=== Lopi is now known as Lopi|idle [10:24] hi, quick question, why does rootstock on karmic insist on qemu even when running under arm? [11:43] sigh, apachelogger had a boring sunday it seems [11:44] * ogra_ wades through bugmail === lilstevie|ZNC is now known as lilstevie [11:55] ogra: heh, any interesting bugs? [11:56] nah, just general noise [11:56] damn [11:56] you should see this weird bug I have had come up since switching to a newer kernel :p [11:57] ^^ [11:57] tell me [11:57] ogra_: more like proper triage :P [11:57] heh, yeah [11:57] the bootloader image is taking over, graphical corruption all round, [11:57] ouch [11:57] basically in samsungs 2.6.35 sourcedrop the fbdev driver has been royally screwed [11:58] yeah, sounds bad [11:58] my login screen is a few gdm elements overlayed over the bootsplash [11:58] did you try to roll back the driver ? [11:58] not yet [11:59] I will try that when I get back [11:59] brb [11:59] food [11:59] Hey, either of you familiar with rootstock? [12:02] I guess I killed the channel again [12:03] while i initially wrote rootstock i havent touched it in more than 12months [12:04] Well, close enough... I had a sorta-generic question [12:05] Is there a armel image that contains enough to run rootstock from a chroot setup? [12:06] Because most of the references I see refer to qemu this or that [12:06] * ogra_ has never run rootstock on arm ... and it wasnt originally created for that [12:06] I can't even apt-get rootstock on karmic because it fails on qemu [12:07] iirc someone patched it to be able to run natively though, but i have zero experience with that [12:07] Hmmm.... [12:07] you shouldnt run karmic anyway [12:07] its EOL [12:07] I know, but that's the image I got to be able to run Ubuntu on an android phone [12:08] Linux localhost 2.6.32.9 #1 Sun May 22 19:44:33 CDT 2011 armv7l GNU/Linux [12:08] what HW is that ? [12:08] epic [12:09] ah, thats samsung, right ? [12:09] htc [12:09] Yeah [12:09] sorry [12:09] no. sammy [12:09] evo is htc [12:10] lilstevie has some samsung experience ... [12:10] k, good [12:10] I want to also sorta-resurrect rhodbuntu for xda soon [12:10] if you want a barebone image, take the omap preinstalled netbook image, grab the content of the second partiton and touch /var/lib/oem-config/run [12:11] hmm, k [12:11] tar it up and untar it on your phone [12:11] well, there lies the issue [12:11] or if you want it more minimal you can du the same with the headless image [12:11] this *is* my computer atm [12:11] *do [12:12] so right now karmic is chroot'ed and running epic4 which lets me be here [12:13] but I want to be able to create something a bit better [12:14] which device? [12:14] I just find it quite amusing that I can subsystem a true GNU/Linux on my phone [12:14] epic 4g [12:14] that is the same family as the i9000 isn't it [12:14] yeah [12:15] galaxys [12:15] only one with a keyboard [12:15] ok, if you give me a bit I will run through some of it with you, I have already gotten things working on the sgs i9000 [12:15] just no modem [12:16] I don't need modem, I just nice chroot /data/local/ubuntu /bin/sh [12:16] eh I don't chroot [12:16] Let android handle the hw... [12:17] chrooting is far from ideal, just fyi [12:17] I know [12:17] Trust me, I know [12:18] heh [12:18] I have a nice setup going, flash kernel and param with heimdall and throw the image on one of the SD cards [12:19] lilstevie, btw, how about providing your flash-kernel patch for upstream inclusion (or at least ubuntu inclusion) [12:19] I just use an ext4 kernel and loopmount the ubuntu image [12:19] ogra_: I dd [12:19] oh ? [12:20] ogra_: the kernel is stored in its own block device that the bootloader reads [12:20] a flash_kernel patch may help for easier updates including command line [12:20] yes [12:21] I'm still trying to figure exactly how but there is a socket in one of the closed source modules that allows writing direct to the non-volatile stash that bootloader params are stored in [12:21] how big is your sd image? [12:21] my SD image is 2GB [12:22] brb gotta do something so the mrs doesn't kill me [12:23] The current "linux" image is woefully inadequate [12:24] the guy that wrote the bootstrap script doesn't even know how to sed and cut [12:25] I got it to the point where it doesn't whine about fs and partiion layouts [12:27] but at least android uses the linux kernel to the point that you can chroot and run a gnu system under it [12:27] back, [12:28] I use an ubuntu initramfs in my kernels [12:28] after playing round with the compression managed to get a natty initramfs to fit too :) [12:29] nice [12:29] lilstevie, oh. you had compression issues ? [12:29] kernel now supports lz right? [12:29] ogra_: yeah the zImage was 8.3MB [12:30] lilstevie, what did you do ? [12:30] i have a similar prob on the ac100 [12:30] but the partition is just over 7 [12:30] I switched to lzma [12:30] thought so [12:30] lz is great except it's a memory hog [12:31] hmm, doesnt help here [12:31] * ogra_ already tried all possible compression methods [12:31] gzip gave me an 8.3MB kernel [12:31] lzo gave me 10mb [12:31] lzma is just fitting weighing in at 6.9MB [12:31] everytime someone posts something in 7z I can't decompress half cuz the dictionary is too big [12:32] ogra_: how big is your partition [12:32] 8M [12:32] hmm [12:32] why so big then? [12:32] but for a full fastboot boot.img [12:32] do you need to mkboot [12:32] the kernel is 3.5M [12:33] so i only have 4.5 left [12:33] the samsung bootloader takes a raw vmlinux [12:33] or zImage [12:33] ah, lucky you [12:34] ac100 uses a fastboot img file to boot ... i need to pre-process kernel and initrd [12:34] I also have a serial shell with the bootloader [12:34] using abootimg [12:34] no serial port here [12:34] eugh, I am not looking forward to that with the transformer [12:34] so a console wouldnt help much [12:34] yep [12:34] well, should be easy [12:35] heh yeah, going to be kernel level for most of it [12:35] i guess it shouldnt take more than one/two days to get a first cut image running [12:35] yeah thats how long I give it [12:35] it is being sent tomorrow, [12:35] properly integrated rather a week or two [12:35] (depending on the kernel) [12:35] but when it gets here I forsee booting within the hour [12:36] I am going to be getting on ASUS ass about GPL [12:36] the kernel source is there [12:36] they still haven't released the 3.1 kernel source [12:36] question is how the nvidia bits are integrated [12:37] much like the xoom [12:37] from what I have looked over so far in the 3.0 source [12:37] if transformer uses nvec as badly as ac100 it will be really really bad to get the peripherials working [12:37] yeah I hope not :/ [12:37] they all need undocumented init codes from nvec here [12:37] are nvidia willing to cooperate? [12:37] so for ac100 it took nearly 6 months to get that properly worked out [12:37] I take it not [12:38] nvidia ? [12:38] lol [12:38] heh well their SoC is horribly documented [12:38] i think the devs would love to ... but management doesnt care [12:38] heh [12:38] samsung are actually taking a big interest in OSS which has been a huge plus [12:38] stevie: do you think using rootstock is the best way to create a new image? [12:39] just getting their 3D driver in a binry form like the x86 one would change the world [12:39] I have been using the omap4 prebuilt [12:39] didn't they just send a few 'pad to kernel hackers? [12:39] hmm, k [12:39] yeah they sent a 10.1 to supercurio [12:40] and natty works on the galaxy? [12:41] and I think karmic was the last one to support arm6 correct? [12:43] yes [12:43] (karmic) [12:43] it is much harder galaxy s is armv7 [12:43] 'er [12:44] shouldnt stop half way in :p [12:44] galaxy s is an armv7 device, cortex-a8 [12:44] Well, gonna use natty on the epic [12:45] ogra_: 3D drivers are something we wish for here too [12:45] But I also want to do something similar for the msm7k2 [12:45] which is arm6 [12:45] lilstevie, well, the nvidia ones exist ... but they are linked and built against unusable xorg versions [12:45] memory might be an issue though [12:45] so you can onyl use them if you also replace half your rootfs [12:46] ogra_: while I can try hijacking the tiomap SGX drivers my gpl-glue is horrible and doesn't support DRI [12:46] ew [12:46] you should talk to jessebarkers team in #linaro :) [12:46] why would they do that [12:46] for the module? [12:47] would omap4 be the right one for karmic too? [12:47] I've been chipping away at patching in the samsung stuff to the DKMS but it has been a slow process [12:48] the module isnt the prob, the xerver is [12:48] i can roll a kernel that ignores all versioning info from that module [12:48] but i cant run userspace without the right xlibs [12:49] well when I say I use the omap4 image, I use the rootfs, rip out all the omap stuff and replace with hummingbird modules and firmwares [12:49] dont use omap4 :) [12:49] use omap [12:49] ah you mean for the ULP GeForce [12:49] what is the difference [12:49] omap4 has the PPA bits [12:49] ah [12:50] so you have a few pieces more to remove to get back to a clean image [12:50] I used the omap4 image cause hardware wise it is closest, ie: same GPU :p [12:50] on omap you can just dpkg -l |grep omap and remove all the bits [12:50] ah nice [12:51] okay [12:51] what's ppa, sorry [12:53] Personal Package Archives [12:53] k [12:54] * ogra_ actually hopes someone from the nouveau people will develop some interest in tegra [12:55] ogra_: I really hope that the keyboard dock for the transformer doesn't need undocumented shit implemented in nvec to work [12:55] that would be a massive failure [12:56] well, that would be what i expect though :) [12:56] from nvidia [12:56] how different is the omap prebuilt to say rootstock-based image? [12:56] it uses a diffrerent build system [12:57] ogra_: lets just hope asus took a bit more control for their dock [12:57] we'll see [12:58] any pro/cons to either, or am I being a n00b [13:00] well, prebuilt gives you a read made system, rootstock *creates* a system [13:01] and given that you cant run it natively apparently, i dodnt really see where you have choice ;) [13:02] eugh, looking at the sources it appears the dock is interfaced through NVIDIA Tegra internal matrix keyboard controller [13:02] yeah, true [13:03] sounds like nvec [13:05] Are there plans to support tegra? [13:07] no, but there will be an ac100 community image [13:07] and probably also one for the eeepad tarnsformer [13:07] *trans [13:07] how much ram is considered decent for arm-based sytem? [13:07] there wont be official tegra support from ubuntu though [13:08] Xofrats, 512M works ... i wouldnt even start trying with anything less [13:08] hopefully my eeepad arrives this week [13:08] 512 is ok [13:08] hmmm, ew... [13:08] wonder how rhodbuntu got it working with 200 [13:08] I am actually thinking of cramfs [13:09] Xofrats, cutting out tons of stuff [13:09] Xofrats, my ac100 uses 130MB for teh desktop on start [13:09] as you can see you can indeed use that on 200M [13:09] just dont start any app ;) [13:10] I guess [13:10] But in this case it'll be sharing memory space with *ndr**d [13:10] yeah I use about 150MB on boot with the tab [13:11] using a browser with 20tabs open, xchat with about 30 channels on several servers, a bunch of terminals and evolution via ssh -X from another machine i use 380M ram and 250M swap here [13:11] 20 tabs... [13:11] according to htop [13:12] ogra_: why wont be official tegra support from ubuntu? any special subject? [13:13] I don't need much for gui from ubuntu anyway [13:13] no collaboration with nvidia atm [13:13] ogra_: k maybe it doesn't I am looking in the wrong config block [13:13] tons of useful console/ncurses app to use [13:14] strobe, ping -f... [13:14] ogra_: wow, nvidia is so close.. :-( [13:14] oh wait, wrong hat *switches hat* [13:14] plm, but there are a bunch of community projects for specific devices [13:15] plm, so we will have community images on cdmiang for some [13:15] *cdimage [13:15] ogra_: yes, I know about OMAP3/4 of TI.. [13:15] i.e. toshiba ac100 ... probably the transformer, likely the nook color [13:15] ogra_: like as pandaboard, bealge ans so one.. [13:15] they are official images [13:16] panda and the omap3 ones are fully supported by canonical [13:16] the tegra2 ones wont be [13:16] ogra_: how does one go about getting a community image on cdimage [13:17] lilstevie, we just work on documenting that [13:17] ogra_: yes.. I would like to use tegra3.. and I would like know if are a there (i not found) a community project for tegra3 like as are there with panda.. [13:17] for now you need to work with a person from the cdimage team (or be in that team) [13:17] ah ok [13:17] lilstevie, though it all starts with having a kernel package in universe [13:18] plm, i dont think you can get tegra3 HW aynwhere yet [13:18] so there is no community for it yet [13:18] ogra_: ok.. but not for tegra2 too :-( [13:19] well, for tegra2 the above applies [13:19] ogra_: ah ok, well my galaxy kernel is nowhere near that standard yet [13:19] ac100 images will be there for oneiric [13:19] lilstevie, what standard ? [13:19] its universe [13:19] ogra_: my standard [13:19] ogra_: above? [13:20] lilstevie, if it doesnt remove all your user data from the HDD and makes most devices work it should be enough [13:20] ogra_: what is that ac100 ? [13:20] indeed if you have a "wipe my disk" bug you ship, thats below any standard :) [13:20] ogra_: I'm looking for in https://wiki.ubuntu.com/ARM [13:20] plm, toshiba ac100 netbook [13:21] yeah, persia wanted to add a page for tegra/ac100 ... apparently thats not there yet [13:21] Mine comes with "nuke from orbit" and locks onto my agps chip [13:21] does that count? [13:22] ogra_: ok.. well.. notebook is so big.. I would like to use power of GPUs but in a little device.. like as omap with dsp... [13:22] ogra_: my standard means it works and can be used [13:23] lilstevie, well, if it does that, roll a package from it :) [13:24] by can be used I mean modem and all [13:24] and modem is far from even close to booting [13:28] ogra_: the more I look at the source for the eeepad the more it looks like it is going to be done in nvec [13:33] anyway, thanks for the info, I'll prolly be back with more n00b q [13:52] welcome back ogra [15:05] w00t! The oneiric daily build from today is booting on the OMAP 4 pandaboard :) [15:07] funny, given that there was no build today [15:08] oh, wait, you mean headless [15:09] ogra_: Yeah, the headless version. :) [15:09] Still going through some filesystem checks at the moment. [15:09] yeah, netbook wouldnt be much fun atm [15:10] given that the whole USB stack seems to be screwed in the current kernel [15:10] Ah. [15:10] That sounds, um, painful. [15:14] ogra_: hey [15:14] ogra_: how much do you use qemu-system-arm with versatile kernels these days? [15:14] lool, me ? [15:14] ogra_: you [15:14] i havent used that since two releases iirc [15:15] Ok; thanks :-) [15:15] rootstock still uses it though [15:15] should be ported to omap or some such [15:15] heh [15:17] hi [15:18] where can i browse the packages available for ARM architectures? [15:18] packages.ubuntu.com seems to have amd64 and i386 only [15:20] launchpad [15:21] generally you can say though that the source package list doesnt differ between x86 and armel, apart from very few arch specific packages like bootloaders [15:22] http://qa.ubuntuwire.org/ftbfs/natty.html has a list of binaries that didnt build on arm [15:22] everything else is available on armel as well as on i386 [15:26] ah.. that's fine [15:28] http://pastie.org/2096114 [15:28] Hmm, OMAP 4 headless image hangs there ^^ [15:30] how big is your SD ? [15:30] the actual write of the resizing happens after that line, it might take a moment with bigger SDs [15:30] 8 gigs [15:30] (you should see the LEDs being busy though) [15:31] Both LEDS are busy [15:31] :) [15:31] good [15:31] then just give it time [15:31] Will do. [15:49] lool, yre you planning to drop versatile completely or just the v7 patch ? [15:50] ogra_: drop it completely [15:51] good [15:51] ogra_: essentially, vexpress should be better in all respects [15:51] in aemu respects too ? :) [15:51] I'm using vexpress these days, and it's buggy, but I believe the same code is used for versatile, so it should be the same really [15:51] *qemu [15:51] Yes; AFAIK it's using the same QEMU drivers [15:51] for SD, fb etc. [15:51] ah, awesome [15:52] but it's not much usable sadly [15:52] the network corrupts or cuts some packets somehow [15:52] worse than versatile ? [15:52] and the SD stalls regularly [15:52] I don't think it's worse than versatile [15:52] k [15:52] oddly, using a Debian userspace makes things better [15:52] not perfect, but better [15:52] I think it's just an accident though [15:53] yeah, i heard the same for the ac100 port, people using debian have less kernel issues it seems [15:53] heh, that reminds me. I want to get debian working on my AC 100 [15:53] (we have a kernel bug that makes input devices die, apparently that shows only very rarely on debian userspace with the same kernel binary) [15:53] (and with similar userspace) [15:54] jeremiah, just bootstrap a rootfs :) [15:55] Cool, I'll do that. :-) [15:57] ogra_: actually something I haven't been able to get straight about the tegra2 [15:57] where does APX exist [15:57] bootloader, or bootrom [15:58] APX ? [15:59] the mode that nvflash talks to [15:59] greetings all. Anyone have a moment to help me out with an ubuntu 11.04 installation on beagleboard-xm rev C? [16:01] in the ac100 thats in rom i think [16:02] hm cool, I am just wondering about flashing risk, etc. if I fuck a flash on the tab when it is doing the bootloader, I am pretty much left with jtag as my only option, other than taking it to samsung [16:02] well, if it is set up similar to the ac100 you cant brick it [16:03] alex12: what issues are you having? [16:03] even if i screw up the MMC completely i can always get it into flash mode and re-do everything [16:03] ok thats what I wanted to hear [16:03] I bricked my tab once, :/ [16:03] thats why i think its in ROM [16:03] lilstevie: define 'bricked' [16:04] ogra_: ok, that tells me what I want to know [16:04] hrw: as in it was a paper weight [16:04] hrw: full definition of brick [16:04] I used canonical's pre-built 11.04 image. Followed instructions to replace uImage and the kernel on the sd-card. Booted the bb-xm rev. c. and a.)colors are off (everything is red) b.) no mouse or keyboard response at the first set-up screen. can't go anywhere. any ideas? [16:04] returned to samsung to fix [16:05] alex12: Which image? Netbook or headless? [16:05] netbook [16:05] hmmm. Haven't heard of that issue offhand. Unfortunately I don't have one of these newer boards to test with. [16:06] okay thanks for your time. i'm going to try the 10.10 pre-built now. [16:07] lilstevie: nice [16:08] hrw: nice that I am not misusing the term, or nice that the device can be permafucked [16:08] ? [16:11] lilstevie: i think he meant "nice" how precisely you defined the term [16:12] heh [16:13] yes [16:13] as lot of people say 'bricked' as 'had to reflash' [16:14] heh soft-brick [16:16] the correct definition of "bricked" depends on the equipment that you have available [16:17] for example an AVR microcontroller can be bricked by disabling the ISP programming interface, but that's only for people who don't have a 12V parallel programmer and a soldering iron [16:17] so its not bricked :) [16:18] bricked for me means something you cant fix ... its a brick [16:19] well that depends on wether "you can't" means "you're unable" or "it's impossible" [16:19] the latter [16:19] my tab required a new logic board [16:23] bricked == have to use jtag or worse - for me [16:24] hrw: agreed [16:25] especially when I lack jtag at home ;D [16:25] I define bricked as fubar'd [16:25] yeah my jtag does not work on the tab [16:25] * ogra_ is happy he doesnt have to work with HW you can actually brick atm [16:26] and i hope that wont change :) [16:26] heh [16:26] samsung screwed something with iROM by not giving a method of halting boot to get into a bootrom recovery [16:27] bitter [16:48] did my bb-xm just die? I have the 5V adapter (from digikey, recommended supply). just powered on, all I'm getting is the D5 LED solid green, no other activity... [16:51] Do you have an SD plugged in? [16:52] yes [16:52] sounds like bootloader or kernel screwup [16:52] okay i'll reimage the SD card, thanks! [17:47] lool: how much mem qemu currently supports for vexpress? 1gb? [17:50] thats what he wrote in the mail [17:51] i'm using bb-xm demo image of ubuntu (netbook-minimal-11.04). after doing an apt-get update, I used 'apt-get install ssh' to install the meta-package. ubuntu was in the middle of it when DVI image just went dead... same thing happened the other evening when trying to install xfce. any ideas why i'd lose video? [17:57] a bug ? [17:57] (would be my first idea) [17:57] :) [18:09] rsalveti: yes [18:10] alex12: could be a power issue too; if you're powering from USB for instance, driving a DVI device + USB + stuff is too much [18:11] lool: thanks, i'll move a device off then [18:44] rcn-ee: thanks for the bb-xm demo image, first time I've been able to get Ubuntu to come alive on my board =) [18:48] cool, your welcome alex12 [18:50] rcn-ee_at_work: so, any ideas why the canonical images don't run right on the -xM rev. C? I know you've been developing the alternatives for quite a while now... canonical still can't keep up? I followed all their instructions, replaced the kernel (which is presumably something similar to what you've done) but I cannot get it to boot. =( Why can't they get it straight? [18:57] alex12: Don't blame us. We have a hard time when board revs change functionality and we aren't informed until users complain. And we don't have them to test/debug. [19:00] alex12, the xM C kinda came out suddenly, i was luckly enough to get one early at a ti training, but by that time it was too late for ubuntu's natty.. the kernel's image was frozen.. with my images, i don't have that limiation.. [19:25] Is it possible to run update-initramfs from within a chroot environment? I got "/bin/df: Warning: cannot read table of mounted file systems: No such file or directory" [19:26] yes, but you need to bind mount /dev from outside the chroot and proc and sys inside the chroot [19:26] Note The chroot is not for the same machine as the running machine [19:27] well, it still uses /dev/null and stuff [19:28] Interestingly the rootstock script mounts proc and devpts from outside the chroot [19:29] hmm ? [19:29] iirc that is [19:29] i dont think so [19:29] it should chroot [19:30] unless someone really mangled my original code [19:30] It has been changed [19:31] never the less, what significance are there concerning mounting proc and sys from the inside? [19:31] the scripts looks at files in there [19:32] *look [19:32] But there's no real difference between mounting proc before running chroot than mounting it afterwards, is there? [19:32] no [19:32] good [19:33] doesnt matter how you mount it, as long as it is populated inside the chroot when you run your command [19:34] yes. And in that context, the rootstock script only mounts proc and devpts, not dev [19:35] Wait a minute, isn't device nodes, like /dev/null, blk nodes, so they dont need /dev as mount to work, or am I misunderstanding? [19:36] null is a char device [19:37] yeah, but still a device node, so if it exists on the fs it will work. It don't need some mounted /dev [19:37] i dont know which devices update-initramfs needs [19:37] null was just an example, just mount it [19:39] Thanks [19:44] There were some talk some time ago about altering update-initramfs to suit uboot image file format. Has this been made? [19:45] surely not [19:45] update-initramfs shouldnt have anything to do with handling bootloader specific bits [19:45] thats the job of flash-kernel in ubuntu/debian [19:46] (on arm at least) [19:46] But some apt upgrade packages triggers update-initramfs which in turn needs to trigger uboot changes, right. [19:46] yes [19:46] so update-initramfs makes a call to flash-kernel [19:47] after it rolled the new initrd [19:47] does flash-kernel handle uSD cards as well? [19:48] it handles whatever was set up for your board in the flash-kernel code :) [19:49] flash-kernel is a pretty horrid script wheer every board cooks its own soup, even if that means to duplicate half the code [19:50] hmm. I see how flash-kernel is setup to be called from initram-fs. It's perhaps easier to just make my small little script to do this part [19:50] just add a patch to flash-kernel ... [19:51] and give it to me so i can include it in oneiric ;) [19:54] * ogra_ needs to do some real life stuff now ... [20:20] How should /etc/fstab look like on OMAP3? Mine is empty. (I refer to non-storage lines, like proc tmpfs etc.) === nielsslot_ is now known as nielsslot === Lopi|idle is now known as Lopi