IRC log of #maemo-ssu for Monday, 2013-06-10

kerioyay00:00
keriodid you downgrade system-services?00:00
merlin1991why should I?00:00
Palimerlin1991: Skry got bootloop after updating to last cssuT00:01
PaliI wrote him to join #maemo-ssu00:02
merlin1991just when I thought we fixed it00:02
Palimerlin1991, ask more Skry, I do not have time now for solving problems...00:04
PaliSkry will contact you00:04
tadzikupdate went flawless here too, cssu-thumb00:07
*** _rd has joined #maemo-ssu00:07
merlin1991hm I've got a bug for arcean00:08
*** _rd has quit IRC00:08
merlin1991a silly looking one00:08
*** NIN101 has quit IRC00:09
merlin1991http://cdnm.at/~christian/maemo/cssu/Screenshot-20130609-230706.png00:12
*** eddyb has quit IRC00:13
*** futpib__ has quit IRC00:20
*** peterpark has joined #maemo-ssu00:35
peetahbootloop again for me with upgrade from CSSU testing thumb to T8.100:50
merlin1991sd in or out, if it boots without sd check if you're relly on 8.100:50
peetahI tried both, and both whent bootlooping00:51
peetahs/whent/went/00:51
infobotpeetah meant: I tried both, and both went bootlooping00:51
merlin1991now this is a bad one00:51
*** nox- has joined #maemo-ssu00:52
merlin1991so it doesn't boot at all?00:52
peetahI'll try again a few times from clean powered off device via bootmenu and then I'll report00:52
peetahit boots but does not get the dots00:53
peetahnope, it really doesnot want to go futher than the "before dots" screen, sd in or out00:57
merlin1991we'll need Pali/freemangordon for this I fear00:58
peetahrestoring backup then ...00:59
merlin1991can you boot into rescueos first?01:00
peetahtoo late sorry01:00
peetahI'll try tomorrow: what do you need ?01:01
merlin1991check if you really got the latest version :)01:03
merlin1991if you boot rescueos, mnt maemo root and "grep -A8 mp-fremantle-community-pr /mnt/maemo/var/lib/dpkg/status"01:03
merlin1991then you should see the version01:04
peetahis there more than one repository on maemo.org for cssu testing, like load balancing or something ? because I fail to see how I could have downloaded T8 instead of T8.101:04
peetahbut I'll do that and will report here01:05
*** Martix has quit IRC01:05
*** BCMM has joined #maemo-ssu01:05
*** kolp has quit IRC01:05
merlin1991peetah: ham doesn't always dl the distribution files, so you might not get the lates stuff unless you refresh the catalogues01:06
peetahok01:06
merlin1991for example Skry had the very same problem today01:07
*** Pali has quit IRC01:08
peetahso I first try to update as usual, then if it bootloops, check version with rescueos, then restore backup, then refresh catalog, then update and ... pray :)01:08
merlin1991yep01:09
*** peterpark_ has joined #maemo-ssu01:21
*** peterpark has quit IRC01:24
*** XDS2010_ has joined #maemo-ssu01:56
peetahcould'nt resist ... it is 8.1 and it is bootlooping with and without sd in. Led color is varying though before rebooting: sometimes white, sometimes yellow, sometimes redish, without a specific pattern02:04
peetahnow restoring backup, then sleep! bye02:06
DocScrutinizer05 LED? merlin1991, you added my suggestion to give progress indicator - particularly for fsck - during boot via indicator LED?02:09
merlin1991nope fsck is disabled now it shouldn't even run02:09
DocScrutinizer05mhm02:10
DocScrutinizer05probably still better than running it against wrong partition, but... .oO(hmmmm)02:10
peetahwhen I say varying, I mean between reboot, not during the same boot02:10
DocScrutinizer05yep, got that02:11
DocScrutinizer05thanks for disambiguation nevertheless02:11
DocScrutinizer05I really wonder if that fstab hack been worth it02:12
merlin1991I really wonder if it is that fstab hack that is causing peetahs bootloop02:17
peetahwhich hack ?02:18
merlin1991the genfstab.awk is replace with code in rcS-late02:18
merlin1991basically from one ugly hack transformed to a less ugly hack02:19
peetahI should have checked what was in fstab before restoring then ...02:19
DocScrutinizer05btw (OT) is there a way to have (globally defined) $var in apache .html/.php? E:G like var OWNDOMAIN=maemo.org    and <a href=http://wiki.${OWNDOMAIN}/flashing>02:20
peetahmod_include02:20
DocScrutinizer05merlin1991: ^^^ :-)02:21
DocScrutinizer05I'm sure I could figure sth to deliver the "right" html content to the right IP ranges then :-D02:22
peetahsomething like <!--#set owndomain=maemo.org --> in your .html02:22
DocScrutinizer05peetah: that would still mean I had to change every single foobar.htm file02:22
DocScrutinizer05as soon as the domain name changes02:23
peetahthat's what sed is usually used for02:24
DocScrutinizer05how would I run sed inside apache?02:26
peetahsorry I did'nt get it right: I thought you had02:27
DocScrutinizer05think of same page getting delivered under two different domains02:27
peetahsorry I did'nt get it right: I thought you wanted to have painless way to change your htmls for various domains02:27
DocScrutinizer05yes02:27
DocScrutinizer05but realtime02:27
peetahthen I don't know, but I'm not an apache expert ..02:28
peetahbackup restored now, so I'll wish you a nice whatever time of the day ! bye02:29
*** BCMM has quit IRC02:41
*** LaoLang_cool has joined #maemo-ssu03:36
*** LaoLang_cool has quit IRC03:43
*** peterpark_ has quit IRC04:16
*** LauRoman has quit IRC04:18
*** nox- has quit IRC05:26
*** amiconn_ has joined #maemo-ssu05:50
*** amiconn has quit IRC05:50
*** amiconn_ is now known as amiconn05:50
*** Milhouse has quit IRC06:00
*** lartza_ has quit IRC06:39
*** lartza_ has joined #maemo-ssu06:46
*** Milhouse has joined #maemo-ssu06:47
*** Milhouse has quit IRC06:53
*** Milhouse has joined #maemo-ssu07:51
*** M13 has joined #maemo-ssu07:52
*** Milhouse has quit IRC07:56
*** Milhouse has joined #maemo-ssu08:09
*** jon_y has quit IRC08:11
*** unclouded has joined #maemo-ssu08:19
*** b-r-a-i-n has joined #maemo-ssu08:24
*** Martix has joined #maemo-ssu09:15
*** futpib has joined #maemo-ssu09:18
*** LauRoman has joined #maemo-ssu10:20
*** Pali has joined #maemo-ssu10:36
*** hardo has joined #maemo-ssu11:02
hardohi11:02
hardosomebody here11:02
freemangordon~ask11:04
infobotQuestions in the channel should be specific, informative, complete, concise, and on-topic.  Don't ask if you can ask a question first.  Don't ask if a person is there; just ask what you intended to ask them.  Better questions more frequently yield better answers.  We are all here voluntarily or against our will.11:04
*** unclouded has quit IRC11:04
*** unclouded has joined #maemo-ssu11:07
hardoare the problems with cssu testing fixed ?11:13
hardobecause yesterday I tried to install cssu testing and it didn't work11:14
freemangordonhardo: refresh in HAM11:14
hardothanks11:15
hardohey I got problems with the cssu11:40
hardotesting11:40
hardoafter installing the update I won't turn on11:41
*** retard has joined #maemo-ssu12:06
Drathirfreemangordon: loop sloved?12:06
Drathirmornin12:06
hardosome people report positive12:07
hardoI have problems12:07
DrathirDocScrutinizer05: maybe is possibility include something that in php ?12:12
*** piscodig has joined #maemo-ssu12:13
*** piscodig has joined #maemo-ssu12:13
*** discopig has quit IRC12:14
*** piscodig is now known as discopig12:32
*** LauRoman has quit IRC12:42
*** BCMM has joined #maemo-ssu12:46
*** jon_y has joined #maemo-ssu12:55
*** jon_y has quit IRC13:01
*** jon_y has joined #maemo-ssu13:03
freemangordonDrathir: it should be13:08
hardoyes everthing is fine13:10
hardoi solved the problem ^13:10
hardobut i have the next problem13:10
hardoi want to update the kernel power13:11
hardoto 5213:11
hardodoes that mean I have to activate the cssu (devel) repositorie)13:11
Drathirhardo: how sloved?13:22
Drathirbotsnack13:22
hardo@drathir13:22
hardoI flashed my n90013:22
hardothen take out the sd card13:23
Drathiroh i seee13:23
hardoand go an webpage and install cssu testing from there13:23
Drathirany errors show when try to update KP 52?13:24
hardoi don t know how to update13:24
hardoi found in ham13:24
hardolinux kernel for cssu13:24
hardoand13:24
hardolinux kernel for power user13:25
hardoi don t know which one i need13:25
hardoi even don t know if i have to activate the repositories from community (devel)13:26
Drathiri have "harder" config because i have also u-boot on board...13:26
hardoi see...13:27
Drathirmenu settings about cssu13:27
hardowhat is there ?13:27
Drathirwhat flavor you have there?13:28
hardo21.2011.38-1Tmaemo8.113:28
hardotesting13:28
Drathirthere only information about cssu installed13:28
hardomy cssu went fine13:29
DrathirVersion: 21.2011.38-1Tmaemo7.213:29
Drathirlooks like updated sucessfull13:29
hardoyes13:29
hardouname -r shows13:30
hardo2.6.28-omap113:30
hardothat means the kernel is not up to date13:31
hardoi got to find a way to update the kernel to 5213:31
DrathirLinux Nokia-N900 2.6.28.10-power52 #1 PREEMPT Sat Apr 6 11:59:23 UTC 2013 armv7l GNU/Linux13:31
hardoso13:32
Drathiridk if this is possible without u-boot...13:32
hardolinux kernel for power user should be right ?13:32
hardoin ham13:32
DrathirDocScrutinizer05 maybe can help more hardo...13:33
Drathiri dont really want to say you something cause boot loop and reflash again a phone...13:35
Drathirbut looks like dependencies works http://pastebin.com/Z5maHz9t13:40
*** kolp has joined #maemo-ssu13:43
*** discopig is now known as piscodig14:01
*** LauRoman has joined #maemo-ssu14:20
*** M13 has quit IRC14:26
*** unclouded has quit IRC14:38
merlin1991hardo: when you don't have u-boot then you should enable extras-deve and install kernel-power-flasher15:09
peetahmerlin1991: are there more users experiencing bootloop even with T8.1 or am I still the only one ?15:26
merlin1991so far only hardo had it but for him it went fine when he reflashed and then installed T8.115:27
hardopeetah my fault was that I forgot about the sd card15:29
merlin1991hardo: well you can put it in now, it shouldn't affect it anymore15:31
peetahok so odds are high that my problem could come from whatever change I could have made in critical files in the past... now which one , I don't know and I really don't want to reflash, so I'll have to postpone this upgrade and dig a little more in my /etc15:31
peetahhardo: what about the sd card ?15:32
hardonothing special15:32
hardotoday i flashed my n90015:32
hardoand after installing cssu my n900 didn't reboot15:33
hardoi think for obvios reseon ^15:33
hardoreason15:33
*** hardo has quit IRC15:56
Drathirhmm now /me thinkink about apply of cssu update ^^16:06
DocScrutinizer05peetah: check rcs-late, preinit, etc16:10
DocScrutinizer05peetah: particularly those that might have changed in T816:10
DocScrutinizer05alas it seems CSSU doesn't (try to) apply patches to those files, but rather ruthlessly replaces them no matter if they are original or customized16:12
*** X-Fade has quit IRC16:12
*** X-Fade has joined #maemo-ssu16:13
peetahafter upgrade, and bootloop, I have a new /var/log/fsck_home.log that says fsck has been cancelled: is this normal ?16:18
DocScrutinizer05well, the T8 fstab hack also touched the already botched fsck mechanism16:24
DocScrutinizer05aiui16:24
peetahdmesg does not give any useful info16:24
peetahI installed sysklogd before upgrading, but it doesn't seems to fill in the /var/log/syslog during the bootloop16:25
DocScrutinizer05and caused fsck to check the wrong MMC since the name of eMMC changes depending on whether or not a uSD been inserted during boot16:25
peetahany other place where I could gather information about the boot process ?16:26
DocScrutinizer05debugging boot process is really hard16:26
peetahbut checking the wrong MMC should simply send a warning, not preventing boot no ?16:27
DocScrutinizer05I think the schema been: do fsck to /home, if it succeeds continue booting; if it had to fix errors try to reboot after fsck exited, to make home remount cleanly16:28
DocScrutinizer05or sth along that line16:28
DocScrutinizer05merlin1991 claimed he kicked out resp disabled fsck completely as a stopgap measure in T8.116:30
DocScrutinizer05it's quite unfortunate that this whole fstab patch made it into CSSU at all, without proper testing and perr review16:31
DocScrutinizer05peer even16:31
merlin1991peetah: check /etc/default/mount-opts16:40
merlin1991and make sure the various _fsck options have a "0" settings16:40
*** piscodig is now known as discopig16:43
peetahmerlin1991: no _fsck options in mount-opts, only home_opts and fat_opts16:45
merlin1991then check if you have any /etc/default/mount-opts.* file16:45
peetahyep dpkg-dist16:46
merlin1991cp yours into a backup mv that into place and merge changes16:46
peetahwow much more options there :)16:46
merlin1991also bitchslap ham for not overwriting it and thus ruining your day16:46
peetahwhat about my own options which do not seem to appear in .dpkg-dist file ? do I copy them there or are they now not used by any scripts ?16:49
merlin1991what are your own options?16:49
peetahhome_opts="rw,noatime,errors=continue,commit=1,data=writeback"16:50
DocScrutinizer05[2013-06-10 15:46:41] <merlin1991> [...] merge changes16:51
peetahfat_opts="noauto, nodev,noexec,nosuid,noatime,nodiratime,utf8,uid=29999,shortname=mixed,dmask=000,fmask=0133,rodir"16:51
peetahok it seems that they are overwritten by rcS-late16:52
DocScrutinizer05bitchslap CSSU for overwriting config files instead of applying diff patches16:52
merlin1991eval home_opts=$common_opts,$root_opts,\$${home_type}_opts16:52
peetahat least for home_opts16:52
peetahmerlin1991: exactly16:53
peetahbut fat_opts is not used anywhere16:53
peetahat least in event.d directory16:53
merlin1991I think it's vfat_opts now16:53
merlin1991hm not used either16:53
peetahso so I have to put them in mount-opts-overwrite16:54
peetah?16:54
*** hardo has joined #maemo-ssu16:54
peetahor just ignore them, they do not seem to be very exotic16:54
merlin1991there is $mydocs_opts now though16:54
merlin1991and yes using mount-opts-overwrite is probably a better choice16:55
DocScrutinizer05md5sum original "obsolete" files, compare md5sums to the master list that comes with "the update", warn if any of the original files been changed wrt what patch developer expected to see, the apply / not apply diff-patches to user's system16:55
DocScrutinizer05it's pretty silly to simply overwrite any existing files16:56
merlin1991well it doesn#t overwrite them at all16:57
merlin1991it places a new file $file.dpkg-dist16:57
DocScrutinizer05that's almost as silly16:57
DocScrutinizer05at least without a FAT WARNING16:57
DocScrutinizer05idealy a diff-patch should get applied16:58
DocScrutinizer05which would do the merges you suggested above16:58
DocScrutinizer05user still should check if the merge went ok16:58
DocScrutinizer05when the original file been different to what devel expected his patch gets applied to16:59
DocScrutinizer05also a backup of original file must get kept16:59
*** Milhouse has quit IRC17:00
peetahstill looping17:00
merlin1991hm check if you have any other .dpkg-dist files on the system17:00
peetahalready did, but nothing except for openvpn which should not influence the boot process17:01
DocScrutinizer05maybe disabling watchdogs could help?17:01
merlin1991then there's nothing left but to add various echo "got past section x" >> /var/log/rcs.log lines into /etc/event.d/rcS-late and see where it stops17:02
peetahthat's what I start to do17:03
DocScrutinizer05tbh those diagnostic prints should be in every boot related script. Maybe even set -vx17:03
DocScrutinizer05exec >>/var/log/rcs.log17:05
DocScrutinizer05err17:05
DocScrutinizer05exec 2>>/var/log/rcs.log ?17:05
DocScrutinizer05the orthodox boot debug procedure is to attach a serial console, btw17:09
DocScrutinizer05that's what the UART-3(?) testpoints under battery are for17:10
merlin1991I bet the kernel  config doesn't have the serial console bootline on the shipped devices17:11
DocScrutinizer05on Freerunner, you got a debug board for that17:11
DocScrutinizer05flasher allows defining new boot kernel cmdline17:11
*** Milhouse has joined #maemo-ssu17:12
DocScrutinizer05so does uBoot17:12
DocScrutinizer05and setting boot options via flasher enables serial console too17:12
DocScrutinizer05http://wiki.maemo.org/Documentation/Maemo_5_Developer_Guide/Development_Environment/Maemo_Flasher-3.517:13
DocScrutinizer05--set-rd-flags=serial-console17:14
DocScrutinizer05obvious, uh?17:15
DocScrutinizer05nfc zhough what's sti-console17:15
*** Milhouse has quit IRC17:20
*** Milhouse has joined #maemo-ssu17:32
*** Milhouse has quit IRC17:38
peetahok, so it seems line "initctl emit MOUNTS_OK" of rcS-late is not in love with my n900... does it ring a bell ?17:43
peetahclock_daemon, hildon-control-panel-personalisation, rc-clean-fail-thumbnail ans xsession start on MOUNTS_OK17:45
peetahwhich disable order would make more sense ?17:46
DocScrutinizer05diable order?18:02
DocScrutinizer05also, what means "is not in love with my n900"18:02
peetahit does not do anything past this line18:03
DocScrutinizer05does it execute that line? does it return from that line?18:03
Drathiri good understand that is possible install KP without u-boot?18:04
DocScrutinizer05sure18:04
Drathirin that case oryginal is only exchanged with KP one?18:05
DocScrutinizer05that's the usual way to do it18:05
peetahit does not print anything after that line18:05
DocScrutinizer05yes18:05
Drathirok thanks now i'll be know that if someone ask about that...18:06
DocScrutinizer05peetah: initctl emit MOUNTS_OK  sends a signal to initctl/upstart, which in turn makes upstart start all scripts that are defined as waiting for that signal. any of the jobs/services started could cause reboot18:06
*** Milhouse has joined #maemo-ssu18:07
DocScrutinizer05peetah: check /etc/events.d for files with "start (on|after) MOUNTS_OK"18:08
peetahthat's why I asked which disable order of the started services cited above would make more sense in order to point the culprit18:08
DocScrutinizer05cd /etc/event.d/; grep MOUNTS_OK *18:10
DocScrutinizer05you won't gain anything by disabling those services18:10
DocScrutinizer05you need to check which one fails on getting started18:11
DocScrutinizer05a cascaded "sleep N" at beginning in any of the services, plus a few echo lines (or logger lines) should help to serialize the jobs and see which one gets started but never finishes18:12
Drathirbut anyway i little scary to helping ppl with kernel questions is too important and critical thing in the system... and too responsible to my knowledge...18:12
DocScrutinizer05I also found >> hildon-control-panel-personalisation.dpkg-new:start on MOUNTS_OK<<18:13
DocScrutinizer05but that's on *my* (obsolete) system, not related to T818:13
DocScrutinizer05peetah: I'd figure e.g. xsession or hildon-control-panel-personalisation or rc-clean-fail-thumbnail might have startjobs that use dsmetool to monitor the process and when the service fails and segfaults for some reason (not unseen for e.g. H-D-widgets when config got messed up) the dsme reboots system after 3 failed respawns in 60s18:19
*** Milhouse has quit IRC18:19
peetahDocScrutinizer05: pfiouuuuu spent too much time on it already ...restoring backup and will look at this sometimes later in the week. thanks for all advices and help !18:21
DocScrutinizer05yw18:22
*** arcean has joined #maemo-ssu18:27
*** Milhouse has joined #maemo-ssu18:32
*** Milhouse has quit IRC18:38
*** NIN101 has joined #maemo-ssu18:40
*** LauRoman has quit IRC18:47
DocScrutinizer05Pali: freemangordon: Jun 10 17:47:14 IroN900 kernel: [85239.571258] OMAP3 ROM RNG: HW init failed: 118:49
DocScrutinizer05on cat /dev/hwrng18:49
DocScrutinizer05:-S18:49
Palibad18:49
DocScrutinizer05is hwrng a feature of PK?18:50
*** Milhouse has joined #maemo-ssu18:50
*** Milhouse has quit IRC18:58
Drathiri have the same [26386.323333] OMAP3 ROM RNG: HW init failed: 119:10
*** Milhouse has joined #maemo-ssu19:11
*** xmlich02 has quit IRC19:11
*** xmlich02 has joined #maemo-ssu19:14
*** Milhouse has quit IRC19:28
*** Milhouse has joined #maemo-ssu19:40
DocScrutinizer05I consider this a possible security threat19:45
freemangordonDocScrutinizer05: which kernel?19:45
DocScrutinizer05IroN900:~# uname -a19:46
DocScrutinizer05Linux IroN900 2.6.28-omap1 #1 PREEMPT Fri Aug 6 11:50:00 EEST 2010 armv7l GNU/Linux19:46
DocScrutinizer05grep PRNG -A6 /etc/event.d/rcS-late19:47
freemangordonno such error here19:49
freemangordonoh, wait19:49
DocScrutinizer05cat /dev/hwrng19:49
freemangordonDocScrutinizer05: do I need to cat /dev/?19:49
freemangordonok19:49
freemangordon[17999.453399] OMAP3 ROM RNG: HW init failed: 119:49
freemangordonhmm, wtf?19:49
DocScrutinizer05and worse: no output of cat /dev/hwrng19:50
freemangordoncat: read error: Input/output error19:50
DocScrutinizer05that stderr19:50
DocScrutinizer05;-P19:50
freemangordonDocScrutinizer05: though cat /dev/random works19:51
DocScrutinizer05that doesn't matter19:51
DocScrutinizer05      echo "Seeding entropy pool"19:51
DocScrutinizer05      dd if=/dev/hwrng of=/dev/urandom bs=$URANDOM_POOLSZ count=119:51
freemangordonnot sure if cat /dev/hwrng is the correct thing to be done19:51
*** amiconn_ has joined #maemo-ssu19:51
*** amiconn has quit IRC19:51
*** amiconn_ is now known as amiconn19:51
DocScrutinizer05grep PRNG -A6 /etc/event.d/rcS-late19:52
freemangordonyep, saw it19:52
* DocScrutinizer05 wonders if that's genuine19:52
DocScrutinizer05or CSSU19:52
* freemangordon checks on gitorious19:52
DocScrutinizer05anyway it seeds entropy with ever-same null stuff19:53
DocScrutinizer05or not at all19:53
DocScrutinizer05which could result in all devices worldwide use same random sequences for tcp stack and whatnot19:53
DocScrutinizer05--> huge security vulnerability19:54
freemangordonDocScrutinizer05: nokia:19:55
freemangordonhttps://gitorious.org/community-ssu/upstart/blobs/14e703ee87703880d80bd14b574cbbc96bfe02fe/debian/event.d/system-services/rcS-late#line13319:55
DocScrutinizer05thought as much19:56
freemangordonDocScrutinizer05: though...19:56
freemangordonif the line in rcS-late fails, the device wotn boot19:57
freemangordonlemme try with dd19:57
freemangordonDocScrutinizer05: dd gives the same error19:58
freemangordoni'll assume that a false alarm19:58
freemangordonif we get the same error in rcS-late, we're in bootloop19:58
*** Vlad_on_the_road has joined #maemo-ssu20:05
kerioDocScrutinizer05: bitchslap HAM for not presenting the standard "config file changed, overwrite or keep?" dialog on package update20:06
keriodiff patches are kinda hard with the way we're doing config files throughout the whole of maemo20:06
DocScrutinizer05well, not everything is a config file20:06
kerioif it's not a config file, it's the user's fault for modifying it under dpkg's nose20:07
DocScrutinizer05meh20:07
kerio(or the package maintainer's fault for not marking it as a config file)20:07
DocScrutinizer05check who's maintainer and what's package of rcs-late20:08
DocScrutinizer05or of mount-opts20:08
DocScrutinizer05that's exactly why CSSU got _invented_20:09
DocScrutinizer05so we can't say we're CSSU and not our fault20:09
kerioanyway, before pali had finalized the latest system-services, i reported a bootloop that i couldn't properly investigate due to lack of time20:09
keriobut my config is way too nonstandard to be of much use :<20:09
*** sixwheeledbeast has joined #maemo-ssu20:10
DocScrutinizer05actually CSSU should try to cope with as many config variations as reasonably feasible20:11
DocScrutinizer05and for sure we MUST NOT assume a completely standard system as granted20:12
DocScrutinizer05rather we MUST check if the system actually _is_ compatible with whatever we try to propagate to all users20:13
keriothis change, if it worked properly, kinda needed some user intervention, in case of a nonstandard config20:13
keriobut that's to be expected, a nonstandard config must modify rcS-late20:14
DocScrutinizer05so indiscriminately replacing whatever rcs-late we find on user's system for sure is a extremely silly and stupid thing, not in line with best practice20:14
keriommh, you can't quite do it differently though20:14
kerionot with the current HAM at least20:14
freemangordonDocScrutinizer05: and what do you propose to do instead?20:14
keriofreemangordon: postinst scripts! :D20:14
DocScrutinizer05you can a) do a md5sum to check if system has unaltered files like rcs-late, and b) apply diffs rather than replace complete files20:15
kerioa diff here would still be catastrophical20:15
kerioit's quite a substantial change20:15
keriothe correct thing to do is to ask the user20:16
freemangordonand what if diff cannot be applied?20:16
keriowhich is what dpkg does20:16
freemangordonwell, but HAM does not20:16
freemangordonso file a bug against HAM20:16
keriois it a bug, though?20:16
keriofeature request20:16
freemangordonstill, I doubt it will be implementd20:17
*** Milhouse has quit IRC20:17
DocScrutinizer05kerio: see what I suggested before: check md5sum, ifequ: (apply patch, exit), warn user, create backup, apply patch, present to user and ask him if that's OK, exit to editor20:19
keriowait, apply patch? at that point you can just replace the file20:20
DocScrutinizer05so you want to ship a diff AND a full replacement file? why?20:21
kerioand yeah, that's pretty much what dpkg does, except that there's no diff between old and new so your choices are to replace, keep or view a diff between current and new20:21
*** hardo has quit IRC20:21
keriocurrent as in the current config file20:21
DocScrutinizer05yes, dpkg should use diffmerge istead of diff20:22
freemangordonBTW is rcS-late really a config file?20:23
kerioit's in /etc20:23
DocScrutinizer05no20:23
DocScrutinizer05that's why it is no HAM bug20:23
freemangordonkerio: I don;t care where it is20:23
DocScrutinizer05ham and dpkg/apt work correctly20:24
DocScrutinizer05it's CSSU maintainers that *handle* it incorrectly20:24
freemangordonso, it is Nokia to blame for not doing it in the correct way initially20:24
freemangordonand we can fix it20:25
DocScrutinizer05we should make HAM/apt *think/ rcs-late was a config file20:25
keriomarking it as not a config file would allow it to be properly dpkg-diverted20:25
DocScrutinizer05(or any other of those scripts, for that topic)20:25
freemangordonDocScrutinizer05: wouldn't a correct fix be to move that script to /sbin and to replace rcS-late with a wrapper to it?20:25
kerioso user modifications will stay, and the user will be responsible for it20:26
keriootherwise dpkg will make rcS-late.dpkg-new in that directory, which fucks up the boot process20:26
DocScrutinizer05kerio: valid point. Yes we probably shouldn't use dpkg/apt at all for it, rather postinst scripts and handle that stuff manually20:27
kerio+1 for anything that removes responsibilities from upstart20:27
DocScrutinizer05freemangordon: ummm - I need to ponder your suggestion20:28
*** Milhouse has joined #maemo-ssu20:29
DocScrutinizer05we basically need sth similar to quilt/bitbake20:30
DocScrutinizer05by structure20:30
DocScrutinizer05the .deb dpkg is fubar on root, by Nokia abusing stuff with their mp-fremantle-pr20:31
DocScrutinizer05and since everybody has mp-fremantle-pr, we can't assume everybody did any modifications s/he needed in e.g. initscripts in a way that's compatible with .deb/dpkg20:32
DocScrutinizer05so all we could do via HAM/apt/dpkg is to deliver "empty" packages that apply a diff patch to system20:33
DocScrutinizer05since basically *any* user modification to any part included in mp-fremantle-pr is messing up the whole dpkg concept20:35
DocScrutinizer05or rather, CSSU would mess up user's system by installing a new mp-fremantle-pr20:35
DocScrutinizer05that's why I always said CSSU shall leave mp-*-pr behind20:36
DocScrutinizer05unpack it into as many independent pkgs as possible20:36
*** Pali_ has joined #maemo-ssu20:38
*** Milhouse has quit IRC20:38
*** b-r-a-i-n has quit IRC20:39
*** b-r-a-i-n has joined #maemo-ssu20:40
*** kerio_ has joined #maemo-ssu20:41
*** BCMM has quit IRC20:42
*** Pali has quit IRC20:42
*** wmarone_ has quit IRC20:42
*** kerio has quit IRC20:42
*** kerio_ is now known as kerio20:42
sixwheeledbeastI have no issues with upstart, why the hate?20:44
kerioit's not sysv20:45
sixwheeledbeastget with the times :)20:48
sixwheeledbeastupstart was made as a better replacement for sysV20:49
DocScrutinizer05upstart as well as systemd have abysmal "UI"20:49
kerioyeah, so it can boot faster by using all the multiple cores the n900 has20:49
kerioOH WAIT20:49
DocScrutinizer05while sysV has a clear "API" by ls /etc/init.d/*, upstart and particularly systemd offer no clean API/UI to find out what's going on and what are the sysop's options to do now20:51
DocScrutinizer05sysV:  for service in /etc/init.d/*; do $service status; done20:52
DocScrutinizer05upstart: ?????20:52
DocScrutinizer05and systemd even abandons runlevels20:53
keriobut runlevels are so unix, DocScrutinizer20:53
keriopeople don't use anything but desktops or tablets20:54
kerioyou should know that20:54
DocScrutinizer05yeah, so let's see if we find some concept that's more windozy20:54
DocScrutinizer05like systemd needing dbus(!!!) to even start up a system, and having his own shell to communicate with it20:55
DocScrutinizer05clearly an ideal proces#020:55
DocScrutinizer05both by system architecture as well as by resource requirements20:56
*** Martix has quit IRC20:56
DocScrutinizer05and here comes my killer argument why both upstart and systemd are hell spawn: upstart is canonical, systemd is poettering20:57
*** Martix has joined #maemo-ssu20:57
DocScrutinizer05both entities are notorious for evil attitude to take unix hostage and turn it into sth that meets their very own design goals and _nothing_ else20:58
DocScrutinizer05generations of linux devels invested eternities of their worktime to make linux a unix-compatible system. Poettering considers this a major *flaw* of linux21:00
DocScrutinizer05plain hybris21:00
DocScrutinizer05"we don't need (easy porting of) *any* of the tools or libs or toolkits or filesystems or drivers of any of the UNIX systems. We are linux, we have everything we need and staying compatible with UNIX blocks our oh-so-devine creativity to have networked audio during user login"21:03
*** amiconn has quit IRC21:22
*** amiconn has joined #maemo-ssu21:22
DocScrutinizer05freemangordon: I don't think the line "" dd if=/dev/hwrng of=/dev/urandom bs=$URANDOM_POOLSZ count=1 "" throwing error would cause a bootloop21:38
DocScrutinizer05after all rcs-late is a upstart script, and there's not much essential after that line in the script section of /etc/event.d/rcS-late21:39
DocScrutinizer05anyway I got 3 lines with RNG in my syslog:21:42
DocScrutinizer05May 31 15:39:58 IroN900 kernel: [    2.623168] OMAP3 ROM RNG: initializing21:42
DocScrutinizer05during regular boot21:42
DocScrutinizer05and then21:43
DocScrutinizer05Jun 10 17:47:14 IroN900 kernel: [85239.571258] OMAP3 ROM RNG: HW init failed: 121:43
DocScrutinizer05Jun 10 17:48:17 IroN900 kernel: [85302.464813] OMAP3 ROM RNG: HW init failed: 121:43
DocScrutinizer05during my former tests21:43
DocScrutinizer05freemangordon: Pali_: could you spot the line in kernel printing that error msg?21:44
DocScrutinizer05o.O https://lkml.org/lkml/2013/3/28/53621:47
*** Milhouse has joined #maemo-ssu21:48
*** wmarone_ has joined #maemo-ssu21:48
DocScrutinizer05I wonder if that's a coincidence that Pali is looking into it recently. I stay with L.J.Gibbs: "I don't believe in coincidences"21:49
Pali_DocScrutinizer05: I ported that driver to 3.9 kernel21:57
*** Pali_ is now known as Pali21:58
*** Milhouse has quit IRC21:58
PaliDocScrutinizer05: http://mxr.maemo.org/fremantle/source/kernel/drivers/char/hw_random/omap3-rom-drv.c#8122:01
Paliit failing call_sec_rom function which is wrapper around smc instruction: http://mxr.maemo.org/fremantle/source/kernel/drivers/char/hw_random/omap3-rom-asm.S22:02
Paliit looks like your omap board is broken22:03
PaliDocScrutinizer05: try to turn device off and on. NOLO has some code which doing something with M-Shield, so maybe it can reset it22:04
DocScrutinizer05my omap board broken?? duh!22:04
Palisecure side of omap cpu returned error22:05
DocScrutinizer05maybe I used a cmd that "broke2 it22:05
DocScrutinizer05this could've been softupd only, for all I can tell22:06
DocScrutinizer05well, ok. I'll give my daily phone another reboot - way too early since it has an uptime of only 10 days22:07
*** Milhouse has joined #maemo-ssu22:10
Drathirhrhr updates knoking to me ^^22:45
Drathirknocking*22:45
*** futpib has quit IRC22:51
merlin1991DocScrutinizer05: upstart has a status check22:55
merlin1991service $service status22:55
DocScrutinizer05uhu, and who tells me about content of $status?22:56
DocScrutinizer05err $service22:56
merlin1991 /etc/event.d22:57
merlin1991but since the upstart on maemo os more than old its not even service ... but "status $service"22:58
DocScrutinizer05hmm I wonder if files like ke-recv.dpkg-new and icd2.dpkg-new in /etc/event.d do any good23:02
DocScrutinizer05http://privatepaste.com/efeb0ad74f23:06
*** macmaN has quit IRC23:10
DocScrutinizer05http://privatepaste.com/a46c5d990c23:11
DocScrutinizer05http://privatepaste.com/910207db6023:12
freemangordonPali: it could be that hwrng cannot be initialized/called more than once23:24
PaliDocScrutinizer05: XYZ.dpkg-new files replacing XYZ after successfull installation/upgrade23:25
freemangordonif it is broken this https://gitorious.org/community-ssu/upstart/blobs/master/debian/event.d/system-services/rcS-late#line223 would fail on boot23:25
Paliif you edited XYZ file, then apt-get will ask you if you really want to update that files23:26
freemangordonPali: so I'd assume we have NFC how it works23:26
Palifreemangordon: original nokia code was "dd if=/dev/hwrng of=/dev/urandom bs=$URANDOM_POOLSZ count=1"23:26
Pali(or /dev/hw_random) I do not know now23:27
Palione is in vanilla kernel and one in maemo patched udev23:27
freemangordonPali: i looked into it, there is no difference23:27
freemangordonit is somewhere in the backscroll23:27
Paliyou can try to add "|| true" so it skip hwrand error23:27
freemangordonPali: there is no error23:28
freemangordonhttps://gitorious.org/community-ssu/upstart/blobs/14e703ee87703880d80bd14b574cbbc96bfe02fe/debian/event.d/system-services/rcS-late#line13323:28
Paliok23:28
freemangordonPali: if there was error (during bootup) the device will enter reboot loop23:28
Paliright23:28
freemangordonDocScrutinizer05 got this error while doing cat /dev/hwrng23:29
freemangordonand I am not sure adding || true is a good idea23:29
freemangordonas that would mean we'll continue to boot on a faulty HW23:29
*** Vlad_on_the_road has quit IRC23:33
*** texed has joined #maemo-ssu23:35
*** arcean has quit IRC23:35
*** texed has quit IRC23:46
*** chem|st has quit IRC23:58
*** chem|st has joined #maemo-ssu23:58

Generated by irclog2html.py 2.15.1 by Marius Gedminas - find it at mg.pov.lt!