Re: 2.6.18 suspend regression on Intel Macs

[Pavel Machek <pavel@ucw.cz>]

Hi!

> > > So what's the plan? Should/Will the ACPI guys remove the bit-preserving
> > > change brought in with the latest ACPICA merge?
> > 
> > 
> > it sounds like a good idea to at least put the workaround back for now,
> > until a more elegant solution (maybe something can be done to make it
> > not needed anymore) is found...
> > (or until it shows it breaks other machines at which point
> > reconsideration is also needed)
> 
> The workaround hasn't been removed. It's still there,
> drivers/acpi/pci_link.c:
> 788 
> 789 /* Make sure SCI is enabled again (Apple firmware bug?) */
> 790 acpi_set_register(ACPI_BITREG_SCI_ENABLE, 1, ACPI_MTX_DO_NOT_LOCK);
> 791 
> 
> The thing is acpi_set_register doesn't permit anymore to write the SCI
> bit since the last ACPI merge. Or maybe you meant that the
> acpi_hw_register_write modifications should be reverted until a better
> solution is found?

Maybe you can just create a patch that modifies ACPI not to mask the
SCI bit? Reverting big chunk of ACPI code is likely not the right
solution.

								Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html