**** BEGIN LOGGING AT Tue Jun 05 02:59:58 2012 Jun 05 10:51:04 ogra_: hi. i just realised that we never really discussed about 12.10 OMAP4 kernel... what do you guys expect? Jun 05 11:16:56 ndec, hmm, not sure, ppisati ^^^ any idea ? Jun 05 12:02:19 ndec: you mean the kernel version? or what else? Jun 05 12:10:42 ppisati: how to you plan to manage the OMAP4 kernel? so it's both the kernel version, and the branch strategy. Jun 05 12:10:51 and also are you planning to make panda images? Jun 05 12:12:32 oh, a "panda tablet" in germany ! Jun 05 12:12:34 http://aldi-nord.de/aldi_ab_mittwoch_0606_48_5_967_15052.html Jun 05 12:13:12 * ogra_ wonders why anyone would go with 4:3 nowadays ... weird decision Jun 05 12:13:38 bah, and it uses only a 4430 Jun 05 12:14:08 ndec: omap4/panda is still our refernce platform wrt arm Jun 05 12:14:16 ndec: so yes, we'll do panda images Jun 05 12:14:43 ogra_: i didn't know about this tablet ;-) Jun 05 12:14:51 ndec: if omap4 delta against upstream is small, we could collapse omap3/4 in master and leave a topic branch for omap5 Jun 05 12:15:14 which kernel version are you shooting for? Jun 05 12:15:34 ndec: 3.5 for sure, but if 3.6 is out in time, we could pick it Jun 05 12:15:54 ok. there will be very minimal enablement effort by us on 3.5. Jun 05 12:16:12 we are going to spend some more time on 3.4 for customer/products constraints. Jun 05 12:16:23 ndec: ah, and we are going to support any omap5* reference/broadly available that comes out before 12.10 Jun 05 12:16:24 that includes the tilt tree. Jun 05 12:16:33 ok Jun 05 12:16:50 ppisati: i don't think OMAP5 has been discussed. Jun 05 12:16:56 i mean between canonical and IT Jun 05 12:16:57 TI Jun 05 12:17:16 nope Jun 05 12:17:18 AFAIK Jun 05 12:17:35 but i pick andy's tilt code, so omap5 support is included Jun 05 12:17:40 yes. Jun 05 12:17:44 cool Jun 05 12:17:59 but there won't be a 3.5 kernel with omap4+omap5 Jun 05 12:18:05 from us nor andy Jun 05 12:18:09 ah Jun 05 12:18:15 so you are going to stay on 3.4 Jun 05 12:18:19 yes. Jun 05 12:18:23 ok Jun 05 12:18:27 about 3.6? Jun 05 12:18:34 too far away! Jun 05 12:18:36 ;-) Jun 05 12:18:39 ok Jun 05 12:18:53 so 3.4 is going to be well supported, i see Jun 05 12:19:04 do you know which kernel will jelly bean use? Jun 05 12:19:26 ndec, i think its based on some lenovo device (medion is owned by lenovo) Jun 05 12:19:34 if you use mainline'ish kernel for OMAP4, you are missing 4460 support (e.g. it will run at 600Mz, instead of 1.2). you are missing PM (DVFS, thermal, ...), and you are missing audio analog. Jun 05 12:19:50 but you should get everything else. Jun 05 12:19:57 including some decent HDMI (display+audio) Jun 05 12:20:39 ppisati: jelly bean, there is no official information on it. there are 3.4 branches in google trees. so i guess that's what they will use. Jun 05 12:20:42 hopefully some of this stuff will enter 3.[5|6] Jun 05 12:20:52 i see Jun 05 12:21:05 but our decision to stay on 3.4 is not correlated with android. we need to fix some PM issues in 3.4 for a product. Jun 05 12:21:30 ogra_: and indeed fakeroot debian/rules clean binary in a fresh quantal-amd64 chroot chokes on tools too Jun 05 12:21:30 i think we should be able to get a 3.6 kernel. but not 100% sure. Jun 05 12:21:48 ndec: ack Jun 05 12:21:54 ppisati, aha, i guess tim should be able to help though Jun 05 12:21:58 ndec: btw there was a nasty bug about freq scaling Jun 05 12:22:05 ogra_: yep Jun 05 12:22:14 in 3.3, right? Jun 05 12:22:23 ppisati: well, what bug? Jun 05 12:23:13 ndec: kernel hangs after a period of forcefull freq scaling (using a dummy script to stress the cpu and then leave it idle) Jun 05 12:24:02 ndec: https://bugs.launchpad.net/ubuntu/+source/linux-ti-omap4/+bug/971091 Jun 05 12:24:03 Launchpad bug 971091 in linux-ti-omap4 "Pandaboard ES freezes with the default CPU scaling governor ondemand" [Medium,Confirmed] Jun 05 12:24:23 ndec: when you have time, can you try it (and leave a panda es overnight) with one of your kernel? Jun 05 12:24:39 * ppisati -> really out to get some food, brb Jun 05 12:50:35 ogra_, what is the status with armel, will we have the same images built with it as with armhf for 12.10? Jun 05 12:52:48 we dont build armel images since precise Jun 05 12:53:10 and by the looks of it armel will completely vanish soon Jun 05 13:11:46 janimo, what do you do with armel that you ask ? Jun 05 13:12:17 ogra_, nothing, just wondering if we should drop the ac100 armel kernels once armhf + graphics work Jun 05 13:12:37 we should drop them with the next upload imho Jun 05 13:12:44 oh right, I forgot there are no images anymore Jun 05 13:12:47 great Jun 05 13:12:56 I am rebasing on 3.1 now Jun 05 13:13:00 yay Jun 05 13:13:04 and I hope to have a working package soon Jun 05 13:13:10 sometimes I hate git too :) Jun 05 13:13:22 great, i'll upload my driver package then after the freeze Jun 05 13:13:23 especially when rebasing between wildly differing repositories Jun 05 13:13:35 heh Jun 05 15:12:45 hmpf, jasper seems unhappy on the current images Jun 05 16:09:19 janimo: did you acutally tried booting a 3.1 kernel with your config? Jun 05 16:09:38 marvin24, yes and it failed :) Jun 05 16:10:06 I guessed so Jun 05 16:10:33 mine works fine over here Jun 05 16:10:33 it was a shot in the dark really Jun 05 16:10:58 I hate merging config files and reviewing dozens of config entry docs :/ Jun 05 16:11:38 you should have kept my packaging ;) Jun 05 16:11:41 yeah, so big thanks for doing it anyway Jun 05 16:12:00 janimo: at least all the CONFIG_*TEGRA* stuff should be enabled Jun 05 16:12:16 I'll give a detailed reply on the list Jun 05 16:12:22 ok, I was not sure how much of it is tegra3 Jun 05 16:12:50 I'll do an updateconfig now but will take all hints from your mail too :) Jun 05 16:18:09 hmof, isntalling PVR on the panda image kills the package manager with held broken packages, fun Jun 05 16:22:12 eeek, infinity we need to patch flash-kernels default bootscr's for all arches i fear... Jun 05 16:22:18 cat /proc/cmdline Jun 05 16:22:19 ro Jun 05 16:22:24 ... Jun 05 16:22:49 indeed debian does neither use quiet nor splash Jun 05 16:23:07 though beyond that it works surprisingly well Jun 05 16:23:10 janimo: you cannot activate something specific for tegra3 if the cpu is not selected Jun 05 16:33:04 ogra_: If it works at all, that's something. :P Jun 05 16:33:21 yeah, it seems to work fine apart from my kernel issues Jun 05 16:33:35 (display isnt detected) Jun 05 16:33:42 Displays are overrated. Jun 05 16:34:49 especially on preinstalled desktop images without sertial output :P Jun 05 16:35:21 ;) Jun 05 16:43:45 lilstevie: do you accept bugfixes for your git tree? Jun 05 16:47:36 janimo: remember that the 3.1 doesn't need a mem=448M@0 option anymore Jun 05 16:47:50 it may fail to boot if you add one Jun 05 16:48:21 marvin24, oh so not only kernel changes but flash-kernel possibly. Jun 05 16:48:29 ogra_, do you know what the deal is there ^ ? Jun 05 16:48:55 f-k doesnt have any special options at all in the booltloader setup Jun 05 16:50:19 abootimg provides a standard bootimg.cfg ? Jun 05 16:51:16 nope Jun 05 16:51:28 f-k in quantal isnt tested on the ac100 yet Jun 05 16:51:40 that will likely need some tweaks Jun 05 16:52:26 i wasnt thinking janimo actually uses quantal on his test ac100 though Jun 05 16:53:48 I am not doing an SRU with 3.1 but I am on the precise userland when testing Jun 05 16:54:27 marvin24, I'll do some tests with the abootimg, I need to flash it anyway as the current kernel does not boot now Jun 05 16:55:15 you dont have sosboot installed ? Jun 05 16:57:29 ogra_, I have 2.2 android I think so lost sosboot quite a while ago Jun 05 16:57:41 I even forgot about it :) Jun 05 16:58:14 ah, sad Jun 05 16:58:23 makes working on the device so much harder Jun 05 16:58:32 I am not sure if I could not go back to 2.1 ar all or it was just too annoying Jun 05 16:59:00 well only bad kernel can mess it up, and I knew I'd have to resort to nvflash if I break it, oh well :) Jun 05 17:32:03 janimo, http://paste.ubuntu.com/1025386/ in case that helps you ... thats my config based on the 3.0 one, works fine since about six weeks for me Jun 05 17:32:34 ogra_: what's you mem line if any? Jun 05 17:32:54 tegrapart=recovery:300:a00:800,boot:d00:1000:800,mbr:1d00:200:800 root=UUID=f8d60f31-32db-4655-8276-332dd4c7e703 mem=512M@0M quiet splash Jun 05 17:33:04 thats my cmdline Jun 05 17:35:11 ok Jun 05 17:35:34 works fine with the binary driver Jun 05 17:39:25 yes Jun 05 17:39:40 I haven't tested mem=448M@0 yet Jun 05 17:40:02 well, mem=512M@0M somewhat sounds like we are stealing away all video ram Jun 05 17:40:15 but it doesnt seem to have any bad effect Jun 05 17:40:17 it is reserved by the kernel now Jun 05 17:40:29 I wonder if we need it at all Jun 05 17:40:41 (I'm on for-next and too lazy to compile a 3.1 kernel Jun 05 17:41:18 i can try that if i'm through with the alpha1 tests tomorrow Jun 05 17:42:27 no need, I can try it myself later on Jun 05 17:42:52 well, its not like we are in a hurry for quantal yet :) Jun 05 17:44:39 who made the stupid decission to release a distro every 6 month ... Jun 05 17:44:57 that's a lot of stress for everyone Jun 05 17:45:14 the initial ubuntu dev team did :) Jun 05 17:45:38 if we had more time, I would give u-boot a try Jun 05 17:45:43 get rid of fastboot Jun 05 17:45:48 and android partitions Jun 05 17:45:57 * ogra_ doesnt care ... Jun 05 17:46:22 if we could have an installer that wouldnt require flashing as the first step, that would intrest me Jun 05 17:46:50 but the few MB you gain by dropping the partitions i dont really care about Jun 05 17:48:58 remember that we still need this tegra/nvidia partitions patch Jun 05 17:49:09 and the ugly command line to support it Jun 05 17:49:47 hmm, no boot woth al those config options added Jun 05 17:50:22 janimo: maybe you should try to lay you config beside and try paz00_defconfig first Jun 05 17:50:47 janimo: can you post your config somewhere? Jun 05 17:50:53 I'll try it here Jun 05 17:51:34 * marvin24_DT just finished porting nvec to device tree and is in danger to get bored Jun 05 17:51:41 yes, I may just try a simple defconfig build Jun 05 17:51:44 I'll pastebin my config Jun 05 17:53:20 marvin24, http://paste.ubuntu.com/1025415/ Jun 05 18:06:27 janimo: Are there enough cookies in the world to bribe you to figure out what's up with mono on armel? Jun 05 18:06:50 infinity, vegan cookies? Jun 05 18:06:57 Of course. Jun 05 18:07:50 I am not suicidal enough from fighting git rebases and the slight twists in kernel packaging on various trees that make me dig for hours so why not? Jun 05 18:08:18 notice I talk like Ncommander when it comes to mono Jun 05 18:09:03 only armel but armhf fine. Ah this is the new armel for v5 ? Jun 05 18:09:59 infinity, is our armel like Debian's now? Jun 05 18:10:29 janimo: Ish. Jun 05 18:10:32 * NCommander bails out of the nearest window head first upon hearing NCommander and mono Jun 05 18:10:38 janimo: Debian is v4t, we're v5t, but close enough. Jun 05 18:11:04 but just enough not to freeload I guess :) Jun 05 18:11:18 janimo: infinity is it a FTBFS or something worse? Jun 05 18:11:32 janimo: Freeloading should, in theory, work. We may have toolchain issues confusing things. Dunno. Jun 05 18:11:40 NCommander, is there worse? :) glibc dobule free causing FTBFS Jun 05 18:11:42 The ARMv7 extensions were at one point #ifdef guarded, but upstream redid that patch a lot which may have broken it Jun 05 18:12:06 janimo: I recommend holy water and a sacrifice to $DIETY Jun 05 18:12:37 * janimo wonders which deities are monkeys Jun 05 18:13:14 Some Hindu ones, probably. Jun 05 18:13:29 infinity, wikipedia confirms, and chinese and japanese too Jun 05 18:13:48 a latino monkey deity would have been preferable though. oh well Jun 05 18:13:56 Heh. Jun 05 18:17:22 janimo: your config booted here Jun 05 18:54:02 how do i debug unaligned memory fault in kernel code? Jun 05 19:01:33 marvin24, oh nice. It may be a command line option then Jun 05 19:02:44 So I may get the boot part, modify the cmdline in some meaningful way and retry Jun 05 19:07:42 marvin24_DT, ^ Jun 05 19:07:57 marvin24_DT, is the nvec DT work required for mainlining? Jun 05 19:52:44 janimo: nvec is already mainlined Jun 05 19:52:55 it is required for the future ;-) Jun 05 19:53:16 tegra arch will remove board files soon Jun 05 20:39:32 * janimo is using a digital camera to film the ac100 screen while booting. Easier than going through all the trouble of figuring out how to get earlier kernel messages from a hung kernel Jun 05 20:40:53 marvin24, here is hangs after Freeing init memory: 336K Jun 05 20:41:36 unless I pass --no-log to the kernel it keeps saying it does not find init and ptys, but I see there are similar bugs in LP Jun 05 20:50:32 marvin24_DT: will the ASUS TF101 (transformer) be supported with this new endeavor? Jun 05 22:30:53 hi everybody Jun 05 22:31:27 nobody here? Jun 05 22:32:17 we're around, what's up? Jun 05 22:32:49 i need some help with the beaglebone. i think it's a stupid problem. Jun 05 22:34:15 I'm not familiar with that board, but I'm sure somebody here will help you figure it if you state what the problem is once they become available ;) Jun 05 22:34:32 thank you in advance :) Jun 05 22:35:16 anyway i just copied the preinstalled image on the sd card (zcat | dd)... put it in the board... and nothing happens! Jun 05 22:36:40 as in nothing on the monitor (if you have one connected) or nothing in the serial console (serial port connection)? Jun 05 22:37:19 I see two serial ports once I connect the usb cable (ttyUSB0/1) Jun 05 22:37:33 nothing appears on minicom Jun 05 22:37:50 it's 115200 8N1 Jun 05 22:38:11 and software and hardware flow control is set to No/None right? Jun 05 22:38:19 exactly Jun 05 22:38:53 thats correct then, strange that you get two USB TTY devices from a single USB to serial adapter Jun 05 22:39:14 what USB adapter are you using? Jun 05 22:39:27 mmm.. the one integrated on the board Jun 05 22:39:41 Ltd FT2232C Dual USB-UART/FIFO IC Jun 05 22:40:01 ahh, an embedded FDTI chip Jun 05 22:40:06 right Jun 05 22:40:36 should i check other serial ports? using ttl to rs232/usb? Jun 05 22:41:09 I'm looking up the documentation on the beaglebone, see if I see something you might have missed ;) Jun 05 22:42:55 thanks... i checked all resources on the web, for ex. elinux.org gives instructions for manual installation but those don't apply to my case Jun 05 22:44:10 interesting trick/suggestion in the readme to use this instead of minicom: screen `ls /dev/{tty.usb*B,beaglebone-serial}` 115200 Jun 05 22:46:12 ls: impossibile accedere a /dev/tty.usb*B: File o directory non esistente Jun 05 22:46:14 ls: impossibile accedere a /dev/beaglebone-serial: File o directory non esistente Jun 05 22:46:22 it means no such file or directory Jun 05 22:46:42 which revision of the beaglebone do you have? Jun 05 22:48:03 i think it's a3 Jun 05 22:49:11 i'm not sure... wait i'm double checking Jun 05 22:49:33 ok is A5 Jun 05 22:49:47 hmmmm Jun 05 22:50:55 ok, try the second ttyUSB in /dev/ if you haven't already Jun 05 22:51:25 already done... nothing Jun 05 22:51:55 what image file did you use to copy from on to the sd card/ Jun 05 22:51:56 ? Jun 05 22:52:20 the one for omap3 server... (ubuntu-12.04-preinstalled-server-armhf+omap.img.gz) Jun 05 22:52:50 that looks right to me Jun 05 22:53:09 do you have another sd card you could try it on instead? Jun 05 22:53:44 yes of course. what do i need to try? Jun 05 22:54:28 try that same image, but on your other sd card and then place that into the beaglebone just to see if maybe its a problem with the first SD card Jun 05 22:55:26 i tried to flash the same sd with Angstrom dist. and it worked like a charm. Jun 05 22:56:01 hmmm, so its just a problem with the ubuntu 12.04 server image huh? :/ Jun 05 22:56:33 mmm... i'll try with the desktop... Jun 05 22:57:31 the desktop image most likely does not have serial console enabled, of which is the case for pandaboard for example, so you'll need a monitor, keyboard, and mouse hooked up in order to make use of that image Jun 05 22:57:57 i don't have the dvi cape :( Jun 05 23:01:04 ok now i flash the sd with the desktop image... Jun 05 23:16:26 it's slower copying the desktop image that the server one Jun 05 23:16:48 despite the desktop image is 200MB less (compressed) Jun 05 23:24:27 ok finally... i'm testing Jun 05 23:25:55 nothing.. uff Jun 05 23:26:12 From what I have been told, beaglebone requires a different kernel. If so, desktop & server will both fail you. Jun 05 23:26:29 (same kernel on both images). Jun 05 23:26:52 oh... and where i can get the working kernel? Jun 05 23:28:19 That I don't know. Try asking on #beagle. Jun 05 23:28:32 thank you a lot. Jun 05 23:28:48 The Ubuntu omap image is (iirc) moving to community supported, so... Jun 05 23:29:30 ah...i didn't know that Jun 05 23:30:14 We'd be happy if that community told us what we needed to do to make it work on the bone. :P Jun 05 23:30:29 Well, don't take my word for it, I no longer work on this project (I just pass on knowledge). Jun 05 23:30:32 But I'm not going to buy the hardware and do it myself. -ENOTIME. Jun 05 23:30:54 infinity == slacker. :P Jun 05 23:31:25 eheheh Jun 05 23:31:30 >:( Jun 05 23:31:39 You should be able to do it in your sleep. Jun 05 23:32:24 I feel like there's a "that's what she said" joke there. Jun 05 23:34:32 ok, thank you guys, i'm going to sleep. bye and thanks again. Jun 06 01:02:23 my Raspberry Pi arrived today. :) Anyone running ubuntu on one yet? Jun 06 01:05:10 heathkid: /topic says don't do that Jun 06 01:05:43 s/don't/can't/ Jun 06 01:05:51 sorry Jun 06 01:06:31 back to my beagle-bone then... **** ENDING LOGGING AT Wed Jun 06 02:59:58 2012