[00:42] rsalveti, Still about? [00:43] rsalveti, I'd like to catch you sometime to talk about TFTP. [00:43] (could be tomorrow if I'm too late today) === tgall_foo is now known as Dr_Who [04:12] persia: sure, paste your questions :-) [04:33] rsalveti, So we have a spec: https://blueprints.edge.launchpad.net/ubuntu/+spec/other-o-arm-uboot-tftp It doesn't seem to have been fleshed out post-UDS, although I remember you saying that some of the work was landing. [04:33] persia: yeah, I have another bp that will cover this one [04:33] Does this spec need to be filled out with stuff to do? Is it already done? [04:33] Ah, cool. [04:33] What's the other blueprint? [04:34] persia: https://blueprints.launchpad.net/ubuntu/+spec/linaro-platforms-o-uboot-new-features [04:34] but it is still missing some bits [04:34] I plan to finish this up until end of this week [04:35] but the work is being done already, and jcrigby will push a new version this week with tftp support for panda [04:35] That covers the TFTP case. Unless you object, I'll mark my spec obsolete and reference your more complete one. [04:35] persia: I was going to do that once I put more details on the other one [04:35] just need to describe the work that was done already [04:36] give me a few days and I should do it [04:36] Ah, OK. In that case I won't do anything at all :) [04:36] yup :-) [04:38] Looking at the wider spec: you don't happen to know if u-boot can provide USB storage gadget, do you? [04:45] good evening, gentlemen [05:06] Matt_O: hey! :-) [05:06] Matt_O: want to help us testing the kernel? [05:06] rsalveti, is it low risk? :) [05:06] Matt_O: should be :-) [05:06] I really need my BB operational so I can to dev work on it (and it is right now) [05:06] but I'd like to help testing also [05:06] Matt_O: do you have ubuntu running on your xM C already? [05:07] yep [05:07] using your kernel in fact [05:07] Matt_O: so it should behave the same :-) [05:07] cat /proc/version [05:07] Linux version 2.6.38-9-omap (root@natty-panda) (gcc version 4.5.2 (Ubuntu/Linaro 4.5.2-8ubuntu4) ) #43xmbc Tue Apr 26 02:54:23 BRT 2011 [05:07] cool [05:07] let me check if the kernel was built already [05:09] Matt_O: yup, just need to install it :-) [05:09] ok [05:09] the installer will make backups right? [05:09] Matt_O: yes [05:09] ok [05:09] so what do I do? [05:10] Matt_O: but please mount the first sd card partition and copy the files as a local backup [05:10] good idea [05:10] doing that now [05:10] Matt_O: add the following lines at your /etc/apt/sources.list: [05:10] deb http://ports.ubuntu.com/ubuntu-ports natty-proposed main restricted universe multiverse [05:10] deb-src http://archive.ubuntu.com/ubuntu natty-proposed main restricted universe multiverse [05:11] Matt_O: apt-get update and apt-get install linux-image-omap [05:12] Matt_O: it should get the version 2.6.38-10.44 [05:12] ok.. working [05:12] this is the one with the fix included [05:13] getting [05:14] Matt_O: then flash-kernel should run automatically, and that will create a new uImage and new uInitrd [05:14] and put your older ones as uI*.bak [05:16] dkms failed for powervr-omap3, installing headers as recommended [05:16] Matt_O: sure, also install linux-headers-omap [05:17] that should make it work [05:18] it's retrying dkms [05:18] dang this thing is slow :) [05:18] (probably my sd card) [05:20] yeah [05:20] ok well I guess I'm ready to reboot and see what happens [05:20] Matt_O: cool, good luck :-) [05:21] phew... I got to the login prompt [05:21] do you want to see the dmesg dump? [05:22] Matt_O: yes, please :-) [05:23] http://home.rulecity.com/browse/dmesg.txt [05:23] Matt_O: still using the older kernel, guess flash-kernel wasn't pulled automatically [05:24] just don't know why [05:24] Matt_O: please call sudo flash-kernel [05:24] and then reboot again :-) [05:24] you sure? the old dmesg didn't detect it as a rev C, it detected it as an unknown [05:24] [ 0.000000] Linux version 2.6.38-9-omap (root@natty-panda) (gcc version 4.5.2 (Ubuntu/Linaro 4.5.2-8ubuntu4) ) #43xmbc Tue Apr 26 02:54:23 BRT 2011 (Ubuntu 2.6.38-9.43xmbc-omap 2.6.38.4) [05:24] Matt_O: from the kernel id [05:24] ok.. trying again [05:25] that's interesting... [05:25] same kernel and now you got Rev C [05:26] ok hit refresh [05:29] looks like it's using the new one now [05:29] Matt_O: cool, can you paste me your dmesg? [05:29] it's the same URL [05:29] just refresh it [05:30] Matt_O: cool, so it worked as expected :-) [05:30] yes [05:30] Matt_O: awesome, can you update the bug report with your results? [05:30] I am breathing large sighs of relief :) [05:30] haha :-) [05:32] update the bug report.. can I just reply to the email thread with a URL? [05:32] Matt_O: sure, I can then post the needed bits [05:33] ok done [05:34] Matt_O: awesome, thanks for helping making ubuntu better! [05:34] I am more than happy to do that, since I love ubuntu :) [05:34] (and it is in my best interest to have a solid BB linux install hehe) [05:34] hehe :-) [05:35] I've been spending the last 2-3 weeks ripping SDL out of my project because I suspected it was interfering with GLES2 on the BB... I'm finally ready to start up GLES2 dev work again with no SDL this time [05:36] in fact, I'm pretty sure it was interfering since SDL likely does glx calls and the powervr stuff uses some weird EGL crap [05:37] oh, got it [05:37] is it working well with the drivers provided for omap 3? [05:37] I wrote a "hello triangle" program that works great [05:38] I did have to manually install the GLES2 headers since I didn't see a package for those [05:38] but.. I was just thrilled to have it work at all :) [05:38] cool [05:38] maybe I will contribute a package to install those headers some day :) [05:39] that sounded bad.. what I mean is.. I'd love to contribuet a package, I would just ned to learn how to do it first [05:40] it should be easy to provide the package for the headers, we just didn't do by default because the broken soname for the libs [05:41] so in the end you have .so and not proper .so.1.7 or something [05:41] ah [05:41] proprietary drivers.. :) [05:41] links doesn't solve the issue because the soname is inside the lib [05:41] yeah [05:42] so if you build against the sgx drivers for omap 3 and switch back to mesa, you'll need the mesa-dev packages just to run your application [05:42] but I believe we should end up and provide the headers [05:42] lot of people are complaining about it [05:42] I don't see why anyone would want to use mesa on the BB... unless they just want to get something compiling fast [05:43] yeah [05:47] have you seen the issue where the s-video out goes blank after a certain period (like 10 minutes) ? [05:48] feels like a screensaver is kicking in but I can still access X via VNC just fine [05:50] could be just the power saving at X side [05:50] even without screensaver [05:50] ahhh interesting === chrisccoulson_ is now known as chrisccoulson === transitlogger is now known as apachelogger === dmart_ is now known as dmart === dmart is now known as Guest72963 === Guest72963 is now known as dmart === dmart is now known as Guest25714 [09:44] guys, can anyone try to compile maverick/ti-omap4? [09:44] i've a problem and it looks exactly like this one [09:44] https://bugs.launchpad.net/ubuntu/+source/linux/+bug/669912 [09:44] Ubuntu bug 669912 in linux "natty gcc-4.5 kernel (2.6.37-rc1) build failure" [Undecided,Fix released] [09:47] maverick didnt have 4.5 [09:48] uhm [09:48] that's why we didn't see this problem before [09:48] oh, wait, it did [09:50] https://launchpad.net/ubuntu/+source/gcc-4.5 [09:50] though i dont remember if it was default [09:51] ogra_: maverick had 4.5 [09:51] as non default [09:51] ah, k [09:51] hrw: not as default i think [09:51] 4.4 was default in maverick [09:51] ok [09:51] 4.5 in natty, 4.6 in oneiric [09:51] k [09:51] so use 4.4 then :) [09:52] i push that fix [09:52] today i've another load of CVE [09:52] then mvl-dove rebase [09:52] and then the 4.6 omap4 usb [09:53] *omap3 [09:53] ah, and i got the ac 100 but didn't have time to play with it :) [09:56] ppisati, https://launchpad.net/~ac100 ... there is also a mailing list [09:56] ogra_: cool [10:15] ppisati, oh, there is also #ac100 in case you are intrested in yet another IRC channel :) [10:26] how many of you use panda with ubuntu kernel? [10:26] tee: ../regina-normal_4.6-4_armel.build: Read-only file system [10:35] all of us ? [10:35] or most at least ... [10:37] I hope that it will not give me /home as RO :) [11:24] hrw, well, if it does, ask janimo, he is in RO too :) [11:24] ok [11:24] * ogra_ thinks hrw didnt get the pun :) [11:24] got, on second read [11:25] :) === tgall_foo is now known as Dr_Who === tgall_foo is now known as Dr_Who === tgall_foo is now known as Dr_Who [15:05] and we're just starting the Linaro arm porting jam! get at #linaro, get the bug list and help us making ubuntu better on arm :-) [15:29] rsalveti: do you want to take a bug on determining the reset button root cause? [15:46] gar, who dropped the cross compile link from the topic ? [16:39] hi, I'm looking for a good ubuntu arm target that's affordable [16:39] any suggestions? [16:40] beagleboard-xm or pandaboard [16:40] thanks [16:40] depends on your definition of affordable [16:40] ~$150 :) [16:40] what suihkulokki said then [16:40] hadn't heard of pandaboard [16:41] oh wow, dual core, nice [16:50] or if you want a more complete project, i think the toshiba ac100 is your only choice [16:50] more complete in hardware-wise [16:50] not software [16:50] If you can find them. They were discontinued. [16:51] and not sold in the us [16:58] NCommander: yes, as the bug seems to be introduced by your patch [16:59] I'm looking for a board with ethernet and serial port mostly [16:59] both should work for that [16:59] ac100 is a netbook (no ethernet) [17:00] rsalveti: well, in theory, I was correctng a bug in FAT filesystems. I'm still kinda scared that having a proper FAT breaks the BootROM :-( [17:01] ac100 looks like a nice little machine, not so good for development.. [17:03] NCommander: that doesn't surprise me [17:04] ugh [17:05] rsalveti: why not? [17:05] What??? A bug in FAT? Say it isn't so. [17:05] because its TI :) [17:05] NCommander: because there was a similar issue with omap 3 in the past [17:06] yeah hehe [17:06] rom code is not bug free [17:06] is like any software [17:06] full of bugs ;-) [17:06] Just the opposite. You get free bugs. :P [17:06] * ogra_ bets rom code is just adapted to x-loader every new HW release [17:06] instead of the other way round [17:06] * GrueMaster bets rom code doesn't change much except for sysboot tables. [17:08] it probably has to stay bug for bug compatible [17:48] persia, rsalveti: What /is/ the difference between rootstock and qmeu-debootstrap? They produce the same rootfs output, don't they? [17:51] rootstock sets some other stuff too [17:51] like create a user, set up oem-config and such [17:51] What happens if I try to boot the rootfs output from debootstrap? [17:53] should boot, but without any user [17:53] could also be lacking some other settings [17:54] How is this done in production installation? The install runs debootstrap and then sets up the system, like rootstock does? [17:55] Worst case, you heard some tearing/crackling noises and have some giant marshmallow man tapping your shoulder. [17:55] sveinse: yes, check livecd-rootfs [17:55] sveinse: but ubuntu is switching for live-build [18:02] It seems we need to find an armel board to run debootstrap native for production [18:03] Any suggestion on an armel board (which is binary compatible (ARMv7a) with OMAP3) which has good storage options, network, and preferably a box in order for it to fit in the server room? [18:04] Needs to be omap3? [18:04] You can get a beagleXM and a case. check beagleboard.org. [18:04] Hmm, no, guess not. Just needs to be able to run ubuntu armel [18:06] What storage options are there? External USB-SATA bridge? [18:07] What are the Ubuntu farm consisting of? -- mostly in respect of storage [18:08] The new panda farm will be using usb<>sata with 160G drives. [18:09] We will use a small (~30G) partition for the chroot environment. faster read/write times. [18:09] The existing systems are mostly babbage3 boards but they are flakey and are being replaced by the panda farm. [18:09] There are also some beaglexm systems. [18:10] The pandaboard should be better for building than beaglexm, shouldn't it? [18:10] dual core, but other than that, not much better. [18:10] I mean, the panda runs ubuntu just as well, doesn't it? [18:10] Yes. Same image just different kernel. [18:11] So, the IO performance is just about the same? [18:11] yes. [18:11] I think someone mentioned 24M/s throughput on usb. Haven't personally benchmarked it yet. [18:11] I guess it will be, since USB2.0 is the slowest link in a USB-SATA setting [18:12] I could be wrong. [18:12] But much faster than SD. [18:12] Well, the panda has twice the memory [18:13] I am also working on a boot method for Panda to pull from the gadget port from a host pc. No SD needed. [18:13] neat [18:14] how does one request rebuilds of packages that failed last time due to build-dep issues [18:14] GrueMaster: gadget port is the linux kernel term for USB OTG, right? [18:15] yes. The mini port on the panda. Should also work on the beagle/beagle XM. [18:15] Use the HW bootloader to load firmware to the panda? [18:19] yes. By default the sysboot is hardwired to 1. usb 2. mmc0 [18:20] For usb to work, it waits for a host to send it a second stage bootloader. Then the second stage bootloader can boot a binary blob from there (u-boot, kernel+initrd, etc). [18:21] We planned on using usb that way for product recovery, but since we need to support the host-side drivers. We just swap sdcards instead [18:21] For details on our buildd farm, check out http://dmtechtalk.wordpress.com/. For the usbboot loader, check out https://github.com/swetland/omap4boot. [18:21] thanks [18:22] For my test pool, I plan on pushing a kernel blob that will flash the sd card & reboot. Then it can boot from sd and I can do testing from there. [18:23] GrueMaster: Is it your project to build the physical farm? [18:23] No. That was my manager's project. He just sent the first one. [18:23] nice project! [18:24] I am writing the software to control the power relays via serial port & download the images through the otg port. [18:24] I'd have a easier job selling native buildfarm to the IT dept if such were available in 19" rack... :D [18:25] The case is a 4u. [18:25] yeah, so I see [18:25] Problem he ran into was too many wires. Probably could have had a custom wiring harness for the usb cables, but oh well. [18:26] Today we are basing it on cross compile on a intel PC. And everyone here in this channel is twitching when I mention that! [18:26] Yea. Too much can fail. [18:26] Wiring (electrical, building installation, network racks) are a craft that must be learned [18:26] Personally, I don't care how it is built. I just complain loudly when it breaks. :P [18:27] (software I mean) [18:27] Yea, the rack stuff can get hard. I designed a rack cabinet for high density cpu validation using desktop boards when I was at Intel. [18:29] Held 16 boards on 8 shelves. Each shelf had 2 aluminum plates with holes for mounting every MB standard available at the time. ITX, ATX, BTX, WTX, Server... [18:30] Even had remote temperature monitoring so systems could be shut down. Helped during Prescott (P4 x86_64) testing. That processor was hot. [18:48] GrueMaster: You're EE? [18:48] Pfft. no. [18:49] Used to be a combat engineer in the army though. [18:54] Anyone happens to know how long time it takes to build qt4-x11 on the ubuntu build farm? (beagleboards?) [18:56] Easiest way to find out is to lookup on launchpad.net. search for the package and go from there. here's the natty updated version: https://launchpad.net/ubuntu/+source/qt4-x11/4:4.7.2-0ubuntu6.1/+build/2458577 [18:56] The buildd names that ends with ceae are all beaglexm. [18:57] I'm trying to benchmark (crudely) the performance of running a native buildfarm vs. using cross compilation [18:57] took 1 day, 15 hours, 13 minutes, 48.2 seconds [18:57] I was going to say 2 days [18:57] iirc, it uses a lot of swap space. [18:58] Here's the link to the build farm. https://launchpad.net/builders/ [18:59] hey guys === MrBIOS-seamicro is now known as MrBIOS [18:59] In the next couple of weeks we should have almost a dozen pandas added. [18:59] whoa... 39 hrs.. That's a bit [18:59] More later in the cycle. [18:59] Yea, not pretty. [19:00] swapping, means that memory is congested. the pandas ought to remedy that -- at least it helps [19:00] I'm doing some testing of Apache performance on a 1.2GHz n ARMv11 board and am seeing an off request/sec wall of 260rps, even serving static content over 40kB in size [19:00] MrBIOS: What do you see on an armv7a system? [19:01] (we don't support older models - sorry). [19:02] sorry, not armv11 [19:02] it's armv7 [19:03] What platform? [19:03] something that's not released :) [19:03] Ah. [19:03] Heh. [19:03] yet [19:04] Cortex A8? A9? [19:04] with that something have sata? ;) [19:04] I read arm7 ;) [19:04] sveinse, maybe ;) [19:05] its cool that there is so much activity on arm going on! [19:05] keep it up! [19:08] * ogra_ bets Daviey or someone else from the server team might be intrested in MrBIOS' test results :) [19:17] ogra_ do you know when Daviey is generally active/around? [19:17] european business hours ... [19:17] he is on uk time afaik [19:18] okay, so I'm eightish hours off from him [19:18] but i just threw in his name as one member of the tem [19:18] *team [19:18] well, is there a list anywhere? ;-) [19:18] they usually hang around in #ubuntu-server [19:19] and arm server is one of our goals [19:19] for 11.10 [19:19] * ogra_ is afk now [19:59] hmm, digikey has no delivery date available for panda boards [20:01] Heh. Not unusual. [20:02] where are you getting your boards? [20:06] Some I got from TI (prerelease), some from digikey. I have one coming from the other source on pandaboard.org. [20:06] Not sure where the boards for the server farm came from. May have been digikey. [20:07] I'm in canada so digikey is good, guess I'll wait [20:15] Canada, eh? Where abouts? [20:17] jburkholder42: if you order from digikey right now, shipping time should be less than two weeks [20:18] jburkholder42: don't wait for them to be in stock [20:18] GrueMaster: Bancroft [20:18] prpplague: ok thanks [20:18] GrueMaster: middle of nowhere ontario [20:19] Ah. But at least you have bandwidth. [20:19] jburkholder42: we generally ship 300 to 500 units per week to digikey, however we had a short supply problem about 3 weeks ago that go use behind [20:19] jburkholder42: so if you order now, you most likely will get it in 10 to 14 days [20:19] GrueMaster: indeed, pretty good high speed up here [20:19] prpplague: cool, thanks for the info [20:19] prpplague: Any news on bamboo? Or at least a pandaboard case? [20:20] GrueMaster: still trying to work out a few details with the chinese company doing the tool [20:20] GrueMaster: pcb is done [20:20] GrueMaster: case design is done [20:20] cool [20:20] GrueMaster: just getting the actual abs parts is the issue now [20:20] GrueMaster: that an my poor bank account crying at the expense [20:20] ouch. [20:21] * prpplague hopes he will turn a profit [20:21] GrueMaster: i can send you a FDM for you to use for now if you want [20:22] Sure. [20:22] GrueMaster: it's no where near as strong as the abs case, but you can use it for now [20:23] GrueMaster: i had hoped that we could find a company in the US to do the tool but the US companies were outrageously expensive [20:24] Will it be clear acrylic? [20:26] GrueMaster: you mean the production case? [20:26] yes [20:27] GrueMaster: no, it will be black matte abs plastic [20:27] Hmm. Still good for making a settop box. [20:27] GrueMaster: you've seen the latest renedering right? [20:28] Not sure. Last one I saw was a few months ago. [20:28] GrueMaster: you have acroread installed on your system? [20:28] Of course. [20:28] GrueMaster: /msg me you email address === 31NAAJ7K3 is now known as sunnydrake [23:53] anybody here know if kexec is known to work on any ARM variant? [23:57] MrBIOS: I think it was being worked on. Not sure if ppisati worked on it. He is in Itally, check with him in the morning.