Isn't Wifi supposed to be off by default?

Andy Green andy at openmoko.com
Wed Jan 14 10:37:21 CET 2009


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Somebody in the thread at some point said:
| Hi,
|   I'm using FSO-unstable with 2.6.28 (from people.openmoko.org/andy Jan 7
| Image). What I've noticed is that with the moredrivers kernel I seem
to get
| sporadic wakeups from suspend. For eg - I never got a full night's
suspend.
| And since I needed to keep auto suspend disabled since FSO suspends even
| when plugged in to a USB cable, I would generally wake with the battery
| pretty much discharged.
|   I was reading about the power consumption issues with WiFi, and so
made a
| small script to switch off WiFi containing
|    iwconfig eth0 txpower off
|    wmiconfig -i eth0 --wlan disable
|    ifconfig eth0 down (probably don't need this)

|   And last night I had an uninterrupted suspend from 22:51 to 05:53! But I
| Lost about 13% power during this time (95% to 82%). Now, that sets me
| wondering about two things :-

Sounds good.  In the packaging kernel now we have wlan as module, but
the stuff to control power properly is waiting on Werner finishing
squeezing as much email as possible out of it and doing the rfkill
stuff, at which point we will rip out the existing arrangements and
leave it alone by default.

|   1. Were most of my wakeups because of WiFi? I should have kept track of
| the reason - but I didn't. Will do so now. But, Isn't WiFi supposed to be
| off by default?
|   2. Why is the phone consuming 13% power in 7 hrs? Seems rather high if I
| look at from the perspective of simple nokia phones.

Balaji is looking at confirming which regulators are up or down during
suspend after the regulator changes.

|  Powertop says Wakeups-from-idle per second : 523.6 interval: 5.0 s
|
| Top causes for wakeups:
|   39.0% (204.0)       <interrupt> : s3c-mci

Yeah WLAN.

|   38.3% (200.6)       <interrupt> : S3C2410 Timer Tick

Hm does it mean anything the WLAN and timer part are approx the same?

|   19.2% (100.4)    ksdioirqd/mmc0 : schedule_timeout (process_timeout)

Hm.  ksdioirqd it looks again like WLAN even though mmc0 sounds more
like Glamo.

Werner, does that triplet look like there is hidden trouble in WLAN / SDIO?

|   I remember reading something about a tickless kernel patch. Has that
been
| incorporated? Any chances of incorporating CONFIG_CPU_FREQ_GOV_ONDEMAND,

Tickless is in AFAIK thanks to Andrzej, CPUFREQ is difficult on s3c2442
since there are no arrangements in the chip to meddle with CPU division
or dedicated PLL independent of peripheral clocks.  This makes for
discontiguities in peripheral action when you change the clock.  There
are patches floating around by Cesar, and Ben Dooks told he was
interested in putting them in platform code.  But it's not there yet.

| CONFIG_ACPI_BATTERY,HPET, USB autosuspend, CONFIG_NO_HZ etc in the
kernels?

We don't have HPET timer.  We don't have ACPI nor would it help, in both
cases these are PC architecture things.  USB autosuspend not sure about.
~ I think NO_HZ went away.

| Of course I understand that the kernel is still in active development
- - just
| wanted to check if these are on the list :-)

As you see your list is not completely applicable, I think most of the
trouble is from WLAN side somehow.

- -Andy
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org

iEYEARECAAYFAkltslEACgkQOjLpvpq7dMouWgCfZuHdqpFAyZy23nbMLGeLiw7o
S0UAn2HE34GO8DHTCk+U0MaJ3sc650ur
=3qO7
-----END PGP SIGNATURE-----



More information about the openmoko-kernel mailing list