/srv/irclogs.ubuntu.com/2010/09/08/#ubuntu-arm.txt

=== bjf is now known as bjf[afk]
* rsalveti interested at what display issue ndec had06:16
=== hrw|gone is now known as hrw
hrwmorning08:28
vstehleogra_cmpc: The daily image looks nice :) I just replaced the kernel and kept everything else (MLO, u-boot, etc...)08:29
=== XorA|gone is now known as XorA
vstehleogra_cmpc: I remark there is no console on UART with the pre-installed image. Don't you think it would be useful for some to have the 'login' prompt on UART too?08:46
persiaThat's semi-intentionally not present.08:48
persiaWe generally don't feel that most end-users are likely to understand the concept of serial console, let alone have any need to use one.08:48
persiaMind you, it might be interesting to make it easier to enable.08:49
DanaGHmm, I set up my beagle boot.scr and user.scr so normal is normal and alternate is development mode.  Normal has splash (had to force with FRAMEBUFFER=Y), aufs, and no serial console (because serial console forcibly disables splash).  Alternate has serial console, and no aufs.08:54
DanaGhttps://bugs.launchpad.net/ubuntu/+source/plymouth/+bug/51682508:55
DanaGAh, maybe I should move my comments to a new bug.08:56
DanaGMy desired behavior would be to show the splash on tty0, but not hide output on ttyS0.08:56
DanaGhttps://bugs.freedesktop.org/show_bug.cgi?id=2223908:57
persiaYeah, you want a new bug.08:57
persiaBut one could strongly argue that having plymouth run *twice*, once in splash mode and once in console mode is a bit messy.08:58
DanaGWhat's odd is that pressing escape toggles the text-as-splash thing... and ruins the serial console.08:59
DanaGCouldn't you just have Plymouth completely disregard the serial console, and handle only tty?09:00
DanaGWhy does it have to hook into both?09:00
persiaDunno.  I'd suggest asking either Keybuk (in -devel), or upstream.09:01
ogra_cmpcvstehle, i have a hack i plan to apply to jasper (the firstboot tool) that will enable a login prompt automatically if a serial console is set on cmdline...09:51
vstehleogra_cmpc: Nice!09:52
ogra_cmpcvstehle, for the reasons persia and DanaG pointed out we wont set a serial console by default in boot.scr though09:52
vstehleogra_cmpc: To continue with the image: I think the 'system testing' tool gets stuck on my board...09:52
ogra_cmpcso that setp is something i expect a dev to be able to do manually09:52
ogra_cmpc*step09:52
persiacheckbox is getting stuck?  How?09:52
ogra_cmpcsounds like a bug09:53
vstehle'Gathering information from your system...' (since one hour now)09:53
vstehleShall I enter a launchpad bug?09:53
persiaHrm.09:53
persiaPlease enter a bug,09:53
vstehleon cdimage?09:53
ogra_cmpcmight be confused when nt finding a pci bus or something, file a bug using ubuntu-bug (that will attach all the needed logs)09:53
persiaThat's what I'm thinking, but I thought that got patched out in karmic09:54
vstehleogra_cmpc: I have no network on the board for now... :(09:54
ogra_cmpcthe package for the test tool is called checkbox09:54
persiaOh, it's the GL check.09:54
ogra_cmpcah09:54
ogra_cmpcmakes sense09:54
vstehleogra_cmpc: Is there a checkbox 'log' I could look at, somewhere?09:54
* ogra_cmpc doesnt know but i would expect so09:55
persiaThere definitely is a log.09:55
persiaAh, found it.  It's probably `lspci | grep -q VGA` that hangs.09:56
persiavstehle, Could you try that command to see if it does hang?09:56
vstehlepersia: It complains but does not hang. (cannot open /proc/bus/pci, etc...)09:57
persiaHow about `grep Loading $XORG_LOG | grep -q "${i}_drv\.so"` ?09:58
* persia isn't seeing much that might be contentious09:58
persiaOh, that won't hit: we'll drop into DRV=$UNKNOWN.  Nevermind09:58
persiaDoes xvinfo work?09:59
vstehlepersia: Xorg loaded two so: fbdev & evdev09:59
persiaRight, but the ${i} is apparently from a whitelist, which escaped me at the first readthrough of the code :)10:00
vstehlepersia: xvinfo is a bit strange: screen #0\n no adpators present10:00
persiaThat's incorrect, but oughtn't break the script.10:00
vstehleBug #633005 :)10:01
persiaLooks like by default checkbox.log drops into .10:04
vstehlepersia: There is a /tmp/checkbox*** dir10:05
vstehlepersia: Hmm. With pipes, after all.10:05
persiaWell, /usr/bin/checkbox-gtk and /usr/bin/checkbox-cli both contain what looks like an attempt to use $CHECKBOX_DATA, falling back to . if absent.10:06
persiaBut I may be mistaken.10:06
vstehlepersia: $HOME/.cache/checkbox/checkbox.log ?10:06
persiaQuite possibly: I may have read that wrong.10:07
persiaAh, indeed.  it's "$XDG_CACHE_HOME/checkbox", so that's the right log.10:08
vstehlepersia: Traces before "hang" are Started firing gather , Calling None.ScriptsInfo.gather() ... , Calling None.BackendInfo.gather() ...10:08
* persia misses the bot10:08
persiaHow did you file that bug?10:11
vstehlepersia: Launchpad, go to package checkbox, click on 'Report a bug'10:12
persiaAh.  That explains the lack of data :)10:12
vstehlepersia: Yes, sorry, no network (I have troubles finding eth switches those days)10:13
persiaWould you mind running `apport-collect 633005` on the affected system (if it has access to LP)?10:13
* ogra_cmpc wonders why we didnt have an image build tonight10:13
persiaOh, right.  Nevermind :)10:13
ogra_cmpcnot even an attempt it seems10:13
vstehlepersia: I'll disconnect another board, hold on10:13
persiaNo worries.  In general, it's best to try to file the bugs from the boards, because that includes more information, but that's more important with crashes than hangs.10:14
persiaThe odd bit is that BackendInfo.gather() only creates some named pipes...10:14
vstehlepersia: Maybe there is a "spawned" process, which exited?10:15
vstehlepersia: And the father is waiting for some data on the pipes?10:15
persiaPotentially.10:16
persiaHad you passed the request for the administrative password?10:17
ogra_cmpchttp://people.canonical.com/~ubuntu-archive/livefs-build-logs/maverick/ubuntu-netbook-omap4/20100908/livecd-20100908-armel.out10:18
ogra_cmpcGRRRRR !!!!!!!!!!!!!!!!10:18
vstehlepersia: I don't remember. More than one our ago is too much for my little brain :)10:18
ogra_cmpcapw, seems its still messed up ^^^^10:18
persiaTry it again.  The request for administrative rights should be quick (or if it isn't, we found the bug)10:18
vstehlepersia: I need to reboot because I cannot kill the window (other bug?) hold on...10:19
persiaNo rush :)10:24
apwogra_cmpc, which kernel was in the archive when it failed10:25
apwdue to a disconnect between tim and myself there were two uploads for that kernel10:25
ogra_cmpcoh, then it might be stuck in NEW10:26
ogra_cmpci'll check as soon as i'm on the other computer10:26
apwogra_cmpc, the fixed version looks to have hit the archive around 3am10:26
apw(my time)10:27
ogra_cmpcthe 903.10 one seems to have built10:27
ogra_cmpc(your time is buildd time too ;) )10:27
apwi thought they were on UTC10:27
apwwhich for half the year we are on10:27
ogra_cmpcisnt BST = UTC ?10:28
ogra_cmpcah10:28
apwGMT == UTC, BST == UTC + 110:28
ogra_cmpcso i have to wait 2 months for my statement to be true :)10:28
persiaBST is within 1 second of UTC half the time, and closer to 3600 the rest of the time.10:28
=== kmargar is now known as markos_
apwno BST is always GMT+1 hour, and GMT is <1s different from UTX10:29
* ogra_cmpc twiddles thumbs10:29
apwUTC10:29
apwogra_cmpc, i am still not sure this damn thing is right10:29
* persia declares everyone should always use numeric timezone designators10:29
apwit looked good in my testing, but i am not sure its built the common headers10:29
ogra_cmpcit doesnt seem to have the dot version in it10:29
apwyou may have what you need in the archive, but this build isn't quite right yet10:30
apwdamn they made a mess when they rebuilt this branch for .35 ... not quite sure how they made it so wrong10:30
apwand 7 hour rebuilds are not helpful10:30
ogra_cmpclinux-headers-2.6.35-903-omap4 is what you built10:31
apwyeah that is right, and it should depend on linux-headers-2.6.35-903, which it did no build10:31
apwit may be in the archive however10:31
ogra_cmpclinux-ti-omap4-headers-2.6.35-903 is what the meta looks for10:31
ogra_cmpcaccording to the image build log10:32
ogra_cmpchow do you manage all that without getting your brain fried ?10:33
apwahh crap is meta fried too ...10:34
apwogra_cmpc, where in the 20 pools do i find the built images ?10:34
ogra_cmpchttp://ports.ubuntu.com/pool/main/l/linux-ti-omap4/ you mean ?10:35
ogra_cmpcand http://ports.ubuntu.com/pool/main/l/linux-meta-ti-omap4/ i think10:36
apwogra_cmpc, ok i think the bits you need happen to be in the archive binary wise though the build needs another fix10:37
apwan old one is there at the moment which i believe the dependancy would use10:37
apwnow the meta package seems broken as well ?10:37
ogra_cmpcwell, the images need it proper to be buildable10:38
apwthe meta package ?10:38
apwPackage: linux-headers-omap410:39
apwArchitecture: armel10:39
apwSection: devel10:39
apwPriority: optional10:39
apwDepends: ${misc:Depends}, linux-headers-${kernel-abi-version}-omap410:39
ogra_cmpci need to check the NEW queue, let me finish breakfast to see whats going on (this classmate isnt powerful enough to open more than one website and xchat)10:39
apwthe meta package seems to point to the correct name10:39
apwogra_cmpc, so i _think_ whats in the archive right now ought to install, though its not quite right10:40
cooloneyapw: i failed to see the bug #632235 in https://bugs.launchpad.net/ubuntu/+source/linux-ti-omap410:40
cooloneyapw: is that because that it is 'Fix Released'?10:41
apwyep, it would have autoclosed on upload10:41
ogra_cmpcapw, i just triggered an image build, lets see if it gets through, might just have been a timing issue10:41
apwnot sure it was in in time to fix the CD though10:41
apwogra_cmpc, how long does that take?  i'll get the remaining issue sorted out shortly10:41
apwbut want to know its the last issue first10:41
persiaThe image is trying to pull ${Kernel-Stem}-omap4 [armel] as a metapackage.10:42
ogra_cmpcapw, if it fails, it fails fast10:42
ogra_cmpcgrr, why is the omap3 buildd dieing all the time10:43
ogra_cmpcssh: connect to host acorn.buildd port 22: No route to host10:43
ogra_cmpcacorn.buildd finished at Wed Sep  8 10:40:56 BST 2010 (failed)10:43
ogra_cmpcNull message body; hope that's ok10:43
ogra_cmpc(omap4 didnt fail yet)10:43
* ogra_cmpc ignores breakfast and goes upstaris to the office10:44
ograok, the NEW queue is empty10:45
ograapw, failed, same message10:50
apwcan you paste the message pls10:50
ograoh, wait, its different10:52
* ogra cant copy paste from w3m which i need to use for reading the logs ... one sec10:52
apwheh always the way10:52
ogralinux-headers-2.6.25-903-omap4 depends linux-headers-2.6.35-90310:52
ogras/25/35/10:53
ograno subarch at all on the latter now10:54
apwyep10:54
=== amitk is now known as amitk-afk
ograalso 20100908.1 should show up soon at http://people.canonical.com/~ubuntu-archive/livefs-build-logs/maverick/ubuntu-netbook-omap4/10:54
apwOH now i understand ... i am getting fooled ...10:55
apwthe build is producing the headers, but in 'all'10:55
apwbut its doing that ... during the armel build ... so they don't get picked up10:56
apwie ... test builds make both, one armel, one all, but the build will only pick up the _armel builds during a real buildd build10:56
ograoh, right10:56
ograyou should make the package "arch: armel" only10:57
apwdammit, i'd claim noone had tested, but ... its very very easy to overlook in our test harnesses10:57
ograthat saves you from all the build failure spam on x86/amd64/ppc too ;)10:57
apwogra, yep, am doing so10:57
* persia encourages folk to use sbuild so that building arch:all packages happens only by intent10:58
apwthe annoying bit is that this was all correctly configured prior to the conversion to .35 and it un10:59
apwunexpectedly all got lost10:59
apwpersia, yeah sbuild would be good, but iirc it11:00
apwit has some assumptions about naming that clashes with naming we use already for different things11:00
apwso its not a simple slot in11:00
persiaapw, The key bit is that sbuild is what runs on the real buildds, so ... :)11:06
persiaAnyway, bug me about it in October, and maybe we can find a way to drop it in to your framework in November.  Sooner is unlikely, what with release and next cycle planning.11:07
apwpersia, don't get me wrong its the right thing, but setting things up its way is more tricky11:07
apwdue to some contraints on our existing framework11:07
apwspecially as we have few machines which can even do an arm build in a sensible time11:08
persiaI figured.  I've just written a number of scripts using the sbuild framework, so thought there may be opportunity for synergy.  I won't complain if you don't want me to look at it :)11:08
apwpersia, should do, at one of the sprints i recon11:08
persiaThat's why I suggested October.  I'm expecting to be at UDS.11:08
apwheh good enough11:09
ogracooloney, who cares for the imx51 kernel beyond smb ?11:24
ograseems we have a bad bug in the lucid kernel that makes java builds fail11:24
ograbug 60504211:24
ubot2Launchpad bug 605042 in eglibc (Debian) (and 10 other projects) "[armel] java fails to start with eglibc-2.12-0ubuntu4 (affects: 2) (heat: 12)" [Unknown,Unknown] https://launchpad.net/bugs/60504211:24
persiaUgh.11:26
ograyes11:26
ogravery ugh11:26
ograapw, do you know who we can assign that to ? its rather urgent else we cant build any java related packages it seems11:27
persiaAnd that blocks a whole bundle of stuff, unfortunately.11:28
apwogra, which arm is it?11:29
ograapw, imx51 ... its the buildd kernels11:29
apwwell i guess we need one of eric/bryan or lee on it, they are are area experts11:30
ograright11:30
apwogra, do you know where the userspace stack is normally on arm ?11:34
ogracan you elaborate11:35
=== amitk-afk is now known as amitk
ograwhich userspace stack ?11:35
apwogra, any/all userspace stacks, what addy are they placed at on arm by defualt11:39
ograapw, i still dont get what you want :/11:42
persiaapw, We don't do it that way.11:42
ogrado you want to know where the packages live ?11:42
ograor where the archive is ..11:42
persiaRather, we pass an initrd to the bootloader (varies by device), and then pass root= as a kernel argument.11:43
apwthe kernel puts the userspace stack in a specific place, the actual address of the actual memory which represents the stack of the CPU11:43
ograoh11:43
apwtrying to work out if these addresses are stack related11:43
* ogra was associating userspace with desktop/console tools etc11:43
apwheh understandable11:43
apwogra, so what makes us think this is a kernel issue, i see nothing other than a change to elibc causes SEGV's11:45
persiaBecause it's only reproducible on imx5111:45
ograit seems to not happen on other arches11:46
apwthat doens't meant its not broken on all arches though11:46
persiaother *boards*11:46
persiahttps://bugs.launchpad.net/ubuntu/+source/eglibc/+bug/605042/comments/2011:46
ubot2Launchpad bug 605042 in eglibc (Debian) (and 10 other projects) "[armel] java fails to start with eglibc-2.12-0ubuntu4 (affects: 2) (heat: 12)" [Unknown,Unknown]11:46
apwthere is no analysis at all as to cause11:46
apwogra, ok i've followed it as far as I can quickly, its a non-trivial arch specific function that i can't easily find ... so we need an imx expert on that one11:58
ograyeah11:58
ograthats why i asked in the beginning :)11:59
apwindeed12:07
apwogra, ok i've pushed and uploaded what should fix that last headers bug12:10
ograyay12:11
ograi'll try as soon as its in the archive12:11
apwogra, its on a buildd already amazingly12:13
ograwohoo12:14
ograthe publisher will take ages though12:14
apwas will the build :(12:14
ograoh, it wasnt the meta ...12:15
ograyeah12:15
apwogra, no sadly its the main kernel12:15
ograwell, 5h plus publisher12:15
* apw will be invistigating why the configuration was not simply carried forward12:16
apwavoiding all this pain12:16
ograavoiding the pain in a real way would mean TI hires 100 monkeys to upstream all that stuff :)12:16
apwheh there is that12:17
apwwe still should have avoided this, not happy12:17
* ogra takes a break12:41
=== freeflyi1g is now known as freeflying
ndecogra: hi... when is the next daily image being planned? the one with es2.0 support, i mean.13:36
ndecpersia: hi! any update on the pulse bug #631362?13:47
prpplaguendec: did you get your audio tested ?13:47
ubot2Launchpad bug 631362 in pulseaudio (Ubuntu Maverick) (and 1 other project) "Include several configuration files (affects: 1) (heat: 14)" [Medium,In progress] https://launchpad.net/bugs/63136213:47
ndechi prpplague... I am just doing it ;-) I can use HMDI, but not headset.13:48
prpplaguendec: which board version do you have?13:48
ndecprpplague: the latest and greatest ;-)13:48
ndecprpplague: 8-layer, es2.013:48
prpplaguendec: any errors?13:48
ndecprpplague: the speaker-test command you sent works for HMDI, I can alsa play gst stream through alsasink to HDMI.13:49
ndecprpplague: but the same command fail with HS. speaker-test tells me : playback open error: -16, device/resource busy13:51
prpplaguendec: most likely you have some other userspace apps accesing the device13:52
prpplaguebrb13:52
ograndec, once the kernel team fixed all issues with dependencies we had due to the version change ... we think this is fixed with the currently building package and i will fire off an image build as soon as thats in the archive (likely later this evening)14:01
ndecprpplague: ok. might be. I am doing an upgrade, when it's done, I will kill GDM and restart the same test.14:01
=== rgreening_ is now known as ghost
=== ghost is now known as rgreening
ograndec, i talked to TheMuso (luke, our pulse maintainer) yesterday in #ubuntu-devel, he said he has a bunch of other fixes too and wants to upload all of them in one go, the multi config file support will be included in that but he couldnt give an ETA for the upload14:02
ndecogra: great. I just wanted to know if it will be committed in 10.10.. looks like yes.14:03
ograyes, promised :)14:03
=== ogra changed the topic of #ubuntu-arm to: Ubuntu ARM Discussion & Development | https://wiki.ubuntu.com/ARM | Want to Submit a Bug? https://bugs.launchpad.net/ubuntu/+filebug | Build a rootfs from scratch: https://wiki.ubuntu.com/ARM/RootfsFromScratch | also visit #linaro | Maverick beta see https://wiki.ubuntu.com/ARM/OMAP | wanna cross build ? see http://idlethread.blogspot.com/2010/09/cross-compilation-redux.html
* ogra makes amitk famous :)14:05
amitkogra: or infamous (since the debuild recipe for ubuntu kernels fails due to perf dependency) :-/14:07
ograbah14:07
ograshould i drop it from the topic again ? or is someone working on a fix for that ?14:08
amitkogra: will need to fix that with some dpkg-cross instructions14:08
ograk14:08
amitkogra: the blog post itself might be useful to advertise the toolchain14:08
ograright, thats what i thought14:08
amitkI'll update the exact kernel instructions14:08
ogragreat14:09
ograin case the url changes through that, please tell me, i'll update it14:09
rsalvetimorning14:11
rsalvetindec: hey, you were having display issues yesterday14:12
ograyesterday ?14:12
vstehleEst-ce que ndec est par ici?14:13
vstehlendec: "We are waiting for the chair person to join" :)14:13
ogralol14:14
* ogra knows that feeling14:14
ograshe has such a nice voice :)14:14
vstehleogra: *Oups* Wrong chan. Now you know that we have our own problems... ;)14:14
ograheh14:15
vstehleogra: (She indeed has a nice voice.)14:15
rsalveti<ndec> ogra: rsalveti: if I turn my display ON a few sec after boot, the resolution is fine. I think i have same problem with race condition and X start...14:21
rsalvetiogra: ^ this issue14:21
ndecrsalveti: thanks for the follow up...14:21
* rsalveti wants to know with which kernel14:21
ograrsalveti, yeah i was just jokiag about the "yesterday" he always has these issues :)14:22
ogra*joking14:22
ndecrsalveti: i am using sebjan's kernel, es2.0, 8-layer. this should be equivalent to what has been uploaded today14:22
* rsalveti still has lots of emails and backlog to ready14:22
ograndec, i think we are one patchset behind14:22
ndecdon't you have the same issue?14:22
rsalvetilet me check our git tree14:22
ndecogra: cooloney pull request was uploaded this morning, i haven't checked but it might be building now14:22
ograthe merge request was only filed today14:23
rsalvetiwe got an updated tree already14:23
ograi think whats building now wasnt a fresh checkout, just a packaging fix from apw14:23
apwndec, ogra, yep ... not his pull request.  if he had mentioned it 2 hours ago we might have put it in the same one!14:24
ndecogra: .11 is building now, and has all patches for 6 and 8-layer, and es214:24
rsalvetiyep, but we're still behind some pm fixes14:25
apwright14:25
rsalvetibut our current kernel will work fine for es2 6 and 8 layers14:25
rsalvetindec: the issue I'm having is: http://www.flickr.com/photos/rsalveti/4932601965/14:25
rsalvetibecause of the disconnect/connect during hdmi detection14:25
ograndec, i was referring to the pull request from 14:49 today (local time) ... all other patches should be in14:26
ndecapw: ogra: rsalveti: right... i missed today's pull request... anyways, .11 should work on 6 and 8 layer ES2.0. the today's pull request is another set of fix14:26
ndecrsalveti: so... i have another one ;-) I only see on the screen the top left part of the desktop. if i turn off HDMI on the TV, let the kernel boot and GDM come, and then turn TV to HDMI resolution is fine. I have a 1920x1080 display (xrandr).14:28
rsalvetindec: hm, interesting14:29
rsalvetican you boot with omapdss.debug=1 and check both boot logs?14:30
* rsalveti never tried to turn on his monitor after boot14:30
* rsalveti will also try that14:30
rsalvetiif gdm (X) is fine, then it got the correct resolution before opening it14:31
rsalvetiso when it doesn't work then it could be a racing condition when starting X and the size_notify (that changes the fb resolution)14:32
rsalvetiogra: hm, no new images to download :-)14:32
ograrsalveti, yeah, kernel package breakage14:33
ograrsalveti, we should have new images by end of my day so you have something to play with for your afternoon14:33
rsalvetihm, that's why a lot of apw's commits14:33
rsalveticool14:33
ograright, he saved out butts :)14:33
ogra*our14:33
rsalveti:-)14:33
ograapw is our new amitk *g*14:34
apwyeah its very difficult to test outside a real build14:34
apwi hope its all resolved now14:34
rsalvetisakoman: cool, thanks for including the patch14:44
rsalvetinow just need to add it to jcrigby linaro's package14:45
ndecrsalveti: i will share the dssdebug later, i am doing an apt-get upgrade ;-)14:45
rsalvetindec: haha, got it working like you!14:46
ograrsalveti, yeah, we'll also need to talk to slangasek and jcrigby about ES2.1 support (since we will need a freeze exception for it)14:46
rsalvetindec: after turning it on later, it worked fine!14:46
* rsalveti can now check his logs14:46
ndecrsalveti: cool... it's actually funny that I found out this use case ;-)14:46
rsalvetiogra: es2.1 support?14:46
ograrsalveti, yep14:46
rsalvetiwhen are we getting es2.1?14:47
ograbeginning of oct. i'd guess14:47
sakomanrsalveti: I'm going to try to sneak that in upstream as a single patch14:47
ndecrsalveti: the thing is I received a new TV, put it out of the box, plug the cable and booted. and after a few seconds I realized that I was not on HDMI, so switch the TV to this ;-)14:47
rsalvetioh, ok, the "final" one14:47
sakomanrsalveti: if they complain I'll break it back into two14:47
ograright14:47
rsalvetisakoman: yep, saw that, cool14:47
rsalvetindec: hah, cool :-)14:48
loolamitk: Would you be tempted to send a patch for fixing cross-compilation of the kernel in the standard way?  currently, it needs -eCROSS_COMPILE; in other packages, debian/rules takes care of setting this kind of stuff so that "dpkg-buildpackage -aarmel" just works; what would need to be changed is setting CROSS_COMPILE to $(DEB_HOST_GNU_TRIPLET)- when DEB_HOST_ARCH and DEB_BUILD_ARCH aren't equal (cross-compilation)15:11
amitklool: let see if I can tempt apw for this ^^^ ;)15:16
apwas in a simple if and set?  that doesn't sound too onerous15:17
=== bjf[afk] is now known as bjf
amitkapw: pretty much15:22
loolapw: Yup, you need to DEB_HOST_GNU_TRIPLET ?=, DEB_HOST_ARCH ?= etc. if they aren't already set somewhere, then ifneq ($(DEB_HOST_ARCH),$(DEB_BUILD_ARCH)) export CROSS_COMPILE=...15:22
loolIn fact, you might want to CROSS_COMPILE ?= in the if body as this allows overriding CROSS_COMPILE as amitk did15:23
amitklool: but what is the std. prefix for CROSS_COMPILE?15:23
loolSo that one can still use non-standard toolchains15:23
lool(arm-none-linux-gnueabi- from CodeSourcery comes to mind, it's not the same triplet as the Debian arch expects)15:24
loolamitk: I don't understand the question15:24
amitklool: I use CROSS_COMPILE=arm-linux- (by linking arm-linux-gnueabi-* to arm-linux-*)15:24
loolamitk: Oh, you should CROSS_COMPILE=arm-linux-gnueabi-15:24
loolarm-linux- is incorrect15:25
loolCROSS_COMPILE is what gets prepended to all toolchain tools, e.g. upstream just calls $(CROSS_COMPILE)gcc and CROSS_COMPILE is empty for native builds15:25
looland the upstream toolchain will prefix all tools with the triplet in cross-toolchains, so CROSS_COMPILE should be triplet suffixed by a -15:25
amitklool: where can I find a good explanation of the 'triplet'? I assume the triplet is the std. prefix15:26
apwisn 't that triplet some kind of gnu standard15:28
loolamitk: GNU triplet is what you configure a toolchain with15:29
loollet me find some referene15:29
rsavoyea triplet is a standard gnu config string15:29
rsavoyeit's heavily used to differentiate between the build, host, and target system15:30
loolapw, amitk: I'm unable to point at a concise definition; multiple places refer to these, and have various lists15:32
loolhttp://gcc.gnu.org/install/configure.html explains --host/--build/--target which all take triplets15:32
looltriplets values are best defined in configure.ac15:32
loolhttp://gcc.gnu.org/install/specific.html has some lists of targets15:32
loolhttp://gcc.gnu.org/onlinedocs/libstdc++/manual/internals.html explains some components of the triplet15:32
rsavoyethe idea at the time was build is the machine you are on, host is what it will be running on, and target is the final output of gcc15:32
rsavoyemost other apps use just build and host15:33
loolapw, amitk: Sorry, can't find a precise reference for GNU triplets, but it's basically a toolchain concept telling the toolchain how to configure itself at the highest level; a Debian/Ubuntu port has only one such triplet, and cross-gcc is expected to be at $triplet-gcc15:43
loolsame for -ld, -objdump etc.15:44
=== bjf is now known as bjf[afk]
rsavoyelool: I can talk about config triplets all you want, what exactly do you need to do ?15:44
amitklool: sounds complicated policy. I'll go through your links...15:45
Nekohey guys15:45
amitkrsavoye: just what they are and why we should care when we set CROSS_COMPILE :)15:45
loolrsavoye: I don't think we need any detail; amitk and apw needed some kind of primer on the concept15:45
Nekoanyone noticed a launchpad bug or anything that describes X taking up 90% of the CPU *all* the time?15:45
loolamitk: It's just a toolchain input to configure it15:45
loolamitk: the output is that the tools are at $triplet-tool  :-)15:45
rsavoyeright, so other configure scripts can find their tools15:46
loolamitk: CROSS_COMPILE is a linux var, it's also used in other projects like U-Boot, but it's not extremely wide-spread15:46
loolNeko: This can happen for a variety of reasons15:46
Nekothe reason here seems to be that I started X and opened a terminal15:46
Nekoit just hit 80%15:46
rsavoyewhich is where issues like arm-linux-eabi or arm-linux-gnueabi are important15:46
NekoI close the terminal and it stays at 80% even if I'm on another VT15:47
NekoX is spinning wildly somewhere15:47
Nekohttps://bugs.launchpad.net/ubuntu/+source/nvidia-graphics-drivers-180/+bug/617994 <-- this does help btw (the disable lcd font smoothing thing)15:47
ubot2Launchpad bug 617994 in nvidia-graphics-drivers-180 (Ubuntu) "Slow performance and high CPU usage on Maverick (affects: 11) (heat: 58)" [Undecided,Confirmed]15:47
loolNeko: you could try tracing it15:47
loolNeko: it could be anything, like a X client doing a lot of server requests15:48
Nekowhat do I use to trace X activity :/15:48
apwNeko, i'd swing by ubuntu-x they'll be better equiped to answer15:48
Nekookay15:48
prpplaguendec: any luck on the audio?16:03
ndecprpplague: not yet... upgrade is taking forever...16:03
prpplaguendec: np, i'll be online all day if you need assistance16:04
=== zyga is now known as zyga-kids
devilhornsso, has anyone had success in making arm images that run w/ qemu in maverick ?? and if so, can I download one ? :)16:09
=== amitk is now known as amitk-afk
=== orbarron1 is now known as orbarron
=== bjf[afk] is now known as bjf
=== hrw is now known as hrw|gone
apwogra, those images are built18:24
GrueMasterHe may be monitoring on his other system.  ogra_cmpc ^^^18:31
ogra_cmpcapw, thanks, but publishing will take a while18:39
* ogra_cmpc waits for .11 to show up at http://ports.ubuntu.com/pool/main/l/linux-ti-omap4/18:41
GrueMasterWill this also fix the missing headers issue?18:42
apwits meant to only sort out those headers18:43
ogra_cmpcGrueMaster, the missing headers were fixed two days ago with the last netbook-meta upload18:43
GrueMasterThe following packages have unmet dependencies:18:44
GrueMaster linux-headers-2.6.35-903-omap4 : Depends: linux-ti-omap4-headers-2.6.35-903 but it is not installable18:44
* ogra_cmpc thinks GrueMaster means a different issue than apw :)18:44
ogra_cmpcGrueMaster, yeah, thats apw's fix18:44
apwahh18:44
GrueMasterCool18:44
ogra_cmpcoh, actually i might have introduced it18:44
ogra_cmpcthe omap4 headers were missing completely from the images until monday18:45
ogra_cmpcapw, so that breakage was always hidden, it could as well have been present from day one, we just didnt see it18:45
ogra_cmpci.e. not introduced with the version change18:46
apwindeed, i see those two issues as one big "headers is bust" problem18:46
apwas you say the bad linkage was hidden by the bad meta linkage and the bad everything else18:47
ogra_cmpcheh, yeah18:47
apwogra_cmpc, is ports publish slower than normal publish ?18:47
ogra_cmpcabout 45 min at least18:47
apwdammit18:47
ogra_cmpcits a bit slower than the main archive ...18:48
ogra_cmpceven though persia always claims thats supposed to be fixed18:48
GrueMasterWill the next build have the new xloader & uboot?  I tried 20100906.2 on ES2 8L, but it hung with an xloader crash.18:50
GrueMaster(could also possibly be my desktop system & flashing SD cards.)18:51
ogra_cmpci would blame your desktop18:52
ogra_cmpcjust installing the linux-image from the archive gets me a working system18:52
ogra_cmpcthough ... hmm, i'm using 6 layer here18:52
ogra_cmpcsince 8 layer has USB issues18:53
GrueMasterYea, I was afraid of that. There appears to be a bug in my system that requires me to reboot weekly.18:53
GrueMasterOtherwise the SD i/O is very slow on my USB multicard reader.18:54
ogra_cmpcapw, BAH !!, we're stuck in NEW18:54
ogra_cmpchttps://edge.launchpad.net/ubuntu/maverick/+queue18:55
apwogra_cmpc, wtf .... arg i guess the headers package may not have been seen before18:55
ogra_cmpcyeah18:55
apwi'll let you wake up an admin!18:55
rsalvetiGrueMaster: it should work with your es 2 8 layer18:57
ogra_cmpclinux-headers-2.6.35-903_2.6.35-903.11_armel.deb  (10.1 MiB) NEW ...18:58
ogra_cmpchmm18:58
rsalvetithe new x-loader got updated at monday18:58
ogra_cmpcsays universe too18:58
GrueMasterI'll try again after a desktop reboot.18:58
rsalvetiGrueMaster: with the latest x-loader you should at least boot your kernel18:59
apwogra_cmpc, looks like its what i'd expect that file to look like at least19:00
ogra_cmpcyeah19:00
ogra_cmpci'm just worried about universe19:00
ogra_cmpcrequires special treatment19:00
ogra_cmpcup to the AA though19:01
ogra_cmpci pinged the admins of the day in -devel19:01
apwogra_cmpc, thats the default isn't it, anywhere else is an override set on accept19:01
ogra_cmpcright, but all other binaries of that package are in main already19:02
ogra_cmpceasily overseen that one goes to universe19:02
apwyeah19:02
apwi think kernel accept is scripted somehow, so hopefully its sorted19:03
ogra_cmpcyep19:03
ogra_cmpcwe'll just have to wait19:03
* GrueMaster is good at waiting.19:03
GrueMasterLots of experience.19:04
ogra_cmpclol19:04
GrueMasterogra_cmpc: Could you trigger a dove build?  NCommander can't ssh in.19:14
ogra_cmpcGrueMaster, running19:16
NCommanderthank you ogra_cmpc19:16
ogra_cmpc90min ...19:16
GrueMasterok19:17
* GrueMaster sets his tea timer accordingly.19:18
ogra_cmpcGrueMaster, NCommander failed19:56
GrueMastersigh.19:57
armin76:D19:59
GrueMasterogra_cmpc: Not seeing an email.  what was the failure?20:00
ogra_cmpccheck the logs20:01
GrueMasterAnd they would be located....20:02
GrueMasterhttp://people.canonical.com/~ubuntu-archive/livefs-build-logs/maverick/ubuntu-netbook-dove/20100908/ is empty20:02
GrueMastergrrr.  Omap4 headers?!?  Really?20:03
NCommander....20:04
NCommander*argh*20:04
armin76lol20:08
=== ian_brasil___ is now known as ian_brasil
beagleeI just installed ubuntu lucid in beagleboard and logged in terminal21:41
beagleewhich command shoud I type to lunch gui21:41
beaglee?21:42
beaglee*launch21:42
GrueMasterbeaglee: Where did you download the image from?  Our live image should boot right into the netbook launcher.21:42
beagleehttp://elinux.org/BeagleBoardUbuntu#RootStock:_Build_an_Ubuntu_root_file_system21:43
beagleei build with rootstock and including xfce4,gdm,xubuntu-gdm-theme,xubuntu-artwork21:44
GrueMasterOh.  Well, that is one way to do it.  Try "sudo apt-get install ubuntu-netbook"21:44
GrueMasterI don't know much about xfce.21:44
GrueMasterYou could also just try startx21:44
beagleeok21:44
beagleedid'nt work21:45
rsalvetiubuntu21:46
rsalvetiargh, wrong terminal :-)21:46
GrueMasterok.  Try "sudo tasksel" and select an installation type.21:46
rsalvetihm, you installed gdm, so trying to load it seems the way to go21:47
beagleedidn't work neither21:51
=== ian_brasil_ is now known as ian_brasil
apwogra_cmpc, any luck with finding an AA ?22:11
=== rtp_ is now known as rtp
ogra_cmpcapw, yes, waiting for publisher now22:26
apwogra_cmpc, excellent, it was there just a 10 mins or so ago, glad to see it gone22:27
ogra_cmpcwell, colin said he did it22:28
apwhe is a star22:29
ogra_cmpcyep22:30
apwogra_cmpc, when do the CDs build, in a couple of hours ?  will you let it run naturally ?22:32
ogra_cmpc1:55 UTC22:33
ogra_cmpci'll monitor the archive though and trigger a build earlier22:33
GrueMasterrsalveti: I'm going back through my backlog of todo's & retests, and was retesting bug 566645.  Doesn't appear to have been fixed.22:41
ubot2Launchpad bug 566645 in linux-ti-omap (Ubuntu Maverick) (and 2 other projects) "OTG configuration is broken on omap kernel (affects: 4) (dups: 1) (heat: 52)" [High,Fix released] https://launchpad.net/bugs/56664522:41
GrueMasterNo errors, but also no OTG functionality.22:41
rsalvetiGrueMaster: hm... with maverick?22:42
GrueMasterYesterday's image.  20100806.222:42
rsalvetiGrueMaster: ok, because of bug 60831222:43
ubot2Launchpad bug 608312 in linux (Ubuntu) "Usb host mode on OTG doesn't work on Maverick with BeagleBoard (affects: 1) (heat: 89)" [High,In progress] https://launchpad.net/bugs/60831222:43
rsalvetiso not having an oops at maverick is a "fix"22:44
rsalvetibut to make OTG work we have this other bug22:44
rsalvetiGrueMaster: another test is to check if this was changed for lucid22:44
GrueMasterOk.22:45
GrueMasterNo oops, so I guess that is fixed at least.22:45
rsalvetiGrueMaster: yep22:45
=== bjf is now known as bjf[afk]
persiaGrueMaster, rsalveti: so USB client is working?22:55
GrueMasterStill need to test that on beagle/XM.22:56
rsalvetipersia: just usb gadget I think22:56
rsalvetihost is still broken at otg (using beagle)22:56
persiaErr, right.  I meant "gadget" :)22:56
persiaWell, cool!  That at least gives us network.  Do we have any other gadget software?22:57
GrueMasterThat's what I meant as well.  Added to my todo testing.22:57
persiaGrueMaster, By "no OTG functionality", you mean gadget isn't working?22:59
GrueMasterI plugged in my OTG cable and should be able to use a USB keyboard or see a USB drive attached.  Nothing at this time.22:59
rsalvetiGrueMaster: but this is host22:59
persiaOh, that's host.  Yeah, 60831223:00
GrueMasterIt does switch from b_idle to a_idle.23:00
rsalvetiit should for a_host23:00
GrueMasterI have not tested gadget stuff.23:00
rsalvetiyep, never used :-)23:00
persiaIt's the connection of a [A <-> mini_B] cable between some host and the device that tests gadget.23:00
rsalvetijust with my n900, but no more than that23:00
GrueMasterLast time I tested USB OTG was at Intel working on Moblin 1.0.23:01
* persia used to use it all the time for usb0 networking, but hasn't played in a couple years23:01
* GrueMaster is kinda rusty.23:01
GrueMasterI have nothing that supports it.23:01
persiaGrueMaster, For Moblin 1.0, we never got gadget working acceptably at all.23:01
GrueMasterI know.23:02
GrueMasterSCH driver issues.23:02
* rsalveti calls it a day, will be back in some hours23:02
persiaOh, was that it?  I remember arguing about lack of software support for various blue-sky use cases, and didn't realise that either test cases were created, or concrete problems existed at the kernel layer.23:03
GrueMasterIt would work somewhat with a Windows PC on the other end.23:03
GrueMasterBut it was unreliable at best.23:03
GrueMasterThat was when usb_gadget support was in its infancy.23:05
persiaHrm?  The userspace stuff hasn't seemed to get much innovation since 2001 or so, and was working fine then.  Have there been improvements kernelside since?23:17
GrueMasterI would assume so.  I really don't know.  It may have just been an SCH issue that I was working with back then.23:18
GrueMasterI have never tested OTG on arm.23:18
ogra_cmpcapw, fyi, build survives the header install23:28
GrueMasterSo, does this mean I'll get new images for omap4 soon?23:32
=== lag is now known as Guest39359
persiaGrueMaster, Needs someone to respin an image.  You might ask in -release in case any of those with access to the button are around.23:42
persiaOtherwise it will automatically get pushed in about 2 hours23:43
GrueMasterAt this point, I can wait.  I have other things to look at until then.23:43
GrueMasterBesides, it is almost 4pm my time.  EOD is coming soon.23:44

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!