RE: [jornada] sound debugged

Just FYI:
This seems to be a good suggestion for the ones not being able to get their
hands on a Flashboard.
I'm doing some battery-uptime testing for myself at the moment, so i can see
how long my Jornada can playback MP3's.
I'm not done yet, but by downclocking the CPU only and turning the display
off I'm getting definitely a lot more playback-time (something like 30% I
think).

Personally I don't think it'll be a problem to catch the press of the
suspend-button, it's most probably controlled by CE normally.

Kind regards,
Simon

-----Original Message-----
From: jornada-bounces_at_handhelds.org
[mailto:jornada-bounces_at_handhelds.org]On Behalf Of Colin Sauze
Sent: dinsdag 15 februari 2005 1:47
To: jornada_at_handhelds.org
Subject: Re: [jornada] sound debugged

What i'm suggesting is that instead of actually attempting to perform a
real suspend/resume operation that pressing the suspend button just
turns off the screen, PCMCIA and sends a SIGSTOP to all processes, a
process is then launched to check for the press of the resume button at
which point the screen and PCMCIA are turned back on and all processes
resumed. I presume its possible to capture this keypress and avoid the
normal suspend routine of the SA1110? If not a different key sequence
could initiate this.

I found this post from the NetBSD jornada 720 port -
http://mail-index.netbsd.org/port-hpcarm/2002/09/16/0004.html

    In it they suggest that you can run for 20 hours without the screen
    on. To be able to maintain a 20 hour uptime and then have to
    shutdown linux or plugin to the mains would probably be acceptable
    for most situations.

I also found this document detailing power management in the SA1110 with
regards to the Sharp Zaurus -
http://people.ac.upc.es/mpericas/zaurus_project-final.pdf

    It mentions (page 11/12) that there is a CPU clock speed scaling
    feature in the SA1110/Linux 2.6. Perhaps this could be used in
    addition to powering off the screen, to basically grind the CPU to a
    crawl and cut power consumption to a fraction of its usual amount.
    Obviously this won't be as good as a real suspend, but as there
    doesn't seem to be anyway of getting this to work without using a
    flashboard and nobody seems to know where to get a flashboard from
    it would seem like a reasonable alternative.

I will have a go at trying to mess with this if/when I get some free
time, but this isn't likely to be until April or more realistically June.

John Ankcorn wrote:

>Colin,
>
>I believe that the problem not precisely with the suspend,
>but indeed with the resume. Please note that when the SA-1110
>comes out of sleep, it jumps to a known vector address, which
>on the Wince version points to the wince 'restore' code. This
>code was written and shipped long before a linux port was
>comtemplated and knows nothing about starting linux back up.
>
>With the flashboard, the restore sequence is almost the same
>as it is for the iPaq and the Jornada 56x, so this is fairly
>'easy' to get working. Lorant proposed a reasonable (although
>technically somewhat difficult) way to get the same
>behaviour when running on wince.
>
>As always, it is just a small matter of programming....
>
>Thanks
>jca
>
>
>On Mon, 14 Feb 2005, Colin Sauze wrote:
>
>
>
>>So basically I can play sound fine until I try to suspend without a
>>flashboard.
>>
>>In some of the 2.4 kernels somebody implemented a hack to turn off the
>>display, it would be nice if the same option where available in 2.6
>>kernels for non-flashboard users (which I presume are the
>>majority??????????).
>>
>>I'm wondering if it isn't possible to further improve the battery life
>>by having some kind of suspend script which just sends a SIGSTOP to most
>>of the processes (e.g. everything but the kernel and something to allow
>>everything else to resume). Then upon resume a SIGCONT is sent to all
>>the processes to bring them back to life again.
>>
>>I don't know how much effect this would have, but i'm sure battery
>>consumption is lower with no processes running as the kernel should
>>issuing lots of HALT instructions during the idle time, so the CPU will
>>draw less power.
>>
>>Is it possible to tell the PCMCIA controllers to sleep and resume
>>without having to power off the rest of the system?
>>
>>If I can make the jornada last the length of my working day on a single
>>battery charge then it makes it pretty much as usable as CE.
>>
>>
>>alexgrosy_at_t-online.de wrote:
>>
>>
>>
>>>Hi,
>>>
>>>of course, the sound chip will also work without a flash board. It's just
>>>suspend/resume that is dependant on one of these boards (as a bootloader
is
>>>required to resume device). When the device is suspended the CPU is
>>>completely reset and without a bootloader wince would just start booting
when
>>>trying to resume without a bootloader.
>>>
>>>kind regards,
>>>alex
>>>
>>>On Sunday 13 February 2005 17:14, Colin Sauze wrote:
>>>
>>>
>>>
>>>
>>>>alexgrosy_at_t-online.de wrote:
>>>>
>>>>
>>>>
>>>>
>>>>>Hi,
>>>>>
>>>>>I have now debugged the sound on the Jornada 720 (kernel 2.6). It is
now a
>>>>>fully featured linux device if you have a flash board (sorry guys, I
can't
>>>>>get you any of them)
>>>>>
>>>>>
>>>>>
>>>>>
>>>>Why does it only work with a flashboard? I don't see how the flash board
>>>>alters your ability to address the sound chip?
>>>>
>>>>
>>>>_______________________________________________
>>>>jornada mailing list
>>>>jornada_at_handhelds.org
>>>>https://handhelds.org/mailman/listinfo/jornada
>>>>
>>>>
>>>>
>>>>
>>>_______________________________________________
>>>jornada mailing list
>>>jornada_at_handhelds.org
>>>https://handhelds.org/mailman/listinfo/jornada
>>>
>>>
>>>
>>>
>>>
>>_______________________________________________
>>jornada mailing list
>>jornada_at_handhelds.org
>>https://handhelds.org/mailman/listinfo/jornada
>>
>>
>>
>
>
>
>

_______________________________________________
jornada mailing list
jornada_at_handhelds.org
https://handhelds.org/mailman/listinfo/jornada
Received on Wed Feb 16 2005 - 16:26:02 EST