diff options
Diffstat (limited to 'meta-amd-bsp/recipes-kernel/linux/linux-yocto-4.19.8/2712-drm-amdkfd-Use-functions-from-amdgpu-to-invalidate-v.patch')
| -rw-r--r-- | meta-amd-bsp/recipes-kernel/linux/linux-yocto-4.19.8/2712-drm-amdkfd-Use-functions-from-amdgpu-to-invalidate-v.patch | 86 |
1 files changed, 86 insertions, 0 deletions
diff --git a/meta-amd-bsp/recipes-kernel/linux/linux-yocto-4.19.8/2712-drm-amdkfd-Use-functions-from-amdgpu-to-invalidate-v.patch b/meta-amd-bsp/recipes-kernel/linux/linux-yocto-4.19.8/2712-drm-amdkfd-Use-functions-from-amdgpu-to-invalidate-v.patch new file mode 100644 index 00000000..bac9e30e --- /dev/null +++ b/meta-amd-bsp/recipes-kernel/linux/linux-yocto-4.19.8/2712-drm-amdkfd-Use-functions-from-amdgpu-to-invalidate-v.patch @@ -0,0 +1,86 @@ +From 0e1eb95195729d57d63fe05abb8f54221ad3758a Mon Sep 17 00:00:00 2001 +From: Yong Zhao <Yong.Zhao@amd.com> +Date: Fri, 12 Oct 2018 17:23:09 -0400 +Subject: [PATCH 2712/2940] drm/amdkfd: Use functions from amdgpu to invalidate + vmid in kfd +MIME-Version: 1.0 +Content-Type: text/plain; charset=UTF-8 +Content-Transfer-Encoding: 8bit + +As part of the change, we stop taking the srbm lock, and start to use +the same invalidation engine and software lock as amdgpu. + +Signed-off-by: Yong Zhao <Yong.Zhao@amd.com> +Reviewed-by: Alex Deucher <alexander.deucher@amd.com> +Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> +Reviewed-by: Christian König <christian.koenig@amd.com> +Signed-off-by: Alex Deucher <alexander.deucher@amd.com> +--- + .../gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c | 37 +------------------ + 1 file changed, 1 insertion(+), 36 deletions(-) + +diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c +index f3323d852c04..4fed797651b1 100644 +--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c ++++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c +@@ -49,17 +49,6 @@ + #include "soc15d.h" + #include "gmc_v9_0.h" + +-/* HACK: MMHUB and GC both have VM-related register with the same +- * names but different offsets. Define the MMHUB register we need here +- * with a prefix. A proper solution would be to move the functions +- * programming these registers into gfx_v9_0.c and mmhub_v1_0.c +- * respectively. +- */ +-#define mmMMHUB_VM_INVALIDATE_ENG16_REQ 0x06f3 +-#define mmMMHUB_VM_INVALIDATE_ENG16_REQ_BASE_IDX 0 +- +-#define mmMMHUB_VM_INVALIDATE_ENG16_ACK 0x0705 +-#define mmMMHUB_VM_INVALIDATE_ENG16_ACK_BASE_IDX 0 + + #define V9_PIPE_PER_MEC (4) + #define V9_QUEUES_PER_PIPE_MEC (8) +@@ -742,15 +731,6 @@ static uint16_t get_atc_vmid_pasid_mapping_pasid(struct kgd_dev *kgd, + static void write_vmid_invalidate_request(struct kgd_dev *kgd, uint8_t vmid) + { + struct amdgpu_device *adev = (struct amdgpu_device *) kgd; +- uint32_t req = (1 << vmid) | +- (0 << VM_INVALIDATE_ENG16_REQ__FLUSH_TYPE__SHIFT) | /* legacy */ +- VM_INVALIDATE_ENG16_REQ__INVALIDATE_L2_PTES_MASK | +- VM_INVALIDATE_ENG16_REQ__INVALIDATE_L2_PDE0_MASK | +- VM_INVALIDATE_ENG16_REQ__INVALIDATE_L2_PDE1_MASK | +- VM_INVALIDATE_ENG16_REQ__INVALIDATE_L2_PDE2_MASK | +- VM_INVALIDATE_ENG16_REQ__INVALIDATE_L1_PTES_MASK; +- +- mutex_lock(&adev->srbm_mutex); + + /* Use legacy mode tlb invalidation. + * +@@ -767,22 +747,7 @@ static void write_vmid_invalidate_request(struct kgd_dev *kgd, uint8_t vmid) + * TODO 2: support range-based invalidation, requires kfg2kgd + * interface change + */ +- WREG32(SOC15_REG_OFFSET(GC, 0, mmVM_INVALIDATE_ENG16_REQ), req); +- +- WREG32(SOC15_REG_OFFSET(MMHUB, 0, mmMMHUB_VM_INVALIDATE_ENG16_REQ), +- req); +- +- while (!(RREG32(SOC15_REG_OFFSET(GC, 0, mmVM_INVALIDATE_ENG16_ACK)) & +- (1 << vmid))) +- cpu_relax(); +- +- while (!(RREG32(SOC15_REG_OFFSET(MMHUB, 0, +- mmMMHUB_VM_INVALIDATE_ENG16_ACK)) & +- (1 << vmid))) +- cpu_relax(); +- +- mutex_unlock(&adev->srbm_mutex); +- ++ amdgpu_gmc_flush_gpu_tlb(adev, vmid, 0); + } + + static int invalidate_tlbs_with_kiq(struct amdgpu_device *adev, uint16_t pasid) +-- +2.17.1 + |