IRC log of #maemo-ssu for Wednesday, 2012-04-11

« Tuesday, 2012-04-10 Index Thursday, 2012-04-12 »
*** lizardo has quit IRC00:03
*** dafox has quit IRC00:07
*** Pali has quit IRC00:14
*** int_ua has joined #maemo-ssu00:44
*** int_ua has quit IRC00:48
*** trbs has quit IRC01:36
*** freemangordon_ has quit IRC01:40
*** freemangordon_ has joined #maemo-ssu01:57
*** freemangordon_ has left #maemo-ssu01:59
DocScrutinizerpali sucks, my tab nick-completion never works for him02:39
DocScrutinizerwell, could be worse, he could have a nick like freemangordon02:39
DocScrutinizer;-P02:39
freemangordonhaha, very funny :p02:40
DocScrutinizerhihihi02:40
*** Raimu-Z has quit IRC03:04
*** Raimu-Z has joined #maemo-ssu03:05
*** arcean has quit IRC03:09
*** KrayonWork has joined #maemo-ssu03:16
*** nox- has quit IRC03:31
*** javispedro has quit IRC04:19
*** M4rtinK has quit IRC04:23
merlin1991hm wft http://talk.maemo.org/showpost.php?p=1190291&postcount=16104:32
merlin1991I think he kinda missed the first post of the stable thread04:33
*** Raimu-Z has quit IRC04:43
*** Raimu-Z has joined #maemo-ssu04:45
merlin1991srly that post made me want to rage05:22
* Lava_Croft hands merlin1991 a stressball05:33
merlin1991hehe thanks :)05:36
*** amiconn_ has joined #maemo-ssu05:40
*** amiconn has quit IRC05:40
*** amiconn_ is now known as amiconn05:40
*** MohammadAG has quit IRC06:46
*** MohammadAG has joined #maemo-ssu06:47
*** LinuxCode has joined #maemo-ssu06:48
*** psycho_oreos has joined #maemo-ssu08:09
*** Pali has joined #maemo-ssu08:39
*** dafox has joined #maemo-ssu09:50
*** DocScrutinizer has quit IRC09:57
*** DocScrutinizer has joined #maemo-ssu09:57
*** BCMM has joined #maemo-ssu11:55
*** arcean has joined #maemo-ssu13:01
*** lizardo has joined #maemo-ssu13:43
*** arcean_ has joined #maemo-ssu14:30
*** arcean has quit IRC14:34
Lava_Croftit is a cruel world where your phone contract is nearly over and there is no single device on the market that you desire to get in combination with a new contract14:49
gulyi'd go with newer iphone, just to regret 9500 communicator14:51
Lava_Croftmy wife has an iphone, thats enough for me14:52
Lava_Crofti only use it to check games14:52
Lava_Croftbesides, the OS is no option14:52
Lava_Croftsame with android :\14:52
Lava_Croftill just scurry off and cry some more14:52
DocScrutinizerget N900 in good condition, maybe offer to swap for $random-phone from your contract15:24
gulywhat about n950 ?15:25
DocScrutinizerthere's definitely no substitute or successor for N90015:25
DocScrutinizereven n950 sucks, I don't use it at all15:26
gulygood devices always die alone15:26
Lava_Crofti have 2 n900s already15:27
Lava_Croftbut they are slow15:27
Lava_Crofti was just a bit whining about how there is no real replacement15:27
gulyi had to wait 6 years after selling my 9500communicator15:28
Lava_Croft5510 here15:29
Lava_Croftn900 replaced it15:29
*** LinuxCode has quit IRC15:34
DocScrutinizerguly: I gather you'll get enough n900 for the next 6 years15:57
gulyi'll do for sure15:59
*** BCMM has quit IRC16:01
*** BCMM has joined #maemo-ssu16:15
*** M4rtinK has joined #maemo-ssu16:19
*** BCMM has quit IRC16:54
*** BCMM has joined #maemo-ssu16:57
*** sigkill_ has joined #maemo-ssu17:18
*** NIN101 has joined #maemo-ssu18:06
*** dafox has quit IRC18:44
*** BCMM has quit IRC18:48
*** M4rtinK has quit IRC18:52
*** M4rtinK has joined #maemo-ssu19:07
*** dafox has joined #maemo-ssu19:20
*** dafox is now known as Guest403319:21
freemangordonDocScrutinizer, Pali ping19:34
Palifreemangordon, pong19:34
freemangordonPali, i am chasing the bug with mmc rename in filemanager19:34
freemangordonare you aware of it?19:34
PaliI did not looked at that bug19:35
freemangordonwell, if you rename a fat partition using filemanager, the voluma label is changed, but even HAL knows nothing about new label and keeps the old one. a reboot is needed so new label to appear correctly19:36
*** int_ua has joined #maemo-ssu19:36
freemangordonmyu question si - could anyone give me some hint where to start from :)19:36
Paliso hal does not update label?19:37
freemangordons/myu question si/my question is/19:37
infobotfreemangordon meant: my question is - could anyone give me some hint where to start from :)19:37
freemangordonyep19:37
freemangordonremains the old one19:37
Palitry to run partprobe ioctl on mmc device19:37
Palihal then should reread partitions from /dev/19:37
Paliand I belive then it update also label19:38
freemangordonbut the voulume is re-mounted19:38
freemangordonby ke-recv afaik19:38
Paliwho is renaming label?19:39
Palike-recv?19:39
freemangordonyep19:39
freemangordonusing mmc-rename.sh19:39
Palisee this: https://bugs.freedesktop.org/show_bug.cgi?id=1036319:40
povbotBug 10363: signature problem19:40
Palishut up povbot!19:40
Palimaybe hal really does not suppot it19:40
freemangordonhmm, could we hack it with some hal_set_property?19:41
freemangordonfrom ke-recv19:41
Palino idea19:41
PaliI think we could really use something to force hal to reread part info19:42
freemangordonPali, what is strange is that the volume is unmounted first and the mounted19:42
freemangordonso HAL should remove the device and re-add it19:42
freemangordonAIUI19:43
freemangordoncould be a problem in ke-recv not unmounting the correct volume?19:43
DocScrutinizertry partprobe19:43
Palifreemangordon, why mmc-rename.sh using mlabel?19:44
PaliI think it should use dosfslabel19:44
freemangordonby historical reasons :D19:44
Palimlabel has stupid windows drive configuration in /etc/mtools.conf19:44
freemangordonyep, but mtools are used everywhere from what I saw19:45
freemangordonsomething to do with USB19:45
freemangordonmass storage mode19:45
DocScrutinizer???19:45
Palialso mtool support only fat19:45
Paliand also only drives configured in /etc/mtools.conf19:46
freemangordonI know, right now we cannot ranme other filesystems using filemanager :D19:46
PaliI think that script should be changed to use dosfslabel and e2label19:46
Pali[fs]label programs has same syntax19:46
freemangordonagree, but lets fix what is broken before doing that19:47
Palifreemangordon, try to run partprobe after renaming label19:47
Palialso partx19:47
Paliif this fixing hal problem19:48
* DocScrutinizer throws up a little at /usr/sbin/mmc-rename.sh19:48
freemangordonno such things here, my busybox is not so fancy as yours :D19:48
DocScrutinizer(me suggests echo partprobe >> /usr/sbin/mmc-rename.sh nevertheless19:49
*** Guest4033 has quit IRC19:49
freemangordonno partprobe here19:49
Palipartprobe and partx calling some syscall19:49
DocScrutinizero.O19:49
Paliyes, maemo5 has no partprobe ....19:50
freemangordonno partx19:50
PaliI see now19:50
Paliseems that messybox is more messy now :D19:50
DocScrutinizeryup, no partprobe19:50
DocScrutinizer:-/19:50
freemangordonhmm, a the remount should do the trick19:50
DocScrutinizernope19:51
DocScrutinizerremount is one level above what you need19:51
freemangordonbut volume labels are not in poartition table AFAIK19:52
* freemangordon is going to ask google19:55
DocScrutinizervolume labels are read when the fs type of partition gets scanned, afaik. Also it seems I recall you label partitions in (*)fdisk19:58
freemangordonhmm, remember DOS format?19:59
DocScrutinizercfdisk has [label}19:59
DocScrutinizerfor each partition20:00
DocScrutinizerI conclude: part labels are related to part table20:00
freemangordonit was asking for volume label at the end, so label is on volume level, and is stored in 1st sector(of the volume), not in partition table IIRC20:00
DocScrutinizerhmmm20:01
DocScrutinizerdd if=/dev/mmcblk1p1 bs=256 count=1|od -c20:02
freemangordonyep, confirmed20:05
freemangordonno need to re-read partition table20:05
DocScrutinizeryou're right, no partition name in MBR20:06
DocScrutinizervolume-labels are primary domain of (mk|tweak)fs20:08
DocScrutinizerstill they are in partition's sector 0 and thus rather low level20:09
DocScrutinizeranyway, depending on what's your error, you need to do the right rereading, and of course you should umount the volume prior to editing label20:10
*** andre__ has joined #maemo-ssu20:10
*** andre__ has joined #maemo-ssu20:10
DocScrutinizerfreemangordon: see blkid, it doesn't care about mounts at all and only knows phynames. still it knows about volume-labels20:11
DocScrutinizeruse blkid -g and/or blkid -c /dev/null to check if kernel has noticed your label renaming20:12
DocScrutinizerif it hasn't, you'll need partprobe, I guess20:13
DocScrutinizerif it knows new label (in blkid), then a remount might suffice20:13
freemangordonblkid knows about label change20:14
freemangordonwithout any arguments20:14
DocScrutinizerdon't use blkid without arguments, it's bogus output20:14
freemangordonit is ok20:14
DocScrutinizerno it's not20:15
freemangordonit shows the correct label for FAT partition on sd card20:15
DocScrutinizersee 24h backscroll20:15
freemangordonyeah20:15
freemangordoni know that it caches volumes20:15
freemangordonbut it seems smart enough to spit the correct volume labels20:15
DocScrutinizermere incidence20:16
freemangordoni.e. not the cached ones20:16
freemangordonwhatever20:16
*** int_ua has quit IRC20:16
freemangordonso, kernel knows about changed volume label, it seems there is something wrong with th remount20:16
freemangordonlet me try to remount by hand20:16
DocScrutinizeryou're free to check your blkid cache and see when and how it changes20:16
DocScrutinizeror you simply trust in blkid -g20:17
*** psycho_oreos has quit IRC20:17
freemangordonDocScrutinizer, i've changed volume label just before running blkid, and it showed me the new one, that is pretty enough for me20:17
DocScrutinizerI dunno if a remount is even possible when file handles are open to the volume20:17
freemangordonnothing is open20:18
freemangordoni'll do unmount/mount20:18
freemangordonnot mount - remount20:18
freemangordonnot mount -o remount20:18
DocScrutinizerfreemangordon: I'm getting bored by your argumentation of "what I see is what I expected, so I don't bother if what I see is correct"20:18
freemangordonDocScrutinizer, not helping much. And TBH i was expecting blkid to give me the cached labal.20:19
freemangordonhmm, hal shows the volume even if unmounted, with the old volume label, after all it seems hal_set_property is unavoidable :(20:22
DocScrutinizerthat's why umount is done via ke-recv20:23
freemangordonyou mean?20:23
freemangordondidn't understand why?20:23
DocScrutinizerthink about removing battery lid20:24
DocScrutinizerwhat's happening in system?20:24
freemangordonforced unmount of sd card20:25
freemangordonAFAIK20:25
DocScrutinizerwhy doesn't hal show uSD after opening bat lid?20:25
*** Pali has quit IRC20:27
freemangordoniirc there was a hck in kernel nadling the situation by unloading the device20:28
freemangordonbut it is different case here20:29
DocScrutinizeryou're sure about it's any different?20:29
freemangordonI think so, we are only re-mountn a volume, but mmc device does not change in any way20:30
freemangordonhowever, I will try to do hal-set-property and will report if it helps20:30
DocScrutinizerit for sure helps20:31
DocScrutinizerthe question is, will it fix the root problem?20:32
freemangordonno20:34
ShadowJKcould someone  summarize the problem?20:34
freemangordonI am not sure if there is such thing like hal-volume-monitor, but if there is, the problem might be there20:34
freemangordonShadowJK, if you rename a volume in filemanager, the new volume name to show you need to reboot the device20:35
ShadowJKdid this work in fremantle?20:35
freemangordonwe are talking about fremantle20:35
DocScrutinizerShadowJK meant did it work on stock fremantle20:36
DocScrutinizerand tbh I doubt it works on my openSuse11 PC20:37
DocScrutinizerShadowJK: the question is: when is who reading the volume labels?20:44
DocScrutinizerwhat needs to be done to retrigger this procedure?20:44
DocScrutinizerI gather partprobe is doing exactly that20:45
* ShadowJK wonders if maemo 1.0 was begun before blkid20:45
DocScrutinizera derived probelm is: when and why is hal reading the volume labels, and from where?20:46
freemangordon:nod:20:46
ShadowJKin a normal linux distro, from before hal was killed, would be getting info about new drives and cards from hal20:47
ShadowJKand ask hal to mount/umount20:47
DocScrutinizerof course you can fix the hal problem by "hacking" hal, by sending a explicit update for a particular label, you even can make hal think the label is "mickeymouse"20:47
DocScrutinizerindeed. On maemo though I guess ke-.recv is doing the (u)mount20:48
DocScrutinizer[2012-04-11 19:23:25] <DocScrutinizer> that's why umount is done via ke-recv20:49
DocScrutinizersee ke-recv-test20:49
DocScrutinizerob - open battery cover signal /  f - format device <arg> /  r - rename device <arg> /  (e - check device <arg>)? /  m - enable USB mass storage20:51
*** dafox has joined #maemo-ssu21:03
*** dafox is now known as Guest6003221:03
*** andre__ has quit IRC21:06
*** Guest60032 has quit IRC21:08
*** Pali has joined #maemo-ssu21:08
*** trbs has joined #maemo-ssu21:08
*** dafox__ has joined #maemo-ssu21:15
*** trbs has quit IRC21:16
*** dafox__ is now known as dafox21:17
*** trbs has joined #maemo-ssu21:17
freemangordonDocScrutinizer,Pali: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=51000121:31
freemangordonseems after all we should inform hal from ke-recv :)21:31
*** dafox has quit IRC21:32
Paliyes, I wrote that hal did not updating label21:34
Paliand seems that method Rescan sounds like good solution21:35
DocScrutinizerI'd guess quite a number of tools already call rescan()21:37
DocScrutinizerparticularly partprobe seems worth looking at it, also ke-recv21:38
*** dafox has joined #maemo-ssu21:40
freemangordonke-recv does not call libhal_device_rescan21:40
DocScrutinizerrather unfortunate21:42
freemangordonwe will make it :)21:42
DocScrutinizerIt seems to me we should include `hal-disable-polling' at beginning, and `hal-disable-polling  --enable-polling' at end of /usr/sbin/mmc-rename.sh21:48
DocScrutinizermight do exactly what we need21:48
freemangordonwill check what exactly hal-disable-polling does21:50
freemangordonhowever we don't need the whole hal database re-scaned, so a programatic call to libhal_device_rescan would suit better21:51
DocScrutinizerman hal-disable-polling - has some nice general discussion about (removable) storage devices21:55
DocScrutinizerprobably even the opposite approach is more correct: enable polling before unmounting, then unmount, rename, mount, disable polling21:56
DocScrutinizerwe don't need constant polling on non-removable devices21:57
freemangordonwell, might work, but my concern is that it would be a little bit less effective re battery consumtion21:57
DocScrutinizerof course an explicit hal-set-property from within that /usr/sbin/mmc-rename.sh might do as well21:57
freemangordonthan a single call to rescan21:57
DocScrutinizerfreemangordon: how's battery affected by a temporary action?21:58
freemangordonDocScrutinizer, no need this to be done through script, ke-recv already talks to hal21:58
freemangordonby a little bit :)21:58
freemangordonbut still21:58
freemangordonanyway, I think we have the reason and the solution, thanks21:59
DocScrutinizerI'd agree leaving polling enabled is something that could impact battery massively21:59
DocScrutinizerbtw this friggin /usr/sbin/mmc-rename.sh is (c)2007 :-S22:00
Palisee git log22:01
freemangordon:D nice22:01
DocScrutinizerseems like it could use a major rewrite22:01
Palisometimes nokia did not extended copyright22:01
DocScrutinizerPali: they'd have if they had changed *anything* in this fsckdup code22:01
DocScrutinizerwhich btw is the explanation why it is using ancient crap like L=`eval grep '\"$DEV\"' /etc/mtools.conf | awk '{print $2}' | sed 's/://'`22:02
DocScrutinizerShadowJK's suspicion it might date from before blkid got invented seems rather correct22:03
DocScrutinizerecho "$0: could not determine drive letter"22:05
DocScrutinizerOMG22:05
DocScrutinizer[2012-04-11 18:46:28] <Pali> I think that script should be changed to use dosfslabel and e2label22:09
DocScrutinizer^^^me agrees22:09
DocScrutinizerbwahaha look into /etc/mtools.conf22:14
*** Pali has quit IRC22:22
*** NIN101 has quit IRC22:34
*** javispedro has joined #maemo-ssu22:45
DocScrutinizerFSTYPE=$(echo $(blkid -c /dev/null -s TYPE $1|sed 's@.*TYPE="\(.*\)"@\1@')); case $FSTYPE in vfat) echo "do dosfslabel $2 here";; ext2|ext3) echo "do e2label $2 here";; *) echo "unknown filesystem:>$FSTYPE<";; esac;22:49
DocScrutinizer^^^ use that as template for a new better /usr/sbin/mmc-rename.sh22:50
DocScrutinizerFSTYPE=$(blkid -c /dev/null -s TYPE $1|sed 's@.*TYPE="\(.*\)" @\1@'); case $FSTYPE in vfat) echo "do dosfslabel $2 here";; ext2|ext3) echo "do e2label $2 here";; *) echo "unknown filesystem:>$FSTYPE<";; esac;22:56
DocScrutinizereven22:56
DocScrutinizermaybe even22:57
DocScrutinizerFSTYPE=$(blkid -c /dev/null -s TYPE $1|sed 's@.*TYPE="\(.*\)" *@\1@'); case $FSTYPE in vfat) echo "do dosfslabel $2 here";; ext2|ext3) echo "do e2label $2 here";; *) echo "unknown filesystem:>$FSTYPE<";; esac;22:57
DocScrutinizercheck $1 is non-empty; add proper return code after echo "unknown..."22:59
DocScrutinizero/23:03
* DocScrutinizer ->dinner23:03
*** Pali has joined #maemo-ssu23:11
*** s1gk1ll has joined #maemo-ssu23:19
*** sigkill_ has quit IRC23:20
DocScrutinizerhttp://maemo.cloud-7.de/maemo5/usr/sbin/mmc-rename.sh23:31
*** nox- has joined #maemo-ssu23:31
DocScrutinizermight still need a bit of polishing23:31
freemangordonthanks, i am just going to test the Rescan() thing23:32
*** dafox has quit IRC23:38
PaliDocScrutinizer, why to use blkid for detecting fstype? We can pass $3 from ke-recv23:50
Palialso e2label working fine for ext4 too23:50
freemangordonPali, there is a critical bug in ke-recv, if you try to format a prtition on external sd card through filemanager, it will always format the first partition :(23:52
freemangordonthe same with rename23:53
freemangordonthe udi passed to hendler functions is NULL23:53
freemangordon*handles23:53
freemangordondamn23:53
freemangordonhandler23:53
Paliuff, ok23:53
Paliat least we know where is problem23:53
freemangordonhttps://gitorious.org/community-ssu/ke-recv/blobs/master/src/ke-recv.c#line43923:54
freemangordonhttps://gitorious.org/community-ssu/ke-recv/blobs/master/src/ke-recv.c#line37723:54
freemangordonso volume_get_num and respectively get_nth_volume will always return "preffered" volume, whichever it is23:56
freemangordonPali,  know if there exists a helper function to ge udi from device id?23:57
freemangordon*get23:57
Palifreemangordon, I do not know now - I'm going to sleep...23:58
PaliI will look at it later - tomorrow23:59
freemangordon:)23:59
freemangordonok, I will fix it23:59
« Tuesday, 2012-04-10 Index Thursday, 2012-04-12 »

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