EDAC/mc: Report "unknown memory" on too many DIMM labels found
authorRobert Richter <rrichter@marvell.com>
Thu, 23 Jan 2020 09:03:00 +0000 (09:03 +0000)
committerBorislav Petkov <bp@suse.de>
Mon, 17 Feb 2020 12:10:14 +0000 (13:10 +0100)
There is a limitation to report only EDAC_MAX_LABELS in e->label of
the error descriptor. This is to prevent a potential string overflow.

The current implementation falls back to "any memory" in this case and
also stops all further processing to find a unique row and channel of
the possible error location.

Reporting "any memory" is wrong as the memory controller reported an
error location for one of the layers. Instead, report "unknown memory"
and also do not break early in the loop to further check row and channel
for uniqueness.

 [ bp: Massage commit message. ]

Signed-off-by: Robert Richter <rrichter@marvell.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Acked-by: Aristeu Rozanski <aris@redhat.com>
Link: https://lkml.kernel.org/r/20200123090210.26933-7-rrichter@marvell.com
drivers/edac/edac_mc.c

index 35d2824925051d6fccfc15c1a7ac7dd15cb3850e..7a1445feb59de909de5c7eda862b83e5bb421b37 100644 (file)
@@ -1243,20 +1243,21 @@ void edac_mc_handle_error(const enum hw_event_mc_err_type type,
                 * channel/memory controller/...  may be affected.
                 * Also, don't show errors for empty DIMM slots.
                 */
-               if (!e->enable_per_layer_report || !dimm->nr_pages)
+               if (!dimm->nr_pages)
                        continue;
 
-               if (n_labels >= EDAC_MAX_LABELS) {
-                       e->enable_per_layer_report = false;
-                       break;
-               }
                n_labels++;
-               if (p != e->label) {
-                       strcpy(p, OTHER_LABEL);
-                       p += strlen(OTHER_LABEL);
+               if (n_labels > EDAC_MAX_LABELS) {
+                       p = e->label;
+                       *p = '\0';
+               } else {
+                       if (p != e->label) {
+                               strcpy(p, OTHER_LABEL);
+                               p += strlen(OTHER_LABEL);
+                       }
+                       strcpy(p, dimm->label);
+                       p += strlen(p);
                }
-               strcpy(p, dimm->label);
-               p += strlen(p);
 
                /*
                 * get csrow/channel of the DIMM, in order to allow