ide-cs problem (was Re: any chance of 2.6.0-test*?)

[Paul Mackerras]

Alan Cox writes:

> The ide release code isnt safe in any context.

Do you mean ide_release in ide-cs.c, or ide_unregister in ide.c?

> The code is currently written on the basis that people don't mangle
> free interfaces (the free up code restores stuff which I grant is

Well, the problem is that there is no way to say that there is a hwif
present without any drives.  The low-level driver (i.e. pmac.c) has to
leave hwif->present = 0, because if you set hwif->present = 1 then the
ide probe code won't go looking for drives.  The probe code then only
sets hwif->present = 1 if it finds a drive.  If hwif->present gets
left at 0 then it is liable to be taken over for a pcmcia card.

What would be good is a way to say "I know this hwif is definitely
present, please probe for devices attached to it, but don't decide
it's not present if you don't find any devices".  I would like pmac.c
to be able to set hwif->present = 1 but still have the probe code go
looking for devices.

Regards,
Paul.