There are three flags being set by default by the PIIX driver for speeds >
PIO 1, and one not being cleared properly on fallback to PIO0.  The most
important one is the prefetch/post write control which only works for ATA
and can do bad things with ATAPI.

The patch does its best to set the flags correctly for drivers/ide.  Its
not 100% perfect but its closer than the original.  100% perfect requires
proper IORDY handling but this isn't critical (and its not right in libata
either ..  yet)

Sergei Shtylyov <sshtylyov@ru.mvista.com> said:

> +					{ 0, 0 },
> +					{ 0, 0 },
> +					{ 1, 0 },
> +					{ 2, 1 },
> +					{ 2, 3 }, };
>
>  	pio = ide_get_best_pio_mode(drive, pio, 5, NULL);

    BTW, there's quite obvious error here which leads to access outside of
timings[] if somebody passes PIO mode 5 (or autotuning code finds out that
drive supports PIO mode 5). Could have been fixed while at it... Those drives
should be rare, though...

> +		}
>  		master_data = master_data | (timings[pio][0] << 12) | (timings[pio][1] << 8);
>  	}
>  	pci_write_config_word(dev, master_port, master_data);

    Actually, there's one more serious issue with piix_tune_drive() -- it
doesn't actually set the drive's own transfer mode.

Signed-off-by: Alan Cox <alan@redhat.com>
Cc: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Cc: Bartlomiej Zolnierkiewicz <B.Zolnierkiewicz@elka.pw.edu.pl>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>