Commit 31e837d2 authored by ganglxie's avatar ganglxie Committed by Alex Deucher
Browse files

drm/amdgpu: handle old RAS eeprom data in non-nps1 mode 



Get MCA address from PA in nps1, then convert MCA address to PA in specific nps
mode.

Signed-off-by: default avatarganglxie <ganglxie@amd.com>
Reviewed-by: default avatarTao Zhou <tao.zhou1@amd.com>
Signed-off-by: default avatarAlex Deucher <alexander.deucher@amd.com>
parent 90237b16

drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c

+14 −2
Original line number Diff line number Diff line
@@ -2886,8 +2886,20 @@ static int __amdgpu_ras_convert_rec_from_rom(struct amdgpu_device *adev, 
				bps->retired_page << AMDGPU_GPU_PAGE_SHIFT))

			return -EINVAL;

	} else {

		if (bps->address) {

			if (amdgpu_ras_mca2pa_by_idx(adev, bps, err_data))

				return -EINVAL;

		} else {

			/* for specific old eeprom data, mca address is not stored,

			 * calc it from pa

			 */

			if (amdgpu_umc_pa2mca(adev, bps->retired_page << AMDGPU_GPU_PAGE_SHIFT,

				&(bps->address), AMDGPU_NPS1_PARTITION_MODE))

				return -EINVAL;



			if (amdgpu_ras_mca2pa(adev, bps, err_data))

				return -EOPNOTSUPP;

		}

	}



	return __amdgpu_ras_restore_bad_pages(adev, err_data->err_addr,

drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c

+23 −0
Original line number Diff line number Diff line
@@ -562,3 +562,26 @@ int amdgpu_umc_mca_to_addr(struct amdgpu_device *adev, 


	return 0;

}



int amdgpu_umc_pa2mca(struct amdgpu_device *adev,

		uint64_t pa, uint64_t *mca, enum amdgpu_memory_partition nps)

{

	struct ta_ras_query_address_input addr_in;

	struct ta_ras_query_address_output addr_out;

	int ret;



	/* nps: the pa belongs to */

	addr_in.pa.pa = pa | ((uint64_t)nps << 58);

	addr_in.addr_type = TA_RAS_PA_TO_MCA;

	ret = psp_ras_query_address(&adev->psp, &addr_in, &addr_out);

	if (ret) {

		dev_warn(adev->dev, "Failed to query RAS MCA address for 0x%llx",

			pa);



		return ret;

	}



	*mca = addr_out.ma.err_addr;



	return 0;

}

drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h

+2 −0
Original line number Diff line number Diff line
@@ -189,4 +189,6 @@ int amdgpu_umc_mca_to_addr(struct amdgpu_device *adev, 
			uint64_t err_addr, uint32_t ch, uint32_t umc,

			uint32_t node, uint32_t socket,

			struct ta_ras_query_address_output *addr_out, bool dump_addr);

int amdgpu_umc_pa2mca(struct amdgpu_device *adev,

		uint64_t pa, uint64_t *mca, enum amdgpu_memory_partition nps);

#endif