I'm an old skool coder (anyone remember BT Wireplay?) who is seriously miffed that all my previous mobile phones have never quite done what I want, have never quite had the right (cf: up to date) software standards that let me develop fun apps and I couldn't do anything about it except spend more cash. Not any more.. mwahaha...
Looking forward to getting a Sat nav, VoIP handset, gaming platform and GSM communication tool in one cool little package :)
Looong time no notes - ah well, life gets in the way sometimes!
My Neo is currently running a build from 27th Dec 2007, with an O2 SIM installed. I took it to the local Linux Users Group meeting recently, which was fun, and today for the first time I've got the GPS running using the oabi hack for Gllin Getting_GPS_console_output_with_gllin. I'm pleased to say that it works, unlike the Bluetooth which seems to be a dead chip  :(
Yay - it compiled :) The rebuild last night took 5 hours and seems to have produced a kernel image, a modules set (24MB!) and a rootfs (40MB). Time to back up the existing rootfs then flash my phone...
Nearly there with my first build - it failed around task 3100 with a compile fault in a library so I decided to try updating and rebuilding... this looks better as it started with task 2911 of 3329, rebuilding the gcc-cross compiler. I wonder though if it now needs to recompile everything as the compiler has been changed. Time to leave my PC on for another night!
More good news: OM 2007.2 build 20070825 seems to have sorted out the power management problem: leaving the phone 'on' with Full PM enabled overnight and it had 80% charge left the following morning :) I've noticed that on resuming from suspend mode, the system clock isn't updated from the hardware clock... I brought the phone into work again yesterday, but the GSM failed to work for me at lunchtime, and I was unable to get any logs without attaching to a USB host.
Setting up the development environment is not so good: The instructions are fine (I'm using MokoMakefile), however the sheer size of the task seems too much for my single core 2.4GHz P4 :(
After downloading for a couple of hours it started building the toolchain (gcc-cross, qemu, etc.) which seems to have gone ok, then after several more hours at 100% CPU it was part way though the glibc build and overheating regularly (my CPU heat sink was full of fluff - it's better now I vacuumed it out!), a rough calculation based on task number (254 of 3362) tells me that it was 7.5% through the build after ~6 hours, so I estimated it would take another 3-4 days to finish. I hit Ctrl-C and let the kids play some games for the rest of the weekend.
According to the Official instructions it should take ~7 hours on a 3.2 GHz Athlon 64, hmmn.
I wonder if there isn't a quicker way to bootstrap the build system? Perhaps downloading someone else's $OMDIR/build tree and updating the .conf file(s)?? I'm not familiar with OE / bitbake (yet!) - time for a question on the development list methinks...
Cool! I put OM 2007.2 build 20070825 on this morning (following the instructions here Phase_1_Software_Testing), and I have a working GSM phone :) I've made a call both ways, incoming seems to work until you hang up at the remote end, after which the phone and / or gsmd doesn't recognise the RING signal. There does seem to be some interference from the GSM radio audible in the handset, perhaps I have a poor signal here and the radio needs more power than other people?
Power management still not done it seems: I had OM 2007.2 20070818 on the phone for a couple of days, and tested leaving the phone running overnight with 'Full PM' enabled - by morning it was flat (about 8 hours).
Jobs for the bank holiday: getting a development environment set up so I can actually have a go at building my own binaries... bye bye disk space!
Thanks to the mailing lists, it seems OM 2007.2 has better power management, so that's going on the phone tonight..
Managed to join the appropriate mailing lists (thanks to whoever resurrected mailman)
Have now discovered just how long the Neo currently lasts without a USB umbilical attached - about 2-3 hours.
Much poking about on the Wiki to see where I could help with the power management, and playing with the power monitoring script available there.
Discovered I can change the backlight level by writing appropriate numbers to the /sys control file - handy when it gets dark and I don't want my eyes fried!
Neo now permanently attached to my PC, since the kids use the PC a fair amount during the day while I am at work, thus keeping my Neo alive :)
Try to join mailing lists - not working it seems, go to IRC channel instead.
Take new toy to office (BT Labs, UK) and show off :)
I finally got my grubby mitts on a Neo1973 today - w00t!
- 0 - turn it on - yay it boots, loads a kernel and stops as expected :)
- 1 - flashing the right rootfs / kernel - used 26th July files from chooseopen.com, held breath and cut/pasted dfu-util commands, then rebooted... it's ok!
- 3 - insert sacrificial SIM (UK T-Mobile) and SD card
- 2 - Test hardware:
- 2.1 - well the CPU and display are ok, as are the Flash and RAM it seems
- 2.2 - Copy mp3 file onto media card (using scp), play back with madplay - audio and SD card interface ok
- 2.3 - Fire up libgsm-tool -m shell, attempt to power up GSM and register SIM - failed :(
- 2.4 - Remove SD card, reseat SIM and try again - ok
- 2.5 - Call mobile number - enjoy moment of nostalgia as RING appears :)
- 2.6 - Answer call, no audio but that's expected - GSM chip ok
- 2.7 - Just for fun call another mobile - also connected ok
- 2.8 - Check time, 02:04hrs go to bed happy!
NB: Must test GPS if possible (looks tricky ATM)