[PATCH] scsi: fix random memory corruption with scsi-mq + T10 PI

[Tony Battersby]

This fixes random memory corruption triggered when all three of the
following are true:

* scsi-mq enabled
* T10 Protection Information (DIF) enabled
* SCSI host with sg_tablesize > SCSI_MAX_SG_SEGMENTS (128)

The symptoms of this bug are unpredictable memory corruption, BUG()s,
oopses, lockups, etc., any of which may appear to be completely
unrelated to the root cause.

Cc: <stable@xxxxxxxxxxxxxxx> # 3.17.x, 3.18.x
Signed-off-by: Tony Battersby <tonyb@xxxxxxxxxxxxxxx>

---

I encountered this problem with a QLogic QLE2672 FC HBA using qla2xxx.
On my system, this would trigger BUG_ON(atomic_read(&bio->bi_remaining) <= 0)
in bio_endio(), or a general protection fault in __sg_free_table()
trying to free prot_sdb, or any number of other weird random problems.
All of this was caused by cmd->prot_sdb pointing to the wrong memory.
To see how the memory is allocated, refer to scsi_mq_setup_tags() in
scsi_lib.c.

For inclusion in 3.19, 3.18.x, and 3.17.x.

--- linux-3.18.0/drivers/scsi/scsi_lib.c.orig	2014-12-08 16:23:28.000000000 -0500
+++ linux-3.18.0/drivers/scsi/scsi_lib.c	2014-12-08 16:24:00.000000000 -0500
@@ -1829,7 +1829,9 @@ static int scsi_mq_prep_fn(struct reques
 
 	if (scsi_host_get_prot(shost)) {
 		cmd->prot_sdb = (void *)sg +
-			shost->sg_tablesize * sizeof(struct scatterlist);
+			min_t(unsigned int,
+			      shost->sg_tablesize, SCSI_MAX_SG_SEGMENTS) *
+			sizeof(struct scatterlist);
 		memset(cmd->prot_sdb, 0, sizeof(struct scsi_data_buffer));
 
 		cmd->prot_sdb->table.sgl =

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/