[PATCH] drm/amdgpu: record correct sram correctable error count on arcturus

Guchun Chen guchun.chen at amd.com
Sun Mar 22 08:32:49 UTC 2020


On arcturus, only sram correctable error is stable after correctable
error injection, so reduce check conditions to record hardware status.

Signed-off-by: Guchun Chen <guchun.chen at amd.com>
---
 drivers/gpu/drm/amd/amdgpu/umc_v6_1.c | 19 +++++++++++++++----
 1 file changed, 15 insertions(+), 4 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v6_1.c b/drivers/gpu/drm/amd/amdgpu/umc_v6_1.c
index 77122a7282e7..57e8f72cc96d 100644
--- a/drivers/gpu/drm/amd/amdgpu/umc_v6_1.c
+++ b/drivers/gpu/drm/amd/amdgpu/umc_v6_1.c
@@ -138,10 +138,21 @@ static void umc_v6_1_query_correctable_error_count(struct amdgpu_device *adev,
 	/* check for SRAM correctable error
 	  MCUMC_STATUS is a 64 bit register */
 	mc_umc_status = RREG64_PCIE((mc_umc_status_addr + umc_reg_offset) * 4);
-	if (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, ErrorCodeExt) == 6 &&
-	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1 &&
-	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, CECC) == 1)
-		*error_count += 1;
+	if (REG_GET_FIELD(mc_umc_status,
+		MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1 &&
+		REG_GET_FIELD(mc_umc_status,
+		MCA_UMC_UMC0_MCUMC_STATUST0, CECC) == 1) {
+		/*
+		 * On arcturus, only sram correctable error is stable after
+		 * error injection, so bypass the check of ErrorCodeExt bit
+		 * field in MCUMC_STATUS register to record correct status.
+		 */
+		if (adev->asic_type == CHIP_ARCTURUS)
+			*error_count += 1;
+		else if (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0,
+				ErrorCodeExt) == 6)
+			*error_count += 1;
+	}
 }
 
 static void umc_v6_1_querry_uncorrectable_error_count(struct amdgpu_device *adev,
-- 
2.17.1



More information about the amd-gfx mailing list