IRC Logs for #crux-devel Tuesday, 2013-08-13

*** acrux_ has joined #crux-devel00:35
*** rmull has joined #crux-devel00:47
*** acrux_ has quit IRC01:20
*** acrux has joined #crux-devel01:23
*** __mavrick61 has quit IRC02:43
*** ___mavrick61 has joined #crux-devel02:44
*** deus_ex has quit IRC04:08
frinnstoh, gcc 2.1806:10
*** Workster has quit IRC07:01
frinnst=======> Package '/usr/ports/pkg/glibc#2.18-1.pkg.tar.xz' is up to date.07:44
frinnstpt_chown was removed because of security issues. can be reenabled if you really need it, do we?07:49
juegood morning08:59
juefrinnst: according to http://linux.die.net/man/5/pt_chown we don't need it08:59
juefrinnst: btw, the curl update seems to be save to me09:01
frinnstyeah i also run it, but i probably havent tested it much09:26
frinnstim rebuilding alles with gcc 4.8.1 (or whatever the pointrelease is) and glibc 2.18 ;)09:26
teK_it always puzzles me why the whole world builds german words in their language/use .. I don't know a single word of either swedish dansk or finnish09:29
*** sepen has joined #crux-devel10:29
frinnstchown("/dev/pts/4", 12352, 5)           = -1 EPERM (Operation not permitted)11:09
frinnsthopefully things will settle down when everything is rebuilt11:09
*** deus_ex has joined #crux-devel11:10
RomsterFika11:14
Romstercoffee and cake in swedish11:14
teK_I know irish coffee.. but fika? never heard of that11:15
Romsterhanging around a few Swedes in another channel i learned a couple of words.11:20
Romsteror i did, i can't remember how to spell them now11:24
Romsterhttp://www.slayradio.org/mastering_swedish.php funny as hell.11:27
frinnstfika is the social thing you do when you have tea or coffee11:33
frinnstlike dinner or something11:33
Romsteryeah pretty much11:33
teK_alright11:37
teK_I always admired 'northern' culture and landscape. Maybe next year I will visit sweden! :>11:37
Romsteri would if i could afford too.11:38
frinnstyes we are awesome12:21
frinnstbut you get a taste of me here everyday, ARE YOU NOT ENTERTAINED?!12:21
frinnstwhy do i still get "urxvt: can't initialize pseudo-tty, aborting." xterm seems to work just ifne12:23
frinnstfine12:23
frinnsti have a hard time figuring urxvt still needs pt_chown12:23
frinnstah, we mount /dev/pts wrong12:38
frinnstor, i do, atleast12:38
frinnstneeds to be mounted with gid=tty or glibc will do wierd things12:43
Romsterfrinnst, does weird things.12:49
frinnstyou know the saying regarding glass houses? :D12:51
frinnsttbh i probably usually spell it "wierd"12:52
frinnstanybody have any strong opinions on the 3.1 toolchain, versions etc?14:03
Romsternot looked into it.14:09
sepenls -al14:29
sepenpfff, really the wrong win ;D14:30
frinnst:D14:41
jaegergah, this new isohybrid bullshit is going to make me insane14:44
frinnstheh, trouble with uefi or what is it?14:45
jaegerNo UEFI involved here, the boot just fucks up with the dual partition thing for some reason14:46
jaegerThe init script tries to mount both /dev/sdd and /dev/sdd1 in this case and both are mountable and *some* file operations work on /dev/sdd but enough of them fail that you can't install properly from it14:47
jaegerI could change the init script not to use full devices, only partitions, but there's probably some edge case where a user would want to be able to do that14:48
jaegermaybe not14:48
frinnstis this a new problem due to some change?14:49
jaegerIt's definitely new but I can't pinpoint what's causing it... perhaps the kernel itself14:49
jaegerI don't recall having this problem with 3.6.11 and the 3.0 release14:51
jaegerIt could also be fixed by generating a USB image via script or manually instead of using isohybrid14:53
frinnstiirc sepen had a script for that14:55
jaegeryes14:55
jaegerIt's a simple thing to do, the appeal of isohybrid was that we didn't have to maintain two images for CD and USB14:55
frinnstsepens script changed the iso, right?14:57
frinnstwe could just go back to a simple iso and make the tool/script available14:57
jaegerI don't believe it did change the ISO, I think he mounted it and copied files from it14:58
jaegerI suppose I should ask someone else to test this issue, maybe my USB drives or computers suck :P14:58
jaegerI'll upload it if you or someone else here has time to test it via USB specifically14:59
jaegerHrmm... looking back in my logs it seems like I already went down the road of altering the init script not to use full devices and the problem still persisted... maybe it's kernel-related after all15:00
frinnstyeah feel free to upload it somewhere15:03
jaegerIt'll take a while, got a shit upstream today for some reason15:03
jaegerIt's going, though15:03
frinnstyay, the rebuild is on xorg, almost finished \o/15:04
jaegernice15:12
jaegerThis image boots 3.9.11, once it's uploaded I'll make one with 3.10.x and test that15:15
jaegerso many kernel releases these days... makes me wonder if we should use a longterm one for the installation media so it works fine with everything like vmware, etc.15:15
teK_this would only make sense if we'd plan to keep updating the kernel on the installation medium. IMHO15:29
jaegerHow so?15:38
teK_LTS means that it will receive updates which is  pointless for a 'static' ISO file15:38
teK_we want a working (at the moment of production) kernel, everything else wrt the future is pure speculation15:38
jaegerI suppose for installation media I don't see a lot of difference between the two15:41
teK_yeah15:41
jaegerfrinnst: ok, upload is done, http://jaeger.morpheus.net/linux/crux/tmp/15:41
jaegerI have run into issues with vmware tools or similar on brand new kernels but I can work around that on my own if it's a problem15:42
teK_mind me testing in qemu?15:42
frinnsti think i'd prefer a lts kernel on our media. it sort of gives one a stable foundation :)15:43
jaegerteK_: can qemu boot as a usb device would? the cd media works fine15:43
jaegerfrinnst: that's kinda what I'm thinking but I'm not dead set on it15:44
jaegeranother option would be to change the Makefile to allow for an array of kernel versions and the boot media would then have multiple kernel options15:44
frinnsteeew15:44
frinnst:)15:44
teK_how is LTS more stable? It is _long_ _time_ _supported_15:44
jaegerwell, the idea is that brand new kernels are not always stable, not that the LTS is *more* stable15:45
frinnstyeah, but say you want to run crux with a lts kernel you *know* it will work (unless we really suck)15:45
teK_the only advantage could be that you probably do not have to adjust the config _iff_ you were about to update the ISO15:45
teK_which we don't, right :P15:45
jaegerI spend lots of time tweaking the config but I imagine I'm the only one, really15:46
jaegerany time I test new ISO features or whatever15:46
frinnstif we'd ship 3.12 or whatever, and the next lts kernel is 3.20 the user has to either use an older kernel (incompatabilites? glibc kernel-flag? or whatever) or wait for a few years15:46
jaegertime spent changing configs is irrelevant to me, the appeal for me would be if, for example, 3.4.57 works perfectly with vmware/virtualbox/hyper-v tools but 3.10.10 doesn't, that sort of thing15:46
jaegerAlso, no user is forced to use the kernel we provide on the installation media15:47
frinnsti doubt anybody does :)15:47
jaegerIt was just a thought, may not be useful or needed15:49
teK_it can15:54
teK_rather slow but it's booting15:54
jaegerI'll admit a part of it is this weird USB crap15:54
teK_got a prompt15:58
teK_installing.15:58
teK_:)15:58
*** sepen has quit IRC16:01
jaegerif the install proceeds then you aren't hitting the problem... when it happens, /media isn't mounted16:04
jaegerI'm guessing that qemu boots it in cdrom mode, though, is that correct?16:04
teK_I dont think so, I run:16:06
teK_% qemu-system-x86_64 -m 1024 -usbdevice  disk:/tmp/crux-3.0-updated.iso -hda update.raw -boot menu=on16:06
jaegerhrmm, ok. what device did it mount when booted?16:11
teK_ /dev/sdb16:18
teK_type iso 966016:18
jaegerweird. thanks16:18
teK_np16:18
teK_I guess that kernel config / installation will  work so I'll stop testing16:21
jaegerwish I could figure out why it doesn't work for me16:22
jaegerwith multiple different USB drives and even different computers16:22
teK_no idea on that, sorry16:25
pitillodo you need to verify something jaeger?16:45
pitilloboot it from an usb stick on my desktop16:45
*** sepen has joined #crux-devel16:46
pitilloI made the stick with sepen's script16:47
jaegerit boots fine for me but doesn't mount the media properly to install17:32
*** _tim has joined #crux-devel17:40
pitillo /dev/sdb mounted on /media -> media shows the files17:45
jaegeralright, thanks. I guess it's just all my hardware, for some reason :(17:55
pitillousb 3.0 booting from usb (there is another option I've never tried, uefi)18:03
jaegerRomster: can you check if usr/bin/pamx is in your netpbm footprint when built? I get that extra file on all my machines18:04
jaegerpitillo: this image doesn't have UEFI boot support anyway, though I've added that in the past18:04
jaegerUEFI and isohybrid are NOT compatible18:04
pitilloI use always sepen's script to build usb sticks18:05
pitillothis is a 16gb new one... removed its only partition and using the script to use the entire driver18:05
jaegerI may go back to that... isohybrid is such a hack18:05
pitillos/using/used18:05
jaegerAh, you didn't use dd to write it?18:06
jaegerWould you mind testing the raw image for me? That's where the problem lies18:06
pitilloyes, sepen's script does a dd directly to a device (not partition)18:06
jaegerCan you try it without the script?18:06
pitillowell, I'm explaining it wrong... sepen's script makes the img and after that, dd to the device18:06
jaegerright... I'm talking about using dd on the iso directly18:07
jaegerdd if=iso of=usb18:07
pitilloyes, I can try18:07
jaegerThat's what isohybrid is supposed to allow... and it boots but the media doesn't get mounted properly18:08
jaegerRomster: libxcb might be the issue, will have to test18:10
pitillojaeger: it boots dd'ing the iso file directly to the device (no partition)18:13
jaegerRomster: never mind the xcb comment, it's required by libx1118:13
jaegerpitillo: is /media mounted?18:13
pitilloyes it is, as iso966018:14
pitillosetup starts18:14
jaegerweird... well, I guess if nobody but me can reproduce this problem, it's not a big deal18:14
jaegerwhat was your dd command line?18:14
pitillosudo dd if=crux-3.0-updated.iso of=/dev/sdb18:15
jaegerok, no block size or any unusual settings18:15
jaegerthanks18:15
pitillonop18:15
pitillonp18:15
pitilloI can reproduce the error jaeger18:18
jaegeroh?18:18
pitillomoving the usb 3.0 stick to a usb 2.0 port it isn't mounting /media18:18
jaegerhrmm, that's extremely odd18:18
pitilloxD18:18
jaegerso maybe an ehci problem18:18
pitillomay be it gives you a clue...18:19
jaegerI'll try that as well18:19
jaegerrather I'll try a usb3 port now18:19
pitillotry to put on a 3.0 usb port...18:19
jaegerI wasn't before because my USB drives aren't usb318:19
pitilloyeah, the problem can be cutted a bit18:19
jaegerargh, that motherboard can't boot from usb318:23
jaegerGA-P55-USB318:23
pitilloummmm18:23
jaegermy laptop only has usb2, will have to try another machine... I guess I can try my workstation here18:24
pitilloit finds the device and media but it doesn't mount it18:24
pitilloummm dd'ing the iso directly to the device (sdb in my case) now when booting from usb 2.0 it shows sdb118:25
jaegerSounds like maybe it's an EHCI problem18:25
jaegeryeah, same behavior I see18:25
pitillois this expected?18:25
jaegerno18:25
pitilloseems it creates the partition as iso966018:25
jaegerI thought that isohybrid was doing something wrong but now I think maybe ehci-hcd is broken18:25
pitillomounting directy sdb on /media works18:26
pitillosdb1 too18:26
jaegerI'll make another one that has ehci disabled18:26
pitillooka, let me now and I'll test directly on hardware here18:27
jaegerok, thanks18:27
pitillomore for you for all your work18:27
jaegerok, made an ISO with no usb2 support, only usb1.1 and usb318:58
jaegerit fails in my workstation's usb2 port as expected, this machine has no usb1.1 support18:58
pitilloI can test it in usb2/usb3 only18:59
pitillomy laptop (usb1) doesn't boot from it if I recall right18:59
jaegerit works properly in the usb3 port except that my keyboard no longer works :D18:59
pitilloummm strange19:00
jaegerwell the keyboard was in a usb2 port19:00
jaegermoved it to a usb3 port and all is well19:00
jaegerSo that works like I expected, will now test in the other machine that has usb1.1 support19:00
jaegerok, so here are the results:19:03
jaegermachine with usb2 and usb3 support only: usb3 works19:03
jaegermachine with usb1.1, usb2, and usb3 support: usb1.1 and usb3 work19:03
jaegerusb2 is disabled on this ISO19:03
jaegerso I think the kernel's usb2 support is the problem now19:04
jaegerThe next thing I'll try is kernel 3.10.619:08
pitillodelimiting19:09
frinnstive not read the backlog, but can it be the order usb1/usb2 drivers are loaded?19:48
jaegerthe kernel bitches if you load uhci/ohci before ehci specifically19:50
jaegerso I've always loaded ehci first19:50
jaegerkernel 3.10.6 has the same problem20:08
jaegerI'll try 3.8.x as well, even though it's not stable, mainline, longterm, or anything20:09
jaeger3.8.13 is also affected20:55
jaegerjue: would you mind adding a symlink to the jre port? usr/lib/jre/bin/ControlPanel -> usr/bin/ControlPanel21:07
sepenguys, what do you think about the new contrib application?21:25
sepenI don't know him before but looks like a guy wanting to work fine21:27
jaegerI haven't looked closely at his ports, myself, but he seems willing to adapt quickly and is friendly enough21:30
sepenyeah I wanted to mean that21:31
sepenwe need new blood for contrib, as someone said ;D21:31
jaeger:)21:32
jaegerI don't have any complaints from watching the emails go by21:32
jaegerkernel 3.4.57 works properly21:36
*** Workster has joined #crux-devel23:49

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