[02:17] Any of you know the proper format is for setting the framebuffer mode on the beagleboard? It seems there are three different formats, and I'm not sure which the ubuntu kernel prefers :| [05:53] L84Supper, there are known issues unfortunately; I've been working with KDE upstream to resolve them [06:06] mcasedevall : good news [06:10] L84Supper, what are you currently trying to build? [06:11] mcasedevall : I was going to try a few 2D graphics apps like Scribus, GIMP, Krita, Inkscape etc this week [06:12] cairo and a few other libs use doubles vs float [06:13] or so I've been warned === bjf is now known as bjf_afk === mcasadevall is now known as NCommander [08:40] high-rez, heh, since there is no ubuntu kernel for the beagelboard its really up to the third party kernel you use ... [08:41] high-rez, http://elinux.org/BeagleBoardUbuntu has a link to properly maintained third party kernels and i would assume the u-boot settings used in the doc match the kernel [08:49] Just fyi: From running build-arm-rootfs script from http://people.ubuntu.com/~ogra/arm/build-arm-rootfs I get two error messages from created installer script within qemu arm: [08:49] http://pastebin.com/d6660cdde [08:49] But seems that installation finishes fine, though [09:23] ogra: From running build-arm-rootfs script from http://people.ubuntu.com/~ogra/arm/build-arm-rootfs I get two error messages from created installer script within qemu arm: [09:23] http://pastebin.com/d6660cdde [09:23] But seems that installation finishes fine, though [09:23] so the script finishes ? [09:23] yes [09:23] then its not critical [09:23] yes, but I just thought to mention it ;) [09:23] if there are actual critical errors the script will die [09:24] yeah, i know about both, thanks ... i'll drop the loopback call that iniscript was dropped at some point [09:25] ok, will drop loopback locally then, too [09:26] the other warning is simply a nonused missing feature in the versatile kernel i rolled [09:26] ok, thanks [09:27] the next generation of the script (called project-rootstock now) will use the packaged versatile kernel instaed [13:03] With http://people.ubuntu.com/~ogra/arm/build-arm-rootfs I built a ubuntu-minimal. Does this include xserver? I'm running it on Overo, and have console on LCD login. [13:04] no, ubuntu-minimal doesntinclude any X bits [13:04] it only installs the bare minimum to boot an ubuntu [13:05] ok, yes, login looks like this ;) [13:05] I have to install ubuntu-destop? [13:05] erm, desktop [13:05] if you want an xserver just install the xorg package if you are lazy [13:05] (that pulls in all possible video drivers though) [13:06] indeed, if you want a full desktop install ubuntu-desktop [13:07] What's the easiest way for it having ubuntu-minimal already. I'm not sure if I can configure Overo for net access easily, so most probably the easiest way is via QEMU again? === cbrake_away is now known as cbrake [16:23] is Gnome or GTK in the plan for Ubuntu-arm? [16:40] I would say that's likely. Ubuntu uses GNOME by default, and there is a Ubuntu Netbook Remix. [16:42] somebody already installed gnome desktop on shivaplug with USB lcd monitor [16:56] L84Supper, our live image for imx51 uses the standard ubuntu desktop [16:56] (the one we released in jaunty) === ogra_ is now known as ogra [17:29] ogra : ICEing the Cortex-a9 kernel on PBX now :) [17:30] ogra : Turns out I can chainload redboot or u-boot from the ARM Monitor [17:30] so, either bootloader will work [17:30] although once I have UEFI running, that's going to be redundant [17:30] lucky you [17:31] ogra : Lucky all of us [17:31] * ogra pastes the famous lasts words of NCommander from #ubuntu-devel [17:31] -rwxr-xr-x 1 mcasadevall mcasadevall 2.8G 2009-07-06 12:30 redboot.bin [17:31] if I can get this working fully, there's a better chance of getting Canonical interested [17:31] The -hell-? [17:31] what did he do to get redboot.bin up to 2Gb+? [17:31] build it with the allowed toolchain [17:32] Martyn, don't try and build redboot with an arm-linux-gnueabi toolchain [17:32] or one variant of the allowed ones we can currently use [17:33] Good lord [17:34] NCommander : But I have done so .. using the CodeSourcery chain... [17:34] M 2009-07-0 [17:34] M 2009-07-0 [17:34] er [17:34] -rwxr-xr-x 1 mcasadevall mcasadevall 497M 2009-07-06 12:35 redboot.bin [17:34] -rwxr-xr-x 1 mcasadevall mcasadevall 1.3M 2009-07-06 12:35 redboot.elf [17:34] I dunno what bugs me more [17:34] That the elf is 1.3M, or the binary is 397M [17:34] *497M [17:35] NCommander : -rwxr-xr-x 1 mbogo users 319022 2009-07-16 18:31 reboot.bin [17:35] Martyn, I can build the right toolchain with normal GCC, its just that isn't acceptable it seems [17:35] NCommander : That's more in line with my experience [17:35] Oop .. wrong binary [17:36] NCommander : -rwxr-xr-x 1 mbogo users 319022 2009-07-01 16:32 redboot_small.bin [17:36] * Martyn hates not having cut and paste between parallels and OSX [17:36] yeah [17:37] Well [17:37] I wonder [17:38] but still, 300M+ is silly [17:38] Martyn, our prob is that we're boud to the toolchain in the archive [17:38] *bound [17:39] ogra, well, I might be able to build the toolchain we need out of the gcc-4.4 package with lool said would be an acceptable alternative if all else failed [17:39] ogra : Ouch [17:39] The question is does the 2.8GB redboot count as "all else failed" ;_) [17:39] ogra : The state of the eabi toolchain needs to be fixed [17:39] NCommander : I think so [17:39] NCommander : Something went very wrong there [17:39] NCommander, depends if you really tried "all else" [17:39] Martyn, so you know the differences between the RTOS/baremetal toolchain, and the linux toolchain :-)? [17:40] NCommander : Some of the differences [17:40] Martyn, right, but you understand that its more or less near-impossible to build redboot with the normal linux one [17:40] (granted a massive universal size fluke) [17:40] NCommander : I'm not tasked to do toolchain fixes at my company until early Aug though [17:41] NCommander : I do, indeed. That's why UEFI is more of a priority, and before it .. u-boot. [17:41] reboot is just going to have to be one of those intractable problems, or simply be left behind. [17:41] Martyn, ok, good, that makes sure I'm not mad [17:41] redboot rather [17:41] ^- ogra [17:41] Martyn, we managed to build redboot with an arm-linux-gnueabi toolchain in Jaunty by linking it against libc [17:41] obviously that was fragile, and now it broke :-) [17:41] Indeed [17:42] it really should be linked against ulibc [17:42] er [17:42] It shouldn't be linked against a C library [17:42] redboot is self-contained [17:42] if you're going to link it, use the smaller one [17:42] we dont support ulibc [17:42] I know, I know [17:42] but you'll get a smaller build. [17:42] th eonly minimal libc we have atm is klibc [17:42] oof [17:42] ogra, newlib :-P [17:43] Martyn, if we had a sane toolchain, the resulting binary won't import any symbols [17:43] the problem is glibc got stupid with the newest release, and started doing fun things w/ ELF files [17:43] well, and newlib in the works which apparently gets you 3G redboots [17:43] Ick [17:44] ogra, er, that was glibc [17:44] newlib fails to link due to static protector [17:44] ogra : In order to suggest a UEFI boot chain for 9.10 .. is there still time to get it working to the point that I could make a case for it? [17:44] since UEFI/arm is almost done [17:45] Apple has put massive resources behind it. [17:46] * NCommander sighs [17:46] -rwxr-xr-x 1 mcasadevall mcasadevall 603K 2009-07-06 12:47 redboot.bin [17:46] Its going the right size [17:46] looks ok for a SD boot to me [17:47] Yep [17:47] might be a bit big for flash ... which we dont use anyway [17:47] ogra : I do. [17:47] in our image builds [17:47] * Martyn only has 64K or 128K of NOR though [17:47] we boot from SD and install to SD [17:47] so it's only big enough for u-boot [17:47] It doesn't boot though [17:48] ogra : One other question: How often does Canonical have a general meet-up to discuss the state and future releases of Ubuntu? [17:48] its probably just slow beacuse its so overweight :) [17:48] 9.10, 10.x, etc... [17:48] Since the ARM side of things is certainly heating up. [17:48] usually about a month to six weeks after the release [17:49] definately [17:49] So 9.10 is already locked in? [17:49] and i would love to see you at next UDS [17:49] mostly, yes [17:49] I also want to be there.. so does my boss. [17:49] (wants me to be there) [17:49] cool ! [17:49] The problem is that we can't use Redboot. [17:50] it's useless on the server platform we're building, and excludes other hardware as well [17:50] I'm hoping to push UEFI [17:50] oooh [17:50] our problem is that we dont have a sane u-boot yet [17:50] That's an improvement [17:50] The board powers on [17:50] But it doesn't print anything [17:50] ogra : What's broken in u-boot? [17:50] NCommander : Working with babbage2? [17:50] misses features [17:50] ogra : Which ones are missing that are needed? (Since I am tasked with u-boot at the moment) [17:50] Martyn, sorta. Bababge2 requires redboot to do some voodoo to let it power on fully [17:50] Martyn, i didnt review it yet, but lool wasnt impressed [17:52] ogra : current u-boot features I have working: Boot from NOR/NAND, boot from SD, ext2 filesystem suport, FAT filesystem support, network adapter (SMCxxxx) support, bootp, dhcp, tftp [17:52] I don't (yet) have USB support working since it's outside my project scope [17:52] on B2 ? [17:53] In software sim for the Realview EB-Cortex A9 [17:53] and Cortex-A8 [17:53] (in uniprocessor and SMP) [17:53] ogra, er, can we publicly talk about that uboot? [17:53] I could trivially get it working on B2 [17:54] NCommander : Although to bring B2 up, I have to do hardware voodoo to get various parts of the board up [17:54] also I don't (yet) have FB console support [17:54] so u-boot would only work on serial, which is annoying [17:55] not more annoying than redboot [17:55] and loads easier to compile [17:56] No, but that's something that would be useful [17:56] Tell you what, why don't I spend tomorrow trying to get it running. [17:56] At worst, I waste 6-8 hours on the experiment [17:56] sure ... put a shiny little dog on the framenuffer during init :) [17:56] ogra : The lack of INT13 is a pain [17:56] ah, damned, someone had that idea before :) [17:56] For someone who is used to having BIOS video [17:57] did you take a look at the beagle code ? [17:57] they have at least a pic up early [17:57] orga_ : ok, so there is support for OMAP and i.mx, what board is everyone using for i.mx development [17:57] ? [17:58] there is no support for OMAP [17:58] framebuffer costs a character bitmap table for ASCII, framebuffer init code, and either USB+keyboard stack(big) or raw USB keyboard stack(smaller) [17:58] there is no official support for OMAP [17:58] right [17:58] orga_ : hmm, somebody else mentioned OMAP [17:59] but that just requires a kernel change. The bootloader might be flexible enough [17:59] ok , just not official [17:59] Martyn, i didnt say frambuffer console :) no need for an ascii table if you just dump a pic on the screen [17:59] OH! [17:59] I'm going to try it on ARM 9 [18:00] Heh, that's just useful to say that you entered the bootloader. [18:00] the beagle just shows a fillscreen bitmap [18:00] *full [18:00] I think pretty bootloader pictures are a waste [18:00] Samsung S3C2440 [18:00] One of my coworkers has the B2 all day, damnit [18:01] steal it ! [18:01] no, he's doing compiler optimization [18:01] * ogra had to send his away to a colleague too :/ [18:01] WAY more important than the work I'm doing [18:01] not for me [18:01] the netbook remix was great unless you had to configure something like networking on a 320x480 display [18:01] tell him your work is more important for me, he should give you the B2 immediately ! :) [18:01] L84Supper : Don't get us started on Netremix [18:01] you're sunk unless you can get panning going [18:02] ogra : LOL .. he's grumbling because I have access to the two PBX'es [18:02] trade them ... [18:02] ogra : Even though he can't do any work on those until I get the kernel working [18:02] who cares about PBX [18:02] I do. [18:02] those are the base platforms //all// Cortex A9 will be based on [18:02] even the NEC, Samsung, and other C-A9 based systems [18:02] pfft ... thats the future ... [18:02] my real only beef is the config screens not fitting vertically in the window so you have to guess at how many times you have to press "tab" to apply or cancel [18:03] * ogra wants something that works *today* :) [18:03] ogra : And it's the future I live in .. so I that I can get U/ARM working on as many platforms as possible [18:03] C-A9 is where everything is going, and everything I'm working on is backportable. [18:03] (like UEFI) [18:04] L84Supper: Don't take this the wrong way, but that discussion really belongs in #ubuntu-devel [18:04] There's lots wrong with the remix, and grumbling about it (without fixing it yourself, or posting patches) is counterproductive [18:05] * Martyn dances! Kernel boot! [18:05] note that its aimed at 1024x576 and wont be much more scalable downwards [18:06] It crashes hard after cpuidle .. but after three days, I finally have debug output. [18:06] WOOT [18:06] I might even get to init() today [18:06] congrats [18:06] now grab the B", quick ! [18:06] *B2 [18:06] Cant :) [18:06] tomorrow, I have it [18:06] :) [18:06] And I'll try getting u-boot working with a set of params [18:07] Martyn : [18:07] I also should figure out a 'safe' u-boot set of parameters, so that someone can set one of user DIP switches to get a sane default setting [18:08] what's the dev board for imx5? [18:08] since I already locked myself out of one development system by accidentally setting delay to 0 [18:08] babbage/babbage2 [18:08] i.mx51 [18:08] and i.mx515 has another one [18:11] lol [18:11] Martyn : any idea where they are for sale? [18:12] s/where/when/ :) [18:15] seems genesi has mx515 devboard orderable https://www.genesi-usa.com/store/details/4 [18:17] everything is open except for the closed graphics libs by PowerVR, I heard that Nokia was asking them to open up [18:18] wow $750 for the dev board [18:18] dev boards cost that or more [18:18] sure I've spent thousands [18:19] * ogra thinks 750 is rather mid-price [18:20] $750 is very reasonable [18:20] Genesi USA appears to be grasping for credibility [18:20] L84Supper: I >will< have 2D drivers for powerVR working by the end of Aug [18:21] L84Supper : 3D is being worked on by other people. [18:21] and will be as nonfree as nvidia [18:22] were still working on the first complete open source laptop, VIA vx855 still looks like it will be the first [18:23] ogra : The 2D drivers will be open and free. Those are reverse engineered [18:23] ogra : The 3D drivers .. who knows. [18:23] yes a pain [18:23] Martyn, well, its equivalent to puolsbo on intel [18:23] ogra : Depends if the Finnish guys figure it out (PowerVR open source) before the the offical closed-source drivers gain traction [18:25] That genesi board looks pretty close to the babbage2 [18:25] just about the same capabilities [18:26] I wish that the genesi board had more RAM [18:26] Martyn, not really, but to another one here on my desk :) [18:26] not to say its 100% the same :) [18:27] just differntly named [18:27] so if you want to do ubuntu stuff, get such a board ;) [18:27] they both look like the Freescale reference design [18:30] so many components on the PCB.. do they really intend to manage to make cheap designs around that? [18:30] it's a kitchen sink board [18:31] it also demos components by partners [18:32] the netbooks will just have the imx soc , an EC, some glue and power supply [18:33] wireless, bluetooth etc will be on separate plug-in modules [18:34] I would assume moving components to separate modules doesn't cut down costs [18:35] it gives the ODM's flexibility for different models [18:35] http://www.via.com.tw/en/initiatives/spearhead/surfboard_c855/index.jsp [18:35] as an example [18:41] can the Flash ROM be reprogrammed by software in-circuit or does it require a separate hardware device? [18:41] on the babbage board [18:42] if it may be done via software we can look at adding it to FlashRom [18:48] in-circuit [18:48] heck, even live from an OS [18:49] ubuntu currently uses SD [18:49] the flash on the B1 was to small to fit in the initramfs [18:49] It shouldn't have to live on that flash [18:50] but we should be able to boot from either the SD or an internal drive [18:50] I still say kernel chainloading is the way to go [18:50] yeah [18:50] no [18:50] why not? [18:50] * ogra isnt a fan of kernel chainloading ... [18:50] Why? [18:50] means you have to maintain a second kernel [18:50] the kernel itself is the ultimate BIOS [18:50] sure, but it's a tiny kernel [18:50] still [18:51] no different than maintaining a bootchain [18:51] and you KNOW you will be able to flexibly boot [18:51] well, until we have UEFI .. that's one of the only ways I can think of to really get a flexible boot [18:52] or HW vendors that agree on a standard :) [18:53] Coreboot [18:53] coreboot on ARM = nonexistent [18:53] yes, we've talked about it [18:53] it would take a LOT more work to get coreboot's stuff going [18:54] well uboot handles the hardware init [18:54] okay back to work [18:54] * ogra saw a coreboot demo at CELF [18:54] on a beagle [18:54] was probably Peter [18:54] is a boot drive chooser still missing? [18:55] what is the command to get a system to execute an ACPI suspend in linux? [18:55] UEFI is another train wreck [18:55] yeah, peter stuge i think [18:56] yes , that him [18:56] carebear in irc [18:57] they were just at LinuxTag as well [19:02] I'd stay with uboot and a chooser for boot location [19:18] ogra: I tried --seed lxde,gdm , this results in "E: Couldn't find package lxde,gdm" === cbrake is now known as cbrake_away