Re: regarding bug #5914 - fs corruption on SATA

[Jens Axboe <axboe@suse.de>]

On Sat, Jan 28 2006, Nicolas Mailhot wrote:
> Le vendredi 27 janvier 2006 à 20:37 +0100, Nicolas Mailhot a écrit :
> > Le vendredi 27 janvier 2006 à 10:50 +0100, Jens Axboe a écrit :
> > 
> > > Of course, silly typo! Thanks for catching that. Update patch below.
> >  
> > ...
> > 
> > a patched kernel reboots before finishing to initialize (Just before it
> > prints a line starting with SCSI - the rest is too fast for me to catch)
> > 
> > Now the kernel base is slightly different from yesterday, so the bug may
> > be in the base not the patch. I'll rebuild a new kernel with the same
> > base and yesterday's patch to check this now
> 
> I can confirm today's patch is not OK. The same baseline with
> yesterday's patch boot fine.

Is this any better?

diff --git a/drivers/scsi/libata-scsi.c b/drivers/scsi/libata-scsi.c
index cfbceb5..07b1e7c 100644
--- a/drivers/scsi/libata-scsi.c
+++ b/drivers/scsi/libata-scsi.c
@@ -1700,6 +1700,31 @@ static unsigned int ata_msense_rw_recove
 	return sizeof(def_rw_recovery_mpage);
 }
 
+/*
+ * We can turn this into a real blacklist if it's needed, for now just
+ * blacklist any Maxtor BANC1G10 revision firmware
+ */
+static int ata_dev_supports_fua(u16 *id)
+{
+	unsigned char model[41], fw[9];
+
+	if (!ata_id_has_fua(id))
+		return 0;
+
+	model[40] = '\0';
+	fw[8] = '\0';
+
+	ata_dev_id_string(id, model, ATA_ID_PROD_OFS, sizeof(model) - 1);
+	ata_dev_id_string(id, fw, ATA_ID_FW_REV_OFS, sizeof(fw) - 1);
+
+	if (strncmp(model, "Maxtor", 6))
+		return 1;
+	if (strncmp(fw, "BANC1G10", 8))
+		return 1;
+
+	return 0; /* blacklisted */
+}
+
 /**
  *	ata_scsiop_mode_sense - Simulate MODE SENSE 6, 10 commands
  *	@args: device IDENTIFY data / SCSI command of interest.
@@ -1797,7 +1822,7 @@ unsigned int ata_scsiop_mode_sense(struc
 		return 0;
 
 	dpofua = 0;
-	if (ata_id_has_fua(args->id) && dev->flags & ATA_DFLAG_LBA48 &&
+	if (ata_dev_supports_fua(args->id) && dev->flags & ATA_DFLAG_LBA48 &&
 	    (!(dev->flags & ATA_DFLAG_PIO) || dev->multi_count))
 		dpofua = 1 << 4;
 


-- 
Jens Axboe