=== Tim-away is now known as TimStarling === pgraner__ is now known as pgraner === jldugger_ is now known as jldugger === maco_ is now known as maco [12:59] I've kernel mod doing debug <7> msg's, however cant seem to get them visible anywhere on intrepid... (used to -work fine earlier into syslog [13:00] killing syslogd and klogd, and doing "sudo cat /proc/kmsg" shows them, but does not do it continuously (again, -used to work fine) [13:02] (e.g. I can see all messages wth cat /proc/kmsg, but have to keep doing ctrl-c and reissuing the command) [13:15] guys, I would like to ask a Q about kernel native aio , io_submit seems to be blocking so ... does anybody have a clue ? [15:15] is this the right place for problems compiling a kernel module under kubuntu 8.04 ? [15:19] i'm trying to install vbox-2.1.4 on kubuntu 8.04. i can compile the kernel module but when i try to load it i get the following error: [ 621.528827] vboxdrv: Unknown symbol __stack_chk_fail [15:20] have tried different kernels and also removing and reinstalling linux-headers [15:20] Build with -fno-stack-protector [15:20] mjg59: sorry not very good at this. how do i do that ? can i change a global option for my system ? [15:21] I've no idea how the virtualbox build system works [15:22] it's got a Makefile. what is strange is that it worked before, something must have changed on my system [15:22] In general I would suspect they use a "normal" out of tree thing... [15:23] What exactly is before. 8.04 but older kernel? [15:23] i think i have a problem loading the closed nvidia drivers too. also a kernel module [15:25] JackWinter, Hm, could it be that this happened after getting 2.6.24-23.48 kernel? [15:25] no i have the problem with any 8.04 kernels i try. before = a few weeks ago with the same kernels i'm trying now. probably my fault, since i've tried to compile my own kernel and failed on something related to the nvidia driver. not sure if it has anything to do with it, but i suspect so... [15:26] smb_tp: could be but happens with .21 .16 etc.... [15:27] Ok, in that case it isn't what I had in mind. [15:27] vbox people tell me that the vbox module should compile with settings from the linux-headers of my kernel, so i don't understand why it doesn't [15:29] Right IIRC those builds take the config from /lib/modules/$(uname -r)/build. So it should be right if nothing changed that contents [15:30] if it helps ls -l /usr/src: http://pastebin.com/d3e120806 [15:31] looks sensible [15:34] and ls -l /lib/modules/2.6.24-16-generic: http://pastebin.com/d1683e7b6 [15:36] could it not be that some global default has changed for my build enviroment, and that the Makefile doesn't take it into account so tries to build the module with stack checking while the kernel doesn't have it ? [15:36] what does "grep STACK /lib/modules/2.6.24-23-generic/build/.config" print for you [15:37] smb_tp: http://pastebin.com/d363cc5a7 [15:38] ok, same here [15:40] and "grep STACK /lib/modules/2.6.24-16-generic/build/.config" shows the same [15:44] JackWinter, A compile for 1.5.6 vboxdrv works here. Maybe you want to pastebin a build log to compare [15:48] smb_tp: here is the build log: http://pastebin.com/d501a6128, but think that works fine, it's when it's trying to load it that there is a problem [15:50] my guess is that something went wrong when i tried to build a kernel. i managed to build a couple of kernels based on the config of my existing kernel. they failed to install from package though due to some conflict with the nvidia driver. i had some problems until i manually removed the half installed kernels, but think i've cleaned the system out... [15:51] JackWinter, I had hoped the log would be a bit more verbose. To see which options are taken. [15:51] no it doesn't really show very much at all :( [15:53] my guess is that for some reasons the modules build with the wrong options and thus can't load. i think i have the same problem with the nvida closed driver, the module builds but can't load. prefer to trouble shoot with the vbox module though :) [15:53] As it saves a little time, yes. :) [16:10] smb_tp: mind you that i have no clue what i'm doing, but i copied the vbox module source from /var/lib/dkms/vboxdriver/2.1.4 to my home and tried to run a make from the build directory, this is the output: http://pastebin.com/m30123aa1 [16:11] JackWinter, Well, good clueless thinking... ;-) [16:15] smb_tp: in any case the beginning of the output. i have no clue as to what i'm expecting, but it doesn't look like all is in order ? [16:16] It looks strange, some things might be testing some environment. Give me a sec [16:19] here is ls -l /usr/local: http://pastebin.com/d7d6df7c2 include is indeed not a dierectory and the date is recent ? and no worries about when, i am happy for all help i can get :) === TimStarling is now known as Tim-away [16:22] on another install i have /usr/local/include is indeed a driectory, so this must be at least one thing borked with my system [16:24] It is an empty directory on my system [16:25] So at least this being a file is quite strange [16:26] looks to be emtpy on my 8.10 system too. do you think i can just delete it and then recreate it with the right owner and access rights ? [16:27] May be look into it (or at least save it as evidence) to know what probably has written there [16:28] It might be worth a try to replace it by an empty dir again, but who knows what else got changed [16:31] hehe, looks like cpp code :) [16:32] must have been written when i tried to install a program from source...:( [16:35] JackWinter, Things like this could happen more likely if /usr/local/include was not there before. [16:37] smb_tp: think it was always there... in any case i removed it, recreated the dir and recompiled the kernel module. it installed without a problem. hopefully that was the extent of the problem, because i don't really feel like reinstalling the whole system... [16:38] Maybe it got its options wrong since the build system got confused. With the older vboxdrv module on my system here, I saw that -fno-stack-protector was used. This seems to be the case for you again. [16:38] am gonna remove the offending source, because that build didn't complete anyway and i don't really need it. hopefully nvidia will also build again... Many thanks for your help! [16:39] JackWinter, Np, I was just pointing a bit :) [16:39] :)) [16:45] Quick instant-triage on this: https://s3.amazonaws.com/pikchurimages/pic_mba_l.jpg - kernel issue? [16:45] (just happened to my PC on Jaunty upgrade) [16:45] Either your hard drive died or it's a kernel issue [16:45] The latter is probably more likely [16:46] hdd dying is a coincidence too far [16:46] I will file a bug soon then [16:52] I noticed the following run-on message in dmesg: [16:52] [ 20.835571] psmouse serio1: ID: 10 00 64<6>elantech.c: unexpected magic knock result 0x5a, 0x03, 0xc8. [16:52] I think there was supposed to be a newline in there somewhere [17:18] probably off topic but since i'm already here and you guys might know. if i install the closed nvidia 180.29 drivers on my kubuntu 8.04 system, kernel 2.6.24, will dkms still work ? [17:59] mdz, what did the next line look like? [18:00] its possible they really are interlieved one from the interrupt handler [18:00] apw: [18:00] [ 20.835571] psmouse serio1: ID: 10 00 64<6>elantech.c: unexpected magic knock result 0x5a, 0x03, 0xc8. [18:00] [ 21.197163] Adding 1023976k swap on /dev/sdb6. Priority:-1 extents:1 across:1023976k [18:00] mdz ta [18:02] mdz indeed that looks like a missing newline [18:03] mdz can you confirm you have an alps touchpad: cat /sys/devices/platform/i8042/serio1/protocol [18:03] apw: no I do not, this is a desktop [18:03] hmmmm [18:03] mizar:[~] cat /sys/devices/platform/i8042/serio1/protocol [18:03] ImPS/2 [18:03] oddness [18:05] ahh i see, this gets dropped out whether the alps driver attaches or not [18:06] mdz i'll get that patched and push it upstream [18:06] apw: thanks [18:33] <_ruben> hmm .. for some reason my module-assistant built kmod wont load .. earlier versions did work just fine .. now i get: scst: disagrees about version of symbol struct_module [18:34] <_ruben> the headers i built against match the running system's kernel [18:45] "Latest kernel upload:" in the topic really really isn't [18:50] <_ruben> bah .. this is annoying .. cant figure out what broke it [19:08] <_ruben> hmm .. apparently using m-a on a 2.6.27-9-server system with 2.6.27-11-server headers installed, doesnt create proper kmods for 2.6.27-11-server systems [19:08] <_ruben> (did use the -l option, and the package name looked ok) === JanC_ is now known as janc === janc is now known as JanC === JanC_ is now known as JanC [20:19] apw: hello? [20:19] maco, hi [20:21] apw: pgraner says you'd be the person to ask about figuring out how to add support for 2 hotkeys that are on my laptop but don't show up in asus_laptop.c [20:23] apw: i disassembled the dsdt with iasl, but i don't really understand it [20:23] i can have a look at the problem, is there a bug for it? [20:25] bug 268429 and bug 327949 ...the web browser key and the video output switch key respectively. [20:25] Malone bug 268429 in linux "asus-laptop module has incomplete support for hotkeys on Asus" [Undecided,New] https://launchpad.net/bugs/268429 [20:25] Malone bug 327949 in linux "Asus Z37E no keypress registered for video output switching key" [Low,New] https://launchpad.net/bugs/327949 [20:26] hrm i should probably attach the dsdt to the bugs [20:26] yeah do that [20:28] ok its attached to 32749 [20:28] er 327949 [20:29] maco, ok will have a look at the pair, and see what i can do, whats your normal around hours UTC? [20:29] i looked about half way through it trying to make sense of it. i saw CPU and LED sections, and dtchen said the hotkeys won't be named, they'll likely be hex values....but that's all i got [20:30] um...16 UTC to 7 UTC, i guess [20:30] ok cool. will see what i can figure out, likely talk to you your am [20:31] ok. do you mind explaining it after you have a look? i'd like to learn [20:34] sure, we can learn together! [20:35] hahaha [20:36] the most sense i can make of it is that it looks rather similar to verilog === lifeless_ is now known as lifeless [21:35] Keybuk: that email you just sent out about stop_machine_create(), you mentioned a 1/2 second delay. is that 1/2 second total for 40 modules, or is that 1/2 second *per* module? [21:36] total for 40 modules [21:36] ok [21:37] was just wondering :) === Tim-away is now known as TimStarling