**** BEGIN LOGGING AT Thu Oct 07 02:59:56 2010 Oct 07 03:27:21 prpplague: which kernel, omap4? Oct 07 03:53:08 hrm Oct 07 05:50:58 hello, i made a rootfs for arm using rootstock and I'm trying to run it in QEMU Oct 07 05:51:13 it boots but never gives me a console Oct 07 05:51:27 just stops after starting sshd and ntpd Oct 07 06:20:23 sebjan: do you know if we have a working x-loader for blaze es2.1? Oct 07 06:20:38 I'm basically using the same tree as panda, but compiled for sdp Oct 07 06:21:09 for some reason it just reboots while starting the kernel boot Oct 07 06:21:49 if I use the old x-loader vincent gave me (probably es1) it'll be even worse, rebooting at the time it tries to boot the kernel Oct 07 06:22:02 without giving any message, even with earlyprintk Oct 07 06:22:28 u-boot should be fine, I'm using linaro's one that's based upstream Oct 07 06:22:56 it could be that we need a fix in the kernel, but I don't even know now if the x-loader is correct Oct 07 07:09:16 rsalveti: hi! Oct 07 07:10:38 rsalveti: for blaze bootloaders: I have some, but they are based on different baselines than yours. Oct 07 07:11:38 rsalveti: regarding u-boot, this may be the biggest issue: blaze is not supported by Linaro u-boot, and the pinmux is made into the Linaro u-boot... Oct 07 07:17:53 morning Oct 07 07:44:49 JamieBen1ett: hi, did you got touchpad working under maverick? Oct 07 07:45:34 hrw: no, its on my list of things to do today. Seems after an apt-get upgrade my wifi has stopped working so now I need to use a wired connection :( Oct 07 07:50:16 JamieBennett: impressive list of failures Oct 07 07:50:35 hrw: right, I'm tempted at a full reinstall as others have reported it working Oct 07 07:50:54 hrw: don't have a lot of time to be investigating it ATM Oct 07 07:50:56 heh... I love that m$ way of thinking.. Oct 07 07:51:02 hrw: heh true Oct 07 07:51:38 my Debian rootfs had 6-7 years when I did reinstall just due to x86 ->x86-64 change Oct 07 07:51:53 hrw: impressive Oct 07 07:52:51 on 6 Nov 2006 I moved to amd64 so Debian rootfs will have 4 years soon Oct 07 07:53:01 but I do not use it since @canonical Oct 07 13:56:41 sebjan: cool, thanks for the email, just replied Oct 07 14:24:10 rsalveti: I am preparing a detailed answer, will be ready soon! Oct 07 14:24:12 f**k, Maverick's Initrd's are compressed by lzma on ARM. I didn't expect this. Is this the same on x86? Oct 07 14:24:29 mopdenacker, yes, everywhere Oct 07 14:24:49 It took me a while to figure out. Had to read the magic number to find this. Oct 07 14:25:29 lzcat initrd_new Oct 07 14:25:34 easily converted Oct 07 14:26:32 * ogra_ac haeds over to the office Oct 07 14:26:46 Hi ogra_ac ! Thanks! LZMA is much slower to decompress... doesn't it slow down the boot process for machines with fast I/O? On OMAP, I can imagine you boot faster because of the slower I/O (MMC for example) Oct 07 14:29:45 Here are the commands to extract the contents of a uInitrd file: dd if=uInitrd of=initrd.cpio.lzma skip=64 bs=1; lzcat initrd.cpio.lzma | cpio -id Oct 07 14:51:15 sebjan: cool, thanks Oct 07 15:28:18 sebjan: oh, huge capacitor Oct 07 15:28:23 we're just removing it Oct 07 15:37:51 ogra_ac: hi! how's dallas! Oct 07 15:38:40 warm :) Oct 07 15:40:26 ogra_ac: question: when is the boot.scr file generated after the partition resize on first boot? is it done in initrd or in real FS? Oct 07 15:41:52 ndec: It is done by jasper in intrd after resize. After that, it comes from /boot/boot.script and is copied to the boot partition by flash-kernel. Oct 07 15:45:07 sebjan: do you know any other difference between sdp and blaze? Oct 07 15:45:27 rsalveti: blaze looks nicer... Oct 07 15:45:34 haha, for sure Oct 07 15:48:48 GrueMaster: thx. Oct 07 15:49:17 GrueMaster: by the way, in the daily image the initrd image is not installed in /boot. is that normal? Oct 07 15:49:53 GrueMaster: does the initial initrd (e.g. the one in the image) have any differences from the next initrd images that are generated upon kernel upgrade? Oct 07 15:51:42 ndec: sebjan: mopdenacker: hey, blaze es2.1 up and running Oct 07 15:51:51 after we removed the capacitor it just boots fine Oct 07 15:51:51 rsalveti: cool! :) Oct 07 15:51:59 with linaro's u-boot and x-loader from my tree Oct 07 15:52:10 wow Oct 07 15:52:13 and our kernel, without any other change Oct 07 15:53:20 ndec: Not sure what you are seeing, but on my system with 20101006 + latest kernel, I have /boot/initrd.img -> initrd.img-2.6.35-903-omap4 Oct 07 15:53:53 rsalveti: Yesssss! Oct 07 15:54:20 robclark: 864x480 Oct 07 15:54:30 yup, that sounds right Oct 07 16:03:39 GrueMaster: One question related to ndec's question. On Blaze, boot.scr gets created with the wrong 'root=' setting. It seems that the 'root' variable if still set to /dev/mmcblk0p2 in the initramfs (scripts/local-bottom/jasper_setup) Oct 07 16:04:27 GrueMaster: do you know where this root variable is defined? Is supposed to have the same value as the 'root' kernel parameter? Oct 07 16:05:42 Is this a fresh image? jasper is supposed to change it to match the uuid of the boot partition, whis won't matter what drive type or location. Oct 07 16:05:48 which Oct 07 16:05:58 GrueMaster: but do you have the file initrd.img-2.6.35-903-omap4? i am seeing a broken link, not actual file Oct 07 16:06:08 Yes Oct 07 16:07:20 Have you run oem-config successfully? One of the last steps is that it runs some of the kernel post-install scripts (one of which generates the initrd.img file). Oct 07 16:07:36 It does this so that it can remove jasper. Oct 07 16:07:53 oh. i see... i was looking at FS before doing the install Oct 07 16:09:36 There are a lot of things that get done on first boot that help to keep the image small for downloading. Normal images wouldn't have these issues, but preinstalled images are a different animal. Oct 07 16:23:47 GrueMaster: I was talking about the initial uInitrd found in preinstalled images (downloaded a fresh one this morning). Oct 07 16:24:33 That is built during image build time and is not part of the image. Oct 07 16:24:45 At least the initrd.img isn't. Oct 07 16:24:53 GrueMaster: do you know where jasper sets this 'root' variable? Oct 07 16:25:22 I'd have to look at the jasper source. Give me a moment. Oct 07 16:28:03 Looking at the source, it pulls this from /proc/cmdline Oct 07 16:29:06 So the initial boot.scr (created during image creation) is where it is initially set. After jasper resizes the rootfs, it uses the UUID. Oct 07 16:29:35 GrueMaster: ahah, very interesting! Oct 07 16:30:22 GrueMaster: So, on the Blaze, even if we don't use the boot.scr file, it is still used as input. Oct 07 16:30:34 That's something we can fix then. Oct 07 16:31:01 GrueMaster: would you be so kind to tell me in which file you've read this, please? Oct 07 16:31:39 What do you need to fix? We have blaze working here with an updated MLO and u-boot.bin on the standard image. Oct 07 16:31:55 Using our boot.scr Oct 07 16:32:52 GrueMaster: ? you at least had to manually change the initial boot.scr to set mmcblk1p2, no? Oct 07 16:33:10 GrueMaster: on blaze the SD card is on mmcblk1, not mmcblk0. Oct 07 16:33:55 Oh. yes. THe image that we booted had already run jasper. Oct 07 16:35:10 GrueMaster: ok. so we are trying to run the image before jasper. e.g. do the resize and install on blaze. Oct 07 16:35:43 so we change the initial boot.scr with /dev/mmcblk1p2, install goes fine. but at the end the new boot.scr is generated with wrong label. Oct 07 16:36:11 According to rsalveti, all you need to do is change the MLO, u-boot.bin, and modify the root= line in the boot.scr. Oct 07 16:36:13 ndec: I didn't touch boot.src Oct 07 16:36:45 ndec: I just booted "manually", typing the commands in boot.scr, but didn't touch this file. Oct 07 16:36:58 ndec: let me check that this works. Oct 07 16:37:02 mopdenacker: ok. but what is your value for root? Oct 07 16:37:34 mopdenacker: settings bootargs or reading the boot.scr is the same, if you use the same values. the boot.scr is not read by jasper. Oct 07 16:38:15 GrueMaster: i agree with this. but mopdenacker tried and it didn't work... but given what's just ^^^ that might be an issue on our end.. Oct 07 16:38:37 ndec: root=/dev/mmcblk1p2 in the bootargs, but I left boot.scr with "root=/dev/mmcblk0p2" (didn't touch this file, since it wasn't used) Oct 07 16:38:46 Or so I thought Oct 07 16:40:31 GrueMaster: rsalveti: i have to go now... but I will come back later in the evening. in the mean time if you could try to run the install on blaze and check the partition label, and verify flash-kernel.conf, that would be nice. Oct 07 16:41:24 I'll check it as soon as Ican pry it away from rsalveti. :P Oct 07 16:44:52 * rsalveti reading Oct 07 16:48:21 yep, will try now :-) Oct 07 16:48:49 yee, second blaze display working! Oct 07 16:48:58 with help from vincent-laptop :-) Oct 07 16:50:05 rsalveti: nice to be working at the same place. I hope you will have opportunities to come to TI in Nice :-) Oct 07 16:50:57 mopdenacker: yeah, that'd be cool :-) Oct 07 16:53:15 GrueMaster: nah, didn't work. Even with a correct boot.src and root= boot parameter (/dev/mmcblk1p2), I still end up with root=LABEL=emmcroot in boot.scr after reboot. Oct 07 16:54:07 hey all --> how do I switch form netbook version to desktop version? (not from GUI) Oct 07 16:54:21 It took the label from /dev/mmcblk0p2, instead of using the one from /dev/mmcblk1p2 (the SD card). There's a bug somewhere, though I don't see where... Oct 07 16:54:35 mopdenacker: checking... Oct 07 16:54:42 mopdenacker: I'm trying right now with a new image Oct 07 16:55:38 GrueMaster: rsalveti : thanks! Gotta go. Good luck! Keep us posted! Oct 07 16:55:47 mopdenacker: sure, see ya Oct 07 17:08:19 orbarron: /etc/gdm/custom.conf Oct 07 17:08:24 (or do it from gui) Oct 07 17:08:38 or if you disable automatic login, you can choose on each login which session you want Oct 07 17:09:57 ahh thanks robclark Oct 07 17:48:27 getting error in installing xfce4 or lxde as no such package Oct 07 17:48:37 apt-get update working fine Oct 07 17:49:00 apt-cache search also not showing package entry Oct 07 17:49:14 what repository need to be add for marvick arm Oct 07 17:49:23 to get these packages Oct 07 17:52:50 any one can help me to get install lxde in arm Oct 07 18:04:42 jo-erlend_igep: you're luck, both network and display patches got applied Oct 07 18:21:17 mopdenacker, so here is the code snippet caring for setting the UUID http://paste.ubuntu.com/508141/ Oct 07 18:21:28 mopdenacker, i dont see how it could possibly be worng Oct 07 18:24:53 ogra, while your here, quick question ... is zeitgeist a "must have" ? or can I implement the searching in a different way ? Oct 07 18:25:17 devilhorns, i think ts a requirement , how else would you implement it ? Oct 07 18:25:50 ogra, not sure yet :) but I noticed that zeitgeist is python based, so I want to implement searching in a different way so it isn't slow Oct 07 18:25:59 (teh zeitgeist packages are properly maintained already, if another solution adds extra maintenance work we wont be able to use it) Oct 07 18:29:02 ogra, I haven't really thought much about a replacement yet, but was just curious if I "can" replace it :) Oct 07 18:29:40 mpoirier, hey man Oct 07 18:30:20 mpoirier, i saw your git pull request about ftrace, that's good, why not for ti-omap4 tree or master only linaro? Oct 07 18:32:04 devilhorns, with something thats maintained by the ubuntu-desktop team and that offers the same functionallity you can Oct 07 18:32:21 ogra, gotcha Oct 07 18:32:24 thanks Oct 07 18:32:39 (i doubt you will find such a thing, good luck :)) Oct 07 18:32:47 hehehe Oct 07 19:04:02 cooloney: sorry I missed you post - I stepped out for lunch. Oct 07 19:04:53 mpoirier, np, man. i saw your ftrace git pull request Oct 07 19:05:10 cooloney: Tim ask me to check it in to linaro first. Also, I haven't tested for omap4 - it should work but I won't submit something I haven't tested. Oct 07 19:05:26 mpoirier, ok, gotcha Oct 07 19:05:47 mpoirier, i will try to rebase you patch on our current omap4 tree Oct 07 19:06:19 cooloney: fantastic but make sure you test - again, I haven't tried on omap4. Oct 07 19:06:52 ogra: thanks! That's the code we've been looking at too. The only thing that could be wrong is the value of the 'root' variable. Oct 07 19:06:54 mpoirier, yeah, sure. i will test. so how did you test it? Oct 07 19:07:01 on omap3? Oct 07 19:07:11 omap3 yes. Oct 07 19:07:23 mopdenacker, right and thats handed over by the kernel to the environemnt Oct 07 19:07:36 I simply enabled the feature and tested a function. Oct 07 19:07:36 mpoirier, good, understand Oct 07 19:08:01 there is another set of patches to enable graph testing but that will be harder to put in. Oct 07 19:08:42 the patches don't apply properly *and* it is deep in assembler macros and obscure symbols... Oct 07 19:10:47 mpoirier, yeah, i tried that before. Oct 07 19:10:53 mpoirier, painful Oct 07 19:11:13 the graph tracing patch set ? Oct 07 19:11:56 mopdenacker: I'm just debugging it, doesn't work here because for some weird reason root must be wrong and then we hang at grep Oct 07 19:12:01 and I can't run the first boot Oct 07 19:13:42 ogra: interesting. I didn't know that the kernel was passing environment variables to the init process (did I get it right?) Oct 07 19:14:01 mpoirier, not try to test, but just apply with dynamic ftrace together Oct 07 19:14:08 many conflict Oct 07 19:14:14 but maybe it changes Oct 07 19:14:24 rsalveti: weird indeed.... Good luck! Oct 07 19:15:00 cooloney: your are right, the first patch set for dynamic tracer weren't applying properly. Oct 07 19:15:50 cooloney: I was able to apply the first patch set after a lot of work. Oct 07 19:16:05 mpoirier, yeah, ftrace is quite important. thx for your work Oct 07 19:16:15 but the final submission by Rabin (the author) applied seamlessly. Oct 07 19:16:30 mpoirier, so how about the upstream status of this dynamic ftrace patchset Oct 07 19:16:40 mpoirier, cool Oct 07 19:16:49 mopdenacker, well, the kernel is *supposed* to hand over the whole cmdline to init ... probably it doesnt Oct 07 19:16:59 cooloney: it has been accepted. Oct 07 19:17:23 mpoirier, it's in rmk's upstream or mainline? Oct 07 19:17:29 cooloney: now I have to get graph tracing in. I'll need help from the implementor - I have to hunt him down. Oct 07 19:17:54 mpoirier, great, author always helped a lot Oct 07 19:18:11 ogra: so each kernel parameter becomes an environment variable? Oct 07 19:18:11 cooloney: http://git.kernel.org/?p=linux/kernel/git/sfr/linux-next.git;a=shortlog;h=80be7a7f642719bf99fc49692fc77d6333f51a73 Oct 07 19:19:48 mopdenacker, each one that has an equal sign Oct 07 19:20:11 mopdenacker: very very cool! :-) Oct 07 19:21:50 bug 586386 Oct 07 19:21:51 Launchpad bug 586386 in linux (Ubuntu Maverick) (and 1 other project) "omap3 kernel should hand over all comdline args to the init environment (affects: 1) (heat: 27)" [Medium,Fix released] https://launchpad.net/bugs/586386 Oct 07 19:21:53 gah Oct 07 19:23:25 rsalveti: dd if=uInitrd of=initrd.cpio.lzma skip=64 bs=1; lzcat initrd.cpio.lzma | cpio -id Oct 07 20:12:45 robclark, http://kernel.ubuntu.com/git?p=ubuntu/ubuntu-maverick.git;a=shortlog;h=refs/heads/ti-omap4 Oct 07 20:13:01 GrueMaster: rsalveti: ogra: hi. i am back. did you try booting the pre installed image on blaze? Oct 07 20:13:20 Working on it now. Looks like a kernel bug. Oct 07 20:16:02 GrueMaster: root is not propagated in init properly? Oct 07 20:16:28 Just found out it is a kernel config option that was missed. Oct 07 20:22:18 ndec, right, as GrueMaster aid Oct 07 20:22:22 *said Oct 07 20:22:30 GrueMaster: which one? Oct 07 20:22:47 this is a bummer... Oct 07 20:22:52 ndec, seems on a rebase the fix for bug 586386 got lost Oct 07 20:22:53 Launchpad bug 586386 in linux (Ubuntu Maverick) (and 1 other project) "omap3 kernel should hand over all comdline args to the init environment (affects: 1) (heat: 27)" [Medium,Fix released] https://launchpad.net/bugs/586386 Oct 07 20:23:29 ndec, its not as bad as it looks though, we'll just offer a fixed uImage to put in place on first boot Oct 07 20:23:40 we're just testing the fix Oct 07 20:23:59 we can't upload the fix? Oct 07 20:24:09 we can, but only as SRU Oct 07 20:24:15 bad timing to find it Oct 07 20:32:24 ogra: yep. sorry about that. Oct 07 20:33:22 ndec, ?? Oct 07 20:33:36 ogra: i wish we found this earlier ;-( Oct 07 20:33:36 ndec, our kernel team screwed it up Oct 07 20:34:00 its good that yo found it at all, there are other bits relying on the feature Oct 07 20:35:06 that option was enabled in our master branch to support this feature for omap3 Oct 07 20:35:21 need to carry it to our ti-omap4 as well Oct 07 20:40:27 cooloney: still building the kernel? Oct 07 20:41:10 mopdenacker: I found one bug here, so we're going to also need a new uImage together with MLO and u-boot Oct 07 20:41:24 but then we can successfully boot blaze Oct 07 20:41:38 and the installer should just work fine :-) Oct 07 20:47:21 rsalveti: and new boot.scr, since it needs to have mmcblk1p2 Oct 07 20:47:31 ndec: yeah, right Oct 07 20:50:37 ndec: vincent-laptop: and about wifi and bt modules, should it work the same way as panda? Oct 07 20:50:57 rsalveti: well... no! Oct 07 20:51:18 sorry, don't know much about blaze hardware Oct 07 20:51:21 blaze has a different wlan chip (1283) instead of 1271 Oct 07 20:51:44 we don't have an up-to-date driver for 1283. but we are trying to get it. Oct 07 20:51:55 ndec: ok then, less things to check :-) Oct 07 20:52:23 rsalveti: well you can play with xorg.conf and create 2 desktops ;-) we tried this already! Oct 07 20:52:50 ndec: yeah, just enabled the second framebuffer, but didn't change X to use both screens Oct 07 20:52:58 there is an accelerometer and proximity sensor on blaze. i think drivers should be there. never tested. Oct 07 20:52:59 had to check the installer problem Oct 07 20:53:47 ndec: do you have the xorg.conf already around? Oct 07 20:53:58 what would be nice is virtual desktop supporting both screens so that we can move windows from top to bottom screen... and also support for multiple touchscreen Oct 07 20:54:15 yes, let me find it Oct 07 20:54:30 yeah Oct 07 20:56:31 rsalveti: http://paste.ubuntu.com/508281/ Oct 07 20:56:59 ndec: cool, thanks a lot Oct 07 20:57:06 will try in a bit Oct 07 20:57:45 cooloney: hey you got that card ready? Oct 07 20:58:18 rsalveti: since we are talking about blaze and x11... blaze has a custom keypad. we probably need to create a new kbd layout for it. for now we have a quick hack and we create /etc/X11/Xmodmap. but that breaks panda ;-) Oct 07 20:58:38 rsalveti: what do you recommend to handle the blaze keypad? Oct 07 20:59:24 prpplague, got some issue after upgrading, please wait for a while. sorry Oct 07 20:59:33 cooloney: doh Oct 07 20:59:45 cooloney: runnign out of time for today Oct 07 21:00:52 prpplague, i tried to dd out my 4G rootfs to a image file, and dd to your card. but cannot boot my panda Oct 07 21:01:17 cooloney: don't have a tar ball of the rootfs i can use? Oct 07 21:01:42 prpplague, so i am copying over our daily image to your SD card, and have to install it on the board. Oct 07 21:01:42 rsalveti: GrueMaster: ogra: I am trying to create pre installed image where our package (which are in OMAP PPA) are also pre installed. So basically we download cdimage, chroot in rootFS and install all our stuff. but do you think we can install new kernel as well? I am not sure if that should/would break the installer. Oct 07 21:01:42 ndec: don't know the right answer now, but probably creating a new key layout for it Oct 07 21:01:59 then we can identify if we're running blaze and load the correct kbd layout Oct 07 21:02:14 ogra: ^^^^ about the blaze keypad, in case you have an idea. Oct 07 21:02:37 ndec, you should be able to do it chrooted in the SD, it gets tricky if you loop mount Oct 07 21:02:45 (replacing the kernel) Oct 07 21:03:01 rsalveti: if you boot on blaze, and look at the X11 log, you will see that the keypad is detected (omap4-keypad). so it's probably just missing a config to load the proper keymap, right? Oct 07 21:03:01 prpplague, i don't have the tar rootfs. and if is the ubuntu minimal rootfs from rootstock, we need to install extra applications. Oct 07 21:03:48 prpplague, so if got the daily image installed properly, please keep the SD for development and testing Oct 07 21:03:54 ndec: probably Oct 07 21:04:09 ogra: in fact I had to loop mount the ext3 and fat32 as well. I created a fake flash-kernel.conf so that flash kernel would flash inside the .img. it seemed to work.. but I wanted to hear from you what you think. Oct 07 21:04:29 ndec: you can update the kernel, but just make sure you also generate a new uinitrd Oct 07 21:04:39 because of the proper kernel modules, if you're changing versions Oct 07 21:04:41 prpplague, actually i tried to tar out the rootfs tar ball, it is quite slow for me Oct 07 21:05:03 yes, otherwise modules aren't found in initrd. we had this the first time ;-) Oct 07 21:05:31 in fact it seems we were able to even flash the new uImage and uInitrd in the .img using flash-kernel ;-) Oct 07 21:05:44 ndec: yep Oct 07 21:06:00 the problem you can easily find on the pre-installed image is lack of inodes and space Oct 07 21:06:07 because you still didn't run the resize Oct 07 21:06:31 so you're kind of limited in what you can install and edit before running the first boot Oct 07 21:10:27 rsalveti: guys i'm dying here, has _anyone_ got an ubuntu file system i can use? Oct 07 21:11:01 prpplague: sure, I have a valid sd card here for panda Oct 07 21:11:28 prpplague, i'm installing the daily image on your SD, just a minutes Oct 07 21:11:58 rsalveti: well... i forgot to mention that we resize it first... well in fact everything is here: http://omappedia.org/wiki/Add_Packages_To_Ubuntu_Preinstalled_Images Oct 07 21:14:18 cooloney: ahh ok Oct 07 21:17:33 ndec: oh, ok Oct 07 21:17:38 ndec: so you're fine Oct 07 21:18:10 ndec, with fake flash-kernel i see no prob Oct 07 21:18:21 just make sure to remove it again ;) Oct 07 21:18:33 ogra: well, we forgot the first time ;-) Oct 07 21:18:34 cooloney: OK, I can confirm that jasper works with this kernel fix Oct 07 21:18:52 rsalveti, thx, man Oct 07 21:19:00 rsalveti: cool! is it cross compiled or native? Oct 07 21:19:16 ndec, my testing kernel is cross compiled Oct 07 21:19:17 given that it only took a few minutes I think I know the answer Oct 07 21:19:25 rsalveti, just tested that Oct 07 21:19:41 cooloney: can you post a native kernel somewhere later in the day? Oct 07 21:19:58 native building kernel? Oct 07 21:20:31 i will try to build it in our schroot environment Oct 07 21:22:52 ndec: I'm just publishing all the needed files for blaze to work Oct 07 21:22:55 with the pre-installed image Oct 07 21:22:59 so GrueMaster can actually test it Oct 07 21:23:04 hold a sec Oct 07 21:34:55 ndec: GrueMaster: http://people.canonical.com/~rsalveti/maverick/blaze/ Oct 07 21:35:35 rsalveti: cool! will test tomorrow... no blaze at home ;-) Oct 07 21:35:41 that should be enough to get blaze up and running with the installer Oct 07 21:36:11 uboot is based of linaro uboot, right? not mainline, nor TI uboot? Oct 07 21:40:25 ndec: based on linaro, but very very close with upstream Oct 07 21:40:40 the only missing part now is the new mmc driver Oct 07 21:40:45 but doesn't affect us Oct 07 21:41:15 rsalveti: can you send your patch to me and sebjan? Oct 07 21:41:37 ndec: what patch? Oct 07 21:41:38 :-) Oct 07 21:41:52 rsalveti: what? uboot linaro works out of the box? Oct 07 21:41:57 the x-loader is my tree on gitorious and linaro is basically from git.linaro.com Oct 07 21:42:04 ndec: yup Oct 07 21:42:23 I just had to build it for omap4430sdp Oct 07 21:42:33 rsalveti: well this is a surprise... so sometimes things work out of the box ... Oct 07 21:42:40 will fill a bug for this on launchpad soon Oct 07 21:43:01 ndec: it's basically because it's quite simliar with upstream Oct 07 21:43:08 so it shouldn't be an issue Oct 07 21:43:14 i am sure ogra which tablet would work out of the box too ;-) Oct 07 21:43:15 but it'd be good if we could check the padmux setting Oct 07 21:43:29 because I know sakoman did change the padmux for panda and not for blaze Oct 07 21:43:36 don't where he got the values for blaze Oct 07 21:43:57 ndec: probably, would be good to test Oct 07 21:44:11 ndec, well, i cant even build the kernel ... the display driver seems to dep on some headers i miss Oct 07 21:44:16 rsalveti: you might know this already, but our TI uboot based on upstream too is here: http://dev.omapzoom.org/?p=bootloader/u-boot.git;a=shortlog;h=refs/heads/omap_upstream Oct 07 21:44:30 ogra: hehe I had told you ;-) Oct 07 21:44:52 it builds if i disable all omap2 dss drivers thogh Oct 07 21:45:05 not sure how useful that still is :) Oct 07 21:45:59 ndec: I don't recommend using that omapzoom u-boot branch Oct 07 21:46:13 it is already obsolete Oct 07 21:46:21 yeah Oct 07 21:46:21 ogra: it's useful for blaze. even though it's called omap2, it means everything after omap2. Oct 07 21:46:31 sakoman: did you also review the sdp pad mux? Oct 07 21:46:39 sakoman: euh? isn't that the tree TI uses for its release? Oct 07 21:46:46 ndec, yeah, i guessed so much Oct 07 21:47:13 ndec, perhaps it is, but that doesn't mean it is the latest and best ;-) Oct 07 21:47:30 sakoman: in fact it probably means, it's not ! Oct 07 21:47:36 ndec: sebjan said TI tested blaze with upstream u-boot Oct 07 21:47:39 rsalveti: I'll take a look at the sdp pinmux and see if anything needs to be changed Oct 07 21:47:49 sakoman: cool, thanks a LOT :-) Oct 07 21:47:57 need to buy you some beers Oct 07 21:48:08 rsalveti: last time I checked it was ok, but perhaps new issues have been found Oct 07 21:48:22 because then we can just use upstream (and linaro) for all omap4 boards Oct 07 21:48:25 even omap3 Oct 07 21:48:33 sakoman: the problem is that TI developers will continue to push into dev.omapzoom.org... Oct 07 21:48:38 rsalveti: I am working hard to make that possible Oct 07 21:48:50 yeah, I know :-) Oct 07 21:49:22 sakoman: you looking at tablet too? Oct 07 21:49:24 rsalveti: I will have relocation patches for mainline panda and sdp in a day or two Oct 07 21:49:42 ndec: I only work on hardware that I have :-) Oct 07 21:49:44 sakoman: cool Oct 07 21:50:06 rsalveti: panda works fine, but sdp still has issues Oct 07 21:50:10 sakoman: that's safer Oct 07 21:50:30 sakoman: what issues are you currently having at sdp? Oct 07 21:51:08 the version with relocation takes a fault after printing the ram size Oct 07 21:51:17 oh, ok Oct 07 21:51:21 in the middle of tracking down the problem Oct 07 21:51:26 sakoman: but do you know any issue with current tree? Oct 07 21:51:44 no issues with the released v2010.09 as far as I know Oct 07 21:51:58 only with top of tree post the forced change to relocation Oct 07 21:52:09 awesome, maybe still some wrong pad mux setting, but here it seems to be running fine Oct 07 21:52:19 but then pretty much all arm boards other than beagle and overo are broken :-) Oct 07 21:52:19 but didn't test bt and wlan Oct 07 21:52:21 lack of drivers Oct 07 21:52:30 :-) Oct 07 21:52:45 and soon panda and sdp4430 Oct 07 21:53:04 the current version of the patch is in my omap4-exp branch Oct 07 21:54:07 about a dozen or so of those patches should hit mainline in the next week Oct 07 21:54:54 nice, too bad I don't have a blaze to test :-( Oct 07 21:55:06 I just got one here because I'm currently at TI Oct 07 21:55:28 my blaze has ES1.0 processor :-( Oct 07 21:55:47 and my panda an already obsolete ES2.0 Oct 07 21:56:10 rsalveti: you can get a blaze: http://svtronics.com/market_omap ;-) Oct 07 21:56:53 sakoman: ES2.0 8L should still work with the release image. Oct 07 21:56:59 ndec: hahaha :-) Oct 07 21:57:06 wayyy too expensive Oct 07 21:57:41 GrueMaster: that doesn't help me doing x-load and u-boot patches for ES2.1 changes :-) Oct 07 21:58:15 No, but you can build them on your 2.0. :P Oct 07 22:36:26 lool, do you know why i cant find any vfp stuff in the -dumpspecs gcc comand ? Oct 07 22:36:38 i though we enabled it by default ages ago Oct 07 22:39:11 ogra: sometimes when installing a package, it triggers a 'system restart required' event. how can i make a package that does this? Oct 07 22:39:57 you dump something into a dir (i have to check the exact path) Oct 07 22:41:16 ndec, touch /var/run/reboot-required i think Oct 07 22:45:06 ogra; well it worked on my laptop... so I just need to do this in postinst script. Oct 07 22:48:57 ndec, the patched native built kernel is here http://people.canonical.com/~roc/kernel/pass_init/linux-image-2.6.35-903-omap4_2.6.35-903.14_armel.deb Oct 07 22:49:18 ndec, http://kernel.ubuntu.com/git?p=roc/ubuntu-maverick.git;a=shortlog;h=refs/heads/ti-omap4 Oct 07 22:49:29 cooloney: thx. how do you build native kernel? qemu or on real hw? Oct 07 22:49:38 ndec, also i pushed the patch in my branch. Oct 07 22:49:42 cooloney: i saw your SRU email too Oct 07 22:50:00 ndec, we setup schroot environment in our built server Oct 07 22:50:24 so i schroot to maverick-armel to do native building Oct 07 22:50:38 but on hw, not qemu? Oct 07 22:55:53 oh, it is on a x86 server's schroot environment. Oct 07 22:56:13 ndec, it looks like it doesn't use qemu to me Oct 07 23:00:13 ogra: arm-linux-gnueabi-gcc -v 2>&1|grep vfp Oct 07 23:05:05 * ogra hugs lool ... awesome, thanks ! Oct 07 23:05:52 ogra: how is it going for you guys? is the latest hardware pleasing? :-) Oct 07 23:08:18 its wonderfully fast Oct 07 23:08:25 we have 400MHz RAM now Oct 07 23:09:16 still enough small issues with 1080p video processing and sound though Oct 07 23:09:32 but release looks really good for panda, blaze and oma4 tablet :) Oct 07 23:09:35 *omap4 Oct 07 23:11:25 rsalveti: is the ability to saveenv to sd cards something of interest (on panda and beagle xM)? Oct 07 23:12:54 (would require a raw partition to be added to the sd card) Oct 07 23:13:01 sakoman: probably, but from the distro pov Oct 07 23:13:05 *not Oct 07 23:13:09 sakoman, could it just create boot.scr right away ? Oct 07 23:13:34 ogra: no, file systems in u-boot are read only Oct 07 23:13:58 ah, sad Oct 07 23:14:03 indeed Oct 07 23:14:07 that would be the most elegant solution Oct 07 23:14:25 read boot-scr as input and write it again on saveenv Oct 07 23:15:01 sakoman, so we modify u-boot to support this feature, heh -:) Oct 07 23:15:12 on the other hand, with a bit of work one might be abble to eliminate the FAT partition and put MLO and u-boot in the raw partition Oct 07 23:16:27 * ogra is not a big fan of raw partitions from an image creator POV Oct 07 23:16:29 cooloney: I suspect that would be difficult to get upstream Oct 07 23:17:17 painful to create in a sane way (you need to wrap a non data partition around it) ... Oct 07 23:18:08 ogra: understood, it would require 3 carefully crafted dd commands Oct 07 23:18:36 and a lot of math for parted (which we use on the image builders) Oct 07 23:18:45 sakoman, it looks like no other choice to me, since we need to update boot.scr which is in SD partition. Oct 07 23:19:11 OK, just wanted to ask Oct 07 23:19:22 cooloney, well, we have ubootenvtools which can read/write the environment from/to NAND already Oct 07 23:19:31 ogra, or is that possible for us to store boot.scr on the board instead of SD Oct 07 23:19:34 *uboot-envtools Oct 07 23:19:35