**** BEGIN LOGGING AT Mon Dec 20 02:59:58 2010 Dec 20 09:01:35 rsalveti, ok second freeze this night. no blinking LED, nothing on the serial console even after reconnecting, I added some info on https://bugs.launchpad.net/ubuntu/+source/linux-ti-omap4/+bug/690370 Dec 20 09:01:36 Launchpad bug 690370 in linux-ti-omap4 (Ubuntu) "Strange out of memory on pandaboard (affects: 1) (heat: 10)" [Undecided,New] Dec 20 09:33:04 rsalveti: Around? Dec 20 09:33:27 mkimage is moving from uboot-mkimage to u-boot; this might impact things like flash-kernel, cdimage and other code Dec 20 09:33:58 The u-boot import doing this happened on the 7th, but failed to build on most arches (except amd64) due to LDFLAGS breaking the u-boot host build Dec 20 09:34:13 I'm fixing that now, so it will break on all arches Dec 20 10:01:33 lool: so now each arch will have to install u-boot package just to have uboot-mkimage? Dec 20 10:05:28 hrw: Yes; uboot-mkimage used to be built from an old fork of the u-boot source tree, so it's an improvement, but I would have preferred if we had a separate package instead Dec 20 10:05:39 my head hurts Dec 20 10:06:25 lool: add uboot-mkimage binary package to u-boot source package Dec 20 10:07:18 hrw: This would be a delta with Debian Dec 20 10:07:35 hrw: I'm not sure how smooth upgrades will be, but I suspect that we need a dummy transitional package here Dec 20 10:07:55 So this should probably be raised to Debian Dec 20 10:08:03 albeit we could add the transitional package on our own Dec 20 10:08:13 lool: or make it and raise it with patch and transitional package? Dec 20 10:08:32 Yup Dec 20 10:08:36 hrw: Would you like to do this? Dec 20 10:10:16 would have first to check packages Dec 20 11:23:28 lool: looks like uboot-mkimage should be transitional indeed Dec 20 11:23:34 I just built both Dec 20 11:26:25 hrw: you built both? I don't understand what you built Dec 20 11:26:52 hrw: I filed a bug in Debian to request a transitional package BTW; I'm happy if we add it in Ubuntu first though Dec 20 11:28:52 u-boot and uboot-mkimage ones for amd64 Dec 20 11:31:08 Ok; one open question is how to name the package Dec 20 11:31:17 Sorry, ignore that Dec 20 11:31:21 I'm confused with another issue Dec 20 11:31:30 (which is that mkimage should be split out of u-boot) Dec 20 11:32:30 ok Dec 20 11:50:42 lool: hi Dec 20 11:51:20 lool: our image is broken for quite a while, since dec 14 because of that Dec 20 11:51:50 the package was removed from the archive, without fixing the u-boot one first Dec 20 11:51:52 rsalveti: I think I've uploaded almost all fixes except for jasper-initramfs where I've sent a merge request Dec 20 11:51:58 rsalveti: But I can't commit to jasper-initramfs Dec 20 11:51:59 lool: I discussed that at #ubuntu-release yesterday Dec 20 11:52:55 rsalveti: Ok; did the discussion raise any other required changes than the ones I prepared? Dec 20 11:52:56 lool: and then we need to move this u-boot to main Dec 20 11:53:13 as it's on universe currently Dec 20 11:53:33 lool: nops, my only problem was that our image is currently broken because of that Dec 20 11:53:38 mainly because of jasper Dec 20 11:53:53 rsalveti: asac/doko removed me from ~ubuntu-mir for some reason, so I can't approve this, but it would be a trivial MIR to bump u-boot to main Dec 20 11:54:09 rsalveti: I'm happy to upload the jasper-initramfs change if you vouch for it Dec 20 11:54:34 I just can't commit it to Bzr, so would either comment out the Vcs-Bzr or move it to the UDD branch Dec 20 11:57:43 lool: I believe ogra is the only one that can write to the bzr branch Dec 20 11:57:59 rsalveti: Yes; that's why I only sent a merge request Dec 20 11:58:05 and as we need it to get our image working, I believe it should be ok to just fix the package at ubuntu Dec 20 11:58:11 rsalveti: Ok; will merge then Dec 20 12:01:48 rsalveti: Uploaded, happy to hear how it works in Ubuntu images Dec 20 12:02:35 lool: ok, then we should ask someone to bump it to main Dec 20 12:04:19 rsalveti: Ping archive admins; tell them u-boot-linaro is already in main Dec 20 12:04:23 rsalveti: ah sorry it's not Dec 20 12:04:34 rsalveti: but tell them uboot-mkimage was in main and wa sa fork ;-) Dec 20 12:10:42 hrw: If you're preparing a transitional package for mkimage, I'd love sponsoring it today while I still remember about this! :) Dec 20 12:11:40 let me first find good example Dec 20 12:12:20 lool: patch for u-boot acceptable? Dec 20 12:12:25 hrw: Yup Dec 20 12:12:34 hrw: debdiff would be ideal Dec 20 12:18:45 rsalveti, I've left my pandaboard in freeze state in case you want me to do something, if not let me know I'll power cycle it Dec 20 12:18:59 guerby: sure, just checking the bug Dec 20 12:19:49 guerby: did you get anything at the uart? Dec 20 12:20:12 guerby: if not, then just reboot it, nothing interesting at the logs :-) Dec 20 12:20:15 rsalveti, nothing on serial (just unresponsive login prompt), I killed screen and relaunched, nothing Dec 20 12:20:16 ops :-( Dec 20 12:20:30 rsalveti, ok I'm power cycling Dec 20 12:21:09 hm, weird that not even a kernel trace Dec 20 12:21:54 rsalveti, may be there are bootargs to tell the kernel to be more verbose? Dec 20 12:22:24 guerby: what are your current boot args? Dec 20 12:23:32 it could be that you had 'quiet' on it Dec 20 12:23:50 rsalveti, text ro elevator=noop vram=32M mem=768M root=UUID=b5d2dfb1-270c-4966-abe6-dfe7a2a17efd fixrtc smsc95xx.macaddr=32:57:F8:93:E1:CD Dec 20 12:25:07 rsalveti, also I noticed the board LEDs were both off after the freeze Dec 20 12:25:09 but quiet just set the log level to warning, you should have seem at least some error Dec 20 12:25:23 guerby: yeah, meaning that the board was dead Dec 20 12:25:35 one led is the heartbeat Dec 20 12:25:44 the other is the mmc activity Dec 20 12:25:57 rsalveti, I had top and tail -f /var/log/messages running through ssh, plus screen on the serial and I saw nothing Dec 20 12:26:00 rsalveti, ok thx Dec 20 12:27:27 nothing is bad :-( Dec 20 12:27:49 lool: http://pastebin.com/g49n8Jde Dec 20 12:28:50 rsalveti, I'm relaunching my bootstrap script Dec 20 12:29:54 ok Dec 20 12:37:01 rsalveti, do you think that my sysctl -w vm.overcommit_memory=2 could have an impact? Dec 20 12:38:32 I also have echo 0 > /proc/sys/kernel/randomize_va_space Dec 20 12:38:58 guerby: but that shouldn't freeze the kernel that way Dec 20 12:39:03 it's probably another bug Dec 20 13:47:26 * rsalveti lunch Dec 20 14:47:10 can someone take a look at bug 688010? Dec 20 14:47:11 Launchpad bug 688010 in ubuntu-netbook-default-settings (Ubuntu) "Missing dependency on gconf2 (affects: 1) (heat: 6)" [Undecided,New] https://launchpad.net/bugs/688010 Dec 20 14:48:05 it is armel only and I provided patch which solves problem Dec 20 14:48:22 hi robclark Dec 20 14:48:52 howdy hrw Dec 20 14:54:32 robclark: is there a kernel for panda which finally gives mem=1G stable? Dec 20 14:55:21 hrw: unsure.. I know some folks are using L24.11 kernel with 1G (minus the hole left for syslink).. Dec 20 14:55:37 but I don't know if that means it's fixed, or just that they haven't had problems.. Dec 20 14:55:56 hi... is there a libgles2-sgx-omap4-dev package? Dec 20 14:56:15 (and L24.11 kernel doesn't work with the L24.9 versions of syslink, ducati, etc from 10.10...) Dec 20 14:56:36 or... where can i get the headers for sgx gles2? Dec 20 14:57:07 demarchi: I'm not a GL expert, but I think the idea is that you should be able to compile things with generic headers from khronos.. Dec 20 14:57:20 although I don't know if those are packaged in any way.. Dec 20 14:57:33 robclark: you mean... the mesa headers? Dec 20 14:58:03 I guess so.. but I've never tried.. Dec 20 14:58:21 the mesa package that contains the headers conflicts with sgx packages Dec 20 14:58:22 assuming mesa headers aren't weird in some way Dec 20 14:59:13 demarchi: http://www.khronos.org/registry/gles/ Dec 20 14:59:42 you can download manually the headers.. perhaps there is a better way, with some deb package, but I don't know Dec 20 15:00:02 maybe vstehle or rsalveti or someone like that has some idea Dec 20 15:07:41 robclark: demarchi: mesa headers should be fine Dec 20 15:08:56 demarchi: libgles1-mesa-dev libgles2-mesa-dev libegl1-mesa-dev Dec 20 15:09:24 at least for omap 3 packages we don't put the headers by default because of the soname mess around the sgx library Dec 20 15:09:51 as we don't want people complaining that built the package against it but only works when using the sgx libs Dec 20 16:12:35 rsalveti: What happened to u-boot-omap4 in maverick? Dec 20 16:12:53 rsalveti: did it get removed? did another u-boot source supersede it? u-boot-linaro maybe? Dec 20 16:13:11 lool: yeah, we're only using the u-boot-linaro Dec 20 16:13:52 lool: can you please file a MIR bug later? Dec 20 16:14:05 I want to get this issue with u-boot fixed so we can generate arm images again Dec 20 16:19:42 rsalveti: LP #692613 Dec 20 16:19:43 Launchpad bug 692613 in u-boot (Ubuntu) "[MIR] u-boot (affects: 1) (heat: 6)" [Undecided,New] https://launchpad.net/bugs/692613 Dec 20 16:19:47 rsalveti: I was filing it, which is why I was asking Dec 20 16:19:53 lool: cool, thanks Dec 20 18:00:55 hello, i want to install the gcc compiler on my beagleboard running with ubuntu 10.10. but i have problems with the "failure http://ports.ubuntu.com/ubuntu-ports/pool/main/g/gcc-defaults/gcc_4.4.4-lubuntu2_armel.deb 416 Request Range Not Satisfiable" i tried to install the gcc compiler with aptitude, apt-get and synapsys. But everytime i got the same message. anyone can help me? Dec 20 18:02:05 rOxx: Sounds like you have some proxy between you and ports.u.c Dec 20 18:03:10 yes i have configured a proxy Dec 20 18:03:43 rOxx: Can you use wget to download the .deb package? Dec 20 18:05:38 @gruemaster yes wget is available Dec 20 18:06:19 so, "wget http://ports.ubuntu.com/ubuntu-ports/pool/main/g/gcc-defaults/gcc_4.4.4-lubuntu2_armel.deb" will download the file without error? Very odd. Dec 20 18:07:05 @lool i have configured the proxy in the Preferences -> network proxy > there manuell proxy config -> entered the proxy data and apply system wide. is this correkt ? Dec 20 18:08:49 @gruemaster now i have the file on my beagleboard. i saved the file with the firefox download manager :-) Dec 20 18:10:13 is this method ok ? Dec 20 18:10:18 rOxx: After setting the system wide proxy, did you log out/log in? That is the only way to ensure the setting gets carried through to all apps that use system environment settings (i.e. bash shells). Dec 20 18:11:55 rOxx: I'm not sure it's enough/correct, nor whether it's because your proxy is being used or is not being used Dec 20 18:11:57 rOxx: using wget as I indicated is a way of testing the underlying environment that apt-get relies on. If wget fails, then we can go further. using firefox, while effective in retrieving the file, doesn't help test apt-get issues (other than basic network connectivity). Dec 20 18:12:07 rOxx: I can just tell you it's related to your proxy :-) Dec 20 18:12:39 rOxx: apt directly talking to ports.u.c works fine, and there are no mirrors, so it's necessarily either your setup being incomplete, or apt misbehaving with your proxy, or your proxy misbehaving Dec 20 18:20:21 @gruemaster i have downloaded the file via wget. it works fine Dec 20 18:21:37 after i changed the proxy settings, i didnt log out, but today i have restarted the system after changing the proxy settings. is this the same ? Dec 20 18:22:43 restarting is just as good as logging out/logging in. Dec 20 18:23:44 I'm running "apt-get install gcc" here on my beagleXM running 10.10, and have not seen a problem (although I do not have a proxy). Dec 20 18:24:58 ok gruemaster and lool, thanks for your help, i go to my neighbour and test it without a proxy connection Dec 20 19:10:56 is anyone familiar with whether or not acm-mode defines a specific use of newline characters? I'm playing with an embedded device that is using the g_serial gadget and by default it uses ACM. I'm not famliar enough with ACM to know whether that standard makes mention of newline treatment. I actually don't want any special newline treatment at all. Dec 20 19:11:11 cdc-acm Dec 20 21:24:34 15:45 < hrw> can someone take a look at bug 688010? Dec 20 21:24:35 Launchpad bug 688010 in ubuntu-netbook-default-settings (Ubuntu) "Missing dependency on gconf2 (affects: 1) (heat: 8)" [Undecided,New] https://launchpad.net/bugs/688010 Dec 20 21:25:03 has anyone tried using usbser.sys and found that it inserts spurious carriage returns (0x0d) when the serial device is actually just sending newline (0x0a) ? Dec 20 21:26:14 0x0d? not 0x0c? Dec 20 21:44:10 yes, 0x0d Dec 20 21:45:50 it seems like when I use my os x machine and configure it to use the gadget serial device as a null modem this problem doesn't exist Dec 20 21:46:00 a wha? Dec 20 21:46:48 it presents an option as Vendor: null modem on os x system preferences and opens up options for the baud rate which was what gave me a good feeling that it would work fine on osx without a custom driver Dec 20 21:47:27 so the embedded device is using the gadget framework and is loading the g_serial module (and using acm) Dec 20 21:48:23 and the windows box to which I'm trying to connected the embedded device using the usb OTG port is configured to use usbser.sys as stated in the linux kernel documentation Dec 20 21:48:33 to which I'm trying to connect* Dec 20 21:51:40 so when I go echo -ne "\n" > /dev/ttyGS0 on the arm box and receive on the os x usb host I just get 0x0a whereas on the windows host I get 0x0a 0x0d or maybe it's 0x0d 0x0a... either way it's putting something in there that shouldn't Dec 20 23:23:39 rsalveti, hmm freeze again after about 7 hours of compile, no message on serial. Let me know if you think of something useful I could do Dec 20 23:24:58 gn Dec 21 01:48:19 guerby: no new ideas for now :-( Dec 21 01:48:32 cooloney: maybe you can help us with this issue Dec 21 01:48:51 guerby: is building gcc at his pandaboard Dec 21 01:49:03 rsalveti: oh, i missed your conversation here Dec 21 01:49:14 rsalveti: need i to help to building gcc? Dec 21 01:49:15 and after a while the board just hang, without giving any extra message at the serial console Dec 21 01:49:17 nothing Dec 21 01:49:33 the heartbeat led stops working Dec 21 01:49:36 so the kernel is dead Dec 21 01:49:41 but no trace at the console Dec 21 01:49:57 cooloney: now how can we debug this kind of issue? Dec 21 01:50:42 maybe some more debugging around the kernel Dec 21 01:51:20 rsalveti: hmm, still have no clue. need a smalll testcase instead of building a whole kernel package Dec 21 01:51:55 rsalveti: highmem might not be a issue. I've tested the linaro kernel which disabled highmem Dec 21 01:52:01 rsalveti: and we still got that issue Dec 21 01:52:11 cooloney: but this is not the one we're used to see Dec 21 01:52:28 in this case the kernel is basically dead, without giving any other message at the console Dec 21 01:53:17 cooloney: see bug 690370 Dec 21 01:53:18 Launchpad bug 690370 in linux-ti-omap4 (Ubuntu) "Strange out of memory on pandaboard (affects: 1) (heat: 10)" [Undecided,New] https://launchpad.net/bugs/690370 Dec 21 01:54:07 rsalveti: oh, my bad, i assume is the old issue Dec 21 01:54:41 rsalveti: thx, man, i will take a look Dec 21 01:55:14 guerby: can you post tomorrow the instructions needed to run the build at the bug? then cooloney can help reproducing this issue with his board Dec 21 01:55:42 cooloney: first issue is the page allocation failure when stressing the memory Dec 21 01:55:51 this is probably caused by the usb driver Dec 21 01:56:07 I believe it's requesting more memory that it should to work normally Dec 21 01:56:19 then the other issue while building gcc is this weird hang Dec 21 01:56:25 no message, no blinking led, nothing Dec 21 01:57:04 guerby: there's also a new kernel if you want to try, that's the one currently used at natty Dec 21 01:57:13 same tree but with some new features Dec 21 02:01:35 rsalveti: thx. i was told before building perl is also hang on panda Dec 21 02:01:56 cooloney: weird, never saw this with my panda Dec 21 02:02:14 we could try natty's kernel and the one based at 37 Dec 21 02:02:50 * rsalveti brb **** ENDING LOGGING AT Tue Dec 21 02:59:58 2010