Re: Re: hx2490 support?

Hi Tvrtko,

On Sun, Jan 28, 2007 at 01:19:52PM +0000, Tvrtko A. Ursulin wrote:
> On Saturday 20 January 2007 04:18, Paul Sokolovsky wrote:
> > I'd vote for it to be called correct in such shape. (And just to
> > remind, I have patches to make even more rigorous DMA shutdown ;-) ).
> Could you send me those because I'd like to try them?

It is possible this may help, but I'm doubtful because Paul's patches
(if I understand them correctly) basically put haret back to the way
it was in 0.3.x series. Since you had the same issue in that series
of haret, I don't think it will help much.

> Today I had a most bizarre (that's if you consider my previous posts where I
> couldn't get 1/20 to successfully start the kernel) experience - 10 out of 10
> attempts to boot were successful to the point mmc block device access. Then
> there were typical errors of which I wrote before. Like this (excerpt):

Any chance some hardware setting was enabled or disabled since your
last set of attempts? For example, bluetooth on/off, irda on/off, pda
plugged/unplugged into usb?

I've seen this pattern before - it was usually some obscure hardware
setting difference.

> Freeing init memory: 88K
> mmcblk0: error 2 transferring data
> end_request: I/O error, dev mmcblk0, sector 162731
> Buffer I/O error on device mmcblk0p3, logical block 1
> lost page write due to I/O error on mmcblk0p3
>
> I'll be talking to kernel guys about it, but still can't shake away the
> feeling that flaky DMA shutdown could be causing it.

You may want to try booting a kernel with the MMC driver disabled and
use an initrd.

-Kevin
Received on Sun Jan 28 2007 - 11:30:40 EST