[Testing] New F11 for the XO-1 build 5-timezones

[Yioryos Asprobounitis]

Here is another strange thing, timezone info is not preserved through reboots (this is 
w/o _any_ network connection) and then hwclock and date although reported in the same timezone they have 
different time eg:
[root at localhost olpc]# hwclock
Fri 14 Aug 2009 05:14:30 PM CEST  -0.042609 seconds
[root at localhost olpc]# date
Fri Aug 14 23:14:53 CEST 2009

although the control panel timezone is in CEST+2 and the /etc/adjtime is polulated with:
[root at localhost olpc]# more /etc/adjtime 
0.000000 1250285471 0.000000
1250285471
LOCAL

Doing an ntpdate update puts th clock in the right timezone without affecting the 
hwclock. 

One strange thing I notice looking for localtime files with mc and hitting Edit by mistake (the first time :) is that both /etc/localtime or /etc/avahi/etc/localtime said "converting from Mac format" when opened for editing. Is this normal?

I do not know to what extend the timezone problem may be related to the rtcwake problems but is strange 
on its own.

--- On Fri, 8/14/09, Paul Fox <pgf at laptop.org> wrote:

> From: Paul Fox <pgf at laptop.org>
> Subject: Re: New F11 for the XO-1 build 5
> To: "Yioryos Asprobounitis" <mavrothal at yahoo.com>
> Cc: "OLPC Development" <devel at lists.laptop.org>, "OLPC Testing" <testing at lists.laptop.org>, "Fedora OLPC List" <fedora-olpc-list at redhat.com>
> Date: Friday, August 14, 2009, 8:31 AM
> yioryos wrote:
>  > 
>  > 
>  > --- On Thu, 8/13/09, Paul Fox <pgf at laptop.org>
> wrote:
>  > 
>  > > From: Paul Fox <pgf at laptop.org>
>  > > Subject: Re: New F11 for the XO-1 build 5
>  > > To: "Yioryos Asprobounitis" <mavrothal at yahoo.com>
>  > > Cc: "OLPC Development" <devel at lists.laptop.org>,
> "OLPC Testing" 
>  > <testing at lists.laptop.org>,
> "Fedora OLPC List" <fedora-olpc-list at redhat.com>
>  > > Date: Thursday, August 13, 2009, 5:50 PM
>  > > hi yioryos --
>  > > 
>  > > yioryos wrote:
>  > >  > 
>  > >  > --- On Thu, 8/13/09, James Cameron
> <quozl at laptop.org>
> wrote:
>  > >  > > 
>  > >  > > rtcwake testing has shown that
> it needs --utc after the
>  > >  > > first shutdown.
>  > >  > > Prior to then the /etc/adjtime
> file is zero.
>  > >  > > 
>  > >  > 
>  > >  > In my hands both time 'rtcwake -um
> mem -a -s 10' and 'time
>  > >  > rtcwake -m mem -a -s 10' are a hit
> and miss.  '-um' is better
>  > >  > (~50%) while without the UCT option
> is hardly 10%.  Usually
>  > > 
>  > > what exactly do you mean, by "50%" and "10%". 
> rtcwake
>  > > doesn't wake at all?  wakes too soon?
>  > 
>  > Either. 10, 50% sucess means wakeup in 11-12sec
> 
> okay.  waking too soon means you were woken by
> something
> other than the RTC.  you can "cat
> /sys/power/wakeup-source"
> to see what it was.
> 
>  > > hmm.  i think we need to step back a bit.
>  > > 
>  > > let's ignore powerd for now -- please disable it
> with:
>  > >    # initctl stop powerd
>  > > 
>  > > then, let's get rid of spurious wakeups, by
> turning off
>  > > most wakeup events:
>  > >    # echo 0 >
>  > > /sys/power/wakeup_events/all
>  > > 
>  > > the lid, the power button, and the RTC alarm
> should still
>  > > be
>  > > capable of waking the system at this point.
>  > > 
>  > > now run:
>  > >    # rtcwake --utc -m mem -s 10
>  > > 
>  > 
>  > As JC reported, under these conditions the XO wakes
> up reliably.
> 
> excellent.  so rtcwake is reliable, except for the
> timezone issue.
> 
> thanks for your testing.  olpc-powerd-10-1.fc11 should
> be
> available soon, which contains a workaround for the
> rtcwake
> issue.  (you can try it sooner by simply adding
> "--utc" to each
> of the rtcwake commandlines in the powerd script.) 
> and quozl and
> i have brought the issue to the attention of the
> util-linux-ng
> folks.  (it actually seems that the kernel might be
> treating the
> daylight savings time field from the RTC a little too
> casually.)
> 
> paul
> =---------------------
>  paul fox, pgf at laptop.org
>