On Fri, 30 Jan 2009, Parag Warudkar wrote:

Excellent! This patch works - tg3 comes back and gets link after resume.

I still think the patch isn't very good. See my previous email. 

The fact that your machine works again is good, though. But before we let 
this lie, I'd _really_ like to know what was broken in the legacy PM path, 
rather than "let's leave it behind". Because a broken legacy path will end 
up biting us for other drivers, and I think the new PM path will need more 
work before it's ready for prime-time.

Are the below differences worth worrying about - especially since post 
suspend some DevID/VendorID and some capabilities seem to be changed?

That's not a devid/vendorid, it's an extended "subsystem" capability that 
we didn't save. We could try to save/restore all capabilities, but right 
now we only do the ones we care about (pcie/pcix/msi, iirc).

I do suspect we might be better off saving everything we can, rather than 
deciding piece-meal to save specific capabilities we know about.

			Linus