[libata] fix oops on non-DMA bmdma hardware
authorJeff Garzik <jeff@garzik.org>
Wed, 22 Mar 2006 02:31:22 +0000 (21:31 -0500)
committerJeff Garzik <jeff@garzik.org>
Wed, 22 Mar 2006 02:31:22 +0000 (21:31 -0500)
Alan noted: "bmdma may be zero but the bmdma_irq_clear function gets
called even in this case during pure PIO operation. Check we have a
bmdma before we use it."

I fixed this by adding a check for zero.  While was I there, I fixed the
non-standard indentation of the small function's code.

Signed-off-by: Jeff Garzik <jeff@garzik.org>
drivers/scsi/libata-core.c

Simple merge