[PATCH 2/2] EDAC/{skx_common,i10nm}: Fix incorrect far-memory error source indicator

Qiuxu Zhuo posted 2 patches 1 month, 1 week ago
[PATCH 2/2] EDAC/{skx_common,i10nm}: Fix incorrect far-memory error source indicator
Posted by Qiuxu Zhuo 1 month, 1 week ago
The Granite Rapids CPUs with Flat2LM memory configurations may
mistakenly report near-memory errors as far-memory errors, resulting
in the invalid decoded ADXL results:

  EDAC skx: Bad imc -1

Fix this incorrect far-memory error source indicator by prefetching the
decoded far-memory controller ID, and adjust the error source indicator
to near-memory if the far-memory controller ID is invalid.

Fixes: ba987eaaabf9 ("EDAC/i10nm: Add Intel Granite Rapids server support")
Tested-by: Diego Garcia Rodriguez <diego.garcia.rodriguez@intel.com>
Signed-off-by: Qiuxu Zhuo <qiuxu.zhuo@intel.com>
---
 drivers/edac/i10nm_base.c |  1 +
 drivers/edac/skx_common.c | 23 +++++++++++++++++++++++
 drivers/edac/skx_common.h |  1 +
 3 files changed, 25 insertions(+)

diff --git a/drivers/edac/i10nm_base.c b/drivers/edac/i10nm_base.c
index e2a954de913b..51556c72a967 100644
--- a/drivers/edac/i10nm_base.c
+++ b/drivers/edac/i10nm_base.c
@@ -1036,6 +1036,7 @@ static int __init i10nm_init(void)
 		return -ENODEV;
 
 	cfg = (struct res_config *)id->driver_data;
+	skx_set_res_cfg(cfg);
 	res_cfg = cfg;
 
 	rc = skx_get_hi_lo(0x09a2, off, &tolm, &tohm);
diff --git a/drivers/edac/skx_common.c b/drivers/edac/skx_common.c
index 52b462899870..6cf17af7d911 100644
--- a/drivers/edac/skx_common.c
+++ b/drivers/edac/skx_common.c
@@ -47,6 +47,7 @@ static skx_show_retry_log_f skx_show_retry_rd_err_log;
 static u64 skx_tolm, skx_tohm;
 static LIST_HEAD(dev_edac_list);
 static bool skx_mem_cfg_2lm;
+static struct res_config *skx_res_cfg;
 
 int skx_adxl_get(void)
 {
@@ -135,6 +136,22 @@ static bool skx_adxl_decode(struct decoded_addr *res, enum error_source err_src)
 		return false;
 	}
 
+	/*
+	 * GNR with a Flat2LM memory configuration may mistakenly classify
+	 * a near-memory error(DDR5) as a far-memory error(CXL), resulting
+	 * in the incorrect selection of decoded ADXL components.
+	 * To address this, prefetch the decoded far-memory controller ID
+	 * and adjust the error source to near-memory if the far-memory
+	 * controller ID is invalid.
+	 */
+	if (skx_res_cfg && skx_res_cfg->type == GNR && err_src == ERR_SRC_2LM_FM) {
+		res->imc = (int)adxl_values[component_indices[INDEX_MEMCTRL]];
+		if (res->imc == -1) {
+			err_src = ERR_SRC_2LM_NM;
+			edac_dbg(0, "Adjust the error source to near-memory.\n");
+		}
+	}
+
 	res->socket  = (int)adxl_values[component_indices[INDEX_SOCKET]];
 	if (err_src == ERR_SRC_2LM_NM) {
 		res->imc     = (adxl_nm_bitmap & BIT_NM_MEMCTRL) ?
@@ -191,6 +208,12 @@ void skx_set_mem_cfg(bool mem_cfg_2lm)
 }
 EXPORT_SYMBOL_GPL(skx_set_mem_cfg);
 
+void skx_set_res_cfg(struct res_config *cfg)
+{
+	skx_res_cfg = cfg;
+}
+EXPORT_SYMBOL_GPL(skx_set_res_cfg);
+
 void skx_set_decode(skx_decode_f decode, skx_show_retry_log_f show_retry_log)
 {
 	driver_decode = decode;
diff --git a/drivers/edac/skx_common.h b/drivers/edac/skx_common.h
index cd47f8186831..54bba8a62f72 100644
--- a/drivers/edac/skx_common.h
+++ b/drivers/edac/skx_common.h
@@ -241,6 +241,7 @@ int skx_adxl_get(void);
 void skx_adxl_put(void);
 void skx_set_decode(skx_decode_f decode, skx_show_retry_log_f show_retry_log);
 void skx_set_mem_cfg(bool mem_cfg_2lm);
+void skx_set_res_cfg(struct res_config *cfg);
 
 int skx_get_src_id(struct skx_dev *d, int off, u8 *id);
 int skx_get_node_id(struct skx_dev *d, u8 *id);
-- 
2.17.1