Warner Losh wrote: >>When I updated to a recent -current, my laptop takes a very long time to >>resume (20 seconds) whereas before it took about 2 seconds. I suspect >>the PCI device probe delay capability you added triggered this. Perhaps >>the PCI resume code queries the register, gets all ones since the bus is >>not active yet, and takes the maximum delay for each device access? > > > You mean enforcing the system software minimum access time delay? At > most I'm waiting 10ms (D3->D0 transition). So you must have 2000 > devices if that results in a 20s delay. There's an implication that I > could halve that value. > > Alternatively, it could be that DELAY doesn't work quite right at this > stage of the resume, so we're sleeping a lot longer than 10ms... > > Warner I think phk's response that it's ATA timeouts may be more correct then. I'll try to do more debugging, I guess. -- NateReceived on Thu Jan 06 2005 - 20:17:35 UTC
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:38:25 UTC