[00:11] NekoXP: I want to use pins for GPIO inside of ubuntu [00:11] it seems like I have to recompile x-loader and change some header or something, in order to do that [00:11] and I just have no idea where to start [00:11] not sure why you'd be required to change x-loader? [00:11] all I want is a couple gpio pins on the physical board I can set high and low [00:12] NekoXP: I think because it's defaulted that the pins are setup as GPMC/i2c and other things [00:12] and not GPIO [00:12] the pins are all 2 or 3-purpose [00:12] you should be able to do that in the kernel [00:12] it's the wrong place but you can do it [00:14] I've not done any omap4 stuff but there should be some kind of pinmux stuff around you can kick.. isn't there a pin header on the board anyway? [00:15] my panda's at the office... [00:18] NekoXP: yes the proper place is the kernel, however pcpower has no experience doing any dev work, he just needs to know where one gpio is already configured [00:19] NekoXP: gpio 139 and 140 are already configured [00:19] schematic :D [00:19] I disagree that pinmuxing should be done so late, but :_ [00:19] if you need to just hack around it's much quicker in the kernel [00:20] NekoXP: schematic doesn't tell you what the software is configured for [00:20] yeah but if you want to change it who cares what it's already configured for [00:21] you just need to find some pins which are on the board but not connected to some other device [00:21] hopefully with no pullups or pulldowns or weird stuff [00:22] NekoXP: thats the whol point, he doesn't want to compile anything, he just wants to know that is already configured [00:22] NekoXP: i already gave it to him anyway [00:25] so I answered him already :) /sys/class/gpio has a list of pins configured as gpio and cross referencing would have found which ones were free and where on the board they are [00:26] NekoXP: no that is no correct [00:27] really? is panda that much different to any other board using gpiolib? [00:28] NekoXP: if the gpio hasn't been exported, it will not show up in /sys/class/gpio [00:29] hmm yeah I guess you have to request it first [00:29] but if 139 and 140 are already configured that's implied [00:32] NekoXP: only the pinmux is configured, doesn't mean it has been requested [00:32] okay so he needs to kick around in the kernel anyway [00:32] ? [00:33] NekoXP: no, he just needs to export the gpio [00:33] GrueMaster: ping [00:33] GrueMaster: where the heck is the x-loader source for maverick omap4 support? [00:35] Couldn't tell you off hand. I'd have to load a system with maverick and then look at a package. [00:35] let me see if I can find it by other means [00:36] https://launchpad.net/ubuntu/maverick/+source/x-loader-omap4 [00:37] GrueMaster: nm, found it [00:37] I was looking for the git tree. [00:37] But that works. [00:39] GrueMaster: getting killed with people wanting to do i2c and gpio from 10.10 lately === asac_ is now known as asac === apachelogger is now known as dslogger [03:39] hello can anyone tell me why ubuntu-omap4-extras-multimedia package fails in ubuntu...I see that two packages have been updated and wont be installed because they no longer exist in the ppa [07:34] Good After noon friends [07:36] need some help to enable my beagleboard XM with touch screen [07:37] it's TSC 2046 Controller [07:38] and I have "insmod" ads7846.ko (2.6.38-8) [07:39] but evtest is not responding for touch events [08:03] any help pls [08:34] persia, u there? [12:00] ogra_, you there [12:00] [ 956.874450] ads7846: disagrees about version of symbol module_layout [12:00] means wrong driver module right ? [12:00] yes [12:01] how did you build it ? [12:01] am wondering how it's happen, i have build the OS image from rootstock [12:01] i mean how did you build that module [12:01] ogra_, it was there bydefault [12:02] in the rootstock build image [12:02] cant be :) [12:02] yes, it is [12:02] :) [12:02] there is no kernel in rootstock images by default [12:02] linux-omap [12:02] so that cant be [12:02] unless you forcefully added one [12:03] i hve added the parameter linux-omap [12:03] ?? [12:04] the message above means that your module was compiled differently from the rest of the kernel ... that simply cant happen in a packaged kernel [12:04] right.. [12:04] i did seed linux omap [12:05] so check what kernel packages you have installed [12:05] while building rootstock [12:05] ok [12:07] ogra_, i have build the rootstock with the parameter --seed linux-omap [12:08] did you check your installation ? [12:08] yes [12:08] it seems like everything is fine [12:09] running on kernel 3.6.38-8 omap [12:10] locate ads7846 [12:10] where is that located [12:10] /lib/modules/kernel/--- [12:10] where exactly [12:11] i dont see it being built in any of our arm kernels [12:11] ogra@panda:~$ locate ads7846 [12:11] ogra@panda:~$ [12:11] /lib/modules/2.6.38-8-omap/kernel/drivers/input/touchscreen [12:11] definitely not enabled on omap4 [12:11] very strange [12:12] oh, i see, on omap4 it is compiled in [12:12] oh.. [12:12] but mine is omap3 [12:12] beagleXm [12:12] right, i dont have one running here so i cant check [12:12] ok [12:13] file a kernel bug then against the linux package (mention omap3 in the description) [12:13] ok [12:13] though that still doesnt explain the error [12:13] and i think only the kernel or touch teams could help here [12:13] ok [12:14] or I have to build the kernel here right? [12:15] you could do that too, indeed, or try one of rcnee's kernels (there is a PPA somewhere with his kernel packages) [12:15] i have tried but failed to enable SGX driver [12:15] anyway will give a try again [12:15] make sure to file bugs if you find issues ;) [12:16] sure will do [14:30] nvflash looks fun [14:30] heh [14:31] nightmare kind of fun [14:31] what for do you need it ? [14:32] ogra_: Hannspree hannspad - cheap android 10.1" tablet [14:33] ah [14:33] well, if you have a good kernel you should only have to use nvflash once [14:33] i.e. a good kernel like the ac100 one wheer you can see the boot partitions from userspace :) [14:35] problem is that there are no good kernels for it. vendor 'forgot' to give gpl source so kernels from community only and they lack few things [14:35] did someone report it as gpl violation already ? [14:36] checking that is on my todolist [14:37] well, generally the chromeos kernels are a good start for tegra stuff [14:40] Hi, I can't figure out why https://launchpadlibrarian.net/76889732/buildlog_ubuntu-oneiric-armel.libvirt_0.9.2-4ubuntu8_FAILEDTOBUILD.txt.gz is failing; I don't *think* it's due to my change. Would somebody be able to have a look. [14:40] ? [14:45] me looks with no promises that he has more clue [14:50] wow, looks really bad [14:55] cjwatson: Is that a python testsuite? [14:55] cjwatson: If so, I imagine it relates to https://launchpadlibrarian.net/76841318/buildlog_ubuntu-oneiric-armel.python-apt_0.8.0ubuntu3_FAILEDTOBUILD.txt.gz [14:56] well, doesnt look like anything used libnl3 on arm before [15:01] infinity: not seeing the link ... [15:01] (but I don't actually know, I'm entirely new to libvirt and was just trying to get everything shifted over to libnl3) [15:01] cjwatson, wild guess: the missing classid file gets pulled by a netlink socket, our kernel on the ancient buildd HW misses bits the nl3 lib uses, so the dtd doesnt end up where it should be [15:02] (really wild guess) [15:03] Oh, I didn't look closely. I'm barely awake here. :P [15:03] Brain just saw two testsuites die. :P [15:10] i dont have a panda with a disk atm, but i can do a testbuild on an ac100 [15:11] to see if its machine related [15:18] Hello, i want to test the uart port on the beagleboard xm with a oszilloscope to check the uboot configuration, i have disabled kernel pinmux. i know that uart2 is ttyO1, are there any possibily to send data over this port ? [15:47] does anyone have any idea why i get inverted colors on my cursor and the xterm text, like this: http://dl.dropbox.com/u/952540/wtf.jpg [15:48] if i swap to a text console and back to X, then they show up correctly [15:49] also, there's no text showing up on any of the apps [15:49] some underscores here and there, but mainly nothing is visible [15:56] what arch is that ? [16:04] omap4, custom maverick rootfs, kernel 2.5.35-980-omap4 [16:05] i have no idea what I've messed up because an earlier rootfs i built didn't have that problem.. [16:12] well, i guess i'll just rebuild the thing for the millionth time. [16:20] before that it'd be nice if i could even once get the video decoding hw accelerated, but there's no info about that on the wiki. or anywhere. i mean, there's the installation instructions for omap4-extras and "use totem", but no information about what videosink should be used when using gst-launch etc. or if any other pads are required. [16:21] it should just work out of the box in maverick [16:21] oh [16:21] there are no codecs provided by TI for natty so dont bother to try it there (or bug TI) [16:21] we should get codecs for oneiric again [16:33] cjwatson, fyi, thats how the libvirt build fails for me in a oneiric chroot on an ac100 http://paste.ubuntu.com/662055/ [16:35] fails a lot earlier in test-chown === dslogger is now known as dickhouse === dickhouse is now known as dragondrop === dragondrop is now known as dslogger [20:44] ogra_, openmax codecs from TI?(re your chat with fisuk above) [22:06] FYI, testing Nathy on Pandaboard, I got I/O errors with a brand new SD card ( Lexar, Class 10, 32GB). I can resize the partition, and at reboot -> nothing. After some investigations, I isolated this issue : http://paste.lisp.org/display/123918 [22:08] all the other cards I tested work as expected. The second 32GB I tested, has other problems : I got lot of I/O issues at runtime (e.g. compiling a big application), and the fs is remounted in read only mode [22:09] after some little search, the relevant code in the kernel is located in drivers/mmc/card/block.c [22:09] at least the error messages (say the what, not the why) [22:09] anybody aware of 1) existing issues ? 2) existing bug reports around that ? [22:11] thanks in advance for any help [22:11] :-) [22:15] You might just have bad cards. I have no issues here with an LX 32G Class 10. [22:15] ericb2: you any further in finding the cause? [22:27] infinity: thanks for the info [22:28] infinity: could this be another issue type, like the resize does something wrong, or I don't know ? [22:28] infinity: the problem is, I got other issues with the same disk image [22:29] infinity: like file system suddenly in read only mode, every time the machine has a good load or something similar [22:29] ericb2: i tried with a manually resized image, that was good. its not resize thats causing the problem [22:30] MrCurious_: thanks [22:30] np [22:30] MrCurious_: you avoided me to try myself :) [22:31] * ericb2 scratching his head [22:34] MrCurious_: so we are two with problematic cards, both class 10 and high speed (at least this is written on the package) [22:35] other possibility could be a wrong high speed card, e.g. identified with high speed parameters and that's not true, causing timeout and worse [22:39] davidm: so it's now official that the scheduler is the culprit? [22:40] TheSeven, it's official it's a general ARM issue, not a Panda USB issue [22:40] likely scheduler but not completely sure about that [22:40] Was best rename I could come up with that fit what we are seeing [22:40] ope to suggestions for a better name [22:41] it would be interesting if this is only related to USB, I/O in general, or even every kind of thread execution [22:42] TheSeven: It's not just USB, and probably not just I/O. It really looks like it's spinlocking breakage. I'm assured that people smarter than me are looking deeper. [22:42] IT also makes a difference on writing to the SD card [22:43] my guess is it has hard coded timings that just work with slow cards, rather than detecting card state, allowing for it to run at max speed [22:43] or a hard coded wait that isnt needed with faster cards [22:43] infinity, I'm trying to get as many people as possible looking at it ;-P [22:43] if i had to "guess" [22:45] MrCurious_: I use a few class 10 cards (one even the same brand as ericb2) with no issues. [22:45] davidm: but compared with what's going on with USB storage, the differences on the SD interface were almost neglegible [22:45] s/with/to/ [22:45] rev c board? [22:46] TheSeven: Yes, but SD is slower by default, so sitting in idle loops 75% of the time doesn't hurt as much. :P [22:46] infinity: I believe you. Just searching how I can proof my reseler it is defective / not good [22:46] at least my board was rev c [22:46] infinity: this card is expensive (76 euros), and I'm not happy [22:47] ericb2: SD cards just aren't intended for compiling things on them [22:47] TheSeven: he he [22:47] TheSeven: I'm testing the armhf build on USB stick [22:48] TheSeven: on USB disk, it works. Trying the usb key now (more easy for traveling) [22:48] ericb2: usb is currently buggy, and possibly 10x slower than it should be [22:48] MrCurious_: promising ^^ [22:49] MrCurious_: I was wondering whether an sata support will arrive some day [22:49] would be fantastic [22:50] sata is reliable, really [22:50] Doubtful on panda. Maybe omap5? [22:51] GrueMaster: indeed, it will probably not occur with Panda . Crossing the fingers [22:51] for omap5 .. in one, two years ? [22:51] * ericb2 hides [22:51] Find me a mobile phone that needs SATA... [22:51] Other than adding it to the USB bus, I wouldn't get your hopes up on panda. [22:51] (Panda's essentially a phone dev platform) [22:52] infinity: I meant more a multimedia one ( double hdmi .. ) [22:52] Panda is the same hardware as the Droid 3. [22:52] + lot of decoders inside. Looks lke a mix of best technos, for hackers and like [22:53] Other than a few NAS systems out there, I don't know of any mainstream arm based products that also have sata. [22:53] There are tons. [22:53] Mostly set-top boxes, PVRs and the like. [22:53] Mainstream? [22:53] Well, depends on how you define "mainstream". [22:54] MrCurious_, ericb2: USB is still way faster than SD [22:54] maybe nfs, as I told previously [22:54] Most of those are PPC based afaik. [22:54] Most people get them bundled with cable service and such, but that's pretty "mainstram", you just don't go comparison shopping. [22:54] * GrueMaster owned a tivo and owns an HD DVR. [22:54] * ericb2 got a powerbook in the room. bought in october 2005 or so [22:55] works well [22:55] davidm: So it doesn't make sense to do further random testing and throwing results at the bug from my side? [22:55] * davidm still owns 2 working tivo series I and has 4 for spare parts [22:56] oops, it's late, and I need to sleep a bit [22:56] Any valid testing might help, the bug is not isolated as yet [22:56] good night all [22:57] Apparently the PVR in my house is MIPS. [22:57] OK I'm off, I'll look at the back scroll later [22:57] infinity: got a gdium too :) [22:58] * ericb2 gone for true [23:02] theseven: i use a ssd on usb, it makes ubuntu on panda tolerable :D [23:03] MrCurious_: Try booting with nosmp and checking your IO performance. [23:07] MrCurious_: a thumbdrive would do as well [23:16] yeah [23:16] gruemaster: is that a opt for boot.scr? [23:17] Yes. Add to the kernel cmdline. [23:17] is that going to kick the speed up? [23:17] but at a cost [23:17] IO speed should be 3-4x but no SMP. [23:18] thats a strange tradeoff [23:18] counterintuitive [23:18] See above about spinlocks. [23:22] doubt spinlocks are broken, more like not optimized