[00:40] using the 1-Aug-2012 omap4+armhf image on a pandaboard ES. it installs the rootfs but leaves the uboot partition empty. is there a work around or shall i file a bug? [02:29] mpls [02:31] mpls ? [02:34] jimerickson, a bug would be helpful indeed [02:34] then a bug it shall be! [02:56] filed bug #1032021 [02:56] Launchpad bug 1032021 in ubiquity "installer does not create a uboot partition on omap4+armhf on pandaboard ES" [Undecided,New] https://launchpad.net/bugs/1032021 [03:07] jimerickson, looks like the installer didnt finish at all, did you get any error messges ? [03:08] just ubi-partman failed with error 141 at one point. i tried it several times. [03:10] right, it didnt proceed after that [03:10] correct [03:11] also yuo seem to have checked the option to install the nonfree stuff (on the first page), this tries to instll the pvr driver which isnt ready for the current kernel yet [03:12] ah ok i can try it again without that if you wish [03:12] (thats unrelated to the partitioner issue indeed) [03:13] no. it shouldnt cause the install to fail or crate partition errors [03:13] ok [03:16] hmm, something created a vfat on your usb drive as first partiton [03:16] did you dothat manually ? [03:17] well at one point i created a fat32 partition in hopes it would populate it [03:18] no, it cant boot off the usb driver [03:18] *drive [03:18] the u-boot partition needs to live on the SD [03:18] (imagine the SD as a bootfloppy) [03:18] right but how else can i install except to a usb hub? [03:19] well, the install as you did it was surely not wrong, but the vfat on USb would be wasted space [03:19] ok [03:20] the bootloader gets installed to the SD card at the end of the installation [03:20] i see [03:20] so you dont need to have a vfat on USB [03:22] i would try without checking the third party driver option and just letting the partitioner take over the whole usb disk automatically (guided partitoning), see if that works [03:22] will do ogra. i will report back in a few. [03:22] ok [03:31] i'm not sure about all the debconf messages in your syslog, will have to ask cjwatson about these tomorrow, no idea if they are fatal [03:32] ok [03:33] its installing right now. no errors so far [03:33] good [03:34] i need to update the install docs on the wiki to actually explain that you need the SD for the bootloader ... [04:12] well it made a 255 mb ext2 partition and the rest ext4. but no fat32. so it does not boot. [04:16] ogra_, is there something else i am doing wrong? [04:27] jimerickson: Hrm? The vfat partition it uses to boot is on the SD card. === Ursinha` is now known as Ursinha [04:27] ok then how do i stop it from booting to the installer? [04:28] jimerickson: It overwrites the installer bits on the card. === rsalveti` is now known as rsalveti [04:28] its not doing that. would you like syslog and partman? [04:29] jimerickson: No, but ogra_ would love more info, I'm sure. :P [04:29] ok [04:33] infinity let me get this straight, the sdcard is the boot disk and i leave a usb stick in the hub to boot from now. no more installing to sdcard? [04:49] nevermind i am confused. [05:50] install halts with /dev/cdrom cannot be unmounted. [05:52] ogra_: mpls means I forgot to change the focus from irc to the terminal. Happens. [06:22] marvin24, on precise kernel+L4T work well indeed. I'll check a newer X [07:24] howdy [07:24] i got a strange problem on my device [07:24] kernel freezes on changing the console [07:25] when i try to change from tty1 to tty2 [07:25] happens not always [07:25] but if system is under load almost always === zyga-afk is now known as zyga [08:19] janimo: I wrote you yesterday, that removing the xorg.conf (on debian) helps. This makes sense as in this case the framebuffer driver is used 8-( [08:39] marvin24, ok. I have not upgraded to quantal or X 12 yet [08:39] I need to make some more GLES measurements with known good setup [08:40] janimo: did you tried multi board support for 3.1? [08:40] that won't work [08:40] yes, does not work [08:40] do you know the reason? [08:40] I fI build in paz00 config even with the actual board files not built-in the image (ifdefed out or removed from the makefile altogether) I get a non-booting zImage [08:41] no idea why [08:42] the only weird diff between the two symbol lists is setup_machine_tags gets a constprop.6 suffix in the PAZ00 enabled image [08:42] on #tegra I was told multiboard setup in mainline 3.1 should work [08:51] janimo: the bootloader is a bit strange I think [08:51] it's not kernels fault [08:51] you could try u-boot though [08:53] mainline u-boot works out of the box on all nvidia boards (including paz00) [08:53] I am trying to boot on tf101 [08:53] which fails if I add PAZ00 to the kernel [08:54] so you think it is the bootloader? Does it do more than just pass a fixed machine ID and let the kernel run? [09:06] janimo: I think it reads the machine id from the kernel [09:07] at least on paz00 [09:07] or does something else weird [09:08] so much for having a standard way of bootloader->arm-kernel interface I guess? [09:09] janimo: fastboot was never designed to boot multiple machine kernels [09:09] each OEM has its own version, fit for a single device [09:09] still, why would it not use the simple way of telling the kernel which mach id is it? To use same fastboot binary on multiple zImages? [09:09] multiple devices that is [09:09] janimo: don't ask ... [09:10] all of the stuff which comes from nvidia is quick-and-dirty hack [09:10] that's the price you have to pay if you want to go quickly to market [09:10] you just cannot do it sane [09:11] we have to accept this [09:18] marvin24, which exactly are the two issues you asked from nvidia? One is the plymouth crasher which we work around via console=tty1. The other? [09:18] segfault [09:18] nvidia seems to know about already and working on it [09:18] so we just have to wait ... [09:21] ah, the X 12 ABI === XorA|gone is now known as XorA [09:24] janimo: if console=ttyS0,.... is specified, is the framebuffer console used at all? [09:25] never tried that [09:25] I think that's a better workaround as it should also kill all the error messages still left [10:45] marvin24, has anyone tried to figure out why all the unity stuff is transparent with unity-gles and the tegra drivers [10:46] lilstevie: I know it is not tegra specific [10:46] try xorg-edgers to find out :-) [10:47] I even don't know if it is driver or unity fault [10:47] and no one seems to care currently [10:48] marvin24, heh I was more asking on the nvidia side of things [10:48] I mean using unity-2d is no issue, just wondered if anyone actually knows what the issue is [10:48] and my answer is that it is propably not nvidias fault this time [10:49] yeah, that was what ogra said the other day too [10:49] nvidia doesn't exactly have an easy way to report bugs either :/ [10:49] e.g. https://bugs.launchpad.net/ubuntu/+source/mesa/+bug/1024768 [10:49] Ubuntu bug 1024768 in unity "[xorg-edgers] Missing icons in unity launcher and app switcher after update" [High,Confirmed] [10:50] lilstevie: yes, that's unfortune [10:50] on the other hand, sometimes you can grab an engineer on #tegra who has access to the internal bug reporting system [10:51] hm [10:51] I am not in #tegra :p [10:51] I just talk with someone [10:51] * lilstevie joins [10:51] he said that the two main problems we are facing on the ac100 (console and abi 12 segfault) are already known [10:51] and they are working on a fix [10:52] ah cool [10:53] I haven't tried quantal here yet, so I haven't seen abi12 issues [11:43] infinity, I was told by Jo Shields Sony has some armhf patch for mono which they do not intend to upstream though. I hope that is good for Debian then for the moment [11:43] janimo: I don't care where the patch comes from, if we can ship it... [11:44] (And if it works) [11:44] well if Debian can ship it we can too [11:44] I'd rather wait till they sort it out [11:44] Well, yes. I meant "if the license is free". [11:44] yes, Jo said GPL [11:45] Sounds shiny. If it's not wildly unauditable, we can probably fix precise with it too. [11:46] why wouldn't they want to upstream it [14:52] infinity, i would appreciate some comments (i surely forgot a few checks or so) http://paste.ubuntu.com/1125314/ [14:53] (thats a wget initramfs script for dumping the images onto the SD and reboot into d-i/ubiquity) [14:53] i need to make sure it doesnt fall over once i left boston :) [14:54] No DNS at that point in the boot, I guess? [14:55] I guess if cdimage moves, they're smart enough to fix it. [14:55] yeah, i would have to add dns clients, but the image server is on the internal net anyway and wont change IP [14:55] and the imageurl option comes from the pxe config so its easily changeable [14:56] imageurl=) <-- ITYM imageurl=*) [14:56] oh, indeed ! [14:56] funny that my test worked ... i shouldnt have had access ot cdimage during the test [14:57] And ${imageurl=##} is much more readable than the form you used. [14:57] ok, i always steal that from the framebuffer script :) [14:57] we should fix it there too then ;) [14:58] Or, wait, I could be on crack. [14:59] I am, but so are you. :P [14:59] heh [15:00] You want ${opt#*=} [15:00] Which suddenly makes much more sense. [15:01] lol, yeah [15:01] * ogra_ blames jetlag :P [15:01] And add a bs=4M to that dd(1) if you don't want it to take a week. [15:01] it takes about 3min [15:02] but yeah, adding it [15:02] Assuming the dd in use takes bs=, but I imagine it does. [15:02] its the normal dd [15:02] no busybox stuff [15:02] ogra@horus:~/tmp$ ls -l bin/dd [15:02] -rwxr-xr-x 1 ogra ogra 3140 Aug 1 22:47 bin/dd [15:02] Kay, other than that, looks okay. [15:03] great, thanks a lot [15:03] * ogra_ rolls the initrd and tries [15:03] ogra_: That's the klibc dd, not the coreutils one. [15:03] oh, ok [15:03] And it doesn't take bs= [15:03] Just tested. [15:03] So, don't do that. :P [15:03] we'll i'll know in a minute if it works [15:04] ah, k [15:04] (base)adconrad@cthulhu:~$ /usr/lib/klibc/bin/dd if=/dev/zero of=/dev/null bs=4M [15:04] /usr/lib/klibc/bin/dd: bad operand `bs=4M' [15:04] Oh, wait, it just doesn't do the "M" bit. [15:04] well, it seems to use a decent blocksize by default ... else it wouldnt be that fast i suppose [15:04] But yeah, 3m is fine. Whatever. ;) [15:04] 3min are not bad for a 600M image [15:07] works like a charm :) [15:08] hmm, the ipconfig output actually shows the dns server [15:14] well, but userspace bits are missing ... [15:14] Which userspace bits? [15:14] /etc/resolv.conf ? [15:14] no idea [15:14] Oh, if you're using a wget linked to glibc, yeah. [15:14] klibc could resolve fine [15:14] ah [15:15] well, not worth the effort [15:15] Probably not. [15:16] * ogra_ now has all bits ready for the automated testing ... just need to glue them together somehow now [15:32] infinity: apt is definately borked WR sources.list arch qualifiers. lp#1032250 [15:33] I've seen it before, ages ago actually, but never really got to the bottom of it before [16:34] bug 1032250 [16:34] Launchpad bug 1032250 in apt "apt chooses wrong architecture if sources.list contains arch-qualified entries" [Undecided,New] https://launchpad.net/bugs/1032250 [16:34] much better. [16:36] ogra_: How goes the automation setup? === calculu5 is now known as calculus [17:02] GrueMaster, well, all pieces are ready, just need to glue them together [17:02] (which i'm working on right now) [17:03] Well, when you are done sniffing glue, have a beer on me. :P [17:05] hehe, will do [17:05] GrueMaster, your usbboot wikipage helped me a lot on monday btw, thanks for that [17:07] No problem. [17:07] Just wish I was still in the game (sometimes). [18:30] ogra_, infinity is quantal's flash-kernel supposed to work on precise to with no issues? [18:30] I need to modify it and I'd rather do it in quantal and then use the same in precise than touch two codebases - even if they may be similar enough [19:01] lilstevie, do you know if the boot part on the asua can be made visible from the kernel? [19:02] with a butchers knife [19:02] aka hacky patch to the gpt driver [19:07] lilstevie, as long as it works why not? [19:07] wouldn't be the first hack in the tf kernel :) [19:08] there are hacks ?!?! [19:09] janimo, f-k from quantal will likely not work on precise withaout a lot of transitioing [19:09] comkpletely different places for all the configs etc [19:10] ogra_, so it is not self-contained? [19:10] and with the switch to drop all boot.scr support you will need to transition to the new configs [19:10] ogra_, but ok thanks will check precise's version then [19:10] (thats something i'm in the middle of in quantal) [19:11] if i'm done the postinst should auto-transition your configs, even on precise (since upgrades will have to work) [19:11] but thats something i earliest start mid next week [19:12] also f-k is currently broken for all but omap/omap4 in quantal [19:13] (there is a hack in place that i will revert as soon as infinity fixed live-installer) === zyga_ is now known as zyga [20:21] using the 2-Aug-2012 omap4+armhf image on a pandaboard ES. it boots to a black screen. sad panda. [20:22] jimerickson, http://inotternews.com/wp-content/uploads/2012/05/sad-panda-podiatrist_medium.jpg?dur=203 [20:54] jimerickson, doesnt it come up with X if you wait long enough ? [20:54] that plymouth is broken is known, you get a black screen instead of a splash [20:54] but once X starts you should get display properly [21:27] well i waited quite awhile now and still a black screen. i will zsync my image tonight and try again later. === XorA is now known as XorA|gone === zyga is now known as zyga-afk