IRC Logs for #crux Saturday, 2011-04-30

*** Zaba has joined #crux00:01
*** jajebiegeniusz has quit IRC00:30
*** jajebiegeniusz has joined #crux00:31
*** SiFuh has quit IRC00:55
*** SiFuh has joined #crux00:56
*** Zaba has quit IRC01:20
*** Zaba has joined #crux01:26
*** DaViruz has quit IRC01:31
*** DaViruz has joined #crux01:32
*** Zaba has quit IRC02:29
*** jue has quit IRC02:44
*** jue has joined #crux02:46
*** ChanServ sets mode: +o jue02:46
cruxbot[opt.git/2.7]: gv: update to 3.7.202:55
cruxbot[opt.git/2.7]: libevent: update to 2.0.1102:55
cruxbot[opt.git/2.7]: msmtp: update to 1.4.2402:55
cruxbot[core.git/2.7]: libcap: update to 2.2102:56
*** mike_k has joined #crux03:26
jueplease test -> httpup sync http://jue.li/crux/ports/#gcc gcc03:47
juethat's gcc 4.5.3 ;)03:48
frinnstjust a versionbump?04:17
frinnstim running 4.6 on my desktop :)04:17
Romstersame 4.6.0 here04:40
jueyes, and the usual footprint update04:42
Romsterwonder if 4.5.3 fixes gtk optimization bug on -O204:42
Romsterfor dragging widgets ie tabs in firefox04:43
juebbl04:44
frinnstRomster: what bug is that? havent noticed04:53
Romstergcc 4.5.2 compiled gtk ok but if you drag a tab in firefox it would segfault04:54
Romsterrecompiling gtk with -O0 solved that04:54
Romsteri can use -O2 on gcc 4.6.0 ok04:55
Romsterso possibly 4.5.3 might of fixed this04:55
tilmaninteresting, lennart reported that same problem04:56
Romsteryeah i had that issue for ages, until recently04:57
Romsteri didn't dig further to find out which optimization flag -O2 enabled that broke it.04:58
Romsterit started happening past some gtk version05:01
Romsterwhich i can't remember05:01
slashbeastanyone of you have ever issue with libtool misdetecting proper CONFIG_SHELL/SHELL?05:30
slashbeastacl-2.2.51 generate libtool with bashism and then it running it with /bin/sh, which failing hard.05:30
Romsteri haven't changed the sh shell from bash yet, highly likely that will break lots of things including what you found, patch it and submit the patch05:33
slashbeastthis is my gentoo system and I have it for a long time, dash (real posix sh).05:38
slashbeastRomster: the solution is CONFIG_SHELL=/bin/sh ./configure05:39
slashbeastRomster: as long as people does not call /bin/sh with bashism, it is all right05:39
Romsteri know that...05:39
*** divadgnol67 has joined #crux05:39
Romsterso submit a but report to the maintainer of acl05:40
slashbeastI don't know is it is acl issue05:40
slashbeastor libtool instead.05:40
slashbeastbut acl 2.2.49 was all right.05:41
slashbeastwhen I will be sure I will report it to ustream05:41
slashbeastfor now I reported it to gentoo05:42
slashbeastmaybe someone will find a real solution.05:42
divadgnol67romster: when i run Xorg -config a new xorgconfig file is created. but it is unuasable.05:44
divadgnol67Is there not a way to choose my video card from a list like xorgconfig used to allow?05:44
slashbeastcurrent xorg should be fine without xorg.conf.05:44
slashbeastand you can create /etc/X11/xorg.conf.d/ and put few files there with keyboard config for example.05:45
divadgnol67i get no usable screen found or something like that05:45
slashbeastsupport for xorg.conf.d was added in 1.8 or 1.9 version of xorg-server05:45
slashbeastdivadgnol67: missing video driver?05:45
slashbeastwhich card you have?05:45
Romsterif it's nvidia there is a tool to configure it for you.05:46
slashbeastnvidia-xconifg, yes05:46
divadgnol67yeah, I know. But when I look in ports it's there05:46
Romsterbut it's trivial to do it manually05:46
slashbeastdivadgnol67: http://paste.pocoo.org/show/IX9Lh8AiWLAKqU9F2fmT/ heres example05:46
slashbeast( for conf in /etc/X11/xorg.conf.d/*.conf; do echo -e "\n\n$conf:" && cat $conf; done ) | lodgeit -p05:46
Romsteryou also need to do gl-select use nvidia05:46
divadgnol67I have a built in intel video card (toshiba laptop)05:46
slashbeastintel/i810 driver05:47
slashbeastxf86-video-intel propably05:47
Romsterok so install the intel driver and change the driver name in the xorg.conf05:47
divadgnol67i thinks is i91505:47
slashbeastdivadgnol67: well05:47
divadgnol67thanks05:47
slashbeastalso, you need KMS05:47
slashbeastcurrent intel drivers need kms.05:47
slashbeasti915 modeset it was iirc05:47
slashbeastin kernel05:47
divadgnol67i have been using openbsd way to long. life has become so easy05:48
Romsterok i'm off to run a pa rig in a pub. i'll be back far later.05:48
divadgnol67kms is a module? or can i compile it into the kernel. sorry for the ignorance05:49
divadgnol67romster - your were on here when i went to bed at 11 last night. you gotta be tired. be carefull05:50
slashbeastdivadgnol67: kms is a 'thing' from kernel05:50
slashbeastin short, after enable it, while booting you will got framebuffer with native res05:50
slashbeastwithout lag like vesafb used to do05:51
divadgnol67slashbeast > dude you rock! thanks for the tip05:51
slashbeastalso siwtching between real console (vt) and xorg will be painless.05:51
slashbeastthis is a driver where current intel driver write output05:51
divadgnol67so let me sumarize05:51
slashbeastyou need KMS-enabled kernel for current xorg-server and intel driver to work.05:52
divadgnol67add kms support in the kernel, add correct driver and then  edit xorg to include intel driver05:52
slashbeastand then do startx05:52
slashbeastwithout xorg.conf05:52
slashbeast21 sentry, you do not need it.05:52
divadgnol67hell yeah!05:52
slashbeastcheck paste above, you can have small files with a part of xorg.conf, like config keyboard05:52
slashbeasthttp://paste.pocoo.org/show/IX9Lh8AiWLAKqU9F2fmT/ the paste05:53
divadgnol67i'm old school unix. 43 years old05:53
slashbeastno idea if xorg-server in crux still use hal or starts using udev.05:53
slashbeasthal is deprecented, the 'new school' use udev.05:53
tilmanudev05:53
slashbeastoh, nice.05:53
slashbeastdivadgnol67: also new xorg is able to start without inputs, try do 'sleep 30; killall X' on another console before startx05:53
tilmancrux never really loved hal much ;)05:53
slashbeastor you will have issue.05:54
divadgnol67thanks again. I'm writing this stuff down.05:54
tilmandivadgnol67: there's no need to specify your video driver in xorg.conf usually05:54
slashbeastnotes are good.05:54
tilmanas slashbeast said, if you're lucky you don't need xorg.conf at all05:55
slashbeastI use it as I need config keyboard with polish layout and synaptics touchpad05:56
slashbeastxorg.conf.d^05:56
tilmani use it to fix my dual-head stuff ;)05:56
slashbeastI think you can do it with xrandr. ;-)05:56
tilmansticking xrandr calls in ~/.xsession is kind of lame though05:56
tilman:]05:57
slashbeastI used to have xrandr --auto in .xinitrc05:57
slashbeastlife saver when my LCD died05:57
slashbeastI could connect anything and I had video.05:57
divadgnol67thanks for the link. i was able to gleen some additional insight05:59
divadgnol67later guys, thanks again.06:05
*** divadgnol67 has quit IRC06:05
*** clare_ has joined #crux06:45
clare_Hi all, wondering if anyone has a way to check on signal strength of mobile broadband accessed via wvdial06:47
frinnstdoubt it, but not sure06:48
deus_exclare_: http://tinyurl.com/3odxkmo07:08
deus_exyou'll most likely need to adapt it for your card, though07:08
deus_exor try umtsmon http://umtsmon.sourceforge.net/07:11
*** lasso has joined #crux07:11
*** DaViruz has quit IRC07:36
clare_deus_ex: thank you, I am looking at both those - will take a while I suspect.07:46
*** DaViruz has joined #crux07:49
*** deus_ex is now known as pedja07:50
frinnsthttp://0pointer.de/blog/projects/why.html07:52
tilmanhttp://lwn.net/Articles/440853/08:34
tilmanowned!08:34
tilman(mezcalero is poettering)08:34
tilmanhttp://lwn.net/Articles/440766/ hahahahhaa08:35
tilmanhttp://i.imgur.com/usftZ.png08:37
tilmanjue: running your updated gcc port now btw. will report back on monday08:38
thrice`I kinda like systemd.  I wonder why people are so pissy towards people free work08:40
tilmanmaybe poettering's advertising plays a role there08:42
thrice`yeah, probably08:43
*** divadgnol67 has joined #crux09:06
divadgnol67slashbeast, you still on?09:11
divadgnol67here is my most current error (EE) intel (0): no kernel modesetting driver detected09:15
divadgnol67also failed to load fbdev09:15
slashbeastdivadgnol67: no modesetting driver - you need enable i915.modeset in kernel09:19
slashbeastdevice drivers > graphics >  somewhere there09:20
*** clare_ has quit IRC09:20
divadgnol67dang it! This is like the 4th time I've had to recompile.09:20
divadgnol67Great learning experience though09:21
slashbeastrecompile, not really09:21
slashbeastit will compile only changes and link image09:21
divadgnol67that explains why its faser the second and third time around09:22
*** treach has joined #crux09:22
divadgnol67it been years since I ran Gentoo so having to compile my own kernel seems foreign09:23
slashbeastif you change for example config_hz it will recompile everything as it have to09:23
divadgnol67I'll be back......09:23
divadgnol67later09:23
*** divadgnol67 has quit IRC09:24
*** root_ has joined #crux09:48
root_slashbeast, there is not an option to add i915 into the kernel.09:48
root_i kwow is something simple and i'm just missing it09:49
juetilman: great, don't think that we have to expect problems, but who knows ;)10:26
jueI'd say that most people don't like the idea that poettering, who has a not so lucky history with avahi or pulseaudio IMO, is now tinkering in a much more important area10:33
*** root_ has quit IRC10:35
frinnsthe's the new schilling10:45
* teK_ *wont* mention his love and affection for Ulrich Drepper :>11:05
*** mrks has quit IRC11:08
*** mrks has joined #crux11:10
*** mike_k has quit IRC11:52
*** fredo696 has joined #crux13:31
fredo696hi all!13:31
fredo696I'm encountering a problem, I just updated my Crux 2.7 with prt-get sysup13:37
fredo696and now when booting, fsck systematically enters maintenance mode13:38
fredo696even though my / is clean13:38
fredo696can anybody help?13:39
*** DaViruz has quit IRC13:50
*** DaViruz has joined #crux13:51
teK_does anyone know why {e,f}grep aren't symlinks to grep(1) the man page says running egrep is the same as running grep -E anyways13:53
tilmanfricking newly fandangled hdds that use 4k sectors14:14
tilmanand don't tell the kernel about it14:14
entemeh, 4k sectors14:15
enteI have no clue how to handle these14:15
tilmanhttp://stream-recorder.com/forum/partition-samsung-hd204ui-ubuntu-linux-hdd-w-t8032.html14:15
tilmanfirst i used cfdisk to partition this drive14:15
tilmanwhich caused random "failed to write blabla" errors14:15
enteespecially not on a raidcontroller which has had the last firmware released in 200814:15
tilmangmane suggests this might be due to misalignment14:15
tilmanand in fact, cfdisk did not put the partition on a 4k alignment14:16
tilmanffs14:16
tilmanand fdisk won't let me put the start of the partition in he first 2047 sectors14:16
tilmanit claims 2048 is the minimum14:16
tilmangrrrrr14:16
entesfdisk?14:19
entegdisk?14:19
tilmannever used these two14:19
enteI think gdisk is for GPT partitions on BSD14:19
entewill have to use it soon, because the disk is >2T14:20
enteso I got a nice integer overflow with the mbr partition scheme14:20
entesuddenly, the size of the partition was negative :D14:20
tilmanwell, i lose 8 MB if i just put it at sector 204814:20
tilmanwho cares really14:20
enteyeah, not about 8M14:20
tilmanwhy not?14:21
tilman2048 * 409614:21
tilmanbytes14:21
tilmanoh14:21
tilmangot you wrong14:21
tilman;)14:21
ente:)14:21
ente[21:20] < ente> will have to use it soon, because the disk is >2T14:22
ente3 2TB disks in a RAID 514:22
tilmanhehe, i just made fdisk overflow an int, too14:24
tilmanlast sector? (2048-3907029167)14:24
tilman> 390702918414:24
tilmani screwed up my math and gave it a value that's outside the range14:24
tilmanand it still accepted it14:24
tilmano_O14:24
entethat's why I use cfdisk14:25
tilmanusually i use cfdisk, too14:25
tilmanbut it seems it will _always_ place the first partition at sector 6314:25
tilmanwhich is no good for these 4k-sector-drives14:25
slashbeastthere is gptdisk for gpt parititons iirc14:25
ente       -b sectorsize14:26
ente              Specify  the sector size of the disk. Valid values are 512, 1024, 2048 or 4096.  (Recent kernels know the sector size. Use this only on old kernels or14:26
ente              to override the kernel's ideas.) Since util-linux-ng 2.17 fdisk differentiates between logical and physical sector size. This option changes both sec‐14:26
ente              tor sizes to sectorsize.14:26
enteaccording to the manpage14:26
entetried that?14:26
tilmanyes14:26
entemeh14:26
tilmansomething's really screwed up14:28
tilmanfdisk doesn't seem to correctly read my answer for 'last sector' at all14:29
InDigo176Try using parted. It can tell you if an partition is properly aligned too.14:29
tilmanif i just hit enter to accept the default, it works okayish14:29
tilmanthanks14:29
tilmani'll give it a try14:29
tilmanin frinnst we trust14:30
tilmanInDigo176: i'm afraid parted might screw up, too, though. linux reports that the disk uses 512 byte sectors :)14:30
tilman(in /proc/yadayada, and in dmesg:14:31
tilmansd 3:0:0:0: [sdd] 3907029168 512-byte logical blocks: (2.00 TB/1.81 TiB)14:31
Romsterdivadgnol67> romster - your were on here when i went to bed at 11 last night. you gotta be tired. be carefull <- heh i[m on alot and sleep and work, now i'm a little drunk and gonna sleep then pack up pa gear.14:33
Romsteri find that util-linux 2.19 works ok but 2.18 does not align correctly.14:34
tilmanmight depend on drive, too14:35
tilmanand what sector size the kernel figures it uses14:35
Romsterprobably reads drive info.14:35
InDigo176tilman: yeah. My 4096-byte sector disk also reports 512 bytes (Western Digital Green). I've discovered that parted tests for 4096 byte alignment (that's called "minimal" in parted) and for 1 MB alignment (that's called "optimal" in parted). Dunno what parted gets 4 KB and 1 MB from14:44
InDigo176Anyway, I put my partition start at sector 8 and it works great.14:45
tilmanfdisk uses atoi() to read stuff into an unsigned long long14:45
tilmanF A I L14:45
tilmanoh14:46
tilmanit's not even an ull, it's an unsigned int14:46
tilmanfaiaiaiaiaiaiiaiaiaiaiail14:46
tilmanand then it goes with MAX_INT (2^31)14:47
tilmanwell, 2^31 - 114:48
entethat's why you need GPT, I guess14:49
tilmanno14:49
tilmans/atoi/strtoul/ will do the trick for me (sizeof(long) == sizeof(int))14:49
entesomething other than fdisk? :P14:49
tilmanso yeah, i found a bug in fdisk14:49
enteah, or that14:49
tilmandidn't think i would14:49
tilman:O14:49
tilmankind of scary14:49
tilmani bet it's fixed in git ;)14:50
tilmanInDigo176: the 1MB thing is what vista uses as default15:04
tilmanInDigo176: and it explains why fdisk wants to use sector 204815:04
tilman2048 * 512 = 1MB15:04
tilman:)15:05
slashbeasttilman: are you talking about 1MB align?15:14
tilmani think so15:15
slashbeastfdisk -S 32 -H 32 did the trick for me15:15
slashbeastthe usb flash memory love it, 2x better performance.15:15
InDigo176tilman: ah15:20
tilmanslashbeast: yeah, i expect a speed-up out of this re-alignment too15:20
tilman:)15:20
slashbeastmisaligned flash memory will die faster15:21
tilmanyeah15:21
slashbeastI found this 'solution' when I was about to buy ssd disk15:21
tilmanslashbeast: i played with -S and -H when i partitioned my SSD15:21
slashbeastwhen I know everything I need I decide to stay with regular hdd15:21
tilman:)15:21
slashbeastthe trim support, and dmcrypt which does not support...15:21
slashbeastas I used to have sda1 boot and sda2 encrypted lvm's pv.15:22
slashbeastand then rootfs over this lvm15:22
* slashbeast is paranoid.15:22
slashbeasttilman: are you happy with ssd?15:23
tilmanyes15:23
tilmani'm not so happy with the btrfs i put on it15:23
slashbeastThe usecase I see for it is my large git repos, where the first status, diff etc take about 10 minutes.15:23
tilmanyeah15:23
slashbeastbtrfs?15:23
tilmanseek time... nonexistent15:23
tilman:D15:23
tilman<315:23
slashbeastThe file system without working fsck. ;-)15:23
tilmanyes15:24
tilmanstay away from btrfs until there is a fsck that can find and repair errors!15:24
slashbeastit become slower and slower, and defrag is useless.15:24
slashbeastI follow btrfs for some time already. The only feature i really like is snapshots and subvols.15:24
slashbeastbut I think before btrfs will have fsck the native zfs will be stable.15:24
tilmanhaha15:25
slashbeastand zfs offer deduplication, sweet.15:25
slashbeastI heard about a gentoo user who use zfs on rootfs.15:25
slashbeastof course you need do a initramfs to support it as zfs is external module, because of CDDL license, argh.15:26
jaegerso there's native linux zfs now?15:26
slashbeastYes sir!15:26
slashbeasthttp://zfsonlinux.org/15:26
treach"you want to remove this 1.5 GB file to get some more space?" "Wow, cool, you've recovered a stunning 1.5 MB!" <- effect of data deduplication :p15:27
slashbeastWell I think dedup is a 'bit' more effective than you said15:27
slashbeast;-)15:27
treach*I* think you completely misunderstood the point. :>15:28
slashbeastgimme a sec to think about it.15:28
slashbeastah, yeah, shame on me.15:28
slashbeastyou never know how much space will be really freed after removing a duplicated file.15:28
treach"need additiona pylons"15:29
treach*additional*15:29
tilmanhaha15:29
*** fredo696 has quit IRC16:20
*** y3llow has quit IRC16:34
*** y3llow has joined #crux16:34
*** treach has quit IRC16:48
*** divadgnol67 has joined #crux18:25
*** divadgnol67 has quit IRC19:57
*** Dudde has quit IRC21:31
*** Dudde has joined #crux21:32
*** lasso has quit IRC21:58

Generated by irclog2html.py 2.11.0 by Marius Gedminas - find it at mg.pov.lt!