Re: RFC: Compact Flash True IDE Mode Driver

[Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>]

On 2/14/06, Kumar Gala <galak@kernel.crashing.org> wrote:
> >> Now I'm confused.  If I understand the code, what I want is for the
> >> "io_32bit" setting to have its RW field set to SETTING_READ, such
> >> that drive->no_io_32bit can NOT be changed.  Additionally, I want it
> >> set to 1 if hwif->no_io_32bit is 1.
> >
> > Yes.
> >
> >> Are you saying that at the end of probe_hwif() I should iterate over
> >> the drives for that hwif and set drive->no_io_32bit to 1 if hwif-
> >>> no_io_32bit is 1?  If so, can I do this in the last loop that
> >> already exists that iterates over the drives?
> >
> > Well, no - this loop is for tuning and is already over-complicated.
> >
> >> Will I not also want to test hwif->no_io_32bit in idedisk_setup() to
> >> ensure that it can only set driver->no_io_32bit to 0 if hwif-
> >>> no_io32bit is 0?
> >
> > No, you want to move this code to ide-probe.c because of the
> > reason given in my last mail: setting drive->no_io_32bit in ide-disk
> > is too late w.r.t. ide_add_generic_settings():
> >
> > init_gendisk()->hwif_init()->ide_add_generic_settings()
> >
> > so drive->no_io_32bit flag needs to be set earlier
> > (probe_hwif() is OK).
>
> Will drive->id->dword_io be valid by the end of probe_hwif()?

Yes.

> > And yes, this IDE stuff is complicated... :)
>
> That it is ;)
>
> - kumar