Hi,
At least for me this problem has been cured by your changes
(Simon) for issue1240. Nice work much appreciated!
If nobody else is still having problems (Hasso ???) I think
you can mark this as fixed in 9f2212d too.
On Tue, 16 Dec 2008 23:47:55 +0100
"Simon 'corecode' Schubert" <[email protected]> wrote:
> Hasso Tepper wrote:
> > There is no solution yet, but I think that I already understand what's
> > going on. The problem is that the fix I committed int atapi-cam doesn't
> > have any effect in case of SATA controller in AHCI (or RAID) mode, so
> > switching IDE controller into IDE or legacy mode might help.
> >
> > I still have two questions ...
> >
> > * How non aligned DMA transfer reaches this level at all. Should higher
> > layers try to avoid this?
> > * Where and how requests should be padded to be aligned to 4 bytes
> > (it's required for SATA ATAPI only?)?
>
> It is an interesting question on where to put this alignment requirement.
> Sephe argued that our busdma functions already provide an alignment
> tool, so maybe it is just being handled not well enough. Apart from
> that, I think the ATA driver should take care of that and not the kernel.
>
> cheers
> simon
>
--
C:>WIN | Directable Mirror Arrays
The computer obeys and wins. | A better way to focus the sun
You lose and Bill collects. | licences available see
| http://www.sohara.org/
