=== ericm-Zzz is now known as ericm === Martyn^ is now known as Martyn [09:41] anyone with a beagle xm rev.c willing to do a test, please contact me [10:10] xranby, thanks for taking over [10:10] * ogra_ hugs xranby [10:10] youre welcome === ericm is now known as ericm-otp === ericm-otp is now known as ericm === ericm is now known as ericm|ubuntu === plars-holiday is now known as plars [12:45] ogra_: what do you think of bug 872178 [12:45] Launchpad bug 872178 in jasper-initramfs "ROOTDEV export not working" [Undecided,New] https://launchpad.net/bugs/872178 [12:54] ndec, hmm, it is used in all other installs [12:54] do you avoid _growroot with some hack or so ? [12:57] ogra_: i asked XavB to join, as he is the one that knows this stuff.. [12:57] XavB: (2:54:37 PM) ogra_: do you avoid _growroot with some hack or so ? [12:57] XavB, it works on all other installs, so there must be something special in the way you use jasper [12:58] ogra_: not sure if other are using mmcblk1 [12:59] ogra_: as soon as you ar eusing mmcblk0, no issue, because a default value is set. [12:59] hmm, i thought we added a way to override that... [12:59] let me take a look at the code, one sec [12:59] if you add trace to jasper_setup, you will see that the export form ROOTDEV is not seen into jasper_setup [13:00] did you try setting root= on the cmdline ? [13:00] ogra_: in fact I reviewed the code and it seems OK, the only problem is the export from _growroot that is not visible into _setup [13:01] ogra_: yes I have root=/dev/mmcblk1p2 [13:01] well, export exports to the full environment, you should see it unless its actually unset [13:01] hmm, that should actually pick up the right bits then [13:01] The _growroot part is working fine, ROOTDEV is computed correctly, the resize of mmcblk1p2 is OK [13:02] but then the UUID is computed on mmcblk0p2 and after the reboot we are using FS on eMMC not SD [13:02] well, i dont see why the export wouldnt work, it is the same shell [13:03] ogra_: adding traces demonstrate that the export don't work, I promise... [13:04] infinity, any idea why an export from one initrd script to another wouldnt work ? even though they should be inside the same environment [13:08] ogra_, so mx5 install actually works if I do not tick automatic login [13:08] just takes very very long [13:08] the packages are removed, adn you get into lightdm [13:08] but only 15 minutes after the ubiquity GUI disappears [13:09] ogra_, re the ROOTDEV bug, I saw jasper setup script (?) set rootdev explicitly in case it is not set in growroot [13:10] which seemed one of the weird places in the code as growroot has exports [13:10] but they may be run from different shell sessions so the exports do not carry over [13:10] janimo: how do you install lightdm [13:10] brandini, it is on the image by default [13:11] how did you enable it? [13:11] start lightdm? [13:11] brandini: it's supposed to be enabled by default. [13:11] what's the problem? [13:11] janimo, thats just a fallback, growroot should actually have exported the right thing [13:11] I don't have any X running on my machine [13:12] ogra_, well the fact there is a fallback shows the export does not always work [13:12] I plug in my TV to the hdmi out and all I see is the login prompt [13:12] brandini: is it intentional? [13:12] so we should either see why it does not work, or just set it anew in setup [13:12] ndec: I'm running 11.10 server so I dunno [13:12] of course not for O :) [13:12] brandini: what do you call login prompt with no X? [13:12] ah... on the console [13:13] yup [13:13] janimo, i just added it to make sure we succeed in any case even if the user adds soemthing weirtd to root= [13:13] i didn't use the server image. is ubuntu-desktop package available? [13:14] * brandini looks [13:15] ndec: you're right :) [13:15] 565MB [13:15] well, better use the task than the metapackage [13:15] apt-get install ubuntu-desktp^ [13:15] *desktop [13:17] :) [13:17] (note the caret, its important) [13:17] too late? [13:18] ogra_: export exports to children, it can't export "up" to the parent. [13:20] argh ! [13:20] damned [13:20] i thought it exports on the same level [13:20] not sure what made me think that [13:50] ogra_: If it was exported by the parent, then children can change it. But children can't export to the parent. [13:50] right [13:50] so i need a tmpfile or some such [13:51] to put the value in and make the second script read it [13:51] ogra_: What are you mangling? [13:51] how ugly [13:51] Bug 872178 [13:51] Launchpad bug 872178 in jasper-initramfs "ROOTDEV export not working" [Undecided,New] https://launchpad.net/bugs/872178 [13:51] infinity, ^^^ [13:51] TI has probs with a blaze install [13:52] Don't we use more or less the same detection routine twice? [13:52] no [13:52] only once [13:52] Or is it just hardcoded in _setup? [13:52] it is hardcoded to a fallback of mmcblk0 [13:52] Ahh. [13:52] if the export doesnt get through [13:53] which doesnt harm us on any installs atm [13:53] since all our images use mmcblk0 ... the blaze cant [13:53] Well, you could do this two ways. You could find it in growroot and write a tempfile, or you could break the detection out into a utility that you install to /bin [13:53] well, i could also add it to initramfs-tools, the value i set in a child should persist for the other child then, no ? [13:54] though it feels very wrong to add jasper vars there [13:54] If you export ROOTDEV= in init, then it would persist. [13:54] But that seems silly. [13:55] no, i mean if i set ROOTDEV to a value in growroot, that should get handed over to _setup [13:55] if the var is owned by init [13:55] Yeah. [13:55] That's what I meant. [13:55] k [13:56] See the list of exports in init. [13:56] well, not the solution for now though [13:56] (and actually a bit to late for a solution in general i think) [13:56] ogra_: why exactly do we have to hardcode mmbblk0 in initramfs? [13:56] Yeah, I'm not sure this is something we should care about for release. [13:56] ndec, its a fallback value [13:57] ndec: We don't have to, and shouldn't, it's a bug IMO. But not an RC one, given that the images all work as designed on their respective targets. [13:57] sure. [13:57] it gets only used if the first script doesnt hand the info to the second (which sadly seems to be a default) [13:57] But yes, the ROOTDEV stuff is all broken/wrong, IMO. [13:57] what do we do in the non fallback solution? [13:57] hack jasper :( [15:41] xranby, was it you who filed that gconf bug where gconf looks in the wrong path ? [15:41] * ogra_ needs the bug # [16:11] I've got a usb to serial adapter I'm trying to use to get to an Arcturus module. For the life of me I can't get minicom to connect to my module. However if I run WinXP in a virtualbox I can get Putty to connect. Have any of seen this before? [16:12] Why use minicom? Use screen, much easier. screen /dev/ttyUSB0 115200 [16:15] Screen won't work either. The adapter is being recognized. Here is my dmesg output. [ 5880.414017] usb 5-1: generic converter now attached to ttyUSB0 [16:16] ok. What does screen output when launched? [16:17] Screen gives me nothing but a flashing cursor. I'm in the dialout group too. [16:18] Used the same cmd you just posted. [16:18] What speed settings do you use in Putty? Any other comm settings? [16:18] 115200 8N1 no flow control. [16:20] I've been all over the web, I'm really at a lost here. [16:21] Very interesting. I am using the same settings on 8 different USB serial cables here with no problem. [16:22] Try sudo screen /dev/ttyUSB0 115200. Also, try unplugging the usb cable from your host and re-plugging it in and checking dmesg to verify it is ttyUSB0. [16:25] Ahh, sudo gives me an error in screen "ioctl failed" [16:26] "Interrupted system call" [16:26] Is that any help? [16:26] Very interesting. [16:26] Anything in dmesg? [16:28] Negative. [16:29] Also I've tried this in 10.04, 11.04, and Backtrack 5. [16:30] This is very odd. My desktop is 10.04, and I have a dedicated serial console (serial-killer) running 11.04, neither have had these issues. [16:31] Make sure your usb device isn't being taken over by the VM. [16:32] Yea, It's really kicking my butt. VM has'nt been open since a reboot. [16:33] I modprobed the adapter myself after boot. [16:34] So dmesg does show the ttyUSB device? [16:35] Yup. ttyUSB0 [16:36] What does "sudo lsof ./dev/ttyUSB0" show? May be modem-manager getting in the way. [16:36] It's 660. Owned by root:dialout. Letme look. [16:37] nromally ubuntu users are in the dialout group by default [16:38] I don't think it is a permissions issue. I think some other app is taking control of the device in the background. [16:39] sudo lsof /dev/ttyUSB0 [16:39] lsof: WARNING: can't stat() fuse.gvfs-fuse-daemon file system /home/wstraus/.gvfs [16:39] Output information may be incomplete. [16:39] lsof should show who/what has the device open [16:40] Ignore the gvfs stuff. [16:41] How do you quit screen? [16:41] -a k [16:43] I had a bunch of screen's open, they were listed in lsof and I killed the pids. [16:44] But after killing all of those nothing gets listed from lsof [16:44] ok, that is a good thing. Now try launching screen again. [16:45] sudo? [16:45] Hi... sorry, dumb question... what are the user+root accounts and pwd for oneiric-core-armel ? [16:45] You may have to do something on your test system as well to get any output (like reset). [16:46] obarthelemy: There is no passwd. You should be able to use it as a chroot environment w/o problem. [16:46] obarthelemy, -core is for building your own images, its just enough OS to run apt, nothing more, no user or configuration [16:47] Just rebooted the module with screen running, got nothing. [16:48] I should see it booting then a login prompt. [16:49] let me try sudo screen. [16:50] Nothing using sudo either. [16:50] I know core is very bare, that' what i want... I'm stuck at the login prompt though, root/blank don't work, blank/blank neither, tried temppwd which was the correct pwd for the minimal-omap rootfs, no luck either [16:51] right, as i said, no users, no passwd, no config [16:52] you would have to create a rootpw before booting the rootfs [16:53] ok, ill dot hat then, thanks [16:53] Do you guys know a good adapter that works well with linux? I can go buy a new one. [16:57] i remember reading about the same problem somewhere [16:57] ThersiT: All of mine are TrendNet USB Serial. [16:58] Essentially anything with a pl2303 converter chip should work well. [16:58] What does dmesg say your USB cable has for a converter? [16:58] hmm, here it was, don't know if this is any help: http://www.trimslice.com/forum/viewtopic.php?f=52&t=335 [16:58] * ogra_ never checked, i'm just buying them as i lose them [16:58] and never had one that didnt work [17:02] GrueMaster: cool I'll go looking. dmesg says "usbserial_generic 5-1:1.0: generic converter detected" but it's a TIUSB3410 [17:02] gildean: I'm looking at that now. [17:15] Thanks guys. I'm just gonna try a new one. [17:16] ThersiT: Make sure you do a little research before you just buy one. You should be able to google the specs easily enough. [17:18] Yea, I'm gonna try to find one with that PL2303 chipset. [17:18] or has an ftdi chipset [17:18] both are reasonably stable [17:19] * GrueMaster has had excellent luck with the TrendNet brand, and a few others that have the pl2303 in them. [17:20] One of my cables is actually a 4 serial port hydra cable. 1 USB<>4 Serial heads. Other than enumerating, it works very well. [17:21] But I doubt a lot of people are using multiple usb-serial adapters on a single system to control 8+ arm systems. [17:22] I read in that forum gildean showed me to short pins 2 and 3 to make a loop back and that failed. I think this adapter in just not linux friendly. [17:24] Highly possible. I also found a bunch of issue reports when I googled for TIUSB3410 Linux. [17:24] To be fair, there "may" be a module that just isn't being loaded for some reason. [17:24] But I doubt that is the issue. [17:25] I did have to add a udev rule to load usbserial. [17:26] ThersiT: Check lsmod to see if ti-usb-3410* is loaded. [17:27] Tried lsusb | grep ti and got nothing relevant. [17:28] Then try "sudo modprobe ti_usb_3410_5052". [17:32] tried it, no change. what is the 5052? [17:33] lsmod says ti_usb is using ttyUSB0 tho. [17:36] Ok, different kernel, different module. [17:39] I'm on Ubuntu Desktop 10.04 2.6.32-34-generic [17:40] Interesting, I'm on the same kernel. [17:41] Well, -pae. [17:43] Yup, been messing with it for a few days. Installed that WinXP vbox last night and hated every minute of it. heh [17:51] On a side note I'm new to these modules. Is uClinux the way to go? [17:52] No idea. Never worked with that. [17:52] does uClinux still exist ? [17:53] i thought that died long ago [17:54] Oh, that would suck. This Arcturus module was bought about 2 years ago and uClinux is what came on it. [17:55] What do you guys use? [17:55] well, you are in #ubuntu-arm [17:55] so make a guess ;) [17:57] Ah heh heh, yup I should have thought of that. [19:50] How can I get a PPA compiled on the ARM farm? [19:50] ( Specifically : http://nginx.org/packages/ubuntu/dists/lucid/nginx/ ) [19:50] compiled in Oneieric [19:52] martyn: Not sure, but I believe heavy bribing with beer may be involved. :P [19:53] Seriously, I wouldn't know. [19:55] I can pull it locally and see if it builds on one of my pandas if that is what you are looking for. === Quintasan_ is now known as Quintasan [20:08] martyn: We already have nginx in the pool (universe). I'm sure it will get resync'd and rebuilt after next week. [20:13] GrueMaster: That would be helpful [20:13] GrueMaster: My buildd is down, having been "borrowed" for a corporate demo [20:13] cool [20:14] So it's migrated from PPA to universe? [20:20] Yes, but it is at 1.05-1. Building 1.08 for you now. Shouldn't take too long. [20:20] btw: Are you coming to UDS-Orlando? === chrisccoulson_ is now known as chrisccoulson === plars is now known as plars-afk