**** BEGIN LOGGING AT Thu Oct 23 02:59:57 2008 Oct 23 03:03:39 bfree: hi! Oct 23 03:03:48 bfree: excellent, another developer! :) Oct 23 03:04:35 yes, using debootstrap in qemu is a valid approach to build a debian rootfs for the palms Oct 23 03:05:10 building the kernel it it will probably take a really, really long time but should also work Oct 23 03:06:16 for the kernel it's usually less painful to use a cross-compiler Oct 23 03:07:06 hi BobOfDoom: I've no idea why but cross-compilers scare me :-P I guess I'm a bit to indoctrinated into the debian ways ;-) Oct 23 03:07:24 cross-compiling the kernel is really easy, just set ARCH=arm and CROSS_COMPILE=/path/containing/gcc environment variables and then make menuconfig or whatever as per usual Oct 23 03:07:48 but yeah, qemu should work fine too, you'll just have to be patient ;) Oct 23 03:07:54 I'd actually fired of a debian install under qemu to see how it went, not too fast is the answer but it's working away Oct 23 03:11:08 Cocoboot is a bit flakey on the 680 at the moment, it often just resets the device Oct 23 03:11:21 I haven't had a chance to look into why yet Oct 23 03:14:52 resets trying to boot or random resets while using it? I assume just on boot? Oct 23 03:15:21 trying to boot Oct 23 03:15:34 palm os has rather poor memory management Oct 23 03:15:43 and suffers heavily from memory fragmentation Oct 23 03:15:44 it has memory management? Oct 23 03:16:10 which means Cocoboot often has trouble mallocing a large contiguous chunk of memory to load the kernel into Oct 23 03:16:22 that's usually the cause of the reset on boot problems Oct 23 03:16:28 thats what happens when u dont have virtual memory Oct 23 03:16:29 :) Oct 23 03:16:31 so minimizing the size of your zImage can help a bit Oct 23 03:16:56 u cant just program the mmu to remap the virtual big blob to lots of smaller blobs Oct 23 03:17:03 yeah Oct 23 03:17:13 it does actually use the MMU, but it only defines some static mappings Oct 23 03:17:22 it doesn't remap stuff on the fly AFAICT Oct 23 03:17:47 yeah Oct 23 03:17:51 thats what it should do Oct 23 03:17:53 but... doesnt Oct 23 03:17:55 mind u Oct 23 03:18:00 even with statixc mappings Oct 23 03:18:04 u still have fragmentation Oct 23 03:18:09 but then its within a userspace app Oct 23 03:18:26 ie your vm space that glibc dishes out to your app when u malloc will get fragged Oct 23 03:18:27 and badly Oct 23 06:09:47 fuck! Oct 23 06:10:05 not now dear Oct 23 06:10:27 i've got a.... turnip Oct 23 06:10:32 I took wife's mobile phone to RING RING again, but she needs it for her wake ups Oct 23 06:10:48 she's late to work :( Oct 23 06:12:57 damn aussies, ignoring problems in life and are happy with turnip Oct 23 06:15:11 * raster throws turnips around Oct 23 06:15:14 TURNIPS! Oct 23 06:15:18 * raster is content Oct 23 06:17:08 at least there is no long thinking about Christmas presents Oct 23 06:23:28 :/ Oct 23 06:24:19 * BobOfDoom doesn't have a turnip Oct 23 06:24:28 ask raster for one Oct 23 06:25:11 http://en.wikipedia.org/wiki/Image:2005carrot_and_turnip.PNG Oct 23 06:25:15 BobOfDoom: no turnip! dude! you need to get out of the ';bra Oct 23 06:25:22 global distribution of carrot and turnip output Oct 23 06:25:47 looks like there's turnips close by Oct 23 06:26:09 or maybe just carrots :( Oct 23 06:27:15 no turnip dot in CR Oct 23 07:10:00 AT+CMUX=? Oct 23 07:10:02 +CMUX: (0),(0),(1,5),(31-127),(10),(3),(30),(10),(1,7) Oct 23 07:10:03 OK Oct 23 07:10:05 BobOfDoom: ^^ Oct 23 07:11:19 ah Oct 23 07:11:32 cool Oct 23 07:12:02 just gotta fix wake then Oct 23 07:12:21 btw. I think it goes sleep much quicker Oct 23 07:13:00 than the 650? yeah sounds like it Oct 23 07:13:05 650 doesn't sleep in CFUN=0 mode Oct 23 07:13:08 they must have changed it Oct 23 07:13:22 I don't assert CFUN=0 every time Oct 23 07:13:25 should I? Oct 23 07:13:31 as part of initialization? Oct 23 07:13:31 CFUN=0 is the default Oct 23 07:13:42 it means minimal functionality Oct 23 07:13:52 virtually everything powered down Oct 23 07:14:10 in that mode it only responds to 5 or so commands Oct 23 07:14:24 so no need to issue that Oct 23 07:14:27 it's alerady in it Oct 23 07:14:36 you want CFUN=1 do anything useful ;) Oct 23 07:15:23 right Oct 23 07:15:25 MRDY 4 :b Oct 23 07:15:38 mm Oct 23 07:16:50 we need some replacement for the hh.org wiki Oct 23 07:16:55 one that lets anyone post Oct 23 07:17:30 thats just silly that the page is locked Oct 23 07:17:30 there must be some way to get rights, isn't it? Oct 23 07:19:10 probably email Jamey Hicks Oct 23 07:19:25 but really there shouldn't be any barreir to entry like that Oct 23 07:19:31 i have to say - no regrets on moving all of e's infra off sourceforge Oct 23 07:19:48 or begged/borrowed/stolen services onto our own Oct 23 07:20:02 havent had problems since then as we fully control our own infra all the way to the hardware Oct 23 07:20:28 yeah, I'm thinking that's what we'll do as well, no reason we can't run some simple wiki software on hackndev.com Oct 23 07:20:42 we previously went with hh.org since it was sorto f central, and the ipaq and HTC ports use it Oct 23 07:21:16 but I think most people are abadoning hh.org anyway for political reasons (the whole trademarks things) Oct 23 07:21:52 BobOfDoom: i would recommend trac myself Oct 23 07:22:26 what is with the handhelds.org trademark thing anyway? Oct 23 07:23:31 * raster reads up Oct 23 07:24:43 I dunno, I didn't pay much attention to it. I think George France incorporated hh.org with the intention of being like Mozilla, trademarked open source stuff. I don't think he consulted many people about it though, and the GPE developers felt like he was stealing their name Oct 23 07:26:07 similar with iPKG and OPIE Oct 23 07:27:37 raster: we actually have a trac install.. but it's been rather neglected Oct 23 07:27:59 it partly wasn't very useful because its running on a slow server (my fault) Oct 23 07:28:50 yeah Oct 23 07:28:57 i heard of the ipkg on earlier this year Oct 23 07:29:00 seems things got nasty Oct 23 07:29:11 and hh.org probably is being shunned by the community Oct 23 07:29:23 and yeah Oct 23 07:29:25 rtrac can be slow Oct 23 07:29:31 look at e's trac Oct 23 07:29:40 i know what that's dooing on the server Oct 23 07:30:06 out svn repo is one of the larger ones out there and trac is being quite punished by it when people commit and browse the tree Oct 23 07:30:23 almost 37,00 revs now Oct 23 07:30:24 and counting Oct 23 07:30:29 wait no Oct 23 07:30:32 just broke that Oct 23 07:30:36 Committed revision 37014. Oct 23 07:31:39 we've got a bugzilla as well that nobody uses Oct 23 07:31:42 but is seems like george was in the wrong Oct 23 07:31:48 hmm.. hackndev.com needs an overhaul' Oct 23 07:31:51 registering a trademark of someone elses project Oct 23 07:32:05 he may be peripherally involved Oct 23 07:32:07 but wrong Oct 23 07:32:15 yeah, definitely Oct 23 07:32:38 i know if someone wanted to trademark e - if they didnt discuss it openly amongst all devs Oct 23 07:32:51 and get a consensus, i'd be bringing out the 2x4 Oct 23 07:33:02 in principle it is probably a wise move Oct 23 07:33:15 though only when done with full backing of the group Oct 23 07:33:42 and then there would be only a handful of the e devs i'd say earnt the trust level to be able to be a steward of such a thing Oct 23 07:33:44 frankly I don't really like either bugzilla or trac. Trac's browsing and wiki are alright, but it's slow and the bug tracker is annoying. Bugzilla I just find confusing with buttons all over the place and no way to just browse stuff Oct 23 07:34:02 anyway - it does mean hh.org will die Oct 23 07:34:03 :) Oct 23 07:34:17 yeah Oct 23 07:34:18 and yeah Oct 23 07:34:25 bugzilla seems to have a following of haters Oct 23 07:34:27 me included Oct 23 07:34:37 trac i like thanks to simplicity for bug filing Oct 23 07:34:44 it's the best thing i've found out there Oct 23 07:34:51 and man have we had bugtraker wars Oct 23 07:35:00 haha Oct 23 07:35:04 but as a bonus we get an svn browser and revision web front-end Oct 23 07:35:07 and a wiki for free Oct 23 07:35:12 so that was fine Oct 23 07:35:26 but our boxes are fat dual-cpu p4-xeon's Oct 23 07:35:31 have 2 of them Oct 23 07:35:44 right nwo though 1 is a cold spare Oct 23 07:35:51 we dont do any loadbalancing Oct 23 07:35:53 ow Oct 23 07:35:59 we're just running off some cheap VPS Oct 23 07:36:04 aaah Oct 23 07:36:05 :p Oct 23 07:36:17 then we probably have a lot more power to throw aroudn than you - thus its ok for us Oct 23 07:36:40 http://fl0rian.wordpress.com/2008/03/28/gpe-trademark-rejected/' Oct 23 07:36:43 hahaha Oct 23 07:36:45 rejected Oct 23 07:36:45 :) Oct 23 07:37:15 XD Oct 23 07:38:03 annnnyway Oct 23 07:38:20 i suspect until i have a streamlined fso Oct 23 07:38:24 or fso streamlines Oct 23 07:38:31 i wont be doing call stuff on my treo any time soon Oct 23 07:38:53 yes Oct 23 07:39:13 I'm thinking of writing a simple "gsmd-lite" Oct 23 07:39:16 someone else is keen to "fix" fso Oct 23 07:39:19 which implements a subset of FSO Oct 23 07:39:22 or replace it as well Oct 23 07:39:26 only the basics Oct 23 07:39:34 i've said i am willing to help Oct 23 07:39:41 i can start the dbus daemon Oct 23 07:39:46 and soem of the calls Oct 23 07:39:58 the problem is just fleshing it out with enough calls Oct 23 07:40:00 and gsm logic Oct 23 07:41:50 I just need to get past uni semester so I have some time to work on this stuff Oct 23 07:42:03 aaah yes Oct 23 07:42:06 i remember that... Oct 23 07:42:16 past papers were the key Oct 23 07:42:17 :) Oct 23 07:42:58 heh, it's not the exams that bother me so much, but assignments at the moment, they take soo much time :( Oct 23 07:43:13 aaah Oct 23 07:43:23 * raster ponders an efso daemon Oct 23 07:43:31 i have an libefso sitting here Oct 23 07:43:40 or the start of it Oct 23 07:43:57 libefso = client wrapper for easy access? Oct 23 07:44:07 fucken busybox :E Oct 23 07:44:11 yes Oct 23 07:44:18 just an easy access lib to wrap the dbus goop Oct 23 07:44:33 the swiss guys did it Oct 23 07:44:38 or have started Oct 23 07:45:57 BobOfDoom: does that debian armel contain GPE? Oct 23 07:46:07 (your image) Oct 23 07:46:55 tail, tr, echo - all these tools changes speed of serial port to 115200 :E Oct 23 07:47:06 so simple terminal like Oct 23 07:47:11 no Oct 23 07:47:12 no X Oct 23 07:47:15 tail -f /dev/ttyS0 | tr '\r' '\n' & Oct 23 07:47:18 it's jsut console Oct 23 07:47:23 while read line; do echo -e "$line\r" > /dev/ttyS0; done Oct 23 07:47:24 although you can apt-get install X Oct 23 07:47:26 if you want Oct 23 07:47:28 just can't work :E Oct 23 07:47:37 eh? Oct 23 07:47:42 that's weird Oct 23 07:48:02 does debian armel contain GPE in general? Oct 23 07:48:15 no idea Oct 23 07:48:20 I haven't touched GPE in ages Oct 23 07:48:23 it's got matchbox though Oct 23 07:48:49 I just need environment with virtual keyboard and terminal Oct 23 07:48:58 ah, it has that Oct 23 07:48:58 and without busybox Oct 23 07:49:43 install X with these instructions: http://hackndev.com/node/212 Oct 23 07:49:51 then apt-get install matchbox-keyboard Oct 23 07:49:58 and you can use that as keyboard Oct 23 07:50:08 and just use xterm Oct 23 07:50:16 or wahtever random temrnial you like Oct 23 07:50:55 debian is huge, but yeah it's useful to have "normal" versions of everything available for debugging and research Oct 23 07:51:41 well, if it is worthy, I can use my 4gb sdhc Oct 23 07:51:50 hah Oct 23 07:51:59 you'll only need 1gb or so Oct 23 07:52:42 dang, it just falls asleep so much :E Oct 23 07:52:47 512m is enough Oct 23 07:53:05 I have to find that GPIOs Oct 23 07:53:20 like me in lectures Oct 23 07:56:55 87 must be power Oct 23 07:57:15 since it makes 35 ready and it changes 14 somehow Oct 23 07:57:39 14 must be HOST_WAKE because no other GPIO is changing Oct 23 07:57:49 but WAKE? Oct 23 07:58:26 79 changes nothing Oct 23 07:58:45 but is chip awaken? Oct 23 07:59:05 ok, what about power down, 79 down, power up, see difference? Oct 23 07:59:40 well when you assert wake it should start responding in minicom again Oct 23 08:00:27 argh, I gotta get this assignment finished so I can give you hand Oct 23 08:01:01 BobOfDoom: but how assert? Oct 23 08:01:04 set high? Oct 23 08:01:05 or Oct 23 08:01:09 set high, set low? Oct 23 08:01:21 set high Oct 23 08:01:25 it won't fall by itself Oct 23 08:01:30 right Oct 23 08:01:38 the way I use it Oct 23 08:01:42 so I have to switch from time to time? Oct 23 08:01:44 is when you want to talk to the chip Oct 23 08:01:47 set high Oct 23 08:01:49 write crap Oct 23 08:01:50 set low Oct 23 08:01:51 but Oct 23 08:01:55 reading Qtopia source Oct 23 08:01:59 I think I might have it wrong Oct 23 08:02:13 in qtopia source they assume it goes to sleep 30 seconds after wake being asserted Oct 23 08:02:28 but I dunno Oct 23 08:02:39 it didn't work like that for me Oct 23 08:02:39 is it relevant for my chip? Oct 23 08:02:41 might have chanegd Oct 23 08:02:57 Greenphone has an even older BCM than the one in the 650 Oct 23 08:02:58 so who knows Oct 23 08:03:31 or maybe it was newer Oct 23 08:03:33 heck I don't remember Oct 23 08:04:44 while true; do sleep 5 ; setl 79; seth 79; done Oct 23 08:04:49 this didn't helped Oct 23 08:06:41 ok, maybe 87 is wake and I don't have power Oct 23 08:06:48 the behaviour can be similar Oct 23 08:06:49 so 35 is RTS and 34 is CTS? Oct 23 08:06:58 I mean Oct 23 08:07:00 the other way around Oct 23 08:07:03 or wahtever it is Oct 23 08:07:08 how can I know :( Oct 23 08:07:23 34: * <- 00000001 Oct 23 08:07:29 35: <- 00000001 Oct 23 08:08:26 shouldn't it have AF Oct 23 08:08:38 hmm Oct 23 08:08:38 no Oct 23 08:08:43 it has afn1 Oct 23 08:08:45 both Oct 23 08:08:47 right Oct 23 08:09:34 34 FFRXD Oct 23 08:09:37 ah ok so 34 FFRXD Oct 23 08:09:40 yeah and CTS Oct 23 08:09:42 35 FFCTS Oct 23 08:10:15 39 is probably FFTXD Oct 23 08:10:24 39: * -> 00000002 Oct 23 08:10:43 41: -> 00000002 Oct 23 08:10:53 FFRTS Oct 23 08:11:01 40 is DTR Oct 23 08:11:25 40: * -> 00000001 Oct 23 08:11:28 afn1 Oct 23 08:11:38 keyboard Oct 23 08:11:54 oh really? Oct 23 08:12:01 that's weird Oct 23 08:12:29 cause your GPIO-tel.txt shows it changing along with the other phone stuff Oct 23 08:12:50 isn't it red button? Oct 23 08:12:58 hmm :( Oct 23 08:13:38 KEY_F6, // "Green/Call" }, Oct 23 08:13:39 KEY_O, // "O" }, Oct 23 08:13:41 KEY_BACKSPACE, // "Backspace" }, Oct 23 08:13:42 KEY_M, // "M" Oct 23 08:13:44 KEY_K, // "K" Oct 23 08:13:45 KEY_I, // "I" Oct 23 08:14:24 mistake Oct 23 08:14:33 that is for IN7 Oct 23 08:14:40 beware, this will kill your browser: http://meshy.org/~ato/tmp/680-tel.html Oct 23 08:14:47 (as it's huge) Oct 23 08:15:19 heh 1.8 is huge? Oct 23 08:15:20 look at frame 356 Oct 23 08:15:28 i've got even 20 Megs :D Oct 23 08:15:34 gah Oct 23 08:15:44 what browser are you using? Oct 23 08:15:52 firefox starts chugging for me Oct 23 08:15:59 maybe I need more RAM in my PC though Oct 23 08:16:52 damn you're right Oct 23 08:17:00 ff 3 works with that Oct 23 08:17:13 konqueror is getting too slow with that Oct 23 08:17:19 (I usually use konqueror) Oct 23 08:20:42 btw. there is GPIO93 in DKIN0 mode and it seems that it is not used Oct 23 08:21:34 could it be the hotsync button? Oct 23 08:21:41 wait, what was DK... direct keypad.. Oct 23 08:21:42 pressing keyboard doesn't affect GPIO40 Oct 23 08:26:55 yep confirmed Oct 23 08:26:58 93 is hotsync button Oct 23 08:27:32 it seems that switching GPIO40 to FFDTR doesn't affect keyboard function :S Oct 23 08:28:33 the only key which should use it 'p' is detected anyway Oct 23 08:30:07 I'm curious what it is attached to then... Oct 23 08:30:19 well, in fact - it can be whatever 1 Oct 23 08:30:35 that's the reason there is no need for GPIO40 Oct 23 08:31:16 hmmm, I'm not sure if there is some dummy GPIO interface :D Oct 23 08:31:51 btw. I can't confirm 93 as Hotsync Oct 23 08:31:54 it's set to AFN0 for me.. Oct 23 08:32:02 you can't? Oct 23 08:32:05 hmm Oct 23 08:32:25 I'm just testing with haret/p Oct 23 08:33:01 90e00008: 0xexxxxxxx Oct 23 08:33:04 90e00008: 0xcxxxxxxx Oct 23 08:33:11 e when hotsync is pressed Oct 23 08:33:15 maybe my math is off Oct 23 08:33:21 that Oct 23 08:33:22 s Oct 23 08:33:23 bit 29 Oct 23 08:33:27 29 + 64 = 93 Oct 23 08:35:09 yeah... 40 definitely AFN0 Oct 23 08:35:29 90e0005c: 60088050 Oct 23 08:35:41 bits 4, 6, 15, 19, 29, 30 set Oct 23 08:35:50 not 15 or 16 Oct 23 08:35:56 I mean Oct 23 08:36:01 not bits 16 or 17 Oct 23 08:36:29 hmm, I think I need some ugly hack for keyboard Oct 23 08:36:43 maybe I can use GPIO_NR_PALMT680_KP_MKOUT5_MD twice :b Oct 23 08:37:07 nope, not out Oct 23 08:37:11 some in :/ Oct 23 08:44:22 gotta go working Oct 23 09:49:10 I moved GPIO for key 'p' to GPIO3 Oct 23 09:49:28 (which is always high) and driver logic stays cool :) Oct 23 09:49:58 changed GPIO40 to FFDTR, flow control is now flawless but I have no response :b Oct 23 09:50:52 but it can be caused by my initialization in _ffuart_configure Oct 23 09:54:12 :/ Oct 23 11:38:43 my dog wants to eat a massive spider that's sitting outside the window :/ Oct 23 11:44:54 :)) Oct 23 12:23:41 hmmm, enabling HW flow control is not working Oct 23 12:23:54 with no flow control it works similarly like before Oct 23 13:27:37 yop Oct 23 13:28:29 yop to you too Oct 23 16:25:45 bonne soirée Oct 23 22:11:38 hey Oct 23 22:41:15 <|miska|> hou Oct 23 22:42:07 :) Oct 23 23:46:02 BobOfDoom: interesting GPIO is maybe also 9 Oct 23 23:46:32 yeah? Oct 23 23:47:06 with phone on it's still high, when phone off - 2 times per second, 1 time per second, 2 times per second Oct 23 23:47:09 ... Oct 23 23:47:30 hmm Oct 23 23:47:46 changing quite slowly Oct 23 23:48:44 I had a look and it can be FFCTS, but AFN is set to 0 Oct 23 23:49:37 hmmm, it's somehow connected with 47 Oct 23 23:50:01 aaah Oct 23 23:50:03 and look at 10 Oct 23 23:50:04 when one is high, the other is too Oct 23 23:50:10 it's output Oct 23 23:50:22 low while phone is off Oct 23 23:50:28 then starts going low/high once phone is on Oct 23 23:50:43 10 looks random Oct 23 23:50:57 at least on my 680 Oct 23 23:51:34 oh Oct 23 23:51:38 hmm Oct 23 23:51:48 I was just looking at your 680-tel Oct 23 23:52:21 oh Oct 23 23:52:26 14 also changes when phone is turned on Oct 23 23:52:31 yup Oct 23 23:52:40 I think that is GSM_HOST_WAKE Oct 23 23:54:36 I added some more http://sleepwalker.hackndev.com/GPIO/ Oct 23 23:57:48 BobOfDoom: you're right Oct 23 23:58:00 it starts to be rando after phone on Oct 23 23:58:25 :D Oct 23 23:58:29 that sounds promising then Oct 23 23:58:33 FFDCD (if AFN1) Oct 23 23:58:36 ? Oct 24 00:26:21 hmm, gpio 10 is independent on gpio 87 (which I thought can be GSM_POWER) Oct 24 00:27:04 strange Oct 24 00:39:09 34 - FFRXD, 35 - FFCTS, 39 - FFTXD, 40 - FFDTR maybe?, 41 - FFDTR, nowhere FFRI, FFDSR, FFDCR, Oct 24 00:39:13 falling asleep Oct 24 00:39:14 gn Oct 24 00:41:52 gn **** ENDING LOGGING AT Fri Oct 24 02:59:57 2008