Update on USB PM state (2.6.15 queue)

Update on USB PM state (2.6.15 queue)

[David Brownell]

[-- Attachment #1: Type: text/plain, Size: 1949 bytes --]

I just thought I'd provide a quick update on the state of USB PM,
at least according to my testing(*)  after the patches that are now
in Greg's 2.6.15 queue(**) ... I just forwarded the patch overviews
to this list, the details are on linux-usb-devel.

  - Suspend/resume of the PCI devices through sysfs behaves again ...
    it seems to have broken badly sometime after 2.6.10 or 2.6.11

  - Likewise it behaves through swsusp "shutdown/reboot", both with
    and without devices attached.

  - I have one system on which S1 suspend/resume works, at
    least if I use PS/2 keyboard to trigger wakeup.

  - On some other systems, both S1 and S3 suspend fail; the
    system dies in ACPI during wakeup processing.

As for remote wakeup, using a USB keyboard, that still looks OK
(no regressions needed fixing) except when ACPI kicks in:

  - For "runtime PM" scenarios where individual devices and
    hubs (including root hubs) are suspended, the relevant
    parts of the device tree wake up just fine.

  - For "system PM" scenarios, and with /proc/acpi/wakeup
    tweaked to enable all wakeup events, mixed results:

        * The system where S1 works seems to run into some
          ACPI bugs; the last messages relate to overflow,
          evidently the wakeup signal is never acknowledged.

        * Those other systems don't have any better luck
          when the S1/S3 resume signal is a USB wakeup than
          any other kind of wakeup.

Compared to the last time I tested this, it's good to at least
see the USB remote wakeup signals waking the system up; those
seem to be ACPI improvements.  But it still looks like the
cases I most want to see working (S1, S3) are failing on PCs.

There are a few small USB PM updates on the way from me (call
it all cleanups), and Alan's planning some updates too.

- Dave

(*)  EHCI and OHCI only; no UHCI boxes.

(**) http://www.kernel.org/pub/linux/kernel/people/gregkh/gregkh-2.6/patches/

[-- Attachment #2: Type: text/plain, Size: 0 bytes --]

Re: Update on USB PM state (2.6.15 queue)

[Pavel Machek]

[-- Attachment #1: Type: text/plain, Size: 586 bytes --]

Hi!

>         * Those other systems don't have any better luck
>           when the S1/S3 resume signal is a USB wakeup than
>           any other kind of wakeup.
> 
> Compared to the last time I tested this, it's good to at least
> see the USB remote wakeup signals waking the system up; those
> seem to be ACPI improvements.  But it still looks like the
> cases I most want to see working (S1, S3) are failing on PCs.

Look at Doc*/power/video.txt, there's long list of working systems in it.
				Pavel
-- 
64 bytes from 195.113.31.123: icmp_seq=28 ttl=51 time=448769.1 ms         


[-- Attachment #2: Type: text/plain, Size: 0 bytes --]

Re: Update on USB PM state (2.6.15 queue)

[David Brownell]

[-- Attachment #1: Type: text/plain, Size: 869 bytes --]

> >         * Those other systems don't have any better luck
> >           when the S1/S3 resume signal is a USB wakeup than
> >           any other kind of wakeup.
> > 
> > Compared to the last time I tested this, it's good to at least
> > see the USB remote wakeup signals waking the system up; those
> > seem to be ACPI improvements.  But it still looks like the
> > cases I most want to see working (S1, S3) are failing on PCs.
>
> Look at Doc*/power/video.txt, there's long list of working systems in it.

Well; that's a list; but sadly, not what I'd call a long one.  :(

And it doesn't address the case of the system I have which handles
S1 just fine  ... except that wakeup through USB wedges (after a
few screenfuls of ACPI complaints about unhandled GPEs).  That's
convincing to me that there's an ACPI bug, one separate from any
issues with video.

- Dave



[-- Attachment #2: Type: text/plain, Size: 0 bytes --]

Re: Update on USB PM state (2.6.15 queue)

[Pavel Machek]

[-- Attachment #1: Type: text/plain, Size: 913 bytes --]

Hi!

> > >         * Those other systems don't have any better luck
> > >           when the S1/S3 resume signal is a USB wakeup than
> > >           any other kind of wakeup.
> > > 
> > > Compared to the last time I tested this, it's good to at least
> > > see the USB remote wakeup signals waking the system up; those
> > > seem to be ACPI improvements.  But it still looks like the
> > > cases I most want to see working (S1, S3) are failing on PCs.
> >
> > Look at Doc*/power/video.txt, there's long list of working systems in it.
> 
> Well; that's a list; but sadly, not what I'd call a long one.  :(
> 
> And it doesn't address the case of the system I have which handles
> S1 just fine  ... except that wakeup through USB wedges (after a

S1 is certainly different problem. I fixed the doc to be more clear on
that.
								Pavel
-- 
if you have sharp zaurus hardware you don't need... you know my address

[-- Attachment #2: Type: text/plain, Size: 0 bytes --]