irclog2html for #htc-blueangel on 20051215

07:50.15*** join/#htc-blueangel rob_w-cgi (n=txuser@dd1010.kasserver.com)
07:50.15*** mode/#htc-blueangel [+o rob_w-cgi] by ChanServ
07:54.35rob_w-cgimorning
09:33.56lkclmorning.  bleh :)
09:35.06rob_w-cgihey
09:35.23lkcljust trying to get the attention of the treo650 linux people...
09:36.44rob_w-cgiwhy that ?
09:36.56lkclthe more the merrier.
09:37.51lkclthey have "working-ish" hardware.
09:37.51lkclability to talk to the GSM radio etc.
09:40.19rob_w-cgi? so do we !
09:41.23lkclyeh - but they're at the "we just got a kernel boot" stage so i thought they'd appreciate some links to thingy.  openembedded etc.
09:44.38lkclrob - did you put the thingy into the initrd about "mkdir -p /oldroot"?
09:46.49rob_w-cginot in the one latest on gnulinux.biz
09:48.50lkclack.
09:52.45lkclzzzz... gpe boot - package initialisation... zzz...
09:55.24rob_w-cgicrap i filled my /etc/modules and now it stucks on boot ..
09:55.44lkcloops...
09:55.45rob_w-cgiseems there is on crappy module which hunts me
09:55.54lkclmuhahahah
09:57.26rob_w-cgisilence
09:57.58rob_w-cginormally i wanna boot now , but i too scary that my new job offer wants to call me ... grgrg i am so keen to get it
09:58.03lkclque, que?  me or the ba?
09:58.14lkcloh dear!!
09:58.24lkclwhat a quandry to be in :)
09:59.31rob_w-cgiyeah i so nervous if that works out .. and i am the worst person on the planet to wait for things  i want !  *falls of his chair *
10:00.17*** join/#htc-blueangel cp15 (n=8fa4660d@www.it-schaller.de)
10:00.17*** mode/#htc-blueangel [+o cp15] by ChanServ
10:00.20rob_w-cgihmm i now have a hint on the current blueangel_battery bug but i dont think its materialisation in the code
10:00.35cp15Good morning
10:00.41rob_w-cgiservus
10:01.08cp15Is lkcl here?
10:01.10lkclmornin
10:01.11lkclyeh.
10:01.13rob_w-cgino
10:01.17lkclno.
10:01.35cp15I have checked the irc logs and have seen you have a problem with sqlite
10:01.39lkcli no exist.  i am green.  you can't see me
10:01.47lkcli am at onnne with my environment.
10:01.49lkclsorry.
10:01.51lkclyes i do.
10:01.53rob_w-cgi*falls over lkcl *
10:02.10cp15It is a problem because sqlite is built on 64 bit
10:02.10lkcli am a squished gecko who thinks he's a chameleon.
10:02.16lkcloops.
10:02.30lkcloh fraggit.
10:02.59lkclbut it's cross-compiled!!!
10:03.15cp15Take a look at btree.c:702 and find out where ptr gets defined
10:03.35cp15There is a strange #ifdef which defines ptr to long long under some circumstances
10:03.47lkclhm - inno said use the other thingy and it'd be fixed, damn :)
10:04.00cp15No, it isn't fixed
10:04.04lkclah.
10:04.27cp15He was just away for too long. I have rebuilt sqlite on my machine and packed it into the image
10:04.55rob_w-cgicp15 i suspect some apm related functions in blueangel_battery to be the bugger .. will try today evening of that works out without them
10:05.41cp15Ok, that would be fine. At least then we have a course battery level. The code to get the fine one is quite complicated
10:08.09rob_w-cgiyou mean the conversion into usable "time left " or so ?
10:09.05lkcl*grumble*.  we manually need to replace config.h
10:09.09lkclthat's doable isn't it?
10:11.17lkclyehh, it's doable.
10:13.45cp15The time left is another thing. The thing I mean is that the battery level from tsc2200 is very coarse (about 32 levels, if I remember corretly) and very nonlinear too
10:16.04cp15Maybe we should send also a note to the sqlite developers that it breaks when cross-compiling it on a 64 bit machine
10:16.29lkclthey already have notes in the code saying "deal with this manually if you're cross-compiling".
10:16.32lkclwhich is very lame.
10:17.12lkclthere's a packages/sqlite/sqlite-2.8.16/main.mk which over-rides their stuff anyway: i'm modifying that to simply say #define SQLITE_PTR_SZ 4
10:17.12rob_w-cgihmm that battery levels are also lame ?? is that because of the cells in the battery ?
10:18.37rob_w-cgiwell ,, whatever .. i am again dying because of nervousity .. *maybe some more coffee helps ?!
10:18.40lkclexcellent, that does the trick.
10:18.53lkclstay off the coffee!  go get some peppermint tea instead.
10:18.56lkclrelax, dude :)
10:19.37rob_w-cgihmm peppermint .. that maybe a try ..
10:22.15lkcli love the effects of peppermint tea: they make my arms go all floppy.
10:22.15cp15The nonlinearity is probably because of the cells... The coarse resolution is a design problem
10:22.51lkcloh, bollocks.  it's just crashed on suspend.  again.  even though i had thingy.  blueangel_suspend
10:23.28lkclbreakfast.  need breakfast.
10:23.39lkclhey cp15 - we definitely need to do a slashdot article.
10:24.01lkclbrb..
10:24.25cp15lkcl: Do you know Arthur Dent?
10:24.36lkcland linuxdevices.
10:24.39lkclnot personally, no
10:24.51lkclgood friend of my mum's though.
10:25.11cp15You sound like him: He always needs a cup of tea
10:25.26lkclah, that's because i'm addicted to caffeine.
10:25.32lkclcccccccc..c.ccccc.
10:26.35rob_w-cgi*looking for any tea in the office *
10:26.38cp15Before we make or work known, I would like to have a easy to use installation system
10:27.36cp15Because currently it is a bit too difficult and we will be overwhelmed with support questions
10:28.29cp15BTW: lkcl, are you planning to come to the CCC congress between christmas and new year? I will hold a little talk about linux on the blue angel and also registered a little space for the xanadux project
10:29.31lkclburble?  where's that?
10:29.32rob_w-cgihmm if my new job works out i maybe able to attend to it !!!
10:30.05rob_w-cgi*found green mint tea *
10:30.08lkcl*shrug* :)  don't be too optimistic: remember, they're going to be thingies - hard-core techies.
10:30.20lkclnow boil kettle, rob - carefully :)
10:30.44lkcland there's a wiki - it's not such a big hairy deal, i promise.
10:31.02rob_w-cgiwell anyway if i fall over the chair ,, at least i will not be concious anymore and dont feel nervous in that  ;-Z
10:31.07lkclplus i don't mind yapping to people.
10:31.40rob_w-cgi* dedicates lkcl as 1st level supported
10:31.41lkcldamn i just picked up the himalaya instead :)
10:31.46lkcleek!
10:31.58lkclbrb....
10:33.26rob_w-cgicp15 .. how about u ? did you think over our last private conversation ??
10:33.50rob_w-cgii might be able to also generate some little projects here and there
10:34.10rob_w-cgicrap my english is bad today *too nervous*
10:34.22lkclwhat sort of funding would your planned business need, rob, and where are you based?
10:35.05rob_w-cgihmm i dont know what sort of but i am open to anything .. based in germany -> south near munich
10:35.42lkclyou talked last week about a plan to set up something based on these devices, running linux.
10:37.40rob_w-cgiwell yeah ... that is a plan ..
10:38.43rob_w-cgibasicly i ve contacts to a american guy who has contacts and offices now in taiwan .. so anther big plan would be to get a finished OEM device into here and get gpe on it  -- selling that to fe smaller gsm providers ..
10:39.09rob_w-cgialso with that american guy  i also do some driver development for tablet pc`s .. already
10:39.45rob_w-cgiwe could off load this discussion from here  and chat in a more different channel
10:40.03lkclack.
10:40.08cp15lkcl: The CCC congress is in berlin, germany. I currently can't give you an url, because it is blocked from here. Take a look at www.ccc.de
10:40.56lkclah.  berlin.  i'm not very rich - but in theory...
10:41.18lkcli have a friend i could possibly stay with... oh, drat, they're going to be in japan.
10:41.25cp15If you can afford it, it is a must. Very interesting lectures
10:41.46lkclwtf??? ba just went blank, no reason.
10:42.35lkclit was doing the packages...
10:43.15cp15If you just need a sleeping place, maybe I can help you out. I have rented a little flat for the congress time. But you probably need a sleeping bag, since the 3 beds in it are already occupied
10:43.21lkcl:)
10:44.14lkcli really really can't afford it - it would be a big financial risk for me to take right now.
10:44.59lkclit's a long story.
10:46.25cp15Ok... Would have been nice, but I see that it is expensive
10:46.25lkcli've just had to cancel all direct debits of all bank payments until i get some money - it's _that_ serious.
10:47.18cp15But you told me you have a 75 GBP per month mobile contract? Isn't that the first place to save some money (but since you ordered the BA now, it is probably too late)
10:48.21lkcl75 GBP is nothing compared to the rest of the outgoings.
10:49.42cp15Yeah, my power bill is higher :-(
10:57.41cp15rob_w: I have to wait for next year to think further about the things we talked about
11:00.58rob_w-cgicool
11:07.03lkcloops - screen is blank!
11:07.10lkclit went into suspend, then back again.
11:07.17lkcli now have a white screen
11:07.20lkclbut i can ssh in.
11:07.32lkclafter pressing the wakeup
11:07.50cp15Hmm... I had a similar problem with my 0x6 ba. I thought it would be fixed, but it isn't
11:08.02cp15Did it already wake up with a non white screen?
11:10.49lkclyes.
11:11.14cp15Hmm... Then I have to re-check the initialisation and shutdown sequences
11:13.15lkclhmmm... the code i've compiled up is OSS not alsa... hmmm...
11:13.21lkclgomunicator, that is.
11:21.58lkclok.  what can i do.  hmmm...
11:25.14cp15What is the problem?
11:30.27lkclrunning gomunicator back to my laptop, the sound isn't working.
11:30.39lkcli was just wondering what to tackle next to see if i can get it working.
11:31.09lkclor maybe just go look at some c-code
11:33.16lkcldarn - of course: i can't log in :)
11:33.33lkclbecause i killed gpe-dm and it's waiting for a login.  drat!
11:33.44lkcldamn and it's just gone to sleep again.  waah!
11:36.36rob_w-cgitry to always get rid of the ipaq-sleep process
11:36.51rob_w-cgior configure the auto-sleep accordingly
11:38.30cp15Ok, that reminds me of another thing that is not working: Sound after sleep
11:38.42cp15I tried to fix it, but it is not yet complete
11:39.02cp15Actually I had problems to get a "go to sleep"/"wakeup" event
11:39.15cp15Try to remove blueangel_audio and load it again
11:41.06lkclcan't attach default card mixer
11:41.50lkclah, right :)
11:43.08rob_w-cgiapmd should deliver those events ( if used ) , right ?
11:43.36rob_w-cgitherefore we could manage that inside /etc/apm/
11:44.28lkclah ha.
11:44.57cp15Better invest the time into fixing blueangel_audio than createing an ugly hack in /etc/apm
11:46.47lkcl_excellent_ - sound actually works, despite getting the mixer problem.  i ran gpe-mixer and turned the volume up :)
11:46.50rob_w-cgiyes sir
11:48.45lkclokay, ipaq-sleep killed - just can't start it again.  oops.
11:50.43lkclokay.  yep.  rmmod blueangel_audio and modprobe blueangel_audio, after a suspend, fixes it.
11:50.56lkclwhat's wrong with a bloody good hack, then??? :)
11:52.50rob_w-cgiwell that should fairly easy to be done in the module
11:52.56lkclsuggestions - cp15: put suspend thingies into blueangel_audio.c?  is that the best way?
11:54.11rob_w-cgii think so ,, 2.6 Agenda tells us that the modules need to take care of their state`s on their own ... those apm sections in /etc are actually left overs from 2.4
11:54.45rob_w-cgifrom what i know afaik
11:55.10lkclack.
11:55.39lkclbtw - what framebuffer's being used - the ATI one or the vga one i hacked?
11:55.47rob_w-cgilkcl would you be interessted to join such a company or would know you where to get  funding for those things ?
11:56.28lkcli know of someone who knows a group of people who have just set up a VC fund, that's all.
11:57.18rob_w-cgihmm what is the target of those funds  ( if there are any ) ?
11:57.28rob_w-cgiIT ? mobil IT ?
11:57.43lkcli don't know.  just that my friend said "if you hear of anything"..
11:57.56lkcland your idea was "anything" :) hence my curiosity.
11:58.16lkclhmmm... there's w100fb.c suspend and resume code already...
12:02.33lkclhmmm... where's the framebuffer initialisation being called, after blueangel_lcd_hw_init?
12:05.06cp15lkcl: Yes, the blueangel_audio should re-initialize the audio after resume
12:06.14cp15The framebuffer initalisation code is called by the kernel itself
12:06.57cp15And the white screen problem is not a problem of the frame buffer, but more a problem of the display or its voltages
12:07.01lkclbefore or after the lcd init?
12:07.28lkclyeh, i remember having a bitch of a time getting the screen to work at all.
12:07.42cp15You can run ati_dump (to be found on gnulinux.biz) and see that fb is working fine. The display is just displaying nothing
12:07.48lkclah.
12:07.50lkclright.
12:07.52lkcloh dear :)
12:07.56rob_w-cgilkcl , well tell them maybe about  ready to deliver  pda smarthphones with a new os ;-) .. for smaller gsm providers .. maybe the get the same feeling
12:08.02lkcloh yes
12:08.05lkclrob_w-cgi: ack!
12:08.22cp15BTW: How did you find out which bit is related to which display voltage?
12:08.40lkclreverse-engineering and guess-work.
12:08.59lkclweeks of it.
12:09.06lkcljust kept on going until it worked.
12:09.24cp15I still know only the bits but not which voltage it controls
12:09.24rob_w-cgicongrats for that lkcl
12:09.30rob_w-cgi~hail lkcl
12:09.32aptACTION bows down to lkcl and chants, "I'M NOT WORTHY!!"
12:09.32lkclecho asdasdasdsd > /dev/fb0 doesn't do nuffink.
12:09.43lkclit wasn't rocket science!
12:10.02lkclack, cp15.
12:11.14rob_w-cgicp15 what is again the properties of the needed ringtone ?
12:12.06lkcli got em...
12:12.33lkclIt has to be 16000 Hz, stereo, format S16_LE
12:12.33lkclraw format.
12:12.34lkclyou have to copy it as /usr/share/gomunicator/R&C_theme.raw
12:13.07rob_w-cgithanks
12:13.10lkcldamn  - bitbake isn't able to do a cvs checkout from sf.net of xanadux.
12:13.16rob_w-cgiso any known tool which will bring me that from a mp3 ?
12:13.29lkclpretty much, yeh.
12:13.55rob_w-cgilike ?
12:14.40lkclyou're asking me??? :) all i did was run wavr and whistle!!
12:15.01lkclah - cvs checkout now decided it _did_ want to work...
12:16.00lkclremember i'm a computing philistine: certain areas (like messing with sound conversion) i just... don't wanna :)
12:22.58rob_w-cgihmm i am a sound newbie ..
12:23.32rob_w-cgibarely do mp3 `s n stuff .. so it looks like converting my magnum.wav wil ltake me some time
12:25.10lkcl.wav is just the same as .raw except for the first few bytes.
12:25.25lkclthat much i know.
12:25.45rob_w-cgioh , interessting
12:25.53lkclthe first few bytes indicate the format - signed, 16/8, channels.
12:25.57rob_w-cgihang on
12:26.00lkcldownload the source code to wavplayer.
12:26.17lkclit's only about 1,500 lines of code.
12:27.43rob_w-cgiok my wav is 16000 16 bit 1 channel
12:27.48lkclxxing ntl advert.
12:27.58lkclokay so if you just used that it would play at double-speed :)
12:28.19lkclntl advert just caused firefox to crash.
12:28.25rob_w-cgiu mean renaming it into .raw
12:28.28lkcli had loads of tabs open, too.
12:28.30lkclyeh.
12:29.00rob_w-cgiso how would i "fix" that double speed ?
12:32.19rob_w-cgii remmeber i had that double speed on my ipaq and i edited some sound conf file to have that fixed
12:32.22cp15rob: Create a wave file, then run sox file.wav -t raw -c 2 -r 16000 -f SB16_LE file.raw (not sure about the exact parameters, you have to check the man page)
12:32.45rob_w-cgi*searches for sox*
12:35.58rob_w-cgihmm all is fine except for the SB16_LE .. is that a "file format" or what parameter is that ?
12:39.08lkcl16-bit and the byte order is swapped
12:39.29lkclah ha - that's the command.
12:39.47lkclleast-order or something.
12:41.46rob_w-cgiok so that then should be  sox  input.wav -c 2 -r 16000 -w -x output.raw
12:41.57rob_w-cgiw for wide = 16bit and x for reverse byte order
12:42.12cp15Ah, the SB16_LE was from aplay
12:42.19cp15Try with and without -x
12:42.28rob_w-cgihmm ok
12:43.12cp15Has somebody here already tried fluidsynth?
12:43.21cp15Can we use it to play midi ringtones?
12:43.34lkcl*snort* :)
12:44.32*** join/#htc-blueangel oxi (i=oxi@153.Red-80-34-17.staticIP.rima-tde.net)
12:44.34oxihi
12:44.44lkclhello
12:46.42rob_w-cgiahm how can i name a file R&C_them `?? the & gives me problems !!
12:48.24lkclOW!  loud!
12:48.29lkclput a \ in front of the &
12:48.38lkclturned the volume up a bit high...
12:49.13oxiim trying to install a linux on my htc universal and i thought i would probably try blueangels image before i move onto porting a base linux myself... so ive got a few questions
12:49.28lkclsure.
12:49.43lkclooo.  htc universal.
12:49.48lkclfuun...
12:50.47oxiim a little worried i could lost my stable os so im wondering whats the best way to perform a complete backup.
12:51.27oxialso, i dont understand very well how could a bootloader be .exe therefore depending on windoze?
12:51.33lkclusing the standard windows sync program, activesync.
12:51.44lkcldo you remember LINLOADER.EXE?
12:51.50lkclfor the x86?
12:51.56lkclit's the same thing.
12:52.16lkclit's not a bootloader program: it's a bootstrap program.
12:52.27oxihmmm
12:52.53lkcltherefore it has nothing to do with your existing OS.
12:53.31lkclthe only thing that you lose is your data because the data is stored in RAM, which is battery-operated.
12:53.34oxianyway is it to risky to try this on my pda? does it overwrite the flash mem?
12:53.50lkclnot really, and no.
12:53.50oxithats not an issue :)
12:54.06oxicool
12:54.29lkclthe only thing is: you can't expect the htc blueangel kernel to run on a universal, it will have different hardware, different GPIO configuration, etc.
12:55.08lkclso you have several months of research ahead of you - using haret.exe (hand-held reverse-engineering tool) before you can expect it to work.
12:55.13rob_w-cgistupid name for our ringtone
12:55.15oxialso, do i need deep knowledge to port the thing? i mean, i know how to write c++ but i dont have a clue about pda hardware... i would crosscompile with oe for arm... and thats all... wouldnt know how to optimise for xscale ...
12:55.41lkclthere are stacks of ports already for xscale for you to base the work off of.
12:56.09lkclanalysing the hardware (taking it apart and photographing it) and finding out what chips are in it is one of the most important things.
12:56.29lkclthen you can look up the part numbers on the internet and find out if there are any datasheets for the chips.
12:56.29oxio_O omg
12:56.44lkcl*shrug* - that's just the way it is!
12:56.57lkclit's been done, already, for both the himalaya and the blueangel.
12:57.35oxidamn manufacturers
12:57.58lkclcp15 - damn it.  what cvs tag is downloaded from sourceforge by bitbake: i just added a couple of lines into the source code to do printks and they're not there.  argh.
12:58.07lkcloxi - yep!
12:58.19lkclwelcome to reverse-engineering :)
12:58.29oxilkcl: thank you for all the info. you really helped me out :)
12:58.51lkclwell, i dunno about that :)  do you _absolutely_ need a universal?  blueangels are nice :)
13:00.04oxiwell i already bought it... :~
13:00.10lkcloops :)
13:00.25lkclthat's what i did with my himalaya.
13:00.34rob_w-cgisf,net naturally takes some "minutes" to tag your changes and therefore you will need to wait unitl you cvs co
13:00.59lkcl18 months later i'm in the fortunate position of the work i did not being wasted because the revision 3 (blueangel) was very similar.
13:01.04lkclack, rob.  ta.  *sigh*...
13:14.05lkclokay - cp15: the w100fb_resume is done BEFORE the blueangel_lcd_hw_init.
13:14.12lkclthat's the wrong way round.
13:15.16lkclthe LCD powerup sequence AFAIK does switch-on stuff that then you need to init the ATI chip... burble, burble...
13:18.32lkclif the w100 framebuffer was a module i'd be able to test this.
13:18.37lkclgrr.
13:18.56cp15Ok, that might be the problem
13:19.18lkclhow to change the order: any clue?
13:20.13lkclor can we deliberately call an init function from inside the LCD
13:20.24lkclin blueangel_lcd_suspend
13:21.18lkclhack, hack :)
13:38.22rob_w-cgii see the mail with oyur printk messages
13:52.10lkclrob - would you be interested to do some work in this country?
14:18.46lkclcp15 - hi.  the thing is, the resume needs to be taken care of by one or the other - not both - or the ordering (registration) of suspend/resume needs to be reversed.
14:20.52rob_w-cgilkcl , well feel free to private message me and gimme some more details
14:27.42*** join/#htc-blueangel geers_ (n=DevzMan@c51008F31.inet.catch.no)
14:46.29cp15lkcl: The public cvs takes some time to get updated from developer cvs
14:46.29cp15But please don't use the cvs to insert debug code, do it in your local tree
14:46.42cp15If I remember correctly blueangel_lcd_hw_init gets called from w100fb_resume
14:46.58rob_w-cgiright you can have a local patch in the oe tree
14:47.45cp15You don't even need a patch, you simply can edit the kernel tee under work and remove the stamps after compile
14:48.13rob_w-cgihehe or so
14:52.58lkclack, sorry.
14:53.43lkclno it doesn't get called.
14:54.26lkclif tg && tg->resume tg->resume(par)
14:56.14lkclhmm, yes, you're right - but the lcd_hw_init fn doesn't show up in the right order in thingy.  dmesg. sooo....
14:57.16lkclmore debug!
15:02.43lkclcp15: yep, you're right.  err... yes.  you're right.  oh.
15:02.44lkclah.
15:03.03lkclyes.
15:03.05lkclokay.
15:03.11lkcli can just reverse the order.  ha ha :)
15:05.02lkcloh btw - i copied run.do_compile.NNNN into linux-2.6-xda as a shellscript, and execute that - it does a quicker job.
15:05.59lkclbitbake is _so_ half-baked...
15:06.10lkcli so can't be bothered to pander to it.
15:08.33lkclhmmm....
15:12.57rob_w-cgiseems like loading blueangel_pcmcia these days locks it up
15:14.17lkclthat's better than the himalaya - there ain't no driver!
15:18.49lkclhmmm....
15:26.03rob_w-cgiok i now have a handmade /etc/modules .. for what ever reason update-modules falls on my image here -- maybe as i just copied over my existing image at one point . .dont wanna loose my apps all the time ;-)
15:33.12lkclyou need 2 blueangels, don't you ;)
15:34.11rob_w-cgime ?
15:34.28lkclyes you!
15:34.39lkclor two SD cards.
15:34.40rob_w-cgiwell ,, hmm
15:35.52rob_w-cgiyeah 2 SD are to be fine .. and i need to raise some money then i do that
15:48.32lkclhmmm...
15:48.57lkclgiving serious consideration to compiling w100fb as a module...
15:53.09rob_w-cgiok so i now called my ba and started vibrating .. but no sound .. but i see on the console mutliple entry`s saying ring_timeout
15:53.19rob_w-cgii used the reversed wav
15:53.29lkclok run the sound mixer
15:53.35lkclalso modprobe snd-mixer-oss
15:53.40lkcland modprobe snd-pcm-oss
15:53.46lkclyeh i know...
15:53.56lkclrun gpe-mixer
15:54.03lkclto check the volume and mic levels.
15:54.47rob_w-cgiok all full level and retry
15:55.33rob_w-cgiCOOL !  works !!! reverse
15:56.43lkclhappy now? :)
15:56.50rob_w-cgii will put the sox command on the wiki
15:57.00lkclcp15 - this is what the "himalaya" initialisation of the lcd looked like:
15:57.01lkcl+               ipaq_asic3_set_gpio_b(&h3900_asic3.dev, 0x0008, 0);
15:57.01lkcl+               ipaq_asic3_set_gpio_b(&h3900_asic3.dev, 0x0080, 0);
15:57.01lkcl+               ipaq_asic3_set_gpio_b(&h3900_asic3.dev, 0x0100, 0);
15:57.02lkcl+               ipaq_asic3_set_gpio_b(&h3900_asic3.dev, 0x0800, 0);
15:57.02lkcl+
15:57.03lkcl+               mdelay(500);
15:57.05lkcl+
15:57.07lkcl+               ipaq_asic3_set_gpio_b(&h3900_asic3.dev, 0x0080, 0x0080);
15:57.09lkcl+               mdelay(17);
15:57.11lkcl+
15:57.13lkcl+               ipaq_asic3_set_gpio_b(&h3900_asic3.dev, 0x0100, 0x0100);
15:57.15lkcl+               himalaya_init_ATI();
15:57.17lkcl+               mdelay(30);
15:57.19lkcl+
15:57.21lkcl+               ipaq_asic3_set_gpio_b(&h3900_asic3.dev, 0x0800, 0x0800);
15:57.23lkcl+               mdelay(30);
15:57.25lkcl+
15:57.27lkcl+               ipaq_asic3_set_gpio_b(&h3900_asic3.dev, 0x0008, 0x0008);
15:57.29lkcl+               mdelay(30);
15:57.53rob_w-cgiok i am off to go home now
15:57.54lkclit wasn't _quite_ just "switch it on and pray"
15:57.59lkclk rob.
15:58.10lkcloops he's gone...
16:01.04rob_w-cgibbl !
16:01.08lkclk rob
16:02.17*** part/#htc-blueangel rob_w-cgi (n=txuser@dd1010.kasserver.com)
16:29.47*** join/#htc-blueangel oho_ (n=oho@calw-d9b91757.pool.mediaWays.net)
17:00.44*** join/#htc-blueangel cp15 (i=vqppip@p549A0CEE.dip0.t-ipconnect.de)
17:00.44*** mode/#htc-blueangel [+o cp15] by ChanServ
17:00.56cp15re
17:04.25lkclheya cp15.
17:04.46lkcldo you know who did the original "init" code for the blueangel lcd?
17:07.52cp15I did it
17:09.07lkclok.  cool.  the himalaya one does 4 gpio lines all off, wait 500ms, switch one on, wait 17ms, switch another one one, do the ATI initialisation, wait 30 ms, switch another one on, wait 30 ms, switch the other one on.
17:09.43lkclwhat you're doing is switching them all on, with ms delays of 0.6, 5, and 5, _then_ doing the ATI initialisation.
17:10.23lkclso - the key question is - really: where did the initialisation come from?
17:10.44cp15The current code: Reverse engineering of the windows driver
17:10.46lkclcos i had to do about a week's worth of reverse-engineering of the wince dlls to get the himalaya initialisation sequence
17:10.54lkclah, okay.
17:10.55lkcl*sigh* :)
17:11.02cp15The old code: Try and error
17:11.29lkclokay.
17:11.48cp15But I think at least part of the initalisation have to be done before ATI, because one of the bits controls ATI power
17:11.57cp15At least I think
17:12.13lkclyes.  that's the way it is with the himalaya.
17:12.39lkclpowerup a couple of lines, wait, do the ATI init, then power the remaining lines.
17:13.01lkcldo you know which ones? :)
17:13.40lkcldo you have some pseudo-code from the windows driver?
17:14.08lkclor, can you send me a dump so i can take a look at it, if you're happy to walk me through it?
17:16.55lkclooo, fun.
17:17.08cp15It is not that big
17:17.17cp15Explanation of the asic3 gpios can be found at http://wiki.xda-developers.com/index.php?pagename=BlueangelASIC3
17:17.43lkclack.
17:18.47lkclok... what am i looking at, here..
17:19.15lkclwhat function am i looking for - or what line number?
17:19.42lkclokay.
17:19.47lkcldispaly_type_toppoly
17:22.34cp15It is a commented disassembly of ahioem.dll which is used by ace_ddi.dll (ATI driver)
17:22.46cp15Don't ask me when which function is called, I don't know
17:22.53lkcl:)
17:25.11lkclokay, after this function display_type_toppoly gets called, there's gpuib_08 switched off, a 10ms delay, gpiob_10 on, a 1ms delay, then gpio20_on.
17:26.25*** join/#htc-blueangel oho_ (n=oho@calw-d9b91749.pool.mediaWays.net)
17:28.22lkcldamn - what's this? sending spi data???
17:30.02oho_Hello
17:31.10oho_cp15: I was and am still really bussy, but should have more time from the weekend onwards
17:33.06oho_I am off again, as I haven't got a permanent internet connection at the moment
17:33.09oho_see you
17:33.50lkclokay.  this is gonna take a while :)
17:34.19lkclthe question is: do i really want to spend that much time, again?  *sigh*...
17:35.51lkclthe vga camera: it's an i2c device?
17:59.08cp15Don't know for what the spi data is. But it is only used on toppoly (0x6) devices
18:00.05cp15The camera is probably i2c with connections to SPI for jpeg image transfer and ATI for moving images
18:17.04*** join/#htc-blueangel apt (i=ibot@pdpc/supporter/active/TimRiker/bot/apt)
18:17.04*** topic/#htc-blueangel is HTC BlueAngel is still under development ! Feel free to ask and check also -> http://www.handhelds.org/moin/moin.cgi/BlueAngel | http://wiki.xda-developers.com/index.php?pagename=BlueangelResearch | http://gnulinux.biz/files/blueangel/sd/linux/ channel chat logs at http://ibot.rikers.org/%23htc-blueangel/ * http://www.userfriendly.org/ *
18:27.16*** join/#htc-blueangel rob_w (n=bob@p85.212.149.228.tisdip.tiscali.de)
18:27.16*** mode/#htc-blueangel [+o rob_w] by ChanServ
18:42.00*** join/#htc-blueangel oxi (n=oxi@212.106.207.206)
18:42.02oxihi
18:44.45cp15Hi oxi
19:07.51rob_wcp15, are you fine to get the GPL into blueangel_kp ?
19:08.38cp15Sure, I will fix it with the next commit
19:09.09rob_wlet me do it !
19:10.12cp15Ok, go ahead
19:18.10oxiI tried your distro on my htc universal but unfortunately it doesnt boot up. Where should i get started for porting a very basic kernel for my pda? got oe already installed and have successfully crosscompiled some packages for blueangel... but i dont know what to do to start porting for my pda. Some guy in this channel told me i should reverse engineer... but how?
19:19.44oxii thought booting off your distro would be a good start point as the processor is very near to mine
19:20.07cp15This can't work since the display is totally different
19:20.37cp15First: Does your device have any external serial connector? Do you have a cable for it?
19:20.44oxiyes i do
19:20.47oxiusb
19:20.59oxithanks btw
19:21.05cp15Not usb, rs232
19:21.33oxiuhh i dont think so
19:21.49cp15Have you already registered an arm machine type?
19:22.04oxino...
19:22.24oxiwhat do you mean by register
19:24.30oxii feel i lack basic knowledge about this... but since i know how to write c++ i was hoping i could maybe port a basic kernel for my device
19:25.20cp15Ok, cr2 has already done it
19:25.27cp15Your machine number is 855
19:26.10cp15Do you have some time next weekend?
19:26.53oxiyes
19:26.55oxiwhy?
19:27.13oxiwhat is that number? i dont get it
19:27.18cp15Then maybe we can put together a at least booting kernel
19:27.27cp15The number is the machine id
19:27.30oxioh that would be wonderful :))
19:28.04cp15I will try to make it boot on my BA first without any special hardware references, so I hope it will work for you too
19:28.20oxithank you :)
19:28.35cp15The machine id is used by the kernel to discover on which machine it is running
19:29.09oxioh ic
19:30.51cp15I hope I can give you a booting kernel and a few tips, but for the rest you are on your own
19:31.43cp15But try to contact cr2 from xanadux-developers.com, he knows the universal very well
19:32.02oxithanthank you :)
19:32.04oxiill do
19:32.16rob_w-1 Invalid module format <- what is that ?
19:34.04rob_woh now i see wrong gcc i use by handcompiles
19:36.46rob_wseems like i need to update my toolchain to gcc 3.3.4
19:49.41lkcloxi - hi.  machine number is a #define in the source code of the kernel - every new bit of hardware needs one, and it's used to enable different features at compile-time and at runtime.  for example, two different Compaq IPAQs may use the same wireless device but they are at different memory locations in the two devices: you'd use the machine_id() function to tell you what address.
19:50.20lkclhttp://wiki.xda-developers.com/index.php?pagename=HTC_Universal
19:50.44lkclhttp://wiki.xda-developers.com/index.php?pagename=UniversalResearch
19:51.30lkclhttp://wiki.xda-developers.com/index.php?pagename=Universal_Technical
19:51.38lkclso - actually, quite a lot has been done already?
19:51.41lkcl!
19:51.57rob_wuse the source , luke    right lkcl
19:54.45lkcloh dear :)
19:55.06lkclhmmm  - it looks like it uses the PXA270 LCD.  wauugh!
19:56.21lkclooo - and it does WCDMA (3G?)
19:58.14cp15The LCD is compared to the BA much worse: Shared memory concept and no acceleration
19:58.24cp15And this with an increased resolution
20:00.01*** join/#htc-blueangel _cronos_ (n=cronos@ron34-1-82-224-149-102.fbx.proxad.net)
20:11.53lkclheyy, that's okay - if it's got no camera, and not using that ATI chip, it's no big deal.  the thing about the ATI chip is i heard it's got MPEG encode/decode built-in.
20:15.38cp15It got even two cameras
20:16.40cp15Yes, the ATI has a dct accellerator. Not really required on 240x320, but quite useful on 640x480
20:22.10oxihey lkcl, sorry i was afk, thanks for the info
20:54.34cp15Good news everyone... I have fixed blueangel_battery so you should now get your battery charge status
20:55.31rob_wohhhh i was that close ..
20:55.32lkclexcellent!
20:55.40rob_wdcc dcc
20:55.48cp15It's in the cvs
20:56.16cp15The trick was that the adc has to be stopped after the conversion is finished
20:57.13rob_whmm let me see
21:01.32rob_wi hate those cvs emails ... as i dont know what the ! stands for  !!
21:02.40rob_woh oh my lcd flickers ?!?!?!
21:03.17cp15Quick, turn it off
21:03.28rob_whmm
21:03.46rob_wsystem is stuck .. no ssh !
21:04.25rob_wthat didnt look nice ..
21:06.15cp15Was the screen blanker active?
21:07.30rob_wno it was during gomunicator was up and the message and number-enter fields where the most flickering
21:08.19cp15Maybe the ati got out of sync
21:09.34rob_wcp15, why is blueangel_asic.h in "the attic" ?
21:10.13rob_wor , what in gods name is "a attic in the kernel source tree " ?!?
21:11.13cp15Attic=Dachboden
21:11.24cp15This is where you store things no longer needed
21:12.09rob_wso far i also got .. aha ,,so then this header will soon be obsolete or "should be made obsolete soon " ?
21:12.58cp15No, I accidentally imported it with a clean kernel source
21:13.16cp15And so I removed it later and then imported it again
21:13.27rob_wahm
21:14.21rob_whmm can i use my account for sf.net in the kernel_cvs.bb
21:15.05cp15Do you use key authentication?
21:15.52rob_wno
21:16.27cp15Then maybe you get a problem with entering your password, but just try it
21:23.21rob_wworks !!
21:27.26lkclalso, if a file is added only in a branch, not in the mainline, it gets put into the attic not into the main directory.
21:28.51cp15So the attic is not only for things you no longer need, but also for things which you might need later :-)
21:30.10lkclyepp
21:30.10lkclso.
21:30.18lkclwhat we gonna do about this frikkin lcd thing?
21:30.55rob_wprevent the software to lock up
21:31.12cp15Have you found anything in the asm?
21:31.30rob_wunfortunatley i had no chance to capture logs
21:31.32lkcli started looking at it, and it reminded me of the time i hurt my brain.
21:31.48lkclhow about creating an NFS-booting kernel?
21:32.16cp15What would it have for advantages?
21:32.24lkclone which doesn't need a framebuffer device - so it can be a module?
21:32.30rob_wi could direct the logs onto the sd , right ?
21:32.34lkclthen it's possible to load/unload
21:32.39lkclyes, that too, rob.
21:32.50cp15Does the framebuffer depend on nfs?
21:33.10lkclif you set up an NFS-booting kernel, you can set up a thingy - root filesystem - which has log files stored there, etc.
21:33.21lkclno, nfs root mount won't depend on framebuffer.
21:34.23cp15Ok, I see... Feel free to add it to the initrd. But I think nfs is currently neither compiled in nor as a module
21:34.37rob_wwell i dont have nfs everywhere
21:34.52cp15Ok, nfs is compiled as a module
21:34.55rob_wi used nfs ?!?
21:35.13rob_wright
22:20.55lkclyou have to specifically select ROOT_NFS option and then you can specify root=nfs:/192.168.0.223/root on the kernel boot line.
23:05.41cp15Will go to bed now
23:05.43cp15Good night

Generated by irclog2html.pl by Jeff Waugh - find it at freshmeat.net! Modified by Tim Riker to work with blootbot logs, split per channel, etc.