Tuesday, 2013-04-09

--- Day changed Tue Apr 09 2013
-!- rikanee [~rika@unaffiliated/rikanee] has quit [Remote host closed the connection]02:04
-!- rikanee [~rika@unaffiliated/rikanee] has joined #ubiboot02:04
-!- rikanee [~rika@unaffiliated/rikanee] has quit [Ping timeout: 264 seconds]02:53
-!- rikanee [~rika@unaffiliated/rikanee] has joined #ubiboot02:54
-!- rikanee [~rika@unaffiliated/rikanee] has quit [Client Quit]02:55
-!- rikanee [~rika@unaffiliated/rikanee] has joined #ubiboot03:13
-!- rikanee [~rika@unaffiliated/rikanee] has quit [Quit: kthxbye]03:27
-!- rikanee [~rika@unaffiliated/rikanee] has joined #ubiboot03:41
-!- rikanee [~rika@unaffiliated/rikanee] has quit [Client Quit]03:45
-!- rikanee [~rika@unaffiliated/rikanee] has joined #ubiboot04:09
-!- TMavica [TMavica@1-36-18-052.static.netvigator.com] has joined #ubiboot05:37
@juicememorning07:17
@juicemeTMavica, did you have time to try the watchdog-kicker version?09:10
-!- valdur55 [~valdur55@box025.mr.datanet.ee] has quit [Read error: Operation timed out]09:10
TMavicanot yet try..12:28
TMavicano time yet12:28
@juicemeok.12:42
JonniI have really wierd problem on one of my N950's, android option always boots into harmattan... really wierd12:52
rikaneeJonni: checked the preinit chain?12:52
Jonnihave to put the device in serial console in the evening :)12:52
Jonnirikanee: well I have rsynced the boot-dir structure from working N950, which boots to android12:53
Jonnithats why its really wierd :)12:53
rikaneemmm.12:53
Jonniwhen I get the device attached to serial console, I should be wiser :)12:53
Jonniso now I have three triple booting N9's and one triple booting N950, and one double booting N950 with android booting to harmattan.12:56
Jonnibut maybe I'll flash the newest kernel, to see if it would help12:57
Jonni0804 is broken atleast for me, boot os/kernel selection failed, most likely drives are not mounted13:10
Jonnior I have wrong cpio13:11
Jonniand menu starts to work again when I downgrade to 050413:23
Jonnijuiceme: 0804 is broken13:23
-!- TMavica [TMavica@1-36-18-052.static.netvigator.com] has quit []13:23
Jonniah I only had corrupted init_nitroid, so I guess I just had forgotten to run sync before rebooting... Im always too fast after unextracting.13:34
Jonniso currently running 0504 with custom cpio and everything working with my custom watchdog kicker. Might be that 0804 tries to kick before the devices have been mounted.13:35
@juicemeJonni, meaning kernel zImage_2.6.32.54-ubiboot-02_080413 ?13:37
@juicemeḧow so?13:37
@juiceme0804 just contains the watchdog kicker you provided.. nothing much else changed from 0504...13:38
@juicemeat what point you start the kicker?13:39
Jonni$SHOWPNG "$IMAGEBASE/topmenu.png" > /dev/null13:44
Jonniwatchdog13:44
Jonni## This is the main loop. If any selection is executed, the script will bail out inside the main menu.13:44
Jonnibasicly I start it in cpio after the menu has been drawn13:44
Jonniatleast is looks like that zImage_2.6.32.54-ubiboot-02_080413 has not mounted p2 and p4 devices, since all the kernels fails to load.13:47
Jonniand my guess is that watchdog is called too early and there is error in boot sequence and if that happens before mount, then things are not that good13:48
Jonnibut anyways just guesses, as Im currently not anywhere near of my serial console env.13:52
@juicemeokay, I'll have to move the initialization down then.14:07
@juicemeOn my device there is no problems, since it works even without the kicker.. I still do not know why.14:08
@juicemeWhen I get the serial debug cable I hope to find out what's so special about my N9.14:08
Jonni(oh and I tested 0804 in N950, so that might have some wierdness too)14:15
@juicemeIt could well behave differently. I guess there is more room for HW differences, as even different hwrevs of N9 are so far apart.14:16
@juicemeJonni, would it be best to start the watchdog kicker if USB cable is connected, so as not to cause device restarts in repair mode?14:23
@juiceme(if you now start it when entering menu...)14:23
Jonnimight be, I just moved it to there since it didnt work on beginning of that script....14:24
Jonniit was just quickhack to get menu working for me14:24
Jonnireal solution would be somewhere inside the kernel image or something14:25
@juicemeI'll try that one and upload the test image for you.14:25
@juicemeOr actually if there is some problems with mounts when it's started, I start it in 2 places, either in the USB cable connected state or just before entering menu.14:26
@juicemeAllright, Jonni, there is a new test version available: http://www.swagman.org/juice/zImage_2.6.32.54-ubiboot-02_09041314:40
@juiceme(md5sum=570e459340baba7dc1a8f07d5567fd38)14:41
@juicemethis one has the WD kicker done so that it starts either in the beginning of a maintanance session or just before the menu is started.14:42
Jonnithanks, I'll try it soon16:15
Jonnisame problem, now Im thinking that maybe its bacause of custon cpio, ie when cpio tries to start watchdog kicker again it just fails.... have to test with vanilla cpio when I get home16:26
Jonnicurrently behind 3g is killer for downloads16:27
Jonniah yes, it was because of cpio, it leads to failure if you try to start kicker twice16:36
Jonniwhen I removed my kludge from cpio, then everything works smoothly16:36
-!- coderus [~coderus@94.181.21.235] has quit [Ping timeout: 256 seconds]16:53
-!- coderus [~coderus@94.181.21.235] has joined #ubiboot16:53
guildwarnn17:17
@juicemeJonni, that's good to know. So there is currently no strange stuff happening with 0904 ?17:55
Jonnijuiceme: looks stable, and all os'es boot up fine17:58
@juicemegood :)17:58
@juicemeBTW, changed BME startup a bit there, I would like to get bmestat client working so I try to start the server so that it creates the socket interface to /tmp/ but it failed for some reason.17:59
@juicemeCan you see the reason why it fails: http://pastebin.com/f7EvD46A18:04
rikaneejuiceme: doesn't bme need dame and sysfs mounted?18:05
@juicemeah, could be18:05
@juicemebut I think it should be mounted, have to check that18:06
rikaneemmm.18:06
@juicememaybe it's mounted at later stage...18:06
@juicemeyes, sysfs is in /etc/fstab:18:10
@juicemesysfs   /sys        sysfs               defaults    0   018:10
@juicemeand before starting "bme_RX-71 -l syslog -v 7 -c /usr/lib/hwi/hw/rx71.so -d" I am doing "mount -a" so it should be there OK.18:11
@juicemeany other guesses ?18:11
@juicemebme_RX-71 should create the socket file /tmp/.bmesrv so that it does not need to exist before, right?18:13
@juicemeHowever I see only /tmp/.bmeevt and /tmp/.bmepmm there...?18:14
@juicemethere are 2 errors logged by BME:18:26
@juicemedaemon.err bme: [bmeipc.c:bmeipc_open():339] unable to connect to server (sd=1): No such file or directory (2)18:26
@juicemedaemon.err bme: [dsme_comm.c:bme_init_dsme_conn():38] cannot connect to DSME socket: No such file or directory (2)18:26
@juicemeso it looks to me like it is trying to access the socket file but for some reason cannot do it. I think that's a bit strange since as it's the provider it should know how to create the socket, right?18:28
@juicemerikanee?18:28
@juicemeJonni?18:28
@juicemewhat's DSME anyway?18:30
JonniI'm in bus with 3g, so my connectivity is a bit limited18:30
@juiceme:*)18:30
@juicemewhat, no full rate 4G @Tampere :) :)18:31
Jonnino full rate on motorway between hki-tre18:32
Jonnidsme - Device State Management Entity18:32
@juicemeokay18:32
@juicemeso there is some other service BME depends on...?18:33
JonniDepends: libc6 (>= 2.4), libcal1, libcreds2, libdsme0.2.0, libgcc1 (>= 1:4.1.1), libiphb018:34
Jonniiphb -> Description: API for IP Heartbeat service18:36
@juicemeI'm fairly sure I have the libraries or it would complain badly.18:38
@juicemeiphb... hm18:38
@juicemeok, on the running Harmattan there seems to be /sbin/dsme-server running just before /usr/sbin/bme_RX-7118:40
@juicemeso I should start that first?18:40
@juicemeyes, all the libraries are there, checked it now.18:44
-!- peterleinchen [~user@dsbg-4d0d35f0.pool.mediaWays.net] has joined #ubiboot22:31
@juicemeHi peter22:41
@juicemepeterleinchen22:42
peterleinchenyep 8@am here, just listening ;)22:43
peterleinchenand using BitchX on N900 (totally new experience ...)22:43
@juicemel8t evening here, 22:5022:43
peterleinchen18t ?22:44
@juicemeBitchX's okay, used a lot of that on the 90's22:44
@juicemel8 = late :)22:44
peterleinchenyes and Coline Stephane compiled it and put into repos :)22:45
@juicememyself, I run irssi under screen on my server, and connect to that either from laptop or n9 ssh :)22:45
@juicememaybe I should try BitchX on N9..22:46
peterleinchenI see I am really old fashioned (bot on l8 and how I use irc :)22:46
@juiceme:)22:46
peterleinchenyes it is funny to see the header blinking up. ansi andconsole style22:47
peterleinchenBut if we continu like this we need to open up a new talk channel???22:48
peterleinchenso I followe discussion and it really looks there are still some HW related probs?22:48
peterleinchenI (rev 1601) do not have problems with menu and watchdog22:49
@juicemewell yes, there are (or were) some watchdog related problems22:49
peterleinchenbut I think Jonnis experience may explain why my booting sometimes fails with wd32_to and I need two attempts22:50
@juicemeand there's one 1603 that works with ubiboot, I've been waiting for somebody to test the newest zImage..22:50
peterleinchenI tried to find out why but could not find any reason22:50
@juicemeI guess you could try to update to 0904 zImage, try if thar helps to the WD resets...22:50
peterleinchenyes I will try as soon as I can /still away22:51
peterleinchendid you check my (small) changes?22:51
@juicemethe change to the boot partition as a variable?22:52
peterleinchenand one question: where is the /mnt folder?22:52
peterleinchenyes22:52
thedead1440Hi peterleinchen :)22:52
@juicemeyes, that looked OK. Didi not implement it uet, though22:52
peterleinchenhi asia!22:52
@juicemethedead1440, hi22:52
thedead1440hi juiceme22:52
@juicemepeterleinchen, /mnt folder... thats on the rootfs of course :)22:53
peterleinchenyes :) BUT where in ? cpio: no in ubifs: no22:53
@juicemeah, yes.22:54
peterleinchenOK, will be the guinea pig for my changes then on my own ;)22:54
@juicemein ubiboot-02, the whole shebang is ramFS22:54
peterleinchenwhat yes ;)22:54
@juicemesee, the filesystem "seed" is built into ramfs in kernel22:55
@juicemeand the cpio is loaded up there to "extend" it in a way.22:55
@juicemequite ingeniuos actually :)22:55
peterleinchenOK expected that but did not know about "seed", rest is known22:56
@juicemewell, look into the source directory on swagman.22:56
peterleinchenOf course I did. and kernel size is not too big? on N900 limit is 2 MB22:57
@juicemethere's the files ubifs_xxxx.tar, those are the ramfs seed22:57
@juicemewell, the reason the rest of the stuff is in cpio is just that22:58
peterleinchenyes and I did not find /mnt in there 9r am I blind?22:58
@juicemeoriginally I built _everything_ into kernel and it got about 20MB of size :P22:58
@juicemethat was too big, caused strange problems22:59
peterleinchen:)22:59
@juicemelook at the file "ubifs/ubifs_list" on the FS seed. That is the definition file which is used by kernel scripts to build the FS. /mnt/ is defined there...23:00
peterleinchenwill upload another v3 of my mods (corrected nit partition number to three) and also includes first attempt for my SIP problems (linking to different directories for aegisfs and openmode)23:01
@juicemenice guide how to do it: http://www.landley.net/writing/rootfs-howto.html23:01
@juicemeokay23:01
peterleinchenThanks, that explains it to Windoof user23:01
@juiceme:p23:02
peterleinchenthedead still here?23:02
thedead1440peterleinchen: yup23:02
peterleinchensearched for openmode and SIP problem but nothing?23:03
thedead1440because there is no reported issue :D23:03
peterleinchendamn!!!23:03
thedead1440i tried sip on my side and didn't face any probs too23:03
thedead1440its maybe sip on n9(00) that's at fault23:03
thedead1440i've read n900's sip sucks too23:03
peterleinchenwith ubiboot and l2fix kernel?23:03
thedead1440yeah i use ubi on mtd5 with l2fix23:04
thedead1440i used getonsip.com w/o issues23:04
peterleinchenthats strange, 'same' here mtd4 and l2fix23:04
peterleinchenand same provider works on stock kernel23:05
@juicemepeterleinchen, about the partition configurability... I still think it should be better to use "\/dev\/mmcblk0p1\/" type sed notation there, instead of just variabeizing the partition number. There is one special case, for example: If you want to make some kernel root to an ubifs partition, for example...23:05
thedead1440>>includes first attempt for my SIP problems (linking to different directories for aegisfs and openmode)<< which dirs?23:05
peterleinchenOK it is your baby!23:05
@juiceme:)23:06
peterleinchen.aegisfs, .account, .activesync23:06
@juicemeBut I'll credit it to you :)23:06
peterleinchenTHANKS :D23:06
peterleinchenthedead, just see my v3 (will upload now)23:07
thedead1440what linkages would you do?23:07
thedead1440oh ok23:07
peterleinchennow attached (23:09
thedead1440attached where?23:10
peterleinchenbriefly: check kernel version -- check link points to correct dir .account_openmode or .account_aegis -- rm and create new23:10
peterleinchenubi thread ? maybe???23:10
peterleinchen;)23:10
thedead1440http://talk.maemo.org/showpost.php?p=1334380&postcount=24923:11
thedead1440found it :D23:11
peterleinchenI knew that you will find ...23:11
thedead1440hehe23:11
thedead1440will do a diff tomorrow when i wake up :D23:13
peterleinchennite23:14
thedead1440night to you too23:14
Jonniwell you can have fully working aegis in open mode, so those hacks are not needed necessarily23:15
peterleinchenyeah? how?23:16
Jonnieasiest way is by never booting to secure mode, that way you dont have to re-create the aegis stores.23:17
peterleincheneasiest not possible :) for your info: I try to create SIP acccount with l2fix and it fails23:17
thedead1440but to remove the warning a first boot into secure mode is still required right?23:17
peterleinchenbut booting stock kernel and SIP works, but after created SIP account and then booting l2fix all mail accounts ask for pass23:18
Jonnithedead1440: no its not. I usally do it like this: full reflash, boot  into secure mode, remove warning, and again full reflash, and trigger openmode without booting23:19
@juicemeJonni, how do you directly boot to open mode without going via closed mode after flashing?23:19
thedead1440Jonni: so full reflash i.e. rootfs+emmc doesn't insert the warning back?23:19
@juicemedo you mean, ater flashing immediately turn the device off, and then flash open mode kernel?23:20
Jonnithedead1440: nope, warning is inserted on 1st boot into secure mode :)23:20
Jonnithedead1440: thats the way to have fully working aegis without warning in openmode :)23:20
thedead1440oh so it means even a full reflash is not able to really reset device?23:20
Jonnithedead1440: it resets the device, but by default device does not have warnings screen images in cal out of the factory23:21
Jonnithey are only inserted when user boots up the device 1st time23:21
thedead1440ah so before doing a full reflash just remove warning then full reflash with openmode kernel is enough?23:22
Jonnithedead1440: nope, full refhash to remove warning, then full refhash and then flash openmode kernel seperately23:22
thedead1440ok23:23
peterleinchenthink I got it, too23:23
thedead1440so then fixed origin to override aegis while still keeping aegis for other things?23:23
Jonniie after removing the varning,  "sudo flasher -f -F main.bin; sudo flasher -a main.bin -k zImage --flash-only=kernel -f -R"23:24
peterleinchenso only "full reflash" means with eMMC, right?23:24
Jonniyep, then you can use fixed origin for everything and aegis for rest23:24
JonnieMMC is not needed, you can change between secure and open mode with main.bin only23:25
@juicemeI'll have to try that sometime. 23:25
Jonniofcourse you can flash emmc if you want, but it doesnt matter23:25
* thedead1440 will try that once back from holidays :D23:25
peterleinchenI do not want hence the question23:26
peterleinchenso eMMC not needed at all? not for getting rid of warning nor for disabling?23:27
Jonniand ofcourse there still remains the issue of locked cal in openmode, I've solved it by installing custom devicelock package that uses aegisfs instead of cal, that way I have fully working mfe and device lock even on openmode :)23:27
Jonnipeterleinchen: yes, emmc is not needed at all.23:27
thedead1440Jonni: any chance of that custom package being "leaked"? :D23:28
Jonniheh, there always is a chance, quite many people have it.23:29
@juicemeis the devicelock package from an earlier PR release?23:29
-!- peterleinchen [~user@dsbg-4d0d35f0.pool.mediaWays.net] has quit [Ping timeout: 256 seconds]23:31
* thedead1440 is awaiting the customlock package showing up on http://maemo.cloud-7.de/HARM/N9/1.3/openmode-kernel/23:34
coderusJonni: and ofc you have devicelock sources and can't share it?23:35
thedead1440coderus: considering who he was, only an idiot wouldn't have made him signed a gazillion NDAs and the like after all ;)23:42
JonniI dont expect seeing sources ever beeing released, unless some subcontractor screws up badly, but you never know, someone leaked twitter.deb, although I have no idea was it subcon or nokian.23:46
-!- peterleinchen [~user@dsbg-4db59f3b.pool.mediaWays.net] has joined #ubiboot23:47
thedead1440Jonni: which twitter.deb? Wasn't that an official update for N950s?23:47
Jonniwell the deb was avail in internet before it was pushed to repoes afaik.23:47
thedead1440oh it was in the /unreleased repo23:47
Jonnior I might have missunderstood, I havent been following twitter things that much23:48
thedead1440iirc it was more/less an official update; not really a leak23:48
Jonniwell n950 pr13 was more or less official image, but never published.23:49
thedead1440there was around a year ago one person who registered on tmo claiming to have sources and scolded profanities against Nokia... I wonder where he disappeared...23:49
thedead1440he deleted his posts i think around 3-6hours afterwards23:50
thedead1440was around the pr1.3 for n950 time iirc23:50
coderusCAL ERROR: nand_write: write (2048 bytes at around 000f7800): Input/output error23:51
coderusthis is the CAL address contains devicelock password :)23:52
Jonniwell the address changes everytime you change the password, and it only holds a copy of the password, its not really used really for checking the device lock code :)23:57

Generated by irclog2html.py 2.12.1 by Marius Gedminas - find it at mg.pov.lt!