**** BEGIN LOGGING AT Thu Apr 14 02:59:58 2011 Apr 14 09:12:30 what's musb? Apr 14 09:13:19 musb is TI driver for OTG port in omap2-4 chip Apr 14 09:14:06 can you use more acronyms please :P Apr 14 09:14:12 :) Apr 14 09:14:41 and, do i need a particular cable for OTG? Apr 14 09:15:08 or any usb should do it? Apr 14 09:15:14 s/usb/usb cable/ Apr 14 09:15:23 which board? Apr 14 09:15:42 beagle Apr 14 09:15:54 which beagle := Apr 14 09:15:56 :) Apr 14 09:16:06 (model) Apr 14 09:16:15 for device you just use normal miniusb cable Apr 14 09:16:16 beaglxm in this case Apr 14 09:16:28 * hrw has only c3/b7 Apr 14 09:16:59 so, can i connect the xm to my pc using the mini-usb connector? the one used to power it? is it ok? Apr 14 09:17:06 i thought linaro is drowning in XMs, they didnt gove you one ? Apr 14 09:17:10 *give Apr 14 09:17:35 anyway, i've a plain beagle too Apr 14 09:18:04 yes, you can connect Apr 14 09:18:17 uhm, k Apr 14 09:18:22 ogra_: do I look like a8 collector? :D Apr 14 09:18:25 because we have 2 issues here Apr 14 09:18:32 heh Apr 14 09:18:34 first, in natty we lost the musb driver Apr 14 09:18:36 ogra_: I do not use beagles now Apr 14 09:18:45 https://bugs.launchpad.net/ubuntu/+source/linux/+bug/759913 Apr 14 09:18:45 who does anyway ;) Apr 14 09:19:18 and the second, is that the musb is kind of broken Apr 14 09:19:22 https://bugs.launchpad.net/ubuntu/+source/linux/+bug/608312 Apr 14 09:19:35 but i can't get the a_idle state with the cable connected Apr 14 09:19:48 anyway, i'll debug it a bit more Apr 14 09:19:58 ppisati, i would recommend talking to koen on #beagle Apr 14 09:20:10 koen? Apr 14 09:20:17 he probably has fixes for such bugs in angstrom already Apr 14 09:20:44 musb is broken in upstream 2.6.38 kernel Apr 14 09:20:50 Angstrom doesnt have a fix Apr 14 09:20:51 that's a patch provided by cooloney, should work in maverick (didnt;'t try yet) Apr 14 09:21:06 XorA|gone, ah, thx Apr 14 09:21:09 but the same patch makes my board hangs at boot in natty Apr 14 09:21:16 XorA|gone: what you mean is broken? Apr 14 09:21:37 ppisati: will go into host mode but will never negotiate a usb address successfully with device Apr 14 09:21:39 XorA|gone: it doesn't work? it doesn't attach? it panics? it format the sd and install freebsd? :D Apr 14 09:22:09 XorA|gone: but i should get the "a_idle" state with the cable connected, right? Apr 14 09:22:50 ppisati: I cant rmemeber the states, it does get you all hopeful and does detect device insertion Apr 14 09:23:02 ok Apr 14 09:23:06 I think agreen was looking into something similar Apr 14 09:23:44 ok Apr 14 09:23:50 anyway, time to debug some more Apr 14 10:16:42 does anyone know if the usb based booting on the panda can pull the kernel via usb too ? Apr 14 10:22:31 ogra_: omap4 bootrom loads xloader from usb and gives control to it - right? Apr 14 10:22:50 i think it loads MLO and u-boot Apr 14 10:23:09 but i'm not sure if u-boot then has access to the usb link to pull the kernel from there Apr 14 10:23:34 probably not Apr 14 10:23:49 then i wonder whats the benefit of it Apr 14 10:24:10 if i need the SD to hold the kernel anyway, it seems to be a bit pointless Apr 14 10:24:43 ogra_: if xloader is able to fetch uboot from usb then it is able to fetch kernel instead of uboot too Apr 14 10:25:24 hmm Apr 14 10:25:58 and if you merge initramfs into kernel image... Apr 14 10:27:03 yes, that could probably work Apr 14 10:28:09 ogra_: There is a tool from TI to do this USB boot stuff Apr 14 10:28:35 lool, i know, i read the doc, but it only talks about loading MLO and u-boot Apr 14 10:29:01 and io was wondering how to then get the kernel if u-boot doesnt have USB support ... Apr 14 10:29:18 though the merged kernel/initrd instead of u-boot loading might work Apr 14 10:29:42 * ogra_ tries to find an alternative way of providing PPAs without having to use the addon board Apr 14 10:30:52 lool, i'm collecting spec ideas ... i was wondering if we should have a flash-kernel rewrite spec so we can break up the stuff into assigned work items Apr 14 10:46:07 ogra_: Dunno Apr 14 10:47:27 janimo, ogra: armel rebuild with GCC 4.6 and binutils trunk started: https://launchpad.net/ubuntu/+archive/test-rebuild-20110419-arm/+builds?build_text=&build_state=failed Apr 14 10:47:33 yes, typo in the date Apr 14 10:48:55 go future Apr 14 10:49:16 looks pretty calm Apr 14 10:49:30 if you stop it now we wont have much work :P Apr 14 13:26:06 cooloney: dude, i can't musb to enter a_idle state Apr 14 13:26:12 flag@omap:~$ cat /sys/devices/platform/musb_hdrc/mode Apr 14 13:26:12 b_idle Apr 14 13:26:26 jcrigby: very belated pong :-) Apr 14 13:26:27 exactly, what do i have to do to reproduce your scenario? Apr 14 13:26:40 cooloney: lp608312 Apr 14 13:26:54 i'm powering my beagle xm via the usb socket Apr 14 13:27:08 is it enough? shall i get a_idle? Apr 14 13:41:34 doko_, i see you still have an open natty WI on the armel tracker Apr 14 13:42:02 do you plan to get to that (testing xvfb on hatty) or should it probably be postponed ? Apr 14 13:42:44 (or is it done already and just missing the paperwork) Apr 14 13:44:34 ogra: pointer? Apr 14 13:45:07 http://people.canonical.com/~platform/workitems/natty/ubuntu-armel.html and https://launchpad.net/ubuntu/+spec/multimedia-desktop-n-xorg-general-planning Apr 14 13:45:32 * ogra_ wonders why that spec is associated at all with the armel team Apr 14 13:54:07 ogra: hmm, don't know why. I appear to show up on different teams ... but it's done Apr 14 13:54:30 :) Apr 14 14:00:35 ogra_: you can boot and get everything you need by the musb Apr 14 14:01:02 rsalveti, ah, sweet, that could be an alternbative to the addon board for PPAs then Apr 14 14:01:07 also, there is a u-boot patchset that enables the smsc usb+eth device Apr 14 14:01:17 indeed we need a central USB server Apr 14 14:01:23 with that we can probably use tftp Apr 14 14:01:46 cool, so lets have a BOF and spec for that in O Apr 14 14:01:48 :) Apr 14 14:02:09 jcrigby: did you ever try this smsc patchset with your panda? Apr 14 14:02:17 * ogra_ added something to the spec ideas wikipage Apr 14 14:04:12 rsalveti, btw, i think "Send the pull request to the kernel team" WI can be closed ... request was sent :) Apr 14 14:04:36 ogra_: haha, sure, will close and put some more comment on the bug Apr 14 14:04:43 :) Apr 14 14:04:52 i guess we have to hope for an SRU here Apr 14 14:04:59 up to the kernel team though Apr 14 14:07:15 yes, I'm planning to make another patchset, and also send it upstream at the same time Apr 14 14:07:37 if we get good reviews, then it'll be easier to do an SRU Apr 14 14:07:43 yeah Apr 14 14:08:14 will just update the sgx drivers again and will get back to this patchset Apr 14 14:18:35 rsalveti, oh ! i thought the WI was only omap3 Apr 14 14:18:52 ogra_: it was Apr 14 14:19:11 "send it upstream, to omap4 branch and to Linaro, to get more feedback" Apr 14 14:19:29 ogra_: it should be included at omap 4 branch Apr 14 14:19:34 ah Apr 14 14:19:37 to have xrandr support Apr 14 14:19:39 i thought it was already Apr 14 14:19:40 that's fine Apr 14 14:19:45 no, not yet Apr 14 14:19:56 rsalveti, patchset? Apr 14 14:20:13 upstream to get more feedback Apr 14 14:20:24 k Apr 14 14:20:24 and linaro just to also add this functionality Apr 14 14:20:50 ogra_: that's why I didn't create more WI for it, because the original plan was just omap 3 Apr 14 14:20:55 sakoman, google is my friend. It was about the broken multiblock xtrers on old beagles Apr 14 14:21:02 jcrigby: the smsc one, that adds usb+eth support Apr 14 14:21:09 sakoman, sent a fix to list last night Apr 14 14:21:19 jcrigby: I'm about to build and try your fix Apr 14 14:21:35 rsalveti, oh good (on both) Apr 14 14:21:38 * ogra_ sees an u-boot upload, does that already include the fix ? Apr 14 14:21:49 rsalveti, is is patchset in history here? Apr 14 14:22:01 jcrigby: nops, saw it at the u-boot m-l Apr 14 14:22:55 rsalveti, oh ok, I'll go look. I was thinking yesterday that it really is a pain to plug/unplug sd card for every kernel or u-boot rebuild Apr 14 14:24:21 jcrigby: yes, at least tftp support would be nice Apr 14 14:25:25 no hurry though :) Apr 14 14:25:27 tftp support would be awesome Apr 14 14:25:31 hatty is nearly done Apr 14 14:25:34 *natty Apr 14 14:35:55 jcrigby: I guessed that from the discussion earlier Apr 14 14:36:08 I'll look at the list for the patch Apr 14 14:41:22 jcrigby: yup, worked fine Apr 14 14:41:28 jcrigby: will comment at the bug Apr 14 14:41:31 rsalveti, fantastic Apr 14 14:41:39 will reply with a tested-by too Apr 14 14:41:46 great Apr 14 14:44:54 jcrigby: ops, one bug Apr 14 14:45:03 jcrigby: xm ES revision is ES1.0 Apr 14 14:45:46 rsalveti, crap Apr 14 14:46:59 rsalveti, I put the ifdef in there to only get OMAP3 but I forgot about different omap3. Apr 14 14:47:44 so if 34xx and old is the condition Apr 14 14:47:55 then we avoid 36xx right? Apr 14 14:48:05 jcrigby: 35xx should be fine (probably this 34xx) Apr 14 14:48:11 jcrigby: afaik yes Apr 14 14:48:38 ok, I'll go rev the patch, may need to ask sakoman for help Apr 14 14:48:38 at least XM ES1.0 works fine with multi-block read Apr 14 14:48:55 right, patch as is will just slow it down Apr 14 14:49:59 jcrigby: was just reviewing the patch and noticed the same thing Apr 14 14:50:30 I think a run time check for cpu type and revision is needed Apr 14 14:50:37 sakoman, right Apr 14 14:50:45 since we have a single binary for both 35xx and 37xx Apr 14 14:51:07 there are still a few places in the kernel that make this same error iirc Apr 14 14:51:37 ok, so I will leave the ifdef to leave omap4 out Apr 14 14:51:44 and change the runtime check Apr 14 14:52:48 yeah, that should do it Apr 14 14:52:54 there is code in sys_info.c to get cpu type and revision Apr 14 14:52:56 I'm also afraid the same issue happens at the kernel Apr 14 14:54:18 sakoman, for kernel it needs to be just runtime checks right? Apr 14 14:54:29 correct Apr 14 14:54:42 since we want a single omap binary Apr 14 14:54:47 right Apr 14 14:55:05 u-boot is at least board specific, so you only need to deal with the board options Apr 14 15:05:11 sakoman, one more dumb question, what are AM35xx devices? Apr 14 15:06:50 jcrigby: I believe that is what TI calls "Sitara" Apr 14 15:07:03 ok, yes Apr 14 15:07:51 http://focus.ti.com/dsp/docs/dspplatformscontento.tsp?familyId=1875§ionId=2&tabId=2643 Apr 14 15:07:53 sakoman, ok changing to if ((get_cpu_family() == CPU_OMAP34XX) && (get_cpu_rev() <= CPU_3XX_ES21)) Apr 14 15:08:25 jcrigby: seems reasonable to me Apr 14 15:08:35 ok, thanks Apr 14 15:30:58 http://www.amazon.com/dp/B004EPV7TK/ref=xs_gb_A3BWIOFSXKB0OY?_encoding=UTF8&smid=A1KWJVS57NX03I&pf_rd_p=441937901&pf_rd_s=right-1&pf_rd_t=701&pf_rd_i=20&pf_rd_m=ATVPDKIKX0DER&pf_rd_r=1JPCN2R605ZZV5R7T0DN - Tegra2 tablet for 280USD Apr 14 15:33:18 cool Apr 14 15:39:45 its been abandoned by nvidia, i dont recommend it :) Apr 14 15:39:49 * Sarvatt has one Apr 14 15:49:12 Why was it abandonded? Apr 14 15:50:38 Sarvatt: harmony? Apr 14 15:50:47 yeah Apr 14 15:51:12 they moved on to a newer platform Apr 14 15:51:16 http://developer.nvidia.com/tegra/forum/honeycomb-harmony Apr 14 15:51:35 For our partners' Android devices, NVIDIA provides support until the hardware partner chooses to no longer support the device. So, for instance, NVIDIA will support the Xoom on all versions of Android Motorola requests until Motorola ceases to support the Xoom. The same goes for ViewSonic with the G-Tablet, Notion Ink with the Adam, Acer with the Iconia, LG with the Optimus 2X and so on. Apr 14 15:52:05 oh good, they updated it yesterday Apr 14 15:52:10 yes Apr 14 15:52:46 aside from that the viewing angles on the screen are horrific on it, it really hinders using it as a tablet Apr 14 15:55:08 haven't been able to find a decent replacement screen for it yet because its a really thin type, something like an asus transformer is well worth the extra 130 bucks Apr 14 16:00:17 * rsalveti lunch Apr 14 17:01:33 grrr. Apr 14 17:01:54 Yea! Apr 14 17:13:55 what is the package name of the new onscreen keyboard Apr 14 17:15:56 zaery: florence Apr 14 17:16:58 thx Apr 14 17:17:17 You should be able to find it in universe. Apr 14 17:19:35 installed, but how do i start it Apr 14 17:27:59 Not sure. Apr 14 17:28:30 There are two packages. florence & florence-applet. Apr 14 17:28:41 I'll load them in a sec and test. Apr 14 17:28:54 what image are you working with? Apr 14 17:29:12 natty x86 Apr 14 17:29:29 keyboard up. Apr 14 17:29:50 can it auto show/hide? Apr 14 17:30:18 I have no idea. I haven't worked on it or with it. Apr 14 17:30:33 I just remember it being packaged by a teammate. Apr 14 17:30:57 If you are using this on x86, why ask on #u-arm? Apr 14 17:31:32 david said to talk to his guys here :) Apr 14 17:31:44 ah. Apr 14 17:33:04 will also be looking at arm tablets soon Apr 14 17:35:58 Amazon is having a fire sale on the Viewsonic tablet. Apr 14 17:40:08 I'm not sure that florence is unity compliant. I installed it on my netbook (amd64) and it doesn't show up in the applets icons. I can launch it through the dash under Universal Access. Apr 14 17:42:23 david couldnt remember the name. he said it was brand new Apr 14 17:42:52 Yea, still very early development. Apr 14 17:47:21 o/ Apr 14 17:47:44 I'm having trouble getting Ubuntu to boot on a BB xM Rev B Apr 14 17:48:15 I did the install and the boot.scr to give me a serial tty login Apr 14 17:48:27 Which image are you using? Apr 14 17:49:15 GrueMaster, the maverick omap 3 preinstalled Apr 14 17:49:25 We have a headless image specifically for this purpose. http://cdimage.ubuntu.com/ubuntu-headless/releases/natty/beta-2/ Apr 14 17:49:41 GrueMaster, no, I want the GUI Apr 14 17:49:44 The preinstalled maverick images assume you have kvm. Apr 14 17:50:12 Then you need to add console=/dev/tty1 as well as serial console. Apr 14 17:51:09 GrueMaster, ok, and what's the deal with cdimage.ubuntu.com/ubuntu-netbook/ports/releases tree having both maverick and 10.10 as child directories? Apr 14 17:51:19 which one are you supposed to select? Apr 14 17:52:16 For maverick on beagleXM, follow the instruvctions at https://wiki.ubuntu.com/ARM/OMAPMaverickInstall. As to the release naming, one is a link to the other. maverick is the codename for 10.10 Apr 14 17:52:38 GrueMaster, ok, so one is a symlink to the other Apr 14 17:52:39 Line natty is the code name for 11.04. Apr 14 17:52:39 cool Apr 14 17:53:41 Also, there is an issue with booting the stock maverick image on beagleXM >rev A3. They changed the hardware right at release time. Apr 14 17:54:02 Requires a kernel fix (which is in maverick updates) to enable video. Apr 14 17:54:40 GrueMaster, that is the same as the one on the OMAPMaverickInstall page, correct? Apr 14 17:54:46 yes Apr 14 17:55:00 downloading the kernel and uimage from rsalveti's page Apr 14 17:55:07 Follow the directions on the https://wiki.ubuntu.com/ARM/OMAPMaverickInstall Apr 14 17:55:20 Very straight forward. Apr 14 17:55:26 GrueMaster, yes, those are the ones I'm following Apr 14 17:55:55 For grins, you should also try the new Natty Beta 2 image (just released today). Apr 14 17:56:30 see /topic for link. Apr 14 18:02:39 GrueMaster, might try it later, right now this is kinda (academic) project critical Apr 14 18:02:41 :) Apr 14 18:02:49 thanks for your help ! Apr 14 18:43:20 GrueMaster, what do you think about a packaging tools profiling spec ? to identify the slow areas of dpkg/apt/update-manager Apr 14 18:43:44 (i think you suffer most from the slowness) Apr 14 18:44:26 I think it might get resolved with armfp, but I don't know. I think it is related to fp emulation as opposed to hw fp. Apr 14 18:44:37 (armhf) ;) Apr 14 18:44:46 yea, that Apr 14 18:45:10 well, i think its a general issue on SD cards as well Apr 14 18:45:21 hardfloat will surely improve a bit here Apr 14 18:46:34 I get varying degrees of performance from different SD cards. For example, package updates are 2x faster on my 8G class 10 than on the microcenter 16G class 6 cards. Apr 14 18:46:59 well, but what means 2x faster ? Apr 14 18:47:15 its still likely 3x slower than say your x86 netbook Apr 14 18:47:32 I'd have to run time sudo apt-get update on both with the same base image to give details. Apr 14 18:48:22 but even flashing the sd cards is a noticeable difference. Apr 14 18:50:41 well, then not only the same base image but the same media Apr 14 18:51:09 well, was just an idea for a possible spec Apr 14 18:52:22 its just that it feels like the whole system is totally responsive but a dist-upgrade with 120 packages takes half a day Apr 14 18:52:32 I think the SD size has something to do with it as well. I flashed the omap netbook image onto a 4G microSD Class 6 at 8.0MB/s and the omap4 image on a 8G class 10 at 4.2MB/s Apr 14 18:52:38 at least thats my impression Apr 14 18:52:48 Yes, that is true. Apr 14 18:52:59 Part of it is download & SD io. Apr 14 18:53:23 i think its more the unpack/configure phase than download actually Apr 14 18:53:31 I see a 4-10x speed improvement using my internal mirror instead of ports.u.c. Apr 14 18:54:07 sure, we cant do much about the download speed ... Apr 14 18:54:09 Depending on activity/time of day. Apr 14 18:54:21 hence the internal mirror. Apr 14 18:54:24 but i think there is room for improvement for the debconf stuff Apr 14 18:54:46 Possibly. What about btrfs support? Apr 14 18:55:05 not sure how well resizing works with it Apr 14 18:55:08 fs benchmarks would be nice Apr 14 18:55:14 we should consider it for sure Apr 14 18:55:35 i was a bit in love with nilfs2 recently Apr 14 18:55:47 I remember it was argued against in maverick as it was still too early, and that carried forward with natty while we still had .35 based kernels. Apr 14 18:55:48 but it has no resize options at all :( Apr 14 18:55:56 ouch. Apr 14 18:56:07 i have seen measurements that show its even faster than btrfs Apr 14 18:56:29 and its a snapshot based fs Apr 14 18:56:29 ext3 faster than btrfs? Apr 14 18:56:37 you can roll back to any state you like Apr 14 18:56:42 Ah. Apr 14 18:56:56 ext3 is slower than btrfs i think Apr 14 18:57:03 at least on SD Apr 14 18:57:14 btrfs has specific SD optimizations Apr 14 18:57:41 I know there is a conversion tool. Wonder if it could be added to jasper. Apr 14 18:57:48 so it would actually be good to switch, but its a question of stability and feature completeness Apr 14 18:58:22 Definately some things to test in early Oneric alpha. Apr 14 18:58:30 we should find out what the actual last blocking factors are in ubuntu Apr 14 18:58:50 if its just some grub issues we dont need to care for example Apr 14 18:59:09 i'm not sure conversion is such a good idea Apr 14 18:59:28 That was on the n900 kubuntu-mobile wiki. Apr 14 18:59:39 you are gambling with the users data ... Apr 14 18:59:41 https://wiki.ubuntu.com/ARM/n900 Apr 14 19:00:13 The conversion could be run after grow-root. No user data to implode. Apr 14 19:01:01 I can do some experiments on it, but I should see about getting alsa-ucm support for beagle first. Apr 14 19:02:41 yeah, thats all oneric ... Apr 14 19:03:03 note it down as idea though :) Apr 14 19:05:15 Added to wiki. Apr 14 19:05:40 awesome Apr 14 19:23:32 GrueMaster, i have changed the monimal size req for headless from 4 to 2G on the wikipage Apr 14 19:23:53 Ok. Apr 14 19:23:54 (for headless thats sufficient) Apr 14 19:23:55 Hrm. Looks like btrfs is still in flux. Apr 14 19:24:00 yeah Apr 14 19:24:06 still not default in ubuntu Apr 14 19:24:22 i think arnd from Apr 14 19:24:23 Beyond that. They haven't finalized the format. Apr 14 19:24:29 #linaro works upstream Apr 14 19:24:31 https://btrfs.wiki.kernel.org/index.php/Main_Page Apr 14 19:25:44 well, the warning is for .31 Apr 14 19:26:02 we'll be at 39 or 40 in oneric Apr 14 19:26:13 * ogra_ goes to make some meeting coffee Apr 14 19:26:24 v2.6.37 (January 2011) On-disk free space cache, asynchronous snapshots, unprivileged subvolume deletion, extent buffer switches from a rbtree with spinlocks to a radix tree with RCU. Apr 14 19:26:41 From the changelog. Apr 14 19:29:47 yep Apr 14 19:31:48 line is up Apr 14 19:31:58 ergh Apr 15 01:14:11 sebjan: how many days are you going to be out for vacation? **** ENDING LOGGING AT Fri Apr 15 02:59:57 2011