[00:04] pbuckley: Looking at the specs, there's a fair chance that any work done by people on the tf201 might Just Work on the tf700, but hard to tell without hardware. [00:04] That display sure looks nice, though. [00:11] cool ill check out whats going on with tf201 and see if i can use any of it as a jumping board [00:11] yeh the display is beautiful [00:11] loving the hardware.. android isn't cutting it though === zyga-afk is now known as zyga [11:55] ogra_: now tried the netboot installer several times an two brand new 16g sd cards. never finished, always failed during the "select and install software" step. [11:56] weird, worked for three people during the past two days [11:56] anything in dmesg or syslog ? [11:57] ogra_: no idea. will try the preinstalled server image now [11:57] (syslog is on tty4 if you do a console install, on serial you need to check the file) [11:57] ogra_: i can look on one of the two cards in a few moment [11:57] there is no preinstalled server image anymore [11:57] its dead [11:57] only a normal daily server [11:58] ogra_: i've got a copy of the 12.04 omap4hf already here. [11:58] (all arm images are now identical to their x86 pendants, desktop -> live, server -> alternate with squashfs, netboot -> well, netboot) [11:59] if you downloaded anything called preinstalled thats several weeks old and 99% positively uninstallable [11:59] ogra_: strange, my ubuntu-12.04-preinstalled-server-armhf+omap4.img.gz here is the only thing that ever worked. [11:59] oh, 12.04 ... sorry [12:00] ok, got one of the cards with failed netboot ready now. [12:01] the file in use was boot.img-serial.gz [12:01] and you zcat'ed it as root to the SD [12:01] ? [12:01] sure. [12:03] ogra_: you meant checkout the second partition on the sd card and there /var/log, right? [12:03] looking at http://ports.ubuntu.com/ubuntu-ports/dists/precise/main/installer-armhf/current/images/omap4/netboot/ the serial image definitely has a partition [12:04] i really have no idea whats broken. it never finished the installer successfully. [12:04] hm, now preinstalled also fails with /dev/mmcblk0p2: UNEXPECTED INCONSISTENCY; RUN fsck MANUALLY. [12:05] something is really going on here. [12:05] booting ... [12:06] *sigh* now giving today's boot.img_serial.gz one more try. [12:06] logged in as root, i do "zcat boot.img-serial.gz > /dev/sdd", then sync [12:06] todays should definitely work [12:07] i'm starting to believe in bad karma, broken feng shui or black magic. [12:07] heh, probably [12:08] hmm, my install doesnt seem to progress [12:08] and i have a suspicion ... [12:08] booted fine, language and locale select fine, network config fine. [12:08] oh, it moved now [12:08] just took a while to do whatever it did [12:08] archive setup fine [12:09] "Loading additiona componnents" now [12:10] user setup done [12:11] partitioning done (picked guided for the usb key) [12:13] user setup done, no home encryption [12:13] same here [12:13] i largely just go with the defaults [12:13] same here. [12:13] its installing the base system now [12:13] guided partitioning, using MMC/SD card #1 (mmcblk0) - 16.1 GB SD SDC [12:14] oh, not sure that will work, good luck :) [12:14] bug 1025420 [12:14] Launchpad bug 1025420 in ubiquity "AssertionError: Returned to partman/choose_partition with nothing to do" [High,New] https://launchpad.net/bugs/1025420 [12:14] though thats a quantal desktop install [12:15] but they both use the same backends [12:15] he picked a pre-created partition on a live image though [12:23] seems like it is stuck doing mkfs.ext4 on mmcblk0p2 [12:26] give it some time ... big slow media etc [12:27] too much time already. [12:27] sd led also shows no activity. [12:27] ah [12:27] my LEDs are both solid green here [12:27] i dont think we have the modules in d-i to use them [12:29] just to be sure, redownloaded qq's boot.img-serial.gz now and starting over again. [12:29] well, installing to mmc will likely have the same issue there [12:29] and i bet if you check the logs you would see a similar error as in the bug [12:36] now its "installing the base system" [12:37] to MMC ? [12:37] yep [12:37] intresting [12:37] nope, installing. [12:37] did you keep the existing partition intact ? [12:37] nope. [12:37] heh [12:37] oh, but you created a vfat in the beginning of the SD i hope [12:38] it wont do that automatically [12:38] it creates an ext2 in the beginning. [12:38] thats /boot [12:38] doesnt help [12:38] you need a vaft to boot from [12:38] ah yes, now we're getting closer. [12:39] if you dont have that partiton flash-kernel-installer will fall over [12:39] it tries to mount -t vfat i think [12:39] LetoThe2nd: I have seen the unexpected inconsistency message every second boot on my tf201, no idea whats causing it but I wonder if there is something going on (booting with single and fscking the volume fixes it) [12:39] so basically the automagic partitioning does not work, and one needs to take care that a vfat partition with, say 64mb or such is in place, right? [12:40] automagic partition only works if you install to USB and leave the SD as "boot floppy/fake NAND" [12:40] lilstevie: there's more going on here. :( [12:40] hm :/ [12:40] if you want a different setup you need to know what you are doing [12:40] ogra_: ok. will do that next then. [12:41] the automatic creation of /boot is plainly silly [12:41] hey i'ma tyrant god emperor. i do not have to know what to do. i have minions for that! *SCNR* [12:41] but the partman code is scary .. i need to be in the right mood to get at it :P [12:41] massively drunk? [12:42] for three days ... yeah [12:43] minimum [12:44] do you think being drunk is enough? [12:45] resizing the terminal during serial netboot install is funny, talking ncurses-interface layout wise. := [12:45] after three days being drunk i dont think anything ;) [12:45] thats the point ;) [12:45] might need some form of hallucinogen to help there [12:45] LetoThe2nd, screen or minicom ? [12:45] lilstevie: please, not engelbert humperdinck records again! [12:45] ogra_: screen [12:46] LetoThe2nd: lol [12:47] lilstevie: "having a bad trip" [12:47] lol === scottb` is now known as scottb [14:58] ogra_: ok, now i've proceeded through the complete netinstaller and kept the first fat partition intact (use largest contiguous space). still does not boot - where to go hinting in the logs? [15:00] pastebin it somewhere (sudo apt-get install pastebinit;; pastebinit /var/log/installer/syslog ;)) [15:00] http://paste.ubuntu.com/1112040/ [15:01] hmm, looks all good [15:02] how does it not boot ? [15:02] what are the symptoms ? [15:02] power on -> led2 solid. [15:02] notihing on serial ? [15:02] exactly that [15:02] thats pretty strange, MLO and u-boot wouldnt have changed [15:02] check the vfat partition? [15:03] yes [15:03] http://paste.ubuntu.com/1112047/ [15:03] the installer only updates boot.scr, kernel and initrd [15:04] right, the timestamps agree [15:04] since MLO wasnt changed you should definitely at least see MLO messages on serial [15:04] negative. [15:05] http://paste.ubuntu.com/1112052/ [15:06] hmm [15:06] * ogra_ is a bit clueless what to say [15:07] from a binary POV nothing should have changed for MLO [15:07] its a fat, files dont move around or anything [15:08] i told ya. bad karma, or broken feng shui. [15:09] boot.img-serial1 * 1 61440 30720 c W95 FAT32 (LBA) [15:09] just retried including starting the terminal over again. led2 solid, nothing else. [15:09] hmm, the data has definitely changed [15:10] i suspect parted rewrites the partition table [15:10] since you created one [15:11] ok, so the guess would be - what? [15:11] oyu should probably pre-create one and select to not format it in the partitioner ... then it shouldnt touch the part. table [15:11] as a workaround [15:11] and the bug would be against partman i guess [15:11] the fat partition? it basically *should* not get touched, as i nowhere told the installer to do so. [15:12] no, not the partition, the partition table [15:12] ogra_: so zcat to sd, manually partition in the pc, then start all over? [15:12] compare my fdisk output with yours for the vfat partition [15:13] right [15:13] my guess would be that this works [15:13] seems the rom doesnt get along with the differently computed partition table [15:13] or some such [15:14] the image is originally created by fdisk ... while the installer uses partman [15:14] (which handles CHS differently for example) [15:14] ok, lets see. [15:16] ok, when repartitioning with gparted after zcat'ing boots. [15:21] my install to USB finished properly btw === XorA|gone is now known as XorA [16:06] ogra_: seems to boot now. [16:06] \o/ [16:53] LetoThe2nd, awesome ! [17:21] marvin24, did you commit yesterday's resume problem fix? [17:22] marvin24, I just checked your new branch compared to what we have in ubuntu, quite a diff. You seem to have added a lot of commits since Jun 8 when the previous package got branched off [17:22] did you add rt2000 driver fixes too besides strictly ac100 stuff? I want to make sure I am not mishandling git log here [17:22] :) [18:12] janimo: use the l4t-r15-ac100 branch [18:12] and yes, I updated some drivers we use [18:12] rt2x00 and cleancache [18:13] marvin24, ok. thanks, I'll use that [18:13] too bad gitorious is so slow [18:13] and sometimes disconnects [18:15] yes, that's the "non-commercial" way ;-) [18:16] janimo: I'm thinking about uboot ... [18:16] what is the prefered way to boot the kernel (where to store, initrd, boot.scr?) [18:17] preferred way with uboot? [18:17] the cases I know have a separate partition for uboot conffiles and kernel+initrd [18:20] I wonder if we should use the boot partitions for this [18:20] if it works I don't see why not. [18:20] it can be restored easily [18:21] and what about an efi partition table? [18:21] We usually store the kernel and initrd on the same flash/partition as u-boot.bin, but that's only for historical reasons, because u-boot used to suck so terribly hard at loading from ext* filesystems. [18:21] There's really no reason we should continue that brain damage, IMO. [18:21] I think it still only understands ext2 [18:22] or ext4 without some features I guess [18:22] For read-only, it's good enough on ext4, afaik. [18:22] As for efi, I don't think u-boot speaks gpt at all yet. [18:23] But that's fine if you load u-boot and the kernel externally, and let the kernel sort out partitions. [18:23] I tried loading the kernel from my ext4 partition, but failed [18:23] it could only see the root fs, but /boot gave garbage [18:23] gpt is supported AFAIK [18:24] chromeos also uses it and likely all new devices also [18:24] * marvin24 goes and checks === XorA is now known as XorA|gone [18:25] seems it should work as long as the storage is smaller than 2 TB ... [19:30] marvin24, use vfat [19:31] ogra_: never ... [19:31] ext2 is fine I think [19:31] for internal mmc [19:31] vfat is the most stable filesystem in u-boot, as sad as it sounds ... [19:31] even read only ext2 doesn't work? [19:31] zzz [19:32] but ext2 is usually ok too, i have seen it failing on arches where iot was supposed to work though [19:32] (some mervell ones, but they used their own u-boot fork i think) [19:32] *marvell [19:46] yeah, we can use mainline :-) [21:21] marvin24, I rebased on your new branch and want to check with you on the config changes introduced as I see it here [21:21] +CONFIG_REGMAP=y and REGMAP_I2C=m are new [21:21] XVMALLOC ZCACHE and ZRAM got turned off [21:23] ogra_, what keys to use to boot into recovery on the ac100? [22:22] ogra_, ah you removed the 30 minutes text from the installer, nice :) [22:23] * RoyK smacks his raspberry pi [22:25] janimo: I think he should have taken an average timing of 5 installs, and ammended the message to "this process may take around 13.73 minutes, ish." [22:26] it is 5-6 minutes in my tests. Not sure if the USB stick or the internal mmc is the bottleneck [22:26] but its not slow at all [22:27] Yeah, I know. I just pulled 13.73 out of my ass. :P [22:28] I figured. Ithought 5 - as I experienced should be representative for all, but forgot the usb stick/sd card used has an influence [22:29] infinity, I'll do some more work on mono tomorrow and in the weekend, today and yesterday I did not do anything.