On Tuesday 28 September 2004 1:56 pm, Alan Stern wrote: > On Tue, 28 Sep 2004, David Brownell wrote:
> > But if there was no power state transition, hence no resume(), > > then a device losing power means a major hardware problem. > > This is getting a little metaphysical... What I meant was that when > resuming from an idle state, a driver has to expect that the device may > have lost power even if it has no recollection of going through a > power-off state transition. Previous discussions in these threads have used resume() as a power state transition, and also talked about "idle" vs "active" as completely orthogonal to power state transitions. There'd be no way for such a thing to happen -- unless "idle" isn't a separate driver state (i.e. orthogonal to device power state). I certainly agree that power state transitions can sometimes involve power loss. Like PCI D3hot morphing to D3cold, But not like USB suspend; by definition (see the USB spec), if power is lost that's just like any other disconnect. So, not metaphysical at all! > > > What about Oliver's objection that support for wakeup depends on the > > > power level? Like he said, it's easier just to try turning > > > on wakeup and > > > see if you get a "not-supported" error. > > > > That's why I specified one bit controlling whether or not > > the device should try to enable wakeup when it changes > > into a suspend state. That copes with PCI restrictions, > > but doesn't try to export them anywhere else (like to USB). > > Well, you actually specified "two bits in the driver model, one saying > if the hardware supports wakeup and the other defining the policy." Just > the second bit should be sufficient, considering that the meaning of the > first bit is not well defined (it depends on the particular suspend > state). Most devices don't support wakeup signaling in any state at all, so it's pointless (and confusing, and misleading) to pretend those devices could ever be wakeup devices. That first bit is well defined: hardware supports wakeup. Even for PCI there's not much confusion ... as a rule, devices that support wake up do it from all the PCI power states they support. The patch is simple enough, I'll send it around when I've had a chance to test it. First, gotta make sure this one hardware bug gets worked around by a per-fifo watchdog timer... then I'll have a few spare moments! - Dave ------------------------------------------------------- This SF.net email is sponsored by: IT Product Guide on ITManagersJournal Use IT products in your business? Tell us what you think of them. Give us Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more http://productguide.itmanagersjournal.com/guidepromo.tmpl _______________________________________________ [EMAIL PROTECTED] To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
