Re: [PATCH] drm/amdgpu : Use XGMI mapping when devices on the same hive v2

["Russell, Kent" <Kent.Russell@xxxxxxx>]

It was merged to 4.19 on Sept 21. It got missed on the 4.20 rebase.

Kent

KENT RUSSELL
Sr. Software Engineer | Linux Compute Kernel
1 Commerce Valley Drive East
Markham, ON L3T 7X6
O +(1) 289-695-2122 | Ext 72122

---

From: Kuehling, Felix
Sent: Thursday, November 15, 2018 12:57:44 PM
To: Russell, Kent; amd-gfx@xxxxxxxxxxxxxxxxxxxxx
Cc: Russell, Kent; Liu, Shaoyun
Subject: RE: [PATCH] drm/amdgpu : Use XGMI mapping when devices on the same hive v2

 

This change is not suitable for amd-staging-drm-next. PCIe P2P was not enabled on amd-staging-drm-next because it's not reliable yet. This change enables it even in situations that are not safe (including small BAR systems).

Why are you porting this change to amd-staging-drm-next? Does anyone depend on XGMI support on this branch?

Regards,
  Felix

-----Original Message-----
From: amd-gfx <amd-gfx-bounces@xxxxxxxxxxxxxxxxxxxxx> On Behalf Of Russell, Kent
Sent: Thursday, November 15, 2018 11:54 AM
To: amd-gfx@xxxxxxxxxxxxxxxxxxxxx
Cc: Russell, Kent <Kent.Russell@xxxxxxx>; Liu, Shaoyun <Shaoyun.Liu@xxxxxxx>
Subject: [PATCH] drm/amdgpu : Use XGMI mapping when devices on the same hive v2

From: Shaoyun Liu <Shaoyun.Liu@xxxxxxx>

VM mapping will only fall back to P2P if XGMI mapping is not available

V2: Rebase onto 4.20

Change-Id: I7a854ab3d5c9958bd45d4fe439ea7e370a092e7a
Signed-off-by: Shaoyun Liu <Shaoyun.Liu@xxxxxxx>
Reviewed-by: Felix Kuehling <Felix.Kuehling@xxxxxxx>
Reviewed-by: Huang Rui <ray.huang@xxxxxxx>
Reviewed-by: Christian König <christian.koenig@xxxxxxx>
Signed-off-by: Kent Russell <kent.russell@xxxxxxx>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 16 ++++++++++++++--
 1 file changed, 14 insertions(+), 2 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index dad0e23..576d168 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -2011,6 +2011,8 @@ int amdgpu_vm_bo_update(struct amdgpu_device *adev,
         struct drm_mm_node *nodes;
         struct dma_fence *exclusive, **last_update;
         uint64_t flags;
+       uint64_t vram_base_offset = adev->vm_manager.vram_base_offset;
+       struct amdgpu_device *bo_adev;
         int r;
 
         if (clear || !bo) {
@@ -2029,9 +2031,19 @@ int amdgpu_vm_bo_update(struct amdgpu_device *adev,
                 exclusive = reservation_object_get_excl(bo->tbo.resv);
         }
 
-       if (bo)
+       if (bo) {
                 flags = amdgpu_ttm_tt_pte_flags(adev, bo->tbo.ttm, mem);
-       else
+               bo_adev = amdgpu_ttm_adev(bo->tbo.bdev);
+               if (mem && mem->mem_type == TTM_PL_VRAM && adev != bo_adev) {
+                       if (adev->gmc.xgmi.hive_id &&
+                           adev->gmc.xgmi.hive_id == bo_adev->gmc.xgmi.hive_id) {
+                               vram_base_offset = bo_adev->vm_manager.vram_base_offset;
+                       } else {
+                               flags |= AMDGPU_PTE_SYSTEM;
+                               vram_base_offset = bo_adev->gmc.aper_base;
+                       }
+               }
+       } else
                 flags = 0x0;
 
         if (clear || (bo && bo->tbo.resv == vm->root.base.bo->tbo.resv))
--
2.7.4

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

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