| Xase | GrueMaster: I think I figured something out. | 01:22 |
|---|---|---|
| Xase | I think the current "NookBuntu" on XDA Developers is also using EvTouch. | 01:22 |
| Xase | And something titled tslib | 01:22 |
| === Jack87|Away is now known as Jack87 | ||
| lilstevie | Xase: is that NookBuntu using an older version of ubuntu? | 02:57 |
| lilstevie | because that seems rather overkill for natty+ | 02:57 |
| Gioyik | HI.. | 03:11 |
| Gioyik | I have a question... | 03:11 |
| Gioyik | I'm interest in use rootstock.. | 03:11 |
| Gioyik | and my question is.. | 03:12 |
| Gioyik | if i modify the OS host.. (Ubuntu) | 03:13 |
| Gioyik | with new packges and wallpapers.. | 03:13 |
| Gioyik | when i build the arm image with the rootstock.. | 03:13 |
| Gioyik | that changes apply for that build? | 03:14 |
| === Jack87 is now known as Jack87|Away | ||
| Xase | lilstevie: I thing it's actually a hybrid of ubuntu and Angstrom though not sure. | 05:55 |
| twb | Xase: what could POSSIBLY go wrong | 05:55 |
| Xase | Several parts say Angstrom... but it looks like and is advertised as ubuntu | 05:55 |
| Xase | I think I'm going to build me a root fs from scratch so I can install the deprecated evtouch... | 05:56 |
| Xase | Since there are rules for it in that nookbuntu lilstevie | 05:56 |
| Xase | Even single touch would be a step in the right direction | 05:56 |
| Xase | I think I should install Ubuntu... I'm tired of Debian =( | 05:58 |
| Xase | A lot of my packages are broken. | 05:58 |
| Xase | I converted a Crunchbang Statler/Squeeze to Wheezy. | 05:58 |
| twb | Xase: testing or- yeah, well, if you run testing you get to keep both halves | 05:58 |
| Xase | twb: I'm an idiot, can you clarify that sentence. | 05:59 |
| twb | testing breaks a lot | 05:59 |
| Xase | Ah, a joke? | 05:59 |
| twb | If you can't deal with that you're not supposed to run testing | 05:59 |
| Xase | Oh I can. | 05:59 |
| twb | Xase: yes the line is "with open source, when it breaks you get to keep both halves" | 05:59 |
| Xase | The breakage is mainly from converting. | 06:00 |
| twb | Fair enough | 06:00 |
| Xase | Testing is actually really nice... though I really wanna give Unity another go now that 11.10 is out. | 06:00 |
| Xase | Ubuntu is great... but so is Debian... but so is Linux in general. | 06:01 |
| twb | Bleh | 06:01 |
| Xase | I flip flop a lot. | 06:01 |
| Xase | Not a unity fan? | 06:01 |
| twb | I'm not a fan of much of anything ubuntu has done to debian | 06:01 |
| Xase | I've been digging gnome 3.0, but I'm still partial to open box/xfce | 06:01 |
| Xase | :p | 06:02 |
| twb | In particular upstart as at lucid routinely fucks me up the arse, no lube | 06:02 |
| Xase | lol | 06:02 |
| Xase | Not a ubuntu fan in general. | 06:02 |
| Xase | ? | 06:02 |
| Xase | s/?/./ | 06:02 |
| Xase | I really need to practice regexps more :( | 06:02 |
| Xase | I'm slacking in my auto-didactic ruby learning :( | 06:03 |
| twb | Well, I will cheerfully take Ubuntu LTS if the alternative is RHEL or SLES or SCO or Solaris or W2k8 or ... | 06:03 |
| Xase | My main dither is I don't have a seperate home partition... | 06:03 |
| Xase | So installing Ubuntu kills all my files... unless I dual boot, copy crap over and chmod/chown a lot of crap. | 06:04 |
| Xase | And then kill my debian install. | 06:04 |
| Xase | Meh, I'd just upload the important stuff if my upload speed wasn't like 0kb/s | 06:04 |
| twb | If your /home isn't separate from the OS, you deserve anything that happens to you | 06:04 |
| Xase | I normally do have a seperate home, but I had a two year old clinging to my back at the moment, and couldn't focus on much of anything. | 06:05 |
| Xase | The only thing I managed to make separate was /boot | 06:05 |
| twb | That's why you hang them on the coat hook by their overalls | 06:07 |
| Xase | Do Dresses work as well? | 06:07 |
| twb | Dunno | 06:07 |
| Xase | I think I'd murder the family member to give my daughter overalls. | 06:07 |
| Xase | Reminds me of my Grandaddy... who always mows the lawn once a week in just overalls. | 06:08 |
| Xase | No shoes, socks, underwear or shit... | 06:08 |
| Xase | shirt* | 06:08 |
| Xase | just overalls. | 06:08 |
| twb | She's two, she probably rolls around in mud and stuff, overalls sound practical to me | 06:08 |
| twb | Or maybe not overalls but a jumpsuit like a mechanic or a pilot wears | 06:09 |
| Xase | Nah, she's sophisticated, she can navigate the ipod, the nook color, and my Inspire, and her mother's laptop. | 06:09 |
| Xase | She's a tomboyish dress wearing nerd in growth :D | 06:09 |
| twb | That doesn't preclude mud | 06:09 |
| Xase | Man she's got a mouth. | 06:09 |
| Xase | The only time she gets near mud is when she's kicking it at ya'. :p | 06:10 |
| twb | I don't have a good solution re. mouth, except maybe wearing noise-cancelling headphones | 06:10 |
| Xase | Eh I only have a x86 install disc =/ | 06:11 |
| twb | PXE FTW | 06:11 |
| Xase | She is the noise cancelling headphones... | 06:11 |
| Xase | You know... I've never given PXE a try... | 06:11 |
| Xase | Give that baby netflix... and out! | 06:11 |
| Xase | Yeah but don't I already have to have something serving the media @ twb ? | 06:12 |
| twb | EH? | 06:13 |
| twb | Oh, yeah, you need a PXE server | 06:13 |
| Xase | I'll just go minimal... | 06:14 |
| Xase | I'll just redownload all my crap, though it will take a lengthy ass time over my connection :D | 06:14 |
| Xase | o.o | 06:14 |
| Xase | Really lengthy. | 06:14 |
| Xase | It'd be faster to just download the whole CD... | 06:14 |
| twb | Whatever | 06:14 |
| Xase | Too bad I can't bypass my ISP and use my shell's internet as my internet :D | 06:14 |
| Xase | 20mbits would be nice | 06:15 |
| twb | "your shell" ? | 06:15 |
| Xase | I have a shell account provided by a friend on their data server. | 06:15 |
| twb | Oh right. | 06:15 |
| Xase | I use it for irssi, and stuff. | 06:15 |
| twb | Sure, just use sneakernet between there and you | 06:15 |
| twb | apt-walk or debmirror or so | 06:15 |
| twb | 20mbps is only about standard ADSL2+ speed, tho. | 06:16 |
| Xase | Sneakernet doesn't quite cut it over two oceans. | 06:16 |
| twb | :P | 06:16 |
| Xase | Man I forgot the line to check bandwidth via ternminal | 06:18 |
| === doko_ is now known as doko | ||
| lilstevie | Xase: the depreciated package is a little overkill | 11:26 |
| lilstevie | it is depreciated cause it was rolled into evdev AFAIK | 11:26 |
| S0NiC | hi guys | 12:02 |
| S0NiC | ogra_ are u there? ;) | 12:02 |
| ogra_ | S0NiC, yes, whats up ? | 12:24 |
| S0NiC | ogra_ do you know rootstock? | 12:30 |
| ogra_ | i wrote it ... | 12:30 |
| ogra_ | its obsolete | 12:30 |
| S0NiC | more obsolete than this https://wiki.ubuntu.com/ARM/BuildEABIChroot? | 12:30 |
| ogra_ | same level :) | 12:30 |
| S0NiC | ah ok, i have two questions | 12:31 |
| XorA | lilstevie: the same dude thats ubuntu on transformer? | 12:32 |
| S0NiC | 1. is armv5tel compatible to arm7? | 12:33 |
| S0NiC | sounds no. right? | 12:33 |
| XorA | ubuntu doesnt have an armv5 build :-( | 12:33 |
| ogra_ | not sure, check wikipedia ... i thougth v5tel was arm11 but i'm usually wrong with the old stuff (and we dont support it anywhere anyway) | 12:34 |
| xranby | can we reduce the default thread stack size in glibc from 8MB to say 2MB or less...? i find it waste of resources on arm to fill up the 512Mb of memory by simply starting some programs that are totally running 64 threads on the system | 12:34 |
| S0NiC | ogra_ ill check it | 12:34 |
| ogra_ | S0NiC, well, wrt ubuntu it wont gain you anything, we only support armv7, cortex-a8 and upwards | 12:35 |
| S0NiC | ogra_ armv7 is what i need | 12:36 |
| ogra_ | well, i was just wondering since you asked about v5 above | 12:36 |
| S0NiC | what you recommend? | 12:36 |
| ogra_ | ?? | 12:36 |
| ogra_ | for pre-cortex-a8 you mean ? use debian, they support older SoCs | 12:37 |
| S0NiC | ogra_ how i get a toolchain with armv7, i tried https://wiki.ubuntu.com/ARM/BuildEABIChroot? and uname -m tells me armv5tel | 12:37 |
| S0NiC | thats why iam wondering | 12:37 |
| ogra_ | then you muss use a very old release in your chroot | 12:37 |
| ogra_ | we dont have any supported release for pre-v7 | 12:38 |
| S0NiC | did you remember we did this yesterday? and you told me that is deprecated | 12:38 |
| ogra_ | karmic and jaunty used to support older HW but both of them are EOL | 12:38 |
| S0NiC | ogra_ ok, how i get a toolchain running arm7. thats the question | 12:38 |
| S0NiC | :) | 12:38 |
| ogra_ | (and we only supported them because immediately switching to v7 was to hard) | 12:39 |
| S0NiC | ah ok | 12:39 |
| ogra_ | use a decebt release for your chroot ... | 12:39 |
| ogra_ | *decent | 12:39 |
| S0NiC | and what you recommend? sorry i have no plan ;_7 | 12:40 |
| lilstevie | XorA: yes, I am he | 12:40 |
| XorA | lilstevie: sweet, one day I shall stop travelling without my Transformer and actually get it installed | 12:40 |
| ogra_ | S0NiC, take natty its the last stable release, fully supports v7 and has the longest life cycle | 12:40 |
| lilstevie | XorA: hehe, I hope on making the process even easier soon :) | 12:41 |
| S0NiC | ogra_ ok ill download it, and how can i get a funcitionally toolchain? | 12:42 |
| XorA | lilstevie: complex Im not worried about as Im an embedded linux guy, but the transformer always being 1000s of miles from my hand is an issue ;-D | 12:43 |
| ogra_ | S0NiC, its is inside there ... you just install the build-essential package | 12:43 |
| XorA | lilstevie: BTW what PMIC is in Transformer do you know? | 12:43 |
| S0NiC | ogra_ thanks. you mean this Ubuntu 11.04 right? is der any stuff i can read especially for this? | 12:45 |
| ogra_ | S0NiC, just use qemu-debootstrap like you did yesterday | 12:48 |
| S0NiC | ogra_ ill try ;D | 12:50 |
| lilstevie | XorA: good question :p | 13:03 |
| lilstevie | XorA: I am really only on the surface still | 13:04 |
| lilstevie | still a lot of work to go before it is 100% | 13:04 |
| XorA | lilstevie: its not one of my PMICs | 13:06 |
| XorA | found it in teardown | 13:06 |
| lilstevie | heh | 13:07 |
| lilstevie | well I haven't noticed any driver issues with the PMIC anyway | 13:08 |
| lilstevie | it wakes up fine, just the framebuffer that doesn't | 13:09 |
| XorA | PMIC probably 90% controlled by its internal power sequencer anyway | 13:10 |
| lilstevie | yeah | 13:12 |
| lilstevie | my only problem with sleep/wake is the framebuffer not coming back up correctly | 13:13 |
| S0NiC | q | 13:38 |
| S0NiC | ogra_ can i bother you again? ;) | 13:39 |
| ogra_ | sure | 13:39 |
| S0NiC | i have to install qemu-arm-static again=? | 13:40 |
| S0NiC | i only bootet now ubuntu 11.04 in my vm | 13:40 |
| S0NiC | booted | 13:40 |
| ogra_ | qemu-user-static is the actual name since a few releases | 13:40 |
| S0NiC | ah ok thx | 13:40 |
| ogra_ | and from that use qemu-debootstrap like yesterday | 13:41 |
| ogra_ | with suite set to natty | 13:41 |
| S0NiC | that was what i thought. thx | 13:41 |
| S0NiC | qemu-debootstrap --arch armel --suite natty illl try this | 13:49 |
| ogra_ | not --suite | 13:53 |
| ogra_ | qemu-debootstrap --arch armel natty <target dir> | 13:53 |
| S0NiC | ogra_ i trie | 13:53 |
| S0NiC | try | 13:53 |
| S0NiC | ogra_ then i can do "sudo chroot <mydir>? | 13:56 |
| ogra_ | right | 13:56 |
| S0NiC | and the i do update? and after that built-essentials? | 13:57 |
| doko | ogra_, what tags are used for armhf ftbfs issues? stll arm-porting-qeuue? | 13:57 |
| ogra_ | doko, ask rsalveti :) i never use tags for that, i usually just pick from ubuntuwire | 13:58 |
| * ogra_ finds using bugs for ftbfs overkill | 13:58 | |
| rsalveti | doko: yes | 13:58 |
| S0NiC | ogra_ iam here again... i cant update/install anything, i get the following error: http://nopaste.info/1edf6eb8ad.html | 14:18 |
| ogra_ | S0NiC, check /etc/pat/sources.list | 14:19 |
| ogra_ | */etc/apt/sources.list | 14:19 |
| S0NiC | mom | 14:20 |
| S0NiC | ogra_ its empty | 14:20 |
| ogra_ | k | 14:20 |
| ogra_ | add the following line to it | 14:21 |
| ogra_ | deb http://ports.ubuntu.com/ubuntu-ports natty main restricted universe multiverse | 14:21 |
| ogra_ | then run apt-get update again | 14:21 |
| S0NiC | if i try ifconfig i get this ifocnfig Warning: cannot open /proc/net/dev (No such file or directory). Limited output. SIOCGIFCONF: Bad address maybe this is a point? | 14:21 |
| ogra_ | that should give you the packages | 14:21 |
| S0NiC | ok | 14:21 |
| S0NiC | ah now it works | 14:22 |
| S0NiC | thx | 14:22 |
| ogra_ | :) | 14:22 |
| S0NiC | that are problems, i have no idea where i have to search for an solution... ;( | 14:23 |
| ogra_ | well, thats basic knowledge about .deb based systems | 14:23 |
| ogra_ | not even ubuntu specific | 14:23 |
| S0NiC | i only run gentoo ;) | 14:23 |
| S0NiC | update / upgrade is done | 14:24 |
| S0NiC | now i try to install build-essential | 14:24 |
| S0NiC | ok worked | 14:24 |
| ogra_ | and then you want the -dev packages for the libs your app links against :) | 14:25 |
| S0NiC | and hopefully it worked after that ;D | 14:25 |
| S0NiC | ogra_ http://nopaste.info/9ea21b319d.html i get an ARM file... hope it works... | 15:10 |
| S0NiC | cross your fingers ;) | 15:11 |
| ogra_ | good luck :) | 15:11 |
| xranby | ogra_: no armhf image sin http://cdimage.ubuntu.com/daily-preinstalled/current/ ? | 15:20 |
| xranby | im looking for a pandaboard armhf omap4 | 15:20 |
| ogra_ | xranby, patience ! | 15:26 |
| ogra_ | ac100 just finished, i'll start omap4 soon | 15:26 |
| xranby | ogra_: ok, patence lasted ~10 min this time.. will the ac100 image appear at the daily-preinstalled location? | 15:35 |
| ogra_ | yes, once the post processing is done ... | 15:35 |
| ogra_ | the livefs build succeeded, waiting for the second part to finish atm | 15:35 |
| ogra_ | its my first testbuild ... might be that it has issues | 15:35 |
| xranby | ogra_: i appreciate the work you do. | 15:40 |
| xranby | will occupy myself with browsing through and triage armhf bugs while i wait | 15:40 |
| ogra_ | hmm, it finished | 15:40 |
| ogra_ | but i dont see it being mirrored | 15:40 |
| ogra_ | http://cdimage.ubuntu.com/daily-preinstalled/20111207.1/ shoudl theoretically have an armhf image for ac100 | 15:41 |
| xranby | which package maintains contains the imager's code? | 15:42 |
| ogra_ | some of it isnt public | 15:43 |
| ogra_ | generally its the cdimage and the debian-cd projects | 15:43 |
| ogra_ | nope. no trace of armhf anywhere | 15:44 |
| * ogra_ tries something else | 15:45 | |
| ogra_ | re-running the second part of the build | 15:46 |
| ogra_ | with some different options | 15:46 |
| ogra_ | (that will take about 30min) | 15:48 |
| ogra_ | AAAND !!! THERE WE GO !!! | 15:49 |
| ogra_ | omap4 started | 15:50 |
| ogra_ | xranby, ^^^ | 15:50 |
| xranby | \o/ | 15:50 |
| ogra_ | should be ready in about 2.5-3h | 15:50 |
| xranby | too bad i cant text you a beer... so here have a ☕ | 15:51 |
| ogra_ | *slurp* | 15:51 |
| ogra_ | :) | 15:51 |
| S0NiC | ogra_ IT WORKS!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! ;D | 15:54 |
| ogra_ | congrats :) | 15:55 |
| ogra_ | seems to be a good day for everyone today | 15:55 |
| S0NiC | ogra_ it was a good idea to join this channel, your help was awesome ;) | 15:56 |
| S0NiC | many many thanks | 15:56 |
| S0NiC | ! | 15:56 |
| ogra_ | youre welcome :) | 15:56 |
| ogra_ | hmm, intresting, the hf image is 20M smaller than the armel one | 16:00 |
| === Quintasan_ is now known as Quintasan | ||
| ogra_ | janimo, did you notice that adam added hf support to the ac100 package ? it not merged into git ... | 16:11 |
| ogra_ | *it is | 16:11 |
| janimo | ogra_, I noticed, yes | 16:12 |
| ogra_ | so before going to 3.0 please merge :) | 16:12 |
| janimo | looked at the quite impressive diff | 16:12 |
| janimo | I need to merge it to git tree of course before 3.0 | 16:12 |
| janimo | :) | 16:12 |
| ogra_ | beyond that see above, ac100 armhf image is available ;) | 16:12 |
| janimo | yup :) nice | 16:12 |
| janimo | looking forward to drop all armel images :) | 16:12 |
| ogra_ | yeah | 16:12 |
| ogra_ | i'm really curious why hf is so much smaller though | 16:13 |
| doko | btw, libreoffice needs some porting work. won't continue with this myself | 16:13 |
| doko | ogra_, well, compare some single packages | 16:13 |
| ogra_ | doko, we lost about 20MB | 16:14 |
| ogra_ | quite impressive | 16:14 |
| janimo | doko, I plan to look at Libo armhf myself | 16:15 |
| doko | ogra_, I don't think this is all code | 16:15 |
| ogra_ | probably not | 16:15 |
| doko | janimo, see bug #900636 | 16:15 |
| ubot2 | Launchpad bug 900636 in libreoffice "libreoffice ftbfs on armhf" [High,Confirmed] https://launchpad.net/bugs/900636 | 16:15 |
| ogra_ | i'm to lazy to compare the manifests before testing | 16:15 |
| ogra_ | will do that later ... | 16:16 |
| janimo | doko, thanks for the diff. I suspect I'll have another few days of 'fun' with libo source tree | 16:16 |
| ogra_ | it didnt error out or anything during the live build phase so there cant be anythng missing, whats seeded will be installed | 16:17 |
| doko | janimo: I think it's only the bridge, passing the arguments in floating point registers | 16:18 |
| janimo | ogra_, does the amrhf gcc have different build option defaults than armel? | 16:18 |
| janimo | doko, indeed, uno bridge assembly and stack wizardry | 16:18 |
| ogra_ | not that i know of | 16:18 |
| ogra_ | beyonf the vfp stuff indeed | 16:18 |
| doko | no, just -mfloat-abi=hard | 16:18 |
| janimo | ogra_, could be we have some optional package deps somewhere which only are active for armel? | 16:19 |
| janimo | so for armhf they are not brought in? | 16:19 |
| ogra_ | shouldnt be anymore | 16:19 |
| ogra_ | we used to ... banshee etc | 16:20 |
| ogra_ | but that was all dropped | 16:20 |
| ogra_ | well, something to examine later | 16:20 |
| doko | could be; currently we only did fix building package to build on armhf, which only built on armel | 16:20 |
| ogra_ | lets see how omap4 looks like | 16:20 |
| GrueMaster | Can we get started on a netinstall image for mx5 soonish? It will help for future testing, and a lot of people with sata will be pleased. | 16:26 |
| ogra_ | GrueMaster, do we have the hf kernel yet ? | 16:26 |
| ogra_ | i dont think we do until linaro uploads the change | 16:26 |
| GrueMaster | No, but we have an armel kernel since Oneiric. | 16:27 |
| * ogra_ is afke for a moment | 16:27 | |
| GrueMaster | We can at least get the prelim work done so when there is a HF kernel, we're ready for it. | 16:27 |
| ogra_ | GrueMaster, oh, netinst ... i guess that wont work, d-i needs the kernel in main iirc | 16:27 |
| ogra_ | anyway, back soon | 16:28 |
| GrueMaster | For all armel platforms, the kernel needs to be preseeded anyway. | 16:28 |
| janimo | doko, what is a good example of a package that has armhf and armel separate codepaths with ifdefs (or whatever is necessary) ? | 16:33 |
| GrueMaster | janimo: If you are looking at the size diffs between packages, it could simply be less instructions needed to perform the same operation determined by the compiler. | 16:35 |
| doko | janimo, I think you would need to look at the sources control file for things like [armel] in dependencies. you won't see this in the packages file | 16:35 |
| doko | janimo, infinity has a modified suite-diff.py which we used to find packages not built on armhf at all. but this won't help much for universe until most of the archive is built | 16:37 |
| doko | ogra_, so what is the status about the ac100 images? | 16:39 |
| xranby | doko: ac100 armhf image status https://bugs.launchpad.net/ubuntu/+source/ubiquity/+bug/901315 | 16:53 |
| ubot2 | Launchpad bug 901315 in ubiquity "crash during background user creation during ac100 armhf precise installer" [Undecided,New] | 16:53 |
| ogra_ | doko, ubiquity is broken, but all other bits seem fine (the bug was apparently sent using firefox :) ) | 17:01 |
| ogra_ | GrueMaster, preseeding isnt the point ... d-i only uses main during its own build afaik ... you need vmlinuz from the .deb to actually roll the netinst images | 17:02 |
| xranby | yes i have sent it using firefox from inside the armhf precise userspace | 17:02 |
| ogra_ | yay | 17:02 |
| ogra_ | bah, but the omap4 image failed to build :( | 17:02 |
| * ogra_ checks why | 17:02 | |
| ogra_ | geez, language-selector explodes | 17:05 |
| ogra_ | Setting up hunspell-en-us (20070829-4ubuntu2) ... | 17:08 |
| ogra_ | Error: update-openoffice-dicts not present or executable. Missing dependency on | 17:08 |
| ogra_ | dpkg: error processing hunspell-en-us (--configure): | 17:08 |
| ogra_ | subprocess installed post-installation script returned error exit status 1 | 17:08 |
| ogra_ | No apport report written because MaxReports is reached already | 17:08 |
| doko | ogra_, a new ubiquity is still building ... | 17:14 |
| doko | * Add armhf support. | 17:14 |
| ogra_ | yeah | 17:15 |
| ogra_ | the hunspell thing looks weird though | 17:15 |
| ogra_ | 3h before the ac100 build succeeded and i dont think the packages changed in that area | 17:15 |
| xranby | janimo: jamvm in openjdk have separate code paths http://git.berlios.de/cgi-bin/cgit.cgi/jamvm/tree/src/os/linux/arm/callNative.S | 17:18 |
| NekoXP | GrueMaster, there's no work required to get an armhf kernel running, the kernel is agnostic as to floating point ABI | 17:27 |
| ogra_ | NekoXP, it needs to match dpkg | 17:28 |
| NekoXP | you just need to change the arch in the packager | 17:28 |
| ogra_ | which means a rebuild | 17:28 |
| GrueMaster | Yes, I figured as such. More a case of adding support for register dumps during task switches and what-not. | 17:28 |
| NekoXP | GrueMaster, no support at all, kernels don't need to change for armhf no matter what | 17:29 |
| GrueMaster | I don't know enough about arm architecture, but I would be surprised if it didn't need a little support added for context switching (same as SSE* on x86). | 17:30 |
| NekoXP | if it supports floating point units at all, it will dump them as it always did, there's no userspace<->kernel ABI difference as floating point registers can't be used for syscalls | 17:30 |
| NekoXP | it's there for softfp | 17:30 |
| NekoXP | you have to save floating point registers regardless, but for hardfp userspace gets to *pass arguments* in floating point registers | 17:31 |
| NekoXP | but that's only if you're passing a float or double, you can't pass floating point to the kernel | 17:31 |
| GrueMaster | Still, if the kernel isnt building for armHF, then some work needs to be done, even if it is only at the package build level. | 17:31 |
| NekoXP | right. that said, I would really really like to see someone take the initiative and decouple the kernel arch from the userspace arch :) | 17:32 |
| NekoXP | surely dpkg could be modified to have knowledge of a "suitable kernel architecture"... | 17:33 |
| GrueMaster | It is more than just dpkg. A lot of other package managers would also need modifying. | 17:34 |
| NekoXP | for instance with this x32 architecture project, that means running 32-bit apps in a 64-bit compatible ABI, it would be another example of different kernels being fully capable of running the same userspace (and x64 already can run a ia32 userspace) | 17:34 |
| NekoXP | actually what would be less weird is if any binaries are required to go into packages, mark the package as that ABI. Since binaries that don't use VFP on armel or armhf can run together anyway it will reduce package building for things that don't touch fp at all | 17:35 |
| NekoXP | that was the original idea, I guess it's difficult though right? | 17:36 |
| NekoXP | but not so different from the differentiation between ${my_running_arch} and "all"? | 17:36 |
| GrueMaster | hey, the pandaboard ES is now on pandaboard.org. Cool. | 17:38 |
| GrueMaster | NekoXP: I can already do that partially on arm. I was running a chroot to an armhf image from an armel image (although I should have used lxc as it left the platform unstable). | 17:39 |
| ogra_ | just because you mounted /proc :) | 17:41 |
| NekoXP | sure, the only incompatibility is if it uses vfp instructions AND uses the vfp abi extension it's basically mutually incompatible for those functions only. If you had libc and libm, libc would be basically completely agnostic.. libm would need to be linked depending on the elf header.. that kind of thing | 17:42 |
| NekoXP | the vast majority of stuff doesn't actually throw around fp arguments, but there's some stuff like OpenGL which definitely does and it makes a hell of a difference | 17:42 |
| NekoXP | and even then you could get the benefit of the improved ABI internally, and use assembler stubs or macros to call out to armel libs if need be (or the other way around). I actually do wonder if gcc would use the vfp abi extension for static functions even if it was built with armel? | 17:44 |
| GrueMaster | Well, I think the goal is to switch to armhf before release, and leave armel to bit-rot, making the whole argument moot. | 17:44 |
| NekoXP | good point | 17:44 |
| GrueMaster | Aven if we keep armel floating along, it should be easy to run armel packages on an armhf system using multilib. | 17:45 |
| GrueMaster | *Even | 17:45 |
| GrueMaster | The opposite is not guaranteed to work. | 17:45 |
| ogra_ | well, moot once we shot down the armel archive | 17:46 |
| * GrueMaster reads up on the sysboot3 switch on the pandaboard ES. Switches boot from USB/MMC1 to UART/MMC1. Useless. | 17:47 | |
| S0NiC | cu guys | 18:01 |
| janimo | xranby, thanks for the pointer | 19:04 |
| apw | GrueMaster, has ppisati given you any 3.2 based ti-omap4 kernels to test ? | 19:46 |
| GrueMaster | I haven't seen one yet. | 19:46 |
| GrueMaster | Nothing on my mirror (updated from ports.u.c bi-hourly). | 19:48 |
| infinity | GrueMaster: It's not uploaded/built yet. apw's just about to do that for us. | 19:50 |
| GrueMaster | ah, ok. | 19:50 |
| apw | GrueMaster, yeah was wondering if you had had a preview | 19:52 |
| GrueMaster | Nope. Loop {everyone}; me. Same as always. :p | 19:52 |
| infinity | You and I entered the loop at the same time. :) | 19:53 |
| infinity | And I'm not sure anyone else is in it. | 19:53 |
| infinity | So, you can claim first this time! | 19:54 |
| GrueMaster | This looks prommisint though: linux-image-3.0.0-1402-omap4_3.0.0-1402.3_armhf.deb on my mirror. | 19:54 |
| MrCurious | anyone here have any experience with asm on arm? | 19:54 |
| GrueMaster | And netinstall images! Santa came early last night. | 19:55 |
| MrCurious | http://pastebin.com/zkP6Zbbs <- trying to understand why it doesnt like REGS_TO_SAVE i think it may be in a invalid format for the __asm__ volatile call | 19:55 |
| GrueMaster | (even if my fingers are cramping up and I can't type). | 19:56 |
| GrueMaster | Ewww. gcc inline assembly. | 19:56 |
| GrueMaster | Brings back bad memories from my early Intel days. | 19:57 |
| apw | GrueMaster, that is the previous kernel, we are taking about jumping that 2 kernel versions :) | 19:57 |
| apw | but if you have working netinstall images finally thats great | 19:57 |
| MrCurious | if i trim regs to save down to 1, then it flies | 19:57 |
| GrueMaster | apw: Understood, but having an armhf kernel for omap4 now means I can have one system play while I way for your spin. | 19:58 |
| apw | GrueMaster, sounds like a good things | 19:58 |
| infinity | GrueMaster: I'm curious if that omap4 netinst plays up for you. | 19:59 |
| GrueMaster | I'll let you know in a few minutes. | 19:59 |
| infinity | GrueMaster: If you get the same "everything works, but I get no login" issue, then it's definitely time for me to get digging on reproducing. | 19:59 |
| GrueMaster | I was able to get further with init=/bin/sh. Something in unity seemed to be hosing the system. | 20:00 |
| infinity | Unity, feh. | 20:02 |
| infinity | Netboot a server install instead of a desktop one? :) | 20:02 |
| GrueMaster | Oops. Meant upstart. | 20:02 |
| infinity | Oh. That's potentially less pleasant. | 20:02 |
| infinity | Could be neverending respawning gettys. | 20:02 |
| GrueMaster | Does ssh spawn a getty? | 20:03 |
| infinity | I think upstart helpfully hides that situation from you, unlike sysv that would spam the console with "YOU'RE FUCKED, MATE". | 20:03 |
| GrueMaster | Heh. | 20:03 |
| infinity | Hrm, no. SSH doesn't run a getty binary. Though, it does still use PTYs. | 20:03 |
| GrueMaster | Crap. Kernel Panic. initrd issue. | 20:03 |
| infinity | Initrd issue? :( | 20:04 |
| infinity | Don't tell me I need to fix it harder. | 20:04 |
| MrCurious | solved :D | 20:04 |
| infinity | Fucksake. It's still putting it in /lib isn't it? | 20:05 |
| infinity | d-i needs a talking to. | 20:05 |
| GrueMaster | Looking | 20:06 |
| GrueMaster | yep. Get out the flogging stick. | 20:06 |
| infinity | I figured moving it in the udeb would be enough. Apparently, it's braindead and just flattens everything to /lib >:( | 20:07 |
| GrueMaster | No arm-linux-gnueabihf in /lib. | 20:07 |
| infinity | Will fix later today. | 20:07 |
| infinity | But if you repack that mess, here's hoping it work. | 20:07 |
| infinity | s | 20:07 |
| infinity | (Works better than omap, that is) | 20:07 |
| GrueMaster | I'll have to wait for a little bit (or cobble the boot process manually). The current setup pxeboots and pulls from my mirror, which is now updating. If I manually flog uInitrd, it will just get erased. | 20:08 |
| GrueMaster | Ok, onward. Netinstall started. Moving on. | 20:20 |
| GrueMaster | grmbl. "Loading libc6-udeb failed for unknown reasons. Aborting." | 20:20 |
| GrueMaster | infinity: ^^^ | 20:23 |
| infinity | ... | 20:28 |
| infinity | GrueMaster: Anything in syslog? | 20:29 |
| GrueMaster | And how do you propose I look when I can't launch a shell or save the log w/o libc6? | 20:31 |
| GrueMaster | (logs are in the ramdisk until post-install). | 20:31 |
| lool | Error: update-openoffice-dicts not present or executable. Missing dependency on | 21:03 |
| lool | dictionaries-common? | 21:03 |
| lool | Looks like armhf images will need libreoffice in some way | 21:04 |
| infinity | We'll get there. | 21:08 |
| infinity | But it also looks like some package is, indeed, missing a dependency. :P | 21:08 |
| infinity | (Or missing an [ -x /foo ] guard. | 21:08 |
| infinity | ) | 21:08 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!