Re: Is the change to IDE probing really necessary?

[Mel Gorman <mel@csn.ul.ie>]

On Tue, Nov 25, 2008 at 12:30:02AM +0000, Alan Cox wrote:
> > > The mask is set to 0 and then the master/slave are detected based upon
> > > the presence of PCI IDE devices on the master/slave ports.
> > > 
> > 
> > This detection was not happening but then again the proper driver was not
> > compiled. The .config I used was very minimal and lacked most drivers.
> 
> The lspci shows the correct thing happened. You have PCI devices and
> didn't load the right driver. No driver, no support, no suprise.
> 

Yeah, luck seemed to have been working up until this point :/. ide-generic
has been working on the machine since at least 2.6.23 for sure and probably
longer than that which is why it had the hallmarks of a regression even though
ide-generic appears to be doing something more safe and sensible these days.

> > It is certainly the case I was not using the correct driver. Using
> > CONFIG_BLK_DEV_PIIX instead of CONFIG_IDE_GENERIC makes everything work as
> > expected without having to pass special options.
> 
> It would do - and you don't want to use pata_legacy/ide-generic for PCI
> devices. There are several with quirks, some even corruptors that the
> legacy drivers do not know how to handle.
> 

Understood.

> So I think this is a case of 'I put best steam coal in my car and it
> didn't go'

Using piix, I won't have to call in Fred Flintstone to push it with his
legs instead.

Thanks

-- 
Mel Gorman
Part-time Phd Student                          Linux Technology Center
University of Limerick                         IBM Dublin Software Lab