libata: zero xfer length on ATAPI data xfer IRQ is HSM violation
authorAlbert Lee <albertcc@tw.ibm.com>
Wed, 5 Dec 2007 07:43:02 +0000 (16:43 +0900)
committerJeff Garzik <jeff@garzik.org>
Wed, 23 Jan 2008 10:24:13 +0000 (05:24 -0500)
Treat zero xfer length as HSM violation.  While at it, add
unlikely()'s to ATAPI ireason and transfer length checks.

tj: Formatted patch and added unlikely()'s.

Signed-off-by: Albert Lee <albertcc@tw.ibm.com>
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
drivers/ata/libata-core.c

index 2b57547bd740478379d6ec9740191ca56240815d..c9e6bd4d0686e1c3a349ec985091231c94aa5ec6 100644 (file)
@@ -5309,12 +5309,15 @@ static void atapi_pio_bytes(struct ata_queued_cmd *qc)
        bytes = (bc_hi << 8) | bc_lo;
 
        /* shall be cleared to zero, indicating xfer of data */
-       if (ireason & (1 << 0))
+       if (unlikely(ireason & (1 << 0)))
                goto err_out;
 
        /* make sure transfer direction matches expected */
        i_write = ((ireason & (1 << 1)) == 0) ? 1 : 0;
-       if (do_write != i_write)
+       if (unlikely(do_write != i_write))
+               goto err_out;
+
+       if (unlikely(!bytes))
                goto err_out;
 
        VPRINTK("ata%u: xfering %d bytes\n", ap->print_id, bytes);