**** BEGIN LOGGING AT Wed Apr 21 02:59:56 2010 Apr 21 07:40:56 morning Apr 21 09:48:49 persia: ping Apr 21 09:51:17 saeed: ? Apr 21 09:51:33 persia hey Apr 21 09:51:59 does hibernate work for you on dove? Apr 21 09:52:41 I don't have a dove board, so both "yes" and "no" seem like the wrong answer to that :) Apr 21 09:53:20 anyway, I noticed that sound settings are not restored after hibernate Apr 21 09:53:22 persia: saeed: going by what was said in the meeting yesterday, no would be the right answer Apr 21 09:53:52 saeed: Which sound settings? Apr 21 09:54:12 volume, mute, etc Apr 21 09:54:40 I mean hw settings Apr 21 09:54:46 Ah, hrm. Apr 21 09:55:34 ericm_: Any ideas? Apr 21 09:55:36 which part of the system should restore those settings after hibernate resume? Apr 21 09:56:21 saeed: add "/etc/init.d/alsa start" to after-resume scripts? Apr 21 09:57:03 hrw: where are those scripts located? Apr 21 09:57:42 /etc/pm/sleep.d/ but I don't see any on my lucid installs, making me thing there is another solution somewhere Apr 21 09:57:58 saeed: no idea - I do not have any machine running ubuntu Apr 21 09:58:27 hrw: Let us limit discussions to Ubuntu approaches here though ;-) Apr 21 09:58:31 monday will be funny day - working on ubuntu/arm without having usable machine for it Apr 21 09:58:51 hrw: Eh, you don't have any laptop/desktop running Ubuntu? Apr 21 09:59:14 lool: Debian on desktop, Debian on laptop Apr 21 09:59:23 if does is an ASoC board then ASoC layer should restore sound exactly as is Apr 21 09:59:26 dove Apr 21 09:59:27 hrw: I personally dual boot Debian on my laptop Apr 21 09:59:38 With a shared /home Apr 21 09:59:47 lool: with 80GB hdd I have dual boot Debian/winxppro Apr 21 10:00:53 considering buying asus ul30 as new platform so kubuntu will land there rather Apr 21 10:00:55 XorA: so `/sbin/alsa --force-reload` shouldn't be required? Apr 21 10:01:03 XorA: the dove has two sound cards (AC97 and i2s), both of them are asoc Apr 21 10:01:10 lool: Do you use Ext2fsd? Apr 21 10:01:16 persia: no, the machine/codec driver should restore Apr 21 10:01:28 * XorA doesnt have a dove or even know what it is Apr 21 10:01:31 kaouthia: No Apr 21 10:01:33 That explains the lack of /etc/pm/sleep.d/50alsa in lucid. Apr 21 10:01:35 but I did used to maintain ASoC Apr 21 10:01:36 never heard of it Apr 21 10:01:59 How do you access your /home from your windows partition? Apr 21 10:02:11 kaouthia: I dont use windows Apr 21 10:02:12 hrw: I do my ubuntu work on fedora here :-D Apr 21 10:02:16 it's a dual boot of debian and ubuntu Apr 21 10:02:31 lool: Ah okay Apr 21 10:02:32 * lool needs to disappear for an hour Apr 21 10:02:39 lool: both reside on same crypted lvm? Apr 21 10:02:58 XorA: How does that work? I didn't know most of our development scripts were available for Fedora (or do you work in a chroot/VM)? Apr 21 10:03:30 hrw: No reason they couldn't, given appropriate bootargs Apr 21 10:06:00 persia: VM combined with native on zoom2 board Apr 21 10:06:29 XorA: ah, yeah, VM is plenty sufficient. Apr 21 10:07:05 Is the Zoom2 fast enough to use for development directly? I was under the impression it was a bit RAM-low. Apr 21 10:07:32 XorA: can you explain how should machine/codec restore settings after hibernate? Apr 21 10:08:11 persia: I recompiled xf86-video-omapfb with no real issues, havent tried kernel yet Apr 21 10:08:32 saeed: look at the drivers in sound/soc/* Apr 21 10:08:43 No need: that you are using it for that indicates it's sufficient :) Apr 21 10:09:16 XorA: what I'm missing, is how can the driver get those parametes from? Apr 21 10:09:30 saeed: because the driver wrote them to the codec Apr 21 10:09:35 saeed: it remembers themk Apr 21 10:10:09 XorA: but after hibernate, the driver will be loaded as from reboot Apr 21 10:12:30 saeed: Im not very familiar with ubuntu on arm, I work on a different distro where we dont power down the chips Apr 21 10:13:46 XorA: but hibernate usually powers down the system, right? Apr 21 10:14:22 Indeed. That's the usual difference between "suspend" and "hibernate". Apr 21 10:14:26 XorA; please note the resuming from standby (suspend to ram) works fine to me Apr 21 10:15:10 * XorA has never hibernated an arm in that case Apr 21 10:15:34 XorA: I think it's the same on x86 Apr 21 10:15:39 hibernate as susped-to-disk-and-power-off? Apr 21 10:15:44 yes Apr 21 10:16:08 never did that on arm either Apr 21 10:16:50 actually mainline kernel doesn't support that, we have added that functionality Apr 21 10:16:56 On boot, we seem to use /sbin/alsa-utils to restore state with alsactl. Apr 21 10:17:00 hrw: ever feel we have walked into a strange new world :-) Apr 21 10:17:16 Might need support for that around hibernate in pm-utils Apr 21 10:17:57 saeed: Do you know if it works on x86? with Ubuntu? You might be hitting a general bug. Apr 21 10:18:10 (or powerpc) Apr 21 10:18:33 persia: no Apr 21 10:19:45 XorA: life without new challenges starts to be boring Apr 21 10:20:09 https://help.ubuntu.com/community/SoundTroubleshooting definitely has notes on working around hibernation issues. Apr 21 10:20:11 XorA: I returned from fosdem with devboard which I did not even connected serial yet Apr 21 10:20:34 hrw: hehe Apr 21 10:20:48 XorA: thinking about giving it for a friend as a gift Apr 21 10:20:50 * XorA is just trying to fix the zoom2 kernel so serial is not needed Apr 21 10:21:12 nice improvement over arm7tdmi which are high tech at his company ;D Apr 21 10:21:43 XorA, have you seen bug 567260 ? Apr 21 10:21:45 Launchpad bug 567260 in xf86-video-omapfb (Ubuntu) "xserver-xorg-video-omap* fail due to no /dev/fb (affects: 1) (heat: 10)" [High,Confirmed] https://launchpad.net/bugs/567260 Apr 21 10:22:02 ogra: thats what I fixed Apr 21 10:22:06 we have another two days for a fix Apr 21 10:22:18 saeed: Do you also lose settings on reboot? Apr 21 10:22:21 ogra: I can post patch to bug Apr 21 10:22:30 so it if it goes into archive tomorrow or friday we can have it in final Apr 21 10:22:32 persia: no Apr 21 10:22:35 its a one liner Apr 21 10:22:39 XorA, cool, please do so Apr 21 10:22:46 * persia looks for another bug Apr 21 10:22:57 gimme a few mins to replace kernel on zoom2 to extract bug Apr 21 10:23:14 oneliners are usually the most amazing changes Apr 21 10:23:54 I remember curses of one developer after his 4 hours of attempts to solve build error which I solved with oneliner for makefile Apr 21 10:24:49 persia, you want a bug ? i have plenty Apr 21 10:25:02 ogra: Do you have one about sound and hibernate? Apr 21 10:25:08 * persia is looking for a *specific* bug Apr 21 10:25:28 ogra: that bug is already detailing my one liner Apr 21 10:25:55 saeed: Try adding /etc/pm/sleep.d/50alsa as indicated at https://help.ubuntu.com/community/SoundTroubleshooting#Getting%20ALSA%20to%20work%20after%20suspend%20/%20hibernate and see if that works. If not, it might give some hints on a path to resolve it. Apr 21 10:26:53 persia, nope, but since i joined the installer team yesterday i can provide you with 100's of installer bugs if you are desperately in need for a bug :P Apr 21 10:26:54 hrw: one particularry fun online fix that took one day to locate.. Apr 21 10:26:59 -char foo; Apr 21 10:27:04 +signed char foo; Apr 21 10:27:34 ogra: I have code that generates lists of unfiled bugs on a regular basis already, if I'm just hunting bugs :) Apr 21 10:28:00 haha Apr 21 10:30:21 persia: I'll try Apr 21 10:31:14 OTG working on boot awesome Apr 21 10:31:16 saeed: If that doesn't work, please file a bug (`ubuntu-bug audio`) Apr 21 10:31:34 XorA, which kernel ? Apr 21 10:32:03 ogra: it works on all linux-omap kernels Apr 21 10:32:08 ogra: specifically on my zoom2 Apr 21 10:32:14 XorA, not ubuntu then Apr 21 10:32:23 ogra: you guys could add it Apr 21 10:33:44 XorA, out kernel team uses mainline only on omap Apr 21 10:34:04 tahst why the omap kernel is actually .33 and not .32 like the rest of ubuntu Apr 21 10:34:16 (except fsl-imx51) Apr 21 10:34:26 yeah Apr 21 10:34:28 ogra: your not allowed to patch at all? Apr 21 10:34:31 ogra: so torvalds tree instead of linux-omap tree? Apr 21 10:34:32 Oh, and -rt Apr 21 10:34:36 persia, which is why i said omap :) Apr 21 10:34:44 hrw, right Apr 21 10:34:49 XorA, we are Apr 21 10:34:51 ogra: "rest of ubuntu" was my picky point Apr 21 10:35:02 amitk, backports/cross ports bits and pieces Apr 21 10:35:04 XorA: Debian policy was: mainline kernel + backports from mainline. probably same/similar for ubuntu Apr 21 10:35:18 persia: it didn't work, I'll fill a bug Apr 21 10:35:19 XorA: pain for omap Apr 21 10:35:34 saeed: Thanks. Apr 21 10:35:40 hrw, well, whatever is maintainable, if amitk decides a certain patch from -omap is valuable that will also be pulled in Apr 21 10:35:57 its up to our kernel maintainer ... Apr 21 10:36:05 i'm only a consumer and tester ;) Apr 21 10:36:06 ogra: http://git.openembedded.net/cgit.cgi/openembedded/tree/recipes/linux/linux-omap-psp-2.6.32/0042-musb-allow-host-io-without-gadget-module.patch Apr 21 10:36:56 eggonlea: you aorund? Apr 21 10:38:56 XorA: the idea is to further the mainlining of omap by using mainline kernels. I've talked to Tony about this and he agrees. linux-omap is more like a buffer to mainline. Apr 21 10:39:52 amitk: I know this, linux-omap diffs have rapidly decreased over mainline Apr 21 10:39:54 NCommander: yes. Apr 21 10:40:08 and yes, i need to find time and pick out the right patches from 34-rc or linux-omap to unbreak OTG Apr 21 10:40:21 amitk: see patch above :-) Apr 21 10:44:07 XorA: thanks, I shall Apr 21 10:44:30 amitk: Im unsure of the pushed to mainline status, but Ajay is the TI USB guy Apr 21 10:44:39 amitk: http://cgit.openembedded.org/cgit.cgi/openembedded/tree/recipes/linux/linux-omap-psp-2.6.32/ is worht checking anyway Apr 21 10:45:35 * XorA really hopes ubuntu speeds the mainlining of zoom2/3 patches Apr 21 10:46:32 zoom and beagle hopefully Apr 21 10:47:41 ogra: beagleboard needs first really stable working version Apr 21 11:01:15 * amitk disappears for lunch Apr 21 11:08:29 * ogra lols about XorA's patch for omapfb Apr 21 11:08:58 XorA, i tought you had added any kind of autodetection Apr 21 11:09:44 i know asac tries to add that before final Apr 21 11:20:31 ogra: I may code that up one day, but for now I just need lucid running Apr 21 11:59:17 hrw: ecryptfs actually Apr 21 12:02:50 ogra: do you know why omapfb instead of fbdev is used as reported in 567260? Apr 21 12:03:06 lool: it crypts files but not filesystem - right? Apr 21 12:03:29 hrw: It's a filesystem on filesystem Apr 21 12:03:38 so it crytps filesystem objects in filesystem objects Apr 21 12:03:44 e.g. a dir to a dir and a file to a file Apr 21 12:04:04 ok, I prefer crypted lvm - handle also swap so I do /boot + crypto-lvm(/ + swap) Apr 21 12:10:08 shashi, its not used by default atm, only if you install the omapfb package Apr 21 12:11:50 ogra: ok thanks....i was inputing some comments/ thoughts in 567260 Apr 21 12:12:37 shashi, thanks :) Apr 21 12:14:20 ok, tickets for uds bought/booked Apr 21 12:15:27 clap clap clap Apr 21 12:15:36 looking forward to meet you there :) Apr 21 12:18:02 same Apr 21 12:24:03 hrw, i was looking at the musb host/device patch, haven't tested it yet, but how's it working for you? Apr 21 12:24:18 rcn-ee: I did not updated my beagle to this kernel yet Apr 21 13:32:13 saeed, could you update your resume from suspend status to bug 551491? Apr 21 13:32:15 Launchpad bug 551491 in linux-mvl-dove (Ubuntu) "BSP update for Marvell Dove kernel LSP 5.1.0 (affects: 1) (heat: 8)" [High,In progress] https://launchpad.net/bugs/551491 Apr 21 13:51:57 WOHOO+ Apr 21 13:52:37 * ogra_beagle wipes this install and starts over with the latest image hoping all is fine now Apr 21 13:53:33 (II) XINPUT: Adding extended input device "bmi-lcd-ts" (type: TOUCHSCREEN) Apr 21 13:53:34 ;D Apr 21 13:53:49 congrats Apr 21 13:54:00 other omap3 board Apr 21 13:59:20 hrw: what board is this? Apr 21 13:59:25 bug 2.0 Apr 21 13:59:29 prototype Apr 21 13:59:36 standard image? Apr 21 13:59:41 or custom kernel? Apr 21 13:59:45 no ubuntu Apr 21 14:00:06 amitk: I am returning device on days so no ubuntu tests for it Apr 21 14:04:06 ogra: you said something about the new image allowing me to install to sd itself? Apr 21 14:05:34 amitk, the netinst one, yes Apr 21 14:06:14 how do i put a custom kernel on those? Apr 21 14:06:28 amitk, pos-install Apr 21 14:06:31 *post Apr 21 14:06:42 you cant do it in the image Apr 21 14:07:05 right Apr 21 14:07:47 d-i is closely built with the kernel Apr 21 14:09:59 ericm_: done Apr 21 14:19:19 amitk: http://wiki.debian.org/DebianInstaller/Modify/CustomKernel Apr 21 14:19:38 zumbi: ? Apr 21 14:20:14 zumbi, heh, thats ancient Apr 21 14:20:14 amitk: are not you asking how to use d-i with custom kernel? Apr 21 14:20:28 from a time where d-i was still on two floppies Apr 21 14:20:50 zumbi: ohh right. :) thanks, i'll take a look Apr 21 14:21:14 amitk, d-i changed completely ... thats even pre-ubuntu Apr 21 14:21:15 amitk: ogra is right, looks old Apr 21 14:21:30 its for 2.4 kernels :) Apr 21 14:21:32 ogra: do you know a more up to date doc? Apr 21 14:21:38 no Apr 21 14:21:46 hehe lol Apr 21 14:21:50 and its a hell lot of work even if i knew a doc Apr 21 14:21:52 amitk: i would like to have efikamx (iMX51) in mainline kernel Apr 21 14:21:59 2.4... Apr 21 14:22:29 ogra: using kernel-wedge? Apr 21 14:22:30 zumbi: want to submit patches? pointers to git tree? pointer to someone that has hardware? Apr 21 14:22:33 installing the existing image and just issuing dpkg -i linux-image .... *.deb is definately easier and a lot faster Apr 21 14:23:00 amitk: i have hardware, the bad thing is they use freescale SDK Apr 21 14:23:29 zumbi: that is fine if you don't mind wiping it over ;) Apr 21 14:23:40 ogra: right, for efikamx, i wanted to add it to flash-kernel which hooks kernel-package Apr 21 14:24:09 ogra: then use somethin similar to rootstock Apr 21 14:24:10 flash-kernel hooks kernel-package ?? Apr 21 14:24:29 flash-kernel is really only to set up the bootlaoder Apr 21 14:25:13 you shouldnt abuse it for other stuff, rather create a new tool that flash-kernel can call Apr 21 14:25:57 flash-kernel (debian version) calls mkimage on vmlinux and initrd Apr 21 14:26:08 yes Apr 21 14:26:24 I thought that was the best way to do for porting a new device Apr 21 14:26:27 but it doesnt do anything with kernel-package Apr 21 14:27:06 kernel-package provides me with a vmlinux and initrd, then flash-kernel could get me the right thing I need (uBoot format) Apr 21 14:27:14 and i dont think we even support kernel-package in ubuntu Apr 21 14:27:31 i might be wrong though, amitk might know better Apr 21 14:27:42 kernel-package is a great tool Apr 21 14:27:55 but i guess it doesnt properly hook into the ubuntu infrastructure Apr 21 14:28:14 -ENOkernel-package Apr 21 14:28:20 uhm.. that's sad Apr 21 14:28:38 you don't even need a debian_dir, it creates it for you Apr 21 14:28:50 yes, thats the scary part Apr 21 14:29:09 zumbi: it might work for mainline kernels, but not for 'managed/supported' kernels. Apr 21 14:29:35 but it is something the kernel team plans to look at in our sprint next week Apr 21 14:29:40 amitk: it works for efikamx, kirkwood sheevaplug, and some others Apr 21 14:29:48 (whether we should support it or not) Apr 21 14:30:02 zumbi, FSVO works Apr 21 14:30:13 what is FSVO? Apr 21 14:30:14 zumbi: sure, but we don't support that HW, so why would we care? (to be blunt) Apr 21 14:30:14 zumbi, as rootstock "works" to create an image Apr 21 14:30:33 you will never get the same as you have with a real install with rootstock ... Apr 21 14:30:47 and you will never get a proper ubuntu kernel package with kernel-package Apr 21 14:31:01 so its "some system that looks like ubuntu" in the end Apr 21 14:31:09 zumbi: though I would love for our kernel to support multiple boards with a single kernel tree. Apr 21 14:31:18 zumbi, FSVO -> for some value of Apr 21 14:31:32 well, I would like to find a way to have an embedded-installer which suits ubuntu and debian Apr 21 14:31:58 which handles all this bootloader, kernel details Apr 21 14:32:23 but it seems hard to find a consensus Apr 21 14:32:58 you cant do it with the concept of a distro Apr 21 14:33:04 you can only get close to it Apr 21 14:33:10 i.e. like rootstock Apr 21 14:33:31 but effectively you will always have a bunch of differences to the real thing Apr 21 14:35:41 yes, it is hard to be generic.. :-/ Apr 21 14:41:26 zumbi: At least in Ubuntu, we explicitly say we aren't building "embedded", so want d-i to just do the right thing. This approach happens to be viable for a huge volume of hardware that others may consider "embedded". Apr 21 14:42:24 zumbi: Rather than building a new embedded installer, what might be helpful is improving the image modification tools in a way that one can easily swap the installed kernel. Apr 21 14:42:46 And improving the kernel packaging scripts to make it easy to convert an arbitrary git tree into a packaging tree. Apr 21 14:43:59 NCommander: ping Apr 21 14:45:10 ogra: what is different compared to ubuntu if you go for a properly seeded rootstock fs and combine that with the kernel/bootloader packages? Apr 21 14:45:36 asac: All the base config stuff. Apr 21 14:45:57 persia: baseconfig like oem-config? Apr 21 14:46:06 asac, debconf Apr 21 14:46:24 rootstock surely doesnt seed all of debconf properly yet Apr 21 14:46:31 ogra: what pieces are missing there? Apr 21 14:46:32 asac: So, the base-config package went away, and the installer handles that. I ended up spending many months fixing a million little issues with moblin-image-creator and we determined it was easier to make ubiquity work for MID than to continue tracking down all the special cases. Apr 21 14:46:40 ogra: so its a matter of tweaking seeds? Apr 21 14:46:43 asac, no idea, i never made an effort to check that Apr 21 14:46:46 asac: No, base-config like the *REALLY OLD* debian package. Apr 21 14:47:00 its a matter of integrating rootstock better with debconf Apr 21 14:47:06 persia: right. but thats now oem-config? Apr 21 14:47:16 and using d-i modules instead of scripts Apr 21 14:47:32 asac: Not at all. There is no relation between those packages. oem-config does *some* of the configuration post-install. Apr 21 14:47:32 ogra: how do we know that d-i and ubiquity do all the same? Apr 21 14:47:32 thats the reason why i switched to oem-config by default Apr 21 14:47:41 persia: i agree on improving the kernel packaging scripts, but which ones? (debian, ubuntu, kernel-package, kernel-wedge,.. just reate yet another one) Apr 21 14:47:45 But oem-config expects a d-i base install. Apr 21 14:47:46 it does a good bunch of that using d-i modules Apr 21 14:47:50 for me it feels like there is deviation anyway, though you would call both systems "ubuntu" Apr 21 14:47:54 persia, it brings it Apr 21 14:48:01 and removes it after it ran Apr 21 14:48:06 zumbi: Ummm. I'm going to have to defer to the #ubuntu-kernel folks at this point :) Apr 21 14:48:28 asac, right, but target has to be to minimize the deviation Apr 21 14:48:31 ogra: No, it runs *some* d-i components. Some stuff never gets run in oem-config. Apr 21 14:48:40 asac, which the step to oem-config did for example Apr 21 14:48:55 ogra: right. just wanted to point out that quite a lot of details are probably not relevant for "ubuntu" Apr 21 14:49:06 ogra: yeah. for me a rootfs + oem-config is ubuntu ;) Apr 21 14:49:06 yeah, might be Apr 21 14:49:24 thats how we ship stuff to production somewhere too, isnt it? Apr 21 14:49:26 nobody researched yet if/what additional pieces are missing in a rootstock install Apr 21 14:49:53 asac: In terms of use of the "ubuntu" term, no, but we expect there to be a large number of small bugs with a rootstock-based install (just as there are many known bugs with vm-builder installs, or any other tools that aren't d-i) Apr 21 14:49:57 asac, not sure, thats something only OSG can answer Apr 21 14:50:24 asac: Nothing called "Ubuntu" ships without a d-i installer to my knowledge. Apr 21 14:50:42 right Apr 21 14:50:44 I know of specialised remixes that use other installers (the one for the Netwalker is an example). Apr 21 14:51:36 and the proper approach for rootstock might also be to run d-i with a preseed file for the second stage Apr 21 14:51:44 guys: if I boot x86-64 from 10.04b2 cd can I make bootable pendrive? Apr 21 14:51:47 instead of the script it uses atm Apr 21 14:51:50 Doing that makes it just a normal install. Apr 21 14:51:55 hrw: Sure. Apr 21 14:52:12 persia: i would like to see those bugs ;) Apr 21 14:52:18 at least one example Apr 21 14:52:30 asac, as i said, nobody actually researched the differences Apr 21 14:52:31 hrw: You might have to play some games to tell usb-creator to make the pendrive based on the cd in your drive though. Apr 21 14:52:46 asac: I'll dig up a URL. Apr 21 14:52:57 asac, you would have to diff the debconf DB between a real install and a rootstock one Apr 21 14:53:17 i'm very sure there *are* differences ... just dont knwo which Apr 21 14:54:43 asac: http://www.sharp.co.jp/support/ex-data/recovery.sh.tar.gz Apr 21 14:55:07 asac: http://www.sharp.co.jp/support/mit/doc/install.html has instructions with screenshots (in case you don't read Japanese) Apr 21 14:57:04 i wouldnt be able to follow that instruction ;) Apr 21 14:57:51 asac: http://translate.google.com/translate?u=http%3A//www.sharp.co.jp/support/mit/doc/install.html&hl=pl&langpair=auto|en&tbb=1&ie=Shift_JIS Apr 21 14:58:23 * hrw hails JS menu in bookmarks toolbar Apr 21 14:58:24 heh, cool. Japanese->English translation with polish controls Apr 21 14:58:27 P Apr 21 14:58:39 bookmarklets saves lot of time Apr 21 14:58:52 persia: s/hl=pl/hl=en/? Apr 21 14:59:39 hrw: If I didn7t already know what the buttons did, yeah :) Apr 21 15:00:17 persia: before google translate I used "Excite Japan" translator Apr 21 15:00:32 it was like 'put url', select 'shelf->something', press button Apr 21 15:00:33 Oh, if you skip the hl= bit entirely, it chooses your default once. Apr 21 15:00:53 hrw: Excite Japan does a slightly better job with the text, but is less good for web pages. Apr 21 15:01:24 time passed, less to do with japanese websites, so google is enough for me Apr 21 15:01:35 ~curse sharp Apr 21 15:02:14 GrueMaster: hmm, I may have spoken too soon on the dove suspend/resume. The first few times it worked for me but seem to be hitting a problem now Apr 21 15:02:20 The Netwalker is actually a nice bit of kit. Apr 21 15:02:29 And they seem to be supporting it some. Apr 21 15:02:49 No idea if it sells well enough to keep that (likely needs more overseas orders to justify worldwide release) Apr 21 15:03:06 netwalker is ugly... Apr 21 15:04:06 Yeah, it's a bit large, and it doesn't convert, and the A key is in the wrong place, but other than that, it works. Apr 21 15:04:36 and it is imx... Apr 21 15:06:14 Well, I guess. I'm more interested in something I can actually put in my pocket and use than which is the best technology inside. Apr 21 15:06:25 I have big aversion to xMB kernel patches Apr 21 15:06:37 As long as all the other SoCs aren't available as 4-5" handhelds, how cool they are isn't so important. Apr 21 15:06:56 I just use the stock kernel, and keep hoping someone will package a newer kernel. Apr 21 15:06:59 persia: I have n900 for playing Apr 21 15:07:14 hrw: You don't even want to get me started about the shortcomings of that device :) Apr 21 15:07:19 Nice internals though. Apr 21 15:07:34 persia: I just hope that netwalker team does not have anyone from zaurus teams Apr 21 15:08:04 I've no idea. Apr 21 15:08:09 2.4.18-crappix (OE/OZ internal name for 2.4.18-embeddix) was total disaster from sharp Apr 21 15:08:45 Dunno. It ships a kernel that works with Ubuntu, which is all I require in a kernel. Apr 21 15:08:54 2.4.18 + arm + pxa + others + bunch of others + wext updates + irda updates + bunch of fixes + bunch another ones Apr 21 15:09:09 and only first 4 parts were from sharp Apr 21 15:09:14 For this one, amitk said he managed to upstream most of the patches. Apr 21 15:09:34 cool Apr 21 15:11:04 * amitk looks up Apr 21 15:11:27 amitk: where is imx51 mainline branch? Apr 21 15:11:37 zumbi: in linus's tree? :) Apr 21 15:11:46 *linus' Apr 21 15:11:55 amitk: we have efikamx stuff http://gitorious.org/efikamx/linux-efikamx Apr 21 15:12:51 amitk: i do not see any mach-mx5 in linus' tree Apr 21 15:13:12 2.6.34-rc? Apr 21 15:13:20 2.6.33 was looking via lxr Apr 21 15:13:34 it went in in 2.6.34 Apr 21 15:13:41 oh! ok, sorry Apr 21 15:15:20 amitk, argh ! Apr 21 15:15:42 amitk, read error on /dev/mtd2: cannot allocate memory Apr 21 15:16:23 thast where ubiquity dies now :/ Apr 21 15:17:07 That's flash-kernel-installer Apr 21 15:18:00 ogra: i thought you got mtd write working... Apr 21 15:19:59 amitk, it works with d-i Apr 21 15:21:12 amitk, worst is that i dont actually read at all Apr 21 15:22:35 ogra: is dd able to read/write? Apr 21 15:22:42 amitk, seems its fw_setenv thats failing to write Apr 21 15:22:47 yes, its dd able Apr 21 15:22:55 but mtd2 holds the uboot config Apr 21 15:23:09 ogra: Are you passing fw_setenv a mtd or a mtdblock device? Apr 21 15:23:13 i only dd to ntd3 and 4 Apr 21 15:23:30 echo "/dev/mtd2 0x0000 0x20000 0x20000" > /target/etc/fw_env.config Apr 21 15:23:34 lool, ^^^^ Apr 21 15:25:01 sounds good Apr 21 15:25:10 ISTR the u-boot tools are using mtd calls for io Apr 21 15:25:18 which would fail on mtdblock Apr 21 15:25:23 i wonder if i need to wipe it like the other ntd devices before writing Apr 21 15:25:38 i dont zero it out at install time Apr 21 15:25:51 which i do with mtd3 and 4 Apr 21 15:27:07 ogra: Have you looked for smarter tools in mtd-utils than dd? Apr 21 15:27:25 dd isn't precisely eraseblock aware. Apr 21 15:27:30 flash_eraseall + nandwrite -p? Apr 21 15:28:02 first cleans mtd, second writes with padding Apr 21 15:28:53 ogra: Oh you're dd-ing to mtdN? Apr 21 15:29:08 ogra: To write kernels and initrd, you'd better write them directly to mtdblockN Apr 21 15:30:29 hrw: Yes, that was what I was thinking, assuming it's NAND. Apr 21 15:32:10 lool, i'm just doing what all other arches in flash-kernel do Apr 21 15:32:14 persia: I do not remember when last time I used NOR Apr 21 15:32:30 mostly NAND or OneNAND recently Apr 21 15:32:50 hrw: My netwalker has redboot on NOR, but I don't have any interest in changing it, since the behaviour is sufficient for me. Apr 21 15:33:11 lool, flash-kernel was minaly a copy/paste job with some adjustments Apr 21 15:33:15 *mainly Apr 21 15:33:49 i can reliably reproduce fw_setenv breakage though Apr 21 15:34:25 andi suspect it is because uboot-envtools was never recompiled Apr 21 15:34:34 ogra: copy&paste of dd against NAND isn't best, even if it's the same as something else. Apr 21 15:35:00 persia, it is the same code for *all* platforms in flash-kernel Apr 21 15:35:19 persia, rewrite flash-kernel if you feel it behaves wrong :) Apr 21 15:35:20 Then I say it's all dangerous at best, and likely bad. Apr 21 15:35:39 Maybe, after I deal with other things. Apr 21 15:35:44 i'm just using the existing functions in there Apr 21 15:35:50 and flash-kernel isnt my problem Apr 21 15:35:51 But dd against NAND significantly reduces NAND life. Apr 21 15:36:07 my problem is flash-kernel-installer Apr 21 15:39:51 http://paste.ubuntu.com/419902/ Apr 21 15:39:58 anyone having an idea ^^^^ ? Apr 21 15:42:22 amm Apr 21 15:42:24 http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=512589 Apr 21 15:42:25 Debian bug 512589 in uboot-envtools "fw_setenv: Cannot malloc -131072 bytes: Cannot allocate memory" [Normal,Fixed] Apr 21 15:48:46 * ogra is desparate Apr 21 15:50:37 lool, any idea ? Apr 21 15:50:56 * ogra doesnt get why it works in d-i Apr 21 15:51:17 i have done about twenty netinst installs now and never hit it Apr 21 15:51:27 and three -server Apr 21 15:51:37 (alternate) Apr 21 15:54:06 persia, btw, looking at the code, flash-kernel translates /dev/mtd to /dev/mtdblock at some point Apr 21 15:54:33 so the writing goes to mtdblock Apr 21 15:55:09 hi. i'd like to be mount the daily .img file on my desktop, instead of using dd to write to SD card, e.g. I want to update the kernel in the .img file. i tried mount -t vfat -o loop xxx and it does not work. how can i do that? Apr 21 15:55:11 * persia is slightly mollified Apr 21 15:55:33 ndec: You need to use losetup and kpartx: there are multiple partitions on the image Apr 21 15:55:39 * persia digs up a reference Apr 21 15:55:43 there arent Apr 21 15:55:51 there is just one Apr 21 15:56:00 ogra: that's what I thought... Apr 21 15:56:06 * persia is confused Apr 21 15:56:10 but hwat persia said is still true ... either mount with offset or use kpartx Apr 21 15:56:24 ndec: http://blog.dustinkirkland.com/2008/10/mounting-kvm-disk-image.html Apr 21 15:56:30 ogra: what is the offset? what is the right command? Apr 21 15:56:31 you need to use an offset that omits the MBR Apr 21 15:56:41 ogra: Ah, so there's only one partition, but it is a partition. That makes sense. Apr 21 15:56:51 i think something like -o offset,512 or some such Apr 21 15:56:59 or offset=512 Apr 21 15:58:00 ndec: use kpartx Apr 21 15:58:02 Oooh, `kpartx -av foo.img` is supposed to just work. That saves effort. Apr 21 15:58:10 ogra: thx. sudo mount -t vfat -o loop,offset=512 is working! Apr 21 15:58:11 Yes Apr 21 15:58:18 (remember to use `kpartx -dv foo.img` later) Apr 21 15:58:24 mount -o loop,offset=512 imagefile.img /mountpoint shoudl work Apr 21 15:58:26 You have to losetup it first Apr 21 15:58:33 ogra, why don't I see omap images here? http://cdimage.ubuntu.com/netboot/lucid/ Apr 21 15:58:40 lool: Not according to the comments on kirkland's blog Apr 21 15:58:57 amitk, because they are ports and not officially supported in lucid Apr 21 15:59:02 persia: actually it works, but you can't tell what name it will have Apr 21 15:59:03 amitk: http://cdimage.ubuntu.com/ubuntu-netbook/ports/daily-live/current/ Apr 21 15:59:06 persia: so it's inconvenient in scripts Apr 21 15:59:14 amitk: http://ports.ubuntu.com/ubuntu-ports/dists/lucid/main/installer-armel/current/images/omap/netboot/omap/ Apr 21 15:59:24 persia, is right as usual :) Apr 21 15:59:56 ogra: That's not a good reason. The issue is that someone in cdimage (hint) needs to add the links to the template script. Apr 21 16:00:21 ogra: I'll dig you up a patch tomorrow if you're stuck (we can't land it until post-RC anyway) Apr 21 16:00:26 ndec: looking for netboot images, not netbook ;) Apr 21 16:00:31 persia, well, if i'm allowed to, i'm not sure what the status on supportability is here Apr 21 16:00:44 amitk: these are in the archive directly Apr 21 16:00:45 ogra: omap is *already* on cdimage. Apr 21 16:00:50 amitk: we dont copy them to cdimage anymore Apr 21 16:00:57 ogra: The issue is that they aren't likely to get to release Apr 21 16:00:58 right, just found out lool Apr 21 16:01:00 amitk: http://ports.ubuntu.com/ubuntu-ports/dists/lucid/main/installer-armel/current/images/ Apr 21 16:01:01 persia, i'm currently more worried about the fs_setenv bug that i cant solve Apr 21 16:01:08 oh right Apr 21 16:01:17 i have no idea whats wrong Apr 21 16:01:20 ogra: Go fix that. I'll give you a patch once the cdimage freeze lifts. Apr 21 16:01:28 and that it works flawless in d-i doesnt help Apr 21 16:01:46 persia, i'd love to, but i have no idea whats going on there Apr 21 16:02:03 amitk: Could you file a bug against ubuntu-cdimage about the missing netboot links? Apr 21 16:03:00 ogra: if I want to change the bootargs in boot.scr, do I just edit the file with emacs or do I need to regenerate something? Apr 21 16:04:19 ndec: usually i remove the stuff from top, and then run mkimage ... again Apr 21 16:04:24 after editing Apr 21 16:04:28 that works quite well Apr 21 16:05:18 GrueMaster, saeed: are either of you able to reproduce the suspend/resume problems I just updated the bug with on dove? Apr 21 16:05:43 plars: bug #? Apr 21 16:05:54 hey. had reconnect Apr 21 16:05:55 GrueMaster: https://bugs.launchpad.net/ubuntu/+source/linux-mvl-dove/+bug/551491 Apr 21 16:05:58 Launchpad bug 551491 in linux-mvl-dove (Ubuntu) "BSP update for Marvell Dove kernel LSP 5.1.0 (affects: 1) (heat: 8)" [High,In progress] Apr 21 16:06:04 if something was written i didnt get it Apr 21 16:06:13 * ogra had a OOM on his laptop Apr 21 16:06:15 sigh Apr 21 16:06:19 lucid sucks Apr 21 16:06:44 ogra: hmm. i didnt hear that ;) Apr 21 16:06:52 https://bugs.edge.launchpad.net/bugs/565981 Apr 21 16:06:56 Launchpad bug 565981 in xorg-server (Ubuntu Lucid) (and 1 other project) "[KMS] gem objects not deallocated (affects: 15) (dups: 2) (heat: 96)" [Critical,Confirmed] Apr 21 16:07:03 pitti just mailed -devel about it Apr 21 16:07:06 plars: What do you mean by "enable audio" Apr 21 16:07:08 ouch Apr 21 16:07:27 yeah Apr 21 16:07:45 i have to reboot once a day and see a lot of sad faces in chromium every day Apr 21 16:08:37 ogra: chromium != lucid ;) Apr 21 16:08:40 asac_, so with the fw_setenv bug i doubt we can have netbook for lucid Apr 21 16:08:46 (intentionally) Apr 21 16:08:51 asac_, chromium is just fallout Apr 21 16:09:04 in the end my whole desktop dies Apr 21 16:09:15 GrueMaster: unmuting the hp jack Apr 21 16:09:20 GrueMaster: so that audio works Apr 21 16:09:38 Ok, might want to clarify that in the bug report. Apr 21 16:09:47 GrueMaster: not sure that it's related, but I think it's about the only thing I did between the time it worked, and stopped working Apr 21 16:09:56 Otherwise it sounds like audio is disabled by default (it isn't). Apr 21 16:10:20 GrueMaster: not truly disabled, no, you just can't hear anything til you do that Apr 21 16:10:35 Yes, you can if you have the right hookup. Apr 21 16:11:12 asac, another reconnect ? Apr 21 16:11:14 I have a 2 RCA-to-mini female cable for just this purpose. Apr 21 16:12:26 Just because the platform is designed with different audio jacks than normal desktops doesn't mean audio is disabled or not working by default. Apr 21 16:13:17 Babbage has a special connector for laptop speakers and defaults to that (I'm still looking into the HP jack single channel issue on it though). Apr 21 16:13:25 have a nice rest of day Apr 21 16:14:22 * ogra has a screwed rest of day now :( after realizing the netbook image wont work Apr 21 16:18:37 ogra: no i was connected through a second irc client until i was able to get my new ip ;) Apr 21 16:18:48 ah Apr 21 16:18:51 GrueMaster: oh, ogra resolved that this morning Apr 21 16:19:08 resolved what? Apr 21 16:19:09 GrueMaster: use the hdphone jack instead, make sure it's set to audio output stereo Apr 21 16:19:19 GrueMaster: on imx51, the single channel problem Apr 21 16:19:40 ok, I'll try it here. Apr 21 16:20:12 GrueMaster, HDSET -> mono phone headset, HDPHONE -> headphones (my guess) Apr 21 16:20:41 you could probably take a look in the docs though to confirm that Apr 21 16:21:06 GrueMaster: seems that there's no good way to get input though Apr 21 16:39:06 plars: Why not plug a headset into the imx51 and try it that way? I have one here I can try. Apr 21 16:39:44 oops, different plug. Apr 21 16:41:23 GrueMaster: hmm? I'm not following. I did try it and it worked for me Apr 21 16:43:01 I was referring to the input. Apr 21 16:43:39 GrueMaster: I tried it briefly without success, but admittedly have not had much time to spend on it yet Apr 21 16:43:50 If that is a headset jack, then it should be able to work with a standard phone headset (proficed the plug is the same size). Apr 21 16:44:08 GrueMaster: ah, no I was just trying with a normal mic Apr 21 16:44:32 normal mic might be wired wrong for this. Apr 21 16:51:02 ogra: based on your previous comments, I assume the FlashKernel errors from Ubiquity on omap are still to be expected? Apr 21 17:05:46 plars: I can get recording working on my babbage using the mic from my babbage 1 plugged into the internal connector. Apr 21 17:06:22 * Martyn has audio working on the tegra2 .. but using a USB adapter Apr 21 17:06:39 I'll need to keep hacking away at the linux driver for the SoC based hardware Apr 21 17:06:52 Martyn: USB audio is a completely different driver set. Apr 21 17:07:17 Still alsa, but not soc audio. Apr 21 17:07:48 GrueMaster: Yep! Apr 21 17:07:57 GrueMaster: But I needed it to work temporarily :) Apr 21 17:08:01 so it was a good interim solution Apr 21 17:08:19 I nearly have the tegra2 built in audio working.. there's an issue with the buffer I'm trying to solve Apr 21 17:08:45 but I definitely have the mic input working now Apr 21 17:34:15 plars, yes and i have no idea how to fix it :( http://paste.ubuntu.com/419902/ Apr 21 17:34:29 it doesnt occur in d-i installs at all Apr 21 17:35:25 ogra: something, probably in ubiquity, is trying to grab a huge chunk of ram it seems Apr 21 17:35:54 ogra: any chance we could fiddle with the compcache percentages and make it happier? Apr 21 17:35:59 plars, well, at that point there is even real swap available Apr 21 17:36:23 ogra_cmpc: but order 5 is 128M right? Apr 21 17:36:45 ogra_cmpc: that is clearly going to fail under our current setup Apr 21 17:36:52 no matter how much swap we have Apr 21 17:36:56 ubiquity calls swapon dirctly after partitioning Apr 21 17:37:19 we use 128M compcache, right Apr 21 17:37:57 i wonder if we need it at all now that we run in only-ubiquity mode Apr 21 17:38:44 RAM should be sufficient to get through to the installer Apr 21 17:38:47 err Apr 21 17:38:49 partitioner Apr 21 17:39:13 ogra: ram only? or back to 25% compcache? Apr 21 17:39:29 plars, if you want to test that you could swapoff ramzswap after partitioning on tty Apr 21 17:39:34 or even before Apr 21 17:40:10 i wonder if we probably hit a general compcache issue here Apr 21 17:40:24 ogra_cmpc: yeah, will give it a try. Is swapoff enough to make it let go of that memory? Apr 21 17:40:32 rmmod probably too Apr 21 18:50:52 ogra_cmpc: ok, I did the swapoff and rmmod early (right after booting) and still got a couple of page_alloc failures early on, but I was able to complete the install! Apr 21 18:54:13 plars, dang ... not sure how to special case that in a nonintrusive way that slangasek accepts Apr 21 18:54:59 i'll think about it we cant do anything before RC anyway Apr 21 18:55:06 ogra_cmpc: I'd like to experiment a bit... I'm wondering if we could tone down the 50% to something like 35% and get away with it Apr 21 18:55:10 plars, thanks a lot for the test ! Apr 21 18:55:33 ogra_cmpc: at the very least, we have a cumbersome workaround Apr 21 18:55:38 sure, play if you like Apr 21 18:55:42 yeah Apr 21 18:55:52 very good work ! Apr 21 18:57:01 ogra_cmpc: I'll open a new bug against casper on this, rather than recycle the old one, unless you feel different? Apr 21 18:57:13 yeah, do that Apr 21 18:57:50 I think circumstances have changed enough that just tacking it on and reopening the old one would be confusing Apr 21 18:58:08 indeed Apr 21 18:58:22 though the question is if slangasek will allow it or not Apr 21 18:58:45 and i also think it might really be specific to the omap kernel Apr 21 18:59:20 i.e. .33 obviously has a new compcache implementation, there is very likely a reason for that Apr 21 19:00:45 would be intresting to see what a babbage board does in the same case if you boot with mem=256M **** ENDING LOGGING AT Thu Apr 22 02:59:57 2010