edac: i5000_edac critical fix panic out of bounds
Tamas Vincze [Sat, 16 Jan 2010 01:01:10 +0000 (17:01 -0800)]
EDAC MC0: INTERNAL ERROR: channel-b out of range (4 >= 4)
Kernel panic - not syncing: EDAC MC0: Uncorrected Error  (XEN) Domain 0 crashed: 'noreboot' set - not rebooting.

This happens because FERR_NF_FBD bit 28 is not updated on i5000.  Due to
that, both bits 28 and 29 may be equal to one, returning channel = 3.  As
this value is invalid, EDAC core generates the panic.

Addresses http://bugzilla.kernel.org/show_bug.cgi?id=14568

Signed-off-by: Tamas Vincze <tom@vincze.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Signed-off-by: Doug Thompson <dougthompson@xmission.com>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

drivers/edac/i5000_edac.c

index 77a9579..adc10a2 100644 (file)
@@ -577,7 +577,13 @@ static void i5000_process_nonfatal_error_info(struct mem_ctl_info *mci,
                debugf0("\tUncorrected bits= 0x%x\n", ue_errors);
 
                branch = EXTRACT_FBDCHAN_INDX(info->ferr_nf_fbd);
-               channel = branch;
+
+               /*
+                * According with i5000 datasheet, bit 28 has no significance
+                * for errors M4Err-M12Err and M17Err-M21Err, on FERR_NF_FBD
+                */
+               channel = branch & 2;
+
                bank = NREC_BANK(info->nrecmema);
                rank = NREC_RANK(info->nrecmema);
                rdwr = NREC_RDWR(info->nrecmema);