[10:50] ogra: do you guys plan to roll Pegatron images too? considering the slight changes required to redboot... [10:51] if i can get the redboot patches in plain form i will include redboot binaries for pegatron and it would be nice to have kernel support too, we wont build specific images though, but we would like to enable oters to do so [10:51] I would like to split the images so that we ship a small device dependent image and load the squashfs from a separate image [10:52] We need to revisit the image format; SD images aren't particularly useful for real lange devices [10:52] yep [10:52] though thats karmic+1, right ? [10:52] That's as soon as possible but not a karmic goal in itself :) [10:52] or do you plan it for this cycle ? [10:52] good [10:53] * ogra would like to have images at all before thinking about a redesign :) [10:54] We do now [10:55] Just with broken casper (Fix committed) and kernel [10:55] right [10:55] once i see it booting to gdm and can use mouse kbd and network i'm happy [10:56] lool, in any case i dont want to have a cycle again where we dont start building images with A1 [10:57] I wish, but I doubt it will be much faster to rebase on top of .33 or whatever is used in 10.04 [10:57] even if that means to carry the kernel of the last release for a while [10:57] Perhaps for dove [10:57] well, with the sperate kernel sourcepackages it should be a bit easier [10:57] so you worked around the squashfs problem or was it fixed? [10:57] no [10:57] neither [10:58] its broken atm [10:58] how can I reproduce it? Since I don't typically install the entire system to test kernels... [10:59] hmm, well, you would need to pul the live image [10:59] then change the cmdline and add LIVEMEDIA=/dev/mmcblk0p2 [10:59] link? [11:00] http://cdimage.ubuntu.com/ports/daily-live/20090812.4/karmic-desktop-armel+imx51.img [11:00] straight dd to SD? [11:00] yup [12:10] argh [12:10] * ogra curses git [12:11] so my day is messed already having to use that crap [12:14] so no matter what i do to get the mvl-dove branch it simply trashes what i have already [12:14] what are you trying to do? [12:15] lool wanted me to do a testbuild and i simply clone the ubuntu-karmic branch, then try to checkout the mvl-dove head (which apparently isnt included) [12:16] ogra: what does 'git branch -r' say [12:16] so i started lookinf for docs that confuse even more, isnt there a simple clone command i can use to get directly what i want ? [12:16] ah, it has origin/mvl-dove [12:16] but note i messed it up already editing .git/config [12:17] and running git pull afterwards [12:17] so just do 'git checkout origin/mvl-dove -b mvl-dove' [12:17] which results in a broken merge [12:17] NEVER EVER use git pull. Don't you bzr heads ever learn :-p [12:18] well, if git would be usable in any small amount we wouldnt need bzr :P [12:18] Here are the commands to fix your setup: [12:18] * ogra starts over with a clean clone [12:18] oh, ok [12:18] * ogra listens [12:18] git checkout master [12:18] git reset --hard origin/master [12:18] * amitk waits [12:19] ok [12:19] Switched to a new branch 'mvl-dove' [12:19] ha ! [12:19] git fetch [12:19] oh [12:19] ok [12:20] what does 'git branch' say? [12:20] ogra@osiris:/var/build/kernel/git/ubuntu-karmic$ git branch [12:20] master [12:20] * mvl-dove [12:21] git reset --hard origin/mvl-dove [12:21] just to be sure you are on tip of mvl-dove [12:21] HEAD is now at 32e3175 UBUNTU: [Config] Bypass udeb generation issues. [12:21] looks good [12:21] now you can carry on building [12:22] yep i guessed so [12:25] amitk, thanks a lot, git is like speaking latin to me [12:26] actually it _is_ due to the archival nature of internet. Everything published on internet should have a expiry date. All the bad docs would disappear then. [12:47] hmm... why does the karmic fsl image switch the console after initramf [12:47] s [12:47] ?? [12:47] can you elaborate ? [12:48] http://pastebin.ubuntu.com/252445/ after this I get nothing on my serial console [12:48] everything is redirected to the DVI output [12:49] where I see initramfs errors [12:50] oh, right, because its a live image [12:50] edit the cmdline [12:50] drop console=tty0 [12:51] though given that upstart has no default setup for serial ttyS i doubt you will get a lot more (beyond the errors) [12:51] lool, the mvl-dove kernel doesnt boot [12:53] lool, whoops, i take that back, wrong mkimage parameters [13:14] lool: you comment on the squashfs bug: babbage image doesn't find its livefs by default (due to a driver renaming its /sys data between 2.6.28 and .31) [13:15] that should be the squashfs driver? [13:15] no, thats the casper bug he was referring to earlier [13:15] so now it does find its livefs by default? [13:16] we currently look for platform-mmc devices but that device was renamed [13:16] the casper fix adds the new name to it i belive (i didnt check it) [13:16] you can override the broken casper detection by setting LIVEMEDIA=/dev/mmcblk0p2 on the cmdline [13:17] the detection isnt realted to the squashfs oops [13:17] its two separate issues [13:17] its almost impossible for me to debug this bug w/o a separate DVI display. Can't keep switching between the board and my devbox. Do you have any other trick? [13:18] sadly no [13:19] dropping console=tty0 didnt help ? [13:23] it helps. But I don't have enought knowledge of initramfs [13:23] where does the stdin: error 0 come from? [13:26] aah, nevermind. passing LIVEMEDIA makes it go away [13:38] amitk: There is a casper fix which isn't uploaded which is why you need the LIVEMEDIA arg [13:38] Main console is on DVI by default but you can easily change that on the kernel cmdline [13:38] lool, dove looks very good so far [13:39] ogra: What did you test? [13:39] its nearly finished doing a kernel compile [13:39] never got that far before, should be done within the next 15min [13:39] ogra: it better be. They delivered a 2.6.31 patchset... [13:40] well, we had weird segfaults and ICEs with all former versions [13:40] nothing like that anymore it seems [13:40] and since i built a .deb from our git tree i'll try initramfs booting next [13:41] it still doesnt see the 512M though [13:44] ogra@dove:~$ sudo hdparm -t /dev/sda1 [13:44] /dev/sda1: [13:44] Timing buffered disk reads: 170 MB in 3.15 seconds = 53.99 MB/sec [13:44] * ogra really loves that [13:47] ogra: did you say there was a dove board to be delivered to me too? [13:47] we had a spare one but you guys were all gone already [13:47] not sure where that went now [13:48] lool, do you remember ? ^^^ [13:52] mage arch/arm/boot/uImage is ready [13:52] HA ! [13:53] lool, i'd consider the stability issues solved [14:23] ogra: Do you have CONFIG_HIGHMEM on your kernel? [14:23] ogra: How much memory do you see in free? [14:23] or /proc/meminfo rather [14:24] ogra@dove:~/temp$ cat /proc/meminfo [14:24] MemTotal: 286664 kB [14:24] # CONFIG_HIGHMEM is not set [14:24] aha [14:24] Hmm [14:24] * ogra tells rtg [14:25] lets get it set for one testbuild [14:25] I don't understand where these 256 MB are [14:26] do you need anything more ? else i'll try a reboot with initramfs [14:27] No I don't thanks [14:47] hmm [14:47] Wrong Ramdisk Image Format [14:47] i wonder why [14:47] i can load it just fine but it panics [14:50] cat /ogra@dove:~$ cat /proc/meminfo [14:50] MemTotal: 514248 kB [14:50] hmm, works if i set mem=512M [14:50] * ogra tries to complie something to see if it stays stable [15:06] ogra: Are you using the latest U-Boot? [15:07] You want the latest one [15:07] ah, no [15:07] do we know how to safely flash it ? [15:07] i dont want to trash my board now that it actually runs stable [15:09] Not sure [15:09] * ogra tries another kernel build with mem=512M [15:10] lets see if it keeps stability [16:11] Anyone know a good ubuntu-arm-vfp rootfs when I can download. [16:27] vfp nope === bizkut-miau is now known as bizkut