**** BEGIN LOGGING AT Wed Nov 21 02:59:58 2012 Nov 21 07:47:41 good morning Nov 21 10:32:20 look, images ! http://cdimage.ubuntu.com/daily-preinstalled/20121121/ Nov 21 10:32:39 (not the slightest idea if anything works with them though) Nov 21 10:40:10 ogra_, when are the images moving over to raring? ;-) Nov 21 10:40:57 these are raring ones ;) Nov 21 10:41:03 oh wow Nov 21 10:41:26 they are unlikely to work yet, just downloading one to test Nov 21 10:41:42 is this stock raring? or are there modifications which still need to get into the archive? Nov 21 10:41:55 the nexus7 settings package Nov 21 10:42:14 that's all? awesome Nov 21 10:42:15 go go go! Nov 21 10:42:24 but first of all i need to get oem-config to work Nov 21 10:42:46 (it might already, cant say until i have tested one :) ) Nov 21 10:43:01 download will still take 45min or so Nov 21 10:43:24 cool - let me know how it goes :) Nov 21 10:43:33 will do Nov 21 10:43:50 i will actually blog once they are usable :) Nov 21 10:44:05 though i guess the nux patch is still missing Nov 21 10:44:12 until thats there the desktop will be a mess Nov 21 10:44:39 (but the installer part up to teh desktop should work at least) Nov 21 10:44:59 ogra_: is the ubuntu-arm branch officially supported? Nov 21 10:45:12 ubuntu-arm branch ? Nov 21 10:45:30 rmhf is an official arch if you mean that Nov 21 10:45:35 *armhf Nov 21 10:46:02 the nexus image is built from universe using a universe kernel .... Nov 21 10:46:30 ogra_: i guess i mixed things up quite a bit :P Nov 21 10:46:33 so there is no official canonical commitment to security of the kernel etc ... i.e. its "community supported" Nov 21 10:46:46 ogra_: ty Nov 21 10:47:21 we have officially supported arches like the panda, highbank (calxeda) and armadaxp (marvell) Nov 21 10:47:58 btw panda ... Nov 21 10:48:09 ppisati, what are your kernel plans for raring wrt panda ? Nov 21 10:48:36 to support the GLES stuff we will still need the ton of patches Nov 21 10:49:01 (unless that changed since quantal) Nov 21 10:50:14 victorp, fyi http://cdimage.ubuntu.com/daily-preinstalled/current/ (untested and likely not working yet) Nov 21 10:51:00 ogra_, \o/ Nov 21 10:51:10 victorp, no blogging yet ! :) Nov 21 10:51:21 does it mean that at least I can ask vanhoof to point his super installer to it ? :) Nov 21 10:51:25 (not before we at least know the installer part works) Nov 21 10:51:30 * victorp puts down wp Nov 21 10:51:35 :) Nov 21 10:52:09 nothing in these images has been tested yet, its all new code so give me a bit to make sure it works Nov 21 10:52:17 ogra_, I know that you are getting the USB creator to work with this, but what does that actually mean for the user? Nov 21 10:52:19 additionally the rootfs img is still to big Nov 21 10:52:37 ogra_, at least something is being spitted out! thanks Nov 21 10:53:01 for the user it means that usb-creator pops up offering him an install as soon as he attaches a nexus7 in flash mode to a raring ubuntu PC Nov 21 10:53:59 fotr pre-raring i guess chris should just go on maintaining the zenity installer Nov 21 10:54:39 i doubt the changes to usb-creator will be easily backportable ( xnox may correct me :here ) ) Nov 21 10:54:58 ogra_: in a ppa, why not =) Nov 21 10:55:05 heh, k Nov 21 10:55:40 ogra_: my hope was to fold the omap4 branch into master Nov 21 10:56:01 ogra_: and import only: dvfs + patches to make pvr-omap work Nov 21 10:56:01 ppisati, please talk to robclark if the GLES bits can workj with that Nov 21 10:56:07 yeah Nov 21 10:56:14 ogra_: i'll do Nov 21 10:56:20 ogra_, the nux fix might be https://code.launchpad.net/~unity-team/nux/nux.depth-texture-detection-support/+merge/134729 (which hopefully lands soon) Nov 21 10:56:34 its just important that these bits work as long as we go on to provide desktop images for pandas Nov 21 10:56:46 robclark: ^ Nov 21 10:57:04 (i wouldnt mind dropping them though, but we need one reference platform and nexus is not fully supported) Nov 21 10:57:07 robclark: we where wondering what it takes to make pvr-omap4 work with vanila upstream Nov 21 10:57:24 robclark: do you have a bare minimum amount of patches that we need to take into account? Nov 21 10:58:01 dholbach, yeah, thats the one Nov 21 11:04:57 dholbach, ogra_, will that land in quantal or raring? Nov 21 11:05:47 lilstevie, I don't know - the folks in #ubuntu-unity should know though Nov 21 11:06:11 well, definitely in raring Nov 21 11:06:26 no idea if any backport SRUs are planned Nov 21 11:06:41 (you could ask on the merge request) Nov 21 11:07:40 ogra_, ok, not a major thing, I think TheMuso and I decided we were going to build tf201 images with gnome anyway for the time being, it can wait until raring for unity I guess :p Nov 21 11:08:11 oh btw ogra_ have you tried using the brcmfmac firmware for the bcm4330 rather than the android one? Nov 21 11:08:52 yes, i didnt try very hard though and couldnt get it to work Nov 21 11:11:30 hm ok Nov 21 12:11:20 yay, it boots and seems to unpack the tarball Nov 21 12:11:29 * ogra-cb_ waits in awe Nov 21 12:16:32 *twiddle thumbs* Nov 21 12:17:49 ogra-cb_, http://www.youtube.com/watch?v=0Wi8Fv0AJA4 Nov 21 12:18:19 haha Nov 21 12:19:08 hmm, k, seems the unpacking finished ... no X yet though Nov 21 12:20:07 doesnt seem to come up :( Nov 21 12:24:41 sniff Nov 21 12:24:49 :c Nov 21 12:24:51 and rebooting gets me into an endless reboot loop Nov 21 12:40:09 hmm, k looks like thats the old plymouth issue Nov 21 12:40:19 whee ! Nov 21 12:40:26 oem-config Nov 21 12:40:56 ogra-cb_, I find it odd that you have an issue with plymouth Nov 21 12:41:39 well, the bootloader sets console=none Nov 21 12:41:48 plymouth picks that up Nov 21 12:41:51 ah, I override that Nov 21 12:41:58 i cant Nov 21 12:42:07 it comes hardcoded from the bootloader Nov 21 12:42:30 we set a second console= arg, but plymouth only picks up the first Nov 21 12:42:49 you could hardcode the command line in the boot.img :p Nov 21 12:43:10 thats exactly what i do Nov 21 12:43:25 then the bootloader set cmdline should vanish 0.o Nov 21 12:43:26 the bootloader still prefixes it with its own stuff Nov 21 12:43:32 the hell Nov 21 12:45:51 k, so there is no touchscreen support at all, X comes up in portrait but beyond that it looks ok Nov 21 12:46:19 i'm nearly through oem-config ... using a kbd/mouse Nov 21 12:46:27 oh Nov 21 12:46:35 suddenly the touchscreen works Nov 21 12:46:50 completely inverted though Nov 21 12:52:38 yay Nov 21 12:52:52 * ogra-cb_ sees a desktop ... garbled unity elements indeed Nov 21 12:55:57 sigh Nov 21 12:56:04 corrupt filesystem Nov 21 12:58:13 ok, so most of this will be fixed by the settings package Nov 21 12:58:56 * ogra-cb_ re-flashes again Nov 21 13:05:02 Hi, I'm having a side project (fluidsynth) where it would be interesting to see if NEON optimisations would speed things up. Assuming those tests are successful, how do we enable those optimisations by default? Nov 21 13:05:23 I mean, we can't just compile with fpu=neon for the standard armhf build, can we? Nov 21 13:05:37 * lilstevie needs to stop procrastinating and finish his project for the tf201 Nov 21 13:05:49 diwic, not all armv7 hardware includes NEON Nov 21 13:06:13 lilstevie, exactly, so what is the standard/recommended way of doing this? Nov 21 13:06:37 so NEON is used automatically where available, and fallback code otherwise Nov 21 13:08:37 diwic, there should be a way to do a runtime check Nov 21 13:08:59 so you can switch code paths at runtime if your code supports it Nov 21 13:09:25 alternatively you can build the binary twice and create a -neon binary (or package) Nov 21 13:09:57 ogra-cb_, hmm, that sounds interesting Nov 21 13:09:58 by default all armhf packages should not be neonized Nov 21 13:10:20 since there are still many a7 SoCs that dont support neon Nov 21 13:10:48 which is sad Nov 21 13:11:53 ogra-cb_, I was hoping there was a wiki page or something about how to best deal with this Nov 21 13:12:11 there might be a linaro page for it, not sure Nov 21 13:13:18 iirc the runtime selection had to do with using /proc/self/auxv Nov 21 13:17:51 hmm, the code for pulseaudio reads /proc/cpuinfo Nov 21 13:21:46 that might work as well Nov 21 13:21:55 just grepping fpr neon i guess Nov 21 13:22:55 but long story short, this is just something every new project has to invent the wheel again Nov 21 13:23:12 yeah, a bit annoying Nov 21 13:23:24 thats why i think there is some better linaro solution Nov 21 13:33:37 xnox, we shoould really look at the startyup time of ubiquty-dm ... not sure if it is so slow in non oem-config mode, but running oem-config i nearly sit a minute at a localhos login prompt before X even starts Nov 21 13:34:04 ogra-cb_: *sigh* Nov 21 13:34:21 it might be less odd if there is a splash Nov 21 13:34:52 ogra-cb_: do these comments mean you have an image with oem-config for me play around with? Nov 21 13:35:10 xnox, a broken one but yeah Nov 21 13:35:20 ogra-cb_: define broken =) Nov 21 13:35:36 a little, a lot, not enough? Nov 21 13:35:49 the tarball installer accidentially unmounts /root after unpacking Nov 21 13:35:57 which leaves you with a black screen Nov 21 13:36:21 rebooting at the point where you are sure it is done gets you into a normal boot though Nov 21 13:36:43 xnox, http://cdimage.ubuntu.com/daily-preinstalled/current/ Nov 21 13:36:47 grab both files Nov 21 13:37:07 ack Nov 21 13:37:07 edit the bootimg with abootimg to point to tty0 instead of tty1 Nov 21 13:37:23 gunzip the img.gz Nov 21 13:37:30 fastboot erase boot Nov 21 13:37:35 fastboot erase userdata Nov 21 13:37:51 fastboot flash boot /path/to/*.bootimg Nov 21 13:38:12 fastboot -S 630M flash userdata /path/to/*.img Nov 21 13:38:20 thats how i flash here Nov 21 13:38:29 and after that: fastboot reboot Nov 21 13:38:38 oh, all of that needs sudo atm Nov 21 13:39:14 what does the "-S" param mean? Oo Nov 21 13:39:27 transfer in chunks Nov 21 13:40:07 xnox, note that the touchscreen is completely broken, you need kbd and mouse Nov 21 13:41:05 well, not broken but inverted Nov 21 13:42:01 fix for the tarball stuff is uploaded already Nov 21 13:42:14 next image wont have the issue anymore Nov 21 13:42:31 for the rest we need the default-settings package now Nov 21 13:47:46 ack. Nov 21 14:00:53 new image build running Nov 21 14:58:03 do you know if there's a place where i can find a maverick armel image? Nov 21 15:00:14 oof Nov 21 15:00:56 oof? Nov 21 15:00:58 woof? Nov 21 15:02:14 ppisati, theoretically on http://old-releases.ubuntu.com/releases/ Nov 21 15:02:20 nope Nov 21 15:02:25 but that doesnt have any arm stuff it seems Nov 21 15:02:25 doesn't arm maverick stuff Nov 21 15:02:38 so probably a cjwatson or infinity question Nov 21 15:03:03 is there a script to recreate an image? Nov 21 15:03:10 since the archives are still there Nov 21 15:03:30 well, there should also be images somewhere Nov 21 15:03:42 i'm pretty sure they havent been thrown away Nov 21 15:04:55 brendand: um, daniel *is* working on fixing the button1 bug Nov 21 15:05:05 infinity: ^ Nov 21 15:05:44 achiang, ok - but In Progress -> Confirmed is not a correct transition then Nov 21 15:05:54 achiang, well, he did set it from in progress to confirmed Nov 21 15:06:37 achiang, In Progress -> Confirmed says something like 'I was working on it, but I give up and can't figure it out' Nov 21 15:06:57 might be better to ask than to accuse Nov 21 15:12:30 it wasn't intended to sound accusatory, probably would have been better phrased as a question Nov 21 15:14:08 ogra-cb_: ah! i found an obscrure mirror that still had it :) Nov 21 15:14:22 heh Nov 21 15:14:45 well, i know GrueMaster had a complete mirror carrying all arm images ever built Nov 21 15:14:52 from Bangkok Nov 21 15:14:56 but he's gone Nov 21 15:18:38 ogra-cb_: lol Nov 21 15:18:44 ogra-cb_: /me curses Nov 21 15:18:49 brendand: ok, fair enough. it might be nice to follow up with daniel and explain that you were just asking him a question. Nov 21 15:20:19 vanhoof, heh Nov 21 15:21:17 achiang, ok followed up on the bug (hopefully politely :) ) Nov 21 15:21:32 vanhoof, achiang .... in case you missed the backlog http://cdimage.ubuntu.com/daily-preinstalled/current/ Nov 21 15:21:36 brendand: thanks. :) Nov 21 15:21:51 not working yet, fixes are uploaded and a new build is running Nov 21 15:22:03 woo hoo! Nov 21 15:22:11 achiang, and apologies again. in my experience it's very common to see people misusing bug states - but i agree that there's no need to be blunt in response Nov 21 15:22:35 you can manually work around the issues though, xnox is just guineapiggin :) Nov 21 15:22:43 brendand: sure, no worries. i just didn't want to demotivate the guy who is fixing one of our hardest bugs. ;) Nov 21 15:24:35 ogra-cb_: so with that, i believe the nux bug is our only remaining issue Nov 21 15:25:00 ogra-cb_: nice! Nov 21 15:25:08 achiang, well, i need to get to the settings, currently you pretty well notice that they are missing Nov 21 15:25:22 and it is still to big until we did the seed change Nov 21 15:26:02 so it wont work for everyone (fastboot -S works fine for me here) Nov 21 15:27:12 i would also like to look into rersizext2fs instead of having to produce a bunch of differently sized images Nov 21 15:28:06 brendand: now it's my turn to apologize to you. guess daniel *did* move on to other things :-/ Nov 21 15:28:52 ogra-cb_: +1 there Nov 21 15:32:34 ogra-cb_: is that image one size fits all? Nov 21 15:33:11 ogra-cb_: or which size is it currently for? Nov 21 15:34:06 xnox, nope, currently its the smalles common denominator ... Nov 21 15:34:10 6G Nov 21 15:34:14 ack Nov 21 15:34:40 we have two options, one is to provide three images ... and the other is to grow the rootfs on the fly Nov 21 15:34:53 i fear the latter will not work properly though Nov 21 15:35:16 by experience i had with the fastboot ext4 version Nov 21 15:37:55 what about providing just update.zip for recovery? Nov 21 15:38:26 feel free to work on something like that Nov 21 15:39:59 for raring the plan is to just to replicate what we did in quantal and improve that ... there are no plans to develop new image types from scratch in the ubuntu infrastructure atm Nov 21 15:40:15 it should be simple, just erase everything and extract boot.img and rootfs.tar.gz Nov 21 15:40:40 well, the files are there now :) Nov 21 15:40:46 feel free to play with them Nov 21 15:41:02 I think I'll do that Nov 21 15:41:18 (though probably wait for the next build, so that you at least get through into the installer) Nov 21 15:41:37 that one *should* get you into oem-config Nov 21 15:41:49 I have to make something so that my multi-boot thing can edit the update-initramfs files Nov 21 15:41:59 ugh Nov 21 15:42:03 that sounds very ugly Nov 21 15:42:05 yeah) Nov 21 15:42:28 but well, I don't really feel like distributing modified ubuntu image Nov 21 15:42:29 do you know that there is initiative for getting grub running under arm ? Nov 21 15:42:32 in linaro Nov 21 15:42:40 whau Nov 21 15:42:42 that will just get ou multiboot for free Nov 21 15:42:56 *you Nov 21 15:43:10 that sounds...complicated to do Oo Nov 21 15:43:25 as complicated as on x86 :) Nov 21 15:43:57 well, on hard drive you can mess with partitions Nov 21 15:44:20 well, you dont need to mess with them Nov 21 15:44:50 I need to separate OS-es somehow Nov 21 15:45:07 no, you just need to be able to switch between boot images Nov 21 15:45:40 the only issue that leaves is that boith OSes want to upgrade the kernel in the same place if it comes to that Nov 21 15:46:03 and thats something thats not easily fixable ... but you will hit that issue in all variants of dial booting Nov 21 15:46:15 hmm, still seeems a bit complicated to do Nov 21 15:46:28 which is why i suggest since day one to put the ubuntu bootimg into recovery and be done Nov 21 15:46:30 switch between boot images == flash new ones? Nov 21 15:46:35 no Nov 21 15:46:38 just switch Nov 21 15:46:56 so...they, like, wanna change the bootloader? Nov 21 15:47:31 they update mmcblk0p2 if a new kernel shows up (or in case of ubuntu even if a package that ships initramfs bits gets ionstalled) Nov 21 15:51:20 ogra_: kexec or similar? :P Sadly, that's the most promising option atm..You can't replace bootloader, even if you had a working u-boot. Nov 21 15:51:48 hmm Nov 21 15:52:20 I still dont really get how they switch between kernels Nov 21 15:52:34 RaYmAn, either kexec or just use what the bootloader offers already (keep android in boot and flash ubuntu to recovery) Nov 21 15:52:37 like, where is the grub stored? Nov 21 15:52:45 ther'es no..grub Nov 21 15:52:56 the grub would be your bootimg Nov 21 15:53:03 ogra_: yeah..but then you lose recovery :) though, not much of an issue with fastboot available Nov 21 15:53:04 the bootloader would chainload it Nov 21 15:53:12 RaYmAn, right Nov 21 15:53:21 it is the price you have to pay for dual boot Nov 21 15:53:35 indeed. It's such a pity they locked down bootloader changes. Nov 21 15:54:34 ask the linaro guys about details for grub on arm :) Nov 21 15:54:44 anyway, it's not like I'll stop now, when it's already working, I am just working on the UI and installation process Nov 21 15:54:51 * ogra_ only heard of it third hand, i havent seen it or anything Nov 21 15:55:28 it's a pitty I can't get kexec working, though :/ Nov 21 15:55:50 well, then you would still have the issue that an android kernel update would overwrite it Nov 21 15:57:11 custom kernels have "AnyKernel" installer, which changes only kernel (and leaves ramdisk as-is), and if you mean OTAs, well, then the user will have to go to recovery and select "inject boot.img" Nov 21 15:58:21 Tassadar: look at kexec-hardboot Nov 21 15:58:28 that'll definitely work. (I have it working on my TF201) Nov 21 15:58:41 It puts some requirements on the 'target' kernel, but it's minor. Nov 21 15:58:48 yes, I did, and I got stuck on "how the hell should I reboot tegra3 via watchdog" Nov 21 15:59:07 you don't need to use watchdog to do it Nov 21 15:59:12 watchdog is disabled in ubuntu kernels iirc Nov 21 15:59:27 caused suspend issues Nov 21 15:59:29 I am simply getting Google white logo =/ Nov 21 15:59:37 with the unlocked padlock Nov 21 15:59:51 hi all, I'm trying to install ubunutu on a N7. I used the ubuntu installer and it installed fine. But now I'm wondering if wifi is supposed to work ? Nov 21 15:59:52 xnox, hmm, that doesnt sound like you flashed bootimg properly Nov 21 15:59:59 vagabon: yes, it is Nov 21 16:00:03 works for me Nov 21 16:00:16 well..using one of the ooold images) Nov 21 16:00:25 Tassadar: hmm how am I supposed to enter the key ? Nov 21 16:00:35 vagabon, with the keyboard Nov 21 16:00:43 that pops up on your screen Nov 21 16:00:53 hmm it doesn't for me Nov 21 16:01:19 try reboot Nov 21 16:01:21 ok Nov 21 16:01:26 it should as soon as the focus goes into the text input field Nov 21 16:01:26 it glitches sometimes Nov 21 16:01:28 Tassadar: volatile u32* reg = (u32*)(0x7000E400); reg |= 0x10; - that should reboot instantly (even from within android or linux) Nov 21 16:01:52 okay, I'll look to that again once I finish this, thanks Nov 21 16:02:30 RaYmAn, instantly as in ... it will take care for syncing your open files etc and flush the memory to disk and properly unmount it ? or instantly like pulling out AC and battery at the same time ? Nov 21 16:03:20 ogra_: instantly as in, instantly.. The idea is that you execute it in the kexec shutdown sequency, so only once it's already almost turned off Nov 21 16:03:21 pulling the battery I guess) Nov 21 16:03:31 it's the same effect as if the watchdog timed out Nov 21 16:03:37 ah Nov 21 16:03:53 it resets the main clock controller iirc Nov 21 16:03:58 Oo Nov 21 16:04:40 Tassadar: my kexechardboot for n7 is here: https://github.com/EnJens/kernel_tf201_stock/tree/android-tegra-nv-3.1-kexec Nov 21 16:04:46 heh Nov 21 16:04:56 err Nov 21 16:04:57 not n7 Nov 21 16:04:58 tf201 Nov 21 16:05:00 I was just writing "could you send me a link..." :D Nov 21 16:05:22 because all I could find was hardboot for ...atrix I think Nov 21 16:05:34 otherwise can I add a static config file in the rootfs somewhere so I can init the wifi connection ? Nov 21 16:05:59 vagabon, you can use /etc7network/interfaces Nov 21 16:06:52 vagabon, http://wiki.debian.org/WiFi/HowToUse#WPA-PSK_and_WPA2-PSK Nov 21 16:07:13 the question is how you will access the rootfs :) Nov 21 16:08:01 unpack it on my host, hack it, repack, reflash... I gues Nov 21 16:08:24 ubuntu installer seems to download the images from somewhere Nov 21 16:09:12 well, wifi works for all of us Nov 21 16:09:30 there are plenty of nexus7 with ubuntu out in the wild and we havent had a single bug report yet Nov 21 16:09:31 ok I'm reflashing the images and give it a new test Nov 21 16:09:39 repacking is quite a task Nov 21 16:10:04 ogra_: does Network manager handle the wireless connection? Nov 21 16:10:10 yes Nov 21 16:10:18 hmm, is it possible to just add some files into that tgz file, or do I have to unpack it and then pack again? Nov 21 16:10:33 you might be able to add Nov 21 16:10:35 not sure Nov 21 16:10:42 hmm I hope I will Nov 21 16:11:05 it's like 2 files only Nov 21 16:11:09 as long as you use the same optiuons that were used for tarring it up it should all be fine Nov 21 16:12:34 ogra_: cool this time the virtual kb pop up :) Nov 21 16:12:47 ah, thats how it sould be :) Nov 21 16:13:04 ogra_: does a sshd is running on the tablet ? Nov 21 16:13:13 ogra_: is a sshd running on the tablet ? Nov 21 16:13:15 nope Nov 21 16:13:24 sudo apt-get install openssh-server Nov 21 16:13:48 why not install it by default ? Nov 21 16:13:51 we used to have it preinstalled, but during UDS there were some 100s of devices around, all using our image with the same user and password Nov 21 16:14:03 so that wouldnt have been clever Nov 21 16:14:14 and after all its just an apt-get away :) Nov 21 16:14:25 using a virtual kb is a PITA :) Nov 21 16:14:45 well, i used the nexus7 for three weeks as my main machine while developing the image Nov 21 16:14:54 you get used to it after a while Nov 21 16:15:02 by the way, I didn't find default root password mentioned anywhere on the nexus 7 wiki Nov 21 16:15:04 its annoying in the beginning, i agree :) Nov 21 16:15:11 you're more skilled than I am :) Nov 21 16:15:16 and, not even the "ubuntu" password, for that matter) Nov 21 16:15:16 Tassadar, thats because there is none ... its ubuntu Nov 21 16:15:24 entering a WPA key is just a nightmare for me Nov 21 16:15:41 the ubuntu password is ubuntu and i'm sure it at least was once mentioned on the wiki Nov 21 16:17:22 just out of curiosity, is the virtual kbd part of unity ? Nov 21 16:20:28 vagabon: package onboard Nov 21 16:20:36 it's part of ubuntu a11y packages =) Nov 21 16:21:11 RaYmAn: when calling the kexec userspace tool, what parameters do I have to load? I mean it allows to specify ramdisk and command line, what if I specify none of these? Nov 21 16:21:23 will it you ramdisk and cmd line from current kernel? Nov 21 16:22:22 not afaik Nov 21 16:22:41 I assume you use the patched kexec tool, right? Nov 21 16:23:00 yeah Nov 21 16:23:02 xnox: which one ? Nov 21 16:23:06 a11y ? Nov 21 16:23:19 vagabon: $ apt-get install onboard Nov 21 16:23:52 ok thanks Nov 21 16:24:16 xnox: actually I'm wondering how the kbd is started automagically... Nov 21 16:24:36 it detects input field focus Nov 21 16:24:50 as soon as an input fiel is focused, it will pop up Nov 21 16:24:58 i think it asks the window manager Nov 21 16:25:03 vagabon: it has been around for a while as our on-screen keyboard as accessibility feature for those who cannot use a regular keyboard. Nov 21 16:25:25 xnox: ok, but I'm just starting to discover ubuntu. Nov 21 16:25:54 xnox: does that mean that the package should work with any DEs ? Nov 21 16:26:01 yes. Nov 21 16:26:02 yes Nov 21 16:26:05 cool Nov 21 16:26:20 I'will try to use it here (xfce4) Nov 21 16:26:22 vagabon: the same way cursor | flashes when you need to input text =)))))) Nov 21 16:34:00 I am not having joy with the daily: either I am dropped into initramfs shell or it doesn't boot. Nov 21 16:36:23 xnox, did you see it unpacking the tarball ? Nov 21 16:37:08 nope Nov 21 16:37:20 hmm, that sounds very wrong Nov 21 16:37:27 it definitely did that here Nov 21 16:37:53 did you do the fastboot erase steps before flashing ? Nov 21 16:39:07 it should clear the screen and then start verifying the tarball structure ... after that it will unpack and tell you that it takes a few min Nov 21 16:39:07 yeap Nov 21 16:39:26 rsalveti, nudge? have any idea on why I might get some XPutImage BadMatch running glmark2? Nov 21 16:39:31 i am getting mounting /dev/mmcblk0p9 on /root failed: Invalid argument Nov 21 16:39:34 the next step was broken for me ... but up to there everything worked as expected Nov 21 16:39:48 but the tarball unpack was not shown.... Nov 21 16:39:53 right, thats the broken step Nov 21 16:40:20 xnox: the kbd doesn't pop up when the application search input field is selected Nov 21 16:40:26 xnox, the quantal image worked for you, right ? Nov 21 16:40:34 ogra-cb_: yeah. Nov 21 16:40:40 vagabon, yep, known bug Nov 21 16:40:42 vagabon: is onboard running in the background? Nov 21 16:40:52 ah, known bug. ok. Nov 21 16:41:00 ok Nov 21 16:41:09 by the way, what will you use to determine the /data partition in tarball-installer? Nov 21 16:41:12 you can go to the onboard settings and switch on "show a floating button" Nov 21 16:41:18 then where can I find a terminal launcher ? Nov 21 16:41:23 then you can force it on if you need it Nov 21 16:41:31 even if the input field detection doesnt work Nov 21 16:41:56 vagabon, press ctrl+alt+t Nov 21 16:42:08 cool Nov 21 16:42:23 xnox, so i guess you should wait for the next build Nov 21 16:42:40 =/ Nov 21 16:42:42 ack Nov 21 16:42:51 ogra-cb_: should it fail? Nov 21 16:42:55 theoretically that shoudl fix the mount issues ... but i'm not really sure whats going on there if you dont see it unpack Nov 21 16:43:08 ogra-cb_: sounds similar to what we saw using -S before if not Nov 21 16:43:22 vanhoof, oh, hmm Nov 21 16:43:36 ogra-cb_: I believe some folks saw that on 16gb n7's Nov 21 16:43:46 xnox, you could trz without the -S option Nov 21 16:43:48 xnox: which nexus 7 do you have? I mean wifi/3g 8/16/32gb? Nov 21 16:43:49 and we yanked -S out of the installer Nov 21 16:43:59 vanhoof, i know Nov 21 16:44:07 but the current image is 750M Nov 21 16:44:12 ah Nov 21 16:44:22 i have 16G and without -S it doesn't flash. Nov 21 16:44:22 wont work without -S until the ubuntu seeds have been split Nov 21 16:44:42 yeah, it exhausts the ram Nov 21 16:44:43 ogra-cb_: im trying on a 8gb now Nov 21 16:44:51 blimey, the vol up/down in the bootloader have shutdown option. Nov 21 16:44:54 my 8G here worked fine with -S Nov 21 16:44:56 \0/ Nov 21 16:45:07 * xnox was stuck in restart loop for a second here. Nov 21 16:45:11 heh Nov 21 16:45:21 getting it started up is sometimes a pain Nov 21 16:45:27 after such a loop Nov 21 16:45:39 ogra-cb_: cat ubuntu.img|adb shell dd of=/dev/block/mmcblk0p10 while in recovery? ;) Nov 21 16:45:53 poor mans fastboot, heh. Nov 21 16:45:56 no adb :P Nov 21 16:46:10 darn Nov 21 16:46:24 we discussed that in advance :) neither update.zip nor any adb way are an option for the official images Nov 21 16:46:25 and it's event funnier when you accidentally press volUp when holding power-up) Nov 21 16:46:25 fastboot boot an image that sets up mmcblk0p10 as UMS? Nov 21 16:46:55 well, fix -S woudl eb the best option Nov 21 16:46:59 *be Nov 21 16:47:01 I suppose Nov 21 16:47:31 or alternatively just dont use oversized images as we do atm :) Nov 21 16:47:50 i was actually pondering to use bzip2 for the tarball Nov 21 16:48:10 but that will break rsync/zsync abilities Nov 21 16:48:46 wouldn't a fastboot boot'ed image work for offline resizing after write? I'm not sure how much actual data is in the .img Nov 21 16:49:12 3-4G Nov 21 16:49:17 iirc Nov 21 16:49:20 ah, nevermind then Nov 21 16:49:58 ah, i lied Nov 21 16:50:10 2-3G actually Nov 21 16:50:50 the pint with resizing is that when i tried any of the ext2fs tools in the past it resulted in an unbootable corrupt fs Nov 21 16:51:11 fastboot does anything to the ext4 Nov 21 16:51:32 else a simple resizext2fs would do Nov 21 16:52:37 have you tried doing it iwth an ext3 image? Nov 21 16:54:08 i havent even tried it with ext4 yet :) Nov 21 16:54:19 i'll get to that next Nov 21 16:54:46 i know that a mkfs.ext4 definitely produces something the kernel cant boot Nov 21 16:54:52 the nvidia bootloaders support for ext4 has been...spotty at best Nov 21 16:55:02 its the kernel Nov 21 16:55:13 the bootloader is long gone at that point Nov 21 16:55:33 i'm usually operating from an initrd shell when trying such stuff Nov 21 16:55:38 hm Nov 21 16:55:56 so what you're saying is that if you do mkfs.ext4 on a device, it refuses to mount it after reboot? Nov 21 16:56:42 it mounts redonly and there is no way to get it properly out of that state Nov 21 16:57:00 peculiar Nov 21 16:57:07 you can force it into RW but it will always be recognized as having errors on boot Nov 21 16:58:01 ogra-cb_: what size are you specifying for -S, /me finally finished download Nov 21 16:59:20 i took 630M iirc Nov 21 17:02:25 http://paste.ubuntu.com/1375201/ Nov 21 17:03:38 cool, yeah didnt have scrollback just wanted to confirm Nov 21 17:03:41 rebooting now :) Nov 21 17:04:04 it will fail with a black screen after untarring, a reboot makes it work though Nov 21 17:04:12 at least that worked for me Nov 21 17:04:29 cool, determining structure now Nov 21 17:04:44 yeah, thats definitely further than xnox got Nov 21 17:05:01 i actually wonder if we can drop the structure check now Nov 21 17:05:28 it takes a lot of time and we dont have any subdirs in the official tarballs ... so there is nothing to strip Nov 21 17:09:30 * vanhoof waits Nov 21 17:17:33 ogra-cb_: flashed, then dumped me to busybox after a few mount errors Nov 21 17:17:59 great, a reboot should get you onto oem-config Nov 21 17:17:59 ogra-cb_: rebooted, not it looks like its attempting to boot but rebooting itself when it looks like plymouth would kick in Nov 21 17:18:10 oh, k Nov 21 17:18:33 well, i changed the cmdline to point to tty0 instead of tty1 Nov 21 17:18:38 that helped here Nov 21 17:19:04 (for console= that is) Nov 21 17:19:10 in the boot.img thats on the daily page? Nov 21 17:19:11 * vanhoof checks Nov 21 17:19:19 the next image will have that and an umount fix Nov 21 17:19:36 ah lemme try that, im still tty1 here Nov 21 17:19:52 i used the bootimg modified, added a break=bottom, chrooted into the rootfs and updated the initrd from there Nov 21 17:24:19 Hi, I am looking to buy an arm system to turn into a PVR. Can I get a usb DVB-T2stick to work in ubuntu under ARM? Nov 21 17:26:10 ogra-cb_: I should read sooner ;) Nov 21 17:26:26 ogra-cb_: just updated boot.img, flashed that and reflashed userdata ;) Nov 21 17:27:11 ogra-cb_: now I wish I was in CPH and got that screen from ya w/ all this fun as of late :) Nov 21 17:27:32 heh Nov 21 17:40:49 ogra-cb_: much further! Nov 21 17:41:03 ogra-cb_: in oem-config now, although with a heavily distorted background Nov 21 17:41:42 yeah, thats not arm specific i think Nov 21 17:41:54 i saw an x86 bug about that yesterday or so Nov 21 17:42:23 chmm Nov 21 17:42:39 ogra-cb_: and portrait by default which comes with its own fun but the update to tty0 got me running, though no actual splash Nov 21 17:43:27 RaYmAn: I applied the kexec-harboot patches, tried to do "kexec --load-harboot /tmp/zImage --mem-min=0x50000000 && kexec -e" Nov 21 17:43:29 but it just reboots Nov 21 17:44:31 I assume the flashed image on boot partition has the patches? Nov 21 17:44:45 and that the target kernel /tmp/zImage at least has the decompressor patches? Nov 21 17:45:07 wait, I used fastboot boot Nov 21 17:45:16 * Tassadar slaps Tassadar Nov 21 17:45:39 it needs to be on flash, because the whole point of hardboot is that it does a full reset, then kernel detects a kexec kernel in memory and jumps to that :P Nov 21 17:46:03 doesnt it use the kernel from recovery? Nov 21 17:50:04 hm, now it's stuck on the google logo, let's see what ramconsole says Nov 21 17:50:53 [ 121.854978] Starting new kernel Nov 21 17:50:54 [ 121.855128] Bye! Nov 21 17:51:06 well, that's disappointing) Nov 21 17:59:44 Tassadar: the value for min-memory can be quite important..I think i used 0xa0000000 or similar Nov 21 18:00:03 honestly, I have no idea what does it mean, so..) Nov 21 18:00:51 http://cdimage.ubuntu.com/daily-preinstalled/20121121.1/ Nov 21 18:00:55 next round Nov 21 18:01:06 vanhoof, ^^^ Nov 21 18:01:28 ogra-cb_: heh Nov 21 18:01:31 * vanhoof grabs Nov 21 18:01:39 that should get you completely through to oem-config Nov 21 18:02:50 ogra-cb_: cool, i'll give it a spin Nov 21 18:03:56 and indeed the rootfs is now zsyncable Nov 21 18:04:08 that should speed up downloading a lot Nov 21 18:04:38 Tassadar: stuck on logo is progress though :P Nov 21 18:05:07 it definitely is, but I also have no debug info :/ Nov 21 18:06:14 ..and I really wonder what is it doing, it does not even open the ramconsole Nov 21 18:07:34 robher: ping, what's ripal's nick? Nov 21 18:15:45 RaYmAn: does normal kexec work on TF201? I mean, on nexus7, it does not even restart the device Nov 21 18:23:58 woot Nov 21 18:24:00 its bootin Nov 21 18:24:03 * Tassadar waits for adb Nov 21 18:24:31 ..but it is the same kernel Nov 21 20:52:38 ogra-cb_: looks good Nov 21 20:52:56 ogra-cb_: flashed booted right up to oem-config Nov 21 21:25:46 * achiang makes an upload of unity-6.12.0 to staging ppa Nov 21 21:25:51 hope it doesn't FTBFS Nov 21 21:27:31 achiang: After the issues you had with diffs to tarballs, I'm thinking that's not actually very 6.12 at all. :P Nov 21 21:27:48 infinity: it's ... "something" Nov 21 21:28:05 probably closer to 7.0 than anything else Nov 21 21:28:22 oh, but... i heard it might FTBFS on armhf Nov 21 21:28:24 le sigh Nov 21 21:58:40 * achiang cries. FTBFS due to bamf API change Nov 21 23:09:51 tassadar_, you should keep an eye on mem-min, that is the minimum memory that it uses to put the zImage and initrd in. I know with the tf201 0x50000000 is outside of real memory so wouldn't work Nov 21 23:23:59 lilstevie: can I somehow find out the memory layout of nexus7? Like, at least what is the "end" address? Nov 21 23:28:04 aah, cat /proc/iomem i believe :ú Nov 21 23:28:31 tassadar_, iomem, but also one of the very first things dmesg will should you is a basic mem map :p Nov 21 23:30:17 and the kexec should load the kernel and initrd somewhere to system ram, preferably outside kernel, yes? Nov 21 23:32:07 well of course, what good would it be loading into the kernel :p Nov 21 23:33:08 hmm, I think I also have to change the hardboot page address, now it is in framebuffer Nov 21 23:35:36 hah yes you probably will Nov 21 23:36:01 I know in the first revision RaYmAn accidentally put it in the framebuffer Nov 21 23:36:35 yes, well, the second put it precisely into nexus 7's frambuffer :D Nov 21 23:42:53 heh Nov 22 00:05:11 no, still freezes Nov 22 00:05:56 I think there is something wrong with "void machine_kexec()", because normal kexec does not even reboot the device Nov 22 00:07:00 did you have a look at the patches Nov 22 00:07:43 yes, on tf2, RaYmAn removed several calls from that method, because they caused tegra3 to freeze Nov 22 00:07:47 there was one to remove an extra flush_cache_all() Nov 22 00:07:49 *tf201 Nov 22 00:07:58 https://github.com/EnJens/kernel_tf201_stock/commit/6e5043854b55a8552d35d56ed9de414d095cd4d6 Nov 22 00:08:03 yes, I removed those Nov 22 00:08:16 the outer_inv_all too? Nov 22 00:08:24 yes Nov 22 00:08:32 spam printks through there Nov 22 00:09:09 thats what we did to figure out exactly where it froze Nov 22 00:09:13 klog does not get disabled somewhere there? Nov 22 00:09:34 afaik it isn't until the very last step Nov 22 00:09:42 oh my god) Nov 22 00:09:58 I have already done that, but thought klog just gets disabled Nov 22 00:10:23 but I don't really remember where was the last printk, have to try that again) Nov 22 00:11:51 the other thing you can do is write to a specific register to force a reboot, move it up one spot each time it works until it fails Nov 22 00:14:16 last printk which gets into klog is before cpu_proc_fin(); Nov 22 00:16:00 hm Nov 22 00:16:06 well, let's try to comment it out I guess) Nov 22 00:49:53 chm Nov 22 00:49:59 no change, still freezes Nov 22 00:50:09 maybe it does not even boot anything Nov 22 00:51:06 the ramdisk console is not touched, there is still output from the kernel in which I did the kexec -e in there Nov 22 01:04:28 tassadar_, pastebin machine_kexec? Nov 22 01:05:40 http://pastebin.com/Zea4zUsp Nov 22 01:07:46 and the last thing you get is exec8? Nov 22 01:08:10 well I would expect that anyway Nov 22 01:09:39 that's right Nov 22 01:13:16 odd Nov 22 01:13:24 if you get to that point it should just work Nov 22 01:13:34 I just looked through the code Nov 22 01:13:56 Im gonna try to mess with --mem-min Nov 22 01:14:15 yeah Nov 22 01:14:26 mem-min was a pain from memory Nov 22 01:14:28 actually Nov 22 01:14:36 does it require any other parametrs? Nov 22 01:14:42 from memory I had similar results until I came across where I am Nov 22 01:14:43 like ramdisk or something Nov 22 01:14:44 um Nov 22 01:15:07 well I have it doing commandline zimage ramdisk mem-min Nov 22 01:15:17 then kexec -e Nov 22 01:15:29 there is no command line in nexus 7's boot image anyway Nov 22 01:15:47 cause it is bootloader set Nov 22 01:15:59 but that isn't the point Nov 22 01:16:05 with the atags patch you need to set it Nov 22 01:16:23 even if your --commandline=`cat /proc/cmdline` Nov 22 01:24:54 now it rebooted Nov 22 01:25:11 yeaaah Nov 22 01:25:13 it works! Nov 22 01:25:20 it just does not have initrd Nov 22 01:25:24 so it reboots Nov 22 01:25:36 I'll put one there Nov 22 01:27:30 ./kexec --load-hardboot ./zImage --command-line="$(cat /proc/cmdline)" --mem-min=0xA0000000 --initrd=./rd.img Nov 22 01:27:34 nice, thank you :) Nov 22 01:27:37 shell@android:/ $ cat /proc/version Nov 22 01:27:38 Linux version 3.1.10kexec4test-gba0df58-dirty (tassadar@nymeria) (gcc version 4.6.x-google 20120106 (prerelease) (GCC) ) #18 SMP PREEMPT Thu Nov 22 01:01:55 CET 2012 Nov 22 01:29:34 there we go Nov 22 01:31:40 * lilstevie goes back to trying to write his boot manager Nov 22 01:32:15 one last thing - it uses kernel from recovery, right? Nov 22 01:34:08 uh? what? Nov 22 01:34:54 https://github.com/EnJens/kernel_tf201_stock/commit/e443f15cd73232b78827dc108c1d2ffbd5bc1fb1#L8R1546 Nov 22 01:36:50 that must be some left over from the original code, you cannot boot to recovery from kernel space (requires some userspace stuff first) Nov 22 01:37:07 but it all depends on what kernel you put in boot, Nov 22 01:37:51 kernel in boot must always have these patches, the one which is kexecd' does not, correct? Nov 22 01:37:52 I put the kexec kernel in boot because I want to go directly to the boot menu, not have to get to recovery to select Nov 22 01:38:07 huh Nov 22 01:38:35 the patches to the decompressor and the patches to do the hardboot kexec are all intertwined Nov 22 01:38:55 so both must be patched, well, that is not good :/ Nov 22 01:39:14 oh wait I see what you mean Nov 22 01:39:24 yes, you require 2 patches to the kexec guest kernel Nov 22 01:40:20 the decompressor optimization patch (otherwise boots take anywhere from 30seconds to 10 minutes) and the atags patch, otherwise the guest kernel just ends up with the initrd and cmdline of the kexec host kernel Nov 22 01:40:49 RaYmAn did say that way back when you started looking Nov 22 01:41:51 Hello everyone Nov 22 01:42:17 I'm trying to install ubuntu server on my pogoplug pro. Is anyone familiar with it? Nov 22 01:44:00 pogoplug pro is armv6, ubuntu does not support armv6 processors Nov 22 01:47:21 Would debian do it then? Nov 22 01:47:29 maybe Nov 22 01:48:24 I saw something on the archlinux wiki which makes commenting on that rather difficult to know Nov 22 01:50:11 rafase282, there does appear to be a third party project that got debian running on it though, just google around Nov 22 01:53:15 yeah I'm googling it around, I'm not a fan of archlinux I prefer ubuntu but i was told it does not support arm6 Nov 22 02:01:31 hm, I should go to bed, it's 3am here Nov 22 02:01:43 so, thanks for help and good night **** ENDING LOGGING AT Thu Nov 22 02:59:58 2012