**** BEGIN LOGGING AT Mon Mar 05 02:59:58 2012 Mar 05 07:31:59 janimo`: I'll be uploading chromium in a couple hours, it would be nice to enable armhf properly :) Mar 05 08:18:20 micahg, well I do not have a proper patch, just handedited the files in question Mar 05 08:18:44 janimo`: can you diff them for me against what's in precise? Mar 05 08:19:20 micahg, ok. I was not sure if a sed call in rules would be ok or a more elegant way that intriduces new variables in gyp Mar 05 08:19:40 janimo`: we usually just patch the files ;) Mar 05 08:20:38 micahg, I am not sure if it was chromium (maybe thunderbird) that made me not want to work with it much. IIRC it did not work with edit-patch but some custom quilt paths and runes that were not documented Mar 05 08:21:33 janimo`: yes, you need to be in build-tree/src for chromium Mar 05 08:22:15 * micahg would love to switch to straight source format 3 and dh7, but that would require time :) Mar 05 08:25:25 micahg, that would be awesome. It would be time gained in the long run Mar 05 08:32:31 micahg, what rule does only the unpacking for chromium? Mar 05 08:32:38 * janimo` tries to come up with a cleanish patch Mar 05 08:33:14 janimo`: debian/rules patch I think Mar 05 08:35:21 Laney, around? Mar 05 08:35:36 oh, did you fix ghc? Mar 05 08:36:02 micahg, yes Mar 05 08:36:18 but just like chromium. Fixed the build, have no idea what the 'proper' upstream patch should be Mar 05 08:36:58 could be a sed in debian/rules if armhf, of some haskell code + config stuff in another unusual build system Mar 05 08:37:04 s/of/or/ Mar 05 08:37:40 I think it needs a normal patch though, upstream assumes softpf and has no config for armhf at all Mar 05 08:43:29 janimo`: nice work! Can you show me the patch? I'll show the ghcarm people what was needed and they should be able to take it from there Mar 05 08:43:50 Laney, just filed a bug in ghc trac. A moment Mar 05 08:43:56 oh ok, nice Mar 05 08:43:58 http://hackage.haskell.org/trac/ghc/ticket/5914 Mar 05 08:44:09 are you uploading to ubuntu? Mar 05 08:44:31 * janimo` praises his habit of using a simple identical password almost everuwhere, so wehn visiting sites every couple years he does not have to scratch his head. Mar 05 08:44:35 * janimo` oopses Mar 05 08:45:02 Laney, no, it is a hack, I;ll leave the proper patch to people who know how to integrate with the ghc build system and make it conditional Mar 05 08:46:13 Laney, I could not figure out from a few minutes of using the code whether all options need to be hardcoded in haskell or can be driven exclusively from some human readable file, like settings in the root dir Mar 05 08:46:26 as this option only needs passinf fo armhf Mar 05 08:46:59 there is mk/build.mk but I am not sure exactly which options it takes Mar 05 08:47:16 Laney, another approach would be making llvm/llc default to hard float when invoked on a hard-float system, like gcc does .Not sure if that is ok with llvm people whough Mar 05 08:48:05 Laney, so if you could point this to arm/ghc people they may fix it properly Mar 05 08:48:39 ok, thanks for the work! Mar 05 08:48:58 Laney, you're welcome Mar 05 08:57:35 micahg, can you help me with the patching of chromium. Finally untarred a sample package. cd build-tree/src Mar 05 08:57:55 quilt push -a Mar 05 08:58:12 ok Mar 05 08:58:13 quilt new fix-armhf-gyp.patch Mar 05 08:59:04 ok Mar 05 09:05:34 micahg, will attach the patch shortly Mar 05 09:05:54 janimo`: thanks Mar 05 09:08:37 micahg, https://bugs.launchpad.net/ubuntu/+source/chromium-browser/+bug/943281 fingers crossed. With restarted package builds and not knowing how and what exactly the build system regenenerates and which bits of config drive what this is a best effor tpatch :) Mar 05 09:08:38 Launchpad bug 943281 in chromium-browser "armhf FTBFS in precise" [High,In progress] Mar 05 09:08:56 janimo`: thanks Mar 05 09:09:04 will look at it in a few minutes Mar 05 09:09:21 micahg, thanks Mar 05 09:09:42 at least the gyp bits i am fairly confident should work. The v8 one, may or may not Mar 05 09:10:25 I hope the TODO item - 'migrate to external v8 lib ' is not too unrealistic :) Mar 05 09:10:45 we won't do that as chromium will move too fast in the stable release Mar 05 09:12:32 TODO: remove v8 entry from TODO file :) Mar 05 09:15:50 janimo`: heh, anyways, patch looks sane and won't break armel builds, I'll upload shortly Mar 05 09:45:59 janimo`: maybe that's why it works in Debian, they use the system v8 Mar 05 09:48:08 janimo`: well, we should find out in ~1hr if it worked or not, I'm about to upload Mar 05 10:23:49 Hi. I'm having some troubles with a custom natty armel installation (on custom HW). One particular device isn't booting properly, while others are. I'm debugging upstart and it seems it locks up in mountall, i.e. it doesn't emit the correct signals for upstart to continue into runlevel/rc-sysv modes. Mar 05 10:24:35 Mountall --verbose show that is indeed stops in mountall, but no errors on the console. How can I debug this? Any ideas? Mar 05 10:28:42 sounds like a kernel issue to me Mar 05 10:34:13 janimo`: armhf seems happy, it's building Mar 05 10:48:07 ogra_: The device had an old HW clock setting. So from what I can see mountall (not fsck) halt if the HW clock is older than the superblock of the fs. If I set the hwclock to newer, it works. Is this expected or a bug? Mar 05 10:48:49 well, if you use an ubuntu initramfs you can easily work around this by using the ficrtc cmdline option Mar 05 10:49:02 it will force the RTC to the past mount time of the disk Mar 05 10:49:13 a kernel parameter? Mar 05 10:49:33 micahg, I still suspect it may fail later on when it tries to link v8 against it all. So in case it does not figure out what the eabi should be another bit in debian/rules may be in order Mar 05 10:50:18 fixrtc I suppose Mar 05 10:51:11 yeah Mar 05 10:51:15 (to both) Mar 05 10:53:18 micahg, amended the bug with a comment in case it fails because of V8 being built without HF Mar 05 10:53:46 janimo`: well, the v8 parts compiled, we'll see if it links later Mar 05 10:54:02 ok. Mar 05 10:54:51 ogra_: It works. Thank you very much! Mar 05 10:55:00 ah qtwebkit-source keeps being ftbfs. But that looks nastier as if it was a kernel/glibc/toolchain issue Mar 05 11:51:17 janimo`: "VFPv3D16" does not imply armhf? Mar 05 11:51:47 Laney, not necessarily afaik. We had that for armel as well Mar 05 11:51:55 it means use a certain set of VFP registers Mar 05 11:52:09 but even in softfp you could use those as long as they do not cross proc call boundaries Mar 05 11:52:11 so not part of the ABI Mar 05 11:52:46 so we only want to use registers D0-D15 and not D16-D31 as found on some chips. Mar 05 11:53:23 so generate code for that VFP config for both armel and amrhf. For armhf use said regs also for passing arguments and return values between calls Mar 05 11:55:24 Laney, we've been passing -mfpu=vfp3-d16 for a while in Ubuntu for armel Mar 05 11:56:28 okey dokey, just thinking how ghc could tell if armhf or not Mar 05 11:57:35 Laney, if it has the equivalent of autogen, it may discover it by seeing gnueabihf in the triplet instead of plain gnueabi Mar 05 11:57:56 if not, it may be good enough to just pass a config option from debian/rules Mar 05 12:00:32 would be nice to be able to drive some options outside haskell code. I see settings.in but am not sure it has sections for passing options to LLVM, as it has for gcc Mar 05 12:01:39 Laney: you can use #ifdef __ARM_PCS_VFP Mar 05 12:01:55 to check for hard-float calling convention Mar 05 12:09:28 xranby, not in haskell code I assume Mar 05 12:10:08 maybe.. depends on if you pass the haskell code through the gcc preprocessor or not Mar 05 12:10:33 xranby, there is a preproc step, not sure if it is gcc's Mar 05 12:14:31 you can ask it to AFAIK Mar 05 12:14:56 some parts of GHC are written in c.. if the arch version detection are done in c code then you can use it Mar 05 12:44:36 janimo`: yep, you called it, linking with v8 broke :) Mar 05 12:46:01 micahg, ok, so the snippet I appended to the bug should help Mar 05 12:46:39 or maybe not , this build system is crazy. I know I got it working by setting that variable to 1 in the sconscript Mar 05 12:47:25 if you know that debian/rules can actually pass that v8_armeabi var reliably to SConscript it would be helpful Mar 05 12:47:41 janimo`: that I don't know Scons is pretty greek to me Mar 05 12:48:57 but, if there's a gyp flag, that's likely to work Mar 05 12:49:14 the script itself is legible, but wherever it is hooked in with gyp and the rest is a mistery to me Mar 05 12:50:31 janimo`: so, I'll drop your scons change and use the gyp flag, how does that sound? Mar 05 12:50:49 sure Mar 05 12:51:17 janimo`: is this worth doing now, or should I wait for the next stable update? Mar 05 12:51:32 although that SCons thing looks suspicious. Why there's no EABI setting like in the other cases simialr in that file - all come in threes soft/softpf/hard in 2 places, buit this one where only soft is specified Mar 05 12:51:37 micahg, next stable Mar 05 12:51:45 ok, thanks Mar 05 12:51:48 progress has been made with this release too :) Mar 05 12:52:02 yes, indeed :) Mar 05 12:52:16 is next stable in 6 weeks? Or do you have intermediate uploads too sometimes? Mar 05 12:52:29 either way, next upload, wherever it is Mar 05 12:52:40 yeah, there are intermediate uploads whenever chromium feels like releasing it's cache of CVE fixes Mar 05 12:52:47 I'll actually do a test run these days, let it build to have a tested fix Mar 05 12:52:51 s/it's/its Mar 05 12:52:58 soon, until I do not forget how to quilt patch the tree Mar 05 12:53:25 * micahg will start using the word attempt with fixing armhf FTBFS :) Mar 05 12:53:58 exaclty - I have been cautious enough to use attempt in many changelogs and bug comments so far :) Mar 05 12:54:13 or the more optimistic - making the build progress Mar 05 14:18:46 /etc/init/ttyS*.conf is set to start on stopped rc, which is rather late in the boot process, which can prevent login if booting fails. Would there be any problems starting getty very early, like on startup (upstart event)? Mar 05 14:19:49 I did this for debugging, and it seems ok. Of cource you are warned of ro rootfs, but that's fine as long as you get a login. Now I'm wondering if we should ship this config permanently Mar 05 14:26:08 Where does the logs from e2fsck run by mountall end up on a natty system? Mar 05 14:53:02 Is it possible to install deb package for armel that depends on other armel packages on armhf system? Mar 05 14:53:41 I tried dpkg -i --foreign-architecture armel but it still complains about missing package despite it's installed Mar 05 15:00:31 GrueMaster, ogra_: oneiric armel images boot fine on my panda board, precise armhf images do not Mar 05 15:00:51 oh ? intresting Mar 05 15:00:55 whats the error ? Mar 05 15:00:55 is that something in the pandaboard hardware that can't do armhf, or is it my low power I've giving the board? Mar 05 15:01:07 ogra_: green LED is the error Mar 05 15:01:34 i mean on your serial console :) Mar 05 15:01:47 I don't have a serial console Mar 05 15:02:28 for development it really helps to have one Mar 05 15:02:42 ogra_: what do I need to buy? Mar 05 15:03:26 an usb to serial adapter, try to get one you can attch directly to the pandas serial connector (female i think) Mar 05 15:15:44 is their any way to CC entire deb packages then transfer the binaries to the host machine? Mar 05 15:19:04 target, sry Mar 05 15:23:13 smplman: Define "cc entire deb packages". Mar 05 15:24:05 smplman: If you mean "attempt to reconstruct it based on what's currently on the filesystem", you want dpkg-repack. But there's no guarantee that's going to give you a sane package. Mar 05 15:24:23 smplman: Usually, you just want to install the same deb everywhere, and then customise your configs accordingly. Mar 05 15:25:04 say i want to install the nano.dpkg on my arm system, but i want to cross compile if first on my x86 system so its faster Mar 05 15:25:19 then transfer the binary over Mar 05 15:26:12 i realize packages are usually platform independent, but i think its probably faster to cc my packages then transfer them over Mar 05 15:26:32 if this is crazy talk just let me know Mar 05 15:26:49 i know how to CC normal source, then transfer the binary Mar 05 15:27:02 im more concerned with ubuntu packaged now Mar 05 15:27:55 i meant *.deb earlier not *.dpkg Mar 05 15:29:09 infinity: if that makes sense Mar 05 15:30:22 smplman: I'm still lost by what you mean by "CC the package". Mar 05 15:30:27 Unless you mean compile? Mar 05 15:30:58 And packages are most certainly not platform independant. Mar 05 15:35:09 cross compile the package on my x86 machine then transfer it to my arm machine Mar 05 15:35:56 Yeah, it's sometimes doable. But not often. And we're still working on making it work better. Mar 05 15:36:22 Anything documented yet? Mar 05 15:36:27 You're really just better off building the package natively, unless you're ready to do a lot of reading on dpkg-cross and the like (and prepared to curse a lot when your use-case fails). Mar 05 15:36:45 Well, all the dpkg-cross stuff is documented. But also in the process of being deprecated for better and shinier methods. Mar 05 15:36:49 its for personal dev so not that big a deal Mar 05 15:37:11 I really do just recommend building natively for most people. Mar 05 15:37:18 i have 11.10 running on both my xoom and beagleboard xm Mar 05 15:37:27 The special case to that rule being people who build a lot of kernels, since that's all set up to "just work" with a cross-compiler installed. Mar 05 15:37:42 yea im good with building kernels Mar 05 15:37:54 it just takes forever to install packages on my xm Mar 05 15:38:08 For most other packages, the odds of it working, and the odds of you wanting to throw things are high enough that building natively is just easier. Mar 05 15:38:50 or just build from source vs using the .deb? Mar 05 15:39:07 I don't see how that changes anything... Mar 05 15:39:20 Wait. Are you talking specifically about dpkg itself taking a long time? Mar 05 15:39:24 There's zero ways around that. Mar 05 15:39:51 I thought you meant building the deb on another system, then copying it over and installing it. Mar 05 15:40:04 There's no sane way to avoid the part where you invoke dpkg. Mar 05 15:58:47 ogra_: You've mentioned ficrtc boot option earlier today. But I can't find docs about it. Was it some typo? Mar 05 16:01:44 int_ua: It is fixrtc, a simple shell script that is part of initramfs-tools. Mar 05 16:03:26 GrueMaster: Thanks Mar 05 16:19:54 are the sudden switch offs i'm seeing on the pandaboard with this weekends daily expected? Mar 05 16:23:35 Hi all, is it possible to run the 12.04 omap4 image under qemu? Mar 05 16:23:45 no Mar 05 16:24:01 ogra_, is there any image that I can run under qemu? Mar 05 16:24:34 the omap3 image might run in the beagle emu Mar 05 16:24:43 qemu simply has no emu for omap4 at all Mar 05 16:27:04 thanks ... are imx images supposed to be running in qemu then? Mar 05 16:27:15 no Mar 05 16:27:41 there is no qemu emulation for any of the images we have apart from omap3 Mar 05 16:27:48 tvoss, qemu is not a platform as such Mar 05 16:28:17 you can roll something yourself based on the vexpress kernel and a self rolled rootfs (or ubuntu-core) Mar 05 16:29:18 ppisati: ping - not sure if you noticed, but I figured out why I didn't have sound & networking on the beaglexm. Mar 05 16:30:13 GrueMaster: network works here Mar 05 16:30:19 GrueMaster: dunno about audio Mar 05 16:30:43 GrueMaster: on the othert hand, while updating this morning i noticed latest kernel hangs before mounting root Mar 05 16:30:48 GrueMaster: did you try it? Mar 05 16:30:55 3.2.0-18.28 Mar 05 16:31:02 I have had spotty network at best. I usually have to unplug, plug in the network cable several times to get it working. Mar 05 16:31:14 GrueMaster: works immediately here Mar 05 16:31:18 rev c and rev a Mar 05 16:31:25 I am just waking up. Haven't even really started anything yet. Mar 05 16:31:28 k Mar 05 16:31:45 GrueMaster: when you are caffeinated enought, tell me which kerbnel version were you running Mar 05 16:31:54 I have a rev b beaglexm. It has had these issues for 2 cycles now. Mar 05 16:32:26 3.2? Mar 05 16:32:29 I documented everything in bug 925094 and bug 838200. Mar 05 16:32:31 Launchpad bug 925094 in linux "No audio on omap (beagleXM) system" [Medium,Confirmed] https://launchpad.net/bugs/925094 Mar 05 16:32:32 Launchpad bug 838200 in u-boot-linaro "No network support on Beagle XM" [High,Confirmed] https://launchpad.net/bugs/838200 Mar 05 16:32:54 3.2.0.18 iirc. I pulled from git. Mar 05 16:33:43 latest one is 3.2.0-18.28 Mar 05 16:33:49 but it seems it's broken Mar 05 16:33:59 previous one -17.27 should be ok Mar 05 16:39:18 Well, it worked in my build of it. Must be a patch since Friday? Mar 05 16:45:19 GrueMaster: i'm booting off the sd FWIW Mar 05 16:45:51 Same here. I just added my kernel to the Beta 1 image. Mar 05 16:46:00 GrueMaster: im running 3.2.9 from https://github.com/RobertCNelson/stable-kernel , good results Mar 05 16:46:35 on my xm rev b ATM Mar 05 16:47:32 smplman: We are using the mainline kernel, not a hacked up version. We really want to be as close to main as possible. Eventually, the omap4 kernel will also be done this way. Mar 05 16:48:03 you wish :) Mar 05 16:48:13 I am well aware of other working kernels. The test image that ships with the XM works perfectly, but not all of it is upstream. Mar 05 16:48:25 ogra_: well, I did say eventually. Mar 05 16:48:38 GrueMaster: I understand, just trying to help Mar 05 16:48:41 yeah, it will take years until the DSS code goes upstream Mar 05 16:48:50 janimo, ogra: can you endorse me on my motu application? https://wiki.ubuntu.com/MarcinJuszkiewicz/DeveloperApplication-MOTU Mar 05 16:58:49 ogra_, GrueMaster: ok here is loading with oneiric successfully and with precise unsuccessfully http://starsky.19inch.net/~jr/tmp/arm/ Mar 05 16:58:55 there's not much in the diff Mar 05 16:59:08 could it be just not showing on the monitor in precise? Mar 05 16:59:13 I have a full power supply now Mar 05 17:00:15 looks like the bootloader runs fine Mar 05 17:00:35 Riddell: omapfb missing maby? Mar 05 17:00:51 smplman: I have no idea what that is Mar 05 17:01:15 smplman, thats compiled in Mar 05 17:01:36 Riddell, it would help to get the kernel bootmessages (which we supress by default) Mar 05 17:01:47 ogra_: how can I do that? Mar 05 17:02:12 Riddell, open the boot.scr from the first partition of the SD card in vi Mar 05 17:02:28 delete the garbage (u-boot header) in the beginning Mar 05 17:02:48 then edit the cmdline in that file and save it as boot.script Mar 05 17:02:59 Riddell: what image are you using? Mar 05 17:03:16 smplman: ubuntu desktop precise beta 1 Mar 05 17:03:18 omap4 Mar 05 17:03:19 ppisati: I am uploading my kernel .deb to http://people.canonical.com:~tobin/linux-image-3.2.0-18-omap_3.2.0-18.28_armhf.deb for you to try. Should be available in ~12 minutes. Mar 05 17:03:45 Riddell, you will then need to install u-boot-tools to get the mkimage command Mar 05 17:04:10 is omap4 backwards compatible with omap3? Mar 05 17:04:14 Actually, save it back as Env.txt and reboot. Newer u-boot should be able to take it from there. Mar 05 17:04:32 ogra_: Riddell ^^^ Mar 05 17:05:03 oh, i didnt know we got that bit from zupstream working nowadays Mar 05 17:05:23 smplman, no Mar 05 17:05:34 different arches Mar 05 17:05:40 ogra_: the whole of the first line is garbage? Mar 05 17:06:20 i know the xm is omap3 Mar 05 17:06:27 well, the stuff vim doesnt show properly :) Mar 05 17:06:41 GrueMaster: so don't save it as boot.scr but as Env.txt ? Mar 05 17:06:50 Riddell, right Mar 05 17:06:59 and put it on the first partiton of the SD Mar 05 17:07:13 ogra_: and what do I need to do with mkimage? Mar 05 17:07:16 not sure, you might need to remove boot.scr from that partition Mar 05 17:07:31 Riddell, if Env.txt works you dont need it at all Mar 05 17:07:42 Yes. Strip the u-boot header (garbage at the beginning), then remove the quiet & splash lines on the bootargs and add "console=ttyO2,115200 console=tty0". Mar 05 17:07:49 else you would need it to add the header back to the file Mar 05 17:08:52 "[ 5.272796] panic occurred, switching back to text console" ouch Mar 05 17:08:59 hrw, done btw Mar 05 17:09:00 how can I get minicom to save to a file? Mar 05 17:09:20 ctrl-a+h Mar 05 17:09:41 ogra_: that saves it? Mar 05 17:09:43 that should give you the list of commands ... there is some way to use a logfile Mar 05 17:10:01 that hangs up Mar 05 17:10:04 i never used logging, i just scoll back in my terminal in screen and copy/paste that Mar 05 17:11:03 this is giving me a bit too much output to do that Mar 05 17:11:50 Riddell: Use "screen /dev/ttyUSB0 (or shat ever the serial port on your pc is), then use screen's logging capabilities. Mar 05 17:12:06 yeah Mar 05 17:12:50 aah Mar 05 17:18:28 ppisati: My kernel is up if you want to test it. Mar 05 17:19:17 GrueMaster, ogra_ http://starsky.19inch.net/~jr/tmp/arm/ubuntu-precisebeta1-armhf-noquiet Mar 05 17:20:06 [ 4.463867] EXT3-fs (mmcblk0p2): error: couldn't mount because of unsupported optional features (240) ? Mar 05 17:20:27 odd indeed. Mar 05 17:21:01 Did we misplace ext4 support somewhere? :P Mar 05 17:21:11 I'm assuming 240 is an ext4 feature flag. Mar 05 17:21:12 * infinity looks. Mar 05 17:21:26 it just loops over the ext filesystems usually Mar 05 17:21:33 thats a red herring Mar 05 17:21:47 Yeah, 240 is an ext4 feature. Mar 05 17:22:35 Something doesn't sound right with his image. I tested all (and I do mean ALL) images for beta 1 on multiple systems. Mar 05 17:22:36 it should be trying ext4 next ... not sure why it doesnt Mar 05 17:22:39 Riddell: Is that a fresh precise install, or something hacked up and/or upgraded? Mar 05 17:23:01 infinity: installed like this sudo sh -c 'zcat ./ubuntu-12.04-beta1-preinstalled-desktop-armhf+omap4.img.gz |dd bs=4M of=/dev/mmcblk0 ; sync' Mar 05 17:23:17 which works fine for oneiric Mar 05 17:23:21 did you let it finish the resize on first boot ? Mar 05 17:23:33 ogra_: I don't know, I had no output Mar 05 17:23:39 if it was interrupted it indeed might corrupt your fs Mar 05 17:23:49 You have no monitor connected? Mar 05 17:24:02 infinity: I do have a monitor connected. it does not show anything with precise Mar 05 17:24:07 Wrong port? Mar 05 17:24:16 Is it HDMI or DVI? Mar 05 17:24:21 (Yes, people mess this up all the time) Mar 05 17:24:28 The HDMI port is the one next to the ethernet. Mar 05 17:24:29 Do you have the monitor on the HDMI port (closest to the USB ports). Mar 05 17:24:35 ie: Not the one on the outside of the board. Mar 05 17:24:50 I have a DVI monitor plugged into the DVI port which worked fine with oneiric Mar 05 17:25:08 might have changed with the new kernel version Mar 05 17:25:11 try HDMI :) Mar 05 17:25:16 Ahh, kay. Something may have broken DVI. I only ever test with HDMI (cause it's all I have the cabling for) Mar 05 17:25:21 I have no HDMI monitor Mar 05 17:25:28 HDMI TV? Mar 05 17:25:32 just plug the DVI into the HDMI port Mar 05 17:25:35 I have no TV Mar 05 17:25:41 should work fine Mar 05 17:25:43 ogra_: That really shouldn't work. Mar 05 17:25:50 ogra_: But I'd be curious to see if it does. Mar 05 17:26:11 or was it the orther way round ? Mar 05 17:26:29 it wont do any harm either way though Mar 05 17:26:35 True. Mar 05 17:26:52 infinity: Yes, it should. hdmi is the same signal as DVI-D. Mar 05 17:27:07 Riddell: Also, despite all protests to the contrary, you can just zcat foo.img.gz > /dev/whatever Mar 05 17:27:13 Riddell: The dd is entirely pointless. Mar 05 17:27:15 right, i thought so Mar 05 17:27:45 GrueMaster: Yes, HDMI is a superset of DVI-D, but it also intentionally leaves out (in many implementations) lots of the possible resolutions. Mar 05 17:27:46 infinity, using a 4M blocksize speeds up the writing Mar 05 17:28:09 GrueMaster: If that's the case on the omap4 kernel (or at the hardware level), I have no idea. Mar 05 17:28:52 I don't know if the dvi port on panda will work w/o the TI drivers (not in FB w/o kernel parameters). Mar 05 17:28:56 i know i used the HDMI port of my monitor with a HDMI cable plugged into the DVI port in the past ... Mar 05 17:29:03 But signal wise, an HDMI port can connect to a DVI monitor just fine. Mar 05 17:29:12 but i dont think i ever tried the other way round Mar 05 17:29:22 The other way around really shouldn't work. Mar 05 17:29:37 But HDMI->DVI will work, if it supports the resolutions you need. Mar 05 17:29:45 The only thing that will be missing is HDMI audio. Mar 05 17:29:49 The audio channels just go off into la-la land. Mar 05 17:30:09 And DVI->HDMI works the same way. Mar 05 17:30:33 Anyhow. Mar 05 17:30:41 Riddell: Using the HDMI port may well make things happy. Mar 05 17:30:47 I gathered :) Mar 05 17:30:56 remaking the image on the SD card Mar 05 17:30:57 Or, we could discuss it more. :P Mar 05 17:30:57 As long as it is all digital. DVI also supports analog (but that is a different discussion and irrelevant to this discussion). Mar 05 17:31:21 infinity, awesome idea ! Mar 05 17:36:17 Riddell: I relooked at your kernel output (above). Why is "rootfstype=ext3" in the kernel cmdline? Mar 05 17:36:37 That isn't normal in our images. Mar 05 17:39:42 hey ubuntu-arm, ld is complaining that 'main.o uses VFP registers, a.out does not' Mar 05 17:40:16 i'm compiling main.cpp with -mfpu=vfp + other args. how can i convince ld that a.out should use VFP registers also? Mar 05 17:47:25 GrueMaster: I have no idea Mar 05 17:47:40 I just used a normal image with the edited boot command edited to remove quiet Mar 05 17:48:24 Well, since we use ext4 images, that line would have given you issues. Mar 05 18:02:53 Riddell: Was this the Ubuntu beta 1 or Kubuntu Beta 1? Mar 05 18:05:05 GrueMaster: Ubuntu Desktop Beta 1 Mar 05 18:06:10 Ok. Checking here, I see no indications of where "rootfstype=ext3" could have come from, unless it is in u-boot and you didn't use the boot.scr. Mar 05 18:06:50 Your console log above was started after u-boot had booted. Mar 05 18:07:21 GrueMaster: I didn't use the boot.scr after I saved it to Env.txt Mar 05 18:07:58 Did it load the Env.txt? Mar 05 18:08:29 dunno reformatting the SD card and trying again Mar 05 18:08:55 Oh, wait. I think I see what is happening. u-boot only looks for the uImage & uInitrd lines in Env.txt. I was told it works the same as boot.scr. Mar 05 18:10:28 So nix the Env.txt idea. Mar 05 18:11:25 At any rate, you should have video on hdmi. Mar 05 18:16:04 damian0815: That sounds like you're using -mfloat-abi=hard on an armel system. Mar 05 18:16:39 damian0815: Which will only work if you link with gcc as well (or pass the right bits to ld), since ld won't magically do what you think it should. Mar 05 18:16:48 damian0815: Alternately, run armhf if you want armhf. ;) Mar 05 18:17:03 infinity: thanks for the reply Mar 05 18:17:13 infinity: i am in fact linking by calling gcc Mar 05 18:17:36 infinity: i guess gcc isn't telling ld to do the right thing? Mar 05 18:17:45 damian0815: On which release is this? Mar 05 18:17:52 infinity: 10.10 Mar 05 18:17:55 Oh, yeah. Mar 05 18:18:04 That's almost certain to just plain not work. Mar 05 18:18:08 rightoh Mar 05 18:18:14 what is this install RELEASE shortcut that's started showing up on my shortcut bar (and how do i get rid of it?) Mar 05 18:18:26 clicking on it just gives me a crash in ubiquity Mar 05 18:18:30 i'm installing 11.10 as we speak. was avoiding before due to lack of omap-extras-graphics. Mar 05 18:18:35 Besides, even if you compiled your application as hardfloat, if it links to anything else, it won't work. Mar 05 18:18:39 (it's still rotating circle on collecting info) Mar 05 18:18:49 damian0815: Install precise/armhf if you want hard-float. Mar 05 18:19:00 damian0815: You'll like it better than oneiric anyway. Mar 05 18:19:06 Riddell: The ubuntu beta 1 image works for me on a non-hdmi DVI monitor here, so you "should" have no issues. If you still do, we can figure it out. Mar 05 18:19:12 GrueMaster: http://starsky.19inch.net/~jr/tmp/arm/ubuntu-precisebeta1-armhf-using-hdmi Mar 05 18:19:27 that's ubuntu beta 1 with a dvi monitor plugged into the HDMI port Mar 05 18:19:43 I'm awaiting instructions :) Mar 05 18:19:57 infinity: what's the status of SGX drivers in precise? Mar 05 18:20:39 er, this is interesting. so ubiquity is crashing with IOError in main() Errno 2, No such file or directory: '/proc/2058/oom_score_adj' Mar 05 18:20:53 why would ubiquity crash because oom_score_adj doesn't exist? Mar 05 18:21:09 Riddell: This is normal so far. You see nothing on screen? Mar 05 18:21:17 GrueMaster: nothing Mar 05 18:21:18 (not serial console, but on the monitor). Mar 05 18:21:18 damian0815: As I understand it, it's all working in the tiomap-dev/trunk PPA. Mar 05 18:21:30 And you have a proper power supply now? Mar 05 18:21:34 GrueMaster: I do Mar 05 18:21:35 damian0815: But depends on installing the 3.1 kernel from the oneiric PPA. Mar 05 18:21:42 damian0815: (That will be resolved soon) Mar 05 18:22:06 infinity: super. i'll see how i go with oneiric. i'm trying to build a straightforward enduser experience for openFrameworks (C++ art+code toolkit) Mar 05 18:22:23 Monitor is using an HDMI->DVI cable and no other conversions? Mar 05 18:22:55 damian0815: Sure. oneiric should work fine. But just stop trying to pass whacky float options to the compiler. Stick with the defaults and your life won't be pain. ;) Mar 05 18:23:22 damian0815: (But, like I said, precise/armhf will treat you better than oneiric/armel, except for the bit where it's not quite done yet) Mar 05 18:23:29 infinity: heh. thx. Mar 05 18:23:52 Riddell: Monitor is using an HDMI->DVI cable and no other conversions? Mar 05 18:24:04 steev_: Is /proc mounted at all? Mar 05 18:24:27 GrueMaster: HDMI-DVI adaptor and DBI cable to monitor Mar 05 18:25:06 Ok. Wanted to make sure there were no DVI->VGA adapters in the mix. Mar 05 18:25:33 Do you have another device with HDMI out that you know works? Mar 05 18:27:07 It is also possible that your cable only handles DVI-A (VGA), although those are rare. Mar 05 18:27:30 GrueMaster: I have no HDMI devices no Mar 05 18:27:47 No PS3 or XBox? Mar 05 18:27:51 no Mar 05 18:27:53 GrueMaster: this setup works fine for oneiric Mar 05 18:28:18 The oneiric image works? Now that is odd. Mar 05 18:29:22 yes Mar 05 18:30:51 Out of curiosity, what rev board is this? Mar 05 18:32:54 GrueMaster: how do I find out? Mar 05 18:33:46 Sticker on the bottom? Mar 05 18:34:32 GrueMaster: Rev A1 Mar 05 18:35:20 Panda or PandaES? Mar 05 18:36:09 I'll look at my stacks and see which is an A1 so I can try to reproduce it here. Mar 05 18:37:13 GrueMaster: Model: PandaBoard Rev A1 Mar 05 18:38:16 Ok. I have one (I have one of each and then some). I'll try it here. Works fine on PandaES and Panda A3 so far. Mar 05 18:38:26 infinity: ping Mar 05 18:39:12 infinity: https://bugs.launchpad.net/ubuntu/+source/qtwebkit-source/+bug/945393 << Any ideas how to fix? (/usr/bin/ld: final link failed: Memory exhausted) Mar 05 18:39:13 Launchpad bug 945393 in qtwebkit-source "qtwebkit-source version 2.2.1-1ubuntu2 failed to build on armhf" [High,Confirmed] Mar 05 18:44:55 * micahg wonders if it's worth filing a toolchain bug for https://launchpadlibrarian.net/95446439/buildlog_ubuntu-oneiric-armel.chromium-browser_17.0.963.65~r124586-0ubuntu0.11.10.1_FAILEDTOBUILD.txt.gz Mar 05 18:48:47 Riddell: I can't seem to reproduce your issue here at all. Give me a minute and I will send you a replacement boot.scr for debugging. Mar 05 18:49:36 GrueMaster: thanks Mar 05 18:51:10 Did it get through the first boot ok? Easiest way to check is to look at the sd card in your desktop and see if the second partition fills the SD. Mar 05 18:52:17 GrueMaster: I think it did but I've since reformatted it ready for your new boot.scr Mar 05 18:53:58 Ok. One sec then. Mar 05 18:55:23 I have to flash it fresh to an SD then modify it. Takes the same time as uncompressing/loop mounting. Mar 05 19:00:38 shadeslayer: It's on my TODO, but no, I don't have any immediate ideas right now. I have a few things on the go. Mar 05 19:01:45 micahg: I wouldn't bother, personally. Mar 05 19:02:20 Riddell: http://people.canonical.com:~tobin/boot.scr Mar 05 19:02:46 Download it and copy directly to partition 1 (fat partition) overwriting existing boot.scr. Mar 05 19:02:54 infinity: ok, well, if there isn't a way to work around it then, Chromium on arm in the stable releases looks slim Mar 05 19:03:01 And make sure screen is set to log before powering on the panda. Mar 05 19:03:34 I need food, will return shortly. Pastebin the console log (or save it somewhere). Mar 05 19:03:44 micahg: I'm much more interested in the LTS anyway, but if we have a spare moment, we can look at the oneiric FTBFS. Mar 05 19:04:32 well, we're 2 hours (build time) closer to havign armhf work, armel already works in precise Mar 05 19:04:42 GrueMaster: what changed? Mar 05 19:08:20 GrueMaster: when I run screen /dev/ttyUSB1 I just get junk output and it causes minicom to get junk output to Mar 05 19:11:39 Works fine here. Mar 05 19:12:54 ogra_: not at cebit are you? Mar 05 19:13:20 GrueMaster: which u-boot/x-loader version do you have with beaglexm? and which cmdline do you use? Mar 05 19:14:00 Riddell, nope, 150km south of it Mar 05 19:14:59 ogra_: you couldn't ship yourself and your office to hannover or Bielefeld for tue/wed/thu could you, just to help me? :) Mar 05 19:16:32 Riddell, hmm not really, what are you doing at CeBIT ? Mar 05 19:17:27 see cebit Mar 05 19:17:31 seeing cebit Mar 05 19:17:34 I've never seen it before Mar 05 19:17:40 ah Mar 05 19:17:52 i have been to each since i was 16 Mar 05 19:18:00 but not since i work for canonical Mar 05 19:18:16 <- grew up in hannover Mar 05 19:18:30 too many travelling days for UDS et al? :) Mar 05 19:19:09 nah, just no interest, i moved south a few years before starting ubuntu stuff, havent been back often Mar 05 19:19:58 if i go, i just do it to visitn my parents, but thats not more than a few hours (though i love hannover, but all friends i had there moved to berlin or elsewhere) Mar 05 19:22:27 * ogra_ is off for the evening Mar 05 19:31:37 GrueMaster: http://starsky.19inch.net/~jr/tmp/arm/ubuntu-precisebeta1-armhf-using-tobiins-boot.scr Mar 05 19:35:29 GrueMaster: now I'm getting a very jumbled login prompt http://starsky.19inch.net/~jr/tmp/arm/ubuntu-precisebeta1-armhf-using-tobiins-boot.scr-login Mar 05 19:35:34 from the serial Mar 05 19:35:38 nothing on monitor Mar 05 19:36:34 hello Mar 05 19:37:38 can anyone direct me to any ubuntu ARM's that might be able to run on the new rasberrypi's? Mar 05 19:40:45 the rasberrypi is an arm11 "armv6" core, ubuntu's current minimal arch for supported dist is "armv7-a".... Mar 05 19:40:58 "use debian" Mar 05 19:41:01 rcn-ee damn.. Mar 05 19:42:04 i guess i would've found that out had i read more on the man for the pi. i haven't bought one yet as they first B boards haven't even shipped yet.. but i heard about it 3 days ago and i'm excited and would like to get my hands on one. so i'm trying to plan my assault on the thing the moment it arrives.. Mar 05 19:42:40 Riddell i was hoping for ubuntu or slax... but i digress... Mar 05 19:44:01 i know there is bt ARMS for android phones.. but obviously an HTC incredible is much more powerful than a pi and i think it uses v11? idk.. Mar 05 19:46:10 the htc incredible is a QSD8650, armv7-a core.. Mar 05 19:47:09 yea.. Mar 05 19:47:29 ARM cortex-A15 cpu Mar 05 19:48:16 for my specific phone.. using the armv7 arch.. anyhow. i know my phone HAS the capability to handle 720p/1080p but is not enabled by default? Mar 05 19:48:22 ah, no.. currently no "cortex-A15" cores are shipping yet.. But shorty.. ;) Mar 05 19:48:26 i need to root and crack this baby open Mar 05 19:48:49 usually. 720p/1080p video is handled by a video coprocessor.. Mar 05 19:49:08 davidhayesbigtoe: you can boot debian tho Mar 05 19:49:12 (the rasperby pi's core has the same feature, that's why they can show it doing video..) Mar 05 19:49:25 blah.. i'm dumb.. the successor to the current scorpion chip i have.. the S4 SoC has the cortex cores Mar 05 19:49:36 also, I'm waiting for my RasPi as well Mar 05 19:49:40 rcn-ee yea, i know that's why i want a pi Mar 05 19:49:41 lol Mar 05 19:49:58 but i also wanna fully crack open my dINC1 now that my contract is up and i'm do for a new phone Mar 05 19:50:25 * shadeslayer has a voucher code, but no way to redeem the voucher Mar 05 19:50:31 i wanna integrate the pi and the droid together.. maybe use it for touch screen display on the pi Mar 05 19:50:33 hah Mar 05 19:52:32 http://lists.qt-project.org/pipermail/qtonpi/2012-March/000215.html ... :( Mar 05 19:54:37 * davidhayesbigtoe :-( Mar 05 19:54:43 i didn't even order Mar 05 19:54:54 i put myself on the mailinglist and sent an inquiry Mar 05 20:07:02 How do I install 12.04 (nightly or beta) on my pandaboard? Mar 05 20:08:22 I've booted a recent nightly and modified it to show a serial console but I dunno how to log in to it- username/password? Mar 05 20:08:57 I've tried ubuntu with no password or ubuntu as password too with no luck Mar 05 20:11:01 There don't seem to be any headless OMAP4 builds of Precise yet, which is what I'd use if I could get some Mar 05 20:13:24 danboid: s/headless/server/ Mar 05 20:13:52 http://cdimage.ubuntu.com/ubuntu-server/daily-preinstalled/current/ Mar 05 20:15:10 infinity, Aha! Thanks v much! Mar 05 21:24:33 Kernel /boot/vmlinuz-3.2.1 does not match your subarchitecture Mar 05 21:24:35 omap4, therefore not writing it to flash. Mar 05 21:24:41 got this on this morning's apt-get dist-upgrade Mar 05 21:26:18 pbuckley: And where did that kernel come from? Mar 05 21:26:24 pbuckley: It's not a distro kernel. Mar 05 21:26:51 hrmm it might have picked it up from ti's repo Mar 05 21:26:57 ill let you know once it finishes Mar 05 21:27:09 dpkg -S /boot/vmlinuz-3.2.1 Mar 05 21:28:42 And it wouldn't be from TI's PPA, they're building 3.1 Mar 05 21:28:42 nm.. i must have built it Mar 05 21:28:52 its not showing up in dpkg Mar 05 21:29:01 Yeah, I suspected as much. Mar 05 21:29:23 If you want flash-kernel to play nicely with local kernel builds, just tack an -omap4 extraver on the end of them. Mar 05 21:29:51 Cause we do machine type to subarch mapping to make sure Bad Things don't happen if someone installed a kernel package for another subarch. Mar 05 23:33:50 https://wiki.ubuntu.com/ARM/RootfsFromScratch Mar 05 23:34:20 What in God's name are you guys using to build that kernel, if I replace it with my own kernel it says rootfs not found. Is there a driver MMC perhaps you guys bake in there? Mar 05 23:34:23 :-) Mar 05 23:44:06 i just use the ubuntu tools for building a kernel Mar 05 23:44:25 let me get you a link Mar 05 23:44:58 https://help.ubuntu.com/community/Kernel/Compile Mar 05 23:47:49 pbuckley, thanks let me read that wiki, if there is a .config file you have somewhere that would be very useful too Mar 05 23:51:19 CodeWar: What platform are you building a kernel for? Mar 05 23:51:36 GrueMaster, vexpress-a9 running on QEMU Mar 05 23:52:16 I can build zImage and load run and have fun with NFS boot all I want but booting off qemu img never worked for me. It appears to work for you guys so diffing whats different :-) Mar 05 23:54:33 I know nothing about the vexpress build or running on qemu, but I believe we have a vexpress kernel in the pool. Mar 05 23:55:03 can you point me to the .config file for whatever kernel works successfully on QEMU? Mar 05 23:55:10 thats all I m looking for anyways :-) Mar 05 23:57:08 Not easily. Either pull the deb package or use the kernel git tree from the link pbuckley pasted. I think the omap kernel is what is used for qemu. Mar 05 23:59:36 hmm... not versatile? :o Mar 06 02:01:22 How choice RAM of qemu-system-arm Mar 06 02:06:14 pnphi, qemu-system-arm -m 256 -append "mem=256M ..." <-- do you want to specifiy how much ram is available? Mar 06 02:06:46 i want > 256 Mar 06 02:06:52 more ? Mar 06 02:07:10 results in segfaults, as far as i know Mar 06 02:07:23 what is segfaults ? Mar 06 02:08:48 pnphi, http://en.wikipedia.org/wiki/Segfault Mar 06 02:20:05 You could use qemu-arm-static instead and just run it in a chroot instead of with a separate arm kernel Mar 06 02:20:14 Then you'd have access to the entire host system's ram Mar 06 02:22:47 how to change Memory of qemu-system-arm ? ? more 256 ?? **** ENDING LOGGING AT Tue Mar 06 02:59:58 2012