On Fri, 30 Jan 2009, Linus Torvalds wrote:

If you can try the patch I just sent out, and use that as a base for 
trying to track down why the heck the legacy code doesn't work, that would 
be great. It might fix it (assuming my guess about "restore_state while in 
PCI_D3 doesn't work" was correct), but quite frankly, it's equally 
possible that it just makes things worse. But it would be really 
interesting to hear..

Sure - will do.

Your machine does seem to be interesting, in that not only does it have a 
PCI-E bridge in it (the eeepc I was playing around with at LCA does not), 
but judging by the lost config state I also suspect that it actually loses 
power during STR.

Not sure what the significance of eeepc is in this case - mine being a 
standard Intel 5400 chipset I would have thought that's the last 
place to look for interesting things!
 

Which is not at all necessarily a given - I suspect it depends on just how 
the power rails are set up on the motherboard. The fact that PCI-E bridges 
have apparently worked for others implies that your problems don't happen 
for everybody, and may relate to that issue.

Do we know this for sure that PCI-E bridges + Suspend have worked for 
others - In this thread at least I think people reported tg3 worked but not 
necessarily with a PCI-E bridge.

Parag