[Gta04-owner] Modem off

NeilBrown neilb at suse.de
Sun Nov 17 08:37:46 CET 2013


On Sat, 16 Nov 2013 11:30:33 +0100 Andreas Kemnade <andreas at kemnade.info>
wrote:

> On Sat, 16 Nov 2013 10:18:40 +1100
> NeilBrown <neilb at suse.de> wrote:
> 
> > On Wed, 13 Nov 2013 15:45:31 +0100 Andreas Kemnade <andreas at kemnade.info>
> > wrote:
> > 
> > > Hi,
> > > 
> > > On Tue, 2013-11-12 at 09:49 +1100, NeilBrown wrote: 
> > > > On Mon, 11 Nov 2013 16:38:31 +0100 Andreas Kemnade <andreas at kemnade.info>
> > > > wrote:
> > > > 
> > > > > > 
> > > > > > I look forward to hearing of your results.
> > > > > 
> > > > > At the moment I even cannot reproduce your the lower currents when the
> > > > > module is removed or unbind is done (when the modem does not show up on
> > > > > the bus). I measured current_now directly after suspend and retried also
> > > > > with that 5-min average using charge_now.
> > > > 
> > > > What current do you measure? 
> > > 
> > > 3.7 with ehci-hcd compiled in
> > > - 32mA without modem, 2 tries
> > > doing unbind
> > > - 45mA (3x)
> > > doing bind
> > > - 54mA
> > > - 24mA
> > > - 24mA
> > > - 54mA
> > > 
> > > doing unbind
> > > 
> > > - 45mA
> > > - 45mA
> > > 
> > > doing bind
> > > - 24mA
> > > - 24mA
> > > - 47mA(?)
> > > - 24mA
> > > - 24mA
> > > - 41mA(?)
> > 
> > So if we ignore occasional high values you get about 24mA when the driver is
> > bound, and about 45mA when not bound.
> > 
> > That is the oppositte of what I get for loading and unloading, which is
> > surprising.
> And it is also surprising that the bind gives lower current than the intial situation
> when the driver is automatically bound during boot.

I've noticed that if I unbind and then bind (or rmmod and then modprobe), that
I can no longer see the modem.
I turn the modem off before hand.  unbind and rebind.  Then toggle the GPIO
line to turn the modem on again and get kernel errors suggesting that it
noticed a device but failed to talk to it properly.

The different power usage could be related to that.

Anyway I've experimented with the omah ehci compiled in to the kernel and not
a module.

Initially (driver bound at boot but modem not on)

30193 1
32351 5

so fairly high, comparable to what you saw.

After unbind:

15098 6

Which is lower than before and much lower than you got.  So I'm definitely
seeing something different to you.

After rebind:

15103 60
17241 10
18260 12
19301 11
19618 1
21497 1
23617 1
26477 1
466083 1
1479484 1

So it doesn't increase significantly.  And I see less current usage after
unbind/bind than I do after rmmod/modprobe.
That is peculiar indeed.  I'll need to try again to ensure that is repeatable.

Even better, with this kernel where omap-ehci is compiled in, if I use it in
normal mode with the GSM modem active I get:

17194 3
18260 12
19301 11
19618 1
21497 1
23617 1
26477 1

which seems to be better than before.  I'm using 10-minute intervals here so
it isn't directly comparable, but I have an app which shows me long-term
average power usage and it was showing an average around 19mA, where as when
omap-ehci was a module the long term was 21mA.

Very strange - needs repeating.

I haven't tried the 3.7-neil-plus kernel yet.  Hope to in the next day or so.

NeilBrown
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 828 bytes
Desc: not available
URL: <http://lists.goldelico.com/pipermail/gta04-owner/attachments/20131117/207e88f5/attachment.bin>


More information about the Gta04-owner mailing list