Re: v3.16-rc1 regression? unexpected usb_autopm_get_interface error

* Re: v3.16-rc1 regression? unexpected usb_autopm_get_interface error
       [not found] <87egyp8f70.fsf@nemi.mork.no>
@ 2014-06-16 11:31 ` Bjørn Mork
  2014-06-16 12:04   ` Bjørn Mork
  0 siblings, 1 reply; 6+ messages in thread
From: Bjørn Mork @ 2014-06-16 11:31 UTC (permalink / raw)
  To: linux-usb; +Cc: linux-pm

[adding the linux-pm list as I suspect this is a more generic issue
affecting usb, quoting all I previously sent to linux-usb for context]

Bjørn Mork <bjorn@mork.no> writes:

> I just booted v3.16-rc1 on my laptop and ended up with an error I've
> never encountered before.  Which makes me suspect that it is related to
> changes in v3.16.  Haven't yet spent any time trying to debug it.  Just
> posting in case it is an already known problem.
>
> I got this in the log:
>
> [  268.689677] usb usb3-port4: not suspended yet
> [  268.696603] cdc_mbim 3-4:1.0: Error autopm - -16
>
> And looking at the latter driver (which I should know ;-), I see that
> this is logged if we get an error from usb_autopm_get_interface when
> attempting to open the chardev:
>
> static int wdm_open(struct inode *inode, struct file *file)
> {
> ..
>         rv = usb_autopm_get_interface(desc->intf);
>         if (rv < 0) {
>                 dev_err(&desc->intf->dev, "Error autopm - %d\n", rv);
>                 goto out;
>         }
>
>
>
> But we shouldn't really hit this, and I cannot remember ever seeing it
> before.  Looking at the device power state, I note that the
> runtime_status is 'error':
>
> bjorn@nemi:~$  grep . /sys/bus/usb/devices/3-4/power/*
> /sys/bus/usb/devices/3-4/power/active_duration:4284
> /sys/bus/usb/devices/3-4/power/async:enabled
> /sys/bus/usb/devices/3-4/power/autosuspend:2
> /sys/bus/usb/devices/3-4/power/autosuspend_delay_ms:2000
> /sys/bus/usb/devices/3-4/power/connected_duration:1523832
> /sys/bus/usb/devices/3-4/power/control:auto
> /sys/bus/usb/devices/3-4/power/level:auto
> /sys/bus/usb/devices/3-4/power/persist:1
> /sys/bus/usb/devices/3-4/power/runtime_active_kids:0
> /sys/bus/usb/devices/3-4/power/runtime_active_time:4040
> /sys/bus/usb/devices/3-4/power/runtime_enabled:enabled
> /sys/bus/usb/devices/3-4/power/runtime_status:error
> /sys/bus/usb/devices/3-4/power/runtime_suspended_time:1519548
> /sys/bus/usb/devices/3-4/power/runtime_usage:0
> /sys/bus/usb/devices/3-4/power/wakeup:disabled
>
>
> Known problem?  Or any suggestions where I start debugging this?
>
> FWIW, I've been using this device with all the latest and greatest
> changes to cdc_mbim for a few weeks already, so I'm pretty sure it isn't
> (only) those changes which trigger this.  I beleive it has to be
> something related to the usb and/or pm core.
>
> If it matters, this device is connected to an internal (mini-PCIe)
> laptop port.  The port state looks OK to me:
>
> bjorn@nemi:~$ cat /sys/bus/usb/devices/3-4/port/connect_type 
> hardwired
> bjorn@nemi:~$ grep . /sys/bus/usb/devices/3-4/port/power/*
> /sys/bus/usb/devices/3-4/port/power/async:enabled
> grep: /sys/bus/usb/devices/3-4/port/power/autosuspend_delay_ms: Input/output error
> /sys/bus/usb/devices/3-4/port/power/control:auto
> /sys/bus/usb/devices/3-4/port/power/runtime_active_kids:0
> /sys/bus/usb/devices/3-4/port/power/runtime_active_time:0
> /sys/bus/usb/devices/3-4/port/power/runtime_enabled:disabled
> /sys/bus/usb/devices/3-4/port/power/runtime_status:unsupported
> /sys/bus/usb/devices/3-4/port/power/runtime_suspended_time:0
> /sys/bus/usb/devices/3-4/port/power/runtime_usage:1

Some more experimenting reveals that this isn't limited to a single
device, or to builtin devices.  I have the exact same problem with *all*
USB devices.  Any USB device which is runtime suspended before it is
used becomes non-functional with a permanent(?) 'error' runtime_status:

nemi:/home/bjorn# grep . /sys/bus/usb/devices/?-?/power/runtime_status
/sys/bus/usb/devices/1-1/power/runtime_status:error
/sys/bus/usb/devices/1-6/power/runtime_status:error
/sys/bus/usb/devices/3-2/power/runtime_status:error
/sys/bus/usb/devices/5-1/power/runtime_status:error
/sys/bus/usb/devices/5-4/power/runtime_status:error
/sys/bus/usb/devices/7-1/power/runtime_status:error

Suspending and resuming the system resets the state temporarily:

nemi:/home/bjorn# grep . /sys/bus/usb/devices/?-?/power/runtime_status
/sys/bus/usb/devices/1-1/power/runtime_status:suspended
/sys/bus/usb/devices/1-6/power/runtime_status:suspended
/sys/bus/usb/devices/3-2/power/runtime_status:suspended
/sys/bus/usb/devices/5-1/power/runtime_status:suspended
/sys/bus/usb/devices/5-4/power/runtime_status:active
/sys/bus/usb/devices/7-1/power/runtime_status:suspended

This doesn't really help though.  Attempting to activate these devices
by opening their respective character devices etc return them to the
same error state:

nemi:/home/bjorn# cat /dev/ttyUSB0
cat: /dev/ttyUSB0: Device or resource busy
nemi:/home/bjorn# grep . /sys/bus/usb/devices/?-?/power/runtime_status
/sys/bus/usb/devices/1-1/power/runtime_status:suspended
/sys/bus/usb/devices/1-6/power/runtime_status:suspended
/sys/bus/usb/devices/3-2/power/runtime_status:suspended
/sys/bus/usb/devices/5-1/power/runtime_status:error
/sys/bus/usb/devices/5-4/power/runtime_status:suspended
/sys/bus/usb/devices/7-1/power/runtime_status:suspended
nemi:/home/bjorn# cat /dev/ttyACM0 
cat: /dev/ttyACM0: Input/output error
nemi:/home/bjorn# grep . /sys/bus/usb/devices/?-?/power/runtime_status
/sys/bus/usb/devices/1-1/power/runtime_status:suspended
/sys/bus/usb/devices/1-6/power/runtime_status:suspended
/sys/bus/usb/devices/3-2/power/runtime_status:suspended
/sys/bus/usb/devices/5-1/power/runtime_status:error
/sys/bus/usb/devices/5-4/power/runtime_status:error
/sys/bus/usb/devices/7-1/power/runtime_status:suspended

But if a device is opened (i.e. marked it "in use") *before* it is
runtime suspended for the first time, then this device will continue to
work.  Also after being runtime suspended.

So the problem is related to runtime suspend before first use. I
strongly suspect 

 aae4518b3124 PM / sleep: Mechanism to avoid resuming runtime-suspended devices unnecessarily

but I haven't been able to verify this yet as it doesn't revert
cleanly.  Will continue to look at it, but any help and/or hint is
appreciated.

Bjørn

^ permalink raw reply	[flat|nested] 6+ messages in thread