drm/amdgpu/gmc9: no need use kiq in vega10 tlb flush

two reasons:
1. there is a spinlock around;
2. vm register is pf/vf copy, vf can access via mmio safely.

Signed-off-by: Xiangliang Yu <Xiangliang.Yu@amd.com>
Signed-off-by: Monk Liu <Monk.Liu@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This commit is contained in:
Xiangliang Yu 2017-02-28 17:06:36 +08:00 committed by Alex Deucher
parent 86d3798af0
commit c7a7266b7d

View File

@ -197,11 +197,11 @@ static void gmc_v9_0_gart_flush_gpu_tlb(struct amdgpu_device *adev,
struct amdgpu_vmhub *hub = &adev->vmhub[i];
u32 tmp = hub->get_invalidate_req(vmid);
WREG32(hub->vm_inv_eng0_req + eng, tmp);
WREG32_NO_KIQ(hub->vm_inv_eng0_req + eng, tmp);
/* Busy wait for ACK.*/
for (j = 0; j < 100; j++) {
tmp = RREG32(hub->vm_inv_eng0_ack + eng);
tmp = RREG32_NO_KIQ(hub->vm_inv_eng0_ack + eng);
tmp &= 1 << vmid;
if (tmp)
break;
@ -212,7 +212,7 @@ static void gmc_v9_0_gart_flush_gpu_tlb(struct amdgpu_device *adev,
/* Wait for ACK with a delay.*/
for (j = 0; j < adev->usec_timeout; j++) {
tmp = RREG32(hub->vm_inv_eng0_ack + eng);
tmp = RREG32_NO_KIQ(hub->vm_inv_eng0_ack + eng);
tmp &= 1 << vmid;
if (tmp)
break;