>
>I understand pci_set_power_state () doesn't wholly work on
>x86 either (doesn't restore some PCI state that can change
>over suspend), but at this point I'd be more concerned that the
>PCI drivers be able to remove the PMAC-specific PM processing
>code. It'd seem like it should be straightforward to have the
>PCI code use the PMAC_PBOOK APIs in one place, rather than have
>every driver use them.
>
>Perhaps the downside of that approach is that it couples
>to the 2.4 PCI driver framework, and PPC still seem rather
>reliant on the 2.2 framework. At some point not too far
>into the 2.4 series, I'd hope such reliance disappears
>(and not just for PPC :-).
Another problem is that the PCI driver would have to be aware of each
device specifics regarding sleep and wakeup. Also, a bunch of PowerMac
devices are not tied to the PCI at all, since they are all parts of one
or two big Apple ASIC, we use the Open Firmware device tree to probe
them, get their address, interrupts, etc...
But yes, in the case of USB, a better long term solution would probably
be to let the OHCI driver do "normal" suspend, and then have the low-
level sleep code handle the mac-spcific turn on/off stuff as part of the
other motherboard specific sleep stuffs.
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
