drm/amdgpu: support ras on SRIOV

support umc/gfx/sdma ras on guest side

Changed from V1:
    move sriov judgment in amdgpu_ras_interrupt_fatal_error_handler

Signed-off-by: Stanley.Yang <Stanley.Yang@amd.com>
Reviewed-by: Tao Zhou <tao.zhou1@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

4 files changed, 44 insertions, 15 deletions

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index 9af8d7a1d011..c45736a902fb 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -5219,6 +5219,10 @@ retry:	/* Rest of adevs pre asic reset from XGMI hive. */
 		r = amdgpu_device_reset_sriov(adev, job ? false : true);
 		if (r)
 			adev->asic_reset_res = r;
+
+		/* Aldebaran supports ras in SRIOV, so need resume ras during reset */
+		if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 2))
+			amdgpu_ras_resume(adev);
 	} else {
 		r = amdgpu_do_asic_reset(device_list_handle, &reset_context);
 		if (r && r == -EAGAIN)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
index 035891ec59d5..2de9309a4193 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
@@ -726,7 +726,9 @@ int amdgpu_ras_feature_enable(struct amdgpu_device *adev,
 	/* Do not enable if it is not allowed. */
 	WARN_ON(enable && !amdgpu_ras_is_feature_allowed(adev, head));
 
-	if (!amdgpu_ras_intr_triggered()) {
+	/* Only enable ras feature operation handle on host side */
+	if (!amdgpu_sriov_vf(adev) &&
+		!amdgpu_ras_intr_triggered()) {
 		ret = psp_ras_enable_features(&adev->psp, info, enable);
 		if (ret) {
 			dev_err(adev->dev, "ras %s %s failed poison:%d ret:%d\n",
@@ -1523,7 +1525,9 @@ static int amdgpu_ras_fs_fini(struct amdgpu_device *adev)
  */
 void amdgpu_ras_interrupt_fatal_error_handler(struct amdgpu_device *adev)
 {
-	if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__PCIE_BIF))
+	/* Fatal error events are handled on host side */
+	if (amdgpu_sriov_vf(adev) ||
+		!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__PCIE_BIF))
 		return;
 
 	if (adev->nbio.ras &&
@@ -2270,10 +2274,14 @@ static void amdgpu_ras_check_supported(struct amdgpu_device *adev)
 {
 	adev->ras_hw_enabled = adev->ras_enabled = 0;
 
-	if (amdgpu_sriov_vf(adev) || !adev->is_atom_fw ||
+	if (!adev->is_atom_fw ||
 	    !amdgpu_ras_asic_supported(adev))
 		return;
 
+	if (!(amdgpu_sriov_vf(adev) &&
+		(adev->ip_versions[MP1_HWIP][0] == IP_VERSION(13, 0, 2))))
+		return;
+
 	if (!adev->gmc.xgmi.connected_to_cpu) {
 		if (amdgpu_atomfirmware_mem_ecc_supported(adev)) {
 			dev_info(adev->dev, "MEM ECC is active.\n");
@@ -2285,15 +2293,21 @@ static void amdgpu_ras_check_supported(struct amdgpu_device *adev)
 
 		if (amdgpu_atomfirmware_sram_ecc_supported(adev)) {
 			dev_info(adev->dev, "SRAM ECC is active.\n");
-			adev->ras_hw_enabled |= ~(1 << AMDGPU_RAS_BLOCK__UMC |
-						    1 << AMDGPU_RAS_BLOCK__DF);
-
-			if (adev->ip_versions[VCN_HWIP][0] == IP_VERSION(2, 6, 0))
-				adev->ras_hw_enabled |= (1 << AMDGPU_RAS_BLOCK__VCN |
-						1 << AMDGPU_RAS_BLOCK__JPEG);
-			else
-				adev->ras_hw_enabled &= ~(1 << AMDGPU_RAS_BLOCK__VCN |
-						1 << AMDGPU_RAS_BLOCK__JPEG);
+			if (!amdgpu_sriov_vf(adev)) {
+				adev->ras_hw_enabled |= ~(1 << AMDGPU_RAS_BLOCK__UMC |
+							    1 << AMDGPU_RAS_BLOCK__DF);
+
+				if (adev->ip_versions[VCN_HWIP][0] == IP_VERSION(2, 6, 0))
+					adev->ras_hw_enabled |= (1 << AMDGPU_RAS_BLOCK__VCN |
+							1 << AMDGPU_RAS_BLOCK__JPEG);
+				else
+					adev->ras_hw_enabled &= ~(1 << AMDGPU_RAS_BLOCK__VCN |
+							1 << AMDGPU_RAS_BLOCK__JPEG);
+			} else {
+				adev->ras_hw_enabled |= (1 << AMDGPU_RAS_BLOCK__PCIE_BIF |
+								1 << AMDGPU_RAS_BLOCK__SDMA |
+								1 << AMDGPU_RAS_BLOCK__GFX);
+			}
 		} else {
 			dev_info(adev->dev, "SRAM ECC is not presented.\n");
 		}
@@ -2637,6 +2651,10 @@ int amdgpu_ras_late_init(struct amdgpu_device *adev)
 	struct amdgpu_ras_block_object *obj;
 	int r;
 
+	/* Guest side doesn't need init ras feature */
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
 	list_for_each_entry_safe(node, tmp, &adev->ras_list, node) {
 		if (!node->ras_obj) {
 			dev_warn(adev->dev, "Warning: abnormal ras list node.\n");
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c
index 8e221a1ba937..42c1f050542f 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c
@@ -124,6 +124,10 @@ int amdgpu_sdma_process_ras_data_cb(struct amdgpu_device *adev,
 		struct amdgpu_iv_entry *entry)
 {
 	kgd2kfd_set_sram_ecc_flag(adev->kfd.dev);
+
+	if (amdgpu_sriov_vf(adev))
+		return AMDGPU_RAS_SUCCESS;
+
 	amdgpu_ras_reset_gpu(adev);
 
 	return AMDGPU_RAS_SUCCESS;
diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c
index d6d79e97def9..18014ed0e853 100644
--- a/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c
@@ -85,9 +85,12 @@ static int psp_v13_0_init_microcode(struct psp_context *psp)
 		err = psp_init_sos_microcode(psp, chip_name);
 		if (err)
 			return err;
-		err = psp_init_ta_microcode(&adev->psp, chip_name);
-		if (err)
-			return err;
+		/* It's not necessary to load ras ta on Guest side */
+		if (!amdgpu_sriov_vf(adev)) {
+			err = psp_init_ta_microcode(&adev->psp, chip_name);
+			if (err)
+				return err;
+		}
 		break;
 	case IP_VERSION(13, 0, 1):
 	case IP_VERSION(13, 0, 3):