**** BEGIN LOGGING AT Wed Mar 13 02:59:58 2013 Mar 13 03:11:25 *may* you also edit his post http://talk.maemo.org/showthread.php?p=1328124#post1328124 to point to that new thread? Or is this sth a mod shouldn't do on tmo? Mar 13 03:13:44 nm, I simply added a post with a note Mar 13 03:14:27 yeah i would have edited it if people seem unsure but since you've posted its enough ;) Mar 13 03:15:06 i do edit posts; like certain threads whereby OP doesn't link to solution or duplicate posts etc... basically all harmless stuff :D Mar 13 03:16:22 do you receive any notifications from TMO anymore? Mar 13 03:16:49 hmm i'm not subscribed to many threads actually Mar 13 03:17:03 i'm active enough to notice posts without a notification :D Mar 13 03:17:47 could you delete your post and post it again? Mar 13 03:17:52 i've subscribed to that thread Mar 13 03:18:02 so we can find out now Mar 13 03:18:07 I'll rather post a new one and delete it Mar 13 03:18:13 ok Mar 13 03:19:03 or rather, you delete it Mar 13 03:21:14 no notification Mar 13 03:21:52 thought as much Mar 13 03:21:56 :-/ Mar 13 03:22:10 after deleting your post it redirected me to: http://213.128.137.28/showthread.php?p=1328691 Mar 13 03:23:11 yup, and that's a moot source for any smtp sendmail Mar 13 03:23:35 literally nobody will accept mail from that server, for now Mar 13 03:23:43 yeah Mar 13 03:23:54 so Mar 13 03:24:24 [GENERAL NOTICE] tmo notification mail is broken. Sorry for the inconvenience Mar 13 03:25:31 eventually tmo admin will set up a banner to let all users of tmo know about the fact, and explain why this issue arised and when it's supposed to be fixed Mar 13 03:25:39 yeah in migration #2? Mar 13 03:26:06 hm? Mar 13 03:26:19 oh you posted here Mar 13 03:26:27 sorry was confused a bit Mar 13 03:27:35 i was thinking of making migration #2 thread sticky and removing the service downtime thread from sticky Mar 13 03:27:45 that way it would be easier to follow on-goings Mar 13 03:28:25 [GENERAL NOTICE] similar isses are likely to occur with other maemo.org services during migration#2, which has its top at Thursday 1700UTC when Nokia is supposed to switch the DNS pointers to new infra Mar 13 03:28:50 thedead1440: sounds fair Mar 13 03:29:16 you do changes to the migration #2 thread title to make it clearer and update the OP with current status; i'll change the sticky Mar 13 03:29:44 k, will look into it Mar 13 03:30:02 your thread is now first thread in General so make it as clear as possible Mar 13 03:30:47 i'm removing the fundraiser thread from sticky too since its pretty moot Mar 13 03:34:00 :nod: Mar 13 03:35:52 * DocScrutinizer05 has a last look at ganglia, shudders, burps, and heads out, waving Mar 13 03:36:07 good night Doc ;) Mar 13 03:36:58 hmm www having excessive load Mar 13 03:55:26 is there a way to clone one n900 to another ? Mar 13 03:55:35 backupmenu sounds like an option Mar 13 03:56:03 ~backupmenu Mar 13 03:56:04 i guess backupmenu is http://talk.maemo.org/showthread.php?t=63975 Mar 13 07:11:44 Does anyone know how to get console map (.map) out of X keyboard map file (like our nokia_vdnr/rx-51) Mar 13 07:11:46 ? Mar 13 07:12:07 I would like to duplicate my hardware keyboard map into recovery shell/similar console mappings Mar 13 08:51:25 21:31:26 Estel_ | divVerent, qhy you constantly like to make your life harder?:P Mar 13 08:51:27 21:32:21 Estel_ | either do as DocScrutinizer05 suggested with vanilla folders and overmounting, or do like me with dead symlinks (which btw, works for most .config things too). Mar 13 08:51:39 Estel_: both ways ARE making the life very hard Mar 13 08:51:52 21:33:26 kerio | because things will get error messages Mar 13 08:51:54 21:33:34 kerio | and you *really* can't open a dead link, afaik Mar 13 08:51:56 21:34:07 Estel_ | thats why I use it, in practice,. But if someone want plausible deniability more, vanilla overmount files are the way to go Mar 13 08:51:58 doesn't work for the browser Mar 13 08:52:07 it silently replaced a dead symlink ~/.bookmarks by a new file Mar 13 08:52:28 tablet-browser-ui is a piece of crap, yes Mar 13 08:52:34 and overmount means having to kill all those processes Mar 13 08:52:39 which is EXACTLY what I call a mess Mar 13 08:53:22 Estel_: and what is this about text2screen getting possibly removed? Mar 13 08:53:32 what happens to bootmenu/backupmenu then? Mar 13 08:56:14 Estel_: but assuming I have to restart stuf... that really sucks... can't I _somehow_ show a prompt on X11 first, and only THEN start up all the stuff? Mar 13 09:01:18 Estel_: also, can you maybe share your current script, regarding the list of all that needs killing/restarting? Mar 13 09:01:31 although I still REALLY don't like the approach, given it's quite easy to go wrong Mar 13 09:01:49 already noticed that if you kill browserd too often, the device reboots Mar 13 09:25:30 divVerent: Meet the watchdog. Mar 13 09:27:08 I just wonder why the WEB browser is seen as THAT critical Mar 13 09:27:15 if it were the telephone subsystem, I would understand Mar 13 09:27:23 but why would I even care if MicroB runs when I want to use e.g. Opera? Mar 13 09:51:46 divVerent: browserd also powers Conversations Mar 13 09:52:03 aka rtcom-messaging-ui Mar 13 09:53:06 which is an awful way :D Mar 13 09:53:18 cant get rid of it to save ram Mar 13 09:53:37 divVerent: ((it silently replaced a dead symlink ~/.bookmarks by a new file)) toldya Mar 13 09:54:24 chown it to root, perhaps chattr +i? :) Mar 13 09:55:13 ((if you kill browserd too often, the device reboots)) toldya Mar 13 09:55:16 DocScrutinizer05: I know, but we still have no SOLUTION... this overmounting stuff has the very same issue (browser keeping to use the previous files) Mar 13 09:55:20 yes, that too Mar 13 09:57:00 anyone know how to use flasher -b to override boot parameters? Mar 13 09:57:06 I tried and failed, don't know why Mar 13 09:57:13 it simply booted up with the original /proc/cmdline Mar 13 09:57:35 I think that never worked Mar 13 09:58:02 that means acting at preinit level is also ruled out for a WORKING encryption setup Mar 13 09:58:12 because replacing the preinit file is bad because that gets lost on upgrades Mar 13 09:58:21 (or patching it) Mar 13 09:58:40 bootmenu hook is your best option Mar 13 09:58:48 but then I depend on bootmenu... but yes Mar 13 09:58:52 that is what I am currently trying Mar 13 09:58:55 but you'll need to edit some initscripts anyway Mar 13 09:59:13 well, maybe not Mar 13 09:59:22 what I basically am trying now: Mar 13 09:59:25 password prompt at bootup Mar 13 09:59:31 provides /home/user first monted to /.home.user Mar 13 09:59:43 on /home, user shall be a symlink to /.home.user Mar 13 09:59:49 and basically, existing init scripts SHOULD be ahppy then Mar 13 09:59:50 the mount-over & fuser -k approach might work without any edits to init Mar 13 10:00:21 that is the other approach I am trying to follow... that is the preferred method when using encfs, of course Mar 13 10:00:52 but fuser -k only kills processes that currently have files open - not such that have read files, and keep their contents in memory Mar 13 10:01:23 also, a general fuser -k -m /home is KINDA unwise ;) Mar 13 10:01:42 it probably is equivalent to kill -- -1 as user... actually, wonder what happens if I do THAT Mar 13 10:02:17 in busybox, kill -15 -1 Mar 13 10:02:34 and no, the init system does NOT handle this well Mar 13 10:02:41 why would any app do open() read() close() open() write() close()? Mar 13 10:02:41 I had hoped upstart would then respawn all that's necessary ;) Mar 13 10:03:01 DocScrutinizer05: would be totally sensible for email Mar 13 10:03:05 first read acccount info Mar 13 10:03:07 close config file Mar 13 10:03:12 then use what you read all the time Mar 13 10:03:18 and when user changes settings, open again, write, close Mar 13 10:03:34 this is probably the usual way to do things with text files Mar 13 10:03:39 while database files normally are kept open Mar 13 10:03:50 then kill -9 Mar 13 10:03:58 instead of kill Mar 13 10:04:08 no, I am not worried about the writing part ;) Mar 13 10:04:20 but that the process may not even be identified by fuser -k in the first place Mar 13 10:04:34 aaah I see, you're worried you miss processes to kill Mar 13 10:04:36 stupid example: browser having file:///home/user/secret/info.txt open Mar 13 10:04:41 the file is closed as soon as it's rendered Mar 13 10:04:57 so overmount/fuser -k -m won't get the browser in this case (if the secret/ dir is all I worry about) Mar 13 10:05:13 but the user would in this case have to manually refresh the browser Mar 13 10:05:24 :nod: Mar 13 10:05:58 you'll find out about that possibly 1 or 2 programs and add it to the list of processes to kill "manually" Mar 13 10:06:05 yes Mar 13 10:07:06 xchat for example is known to never write back config edits unless properly terminated from GUI Mar 13 10:07:34 but really too bad that kill -15 -1 does not do what it should ;) Mar 13 10:07:43 not even gentle device shutdown will make it write back edits Mar 13 10:07:46 on most desktop linux, killing ALL processes you own just brings you back to the login screen Mar 13 10:08:35 and even killing all on the X-owned tty would happily restart the login session [but then you lose text mode...] Mar 13 10:09:23 * DocScrutinizer05 idly ponders if ACT_DEAD could do something useful for divVerent Mar 13 10:09:37 hehe, I found this one too... but have not yet seen the way to get OUT of that mode ;) Mar 13 10:09:48 lol Mar 13 10:09:49 anyway, suppose I want to overmount whole /home/user... wonder how to best find out what I need to kill then Mar 13 10:10:04 an alarmd event should do the trick Mar 13 10:10:06 fixing MyDocs is trivial, so let's ignore that Mar 13 10:10:55 (for MyDocs, either umount/mount, or two levels of mount --bind would do the trick) Mar 13 10:10:56 lsof is a first good approach Mar 13 10:11:04 yes, lsof can do a bit more than fuser Mar 13 10:11:27 fuser -m is no option because /home/user is not a file system :P Mar 13 10:11:46 we also got inotify Mar 13 10:12:18 wonder if doing mount --rbind /home/user /home/user done "early enough" would fix fuser -m Mar 13 10:13:27 inotifywatch/inotifywait could help you create a list of all processes ever touching some files you want to know about Mar 13 10:14:34 would also catch those that do open() read() close() Mar 13 10:16:49 wonder if there is a more clever way than lsof +D Mar 13 10:17:00 inotify... works if I start it early enough Mar 13 10:19:09 kill `lsof +D /home/user -t` Mar 13 10:19:13 is still a bad idea Mar 13 10:19:16 (reboots) Mar 13 10:21:54 yeah, since it most likely kills hildon-* Mar 13 10:22:15 in a way that is what I want/need to do Mar 13 10:22:20 as it DOES use info from the homedir Mar 13 10:22:21 mounting /home, running shit, and remounting /home is a silly idea Mar 13 10:22:40 kerio: right... I am basically pursusing two separate paths, to see which one will work better Mar 13 10:22:52 1. separate dmcrypt partition; /home/user then gets entirely used from that Mar 13 10:23:02 2. encfs at runtime, reinitializing user processes Mar 13 10:23:26 "problem" of 1. is that it is obviously not deniable, whatever you do Mar 13 10:23:30 as you WILL get a password prompt at boot Mar 13 10:23:39 no you won't Mar 13 10:23:51 how to ask for the key then? Mar 13 10:23:54 uboot that needs a held key combination Mar 13 10:24:03 ah, tricks like that Mar 13 10:24:05 and the whole system on uSD Mar 13 10:24:09 or luks setup timeout, and no screen output ;) Mar 13 10:24:32 for deniablity you also need a real, stock system behind Mar 13 10:24:39 right Mar 13 10:24:42 and I don't need that Mar 13 10:24:48 where I live, encryption is (still) legal Mar 13 10:24:55 (and giving out the key is not required) Mar 13 11:14:54 hm... what am I doing wrong when my alternate linuxrc script isn't started at all by bootmenu? Mar 13 11:15:22 I am pretty sure it is not being started at all, as I have text2screen stuff at the top Mar 13 11:16:08 and modprobe fbcon next :P Mar 13 12:16:53 hm... it tries to find /mnt/initfs/$LINUXRC... that of course won't work Mar 13 12:17:13 (debugging it now by having added "modprobe fbcon" to preinit) Mar 13 12:21:42 no it doesn't, does it? Mar 13 12:21:50 there's no /mnt/initfs in stock maemo Mar 13 12:21:53 exactly Mar 13 12:22:07 because there is by default no pivot_root switch Mar 13 12:22:16 but the rootfs is what the kernel sees at initialization Mar 13 12:22:24 yep Mar 13 12:22:29 in /sbin/preinit Mar 13 12:22:30 at least on the N900, I do see the scripts have support for other setups Mar 13 12:22:38 hardly Mar 13 12:22:54 also, that should be the bootloader's job Mar 13 12:22:58 what you see is bit rot Mar 13 12:23:00 SHOULD ;) Mar 13 12:30:58 ok, found ONE possible solution... apparently, all the items need to have the same ID Mar 13 12:31:11 namely, "nand" Mar 13 12:34:44 oh well, at least someone is going to test the unused cruft in preinit Mar 13 12:34:53 hehe Mar 13 12:42:18 where can I get the 20.2010.36-2 Vanilla image for the N900? The nokia.com server is broken… Mar 13 12:42:22 does anyone have them? Mar 13 12:43:11 ~skeiron Mar 13 12:43:12 i guess skeiron is the semi-official backup and emergency standin for all internet borne maemo resources: http://skeiron.org/tablets-dev/ http://talk.maemo.org/showthread.php?p=1315143#post1315143 Mar 13 12:43:18 madduck: ^^^ Mar 13 12:43:28 thedead1440: thanks! Mar 13 12:43:34 this is preferable to some .ru domain ;) Mar 13 12:43:40 :D Mar 13 12:45:29 HAHA, HERE [ -f ] is NO good idea ;) Mar 13 12:45:35 i really hope this will get my n900 completely back to factory as there is so much stuff on it from the previous owner, I really don't want to deal with it. Mar 13 12:45:48 i.e. [ -f /dev/mmcblk0p4 ] just won't ever match ;) Mar 13 12:46:01 madduck: it will, and ONLY this will Mar 13 12:46:13 I had the same problem when I got my N900... I didn't even TRY if doing less would have achieved it too Mar 13 12:46:16 just reflashed both images Mar 13 12:46:32 so emmc and then vanilla? Mar 13 12:46:40 eh, emmc and then combined Mar 13 12:46:41 read the page, I already forgot the proper order Mar 13 12:46:49 and there then was some weird reboot trick you have to do as well Mar 13 12:46:54 RX-51_2009SE_10.2010.13-2.VANILLA_PR_EMMC_MR0_ARM.bin followed by RX-51_2009SE_20.2010.36-2_PR_COMBINED_MR0_ARM.bin it seems Mar 13 12:46:58 it's all explained and it worked for me, but I cannot remember exactly Mar 13 12:47:23 also, if you do wrong, the device won't be bricked - you will just have to start over if I understand it right Mar 13 12:48:02 because the USB flasher mode (holding U key thing) acts before kernel is loaded Mar 13 12:48:12 BTW, do NOT do the boot loader reflash. You won't need it. Mar 13 12:48:27 that is the only part that CAN brick it ;) Mar 13 12:50:09 HAHA; my encryption prompt now SOMEWHAT works... need to fix keymap though Mar 13 12:50:14 or passwords can only use lowercase letters :P Mar 13 12:52:02 now the only part that may require editing init scripts may be bind-mounting /.home.user to /home/user Mar 13 12:52:05 what is hwid? what is the type label of the device? I cannot see anything on the back Mar 13 12:52:16 b elow the battery? Mar 13 12:52:24 no 4 digit code there Mar 13 12:52:31 0434? Mar 13 12:52:34 where are you asked for it? Mar 13 12:52:39 cold flash Mar 13 12:52:53 why do you need to cold flash? Mar 13 12:52:56 don't cold flash Mar 13 12:53:08 13:48:12 divVerent | BTW, do NOT do the boot loader reflash. You won't need it. Mar 13 12:53:10 13:48:27 divVerent | that is the only part that CAN brick it ;) Mar 13 12:53:12 that's what I meant by this :P Mar 13 12:53:18 aha Mar 13 12:53:28 no settings or user data are stored on the boot loader area Mar 13 12:53:31 so there is no need to overwrite it Mar 13 12:53:42 well, i flashed the emmc and then the combined image, but the combined image didn't do any writing, just quit with exit code 0 after printing a lot Mar 13 12:54:12 then you did it wrong. Your device probably will fail booting now, which is ok. Maybe reboot and hold U again? Mar 13 12:54:16 then retry the other image? Mar 13 12:54:29 http://slexy.org/raw/s29zd3b683 Mar 13 12:54:30 BTW, for rebooting, USB may not be plugged in Mar 13 12:54:45 that's all the output? Mar 13 12:54:54 yes Mar 13 12:54:59 i get a dialog to unlock, all messed up, and then am told there's a missing image Mar 13 12:55:05 which image is missing? emmc or combined? Mar 13 12:55:17 reflash the combined now, I would say Mar 13 12:55:30 the EMMC upload worked, given the output Mar 13 12:55:59 yeah, but same for combined flash Mar 13 12:56:02 nothing is uploaded Mar 13 12:56:17 maybe it detects that it's already uploaded? maybe i need to flash older stuff first, then try again with the latest? Mar 13 12:56:28 no, sure not Mar 13 12:56:31 it's not that clever Mar 13 12:56:53 ah, now the combined flash works, magically Mar 13 12:57:01 just for reference, I had used these images: Mar 13 12:57:02 let's see how far this gets me now… Mar 13 12:57:03 RX-51_2009SE_1.2009.41-1.VANILLA_PR_EMMC_MR0_ARM.bin Mar 13 12:57:05 RX-51_2009SE_20.2010.36-2_PR_COMBINED_MR0_ARM.bin Mar 13 12:57:24 okay, i have a newer emmc but the same combined Mar 13 12:57:48 should make no difference anyway Mar 13 12:58:21 there'll be some difference, albeit minute Mar 13 12:58:45 a difference the first software upgrade run should fix ;) Mar 13 12:59:08 yay apt Mar 13 12:59:16 alright, let's see… Mar 13 12:59:21 booting now Mar 13 13:00:25 gosh it's slow Mar 13 13:00:49 the lock code is still the one I set, not 12345 Mar 13 13:01:32 the lock code is in CAL Mar 13 13:01:42 it's not flashed Mar 13 13:02:38 okay, good. Mar 13 13:03:22 yay, it seems vanilla now! Mar 13 13:03:46 now i need to find a way to import contacts via caldav, set up jabber, email, and find a way to monitor data usage Mar 13 13:03:49 (suggestions welcome) Mar 13 13:04:20 carddav even ;) Mar 13 13:07:27 is extras-devel still offline? i heard it's been back up for a while now, but i keep getting error "Failed to fetch http://repository.maemo.org/extras-devel/pool/freemantle/free/f/frontcamera/frontcamera_1.0.0-2_armel.deb: Hash Sum mismatch" what gives? Mar 13 13:10:37 it's even 404ing in the browser Mar 13 13:11:16 something about an nfs bug corrupting files... there's a mirror that works Mar 13 13:11:26 (as nobody who knows better answered) Mar 13 13:11:56 ~skeiron Mar 13 13:11:56 extra, extra, read all about it, skeiron is the semi-official backup and emergency standin for all internet borne maemo resources: http://skeiron.org/tablets-dev/ http://talk.maemo.org/showthread.php?p=1315143#post1315143 Mar 13 13:30:43 r00t^home: thanks Mar 13 13:31:48 doesn't contain frontcamera but at least it seems to be working Mar 13 13:43:04 I must admit that opening my calender in the desktop as sql is a joy to someone not dealing with SQL too frequent Mar 13 14:03:46 ~ping Mar 13 14:03:47 ~pong Mar 13 14:04:35 by the way, as I see so many ppl here, did anyone manage to get an event service? Mar 13 14:04:37 ~botsnack Mar 13 14:04:37 thanks, DocScrutinizer51 Mar 13 14:04:56 such that colelcts all sort of inputs as, gps, wifi connection etc Mar 13 14:05:13 then you can issue an action by the events collection... Mar 13 14:05:31 what launches /usr/sbin/osso-mmc-mount.sh? Mar 13 14:05:54 divVerent, ke-recv Mar 13 14:08:13 I am lookign for the logic how it finds the device name to use for MyDocs Mar 13 14:08:23 because, it now seems to detect /dev/dm-0 instead of mmcblk0p1 Mar 13 14:09:19 divVerent, it is from HAL Mar 13 14:09:25 or from /etc/fstab Mar 13 14:09:31 sure not from fstab Mar 13 14:09:42 hal, how to check Mar 13 14:09:47 ke-recv use HAL Mar 13 14:09:52 $ lshal Mar 13 14:09:56 cityLights: there's dwimd and profile manager Mar 13 14:10:34 in HAL I only see the right device Mar 13 14:11:08 the other does not have MyDocs in its info Mar 13 14:11:15 my question is rather, where is the decision made which device to use Mar 13 14:11:20 the data is pulled from HAL, but then? Mar 13 14:13:15 ke-recv listening for HAL changes, then it read device from /dev/, read filesytem and mount it. mmc device is mounted to /media/mmc*, usb disks are mounted to /media/usb/* Mar 13 14:13:55 but what becomes /home/user/MyDocs then? Mar 13 14:15:48 maemo system has hardcoded mmcblk for MyDocs Mar 13 14:15:53 except it does not Mar 13 14:15:58 I _want_ it to use mmcblk0p1 Mar 13 14:16:03 but for some reason it now uses /dev/dm-0 Mar 13 14:16:09 when I changed preinit to open a crypto device Mar 13 14:16:11 what is dm-0 ?? Mar 13 14:16:15 device mapper stuff Mar 13 14:16:23 did you installed 3rd app? Mar 13 14:16:27 cryptsetup Mar 13 14:16:29 and my own scripting Mar 13 14:16:29 did you changed something? Mar 13 14:16:33 so I want a way to fix it Mar 13 14:16:35 of course I did Mar 13 14:16:40 maybe that 3rd app messing it Mar 13 14:16:42 so I want to find out how to fix this ke-recv part Mar 13 14:16:50 sure, by creating the dm-0 device Mar 13 14:16:59 so I want to find out how to make ke-recv use the right device EVEN if dm-0 exists Mar 13 14:17:04 because I cannot avoid dm-0 existing Mar 13 14:17:06 divVerent, do you have cssu? Mar 13 14:17:08 yes Mar 13 14:17:45 you can use blacklist script: /etc/default/osso-mmc-blacklist.sh Mar 13 14:17:55 that may be an idea Mar 13 14:18:29 hm... so it all depends on mmc_host.slot_name Mar 13 14:18:33 in $PDEV is full path to device in /dev Mar 13 14:18:33 (now having ke-recv sources open) Mar 13 14:18:36 THAT one I cannot change... Mar 13 14:18:41 in $MP is mount point Mar 13 14:19:03 the device mapper device apparently inherits it from its base device Mar 13 14:19:07 slot_name is exported from kernel Mar 13 14:20:37 and what prevents it from also mounting the OTHER mmc devices? Mar 13 14:20:44 like the root partition Mar 13 14:22:00 in /etc/default/osso-mmc-blacklist.sh you can check if $PDEV is /dev/dm-0 and then exit Mar 13 14:22:06 yes, I know Mar 13 14:22:16 but I want to see if there is a better way, as someone may already have put a blacklist there Mar 13 14:23:00 divVerent, I'm patching ke-recv and also rcS-late which generate /etc/fstab Mar 13 14:23:15 I'm adding support for real /etc/fstab which will not be overwritten Mar 13 14:23:28 so this can fix your problem Mar 13 14:23:49 but it is not finished & tested yet Mar 13 14:25:28 nice Mar 13 14:32:47 then using the blacklist for now is good enough Mar 13 14:33:45 real fstab suppoort would indeed be good Mar 13 14:34:03 /dev/mapper/.home.user Mar 13 14:34:04 1030548 1304 976896 0% /.home.user Mar 13 14:34:06 I have now this mounted... Mar 13 14:34:16 and eventually, I want to mount this over /home/user Mar 13 14:34:22 for which I currently WILL have to patch rcS-late Mar 13 14:36:09 to blacklist a device, what is the blacklist script supposed to do? Mar 13 14:36:12 exit 0? or exit 1? Mar 13 14:36:46 exit 0, I suppose Mar 13 14:38:13 at least that's what the comment in the calling script says ;) Mar 13 14:40:14 divVerent, that script is loaded by "source" command Mar 13 14:40:42 so it does not depends on exit value Mar 13 14:42:21 it has to exit for the blacklisting to succeed Mar 13 14:42:26 which also exits the calling script then Mar 13 14:44:52 if [ -d /.home.user/MyDocs ]; then Mar 13 14:44:54 mount --bind /.home.user /home/user Mar 13 14:44:57 let's see if THAT will work Mar 13 14:45:44 fsck is a bit unhappy that this early at bootup, clock isn't set yet :P Mar 13 14:49:13 ok, did a small edit on hildon desktop... rebooted without crypto disk mounted Mar 13 14:49:37 widget is gone, great Mar 13 14:51:33 now rebooted with crypto disk active again... and it is there again Mar 13 14:51:51 actually... apparently I didn't just implement encrypted homedir... but switchable user accounts ;) Mar 13 14:52:00 in boot menu, if I boot unencrypted, I can get a totally different environment Mar 13 14:53:17 only catch, MyDocs is no longer mounted... argh Mar 13 14:55:24 (account switching) which been the whole purpose Mar 13 14:56:12 if you want some sort of deniability, yes Mar 13 14:57:16 mounting mydocs is simple Mar 13 14:57:30 I know Mar 13 14:57:38 but I want to find out why it no longer works automatically Mar 13 14:57:49 I only blacklisted /dev/dm-* and /dev/mapper/* Mar 13 14:58:12 ITEM_NAME="IBM OS/2 Warp 5 Mobile" Mar 13 14:58:18 wonder if THAT is still "plausible" ;) Mar 13 14:58:23 ooh blacklist? my little hackl? Mar 13 14:58:37 yes, ebcause for some reason with a dm-crypt device opened at boot Mar 13 14:58:52 the ke-recv thing tries to mount /dev/dm-0 and not /dev/mmcblk0p1 as MyDocs Mar 13 14:59:07 so I tried blacklisting /dev/dm-0 Mar 13 14:59:14 but now nothing gets mounted as MyDocs Mar 13 14:59:43 /dev/dm-0 is a device node created by device-mapper, I cannot REALLY avoid creating it Mar 13 14:59:57 even if I deleted it, it'd still appear in /proc/partitions and the like Mar 13 15:01:53 I also noticed another issue... something is wrong with the parted package Mar 13 15:01:56 it cannot resize FAT partitions Mar 13 15:02:04 I ended up rather doing it via backupmenu's usb mass storage support Mar 13 15:02:14 it claims to do it, "moving data", but nothing changed in the end Mar 13 15:03:11 lol Mar 13 15:03:24 BTW, the blacklist does NOT work properly... "mass storage mode" of USB still is broken Mar 13 15:03:26 [22439.310163] sd 29:0:0:0: [sdc] 2094072 512-byte logical blocks: (1.07 GB/1022 MiB) Mar 13 15:03:33 that's not the VFAT partition, that's my encrypted drive Mar 13 15:03:39 i.e. still /dev/dm-0 Mar 13 15:03:54 but there MUST be ONE common place where it decides which device to use for MyDocs Mar 13 15:05:00 well, the blacklist script is just a suggestion Mar 13 15:06:05 yes, it was a good idea Mar 13 15:06:08 just didn't work out Mar 13 15:08:00 it been intentionally simplistic Mar 13 15:11:41 hal-set-property --udi /org/freedesktop/Hal/devices/volume_uuid_eba1177e_6662_4421_86ab_2dedb96b74bb --key block.storage_device --string "" Mar 13 15:11:44 after this command it "works" Mar 13 15:12:08 well, simplistic... where DOES it make the decision which device to use? Mar 13 15:12:25 [22995.255798] sd 30:0:0:0: [sdc] 54536128 512-byte logical blocks: (27.9 GB/26.0 GiB) Mar 13 15:12:28 also right one for USB Mar 13 15:15:03 ULOG_WARN_F("%s has more than one children, " Mar 13 15:15:04 "using the first one", udi); Mar 13 15:15:07 argh... ok, this tells it all Mar 13 15:15:22 it simply uses the first device HAL returns, and HAL has no specific ordering Mar 13 15:16:41 can the order in which HAL returns stuff be influenced somehow? Mar 13 15:17:46