Power management report

Hello h4100-port,

  Port is nearing completion, and I started to whip PM and DPM into
shape. I've found another power leak - during running state though -
due to RS232 transceiver being always on. As for suspended state, it
appears that proper LCD powerdown was major win. I made careful review
of ASIC3 GPIO state during suspend, and cursory - of CPU GPIO, and it
appears to be correct. Over last 24hrs, I performed clean test of power
consumption during suspend, on a kernel with bare minimal initrd. I
started with fully charged battery, suspended device and woke it up
just few times to take metrics:

07:09 4017 84 2121 1896 1832 0
15:33 4010 97 2123 2095 1832 0
15:47 3984 98 2133 1973 1831 0
07:27 3970 97 2100 1616 1830 0

  Second column is voltage in raw units, third is current also in raw
units. One thing to note here is that current immediately after boot
from wince is lower than later when suspended and resumed in Linux.
So, something is still missing here, I guess, CPU frequency settings.

  Anyway, as you see, battery didn't discharge too much, even in terms
of raw units. Immediately after last sampling I rebooted to wince, and
battery indicator showed 85%. Nowhere like bad, that should give 5-6
days' life in suspend. And I yet have to test what wince gives me -
device was bought spring 2004, so battery is not at its best by now.

  I'll commit soon battery classdev support which will allow detailed
sampling of battery parameters like above, and everyone can join with
the hunt for the better PM. Actually, something can be done even
without Linux - exactly measuring how much juice is burnt in suspend
by wince. If anyone will want to make such test, please charge the
device fully, hard reset (important), and keep suspended exactly
24hrs, and report what battery meter will show.

  My device is in production use, so it will take some time before
I'll be able to make this test myself ;-)

-- 
Best regards,
 Paul                          mailto:pmiscml_at_gmail.com