From 5c39d600e315ce1a0d13c491693c0390d40c94f3 Mon Sep 17 00:00:00 2001 From: Le Ma Date: Fri, 22 Nov 2019 18:39:11 +0800 Subject: [PATCH] drm/amdgpu: clear uncorrectable parity error status bit This should be cleared during every nbif uncorrectable error cleanup work. Signed-off-by: Le Ma Reviewed-by: Hawking Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c b/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c index 9a3a65a0691c..bb701dbfd472 100644 --- a/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c +++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c @@ -482,10 +482,12 @@ static int nbio_v7_4_init_ras_err_event_athub_interrupt (struct amdgpu_device *a return 0; } +#define smnPARITY_ERROR_STATUS_UNCORR_GRP2 0x13a20030 + static void nbio_v7_4_query_ras_error_count(struct amdgpu_device *adev, void *ras_error_status) { - uint32_t global_sts, central_sts, int_eoi; + uint32_t global_sts, central_sts, int_eoi, parity_sts; uint32_t corr, fatal, non_fatal; struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status; @@ -494,6 +496,7 @@ static void nbio_v7_4_query_ras_error_count(struct amdgpu_device *adev, fatal = REG_GET_FIELD(global_sts, RAS_GLOBAL_STATUS_LO, ParityErrFatal); non_fatal = REG_GET_FIELD(global_sts, RAS_GLOBAL_STATUS_LO, ParityErrNonFatal); + parity_sts = RREG32_PCIE(smnPARITY_ERROR_STATUS_UNCORR_GRP2); if (corr) err_data->ce_count++; @@ -505,6 +508,11 @@ static void nbio_v7_4_query_ras_error_count(struct amdgpu_device *adev, /* clear error status register */ WREG32_PCIE(smnRAS_GLOBAL_STATUS_LO, global_sts); + if (fatal) + /* clear parity fatal error indication field */ + WREG32_PCIE(smnPARITY_ERROR_STATUS_UNCORR_GRP2, + parity_sts); + if (REG_GET_FIELD(central_sts, BIFL_RAS_CENTRAL_STATUS, BIFL_RasContller_Intr_Recv)) { /* clear interrupt status register */ -- 2.11.0