Oliver Neukum wrote: > Am Montag 05 Mai 2008 17:06:31 schrieb Alan Jenkins: > >> Oliver Neukum wrote: >> > > >>> I've written better support for autosuspend. It's not fully finished and >>> I've >>> got reports about deadlocks with keyboards. Do you want to test? >>> >>> Regards >>> Oliver >>> >> Yes, I'd love to test that. Let me know if I should build with lockdep, >> or any other particular debugging option. >> > > Lockdep and CONFIG_USB_DEBUG please. > It's against 2.6.26-rc1 > > Regards > Oliver > > --- > I had to enable autosuspend manually (which is a good). It doesn't implement "don't suspend when keyboard LEDs are set", but I guess you already knew that. Powertop shows it's effective at allowing my USB device tree to suspend, avoiding UHCI wakeups. I didn't get any error messages or lockdep warnings for you to look at, sorry. I can comment on the experience and what I assume is my buggy hardware though :-).
There's a noticeable delay, somewhere below a second, for it to resume (in response to a keypress). If I start typing at full speed I can loose quite a few characters. The first keypress - the one which triggers the resume - isn't be lost, but the next few are. It suggests my keyboard only buffers a single event, or that events are lost during the resume process (but not at the start). If I set the timeout to 0, I can't type at all. The keyboard LEDs stay on while I'm typing, but nothing appears on the screen. It suggests the keyboard takes a minimum amount of time to resume, and it's not being allowed to do so - though maybe it's reporting that it's resumed too early. Verdict: It's not "transparent" on my shiny new Logitech USB keyboard. However, I might still choose to use it with a timeout set at a similar length to my screensaver. I find it useful that it turns the LEDs off and on, as a visual reminder of what's going on. (So personally, I wouldn't want it to implement "don't suspend when keyboard LEDs are set"). Alan _______________________________________________ Discuss mailing list [email protected] http://mail.lesswatts.org/mailman/listinfo/discuss
