From: Yazen Ghannam <yazen.ghan...@amd.com> We should move away from using AMD-specific amd_get_nb_id() to find a node ID and move toward using generic Linux methods. We can use cpu_to_node() since NUMA should be working as expected on newly released Fam17h systems.
Replace call to amd_get_nb_id() and related shifting by a call to cpu_to_node() which works as expected. Signed-off-by: Yazen Ghannam <yazen.ghan...@amd.com> --- Link: http://lkml.kernel.org/r/1486760120-60944-1-git-send-email-yazen.ghan...@amd.com v1->v2: - Just use cpu_to_node() instead of trying to make amd_get_nb_id() work. drivers/edac/mce_amd.c | 6 +----- 1 file changed, 1 insertion(+), 5 deletions(-) diff --git a/drivers/edac/mce_amd.c b/drivers/edac/mce_amd.c index ba35b7e..dbccf40 100644 --- a/drivers/edac/mce_amd.c +++ b/drivers/edac/mce_amd.c @@ -878,12 +878,8 @@ static void decode_smca_errors(struct mce *m) pr_cont("%s.\n", smca_mce_descs[bank_type].descs[xec]); } - /* - * amd_get_nb_id() returns the last level cache id. - * The last level cache on Fam17h is 1 level below the node. - */ if (bank_type == SMCA_UMC && xec == 0 && decode_dram_ecc) - decode_dram_ecc(amd_get_nb_id(m->extcpu) >> 1, m); + decode_dram_ecc(cpu_to_node(m->extcpu), m); } static inline void amd_decode_err_code(u16 ec) -- 2.7.4
