*** NIN101 has quit IRC | 00:08 | |
*** xes has joined #maemo-ssu | 00:29 | |
*** nox- has joined #maemo-ssu | 00:29 | |
*** usakc has quit IRC | 00:58 | |
*** Martix has joined #maemo-ssu | 01:24 | |
*** xes has quit IRC | 01:32 | |
*** jonwil has joined #maemo-ssu | 02:10 | |
*** MohammadAG has quit IRC | 02:37 | |
*** trx has quit IRC | 03:05 | |
*** trx has joined #maemo-ssu | 03:06 | |
*** trx has joined #maemo-ssu | 03:06 | |
*** Martix has quit IRC | 03:08 | |
*** DrCode has quit IRC | 03:18 | |
*** kolp has quit IRC | 03:21 | |
*** DrCode has joined #maemo-ssu | 03:29 | |
*** DrCode has quit IRC | 04:05 | |
*** DrCode has joined #maemo-ssu | 04:09 | |
*** nox- has quit IRC | 04:13 | |
*** dos1 has quit IRC | 04:31 | |
*** LauRoman has quit IRC | 05:35 | |
*** amiconn_ has joined #maemo-ssu | 05:41 | |
*** amiconn has quit IRC | 05:41 | |
*** amiconn_ is now known as amiconn | 05:41 | |
*** trx has quit IRC | 07:39 | |
*** DrCode has quit IRC | 08:31 | |
*** DrCode has joined #maemo-ssu | 08:34 | |
*** luf has joined #maemo-ssu | 08:56 | |
*** DrCode has quit IRC | 08:59 | |
*** DrCode has joined #maemo-ssu | 09:01 | |
*** freemangordon_ has joined #maemo-ssu | 09:07 | |
*** freemangordon has quit IRC | 09:07 | |
*** tom has quit IRC | 09:18 | |
*** tom has joined #maemo-ssu | 09:39 | |
*** DrCode has quit IRC | 09:41 | |
*** DrCode has joined #maemo-ssu | 09:42 | |
*** LauRoman has joined #maemo-ssu | 10:02 | |
*** tom has quit IRC | 10:17 | |
*** tom has joined #maemo-ssu | 10:17 | |
*** X-Fade has quit IRC | 10:32 | |
*** Martix has joined #maemo-ssu | 10:38 | |
*** LauRoman has quit IRC | 10:41 | |
*** Pali has joined #maemo-ssu | 10:49 | |
DocScrutinizer05 | http://talk.maemo.org/showthread.php?p=1372710#post1372710 | 10:52 |
---|---|---|
*** Martix has quit IRC | 10:53 | |
*** MohammadAG has joined #maemo-ssu | 11:03 | |
FatPhil | Damn, a second of my n900s has a broken USB :( | 11:22 |
*** X-Fade has joined #maemo-ssu | 11:24 | |
FatPhil | "What we need for infra are volunteers, raw manpower" - such as? | 11:30 |
*** dhbiker has quit IRC | 11:38 | |
*** dhbiker has joined #maemo-ssu | 11:43 | |
*** joshgillies has joined #maemo-ssu | 11:43 | |
joshgillies | anyone in here developing on an ARM chromebook? | 11:44 |
joshgillies | looking for pointers toward getting the sdk up and running | 11:45 |
Pali | no chromebook here | 11:48 |
freemangordon_ | Pali: we have a problem with omapfb :( | 11:48 |
freemangordon_ | and dss driver too | 11:48 |
Pali | which problem? | 11:48 |
Pali | cma not working? | 11:49 |
freemangordon_ | CMA actually doesn;t seem to work reliably, so it can't allocate framebuffer memory | 11:49 |
*** freemangordon_ is now known as freemangordon | 11:49 | |
freemangordon | Pali: yes | 11:49 |
Pali | and why preallocate without cma not working? | 11:50 |
freemangordon | Pali: it is removed ;) | 11:50 |
freemangordon | no more vram= boot parameter | 11:50 |
Pali | ah... | 11:50 |
freemangordon | and dss: | 11:50 |
freemangordon | Jan 1 06:35:24 Nokia-N900 kernel: [ 2172.539093] omapdss DISPC error: horizontal timing too tight | 11:50 |
freemangordon | Jan 1 06:35:24 Nokia-N900 kernel: [ 2172.545074] omapdss APPLY error: dispc_ovl_setup failed for ovl 1 | 11:50 |
freemangordon | this happens for both 720p and lower res videos | 11:51 |
freemangordon | (in case CMA succeeds :D ) | 11:51 |
freemangordon | Pali: but on the bright side, seems DSP is tock stable | 11:52 |
freemangordon | *rock | 11:52 |
Pali | ok, so if you disable cma low res videos working? | 11:55 |
freemangordon | Pali: makes no difference, as then memory cannot be allocated because of the fragmentation | 11:55 |
freemangordon | so CMA is the way, but it seems broken | 11:56 |
Pali | and reverting preallocate? | 12:00 |
freemangordon | didn't try it, but I guess it will fix the problem | 12:02 |
*** LaoLang_cool has joined #maemo-ssu | 12:26 | |
*** LaoLang_cool has quit IRC | 12:32 | |
*** kolp has joined #maemo-ssu | 12:43 | |
*** Pali has quit IRC | 13:09 | |
DocScrutinizer05 | FatPhil: such as somebody looking into tmo karma and fixing it | 13:28 |
kolp | I wonder if the size of the current active community really necessitates the existence of a community-voted council. Wouldn't it be enough to have an admin team appointed by HiFo org? Just musing... :) | 13:31 |
kolp | Hm, wrong channel.. | 13:32 |
*** lizardo has joined #maemo-ssu | 13:37 | |
*** joshgillies has quit IRC | 13:45 | |
*** joshgillies has joined #maemo-ssu | 13:53 | |
*** Pali has joined #maemo-ssu | 15:25 | |
*** LauRoman has joined #maemo-ssu | 15:39 | |
*** joshgillies has quit IRC | 15:49 | |
freemangordon | ~ping | 16:13 |
infobot | ~pong | 16:13 |
freemangordon | Pali: how am I supposed to report that omapfb problem on lkml? | 16:26 |
jonwil | ok, so at this point for audio, there are 4 things I dont know and want to know, firstly is how the audio system (PulseAudio specifically) knows the difference between e.g. music playback and VoIP audio, second is how the Bluetooth audio is fed into PulseAudio (or fed out of PulseAudio), the third is how the FMTX/FMRX audio is fed into/out of PulseAudio and the last is whether there exists a... | 16:34 |
jonwil | ...nice "Pulseaudio for Dummies" type book out there :) | 16:35 |
Pali | freemangordon: run ./scripts/get_maintainers.pl -f /path/to/omapfb and check who is maintainer | 16:42 |
Pali | and send email to maintainers plus maybe cc lkml | 16:42 |
freemangordon | Pali: ok, thanks | 16:42 |
Pali | jonwil: first: it using library libplayback | 16:43 |
Pali | in this library you can tell type (voice call, music, other...) | 16:43 |
Pali | libplayback using dbus to inform ohmd about type and also set some PA property on used PA stream | 16:44 |
freemangordon | Pali: hmm, do we have anything major but cameras that doesn't work in 3.12? | 16:44 |
Pali | jonwil: second: via native bluetooth PA plugin | 16:45 |
Pali | PA somehow connect to bluetoothd (dbus/socket) | 16:45 |
jonwil | so libplayback specifically talks to pulseaudio? | 16:45 |
Pali | this is fully documented somewhere and it is standard solution on linux desktop | 16:45 |
jonwil | or does ohmd talk to pulseaudio somewhere? | 16:46 |
Pali | jonwil: probably yes | 16:46 |
jonwil | ok | 16:46 |
Pali | ohmd can talk to PA | 16:46 |
Pali | and also libplayback can talk too | 16:46 |
Pali | freemangordon: camera, sound, maybe jack | 16:46 |
jonwil | what about FM? | 16:47 |
Pali | and fmtx/fmrx is not tested | 16:47 |
jonwil | ok | 16:47 |
Pali | but drivers are there | 16:47 |
freemangordon | Pali: I was able to get sound through BT :) some days ago | 16:47 |
Pali | fmrx driver is same as in fremantle kernel | 16:47 |
Pali | (I checked it) | 16:47 |
jonwil | I mean how does FMTX/FMRX end up in PulseAudio? | 16:48 |
Pali | and bluetooth driver was tested by skry and it worked | 16:48 |
freemangordon | Pali: though I see lots of "unrecognized HCI packet) | 16:48 |
Pali | jonwil: fm and PA on n900 is totally unknown for me | 16:48 |
jonwil | ok | 16:48 |
jonwil | FMTX goes though fmtx-middleware | 16:48 |
jonwil | so I should probably start there | 16:49 |
freemangordon | jonwil: RE fmtxd :) | 16:49 |
Pali | but I do not understand why middleware is needed | 16:49 |
Pali | fmtx should be /dev/radio device | 16:49 |
freemangordon | Pali: it controls stuff like USB cable connected, etc | 16:49 |
Pali | and alsa has radio plugin | 16:49 |
freemangordon | also regional frequencies | 16:49 |
Pali | so PA can use normal alsa interface for radio support | 16:50 |
Pali | and it should act as sound card... | 16:50 |
Pali | freemangordon: only that? or it also routing audio? | 16:50 |
Pali | freemangordon: also need to test bluetooth if working | 16:50 |
freemangordon | not sure, but I *think* it is not fmtxd to route. might be wrong, didn;t look at it | 16:51 |
Pali | and need to test if wifi adhoc mode working | 16:51 |
Pali | all audio on maemo is big mess | 16:51 |
jonwil | btw, I have thought a bit and have realized that the idea to pretend that the Option modem audio is voip and not cellular audio (at least to Pulseaudio) is a good idea. | 16:51 |
Pali | alsa, PA, PA modules, alsa policy, PA policy, alsaped, pasr, ohmd, ohm plugins, libplayback, swi-prolog, prolog code, dresd, dres daemon... | 16:53 |
Pali | and other stuff which I forgot | 16:53 |
jonwil | btw pali are you 100% sure that your decompile of the policy-settings-rx51 prolog code is functionally identical to the stock compiled prolog code? | 16:55 |
Pali | I'm 99% sure | 16:55 |
jonwil | ok, in that case all we need to do is to write a decompiler for policy.dresc :) | 16:56 |
freemangordon | http://talk.maemo.org/showpost.php?p=1379819&postcount=72 | 16:56 |
Pali | there is 1% chance that I forgot something (some non predicate prolog code) | 16:56 |
Pali | jonwil: do we need it? | 16:56 |
freemangordon | feel free to correct me if I've write something which is not correct | 16:56 |
freemangordon | *written | 16:57 |
Pali | jonwil: one nokia dev wrote on TMO some info about dres and dres compiler | 16:57 |
jonwil | yes he did | 16:57 |
jonwil | I remember it | 16:57 |
Pali | jonwil: you can try contact him again, if he does not have any usefull stuff (e.g. decompiler :-)) | 17:01 |
kolp | Getting rid of prolog sounds like a worthwile goal :) | 17:01 |
jonwil | I doubt he would have a decompiler as there would be no reason to write one | 17:01 |
FatPhil | freemangordon: where can I pull your kernel? | 17:01 |
Pali | jonwil: instead writing decompiler it should be easier to load that dres binary file into dres VM and dump factoid info into debug console | 17:02 |
Pali | jonwil: if I remember correctly in dres file are only some functions and lot of static data... | 17:03 |
Pali | and that debug dump command doing some "disassembling" of current memory state | 17:04 |
Pali | and dres VM is stack based - this could help you | 17:04 |
Pali | freemangordon: what is needed is to patch all maemo daemons to work without that kernel patch for sysfs | 17:06 |
freemangordon | Pali: which patch? | 17:06 |
freemangordon | gpio? | 17:06 |
freemangordon | ooh | 17:06 |
Pali | no select syscall | 17:06 |
freemangordon | yeah | 17:07 |
freemangordon | FatPhil: a second | 17:07 |
Pali | but that omap platform gpio driver too | 17:07 |
Pali | needs to be rewritten | 17:07 |
Pali | or removing using it from maemo apps... | 17:07 |
freemangordon | FatPhil: this is upstream (Pali :P ) https://gitorious.org/linux-n900/linux-n900/source/683cc15ed3e6bceb51472a530244f72fa8d40b52: | 17:07 |
freemangordon | FatPhil: this is mine tree ( a little bit ahead) https://gitorious.org/linux-n900/freemangordons-linux-n900/source/777f4ace4c58799575391833db48050ceefe6fd6: | 17:08 |
jonwil | Which maemo apps are using this particular driver? | 17:08 |
jonwil | the GPIO one | 17:08 |
Pali | jonwil: sscd :-( | 17:08 |
jonwil | ugh | 17:08 |
Pali | mce | 17:08 |
Pali | preinit | 17:08 |
freemangordon | Pali: sscd is small binary | 17:09 |
freemangordon | and we know what it does | 17:09 |
Pali | sscd is using it for cmt_ gpios for init | 17:09 |
freemangordon | mce... well... :) | 17:09 |
freemangordon | Pali: :nod: | 17:09 |
jonwil | I wouldn't call 100k a small binary in my book | 17:09 |
freemangordon | jonwil: sscd is 100K? | 17:09 |
Pali | and every app which using any gpio directly | 17:10 |
jonwil | yes its 108,836 bytes | 17:10 |
freemangordon | omg, my bad | 17:10 |
Pali | (do not know which other) | 17:10 |
freemangordon | Pali: well, I guess we can live with that gpio | 17:10 |
Pali | just call grep /sys/something on maemo | 17:10 |
Pali | ok | 17:10 |
freemangordon | for the select thingie, we can LD_PRELOAD | 17:11 |
Pali | ok | 17:11 |
Pali | freemangordon: what is problem with front camera? | 17:12 |
Pali | still green rectangle? | 17:12 |
freemangordon | yep | 17:13 |
Pali | or something changed after fixing IRQ? | 17:13 |
freemangordon | nothing | 17:13 |
freemangordon | ISP irqs are just fine | 17:13 |
freemangordon | it is something in the ISP driver, I hope sailus ti help us find it | 17:13 |
Pali | freemangordon: I remember that somebody wrote that camera worked on meego with meego 37+ kernel | 17:13 |
Pali | maybe looking into meego n900 kernel tree... | 17:14 |
freemangordon | Pali: there is a long way between 2.6.37 and 3.12 as you probably know :P | 17:14 |
Pali | yes | 17:14 |
freemangordon | Pali: looked, the driver is the same | 17:14 |
Pali | but checking if we not miss some patch which is not in upstream... | 17:14 |
Pali | ok | 17:14 |
freemangordon | Pali: well, it is possible I've missed something | 17:15 |
freemangordon | Pali: BTW 3.12 boots the screen in 32bpp by default :) | 17:17 |
*** dos1 has joined #maemo-ssu | 17:17 | |
freemangordon | I have fix, still not pushed | 17:17 |
Pali | and what should be correct? | 17:17 |
freemangordon | 16 | 17:17 |
freemangordon | or even 15, not sure | 17:18 |
Pali | ~seen pavelm | 17:18 |
infobot | pavelm <~pavel@199.214.broadband13.iol.cz> was last seen on IRC in channel #maemo, 246d 1h 59m 41s ago, saying: 'Wow. Thanks a lot, it is installing now,'. | 17:18 |
freemangordon | Pali: we need omapfb.mode=lcd:848x480-16 in cmdline | 17:18 |
Pali | cannot be mode passed from board data? | 17:19 |
freemangordon | what about DT? :P | 17:19 |
Pali | or DT? | 17:19 |
freemangordon | not sure about DT | 17:19 |
Pali | I think resolution is already in board file | 17:20 |
freemangordon | I dont see it https://gitorious.org/linux-n900/freemangordons-linux-n900/source/777f4ace4c58799575391833db48050ceefe6fd6:arch/arm/mach-omap2/board-rx51-video.c | 17:21 |
freemangordon | Pali: BTW leds still doesn't work | 17:24 |
freemangordon | well, RGB led | 17:24 |
freemangordon | Pali: and we have a problem that after rebooting from 2.6.28 to 3.12 it seems one of the i2c busses doesn't work, at least cameras and audio drivers refuse to load | 17:26 |
jonwil | ok, I think my next target is to figure out whats going on inside libplayback. Which might not be that hard since its got x86 binaries plus a -dev package that contains a libplayback.a static librariy | 17:30 |
jonwil | libplayback-1.a that is | 17:30 |
jonwil | and the x86 binary .so file is only 21k | 17:31 |
jonwil | all I have to figure out now is if there is any useful debugging info in that .a file | 17:31 |
Pali | jonwil: if you RE libplayback please write some documentation... | 17:34 |
Pali | see maemo wiki, there are already some links for libplayback | 17:35 |
Pali | freemangordon: I will look on led driver | 17:35 |
freemangordon | FatPhil: but in order to boot maemo you need some userspace packages installed from cssu-devel and some help from Pali :) | 17:45 |
freemangordon | and maybe from me | 17:45 |
Pali | s/some/all/ | 17:46 |
freemangordon | :D | 17:46 |
freemangordon | Pali: well, clock-ui is not needed I guess. | 17:46 |
jonwil | god I hate when libraries have an internal data structure and just pass around pointers to it. | 17:48 |
Pali | welcome to OOP world :-) | 17:49 |
jonwil | interesting, libplayback says call and voip audio are the same thing. So how does pulseaudio know whether to route audio to the cellular modem or through the voip logic? | 17:50 |
Pali | application name... | 17:51 |
jonwil | what handles that though? pulseaudio? ohm/policy/prolog? something else? | 17:51 |
Pali | we already know that application with name "mediaplayer" can something do what other cannot | 17:52 |
Pali | if I remember it is alsaped together with prolog and dres | 17:52 |
Pali | and there also is some config ini file where is mediaplayer written | 17:52 |
freemangordon | hmm, there is a huge plicy file somewhere in /etc | 17:52 |
freemangordon | *policy | 17:52 |
Pali | jonwil: and maybe really voip and gsm calls are same for PA | 17:53 |
jonwil | no they are not, PA needs to know the difference to know whether to feed the audio to libcmtspeech or not | 17:53 |
jonwil | aha, Policy-application-detector relates to this | 17:54 |
Pali | right that detector is other audio application | 17:54 |
jonwil | its what grabs the name and feeds it to the policy ststem | 17:54 |
jonwil | system | 17:54 |
Pali | jonwil: if I remember correctly it talk to some udp port on localhost to ohmd | 17:54 |
Pali | and it is small daemon | 17:54 |
Pali | can you look if you can RE it? | 17:55 |
jonwil | I will look | 17:55 |
jonwil | aha, the big conf file you mentioned looks like xpolicy.conf | 17:55 |
Pali | it sending something like "pop up" or "pop down" via udp | 17:55 |
Pali | and it checking for X window number... | 17:56 |
Pali | that config file is maybe in git repo where is decompiled prolog file | 17:56 |
Pali | and I think I already modified it | 17:57 |
jonwil | ok, yeah the file is definatly xpolicy.conf | 17:59 |
jonwil | going to look at policty-application-detector | 18:00 |
jonwil | ok, app-detect seems to be looking for rtcom-call-ui and systemui by name | 18:01 |
jonwil | its sending to port 3001 and 3002 | 18:04 |
Pali | you can download ohms sources (+ plugins) | 18:06 |
Pali | look at fremantle closed packages for correct git commit | 18:06 |
Pali | and you can see what ohmd doing with it | 18:06 |
jonwil | yes I know | 18:09 |
jonwil | ok, looks like I will need to find a good tutorial or guide on sources and sinks in pulseaudio | 18:19 |
*** jonwil has quit IRC | 18:19 | |
FatPhil | Pali: I'm going to reserve one of my "junker" n900s for kernel hacking. I'll nuke it before I do anything - what FIASCOs should I start with? | 18:29 |
* FatPhil is unhappy that 2 of his n900s are "junker"s :-( But at least one is fine for dev work presently. | 18:31 | |
freemangordon | FatPhil: flash it with stock pr1.3, install cssu-testing(or -thumb) and use Pali's script to copy maemo to /home/maemo5 | 18:31 |
*** LaoLang_cool has joined #maemo-ssu | 18:31 | |
FatPhil | freemangordon: OK, I'll get back to you in a few hours... | 18:32 |
FatPhil | What's the week/release number of "stock" pr1.3? | 18:32 |
freemangordon | FatPhil: oh, and before executing Pali's copy script, enable cssu-devel repo and do apt-get update/upgrade | 18:32 |
freemangordon | 2-36 iirc | 18:32 |
freemangordon | or was it 36-2?!? | 18:32 |
*** XDS2010 has quit IRC | 18:33 | |
freemangordon | FatPhil: keep in mind that both trees (mine and Pali's) lack an et8ek8 patch, i'll provide it to you via pastebin | 18:33 |
kolp | 36-2 | 18:33 |
*** LaoLang_cool has quit IRC | 18:34 | |
freemangordon | FatPhil: http://pastebin.com/Xw71qprL | 18:34 |
*** lenoch has joined #maemo-ssu | 18:35 | |
FatPhil | I swear I did that on my daily device, but can't find the fiascos now | 18:36 |
FatPhil | brain (or filesystem) like a sieve | 18:36 |
freemangordon | flashing | 18:38 |
freemangordon | ~flashing | 18:38 |
infobot | it has been said that maemo-flashing is http://wiki.maemo.org/Updating_the_tablet_firmware | 18:38 |
Pali | tablets-dev.nokia.com | 18:38 |
FatPhil | which would you prefer I do, -thumb, or non-thumb? | 18:38 |
freemangordon | FatPhil: -thumb is faster | 18:38 |
FatPhil | I'd rather be at the cutting edge, so -thumb | 18:38 |
freemangordon | otherwise it doesn;t really matter | 18:38 |
*** NIN101 has joined #maemo-ssu | 18:39 | |
FatPhil | IT would be good to test on both thumb and non-thumb, I'll try to find (or solder together) another dev device so I can do both | 18:39 |
freemangordon | FatPhil: test what? | 18:40 |
freemangordon | it is not even in alpha stage (kernel) | 18:40 |
FatPhil | test the kernel, test it boots to /init, after that, the kernel's job is over ;-) | 18:50 |
FatPhil | Userspace is evil and should be killed! | 18:50 |
freemangordon | with fire :D | 18:52 |
FatPhil | with signal number 9, why give it any warning!? | 18:56 |
FatPhil | This is going to sound silly - how do you flash a device that's got uboot on it? Flasher never sees the USB device, it just boots to userspace | 18:59 |
freemangordon | FatPhil: hmm, u-boot is after nolo so it shouldn't make any difference | 19:00 |
freemangordon | FatPhil: do as I say: | 19:01 |
freemangordon | 1. remove the battery | 19:02 |
freemangordon | 2: plug the USB cable | 19:02 |
freemangordon | press'n'hold U | 19:02 |
FatPhil | I'm on the jig, so battery removal is a switch | 19:02 |
FatPhil | ooooh | 19:02 |
freemangordon | 4. insert the batteru | 19:02 |
freemangordon | missing the 'U' step? | 19:02 |
FatPhil | what R&D flag state should I be in? | 19:04 |
freemangordon | doesn;t matter iirc | 19:04 |
FatPhil | U's not making any difference | 19:05 |
FatPhil | AH - cold flash! | 19:05 |
FatPhil | I'd forgotten about thayt | 19:05 |
freemangordon | why cold flash? | 19:05 |
freemangordon | your NOLO is broken? | 19:05 |
FatPhil | Hmm, I don't have a signed second | 19:06 |
FatPhil | Not sure, but flashing ain't working | 19:06 |
freemangordon | FatPhil: if flasher can;t see it, try another USB port or even USB hub | 19:06 |
FatPhil | flasher can read the R&D flags | 19:07 |
freemangordon | but,but... then it sees it, what is the problem? | 19:07 |
FatPhil | it doesn't see it when I try to flash | 19:07 |
freemangordon | wrong flasher parameters? | 19:07 |
freemangordon | flasher -F fiasko -f -R | 19:08 |
freemangordon | (iirc) | 19:08 |
FatPhil | yeah | 19:08 |
freemangordon | doesn;t make any sense | 19:08 |
Pali | nolo checking for BSI of battery | 19:08 |
freemangordon | oh :) | 19:08 |
Pali | and if BSI or charge level is loo low refuse flashing | 19:09 |
* freemangordon thinks Pali is a smart guy | 19:09 | |
freemangordon | :D | 19:09 |
Pali | s/charge level/voltage level/ | 19:09 |
infobot | Pali meant: and if BSI or voltage level is loo low refuse flashing | 19:09 |
Pali | I wrote new version of 0xFFFF flasher :-) | 19:10 |
freemangordon | Pali: hmm, IMO those checks are sane | 19:10 |
Pali | but if you load NOLO from cold flash, then it disable these checks | 19:10 |
freemangordon | we don;t know what will happen if you run out of juice in the middle of cmt FW flashing | 19:11 |
freemangordon | broken BB5 I guess | 19:11 |
FatPhil | yeah, I it doesn't always do the query-rd-mode either | 19:13 |
FatPhil | [ 0.145] Battery voltage 4.104 V, BSI: 1021 | 19:14 |
FatPhil | hmm, well, that's a jig | 19:15 |
FatPhil | ANyone got a copy of secondary for a cold flash? | 19:15 |
kolp | When flashing the N9 the flasher waits and charges the battery for a while if it doesn't have enough juice. Is there any chance something like that can be done for the N(eo)900 in the future? | 19:16 |
*** luf has quit IRC | 19:16 | |
freemangordon | FatPhil: secondary is in fiasko, you can extract it | 19:17 |
dos1 | kolp: NOLO is propertary, so "nope" for N900 | 19:18 |
dos1 | but in Neo900, whole stack will be open as in GTA04, so it will be there if someone implements it | 19:18 |
kolp | Oh, it's a nolo thing | 19:18 |
kolp | Ok, thx :) | 19:18 |
FatPhil | I think it's dodgy USB connector :-( | 19:27 |
kerio | :( | 19:32 |
FatPhil | NP, I never gave my g/f her replacement device - I've just made it my dev device! | 19:33 |
FatPhil | I've promised to do surgery to her device to put a new screen and frame on it, as otherwise her device is fine | 19:34 |
FatPhil | I'd like to dedicate this flashing to my ex-boss, Timo Jokiaho, who sold me this device for 8 (rather good) beers! | 19:35 |
FatPhil | He doesn't have to worry about any of his personal information still being on it anywhere, put it that way | 19:36 |
merlin1991 | well for that you'll have to flash emmc aswell | 19:36 |
freemangordon | merlin1991: hi! | 19:37 |
freemangordon | merlin1991: nuw cssu soon? | 19:37 |
freemangordon | *new | 19:37 |
Pali | merlin1991 are you alive? :-) | 19:38 |
freemangordon | Pali: that could be bot | 19:38 |
merlin1991 | Pali: I've been reading this channel regulary since a week or so | 19:38 |
FatPhil | merlin1991: did that first | 19:38 |
merlin1991 | never had a point though where I could jump in :D | 19:38 |
merlin1991 | for new cssu I'll start gathering stuff this weekend | 19:39 |
freemangordon | Pali: lets use turing test on him :) | 19:39 |
merlin1991 | my storage unit isn't endless | 19:39 |
Pali | turing test has nothing with storage size :-) | 19:39 |
freemangordon | :nod: | 19:40 |
freemangordon | ~turing | 19:40 |
infobot | ...but I *AM* Turing complete!!!! | 19:40 |
merlin1991 | iirc If a machine is turing complete you still need storage for instructions relative to the complexity of your problem | 19:40 |
freemangordon | infobot: no, you are not | 19:40 |
infobot | freemangordon: I think you lost me on that one | 19:40 |
freemangordon | merlin1991: turing test doesn;t mean turing complete (iirc) | 19:41 |
freemangordon | http://en.wikipedia.org/wiki/Turing_test | 19:41 |
Pali | in turing test you have to decide if you are talking with robot or real peron | 19:41 |
merlin1991 | was turing test then the one with humans? | 19:41 |
merlin1991 | ah yep :) | 19:41 |
freemangordon | yep | 19:41 |
merlin1991 | btw all this kernel hacking, do you plan on bringing our userspace running on upstream+patches, or what is the ultimate goal | 19:45 |
freemangordon | I guess | 19:45 |
freemangordon | merlin1991: *if* we had to support neo900, I don;t see why we shouldn't support n900 as well | 19:46 |
freemangordon | not to say it will be much easier | 19:46 |
freemangordon | merlin1991: does it make sense to you? | 19:47 |
merlin1991 | in a way | 19:47 |
freemangordon | merlin1991: I really hope neo900 will bring some developers back to fremantle | 19:48 |
merlin1991 | heh, I'd happily go back to a n900 that is faster | 19:48 |
merlin1991 | I'd finally have my openvpn relay again, proper ssh with a keyboard, ... | 19:48 |
freemangordon | merlin1991: whaaat? n900 is not your primary device ?!? | 19:49 |
merlin1991 | nope I'm on the dark side (n9) | 19:49 |
freemangordon | hmm, this isn't dark side IMO, this is grey side | 19:50 |
merlin1991 | it's almost black hat xD | 19:50 |
freemangordon | you have the worst from the both worlds :D | 19:50 |
merlin1991 | my main reason for using the n9 is, that the modem in there still works properly :D | 19:50 |
merlin1991 | with my n950, and all 3 n900s I loose connection all the time | 19:51 |
freemangordon | merlin1991: but wy don;t buy another n900, those are cheap these days, I heard of people getting one fo a couple of beers :D | 19:51 |
freemangordon | merlin1991: hmm, tried to change the SIM? | 19:51 |
merlin1991 | I have 2 sims, both work fine in the n9 but fail in all the others | 19:52 |
freemangordon | weird | 19:52 |
FatPhil | My daily device was bought for 7 beers, the development device was bought for 8. ~30e each in real money | 19:53 |
freemangordon | what?!? 4 euros a beer | 19:53 |
FatPhil | YEah, I bought them in Estonia rather than Finland. | 19:55 |
* FatPhil spent 120e on beer on thursday night (last but one day at work in Finland) | 19:56 | |
FatPhil | still got to work bright and early on my last day, clearly didn't drink enough | 19:56 |
freemangordon | :D | 19:56 |
merlin1991 | hm 120€ in beer | 19:57 |
merlin1991 | I'd be dead if I try that here | 19:58 |
freemangordon | well, what are friends for? | 19:58 |
merlin1991 | that would be like 40 beers | 19:58 |
freemangordon | (about 120 here :P) | 19:58 |
freemangordon | ina abar | 19:59 |
FatPhil | Fancy imported beers, up to 25e a bottle | 19:59 |
freemangordon | OMG | 19:59 |
FatPhil | Welcome to Finland | 20:00 |
merlin1991 | been there, saw the prices :D | 20:00 |
FatPhil | Are devices with broken USB of use to any developers? How do you flash if not over USB? (there ought to be a SSD boot!) | 20:02 |
dos1 | µsd boot works with u-boot | 20:04 |
dos1 | (but probably not with fremantle :)) | 20:04 |
merlin1991 | I still need to fix my sisters usb port, so I can get my only functioning n900 back xD | 20:05 |
merlin1991 | I have the parts, but not the soldering equipment :/ | 20:06 |
FatPhil | me too. now have 1 broken, and 1 very dodgy usb device | 20:08 |
Pali | there should be way to flash device via serial console | 20:17 |
*** X-Fade has quit IRC | 20:17 | |
Pali | at least flasher-3.5 has support for it | 20:17 |
Pali | flasher-3.5 can cold flash device via serial console | 20:17 |
Pali | but it maybe needs some (unreleased) 2nd x-loader image which can load secondary (nolo) image via serial port too (and not via usb which doing normal cold flashing) | 20:19 |
FatPhil | Well, in order to dick with package repo's easily, I'm gonna need wifi, and that means a pub! | 20:30 |
FatPhil | Pali: gonna have to look into that. Not sure how to control flasher, it asks for USB ids. | 20:32 |
*** XDS2010 has joined #maemo-ssu | 20:33 | |
sixwheeledbeast | did someone say pub? :P | 20:33 |
Pali | you need to use cold flashing (-c param) | 20:35 |
DocScrutinizer05 | Pali: I guess xloader is capable of serial, but I dunno for NOLO | 20:52 |
DocScrutinizer05 | ooh wait, freemangordon told me it's a combo-xloader-nolo image that gets simply extended on loading | 20:53 |
DocScrutinizer05 | or was it jacekowski? | 20:54 |
Pali | cold flashing: omap bootrom will do handshake via usb with computer and then computer send signed bootable image (2nd) and boot it | 20:55 |
Pali | 2nd will read from usb secondary image which will flash to nand | 20:55 |
Pali | and then also boot it and do not check voltage and bsi | 20:56 |
Pali | and immediately enter to flashing mode (no need for U key) | 20:56 |
Pali | omap bootrom support more boot methods (nand, mmc, usb) and one is also serial console | 20:57 |
Pali | this is written somewhere in public trm | 20:57 |
freemangordon | at the end :D | 20:58 |
Pali | I used all this info when I was implementing cold flasher in 0xFFFF flasher | 20:58 |
Pali | and if I remember it is possible to change order of booting | 20:58 |
Pali | so maybe we can boot (signed) xloader from eMMC/SD | 20:59 |
DocScrutinizer05 | you can't change boot order. it's "hardwired" | 21:00 |
Pali | but due to that there is no doc for HS devices we do not know format in which xloader needs to be stored (and where) | 21:00 |
Pali | for GP devices you *can* configure boot orde | 21:00 |
Pali | it is written in trm | 21:00 |
DocScrutinizer05 | and we know that xloader needs to be signed, so any docs on format are useless | 21:00 |
Pali | but do not know if this also work for HS | 21:01 |
Pali | no we need docs where and how to store signed xloader | 21:01 |
DocScrutinizer05 | afaik "bootorder" is a few bits in EFUSE | 21:01 |
DocScrutinizer05 | only relecant for BOOTROM | 21:01 |
Pali | for example for GP devices, bootloader needs to be stored on mmc first fat partion under name MLO | 21:02 |
DocScrutinizer05 | yes, that will be same I guess | 21:02 |
Pali | but I think for HS devices it will be another name... | 21:02 |
DocScrutinizer05 | and not mmc, but uSD | 21:02 |
Pali | in trm was any mmc device | 21:03 |
DocScrutinizer05 | no, it only checks mmc1 | 21:03 |
Pali | mmc1 is eMMC | 21:03 |
DocScrutinizer05 | which happens to be uSD, for exactly this reason | 21:03 |
DocScrutinizer05 | no, mmc1 is uSD | 21:03 |
Pali | mmc0 is SD | 21:03 |
DocScrutinizer05 | it gets renamed later | 21:04 |
Pali | only maemo has swapped it | 21:04 |
DocScrutinizer05 | hw its mmc1 and mmc2 | 21:04 |
DocScrutinizer05 | iirc | 21:04 |
Pali | ok, maybe numbering starts from 1 (and not 0) | 21:04 |
DocScrutinizer05 | anyway bootrom looks on FIRST mmc jw interface for a file called MLO | 21:04 |
Pali | but first is SD card | 21:04 |
Pali | in trm is written that it can load also from raw partition | 21:05 |
DocScrutinizer05 | maybe | 21:05 |
DocScrutinizer05 | the EE design rationale been that you can boot from uSD | 21:06 |
DocScrutinizer05 | that's why first mmc hw interface is uSD and not eMMC | 21:06 |
Pali | via usb you can tell omap bootrom to load/boot from mmc | 21:06 |
Pali | like to load image via usb | 21:07 |
DocScrutinizer05 | via USB you can't tell anything, you just can load xloader from USB | 21:07 |
Pali | maybe we can try to dd xloader to mmc as raw partition if it work... | 21:07 |
Pali | you can do it! | 21:07 |
*** X-Fade has joined #maemo-ssu | 21:07 | |
Pali | look at my cold flasher source code and comments | 21:07 |
Pali | and trm | 21:08 |
DocScrutinizer05 | BOOTROM simply checks the "hardwired" (by EFUSE) devices for xloader, and loads the fist it finds | 21:08 |
DocScrutinizer05 | first* | 21:08 |
DocScrutinizer05 | everything beyond that is a matter of xloader you load, afaik | 21:08 |
Pali | but before that it waits some time for usb commands | 21:08 |
DocScrutinizer05 | sure you can load an xloader that tries to load NOLO via WIFI | 21:09 |
freemangordon | and that is how coldflashing works | 21:09 |
Pali | no no | 21:09 |
DocScrutinizer05 | basically, if xloader wasn't that tiny | 21:09 |
Pali | you can tell bootrom before booting xloader to change boot order | 21:09 |
Pali | and boot from other device | 21:09 |
freemangordon | Pali: no? in coldfalshing mode n900 appears like some strange USB device | 21:09 |
DocScrutinizer05 | I had an extensive look at TRM | 21:09 |
Pali | and if you want to load image via usb, you send command via usb thay you want to boot from usb | 21:10 |
Pali | I written that code in 0xFFFF, so I know it | 21:10 |
DocScrutinizer05 | and at http://www.omappedia.org/wiki/Bootloader_Project | 21:10 |
Pali | and know how cold flashing work | 21:11 |
freemangordon | ok :) | 21:11 |
DocScrutinizer05 | SYSBOOT Pins | 21:11 |
freemangordon | I was under impression that coldflashing is doen with the help of bootrom, obviously wrongly :) | 21:12 |
DocScrutinizer05 | The internal ROM Code can attempt to boot from several different peripheral and memory devices, including, but not limited to: Serial (UART3), SD Card, eMMC, NAND, and USB. The order in which these devices are searched for a valid first-stage booting image (x-loader) is determine by a set of GPIO configuration pins referred to as SYSBOOT. | 21:12 |
DocScrutinizer05 | so not even EFUSE | 21:12 |
DocScrutinizer05 | actually hardwired | 21:12 |
Pali | but before that is used usb! | 21:12 |
DocScrutinizer05 | nothing you can 2tell it2 via USB | 21:12 |
Pali | and if you send usb command it can overwrite that is in sysboot | 21:13 |
DocScrutinizer05 | no | 21:13 |
Pali | yes | 21:13 |
freemangordon | Pali: wait, it could be hardwired to try USB first | 21:13 |
Pali | did you programmed already that? | 21:13 |
DocScrutinizer05 | [citation needed] | 21:13 |
Pali | I yes | 21:13 |
freemangordon | or there is a backdoor which is not in publick TRM | 21:14 |
Pali | it was written in version of trm which I read when I worked on 0xFFFF | 21:14 |
DocScrutinizer05 | I never heard of such command, nowhere | 21:14 |
Pali | download 0xFFFF sources | 21:14 |
freemangordon | Pali: which TRM is that? | 21:14 |
Pali | go to src/cold-flash.c | 21:14 |
DocScrutinizer05 | and I read the TRM as well as pages like http://www.omappedia.org/wiki/Bootloader_Project | 21:14 |
Pali | there should be written version of TRM | 21:14 |
Pali | and also all commands for usb!!! | 21:15 |
DocScrutinizer05 | ROMBOOT doesn't take commands | 21:15 |
Pali | as #define | 21:15 |
DocScrutinizer05 | that's maybe xloader, or more likely even NOLO | 21:15 |
Pali | try to find in trm something like peripheral booting | 21:15 |
Pali | it is not xloader/nolo | 21:16 |
DocScrutinizer05 | I *READ* peripheral booting in TRM | 21:16 |
Pali | it is bootrom and how coldflash working | 21:16 |
Pali | find something like "boot next" | 21:16 |
DocScrutinizer05 | it has a weird device table | 21:16 |
Pali | there is "boot next" command which skip entry from device table | 21:17 |
DocScrutinizer05 | so what for? our first entry is uSD anyway | 21:17 |
Pali | find big diagram in trm | 21:18 |
DocScrutinizer05 | see our particular wiring of SYSBOOT pins | 21:18 |
Pali | there was also written how sysboot pin cn be skipped | 21:18 |
DocScrutinizer05 | WHAT FOR??? | 21:18 |
Pali | and start booting from device specified by usb command | 21:18 |
DocScrutinizer05 | even if such thing exists, it's vastly useless | 21:19 |
DocScrutinizer05 | since we already have boot sequence USB,MMC1,NAND | 21:19 |
DocScrutinizer05 | but that's for xloader only | 21:19 |
DocScrutinizer05 | and it's hardwired | 21:20 |
DocScrutinizer05 | oops | 21:21 |
DocScrutinizer05 | sys_boot[5:0]: 0x10 | 21:21 |
DocScrutinizer05 | Boot order: OneNAND USB UART3 MMC1 -- | 21:21 |
Pali | /* Omap MMC/SD2 boot message */ # static const uint32_t omap_mmc2_msg = 0xF0030506; | 21:21 |
DocScrutinizer05 | IroN900:~# omap34xx-boot-order | 21:21 |
DocScrutinizer05 | sys_boot[5:0]: 0x10 | 21:21 |
DocScrutinizer05 | Boot order: OneNAND USB UART3 MMC1 -- | 21:21 |
Pali | /* Omap Peripheral boot message */ static const uint32_t omap_peripheral_msg = 0xF0030002; | 21:23 |
Pali | here are two usb commands for omap bootrom | 21:24 |
Pali | usb commands exists!!!! | 21:24 |
Pali | Omap Boot Messages: spruf98v.pdf (page 3444): OMAP35x Technical Reference Manual - 25.4.5 Periheral Booting | 21:25 |
Pali | this is in 0xFFFF comments | 21:26 |
DocScrutinizer05 | gosub @gc; POP; POP; print " anyway my original statement that you can boot a device with broken USB from uSD been incorrect, as far as xloader goes " | 21:27 |
DocScrutinizer05 | Pali: I stand corrected | 21:27 |
DocScrutinizer05 | spruf98v? W*T*F?? | 21:28 |
DocScrutinizer05 | I know rev F maybe, not V | 21:28 |
Pali | this is what I found and used when worked on 0xFFFF | 21:28 |
Pali | and USB in bootorder is probably usb flash disk connected in usb host mode | 21:29 |
Pali | not usb in peripheral mode... | 21:29 |
Pali | but do not remember | 21:29 |
FatPhil | [in pub] hmmm, cant' connect to any of the default repos (e.g. nokia and extras) | 21:30 |
FatPhil | however, I'm doin the CSSU installation without them | 21:30 |
Pali | change time back to year 2010 | 21:30 |
DocScrutinizer05 | Pali: do you know which version of ROMBL this refers to? and which version N900 has? | 21:31 |
Pali | otherwise nokia repo will not work | 21:31 |
FatPhil | some SSL negotiation issue | 21:31 |
Pali | DocScrutinizer05: do not know version | 21:31 |
Pali | and I cannot opat that big pdf on n900... | 21:31 |
FatPhil | Pali: can you add that to the wiki if it's not there | 21:32 |
Pali | FatPhil: maybe you need to activate internet via web browser? | 21:32 |
*** LauRoman has quit IRC | 21:32 | |
Pali | FatPhil: I'm on n900 and copy/paste lot of parts of text is hard here... | 21:33 |
DocScrutinizer05 | Pali: I still think I haven't noticed those commands in SPRUF98D | 21:33 |
DocScrutinizer05 | they *might* be a feature of later ROMBL versions | 21:33 |
Pali | DocScrutinizer05: I tested that peripheral command and it worked and also next command worked | 21:33 |
DocScrutinizer05 | or I ignored them since I thought they're useless | 21:33 |
DocScrutinizer05 | ooooh :-D | 21:34 |
Pali | peripheral command is used for cold flashing | 21:34 |
Pali | and next to skip cold flashing | 21:34 |
Pali | https://gitorious.org/0xffff/0xffff/source/master:src/cold-flash.c | 21:35 |
DocScrutinizer05 | hah. awesome | 21:35 |
Pali | look at cold flash source code | 21:35 |
Pali | and you can see what is handled by omap bootrom and what by xloader | 21:35 |
DocScrutinizer05 | /note to self: don't mess with Pali ;-) | 21:36 |
FatPhil | Pali: internet's flakey, but I managed to browse to the wiki, install the repo installer, and download 50MB of CSSU, which is happily installing currently | 21:36 |
Pali | do you know if it is possible to detect from asis id (sent by omap bootrom) if device is n900 or not? | 21:39 |
DocScrutinizer05 | Pali: is result of omap34xx-boot-order binary correct? | 21:39 |
DocScrutinizer05 | or is that somehow covered/obfuscated when system got booted up? | 21:40 |
Pali | on GP device should be | 21:40 |
Pali | but I think it could be correct also on HS | 21:40 |
Pali | I think that obfucation does not make sense | 21:41 |
Pali | still it allow booting only signed image... | 21:41 |
*** dos1 has quit IRC | 21:42 | |
*** LauRoman has joined #maemo-ssu | 21:44 | |
*** LauRoman has quit IRC | 21:46 | |
DocScrutinizer05 | ah, maybe some memory re-mapping kicks in after some crap in HS been set | 21:47 |
DocScrutinizer05 | who knows, I don't | 21:47 |
DocScrutinizer05 | Boot order: OneNAND USB UART3 MMC1 -- makes no sense to me | 21:47 |
*** Vlad_on_the_road has joined #maemo-ssu | 21:47 | |
DocScrutinizer05 | lemme check the damn schematics | 21:48 |
Pali | why? | 21:49 |
Pali | its ok | 21:49 |
Pali | it booting from onenand | 21:49 |
DocScrutinizer05 | sys_boot4:1 6:1 5:BOOTMODE | 21:49 |
Pali | usb peripheral boot for cold flash is started by special usb command before that boot order | 21:50 |
DocScrutinizer05 | hmm, makes sense then | 21:50 |
Pali | and usb in that boot order is probably booting from usb flash disk in host mode | 21:51 |
DocScrutinizer05 | LOL, 5:BOOTMODE is determined by CHRG_IND | 21:51 |
Pali | which cannot work on n900 where is no usb host mode port | 21:51 |
DocScrutinizer05 | :nod: | 21:52 |
kolp | Will anything break if I switch on host mode while wall-charging, without having anything else connected to the usb port? | 21:54 |
DocScrutinizer05 | IOW CPU cant boot "normal" as long as steady yellow emergency chage indicator. It's not CPU bootup that stops emergency chaging, it's emergency charging end that boots CPU | 21:54 |
FatPhil | DocScrutinizer05: you understand correctly | 21:55 |
FatPhil | yellow = NOLO charging, and it won't exit that till it's happy | 21:55 |
Pali | where you found that? "DocScrutinizer05: LOL, 5:BOOTMODE is determined by CHRG_IND" | 21:57 |
DocScrutinizer05 | schem p. 4 & 6 & | 21:58 |
DocScrutinizer05 | SoC sys_boot:5 -> (BOOT_MODE) -> N1302 | 21:59 |
DocScrutinizer05 | -> (CHRG_IND) -> N1140 bq24150 | 22:00 |
DocScrutinizer05 | CHRG_IND also directly drives red and green (=yellow) indicator LED | 22:00 |
DocScrutinizer05 | as long as BQ24150 is in emergency charge mode, CPU can't boot "normally" | 22:01 |
*** lenoch has quit IRC | 22:02 | |
Pali | and when bq2415x is in emergency mode? | 22:02 |
Pali | when timer is not resetted in 30s? | 22:02 |
DocScrutinizer05 | yes, and when battery voltage below 3.7V or sth | 22:05 |
DocScrutinizer05 | and for max 30min | 22:05 |
DocScrutinizer05 | FatPhil: I wouldn't call that NOLO charging, since it's an autonomous mode of BQ24150 charger chip, CPU not involved | 22:07 |
DocScrutinizer05 | maybe "NOLO tolerated autonomous hw charging" | 22:08 |
FatPhil | there's a bq24150 driver in it, I presume nolo sets it up, and lets it run | 22:08 |
FatPhil | has anyone decompiled nolo? | 22:09 |
Pali | jacekowski only some small parts | 22:11 |
Pali | its funny that nolo cannot change battery when you want to enter flash mode... | 22:12 |
Pali | but has charging driver... | 22:12 |
Pali | so driver is either useless or not used | 22:12 |
Pali | or not working | 22:12 |
Pali | or need to send some special command to start it? | 22:13 |
DocScrutinizer05 | ~bootloop | 22:18 |
infobot | i heard bootloop is when your device has broken rootfilesystem, so during reboot it fails on some service startup or kernel module load and thus reboots. This *drains* battery! And you can't reflash to stop bootloop when battery is drained. Recharge your battery by other means before reflashing. | 22:18 |
DocScrutinizer05 | ~flatbatrecover | 22:18 |
infobot | Remove battery for 1 minute. Insert battery. Plug powered ***NOKIA WALLCHARGER*** to device. Watch steady amber. Let sit and charge. Do NOT try to boot. After 30 min, you got either a) a booted up N900, b) flashing amber which means you can boot, c) steady amber going off - in this case start over again with ~flatbatrecover. CAVEAT! Only works when ~rootfs is OK (no ~bootloop)!. | 22:18 |
kolp | DocScrutinizer05: in Neo900, would it be possible to export the root fs via USB when powered down? | 22:20 |
kolp | Hm, prolly not because of nand or whatever | 22:21 |
DocScrutinizer05 | that's a fptf question and issue | 22:21 |
kolp | A software thing only? | 22:21 |
DocScrutinizer05 | no hw can export anything when "powered down" | 22:21 |
Pali | if you use uboot, then yes | 22:21 |
kolp | Ah, well, wrong terminology then | 22:21 |
Pali | uboot can support "reverse flash" | 22:22 |
Pali | but uboot does not have working support for n900 onenand yet... | 22:22 |
DocScrutinizer05 | Pali: are you sure N900 NOLO has anything to do with bq24150? | 22:22 |
kolp | Would be useful for fixing bootloops if you could access the root fs without booting to UI | 22:23 |
kolp | And a security issue... | 22:23 |
Pali | I know that n900 NOLO cannot fix dead battery | 22:23 |
Pali | and cause bootloops instead charging enough for flashing | 22:23 |
Pali | DocScrutinizer05: this is why i wrote that nolo bq support is useless... | 22:24 |
FatPhil | Pali, I may be thinking of nolo-h, not nolo-r | 22:24 |
DocScrutinizer05 | Pali: I doubt there *is* *any* bq24150 support in N900 NOLO | 22:25 |
DocScrutinizer05 | kolp: rescueOS can be booted from USB to RAM, and access maemo rootfs | 22:26 |
DocScrutinizer05 | allegedly rescueOS even can charge a battery | 22:26 |
DocScrutinizer05 | and also allegedly it can get ram-loaded even when nolo refuses to enter normal flashing mode | 22:27 |
DocScrutinizer05 | I don't see that | 22:27 |
DocScrutinizer05 | FatPhil: bq24150 doesn't need setup for emergency charging | 22:28 |
DocScrutinizer05 | that's the whole point of that mode | 22:29 |
DocScrutinizer05 | as soon as anything talks to bq24150 via i2c, emergency charging is stopped | 22:29 |
DocScrutinizer05 | and will resume only on chip reset, due to reset command or i2c watchdog (32s) timeout | 22:30 |
FatPhil | ~bootloop | 22:30 |
infobot | bootloop is, like, when your device has broken rootfilesystem, so during reboot it fails on some service startup or kernel module load and thus reboots. This *drains* battery! And you can't reflash to stop bootloop when battery is drained. Recharge your battery by other means before reflashing. | 22:30 |
DocScrutinizer05 | ~factinfo bootloop | 22:30 |
infobot | DocScrutinizer05: there's no such factoid as bootloop | 22:30 |
DocScrutinizer05 | sucker??! | 22:30 |
DocScrutinizer05 | ~bootloop | 22:30 |
infobot | i heard bootloop is when your device has broken rootfilesystem, so during reboot it fails on some service startup or kernel module load and thus reboots. This *drains* battery! And you can't reflash to stop bootloop when battery is drained. Recharge your battery by other means before reflashing. | 22:30 |
DocScrutinizer05 | ~literal bootloop | 22:30 |
infobot | "#maemo bootloop" is "when your device has broken rootfilesystem, so during reboot it fails on some service startup or kernel module load and thus reboots. This *drains* battery! And you can't reflash to stop bootloop when battery is drained. Recharge your battery by other means before reflashing." | 22:30 |
DocScrutinizer05 | ~factinfo #maemo bootloop | 22:31 |
infobot | #maemo bootloop -- created by DocScrutinizer05 <~saturn@openmoko/engineers/joerg> at Thu Oct 3 14:52:17 2013 (7 days); last modified at Thu Oct 3 15:06:02 2013 by DocScrutinizer05!~saturn@openmoko/engineers/joerg; it has been requested 4 times, last by DocScrutinizer05, 17s ago. | 22:31 |
DocScrutinizer05 | FatPhil: ok, it's a bit fuzzy - actually emergency charging *will* work even with broken rootfs, but it will not result in NOLO allowing you to flash | 22:32 |
DocScrutinizer05 | since NOLO threshold to allow flashing > bq24150 threshold to stop emergency charging | 22:33 |
DocScrutinizer05 | iirc! | 22:34 |
Pali | nolo is not signed... so maybe we can try to find hex number of that value and change it? | 22:34 |
DocScrutinizer05 | yes, I suggested that several times | 22:35 |
Pali | but there is another problem which I tested: | 22:35 |
Pali | nolo loaded by cold flashing (it always enter to flash mode!!) has unstable usb when battery is too empty | 22:36 |
Pali | which caused that I was not able to flash | 22:36 |
DocScrutinizer05 | quite possible | 22:36 |
Pali | because usb transfer was not possible | 22:36 |
DocScrutinizer05 | :nod: | 22:36 |
Pali | so hexchaning it will not help | 22:36 |
Pali | we need either proper charning in nolo or increase bq treshold | 22:37 |
DocScrutinizer05 | NOLO should learn my dirt-simple bq24150-charging | 22:37 |
*** xes has joined #maemo-ssu | 22:37 | |
DocScrutinizer05 | 5 i2c commands to configure, one timer plus one i2c infinite loop to tickle WD | 22:38 |
Pali | but omap bootrom was still able to transfer 2nd and secondary image and boot it! | 22:38 |
DocScrutinizer05 | bq24150 is sane and safe | 22:38 |
DocScrutinizer05 | weird | 22:38 |
Pali | so if we are able to write ASM code for charning we can send 2nd and that asm code instead nolo | 22:39 |
DocScrutinizer05 | maybe because ROMBL doesn't initialize RAM and stuff | 22:39 |
Pali | and 2nd image can boot it | 22:39 |
Pali | because nolo is not signed | 22:39 |
DocScrutinizer05 | :nod: | 22:39 |
Pali | but somebody must write that pure ASM code for bq charging | 22:39 |
DocScrutinizer05 | sounds feasible | 22:39 |
DocScrutinizer05 | that isn't hard | 22:39 |
Pali | plus in this state lot of hw parts are not initialized | 22:40 |
DocScrutinizer05 | i2c is a hw interface | 22:40 |
Pali | and need to stop twl watchdog | 22:40 |
DocScrutinizer05 | :nod: | 22:40 |
Pali | DocScrutinizer05: are you able to do it? | 22:40 |
DocScrutinizer05 | and possible even BB5 WD | 22:40 |
DocScrutinizer05 | I'm _able_ to do that, but no time atm | 22:40 |
Pali | in uboot I turned off only twl | 22:41 |
Pali | and no reset anymore | 22:41 |
DocScrutinizer05 | ok | 22:41 |
DocScrutinizer05 | then BB5 only gets upset after it got initialized | 22:41 |
DocScrutinizer05 | sounds ane | 22:41 |
DocScrutinizer05 | sane even | 22:42 |
Pali | and thanks to bq24150 support in qemu we can debug code :-) | 22:42 |
DocScrutinizer05 | there's a reset wire from cpu to bb5, and one line the opposite direction | 22:42 |
Pali | check and dump bq registers from qemu | 22:42 |
DocScrutinizer05 | you already have all that in my charge draft, which ShadowJK based his charge21.sh on | 22:43 |
DocScrutinizer05 | it's dirt simple | 22:43 |
DocScrutinizer05 | main obstacle is talking to I2C IP lowlevel | 22:44 |
Pali | but my low level hw programming skills are not good... | 22:44 |
DocScrutinizer05 | on register level | 22:44 |
Pali | I never wrote any i2c driver in asm | 22:44 |
DocScrutinizer05 | I guess you first need to initialize it | 22:44 |
DocScrutinizer05 | speed and power | 22:44 |
DocScrutinizer05 | then simply read buffer, or write buffer | 22:44 |
DocScrutinizer05 | rest is hw | 22:45 |
DocScrutinizer05 | I never looked into ABI/API of I2C IP | 22:45 |
DocScrutinizer05 | but I'm EE for knowing how it must look like | 22:46 |
DocScrutinizer05 | some 2..5 registers | 22:46 |
Pali | anybody who is able to do that and has time? | 22:46 |
DocScrutinizer05 | btw you wouldn't need asm, use C and get the asm intermediate | 22:46 |
DocScrutinizer05 | or use C straight away | 22:47 |
DocScrutinizer05 | C obj code | 22:47 |
DocScrutinizer05 | C is (and always been) a glorified macro assembler | 22:47 |
Pali | you also need linker scripts/hacks to generate noloscnd header plus start code at specific address | 22:48 |
DocScrutinizer05 | no clue about that | 22:48 |
Pali | or code must be position independent | 22:48 |
DocScrutinizer05 | it is | 22:48 |
Pali | in asm you can add .ascii "noloscnd" | 22:49 |
DocScrutinizer05 | some 50 ARM opcodes | 22:49 |
DocScrutinizer05 | you can use ASM in C | 22:49 |
DocScrutinizer05 | inline | 22:49 |
DocScrutinizer05 | don't ask me how, last I did this been 1990 | 22:50 |
DocScrutinizer05 | last I seen this must be 2 years ago | 22:50 |
DocScrutinizer05 | you frequently see this stuff in highly optimized low level hw drivers | 22:52 |
DocScrutinizer05 | but nowadays you almost don't need it anymore, with all that new smart stuff that C allows, like defining absolute addr or register for variables | 22:53 |
DocScrutinizer05 | Pali: check SPRUF98 for I2C hw block (IP), if any questions how to handle it, ask me | 22:54 |
DocScrutinizer05 | I can tell you what to do with it once you found out (with a little help from my side) _how_ to talk to I2C | 22:55 |
Pali | DocScrutinizer05: I cannot look now (opening that pdf on n900 is impossible) and also Impossible) I do not have time for writing that code... | 22:55 |
DocScrutinizer05 | the "code" is dirt simpla. I2C is dirt simple. You send chip addr, then send register addr<<1&(read?0:1), then write or read the register value | 22:58 |
DocScrutinizer05 | or was it chipaddr<<1&(read?0:1), then register addr | 22:59 |
Pali | and for charging you need to enable some gpio | 23:00 |
DocScrutinizer05 | anyway a register write are three atomic elementary writes to the i2c hw interface, a register read are two writes and one read | 23:00 |
Pali | tx reset or something... | 23:00 |
DocScrutinizer05 | huh? | 23:01 |
Pali | some gpio, freemangordon already wrote it here... | 23:01 |
Pali | look into log | 23:01 |
Pali | some #define | 23:01 |
DocScrutinizer05 | sorry, I didn't get it when fmg asked you about that | 23:02 |
DocScrutinizer05 | and I don't get it now | 23:02 |
DocScrutinizer05 | there is no GPIO I know of that needs to get set for charging | 23:02 |
DocScrutinizer05 | there is some nasty foo for detecting charger | 23:03 |
DocScrutinizer05 | charging is only bq24150, and that is only attached to CPU via I2C | 23:04 |
DocScrutinizer05 | and it needs only I2C commands to this chip to keep it charging | 23:04 |
DocScrutinizer05 | ~5 for setting up max voltage, max USB current, trickle charge, etc, and one recurring register access for tickling WD, every 15s | 23:05 |
DocScrutinizer05 | that's all it needs | 23:05 |
DocScrutinizer05 | ~50 ARM opcodes | 23:05 |
DocScrutinizer05 | incl setup of the I2C hw interface in SoC | 23:06 |
DocScrutinizer05 | more if you want to do this in a multitasking way, allowing other tasks while tickling WD | 23:06 |
*** Pali has quit IRC | 23:44 | |
FatPhil | Does the bot have a way of remembering messages for people who've left the channel... | 23:45 |
FatPhil | he says glaring at Pali... | 23:45 |
*** Vlad_on_the_road has quit IRC | 23:46 | |
FatPhil | oi, silly bot, tell Pali about this: http://www.spinics.net/lists/linux-omap/msg11786.html | 23:46 |
*** xes has quit IRC | 23:48 | |
*** lizardo has quit IRC | 23:55 | |
FatPhil | ANyone know what http://elinux.org/N900 and https://gitorious.org/linux-n900/linux-n900 are ? | 23:55 |
Generated by irclog2html.py 2.15.1 by Marius Gedminas - find it at mg.pov.lt!