Re: [BUG] IT8212 libata driver *still* hard-freezes system on boot on 2.6.22 final.

[Rodney Gordon II <meff@spherevision.org>]

On Mon, 2007-07-09 at 13:16 +0100, Alan Cox wrote:
> > > It would be useful to know if a kernel built with no DMA support for the
> > > IT8212 behaved as that might give us a clue about what is choking. If you
> > > look in drivers/ata/pata_it821x.c:it821x_init_one you'll see a pair of
> > > lines that say
> > > 
> > > 	mwdma = 0x1f,
> > > 	udma_mask = 0x7f
> > > 
> > > 
> > > remove, rebuild and report..
> > 
> > Alright.
> > 
> > Attachment [0] is a successful boot with those lines removed. (This a
> > good thing or a bad thing? :)
> 
> Thats an interesting question in itself. It does seem to indicate the
> problem is DMA tuning or DMA related and not IRQ routing - which is good
> as of itself.
> 
> > Attachment [1] is a successful boot with the old ata subsystem using the
> > IT8212.
> 
> For the next trick set mwdma_mask = 0x07 but leave UDMA off and see what
> happens. 

Lock.

> Then try with udma_mask = 0x02 which should force the 50MHz
> clock on not the 66MHz clock just in case its clock related.

Lock. :(

> How much RAM do you have ?

1536MB, using highmem 4GB. Please refer to my dmesg attachments in the
last post for more RAM info :)

> Alan

-r