> In article <Pine.LNX.4.00.9807251517560.243-100000@billabong.demon.co.uk>,
> John Hayward-Warburton <linux@billabong.demon.co.uk> wrote:
> >
> >Saw in drivers/block/Config.in that IDE DMA code was buggy and had
> >not been enabled in 2.1.111. I must confess to having seen no
> >problems on this machine (can't remember whether it's HX or TX
> >Intel chipset), Quantum Fireball SE 8.4A using Ultra-DMA on /dev/hdc
> >and an older Seagate ST32140A on /dev/hda, Intel P233MMX, Gigabyte
> >GA-586TX2 m/b, no SCSI anywhere.
> >
> >Was this intentional? Config.in hacked here to enable DMA again.
>
> It was very much intentional.
>
> People claim that the DMA problems may be due to having IDE cables of
> bad quality or too long for the spec. HOWEVER, that claim misses the
> point: even if the cable is bad it's not ok to have silent corruption,
> so there is something wrong with the driver with DMA enabled. The newer
> IDE standards are supposed to have CRC checks, and obviously they aren't
> triggering for some reason.
>
> This I disabled DMA until somebody can tell me what is up, rather than
> just suggest people try shorter cables.
>
> Linus
The configuration on which the filesystem corruption was observed is a
Tyan Tomcat III board + WD 2.2GB drive; that motherboard is using the
430HX chipset, which pre-dates Ultra-DMA and doesn't support CRC check.
Gadi
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.altern.org/andrebalsa/doc/lkml-faq.html