoted 5 lines

NAK, that checking is actually correct and points out that this doesn't work correctly.

Fix the function to be able to handle the case when there are already mappings in the VM.

KFD and the Thunk assume that they own the entire address space. If something is already mapped, KFD and the Thunk would not be aware of

it.

Quoted 3 lines
Quoted 3 lines

amdgpu_allocate_static_csa runs during driver initialization, before there are any VMs. Looks like amdgpu_map_static_csa is the more important function here, which currently runs in

amdgpu_driver_open_kms.

Quoted 6 lines
Quoted 5 lines

Could amdgpu_map_static_csa be delayed to the first time user mode maps memory into the VM through GEM APIs? That way it would allow turning a VM into a compute VM that hasn't been used for any GEM memory mappings.

Yes. Frank and I found, while we do acquire_process_vm(), the vm->va RB

trees are always not empty under SRIOV, because amdgpu_map_static_csa() will be called while vm inits under SRIOV, that will actually map the CSA BO to the VM and insert the node into vm->va before it acquires process VM in KFD.

Yeah and exactly that will clash with the KFD.

Could be that it still works initially because the CSA is not mapped at the beginning of the address space IIRC, but we need to clean that up in the long term.

Quoted 6 lines
Quoted 2 lines

We should either delay adding the CSA BO to the VM (probably best done during context creation).

Yes, we planed to create sdma csa buffer when context creation.

Quoted 1 line

Or we start to make nails with heads and move the CSA BO into the context

So we can add the sdma_csa bo/static_csa in the struct amdgpu_ctx?

Yes, exactly that's my idea as well.

Regards,
Christian.

Quoted 59 lines

Best Regards
Rex

Quoted 55 lines

altogether. We are going to need this anyway for MCBP, so that would be a good first step.
Ray/Frank can anybody of you take care of this?

Thanks,
Christian.

Quoted 44 lines

Thanks,
Ray

Quoted 41 lines

Regards,
  Felix

Quoted 26 lines

Christian.

Quoted 24 lines
Change-Id Id30b86ad15ae509aeed9ed8ab60c259c88af3df5
Signed-off-by Frank.Min <[email protected]>
---
    drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 9 +++++----
    1 file changed, 5 insertions(+), 4 deletions(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +0 -0
index 6904d79..6066f87 100644
@@ -3100,11 +3100,12 @@ int amdgpu_vm_make_compute(struct

amdgpu_device *adev, struct amdgpu_vm *vm, uns
          return r;
        /* Sanity checks */
-    if (!RB_EMPTY_ROOT(&vm->va.rb_root) || vm->root.entries) { -        r = -EINVAL;
-        goto unreserve_bo;
+    if (!amdgpu_sriov_vf(adev)) {
+        if (!RB_EMPTY_ROOT(&vm->va.rb_root) || vm->root.entries) { +            r = -EINVAL;
+            goto unreserve_bo;
+        }
      }

Quoted 8 lines
Quoted 4 lines

-
      if (pasid) {
          unsigned long flags;

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx