[00:04] well there is something i didnt notice before [00:05] in the 28 kernel, even if my load goes above 100 its still peppy and responsive, not sluggish at all === chuck_ is now known as zul [00:12] my load average is now 1.63 [00:12] still pretty peppy [00:25] so i dont know if this will help, but [00:26] when running the 29 or 30 kernels it would bog down and the load would rocket with just 2 terminal windows open [00:26] in the 28 kernel, it took me openingup 10 apps to get the load to 1.60 and even then it was peppy [00:26] you could tell it was under some load but performed decently [00:57] when building my own kernels for testing, they are quite large. what do you have to do to have the debug symbols stripped to cut the size down to a reasonable level? [01:06] psusi: they are stripped if you go through the regular ubuntu build processes [01:07] DrDetroit: kernel.ubuntu.com/~jj/linux-image-2.6.32-29-generic_2.6.32-29.57~lp740549_i386.deb [01:07] download that [01:08] and install it with dpkg -i linux-image-2.6.32-29-generic_2.6.32-29.57~lp740549_i386.deb [01:08] then you can reboot, hold down left shift key to select it from the grub menu [01:08] and let me know how it performs [01:17] jjohansen1, I'm just doing the old make ; make install way from my git tree [01:18] ok let me copy those instructions down [01:27] ok got it downloaded [01:27] will try installing it now [01:30] ok it says it installed [01:30] when i choose it , will i be choosing 2.6.32.29-generic? [01:31] or is it the longer name [01:31] 2.6.32-29-generic_2.6.32-29.57~lp740549_i386.deb [01:32] jjohansen1: will i be choosing 1.6.32-29-generic or 2.6.32-29-generic_2.6.32-29.57~lp740549_i386.deb [01:33] ok its all set to reboot [01:33] I guess I will just find out when i see the menu [01:33] hehe [01:33] back shortly with a report [01:39] jjohansen1: ok i am rebooted [01:39] will give this a test run [01:39] so far i have seen a 2.10 load but still works ok [01:40] my update manager says there is a kernel update of 2.6.32-29 i will ignore that one [01:44] so far so good [01:45] i will let it run for a bit on its own [01:50] yeah the versioning stuff is a bit weird [01:50] I'll assume it is good and start building the next one [01:51] i think its ok [01:51] i will activate my other screensaver [01:51] i can usually see it slow down then [01:52] i am not sure how many of these i can do tonite [01:52] I am an old man and fall asleep easily [01:52] hehe [01:53] ok got hypertorus on a 10 min screensaver [01:53] turned it to 3min [01:53] that should give us a excellent idea [01:54] DrDetroit: np, we can stop when ever you want, though there are only 4 or 5 left [01:54] i will try and do them if i can stay awake [01:54] I'll try to make sure they coming as fast as possible [01:55] its ok [01:55] take the time you need [01:55] and I am not your only concern i bet [01:55] dont forget family and work etc [01:57] DrDetroit: no, but its fairly easy to kick off a new build, and then let it run while I do something else [01:57] I copied and pasted your instructions, it worked very well, thanks [01:57] also as we zero in most of the kernel will already be built and the incremental compiles can go fast [01:58] I am 60, but i still enjoy computers [01:58] * DrDetroit chuckles [01:59] case in point, new kernel up same place, same instructions :) [01:59] hehe, well may dad is 70 and he enjoys them too [02:00] DrDetroit: just in case you missed ^ [02:03] same name? [02:05] yep, it will install right over the old one [02:05] ok getting it now [02:05] I could bump the name if you want, but this leaves less mess for you to uninstall [02:05] i think its cool that yhour dad likes computers [02:06] downloading now [02:08] and installing [02:13] and rebooting [02:13] back shortly [02:13] ogasawara, around? [02:17] bjf: unlikely she started pretty early today [02:19] jjohansen1, that's what i figured but i thought i'd check anyway [02:28] jjohansen1: ok got it running [02:28] \o/ [02:28] I think, if its ok with you, I will run this for awhile [02:28] sure [02:28] I know that sometimes when I rebooted the machine, it took a bit for the problem to appear [02:28] but I will remain logged in here [02:29] sounds good [02:29] maybe we could continue tomorrow? [02:29] I can be here in the morning, or whatever time you are here [02:29] even later on this evening [02:30] DrDetroit: sure what ever time you want to be around [02:30] I live in Arkansas so am on Cental time [02:30] brb [02:31] DrDetroit: heh, well I am west coast but run real weird schedules so I am around just about any time, and if I am not around I can hand off to sconklin [02:33] ok all set up [02:33] will let this run for a bit [02:34] i want to make sure i run all the stuff i would normally run during a session [02:35] Linux bashful 2.6.32-29-generic #57 SMP Wed Mar 23 00:51:49 UTC 2011 i686 GNU/Linux [02:35] [bobp@bashful:]$ w [02:35] 21:35:00 up 2 min, 2 users, load average: 1.60, 0.73, 0.28 [02:35] perky though [02:48] jjohansen: I think this one is ok [02:48] DrDetroit: okay, I'll kick the next one off, should be ready in 10 min or less [02:50] ok [02:50] i can do one more [02:59] DrDetroit: same place, same instructions [03:03] ok [03:03] thanks [03:03] i hope we find it [03:03] * DrDetroit chuckles [03:03] what happens if we don't? [03:04] oh and i have a stupid question [03:04] well then, we retry to make sure we didn't make a mistake :) [03:04] but we will find it [03:05] I notice that on the chat box title it save kernel version 2.6.38 mine is not that currnet, but i thought i had the most current for my dist [03:06] and installing the next one [03:07] hrmm, I am not familiar with chat box, you could run uname -a from a shell and see what it says [03:07] i meant at the top of my chat screen it says [03:07] Natty Kernel Version 2.6.38 [03:08] i was just curious since i thought the one i had was the current kernel version [03:08] ok finished installing [03:08] back after the reboot [03:12] and back [03:12] will let this one run for a bit [03:13] DrDetroit: can you run uname -a in a terminal [03:14] sure [03:14] Linux bashful 2.6.32-29-generic #57 SMP Wed Mar 23 00:51:49 UTC 2011 i686 GNU/Linux [03:15] load average: 1.60, 0.97, 0.42 [03:15] currently 1.60 [03:15] okay that looks good [03:15] but perky [03:15] not really doggy [03:15] lets let this one run for a bit [03:16] this one might be it [03:16] let me watch it for awhile [03:48] jjohansen i thnk this one is ok [03:49] DrDetroit: okay I will kick off the next one [03:50] ok i will probbly download it but not test it till tomorrow if that is ok [03:50] i am starting to fade into the sunset [03:51] DrDetroit: perfectly fine, any idea when you will be on tomorrow? [03:54] what times are you here? [03:55] like I said its all over the place, I expect I won't be on much longer tonight but will be on by 5 or 6 tomorrow [03:59] i will try and catch you in the early am [04:03] DrDetroit: okay, same place, same instructions when ever you can get to it [04:04] ok on my way [04:06] i will report back tomorrow morning [04:06] early [04:06] sounds good [04:08] night and thank you very much === DrDetroit is now known as DrD_zzzzz [04:10] good night [04:44] So... Why will my display manager not work if CONFIG_PRINTK is disabled? [04:44] That doesn't really make sense to me.. === _LibertyZero is now known as LibertyZero [07:31] hi all [07:35] hi [07:43] Hi, Can I force a release version for a compiled kernel? [07:47] hi jjohansen [07:47] fairuz: hi [07:47] jjohansen: I already has a question for today :D [07:47] hehe :) [07:48] Can I force a release version for a compiled kernel? Because when I compiled the kernel, the uname -r is something like 2.6.35-980-aaa123ccc. And there are some modules that are not loaded because of this ( i see the messages during the boot ) [07:49] jjohansen: I see that they took the release from kernel.release file, but how is this file is generated and from where [07:50] fairuz: search for the EXTRAVERSION config symbol in Kconfig [07:50] fairuz: the ubuntu kernel build scripts overwrite it [07:50] fairuz: the release version is generated, the place you can easily tweak is debian/changelog [07:51] amitk: Ah, I thought it's EXTRAVERSION in the Makefile [07:51] at the top of debian/changelog you get something like [07:51] linux (2.6.36-1.7) natty; urgency=low [07:52] you can manipulate the release pocket (natty) [07:52] and the numbers when I do custom builds I usually add ~jj or ~lp##### [07:53] the will just show up in the package numbering eg. (2.6.36-1.7~jj) [07:53] if you use a - it will break the scripts [07:54] jjohansen: ok [07:54] now if you start changing the 1.7 you may need to pass extra parameters to your kernel build [07:54] like skipabi=true skipmodules=true [07:54] which do checks against the abi (numbers you just changed) [07:55] you can of course update the contents of debian.master/abi instead [07:56] of and one more thing, debian/changelog is copied from debian.master/changelog when you do fakeroot debian/rules clean [07:56] so you can make the change in debian.master/changelog if you want it to stick around [07:57] jjohansen: emm I never use fakeroot command :-/. Until now I just do the classic make [07:58] fairuz: ah, so let me change the instructions a bit :) [07:59] jjohansen: :D [07:59] fakeroot debian/rules (I alias this to fdr) is the traditional way of building an ubuntu kernel package [07:59] fdr clean sets up for the environment [07:59] fdr prepare- [08:00] setups the build (called by binary- if it hasn't been done) [08:01] fdr binary- builds the kernel, and package it up into a linux-image.XXXX.deb [08:01] jjohansen: ah ok [08:01] of course you don't need to do this if you are just building a kernel [08:01] fairuz: I can point you at a wiki page if you are interested [08:02] jjohansen: sure [08:02] jjohansen: it will be good to discover that [08:02] fairuz: first up generic jump off page ubuntu kernel team uses https://wiki.ubuntu.com/Kernel/Dev [08:03] https://wiki.ubuntu.com/Kernel/BuildYourOwnKernel [08:03] is the specific page [08:03] lots of useful information in those pages [08:04] anyways, ubuntu uses the CONFIG_VERSION_SIGNATURE entry in the .config [08:05] eg. [08:05] CONFIG_VERSION_SIGNATURE="Ubuntu 2.6.36-1.7-virtual 2.6.36 [08:06] this gets cons together by the build scripts from the debian/changelog and some vars [08:06] jjohansen: ok [08:06] if you are just doing make you can set that and just build [08:07] * jjohansen likes straightup make for incremental development, as there is no package to build/reinstall [08:07] its quick to just make ; make install [08:08] jjohansen: ok. On my compiled kernel. uname -r gives me something like serial number. Does it came from debian/changelog too? [08:09] jjohansen: v2.6.38-366-g65e01eb to be exact [08:09] jjohansen: sorry mistake.. this one 2.6.38-00366-g65e01eb [08:10] fairuz: well that one is strange to me, your build arm aren't you [08:10] jjohansen: yes [08:11] okay, I actually haven't done an arm lately so I am not sure I will have to dig a little bit [08:11] jjohansen: I just clone the git tree, then prepare the .config, then just do make ARCH=arm CROSS_COMPILE=arm-none-linux-gnueabi- uImage [08:11] * jjohansen needs to learn anyways as I got my affika mx today and it needs a new kernel :) [08:11] jjohansen: cool :D [08:12] jjohansen: and since the kernel release is like that, when it boots, there are errors like /lib/modules/2.6.38-00366-g65e01eb/xxx/xxx/xxx cannot be found etc [08:13] did you do a make modules_install? [08:14] what is the concept of designing the multiprocessor for executing a single program?.. please help..:) [08:14] update the initramfs -uk 2.6.38-00366-g65e01eb [08:14] aakshay: not sure I understand the question? [08:17] jjohansen: no. Never heard of module_install before [08:18] fairuz: make that modules_install [08:18] s on the modules [08:18] after doing a make install, I always do a make modules_install [08:18] jjohansen: Before or after the actual compilation? [08:18] it will copy all the modules into /lib/modules/XXXX [08:18] after [08:18] ok [08:19] make ; make install ; make modules_install [08:19] sudo update-initramfs -ck or -uk [08:19] then update-grub on x86 [08:20] jjohansen: If I cross compile, is it the same? [08:20] jjohansen: or I have to compile on the arm board ( which I think will take forever ) [08:20] fairuz: I assume so, as long as all the targets are set right [08:21] fairuz: the make builds the modules, make modules_install just copies [08:22] jjohansen: wait there is something I dont quite understand. You said that it will copy to /lib/modules/xxxx [08:22] well install to, yes [08:22] jjohansen: but that is my filesystem on the board ? [08:22] right [08:23] you need to get the modules in location some how [08:23] how do you install the kernel to /boot? [08:23] i just scp it to the board [08:24] since both machines are on network [08:24] jjohansen: not the right way to do it i suppose? [08:25] jjohansen: thanks for considertaion.. i would like to design the multirocessor system to chek the memory acces(using array as memory).. so can u please guide me how to code? [08:25] ah then you are going to have to do something similar for the modules [08:26] jjohansen: ah ok [08:26] fairuz: no it works, make install is just a convenience that copies to the standard location and has a little bit of extra magic baked in [08:26] jjohansen: then i just copy from this standard location to my board's filesystem? [08:27] as long as you build with your compile target separate from the source directory you should be able to just scp your built modules [08:27] yeah [08:27] fairuz: look in the Makefile for modules_install [08:28] its a pretty simple target [08:29] jjohansen: i saw it yes [08:29] aakshay: hrmm, I'm not sure I under stand what you trying to do still. Are you trying to emulate a multiprocessor system running multiple threads accessing memory at the same time? [08:31] aakshay: our right a multi-threaded program that is running in parallel on a multi-processor system? And what do you mean by check memory access [08:34] jjohansen: yes I want to emulate a multiprocessor system running multiple threads accessing memory at same time. [08:34] jjohansen: so please tell how to implement it? [08:35] jjohansen: let me resume what I have to do as it's not yet very clear. http://pastebin.com/ncccv0vb [08:39] aakshay: well that sounds like an assignment :) The basics are to emulate an instruction per processor, and each processor will do a memory lookup on the array, your code will have to deal with the processor specifics about the emulated memory cycle if you want to be accurate with how smp system deal with memory accesses, collisions, etc [08:40] aakshay: I really don't have enough info even to say much beyond that, and being what looks like a home work assignment I am hesitant to say to much beyond general hints [08:40] fairuz: no [08:41] kamal: which was the command you suggested after fdr binary-generic? [08:41] kristian-aalborg: binary-headers ? [08:42] fairuz: you would only run make modules_install if the build was on the machine you are installing on [08:42] ah, found it... binary-headers [08:42] jjohansen: ok that make sense [08:43] fairuz: so make config ; make ; cp uImage to boot partition ; cp modules to boards file system [08:43] @cp -f $(objtree)/modules.order $(MODLIB) [08:43] @cp -f $(objtree)/modules.builtin $(MODLIB) [08:43] this time, I turned on skipmodule=true before doing anything, and it finished in the first try [08:44] that is from the modules_install target, it copying from the build tree to the target fs [08:44] kristian-aalborg: \o/ [08:44] I hope it will make a difference... [08:44] jjohansen: ok [08:45] fairuz: modules_install also can do firmware stuff but we will deal with that later if needed [08:48] jjohansen: yes. its an assignment.. :P... i am trying to code this but not finding the right sources to start with? :( [08:48] the proper way to remove kernels installed with dpkg -i is dpkg -r, correct? [08:50] aakshay: think of each cpu as a thread, running simultaneously. You can emulate this by stepping each cpus instruction pointer one instruction at the same time. Running the instruction, and potentially not commiting results of the instruction. That will all depend on the architecture characteristics you are trying to emulate. I don't even have enough info to tell you more [08:51] kristian-aalborg: that or --purge [08:58] jjohansen: thanks for the help. :).. one more thing what is the use of "%*s" in " fscanf(fd, "%*s%d", &num_of_regions);" ? [09:01] aakshay: %* surpresses the match string s matches a non-whitespace string of chars [09:01] %d matches numbers, and stores it in &num_of_regions [09:02] aakshay: do man fscanf (or google it) [09:06] jjohansen: googling it.. :) [09:15] it *still* just reboots :( [09:15] is there a kind sould who'd try building my .config? [09:16] or have a look at it... it should be fairly basic [09:17] kristian-aalborg: I can look at your config, but if its building for you I doubt that building it else where will reveal much except a successful build [09:17] kristian-aalborg: so you aren't even dropping to an initramfs just going straight to reboot [09:18] yes [09:18] the screen just goes black, then the machine reboots [09:18] kristian-aalborg: can you remind me what type of machine [09:18] a ThinkPad 770, of course :) [09:19] kristian-aalborg: grub, grub2? [09:19] grub2 [09:19] and lucid [09:20] kristian-aalborg: if you hold down the left shift key after the bios screen does the grub boot menu come up [09:21] http://pastebin.com/dGuzdH9N [09:22] I have 10 seconds of grub by default [09:24] kristian-aalborg: okay, so we at least know its getting to grub [09:25] yes, but when I choose the kernel the described mishap takes place === smb` is now known as smb [09:31] kristian-aalborg: is your cpu a pentium-mmx? [09:34] yes [09:34] kristian-aalborg: do you have a working .config to diff against? [09:35] not apart from the one from git [09:35] http://www.thinkwiki.org/wiki/Category:770 <--- the box in question [09:36] * smb has not read the full scrollback but wonders whether pae might be involved. Or our processor settings... [09:36] kristian-aalborg: so you don't have kernel that boots on the machine or you don't have the config for the kernel? [09:36] ah, sorry, I mistunderstood what you meant [09:36] smb: yeah that is what I am trying to figure [09:36] * misunderstood [09:36] yes, I have the generic kernel on it, from a regular update - it boots fine [09:37] kristian-aalborg: okay, so the -generic i386 works, got it [09:39] what I did was make oldconfig then a few tweaks with make menuconfig - then cp to the machine that I build on [09:40] kristian-aalborg: do you update the initramfs? [09:40] jjohansen, Hm wait. make oldconfig? on what? [09:41] smb: good question [09:41] make oldconfig on the 770 then move it to a newer machine === ogra is now known as Guest42038 [09:42] What I am wondering is that if you use the debian/rules process, it will create a new config under debian/build/... === Guest42038 is now known as ogra_ [09:42] smb: I copied .config after that, per jjohansen's instructions [09:43] Ah ok. Did miss those instuction parts [09:43] this was a few days ago [09:43] smb: it was a few days ago [09:43] heh ok [09:43] Rome wasn't built in a day :) [09:45] Note, that if you want to tweak a config, you may also run "fakeroot debian/rules editconfigs" It might be that older releases do not have the y/n questions for each flavour but one can ctrl-c after the config that was of interest [09:46] kristian-aalborg: these configs are significantly different [09:46] which configs? [09:46] yours and the -generic kernel [09:46] hurm.. [09:46] anything "risky"? [09:47] * kristian-aalborg has to go to work shortly [09:47] kristian-aalborg: different enough all bets are off [09:48] hmmm [09:48] kristian-aalborg: can you redo your config changes using fakeroot debian/rules editconfigs [09:48] that will call make menuconfig, and then save off your changes [09:48] I'll give it a shot, but it will be later [09:48] however, changing .config is what it's all about...? [09:49] kristian-aalborg: sorry, something didn't take right when building your config [09:49] kristian-aalborg: right, but you want to work from a know working config and then tweak it to what you want, its much easier that way [09:49] I *think* I apt-got the source on the 770 - the machine I build on got it by git [09:50] make oldconfig it not sure-fire to work? [09:50] kristian-aalborg: shouldn't matter [09:50] kristian-aalborg: no it isn't [09:50] especially in this case [09:51] ah, that might be it [09:51] it's also fairly new, I think [09:51] so - next step: build a generic kernel, move and test - then tweak [09:52] perhaps compare to the config from makeconfig [09:52] kristian-aalborg: you don't need to build the generic kernel [09:52] jjohansen: did you try building it? [09:52] kristian-aalborg: no [09:52] k [09:53] kristian-aalborg: fakeroot debian/rules clean [09:53] kristian-aalborg: fakeroot debian/rules editconfigs [09:53] jjohansen: I have some slavery to do :( [09:53] this will edit your build configs [09:53] see you guys later - I feel like there's some kind of progress [09:53] then fakeroot debian/rule binary-generic [09:53] or what ever build target you were using [09:54] work > geekery, unfortunately [09:55] yeah [09:55] see ya [09:56] * smb thinks jjohansen also has something important to do :) [09:56] smb: really what could that be ;) [09:57] jjohansen, Weeell, there are many sheep to be counted. :-P [09:58] smb: oh shoot, yet another thing that needs to be done. Let me guess this is a rather high priority task :) [09:59] smb: how often do you build on zinc? [10:05] jjohansen, that goes near nill [10:05] jjohansen, The mainline builds used to be done there though [10:06] smb: okay, thats what I thought. /me was trying to figure out how to make a bisect public without doing the builds there, /me just ended up using tangerine and making the debs public [10:06] jjohansen, Yep, much quicker that way [10:19] jjohansen: sorry to ask again but what the files that should I transfer to /lib/modules of the arm board after I cross compile the kernel? [10:25] fairuz: the actual list varies based on the build, modules_install does [10:25] @cp -f $(objtree)/modules.order $(MODLIB)/ [10:25] @cp -f $(objtree)/modules.builtin $(MODLIB)/ [10:25] and then some magic to get the list of .kos to copy [10:26] fairuz: have you ever used sshfs [10:26] jjohansen: nope -( [10:27] fairuz: it allows you to mount a remote machine over ssh, so you can do things like make install and make modules_install to the remote machine [10:27] jjohansen: right now the kernel boots but when i tried to insmod an external module, it gives me invalid module format error. So it must be because of /lib/modules files right? [10:27] sadly its been a while since I used it so I don't remember the details [10:29] fairuz: likely, we are getting to the level I would have to fiddle with things to see what is happening [10:29] manually installing modules isn't something I have had to do for a long time [10:30] fairuz: I would google sshfs and take a look at using that [10:31] its probably been 6 months since I've used it for kernel installs but it worked well for me [10:31] jjohansen: ok. it's different from normal ssh? [10:32] fairuz: its fuse module that leverages normal ssh [10:32] it makes the remote machine show up as a mount on the local machine, so you can treat it like a local disk [10:33] then you can set your make install target and install to your remote machine [10:34] jjohansen: ok [10:35] sorry I am not more specific, right now I just have to resort to google to find out how to use it again, but I remember it working well for me [10:36] IIRC I did it from the target machine to the build machine, so the build machine mounted as a directory on the target, [10:37] I cd into the build machine directory on the target machine [10:37] and ran make, make modules_install, so the remotely built kernel installed to the local (target machine) [10:38] that way I didn't even need to specify make variables to set where the target should be installed [10:39] jjohansen: ah ok [10:39] jjohansen: i just did the opposite [10:39] :D [10:40] :) [10:47] jjohansen: but if I do make like that, it will use which compiler? target machine or build machine? [10:48] fairuz: do your make on the build machine like normal, after you have built, mount the build machine from the target, and then do make install, make modules_install [10:48] jjohansen: aha, i got the idea [10:48] it will use the targets make but everything is already built, so its just running the scripts, copying files etc [10:49] jjohansen: i'll try that after lunch :D thanks a lot for the help [10:49] fairuz: np, hope it works for you === DrD_zzzzz is now known as DrDetroit [10:55] and rebooting to test out last nights download [11:01] ok will let this one run for a bit [11:01] DrDetroit: sounds good, let me know when your ready for the next one [11:02] will do, just woke up, so have to think about coffee (maybe w/irish cream) and feeding the felions [11:02] hehe [11:44] jjohansen: I think this one is ok [11:45] DrDetroit: alright, I'll get the next one building [11:46] btw if you need someone to do this kind of thing for other reasons, I don't mind being a guinie pig [11:46] DrDetroit: hehe, we like guinie pigs :) [11:46] hehe i used to be a beta tester for lucent technologies [11:47] we tested their max3000 and the maxTNT [11:47] when I owned and ran the rural isp here [11:56] DrDetroit: same place kernel.ubuntu.com/~jj/linux-image-2.6.32-29-generic_2.6.32-29.57~lp740549_i386.deb [11:56] same install [11:57] ok [11:58] 1:01, 2 users, load average: 2.17, 1.39, 1.09 [11:59] even with that, its not mushy or slow to respond [12:00] maybe just a tad, but thats because i was dl'ing [12:00] well lets give it a few minutes [12:01] nod [12:01] i am just installing the new one now [12:01] that was with the old one [12:02] while i was downloading [12:02] ah, okay [12:05] back in a min, rebooting [12:11] and back [12:11] will let this one run for a bit [12:24] Q: on Hardy, the package linux-image for Xen should update /boot/grub/menu.lst, correct? [12:33] jjohansen: Before doing install_modules, should i backup the /lib/modules folder? because I dont want to ruin a working setup [12:33] fairuz: you don't need to if the kernel being installed has a different version, it will create its own directory, so you can multiple kernels installed [12:34] jjohansen: yes but in my case they have the same name [12:34] jjohansen: i can just rename the old one right? [12:35] fairuz: hrmm, that is problematic, you can rename the old one but you might not be able to boot then [12:35] that is boot into the old one [12:36] jjohansen: so i should recompile with different name? [12:36] fairuz: that would be best, you can do an incremental compile [12:36] remove the debian/stamps/stamps-build-XXXX file [12:37] jjohansen: what's that file for? [12:37] change the name and then do fakeroot debian/rule binary-XXXX [12:37] the build scripts use it for time stamps of which stages completed [12:37] remove the build stamp and it will rebuild the kernel [12:38] but if you still have all the .o around from a previous build, make will only compile the parts that have changed [12:38] fakeroot debian/rule binary-uImage is the same as make uImage? [12:38] it makes for a fast kernel compule [12:38] fairuz: doh I keep forgetting you are using arm [12:39] fairuz: no, you don't need to remove the stamp file, you shouldn't have it for the way you are building [12:39] jjohansen: i have a debian and also debian.ti-omap4 folder in my source folder [12:39] just change the name and use make uImage [12:39] jjohansen: ok. sounds easier :D [12:40] fairuz: right but aren't using them to do the build are you? [12:40] fairuz: you aren't making a .deb that you then install with dpkg [12:40] jjohansen: nope [12:40] so just use make [12:40] jjohansen: I just need the kernel image [12:40] jjohansen: ok [12:42] jjohansen: Should I do this module_install systematically each time I compile the kernel or just in certain case? [12:44] fairuz: depends, if you make a change that could affect the modules yes, update an include etc. other wise as long as the abi doesn't change there should be no need [12:46] jjohansen: ok. btw, received your arm box yet? [12:48] fairuz: yep it came yesterday, updated it to natty but I haven't had time to really do anything with it yet [12:50] jjohansen: I am going to run to town soon to do some errands, I will let the current kernel run for the time I am gone, if that is ok with you [12:50] if it does not mess up, then we can move on when i get back [12:50] DrDetroit: sounds good [12:50] thanks === jjohansen is now known as jj-afk [12:51] back on in a couple hours [12:51] * DrDetroit nods === DrDetroit is now known as DrD_away [13:05] kernel team -- I need to know if the hardy xen package should update /boot/grub/menu.lst; if it should, the -proposed does not do it [13:10] hggdh, Now I see it... [13:11] When I do make modules_install, it gives permission error, same with sudo make modules_install. Should I install it to another writable folder first then move them to /lob/modules? [13:11] *lib [13:16] tgardner: so which build box are you doing armel test builds for natty since tangerine doesn't have the natty-armel chroot? [13:18] ogasawara, you can use urbana which is running Natty, and has a functional armel schroot for natty. Otherwise, I'm just cross compiling. [13:19] do i smell an omap4 upload for today ? :) [13:19] ogra_, shortly [13:19] awesome !! [13:20] ogasawara, you can also upload to the c-k-t PPA, but the builds take awhile. [13:22] ogra: I'm still not able to get a functional armel qemu schroot on a Lucid host. It does work on a Natty host. Any thoughts ? [13:23] do you use a backported version of qemu-linaro on lucid ? [13:24] ogra_, as far as I know. I've got -backports enabled === jdstrand_ is now known as jdstrand [13:24] i dont think its in backports, the linaro gusy have their packages in a ppq [13:24] *ppa [13:25] but there should be a lucid version, i'd try if that works better [13:26] ogra: I've haven't looked in a couple of weeks, so I'll try again. === sconklin-gone is now known as sconklin === DrD_away is now known as DrDetroit [14:00] and back [14:00] * ogasawara back in 20min [14:01] sconklin, I'm loving the pictures :) [14:03] JFo, You got pictures of sconklin? :-P [14:03] smb, I've seen some [14:03] ooold ones :) [14:04] JFo, there are more you'll never see ;-) [14:04] :-( [14:04] * JFo is sad now [14:04] Ohh, the kind of that mothers usually take out at the most inappropriate times? :D [14:04] smb, hah! not quite that old :p [14:04] I've been starting to scan the film that I shot in the 70's and 80s [14:05] JFo: I've probably still only even glanced at about 5% of my files [14:05] I can only imagine [14:06] I agree with the sentiment that you should start back taking more, though. [14:06] great stuff [14:06] it's kinda sad. Some of the happy young folks in those photos are now broken down addicts. Or not around [14:07] the unfortunate story of life [14:07] When I get more I'll put them in sets on flickr or picasa or something. Until then, Facebook is a fun way to get reactions to them [14:08] I found another box of various band photos last night, need to go through those [14:08] strolling down amnesia lane as a friend would say. :-) [14:10] I find a lot of things I forgot about, and can't find the ones I am specifically looking for. That's the way, isn't it . . . [14:10] unfortunately :-/ [14:24] ok i guess i can go out and mow the lawn or something while I wait for jjohansen to return === DrDetroit is now known as DrD_away [14:25] pretty sad to be mowin in March [14:25] * DrD_away shakes his head === diwic is now known as diwic_afk [14:59] Hi, when the modules.dep are generated? because when I do modules_install, it's not there. [15:08] Hi, can we generate a header folder from a source? [15:08] (kernel source) === herton is now known as herton_lunch === BenC_ is now known as BenC === cmagina is now known as cmagina-lunch === jj-afk is now known as jjohansen [16:53] hi again [16:54] jjohansen: I am ready again when you are === DrD_away is now known as DrDetroit [16:54] DrDetroit: how did the last kernel work out [16:54] it seems to be ok [16:54] i am using it now === cmagina-lunch is now known as cmagina [17:02] jjohansen: it seems to be ok [17:02] i am using it now [17:03] DrDetroit: hehe, yeah sorry, next kernel is ready :) same place [17:04] no probelm [17:04] no hurry [17:05] I am glad your back cause my lawnmower started right up, so if you were not here, I would have to mow [17:05] hehe [17:05] DrDetroit: well we can't have that [17:06] ogasawara, did you see that there has been another gcc upload? You should ask doko if there is any code generation impact wrt the kernel. [17:06] tgardner: argh, that'll screw us for beta freeze if we have to perform another kernel upload [17:07] ogasawara, well, the gcc upload has an enormous changelog. I can't tell what impact it might have. [17:10] ppisati: ping. You will need to retest bug 732700 as I haven't come across this issue before. Other than that, the 216/416 kernel looks good to me for dove. [17:10] Launchpad bug 732700 in linux-mvl-dove "apparmor_parser triggers a kernel panic" [Undecided,Fix committed] https://launchpad.net/bugs/732700 [17:10] http://ubuntuforums.org/forumdisplay.php?f=332 <--- I changed the scope of my little project :) [17:12] GrueMaster: pong [17:13] GrueMaster: actually that problem triggered using, for example, a maverick kernel on a lucid userbase [17:13] GrueMaster: but yeah, i can confirm it has been fixed [17:15] mpoirier: ping [17:16] ppisati: pong [17:16] mpoirier: do you have any OMAP3 board? [17:17] ppisati: not that belongs to the company no. [17:17] mpoirier: uhm k [17:17] mpoirier: because i'm looking for a OMAP3 board [17:17] mpoirier: to cleanup lucid/ti-omap [17:18] ppisati, i had several beagle board flavours, i think they got shipped to either eric or brian [17:18] ppisati: What on lucid needs cleaning up? It wasn't a fully supported release iirc. [17:18] bjf: i'll ask them, thanks [17:18] GrueMaster: it seems my duty here is to cleanup all the arm-related bits, so... :) [17:18] ppisati, its likely easier to just buy one rather then have it shipped from Asia === sforshee is now known as sforshee-lunch [17:19] jjohansen: ping [17:20] kristian-aalborg: pong [17:20] ppisati: Do you have specific bugs you are looking at? [17:20] see how I just went open source ;) [17:20] GrueMaster: https://bugs.launchpad.net/ubuntu/+source/linux-ti-omap [17:21] GrueMaster: as many as i can [17:21] GrueMaster: i'm ping-pong-ing with saeed about mvl-dove, i'm one step closer to get audio working [17:21] GrueMaster: let's see... [17:21] cool. [17:22] GrueMaster: in the mean time, i'm trying to be productive on another front [17:22] kristian-aalborg: which thread? [17:22] GrueMaster: but my XM doesn't collaborate... [17:22] GrueMaster: from time to time, it spontaneously reboot [17:22] GrueMaster: eth doesn't work anymore [17:22] As to the lucid omap issues, I can see if they are reproducable. You need a Beagleboard C4 for those. [17:22] GrueMaster: etcetc [17:23] btw, ti-omap4 in maverick, there're some CVEs open [17:23] but it says "no rebase on master" [17:23] so, do i cherry-pick the fix from master an apply locally? [17:23] ppisati, yep [17:23] I don't think they were pulled into a kernel update for omap4. [17:23] tgardner: goocd [17:23] kristian-aalborg: never mind found it [17:24] ppisati: Since I am the QA guy for the canonical-arm team, if you need something verified before you attempt to dive in, just let me know. I have all of the supported platforms online and all of the images for testing. [17:25] GrueMaster: ok [17:25] We can work together to clean up the cruft. [17:25] GrueMaster: perfect [17:26] GrueMaster: but let me first find a beagle C4 [17:26] ogasawara: I thought the issue was the kernel had a naive check for compiler version number that prevented out-of-tree modules from rebuilding after *every* gcc rev, whether or not there are code changes? [17:26] tgardner: ^^, is that correct? [17:27] * ogasawara is not familiar enough to know [17:27] ogasawara, I think thats right. Lemme figure it out. apw could tell me off the top of his head. [17:30] bjf: you mean, ericm and bdmurray, right? [17:30] hurm, looking at /boot/config - I thought the distro kernel was just downloaded and installed, not individual? [17:30] ppisati, i meant ericm and cooloney [17:31] bjf: k [17:32] ogasawara, gcc versioning changed from 4.5.2-6ubuntu5 to 4.5.2-7ubuntu1, so I think we're OK [17:34] ogasawara, on the other hand, why did you have to do a no-change rebuild last Friday? Only the minor version changed. [17:37] tgardner: my understanding from what slangasek said is that any gcc revision forces us to re-upload. [17:38] tgardner: he noted "the whole problem we have with needing kernel rebuilds after gcc updates is that having the kernel check for the things that actually matter at build-time was too hard so there's a compiler version 'fingerprint' check instead" [17:38] if the latter is correct, I need to move my /boot/config from the old machine before editing it? [17:38] ogasawara, and the fingerprint includes then _whole_ version? [17:39] tgardner: that's what I'm guessing. [17:39] tgardner: I'm adding this as a discussion point at UDS [17:39] ogasawara, well then, that seems to indicate yet another upload. [17:39] tgardner: indeed. I've already given skaet a heads up we'll be violating the beta freeze [17:40] I'm trying to figure out why when you detach a loop device sometimes it'll error out reporting device or resource busy. I have a script that can reproduce it about 60-70% of the time. Adding a command that doesn't write to the loop device makes it much less likely to happen. I thought the issue might be that the kernel still has things to write buffered and that a sync would fix the issue. However, after running the script 11 [17:40] 86 times it failed. [17:40] ogasawara, we don't have any external modules right now in Natty that we care about. Everything else should be DKMS. [17:41] I wonder if the DKMS rebuild check looks at the finger porint, or just the kernel ABI. likely the latter. [17:45] cody-somerville, do you have to reboot in order to fix it ? [17:45] tgardner, no. I can losetup -d immediately. [17:46] cody-somerville, well, thats not a lock imbalance then. [17:46] tgardner, It appears there is some sort of race condition or something after writing to a loop device. [17:47] cody-somerville, have you talked to Surbhi? She's done some work on mountall races recently. [17:47] I have not. [17:48] cody-somerville, given that losetup -d cleans up the issue, it sound a bit like an app space race. [17:52] tgardner, I was thinking the issue was maybe lo_refcnt being > 1 when loop_clr_fd is called. [17:53] tgardner, ie. kernel is trying to write out buffer but due to thread/lock contention its slightly delayed. [17:54] cody-somerville, can you tell who still has file open on the loop device by using fuser? [17:54] has a file* [17:55] tgardner, I haven't been able to get fuser to show anything. I've tried running fuser before losetup -d, right after losetup -d when it fails, etc. and it never shows anything. [17:55] tgardner, I'm pretty sure it isn't a user space process accessing the loop device. [17:56] cody-somerville, how about repeating 'losetup -d' until it succeeds? (seems like it should block until the underlying device is done) [17:57] tgardner, I was thinking of patching losetup to do that, yea. If I can't fix the underlying problem, I'll have to keep retrying losetup -d as you suggest. [17:57] cody-somerville, how emailing your script and I'll see if I can repro ? [17:58] tgardner, Is pastebin okay? [17:58] sure [17:58] tgardner, http://pastebin.ubuntu.com/584410/ [17:59] tgardner, comment out all the sync commands if you want to reproduce easily. [17:59] cody-somerville, ok, gimme awhile. [17:59] tgardner, FYI, I wrote the script initially to test a bug in parted (hence all the calls to fdisk -l), [18:00] You'll need to run the script as root naturally. [18:04] jjohansen: looks like this one is ok also [18:05] DrDetroit: okay 1 more kernel [18:05] pooh, what happens if it doesn't fail? [18:05] maybe we have inadvertenly fixed my problem [18:06] i guess i can always run the 30 for awhile to see if what ails me is gone [18:06] hrmm, well looking at the commit, it can't be the problem [18:06] i am not smart enough to know [18:07] DrDetroit: can you retry the known bad kernel and see if you can replicate the problem [18:07] that is what i meant by running the 30 kernel [18:07] that is where i first noticed the issue [18:07] right [18:07] well i am ready to try the last one on this run i guess === sforshee-lunch is now known as sforshee [18:11] jjohansen: i am ready for the last kernel when you are [18:11] jjohansen: if that passes then i will run the 30 kernel for awhile [18:12] DrDetroit: it will pass, its a book keeping commit only changing changelogs, etc [18:12] jjohansen: then i will reboot to the next kerne; [18:13] thank you for all your help, i will continue to be logged in here as i run the other kernel [18:13] I am rechecking the bisect range, maybe I got the wrong tag [18:14] DrDetroit: I'm surprised anyone remembers that forgettable movie ;) [18:14] hehe [18:14] it has been my "game" name for many years now [18:15] I started using it as a MUD charater name back in 98 [18:16] the funny thing is, Ackroyd would end up being in much more embrassing things later [18:16] hehe [18:16] when your selling yourself, apparently the only issue is price [18:17] * DrDetroit chuckles [18:18] Dragnet is cool, though [18:18] brb, rebooting === tgardner is now known as tgardner-lunch [18:25] jjohansen: I am rebooted and will let this one run while I do lunch and afternoon chores [18:25] I will watch it to see if the problem reappears [18:25] DrDetroit: sounds good [18:26] thank you for helping me [18:28] How can I trace a kernel thread? === herton_lunch is now known as herton [18:43] aha! [18:43] the vanilla build of the kernel killed my 770 also, thanks to whoever suggested trying it [18:49] I cp'd the working config over, now I need to tweak and build === sconklin is now known as sconklin-afk === DrDetroit is now known as DrD_away [18:56] * bjf -> lunch === bjf is now known as bjf[afk] [19:23] * kristian-aalborg builds; eats yogurt... [19:23] * jjohansen -> lunch [19:33] anybody got a nice script for building? fdr this, fdr that... [19:33] jfk this, gwb that... === bjf[afk] is now known as bjf [19:45] kristian-aalborg, https://wiki.ubuntu.com/Kernel/BuildYourOwnKernel [19:47] I think I'll make a script of that === tgardner-lunch is now known as tgardner === sconklin-afk is now known as sconklin [21:21] jjohansen: good news, building from the copied config seems not to halt [21:22] kristian-aalborg: good to hear [21:23] yup... if everything goes well, I'll be using that as my base [22:01] tiny script: http://pastebin.com/sUVB3baF ... comments welcome [22:10] jjohansen: I have left it run for 3.5 hrs while outside and the performance has degraded === DrD_away is now known as DrDetroit [22:10] up 3:49, 2 users, load average: 1.36, 1.44, 1.39 [22:11] DrDetroit: okay, I think we are going to bisect a slightly different range this time [22:11] can i wait a bit [22:11] jjohansen: I have to do evening chores === sconklin is now known as sconklin-gone [22:12] jjohansen: this was the 30 kernel [22:12] DrDetroit: np, I'll set something up and when your ready I'll have a kernel waiting for you [22:12] ok [22:12] jjohansen: may I ask a question? [22:13] jjohansen: Are we going to do the 29 again? or are we going to do the 30? [22:14] jjohansen: I am also wondering if I should let the 29 kernel we just finished run for a few hours like i did this one [22:15] well I was thinking .28 - .30, which would cover everything, and we let them run longer [22:15] so yeah [22:15] let me go ahead and before we start, let me run the one we just finished for a couple hours [22:16] maybe we can start on this this evening or morning [22:18] jjohansen: brb gonna reboot back to the 29 we just finished [22:22] jjohansen: ok back on the 29 kernel we just finished [22:22] I will let it run for a couple hours [22:22] and see what happens [22:22] yeah, sounds like a plan [22:23] too weird [22:27] jjohansen: I know yhou have better things to do that hold my hand, but I sure do appreciate it [22:28] DrDetroit: its no problem, you are doing most of the work, I'm just cranking up the bisect as needed. Your performance bug is a pretty serious regression and your not the only one who has reported seeing the behavior so it would be nice to track it down :) [22:30] jjohansen: it's nice to know I am not the only one, and I will be glad take whatever time we need to figure it out [22:32] DrDetroit: are you sure you used the correct commands? [22:32] I just waited an hour for nothing because of a typo.... [22:35] kristian-aalborg: I am just testing out kernels that jjohansen is providing to me [23:22] jjohansen: back on the 28 kernel, will run it for a couple hours [23:22] the 29 was starting to slow down [23:22] I will report back after this one runs for awhile [23:23] DrDetroit: I though .28 is the one you were running that was good and you have been running that one for hours already [23:23] ie. it was in general use until .29 without problems [23:25] no i was running the 29 [23:25] first i ran the 30, then the 29 and i just rebooted to the 28 [23:25] and so far as i can tell atm the 28 is good, but i thought i would run it again for an hour or so while i do evening chores [23:26] I can go back and run the 29 again for a while [23:26] i dont mean to confuse you [23:27] I just wanted to double check that 28 was good [23:30] Once I get back from chores, I will run the 29 again that we just finished [23:36] DrDetroit: okay thanks no worries, far better to take our time and make sure we got the right results this time