| 00:00.01 | cr2 | platform_asic3=&blueangel_asic3; |
| 00:00.11 | cr2 | replace these to himalaya. |
| 00:00.39 | cr2 | #include <asm/arch/blueangel-gpio.h> |
| 00:01.01 | cr2 | #include <asm/arch/hx4700-asic.h> |
| 00:01.11 | lkcl | you want me to just assume it's a blueangel_asic3, yes? |
| 00:01.13 | lkcl | reckon? |
| 00:01.20 | cr2 | here is the SD base register. |
| 00:01.37 | cr2 | not only that. |
| 00:02.05 | cr2 | there is also a bus shift, check if cp15 ifdef'd it correctly. |
| 00:02.20 | cr2 | the main thing is the correct SD base register. |
| 00:03.02 | cr2 | hx4700 has the same value as blueangle |
| 00:03.21 | cr2 | himalayas is in xda-dev wiki. |
| 00:03.46 | cr2 | asic5 is for the backpack. imho irrelevant. |
| 00:04.04 | lkcl | TC Himalaya normal 0x0d800000 0x0c800000 |
| 00:04.13 | lkcl | GPIO_BASE = 0x0d80000 |
| 00:04.19 | lkcl | SDIO_BASE = 0x0c800000 |
| 00:05.32 | cr2 | yes. |
| 00:07.17 | cr2 | i wish this file will be portable. we need it for the universal too. |
| 00:11.42 | cr2 | iclem: gpio 11 is not an interrupt. |
| 00:12.04 | cr2 | gpio10 looks better. |
| 00:12.36 | iclem | wow i'm lost ;) |
| 00:13.32 | iclem | where can I put correct himalayan values without breaking blueangels ones ? |
| 00:14.39 | cr2 | himalaya-asic.h ? |
| 00:15.03 | cr2 | gpio10 is the only unassigned interrupt. |
| 00:15.12 | cr2 | do an #ifdef for now. |
| 00:15.23 | cr2 | for portability. |
| 00:16.11 | iclem | I've got this line : #define GPIO_NR_HIMALAYA_SD_IRQ_N (11) |
| 00:17.04 | lkcl | in blueangel-gpio.h |
| 00:17.11 | iclem | should I replace 11 by 10 ? |
| 00:17.27 | lkcl | yep. |
| 00:17.39 | iclem | ok ;) it's not that difficult |
| 00:22.50 | iclem | Warning: "himalaya_asic3" [drivers/mmc/asic3_mmc.ko] undefined! |
| 00:23.50 | cr2 | hack alert. |
| 00:24.03 | cr2 | it is h3900_asic3 for himalaya. |
| 00:24.26 | cr2 | it is better to replace this name. |
| 00:25.25 | cr2 | in all himalaya*.c files. later. |
| 00:38.46 | iclem | what value must I replace for GPIO_BASE & SDIO_BASe ? |
| 00:38.53 | iclem | is it : #define HX4700_ASIC3_PHYS PXA_CS3_PHYS |
| 00:39.03 | lkcl | <lkcl> GPIO_BASE = 0x0d80000 |
| 00:39.04 | lkcl | <lkcl> SDIO_BASE = 0x0c800000 |
| 00:39.29 | lkcl | actually it's 0x0d80 0000 |
| 00:39.30 | lkcl | and 0x0c80 0000 |
| 00:39.42 | lkcl | i missed out a 0 on the end of GPIO_BASE |
| 00:41.13 | iclem | yes thanks for the values but I don't know what to (re)define with these |
| 00:42.00 | lkcl | lemmeavvalook... it's on the xda-developers wiki... |
| 00:42.34 | lkcl | i just saw it 2 hours ago |
| 00:43.27 | cr2 | damn. tell me about portability ;) |
| 00:43.31 | cr2 | <PROTECTED> |
| 00:43.43 | iclem | ok |
| 00:43.54 | lkcl | hmmm.... |
| 00:44.10 | lkcl | well, portability is a good idea :) |
| 00:44.55 | cr2 | then this address should be in *_asic3 struct. |
| 00:45.56 | lkcl | damn what's happened to all this code, cos the asic3 stuff has all warped since 2.6.3 kernels... |
| 00:46.22 | cr2 | it's more political. |
| 00:46.41 | lkcl | hmmmm |
| 00:47.04 | cr2 | the asic3 sd works, because the hx4700 made it work. recently. |
| 00:47.20 | cr2 | the sdio is a real can of worms. |
| 00:47.51 | iclem | => ioremap(SDIO_BASE , ... |
| 00:48.04 | iclem | it's ok ? |
| 00:48.19 | cr2 | yes. as a temporary solution. |
| 00:48.24 | iclem | thx |
| 00:48.55 | lkcl | can you work it out from this: http://wiki.xda-developers.com/index.php?pagename=ASIC3 |
| 00:50.48 | cr2 | i have gathered this file together because there were no usable docs about asic3 |
| 00:52.48 | lkcl | good _grief_. no screen, but i have a usbnet! |
| 00:52.55 | lkcl | on a himalaya boot! |
| 00:54.34 | iclem | with the kernel i provided ? |
| 00:54.41 | lkcl | no, with one i just built. |
| 00:54.54 | iclem | ok, do yo a diff ? |
| 00:55.20 | lkcl | quee |
| 00:55.22 | lkcl | que? |
| 00:55.32 | iclem | do you want a diff with the last changes ? |
| 00:55.38 | lkcl | yesplease. |
| 00:56.19 | lkcl | what's the status that you've managed to get to (without the asic3 stuff)? |
| 00:56.37 | lkcl | have you got touchscreen and usb0 working, for example? |
| 00:56.51 | lkcl | i can't quite get over the fact that this might work |
| 00:57.18 | iclem | yes, ts and usb0 works |
| 00:57.33 | iclem | http://iclem.net/himalaya/K2-6-12-hh2-xda0-ba0-himalaya-20051228.diff |
| 00:57.36 | lkcl | _excellent_ - then i shall nick your kernel :) |
| 00:57.55 | iclem | all the work has been made by llange |
| 00:58.28 | cr2 | don't forget the sound and DoC :) and IR!!!!!!!!!!!!!!!!!!!!! |
| 00:58.51 | lkcl | thanks cr2 |
| 00:59.16 | iclem | problem is battery is not charging under linux |
| 00:59.37 | cr2 | you should thank cp15. i would have never written the sound driver ;) |
| 01:01.46 | cr2 | really. try do do something about IR for the poor guys who have no serial cable... |
| 01:02.03 | lkcl | one thing at a time... |
| 01:02.21 | cr2 | :) i'm leaving. |
| 01:02.37 | cr2 | good night. |
| 01:02.41 | lkcl | k cr2. |
| 01:03.31 | iclem | bye |
| 01:03.51 | iclem | what should happen when insmoding asic3_mmc ? |
| 01:05.04 | iclem | no block device in /dev/mmc, but no error msg at all |
| 01:05.57 | iclem | and touchscreen not working anymore after insmod |
| 01:07.01 | lkcl | let me catch up. there's a bug in the himalaya_lcd_hw_init() code - as in: it's unlikely to work (from a resume, definitely not). |
| 01:07.23 | lkcl | the initialisation sequence doesn't have the ATI stuff in the middle of it. |
| 01:08.36 | lkcl | wtf has llange done to the himalaya lcd initialisation??? that stuff is there for a xxxxing good reason. |
| 01:09.03 | lkcl | the delays are _very_ specific. the initialisation of the ATI is _absolutely_ necessary. |
| 01:10.32 | lkcl | piss. |
| 01:10.58 | iclem | the LCD works, don't know about resume, have not tried |
| 01:11.15 | lkcl | it _won't_ work: if it does, it's because wince has already initialised it. |
| 01:11.44 | iclem | yes |
| 01:11.49 | lkcl | and as long as this "pseudo" initialisation never turns off the LCD power, that initialisation by wince will still be valid. |
| 01:12.12 | lkcl | the moment you kill the power to the LCD, this code stands _zero_ chance of getting the ATI to work again. |
| 01:12.18 | iclem | it's the same as BlueAngel isn't it ? |
| 01:12.37 | lkcl | other than the blueangel doesn't work either, but yes, it's the same. |
| 01:13.27 | lkcl | i spent about 8 weeks getting that code to at least work. |
| 01:15.14 | lkcl | ok. |
| 01:15.19 | lkcl | i have something that might work. |
| 01:15.50 | lkcl | split the two sections up into pre and post, just like for the blueangel. it's using the w100fb so... yehh. |
| 01:20.33 | lkcl | ok. compile underway... whoops :) |
| 01:21.18 | iclem | have you any idea about battery not charging ? |
| 01:21.35 | lkcl | nooo |
| 01:21.45 | lkcl | w4xy might. |
| 01:21.58 | lkcl | he worked on the usb stuff. |
| 01:25.42 | iclem | what's your board id ? |
| 01:27.30 | lkcl | 5 probably. |
| 01:27.56 | lkcl | how do i check? (once i have a boot, of course - need to get a boot first!) |
| 01:28.11 | iclem | dmesg, there is a printk |
| 01:28.16 | lkcl | ahh. |
| 01:28.27 | lkcl | lemme get a boot first, then i can tell you :) |
| 01:28.29 | iclem | llange has 4, i've 6 |
| 01:28.52 | lkcl | mine is 2 years old |
| 01:29.02 | iclem | mine is brand new ;) |
| 01:39.27 | lkcl | here's a likely reason why the modprobe asic3_mmc causes the screen to go blank: |
| 01:39.48 | lkcl | it should be loaded first (because it flips various multiplexed GPIOS and stuff) |
| 01:40.08 | lkcl | and it probably spikes the power - or resets - the LCD power. |
| 01:40.34 | iclem | ok |
| 01:40.57 | iclem | actually screen do not blank, but touchscreen do not work anymore |
| 01:41.23 | lkcl | oh right :) oh yes. .... *thinks*... touchscreen... ssp... hmmm... doesn't quite make sense. |
| 01:41.44 | lkcl | the touchscreen is on the ssp interface (one-of). |
| 01:42.05 | lkcl | unless there's an incorrect address and some random memory is getting overwritten. |
| 01:42.25 | iclem | yes |
| 01:42.48 | iclem | i'm not sure of what i've done, will check that tomorrow |
| 01:42.58 | lkcl | ack. |
| 01:43.14 | lkcl | okay - let's try another boot, this time with the same kernel you're using.... wheeeeee |
| 01:43.44 | lkcl | boot! |
| 01:43.51 | iclem | of course ;) |
| 01:43.52 | lkcl | screen! |
| 01:43.58 | lkcl | usbeeeee! |
| 01:44.02 | iclem | all the screen ? |
| 01:44.07 | iclem | no margin ? |
| 01:44.12 | lkcl | familiar login... hmmmm.... |
| 01:44.42 | lkcl | no keyboard, so i can't type root :) |
| 01:44.55 | lkcl | <PROTECTED> |
| 01:45.02 | lkcl | no modules.dep... |
| 01:45.06 | iclem | you can use the ts |
| 01:45.15 | iclem | no modules nedded |
| 01:45.24 | iclem | built-in kernel |
| 01:45.33 | lkcl | it's not booted up with thingy. familiar. |
| 01:45.56 | iclem | oh yes |
| 01:45.58 | lkcl | i have /etc/modprobe.d set up from the universal... |
| 01:46.32 | iclem | i'm using a ramfs with gpe |
| 01:46.52 | lkcl | yeh. it seems i still have a bit of work to do. |
| 01:46.58 | iclem | ;) |
| 01:47.13 | lkcl | could not load /lib/modules/2.6.12-hh2-xda0-un0/modules.dep |
| 01:47.17 | lkcl | note "un0".... |
| 01:49.07 | iclem | are you using nfs root ? |
| 01:49.10 | lkcl | no. |
| 01:49.53 | iclem | handy to test modules |
| 01:49.57 | lkcl | yehh. |
| 01:50.44 | lkcl | you're loading stuff not from the SD-card but from ramdisk, yes? |
| 01:51.00 | lkcl | so you have gpe in ramfs - in the initrd???? :) |
| 01:51.37 | iclem | yes, copying gpe image in ramfs and pivotroot after initial nfs root |
| 01:51.50 | iclem | copying via nfs |
| 01:52.01 | lkcl | whatwhatwhat??? :) |
| 01:52.08 | iclem | huhu |
| 01:52.09 | lkcl | oh dear - sounds fun. |
| 01:52.23 | iclem | standalone himalaya |
| 01:52.28 | iclem | without SD |
| 01:52.38 | iclem | but cannot be charged anymore |
| 01:52.41 | lkcl | can you do me a favour: could you possibly document it on a wiki page somewhere, pleeeease |
| 01:52.49 | iclem | of course |
| 01:52.53 | lkcl | what i would need to do to set that up. |
| 01:53.18 | lkcl | i might have to do that. urk :) |
| 01:53.55 | lkcl | what i ended up doing originally was to copy files and modules into the initrd - including, in the end, Xfbdev |
| 01:54.13 | lkcl | i think i even put dropbear in it. unpacked from the arm .debs :) |
| 01:54.37 | lkcl | mad. |
| 01:54.39 | iclem | wow |
| 01:54.40 | iclem | yes |
| 01:55.06 | iclem | you seem not to be discouraged easily |
| 01:55.07 | lkcl | well - it worked. but i had to specify a _really_ large initrd - 12mb or so. |
| 01:55.19 | lkcl | not really :) |
| 01:55.42 | lkcl | only by myself. |
| 01:56.08 | lkcl | okay. think. what can i do. |
| 01:56.19 | iclem | nice to see you back on htc hacking then |
| 01:56.20 | lkcl | try building in asic3_mmc. |
| 01:56.25 | lkcl | he he. thanks :) |
| 01:57.08 | iclem | i must go to bed now, eyelids are really to heavy. I will document my setup tomorrow |
| 01:57.15 | lkcl | okay iclem. |
| 01:57.19 | lkcl | thanks for sticking around. |
| 01:57.32 | iclem | thx and good night |
| 01:57.36 | lkcl | night dude. |
| 02:28.12 | lkcl | ~botmail for iclem: http://wiki.xda-developers.com/index.php?pagename=HimalayaLinuxBootingViaRootNFS |
| 02:28.20 | lkcl | ~botmail for iclem: found it... |
| 04:01.12 | lkcl | ok. hmmm.... bus shift in the asic3_platform_data struct i tried setting to 1 (_complete_ LCD wipeout :) and to 2 made no difference... hmmm... ARGH. |
| 04:13.22 | lkcl | i want my python code back! the code that runs telnet to haret and issues GPIO and pd commands! |
| 04:13.24 | lkcl | waaah! |
| 04:21.03 | lkcl | ~botmail for iclem: GPIO3_CH_TIMER - that's the charger. |
| 04:21.59 | lkcl | the asic3 stuff for the himalaya is screwed. the charger isn't initialised on, etc. much has been messed about since 2.6.3 - oh well, it's probably for the best :) |
| 08:26.40 | *** join/#htc-blueangel rob_w-cgi (n=txuser@dd1010.kasserver.com) |
| 08:26.40 | *** mode/#htc-blueangel [+o rob_w-cgi] by ChanServ |
| 09:27.17 | rob_w-cgi | lkcl asic3_mmc is broken to compile for BA |
| 09:34.47 | lkcl | YAAY mda III arrived this morning. |
| 09:34.55 | lkcl | oh. sorry, rob - which line? |
| 09:35.10 | lkcl | blueangel? oh bugger :) |
| 09:36.31 | lkcl | damn i forgot to do a cvs commit last night. |
| 09:38.32 | rob_w-cgi | ahm 1045 |
| 09:38.46 | rob_w-cgi | but inside that file that define looks good ?!? |
| 09:38.58 | lkcl | it's the CONFIG_ARCH_BLUEANGEL |
| 09:39.03 | rob_w-cgi | yep |
| 09:39.04 | lkcl | change that to CONFIG_MACH_BLUEANGEL |
| 09:39.12 | rob_w-cgi | oh |
| 09:39.25 | lkcl | i just cvs committed that. sorry, was up a bit late. |
| 09:39.31 | lkcl | am going back to bed now :) |
| 09:39.33 | rob_w-cgi | np .. thanks |
| 09:44.13 | lkcl | oh - rob? |
| 09:44.26 | lkcl | have a look at himalaya_lcd.c, as it stands. |
| 09:44.43 | lkcl | the boardids, according to iclem, are the same for the himalaya (5) and your device (5). |
| 09:44.53 | lkcl | so take a look at bluangel_lcd_init_pre() |
| 09:44.55 | lkcl | and post. |
| 09:45.00 | lkcl | and himalaya_lcd_init_pre() |
| 09:45.01 | lkcl | and post. |
| 09:45.32 | lkcl | copy the principle. and the timings. |
| 09:46.11 | lkcl | oh - and also, have a look at himalaya_lcd.c as it USED to be: there's an initialisation of all four GPIOs to 0, followed by a mdelay(500) - all of which llange ripped out. |
| 09:46.26 | rob_w-cgi | hmm wait wait |
| 09:46.50 | lkcl | it's absolutely essential, for a proper initialisation, that the zeroing and the mdelay(500) is there. |
| 10:01.31 | *** join/#htc-blueangel JuNuKN (n=JuNuKN@Me300.m.pppool.de) |
| 10:01.41 | rob_w-cgi | yeah i was about to put back "the old " initialisation which was used on any board id |
| 10:02.50 | JuNuKN | morning cr2 and lkcl ... and the rest I didn't contacted personally right now ;o) |
| 10:03.25 | rob_w-cgi | lkcl so you say himalaya_lcd is the golden rule to be used also in blueangel_lcd ? |
| 10:03.35 | JuNuKN | I read the log from last night, ... but it break at the dayswitch, where it was very interresting! Does iclem and you got the mmc on himalaya to work? |
| 10:04.30 | rob_w-cgi | JuNuKN : they had a successfully day yesterday and i see cvs commits from 2 o clock in the moring also concerning mmc ! |
| 10:05.19 | JuNuKN | :o) great ... ! And DoC stuff ? # |
| 10:06.43 | JuNuKN | Please explain me which to be checked out to see all relevant things ... I understood, that Head isn't the right one! ? So any device has it's own branch? Must I check out all branches to follow all progresses? |
| 10:08.58 | rob_w-cgi | to my knowledge there is now himalaya branch but they use the ba branch . .so to checkout is to cvs -z3 -d:pserver:anonymous@cvs.sourceforge.net:/cvsroot/xanadux co -ko -r K2-6-12-hh2-xda0-ba0 linux-2.6-xda |
| 10:22.36 | JuNuKN | Ok, I am on the go to checkout this branch under Windows ... to visualize it to myself via WinCVS. What programm do you use to see the branches and diffs in a visualized manner under linux? pure cvs is commandline based, which is nothing for me to get a rought overview, while fighting with the commandline# |
| 10:24.42 | rob_w-cgi | ahm to visualize i only use viecvs |
| 10:24.52 | rob_w-cgi | but not on my host but on the sf cvs |
| 10:24.57 | JuNuKN | And which is the "tree-path" to see the mmc changes from last night ... first orientation! :o( sorry |
| 10:25.16 | rob_w-cgi | drivers/mmc/asic3_mmc |
| 10:27.27 | JuNuKN | ... the checkout is not complete right now, but at the moment I only see the root nodes ,,, documentation and linux-2.6-xda and the last I expect somewhere the himalaya stuff ... but probably I am wrong and have to wait for the rest of checkout |
| 10:30.09 | JuNuKN | Oh I was to fast .... under the node linux-2.6-xda now driver appeared !! Thanks rob_w ! I will check sources now ! |
| 10:43.36 | rob_w-cgi | crap sf , the sync to anoncvs takes too long |
| 10:44.44 | JuNuKN | OK, I saw the changes now, Thanks ... It was a evel hack for the BA only, which was corrected this night ;o) Think we have to find more of them ;o) |
| 10:48.08 | JuNuKN | I "somedays ago" downloaded a complete tarball of a SF project, which took also a while, but the diffs between the Versioned files was much faster ;o) The tars are build every night ... maybe it is from help for you |
| 10:50.39 | JuNuKN | Where must I search for the framebuffer stuff and console stuff for the devices? |
| 11:11.02 | JuNuKN | botmail read |
| 11:11.31 | JuNuKN | ~botmail read |
| 11:12.02 | JuNuKN | ~botmail read |
| 11:22.11 | JuNuKN | ~botmail help |
| 11:28.28 | JuNuKN | ~seen cp15 |
| 11:28.33 | apt | cp15 <i=tkejci@p549A4246.dip0.t-ipconnect.de> was last seen on IRC in channel #htc-blueangel, 9d 12h 8m 4s ago, saying: 'So bye'. |
| 11:32.55 | JuNuKN | ~botmail for iclem : Thanks for the new kernel ... I will test it later ... I have always problems, when I restored a WinCe Backup. It doubles some contacts and dletes some Files, to be synced with PC, therefore I am a little bit restricted with my tests ... : Sad, that the GPE Image isn't available within a initRD .... It would make it much more reproducable ... But perhaps you can give me some |
| 11:32.55 | JuNuKN | <PROTECTED> |
| 11:40.15 | JuNuKN | ... away for now |
| 11:46.52 | lkcl | ~botmail for JuNuKN we gave it a shot (mmc) - didn't come off. pain. i also heaved in some patches that should have been merged a long time ago. |
| 11:48.21 | JuNuKN | ... i am still here ... |
| 11:48.23 | lkcl | ~botmail for JuNuKN: http://wiki.xda-developers.com/index.php?pagename=HimalayaLinuxBootingViaRootNFS here's a start if nothing else. you will then have the initrd unpacked, you can modify the linuxrc, add in a mount -t nfs of the new rootfs, and do a pivotroot on that. |
| 11:48.26 | lkcl | oh - are you? :) |
| 11:48.34 | lkcl | who signed off, then? :) |
| 11:49.42 | lkcl | okay - if you're still hanging around for stupid cvs then try this: http://hands.com/~lkcl/public_html/xanadux.patch |
| 11:49.54 | lkcl | ah, it was rob. |
| 11:50.03 | lkcl | food. gotta go get food. |
| 11:50.07 | JuNuKN | rob_w i only say good by an let irc running. but was bind in the room by my son and saw your info |
| 11:51.40 | JuNuKN | that was also my reason .... wish you a good meal and perhaps we met this night ... bye ... offline |
| 11:52.26 | JuNuKN | but the patch link doesn |
| 11:52.31 | JuNuKN | #t work ... |
| 11:54.07 | lkcl | eep. |
| 11:54.51 | lkcl | http://hands.com/~lkcl/xanadux.patch - sorreeee :) |
| 12:02.14 | *** join/#htc-blueangel rob_w-cgi (n=txuser@dd1010.kasserver.com) |
| 12:02.14 | *** mode/#htc-blueangel [+o rob_w-cgi] by ChanServ |
| 12:02.18 | JuNuKN | Thanks, I will see whether I am able to use the patch and compile a kernel myself ... for the first test I got a new kernel by iclem to download and test with console and Touchscreen working ... |
| 12:48.37 | *** join/#htc-blueangel rob_w-cgi (n=txuser@dd1010.kasserver.com) |
| 12:48.37 | *** mode/#htc-blueangel [+o rob_w-cgi] by ChanServ |
| 13:20.44 | *** join/#htc-blueangel iclem (n=clement@iclem.net) |
| 13:23.16 | *** join/#htc-blueangel rob_w-cgi (n=txuser@dd1010.kasserver.com) |
| 13:23.16 | *** mode/#htc-blueangel [+o rob_w-cgi] by ChanServ |
| 13:23.31 | rob_w-cgi | ping lkcl |
| 13:25.02 | iclem | hi all |
| 13:25.42 | iclem | according to the hour of the last botmail of lkcl, he may be sleeping ;) |
| 13:26.16 | rob_w-cgi | hmm |
| 13:26.41 | rob_w-cgi | iclem are you familiar with his lcd voodoo ? |
| 13:28.12 | iclem | a little bit, what is your question |
| 13:28.54 | rob_w-cgi | iam trying to get this lcd stuff straight and work for the ba with board id 5 .. i ve a assumption that this board id was forgotten abit |
| 13:29.29 | rob_w-cgi | so lkcl told me today i need to check on himalaya_lcd and how it treats all .. but i am unsure if he meant that i can copy that or so ?? |
| 13:32.04 | iclem | I think you can't copy it straight forward |
| 13:32.54 | iclem | in himalaya_lcd there are to structs depending on board ID |
| 13:33.06 | iclem | two structs |
| 13:33.13 | rob_w-cgi | yeah i didnt meant that .. aha |
| 13:34.07 | rob_w-cgi | hmm maybe i go for the a merge of the blueangel_lcd before all that changes and now .. i think i remmeber now that lkcl said the timing will be the problem |
| 13:34.32 | iclem | it's the same in blueangel_lcd |
| 13:35.40 | lkcl | rob_w-cgi: the sequence is: four GPIO clears-to-zero, wait 500ms |
| 13:35.46 | lkcl | two GPIO sets, wait 30ms |
| 13:35.54 | lkcl | sorry - one GPIO set, wait 30ms |
| 13:36.00 | lkcl | another GPIO set, wait 30ms. |
| 13:36.06 | lkcl | then the ATI initialisation. |
| 13:36.12 | lkcl | then another GPIO set |
| 13:36.17 | lkcl | then wait 30 ms |
| 13:36.21 | lkcl | then the final GPIO set. |
| 13:36.29 | lkcl | that's what you've got to have happen. |
| 13:36.49 | lkcl | exactly _which_ GPIOs is defined, already, in blueangel_lcd.c |
| 13:36.54 | lkcl | in the correct order. |
| 13:37.04 | rob_w-cgi | ok great |
| 13:37.20 | lkcl | so you need to cut/paste the two GPIO sets into the blueangel_lcd_init_pre() at the appropriate if statement |
| 13:37.27 | rob_w-cgi | i just had that 30ms waits in between .. but will i need to make the clear to zero |
| 13:37.33 | lkcl | and leave the other two GPIO sets. |
| 13:37.54 | lkcl | well, strictly speaking, the clears-to-zero and the 500ms wait _might_ not be necessary |
| 13:38.05 | lkcl | because in the suspend, they will have already been set to zero. |
| 13:38.35 | lkcl | however, it depends what blueangel_resume.c does as to whether they get initialised to garbage, or 1s, i just don't know. |
| 13:39.14 | lkcl | to be _safe_, clear all 4 GPIOs to zero and wait for 500ms, in the lcd_init_pre() function, and _then_ do the first one, wait 17ms, second one. |
| 13:39.27 | lkcl | whatever. |
| 13:39.56 | rob_w-cgi | ok now it looks like ( all for board id 5 ) gpio_lcd_pwr1_on - wait 30 - pwr2 on - wait 30 pwr3 on - wait 30 pwr4 on - wait 30 .. all in pre_init .. post init is still empty for id 5 |
| 13:40.51 | rob_w-cgi | oops siad not pwr4 but pwr 1 again |
| 13:41.53 | rob_w-cgi | hmm |
| 13:42.36 | rob_w-cgi | from your above messages i get out that i set the frst 2 gpio in pre init and then the other 2 gpio`s in post init .. all with 30ms pause , right `? |
| 13:49.56 | rob_w-cgi | ok i am confident with taht and give that a go |
| 13:50.04 | rob_w-cgi | hoepfilly my lcd will not die |
| 13:53.37 | lkcl | that's correct, rob. |
| 13:54.04 | lkcl | check the exact required sequence - of GPIOs and pauses - in himalaya_lcd.c |
| 13:54.22 | rob_w-cgi | ok cool ... |
| 13:55.02 | lkcl | i have a blueangel! hurrah. an MDA III |
| 13:55.07 | rob_w-cgi | i that baby is then again usable to suspend and resume .. |
| 13:55.13 | rob_w-cgi | oh how cool... |
| 13:55.19 | rob_w-cgi | u like it ? |
| 13:55.40 | lkcl | yeh - now all i gotta do is work out what to do with it :) |
| 14:00.27 | *** join/#htc-blueangel dullard (n=jim@adsl-static-1-30.uklinux.net) |
| 14:25.58 | rob_w-cgi | ok now i am booting |
| 14:26.49 | lkcl | whoo-hoo :) |
| 14:27.30 | rob_w-cgi | oh no .. i wasnt able to read the startup.txt and now it is stuck on black screen turned of ... bah |
| 14:27.42 | lkcl | oops - another boot :) |
| 14:28.30 | rob_w-cgi | not possible .. there is a bug that you cannot turn on the unit anymore if you dont have AC .. which i dont have here |
| 14:29.17 | lkcl | oh, damn - i deleted the modules on the SD card for the ba. |
| 14:29.20 | lkcl | rats. |
| 14:29.42 | rob_w-cgi | so we only know today evening , when i got home .. .craaap |
| 14:29.51 | rob_w-cgi | ~kill HTC for this bug |
| 14:29.53 | apt | ACTION shoots a super-inverse anti-meson gun at HTC for this bug |
| 14:31.01 | lkcl | ~kill bitbake for time-wasting dependency stupidity |
| 14:31.03 | apt | ACTION shoots a charged pseudoquark gun at bitbake for time-wasting dependency stupidity |
| 14:31.52 | rob_w-cgi | lkcl if you have lots of RAM you can use psyco |
| 14:32.10 | lkcl | can't use psyco it doesn't help (plus i have amd64). |
| 14:32.34 | rob_w-cgi | oo |
| 14:32.49 | lkcl | i had to do a total rebuild. |
| 14:33.09 | lkcl | amd64 processors are only £75 and the motherboards only £45 (on ebuyer.com) |
| 14:34.24 | rob_w-cgi | scary |
| 14:34.50 | lkcl | SHRIIEEEK!!! an AMD64 dual-core processor - 4800+ (2.4 ghz) is £470 + VAT!!!!! |
| 14:35.09 | lkcl | oh my god that's just an insane price. |
| 14:36.02 | lkcl | hmmm... the 3000+ isn't listed on ebuyer.com any more... hmmm. |
| 14:36.10 | rob_w-cgi | hehe |
| 14:37.10 | lkcl | okay - the 754 pin ones, yes, they're still available - all the way down to 1.6ghz. |
| 14:37.33 | lkcl | and only £45. |
| 14:37.46 | lkcl | but the 939 pin ones only start at 2ghz - and £125. |
| 14:37.54 | lkcl | damn was i lucky. |
| 14:40.26 | lkcl | complete rebuild. _again_. damnit. |
| 14:41.33 | rob_w-cgi | why |
| 14:43.35 | lkcl | an error due to bitbake - FILESPATH missing |
| 14:43.51 | lkcl | an error due to bitbake - FILESPATH missing |
| 14:44.03 | lkcl | <PROTECTED> |
| 14:44.47 | iclem | hi lkcl |
| 14:45.16 | iclem | if I follow you ahve now three HTC devices ? |
| 15:24.16 | *** join/#htc-blueangel iclem (n=clement@iclem.net) |
| 15:32.37 | lkcl | hiya iclem. yep. |
| 15:34.12 | lkcl | my old himalaya. a new universal ("free" under contract - yeahright) and a 2nd-hand blueangel. screen quality is _really_ poor on the mda III. |
| 15:34.54 | rob_w-cgi | lkck what board id is |
| 15:35.00 | rob_w-cgi | the ba ? |
| 15:35.16 | lkcl | no idea yet. i need to rebuild the modules, and am having difficulties with bitbake |
| 15:36.15 | lkcl | some python code appears to have found its way into a shell script |
| 15:39.12 | rob_w-cgi | oops |
| 15:41.50 | lkcl | ahh... someone missed out the word "python" in front of something somewhere... i might be able to fix it in the .bb file... |
| 15:43.11 | lkcl | there - sorted it. |
| 15:46.47 | rob_w-cgi | try to see if you are running the latest checkout .. i ve many problems which where only by me .. runing old versions or screw ups in between version jumps |
| 15:47.21 | lkcl | que?que? drivers/serial/pxa.c:768: error: initializer element is not constant - where did _that_ come from? :) |
| 15:47.29 | lkcl | hmmm time to look at cvs logs... |
| 15:50.13 | lkcl | don't tell me the version of arm-linux-gcc has gone up... |
| 15:52.27 | rob_w-cgi | well i would not bet on it stayed |
| 15:53.05 | rob_w-cgi | ok i am off to go home |
| 15:53.07 | rob_w-cgi | cu |
| 16:39.59 | *** join/#htc-blueangel JuNuKN (n=JuNuKN@Me300.m.pppool.de) |
| 16:40.34 | JuNuKN | ~botsmail read |
| 16:41.01 | JuNuKN | ~botmail read |
| 16:41.39 | JuNuKN | ~botmail read |
| 16:42.55 | iclem | hello JuNuKN |
| 16:43.38 | JuNuKN | Hello iclem .... thanks for the PM ... and the Wiki stuff ... sounds fantastic ! |
| 16:47.27 | iclem | hehe, it is a simple way to show himalaya running linux to friends |
| 16:47.35 | iclem | not much more |
| 16:49.31 | JuNuKN | Yes, and to get myself addicted to the possible feautures... |
| 16:50.15 | JuNuKN | I have to prepare my newley installed linux machine (PC) with NFS , toolchain, sourcecode and so on ,,,, |
| 16:50.46 | iclem | got rid of your corruptions problems ? |
| 16:52.37 | JuNuKN | You did really a great job with this doku ! Maybe it is also possible somehow (maybe dd) to store the newley created Image later back into a initRD ? Then it is a kids work to have linux on the device, for get addicted to participate in developement ;o) |
| 16:53.31 | JuNuKN | No, .-... I did not found the problem ... I only installed linux newley and bought a fan for the Harddisk .... Sorry have to go to eat now ... will come later! |
| 16:53.45 | JuNuKN | Shure ! |
| 16:54.10 | iclem | cu |
| 17:06.00 | lkcl | heya iclem. |
| 17:06.38 | lkcl | JuNuKN - do consider deliberately down-clocking your processor, if it's possible for your box. |
| 17:06.57 | lkcl | iclem: did you get the message about the himalaya nfs page i found, on xda-developers.com? |
| 17:07.31 | iclem | no, was it about HimalayaLinuxBootingViaRootNFS ? |
| 17:13.41 | lkcl | yep - that's the one. |
| 17:14.56 | lkcl | oh - big joke (on me) - the initrd you suggested i unpack? the reason it contains Xfbdev and dropbear is because it _is_ the initrd i made up nearly 2 years ago :) |
| 17:15.40 | iclem | hehe |
| 17:15.51 | lkcl | what goes around comes around, i suppose. |
| 17:16.14 | lkcl | what an awful hack job. it's why the initrd is nearly 5mb and unpacked is over 12. |
| 17:17.56 | iclem | yes, we could have cleaned it a bit, but it works, and time is used for something else :) |
| 17:21.54 | lkcl | oo, how exciting: bitbake gpe-image uses xserver-xor-6.8.99.11-r0 |
| 17:43.41 | iclem | lkcl: thx for you botmail about charger |
| 17:44.51 | iclem | This register seems to be already triggered by himalaya_udc_command() |
| 17:45.23 | lkcl | whatwhatwhat? i sent you stuff about a charger? oh - yes. i did, didn't i. yeh, w4xy spent some time getting it right - it appears that it's been hacked about since then. |
| 17:46.44 | iclem | ok, I willl dig into old code to see what has changed |
| 18:11.11 | *** join/#htc-blueangel rob_w (n=bob@p85.212.148.147.tisdip.tiscali.de) |
| 18:11.11 | *** mode/#htc-blueangel [+o rob_w] by ChanServ |
| 18:36.20 | lkcl | okay - looking at http://wiki.xda-developers.com/index.php?pagename=ASIC3 |
| 18:36.36 | lkcl | that's the one that says that hx4700 and blueangel are "normal" registers. |
| 18:36.51 | rob_w | lkcl, did not work ,. tux of haret did not went away and startied corruptiong .. but kernel did boot |
| 18:37.31 | lkcl | whatwhatwhat, rob? |
| 18:38.16 | lkcl | and that means that bus_shift needs to be set to 1 in the asic3_platform_data. |
| 18:38.41 | lkcl | gimme 2secs rob - i have been trying to do a bitbake recompile since some time last night. |
| 18:39.09 | lkcl | there are several bugs in the oe.familiar dev monotone stuff that have stopped a build. |
| 18:39.22 | lkcl | grr, arg :) |
| 18:40.06 | lkcl | rob - you sent too much: bitchx (my irc client) just put you on auto-ignore, for flooding :) |
| 18:40.52 | rob_w | <PROTECTED> |
| 18:40.54 | rob_w | ok hang on |
| 18:40.56 | rob_w | http://pastebin.com/483159 |
| 18:41.02 | lkcl | ah - you're on |
| 18:42.20 | lkcl | interesting. |
| 18:42.41 | lkcl | it's actually switching GPIO LCD_PWR_1 _off_, and LCD_PWR2 _on_ |
| 18:42.54 | lkcl | then PWR_3 on |
| 18:42.58 | lkcl | then PWR_1 on. |
| 18:43.22 | lkcl | that's radically different from the himalaya_lcd stuff, which does LCD_PWR 1,2,3,4 off |
| 18:43.36 | lkcl | then LCD PWR_1 on, delay, LCD_PWR2 on, delay |
| 18:43.42 | lkcl | ATI 3200 initialisation |
| 18:43.43 | lkcl | delay |
| 18:43.50 | rob_w | brb ,, lets get this today |
| 18:43.59 | lkcl | LCD_PWR_3 on, delay, LCD_PWR_4 on. |
| 18:46.45 | lkcl | i'm inclined to say that you really need to look at the assembler, or to run the same emulator that cp15 was running. |
| 18:51.47 | *** join/#htc-blueangel expose (n=carstenh@hacker-35-225.congress.ccc.de) |
| 18:52.01 | expose | hi |
| 18:52.06 | lkcl | heya |
| 18:53.42 | expose | questions: |
| 18:53.47 | lkcl | sure. |
| 18:54.04 | expose | does opie work? does gsm work? what major thing doesnt? |
| 18:54.35 | expose | aka "is it useable as a all-day mobile phone yet?" using calls and sms |
| 18:54.46 | lkcl | opie works. gsm works. resume doesn't, because we're trying to get it to work "properly". |
| 18:55.22 | lkcl | yes, if you revert to the cvs as of two weeks ago, which contains work-around suspend code that blanks the screen instead. |
| 18:55.32 | lkcl | that will give you a battery life of about 18 hours. |
| 18:56.07 | lkcl | gsm interrupts on incoming phone calls will wake it up, gomunicator will take the call. |
| 18:56.30 | lkcl | what we're battling with at the moment is to get the LCD to reactivate after a resume. |
| 18:57.04 | lkcl | the initialisation sequence is _very_ specific, including ms and usec delays on _four_ GPIO power-up lines. |
| 18:57.20 | lkcl | it took me 8 weeks to get the himalaya's sequence right, two years ago. |
| 18:57.41 | lkcl | ended up with about 50 lines of friggin code, after 8 weeks. |
| 18:58.08 | expose | which means standby times are..? |
| 18:58.08 | expose | 1/2 the original ones, or a quater or |
| 18:58.14 | *** join/#htc-blueangel expose (n=carstenh@hacker-35-225.congress.ccc.de) |
| 18:59.39 | expose | which means standby times are..? |
| 18:59.41 | lkcl | dunno - it's not ideal, but it works. it's "standby" rather than "suspend" basically. |
| 19:00.09 | lkcl | there'll be some web site from o2 with the details... lemme see if i can find something (my wife made me throw the box away) |
| 19:00.53 | lkcl | standby time is supposed to be 168 hours. |
| 19:01.05 | lkcl | according to this: |
| 19:01.06 | lkcl | http://reviews.zdnet.co.uk/hardware/handhelds/0,39023879,39193995,00.htm |
| 19:06.49 | lkcl | hope that helps. it's also why we're trying to get the damn lcd wake-up sequence right. it's kinda crucial :) |
| 19:08.36 | JuNuKN | By the way, did someone heard the presentation from cp15 on ccc conference? |
| 19:13.42 | expose | thanks |
| 19:13.50 | expose | i just did |
| 19:13.53 | expose | still sitting here :-) |
| 19:14.02 | expose | while the presentation is over yet for a while |
| 19:14.13 | expose | JuNuKN: |
| 19:22.27 | lkcl | ooo - how's it going? |
| 19:22.42 | lkcl | can you let cp15 know that my mda III arrived? |
| 19:31.03 | rob_w | expose, how was it ? |
| 19:33.42 | rob_w | lkcl, so u say before the lcd was only blank but the ati didnt "sleep" ? is that the goal we are fighting ? |
| 19:35.43 | JuNuKN | Ooops was away, to bring the kids to bed .. |
| 19:36.45 | JuNuKN | expose: How was it? A lot of interested people? |
| 19:39.52 | lkcl | not quite, rob. 2 years ago, suspend wasn't even on the cards, it was too early. _but_ i did get the himalaya LCD initialisation sequence. a patch by llange which i committed last night without double-checking it trashed the initialisation sequence i'd written, with something that would never work (except if you booted from wince, and you never suspended the device). |
| 19:40.14 | lkcl | and the himalaya has the same problem with its LCD initialisation sequence. |
| 19:40.46 | lkcl | so, what cp15 did was to put in "black" on-screen, temporarily, and to keep the power on. |
| 19:41.03 | lkcl | so, yes - now i understand what you're asking: yes, you are correct. |
| 19:41.39 | *** join/#htc-blueangel expose_ (n=carstenh@hacker-35-225.congress.ccc.de) |
| 19:48.02 | JuNuKN | expose_ , are there a lot of people interested in the Blueangel status? |
| 20:01.07 | expose | well, it seem like |
| 20:01.55 | expose | i cannot tell of course, but people where staying there, and listened although the laptop <--> projector stuff didnt work wich is a huge drawback of course. |
| 20:07.34 | JuNuKN | shit happens ... so nobody has seen the device in action ? |
| 20:25.40 | expose | nope |
| 20:25.58 | expose | but the EZX motorola boards |
| 20:26.31 | expose | via the usb-link and telnet and stuff, didnt see that too good since it was small and bad |
| 20:26.50 | JuNuKN | shit, so probably nobody will be addicted to participate on the developement |
| 20:27.28 | expose | well, possibly. duno. but it wasnt really good due to that projector problem. |
| 20:28.56 | JuNuKN | ok, we will see what he sais, when he returns ., |
| 20:30.52 | JuNuKN | ... if he returns ... |
| 20:36.09 | *** join/#htc-blueangel TelePuzik (n=UNIX@80.81.39.150) |
| 20:36.22 | TelePuzik | Hi Guys |
| 20:39.04 | JuNuKN | Hello |
| 20:39.51 | TelePuzik | Merry Christmass |
| 20:40.14 | JuNuKN | <PROTECTED> |
| 20:40.21 | TelePuzik | :) |
| 20:40.53 | TelePuzik | still got to celebrate on the 6th and 7th;) |
| 20:51.48 | lkcl | for goodness' _sake_ somebody's been messing with the source code on gpe-image dependencies. *sigh*. stuff that built 2 days ago now doesn't. *grumble* |
| 20:54.09 | JuNuKN | Sounds bad. Hello again lkcl. |
| 21:00.49 | JuNuKN | Everybody bussy ... |
| 21:02.10 | *** join/#htc-blueangel expose (n=carstenh@hacker-35-225.congress.ccc.de) |
| 21:24.22 | JuNuKN | So I will try to compile tomorrow .... I downloaded a CrossCompileChain and the sources ... But how to proceed with the Crosschain ? |
| 21:42.28 | *** join/#htc-blueangel prh (n=paul@212.13.203.80) |
| 21:42.38 | prh | so linux is running on the universal? |
| 21:42.47 | prh | hmmm! |
| 21:50.39 | iclem | JuNuKN: untar it and add bin location to path |
| 21:51.27 | iclem | then "cd" to your source dir and "make himalaya_defconfig" |
| 21:53.32 | iclem | be back later |
| 21:54.28 | JuNuKN | Hey iclem ,,,, which toolchain do you use? I took the official familar gcc-3.3.2.tar , but it reported problems by untar ... directories, where files expected ... |
| 21:57.19 | JuNuKN | I try now the arm-linux-gcc-3.4.1.tar.bz2 |
| 22:03.46 | *** join/#htc-blueangel expose (n=carstenh@hacker-35-225.congress.ccc.de) |
| 22:14.48 | JuNuKN | OK, this Chain can be unpacked without errors ... |
| 22:15.08 | JuNuKN | make himalaya_defconfig worked also |
| 22:16.41 | JuNuKN | What now ... which bin path of the toolchain has to be included ? must it be copied into the root directory, or is it enought to have it in the user directory ? How to compile than ? |
| 22:23.40 | *** join/#htc-blueangel expose (n=carstenh@hacker-35-225.congress.ccc.de) |
| 22:23.48 | lkcl | BITCH. |
| 22:23.57 | lkcl | damn stuff has moved over to gcc 4.0 |
| 22:24.02 | lkcl | bitbake builds. |
| 22:24.04 | lkcl | damnit. |
| 22:24.17 | lkcl | that means that osb-jscore wouldn't compile without some hacking. |
| 22:24.21 | lkcl | *sigh*. |
| 22:25.13 | lkcl | well, that small nightmare is over, now. |
| 22:36.53 | lkcl | friggin thing. this is going to cause grief. |
| 22:37.18 | lkcl | does anyone know how to make this damn stuff move back to arm-linux-gcc-3.4-or-thereabouts? |
| 22:37.48 | lkcl | i _so_ don't need the grief of gcc 4.0.2 |
| 22:45.02 | JuNuKN | Bye iclem ... i have to go to bed for today, maybe you can post me the answer per botmail ... Many thanks in advance. |
| 23:22.21 | iclem | lkcl: looks like using oe is like a moving target |
| 23:45.38 | iclem | somebody here ? |