=== kklimonda1 is now known as kklimonda === _LibertyZero is now known as LibertyZero [08:40] which process has pid = 0? === smb` is now known as smb [08:40] itih [08:40] fairuz: init. [08:41] RAOF: ty [08:41] (Which, in Ubuntu, will be upstart) [08:42] RAOF: does this process will run with the kernel? [08:42] RAOF: or just at the beginning [08:43] I'm not sure what you mean? [08:43] RAOF: when I do ps -A, I see that there is no pid 0, init is at pid 1 [08:44] Ah. Then maybe I meant “there is no pid 0, but init is at pid 1” :{ [08:44] Ah. Then maybe I meant “there is no pid 0, but init is at pid 1” :) [08:44] so pid 0 doesnt exist at all? [08:45] Now that I think of it, ‘0’ is probably used as a flag value for pid. [08:45] init is 1 [08:46] yes, I see that init is pid 1 [08:46] i just wondering, since I saw a code that set the affinity of the processor [08:46] and it passes 0 for the pid argument [08:47] And I just wonder which process does have pid 0 [08:47] for syscalls, passing 0 as a pid usually means "myself" [08:47] ah ok [08:47] that make sense [08:48] "If pid is zero, then the calling process is used." -- SCHED_SETAFFINITY(2) [08:49] Krunch: RAOF: ok all make sense now. Thanks [08:51] if we set sched_setaffinity to current process [08:52] that doesnt mean that if we start another process after that, it receive the same affinity? [08:52] it just affects the process that call the setaffinity, right? [08:54] read the friendly manual [08:54] A child created via fork(2) inherits its parent’s CPU affinity mask. [08:56] Krunch: yes i read it. Thanks. But my concern is if I launch a process A to start a hardware counter, then start the process B, and recall process A to stop the counter [08:56] so if I set the affinity in process A, it does not affect process B? [08:57] (process B is not a child of process A), consider its a random test case [08:57] i don't see why it would [08:57] ok [08:57] but just try [09:29] does mmap in user space do the same thing as ioremap in kernel space? === soren_ is now known as soren [10:09] fairuz, nope [10:10] hi, got a new HP Envy and I am seeing some issues. brighness control keys are not working, touch pad is not wokring correctly and I also not sure turbo boost is working as it should (I see scaling up to 1.73ghz instead of up to 2.93ghz). [10:10] apw: ok. So is there a way to get a virt addr for a known phys addr in user space? [10:11] I am currently on the stock 10.10 kernel (2.6.35) [10:11] any recommendations? [10:11] fairuz: i doubt there is an API for that [10:12] fairuz: what you actually trying to do? [10:12] Krunch: just trying to write to a register in userspace [10:12] I want to avoid writing a device driver if possible [10:13] have you looked at ptrace? [10:15] Krunch: I dont think I get what you mean by using ptrace [10:15] the ptrace() system calls allows you to observe and modify processes, there is a flag specifically to change the registers [10:16] or you meant a hardware register or something? [10:17] s/hardware/pci or otherwise not general purpose/ [10:17] yes i mean hardware registers, sorry. [10:18] i doubt you can go and mess the hardware from userland without writing a device driver [10:18] but there are some facilities like libpci [10:18] Krunch: ok. [10:19] You probably can go and mess with (at least PCI, probably more of) the hardware from userspace; X drivers used to be user-space, for example. [10:21] well, in principle you can do pretty much anything you want from userspace by poking around /dev/kmem :) [10:21] do you mean /dev/mem? [10:21] either [10:21] but don't do that [10:22] it's dangerous i assume? [10:22] it's tricky and it will break on every update [10:22] What sort of hardware are you trying to prod, anyway? [10:23] you can see an example of using libpci to change CPU MSR here http://timetobleed.com/enabling-bios-options-on-a-live-server-with-no-rebooting/ [10:23] that's a much saner approach [10:23] and i am confused, the MSR is not changed through libpci [10:24] depend what you are trying to do exactly [10:26] RAOF: it's PL310, a L2 cache controller for ARM [10:28] Funky. No PCI for you! [10:29] RAOF: yes =) [10:29] X did (and still does, I believe) also have userspace I2C drivers, which might be closer to what you're after. [10:30] ok thanks [10:30] If it's too complicated, I'll just write a device driver for that [10:30] easier i guess === ericm_ is now known as ericm [10:58] apw, ba04c7c93bbcb48ce880cf75b6e9dffcd79d4c7b === zul_ is now known as zul [14:25] tgardner, http://www.linuxpromagazine.com/Online/News/Ubuntu-11.10-Named-Oneiric-Ocelot [14:26] * smb wished they would get back to attributes he does not need a dictionary for [14:26] I'll be referring to it as 'O' [14:26] smb :) [14:29] Hm, ok something as dreamlike. Could be good or bad. :-P [14:29] http://who-t.blogspot.com/2011/03/how-to-dos-developer.html [14:35] Hi - bug 714335 was supposedly fixed in the -generic but not -server kernel. Is that possible? Or does rebuilding -server and -virtual kernels require a separate trigger? [14:35] Launchpad bug 714335 in linux "KVM SMP Linux Guests Hang on AMD" [Medium,Confirmed] https://launchpad.net/bugs/714335 [14:38] hallyn, What makes you think that it should be fixed? IIRC that is one thing I had on my things to review still [14:38] smb: just that the bug submitter said it was fixed :) [14:39] smb: great, thanks I'll leave a comment to the effect that it's awaiting review [14:40] hallyn, One never knows what exactly was fixed where. Yes, and I try to get back to that next thing today [14:42] hallyn, I beleieve I started and was wondering about patch #1 which said something about moving a function but the patch looked like adding (probably because of the backport). And before I got to check against the upstream patch I must have been interrupted by something else [14:43] smb: yeah the original commit did the same thing and i didn't want to rock the boat. (it 'moved' a function without removing the original one :) [14:43] smb: thanks. [14:43] hallyn, Ah ok. :) === ericm is now known as ericm-Zzz === diwic is now known as diwic_afk === smoser` is now known as smoser === ivoks_ is now known as ivoks [18:11] <-food [18:20] hi, i was wondering if bug 621265 could get some attention. its had a fix ready for over a month yet still hasn't been backported to maverick. [18:20] Launchpad bug 621265 in linux "Slow Wireless Connection in Intel 3945abg" [High,Confirmed] https://launchpad.net/bugs/621265 [18:22] * jjohansen looks [18:24] LLStarks: generally fixes like this wait on them moving into the upstream stable tree, of course the .35 tree is a bit different story [18:27] even usability issues? [18:28] how do i request an sru? [18:31] LLStarks: you can send a mail to the kernel team mailing list [18:32] smb, around ? [19:33] apw: natty 2.6.38-6.33 is FTBFS (due to toolchain breakage?): https://launchpad.net/ubuntu/+source/linux/2.6.38-6.33/+buildjob/2306132 [19:41] kamal, likely he is out with pgraner and the new folks since they are doing the 'new guy sprint' [19:44] JFo: yup, well apw will surely enjoy knowing that natty kernel is borked when he returns from whatever den of iniquity they're um... "sprinting" at. ;-) [19:56] lo. was wondering how to compile the new kernel on 11.04 ? what packages do you need, ? I have done apt-get build-dep linux [20:13] is there anyone here ? [20:14] yryary [20:16] what packages are required to compile the present kernel ? just that I am getting errors with .size on assembly.. which could be a funny nasm ? [20:16] genux: this is natty? [20:16] yep [20:17] genux: friday we started noticing build failures, it seems to be a tool update issue [20:18] genux: looks like this?: arch/x86/kernel/entry_64.S:1544: Error: .size expression does not evaluate to a constant <--- the toolchain problem [20:18] arh k. do you have any idea on what tool ? is it because of library's 32bit/64bit ? or is it nasm ? [20:18] yep [20:18] atm you can get around this by building in a maverick chroot, but its an issue that will need to be resolved this week [20:19] genux: I'm not sure yet, I didn't pursue it friday, and haven't looked at it today yet either [20:19] k.. I wiped off maverick :(. and installed alpha 3. [20:19] but we do know that it builds fine in a tool chain from about a week ago, and on maverick [20:19] genux: can you install a chroot env [20:20] if you have space its easy, using a kernel team script [20:20] k cool :). shall do that instead :). thanks very much for the help. [20:20] ask a silly question but where would I get the kernel team script ? [20:20] genux: just a sec I am looking that one up [20:20] yeah I have the space.. 120GB for the install [20:20] jjohansen: thanks :) [20:23] jjonansen: I would really like to learn more about the kernel process, is there any advice ? I have kernel device drivers book, but would like to re-write the acpi allocation for devices that *believe* the BIOS is saying where something is, but the BIOS has got it wrong. (a problem on my laptop tbh) [20:23] genux, jjohansen: https://wiki.ubuntu.com/Kernel/Action/BuildChroot [20:24] kamal: thanks, I have the worst time finding things since the wiki was reworked [20:24] jjohansen: me too ;-) [20:24] jjohansen: thanks very much [20:25] genux: as for kernel dev, Linux Device Drivers 3rd edition is a good starting point (its a bit dated, but its free). The new edition of robert loves kernel book is fairly up to date as books go and is good [20:25] kernel newbies is also a good start [20:26] for Ubuntu specific stuff [20:26] https://wiki.ubuntu.com/Kernel/Dev [20:26] yeah.. I have subscribed to the kernel newbies :). [20:26] thanks for the book (I have got the 3rd edition :) ) and shall get the other one :) [20:26] thanks for the ubuntu link. [20:26] thanks very much for the help [20:27] unforntunately after a certain point its just diving into the code and trying to figure out what is going on (especially on linux, documentation is somewhat lacking) [20:28] genux: also the most up to date kernel doc is in the linux kernel it self in the Documentation directory [20:28] genux: np [20:28] tbh.. that is something that really interest me.. I have done c/c++/java/.net/php etc etc. and really want to get to learn the internals of the OS :) thanks very much [20:29] jjohansen: thanks, you have been really helpful :). [20:29] genux: glad to be of some help [20:33] jjonansen: btw the wiki link, there is a scripts directory within the chroot-setup, chroot-setup/scripts/build-mkschroot, should that be a ln within the chrootsetup directory ? [20:36] genux: ?. I am not sure I follow, you can copy the script to your bin dir, or add it to your path element or execute it in place [20:37] so I would normally do ./chroot-setup/build-mkschroot --arch=i386 maverick maverick-i386 http://archive.ubuntu.com/ubuntu [20:37] oh and make sure I am in the directory where I want the chroot to be created [20:37] and then to enter the chroot I use schroot [20:37] schroot -c maverick-i386 [20:38] s/i386/amd64/ if you want to do a 64bit environment [20:38] jjohansen: thanks [20:38] :) [20:38] the chroot is nice in that it carries processor personality so that you can have both i386 and am64 chroots on the same machine [20:38] very cool :) [20:38] ie. build i386 and amd64 kernels and not have to think about it [20:39] thanks very much [21:27] quick question, can i make the squashfs of a maverick live cd from natty? [21:27] i mean will it be compatible> [21:28] we have squashfs-tools 4.1 in natty and 4.0 in maverick [21:31] ( i'm trying to get a fully updated maverick live CD with backports and updates enabled )