[02:23] i think i have spotted a problem with II breaking other releases. is a kernel dev available to discuss? I was at ubuntu+1 but no joy there [02:23] think it is dkms related [02:25] when i installed II on sda2, i got xorg breakage on sda3 gutsy install. nvidia driver would not load on the gutsy partition after reboot. [02:25] if this is the wrong channel to discuss, I apologize [02:26] i have accel driver roking now, not asking for how to fix it [02:26] s/roking/working === Nicke_ is now known as Nicke === ivoks_ is now known as ivoks [10:35] hi guys [10:35] /boot/config-2.6.27-4-generic:CONFIG_SYSFS_DEPRECATED=y [10:35] /boot/config-2.6.27-4-generic:CONFIG_SYSFS_DEPRECATED_V2=y [10:35] how have these options crept back in as "y" ? [10:35] https://wiki.ubuntu.com/DesktopTeam/Specs/HardyHardwareDetection [10:35] we agreed to change them to "n" (and indeed did so) === nm-rocker is now known as asac [12:25] Keybuk: I can't comment on the main kernel but I note these options aren't set as desired in the ports kernel (http://paste.ubuntu.com/54603/) [12:25] I can raise and issue and get them adjusted for the ports too. [12:37] pgraner: ^ [12:37] (I can see you're up - you're replying to e-mail :p) [12:38] Keybuk: bug raised for ports LP: #279019 [12:38] https://bugs.launchpad.net/ubuntu-ps3-port/+bug/279019 [12:57] munckfish, Keybuk: the kernel team is not maintaining anything other than i386/x86_64 so I can't speak to those, as for getting them fixed up for 386/64 not a problem. I'll ping smb_tp since he is maintaining the Hardy tree [12:58] pgraner: I'm working on the PS3 Port Project so I care about and will address the change for the ports kernels and submit a pull request [12:58] munckfish: thanks [13:18] pgraner: I was more referring to the fact those options are _wrong_ in the x86 kernel ;) [13:18] and weren't in hardy [13:18] Keybuk: yep, we will get it fixed [13:19] pgraner: great, thanks [13:19] was tracking down a race condition that got increasingly familiar the closer I got [13:19] "but we *fixed* this! ... check config ... hmm, someone *unfixed* it" :p [13:20] Keybuk: is there an LP ticket for the race condition (just so I can read background info)? [13:21] munckfish: not a currently open one, there's an old one but I forget the # [13:21] it can be simply summarised [13:21] "raw usb devices don't show up at the same time as cooked ones if CONFIG_SYSFS_DEPRECATED=y" [13:21] Keybuk: thx [13:22] current udev isn't even tested with deprecated=y, so there may be others too [13:22] in fact, I think the latest udev upstream won't even start if that config option is set [13:22] Keybuk: ok so it's just much much safer to get that opt out [13:22] understood [13:59] Oct 6 13:59:17 quest kernel: [93700.720031] ata2: EH complete [13:59] what's EH? and why do I need to be notified that it's complete every two seconds? [14:03] Keybuk: looks like you're getting a SCSI layer error, its coming from drivers/ata/libata-eh.c:ata_scsi_error() [14:03] is there a way for me to debug further? [14:03] there are no other messages [14:04] oh, there is another in dmesg [14:04] [93989.536032] ata2: EH pending after 5 tries, giving up [14:04] [93989.536040] ata2: EH complete [14:04] [93991.096042] ata2: EH pending after 5 tries, giving up [14:04] [93991.096050] ata2: EH complete [14:04] Keybuk: EH is a bit cryptic. I assume "Error Handler". [14:05] All of these messages are coming from the same function. [14:06] [ 3.564507] ata2: SATA max UDMA/133 cmd 0x970 ctl 0xb70 bmdma 0xd808 irq 23 [14:07] I'm guessing that's sata_nv, though no dmesg makes a direct association [14:07] Keybuk: lspci show the device as sata_nv? [14:07] [ 3.558356] pata_acpi 0000:00:07.0: PCI INT A -> Link[APSI] -> GSI 23 (level, [14:07] low) -> IRQ 23 [14:07] [ 3.562060] sata_nv 0000:00:07.0: PCI INT A -> Link[APSI] -> GSI 23 (level, l [14:07] ow) -> IRQ 23 [14:07] rtg: the board is sata_nv yeah [14:07] funny thing is [14:08] I don't think there's anything actually on ata2 [14:08] [ 4.036232] ata1.00: ATA-7: SAMSUNG SP2504C, VT100-33, max UDMA7 [14:08] [ 4.837197] ata3.00: ATA-6: WDC WD740GD-00FLC0, 33.08F33, max UDMA/133 [14:08] [ 5.329197] ata4.00: ATA-6: WDC WD740GD-00FLC0, 33.08F33, max UDMA/133 [14:08] Keybuk: I wonder if something is trying to probe ata2 periodically? [14:09] [ 6.184293] ata5.00: ATAPI: _NEC DVD_RW ND-3550A, 1.05, max UDMA/33 [14:09] [ 6.184304] ata5.01: ATAPI: LITE-ON COMBO SOHC-5236V, R$0A, max UDMA/44 [14:09] heh [14:09] it's not even either DVD drive [14:09] rtg: how could I tell? [14:09] Keybuk: good question. I'm not at all familiar with the ATA library. [14:10] Top causes for wakeups: [14:10] 43.9% (101.1) : sata_nv, eth0 [14:10] wouldn't have thought anything from user space would be probing [14:10] in fact [14:11] I don't even think there's an interface for anything from user space *to* probe [14:11] No udev rules? === ivoks_ is now known as ivoks [14:11] /sys/devices/pci0000:00/0000:00:07.0/host1 has no target [14:11] just power/wakeup [14:11] (and the link to the scsi host) [14:12] rtg: udev only does things to devices which have a /dev entry [14:12] Keybuk: Garzik is the maintainer. maybe you should look on bugzilla and see if anyone else is having similar sata_nv issues. [14:12] there's no device on that ATA port, so there's nothing in /dev :p [14:15] rtg: http://bugzilla.kernel.org/show_bug.cgi?id=9010 maybe? [14:15] not sata_nv tho [14:15] quite different though [14:15] nothing exactly matches [14:17] Keybuk: and its a pretty old issue. It started against a 2.6.23 kernel. [14:20] yeah [14:20] I've only seen this since upgrading to Intrepid from Hardy [14:21] Keybuk: I assume you're running 2.6.27-5 ? Its an (-rc8) based kernel, so its worth reporting upstream. [14:30] 27-4 ? [14:30] when did 27-5 come out? [14:33] Keybuk: iver the weekend, uploaded Friday. === lamont` is now known as lamont [16:47] Keybuk: I think EH is error handler, but it usually means you have medium or hardware issues [16:48] oop, I'm late on that convo [16:48] * BenC kicks pidgin's scroll [16:49] BenC: right, but how do you figure out what error? In Keybuk's case, there is nothing attached to that port. [16:51] I surely do hate nVidia resume issues [16:51] wow. do you actually attempt to debug them? [16:52] well, somewhat feebly [16:54] maybe the error is it keep trying to probe something that is not there [16:54] or the card is producing errors that we should just ignore [16:55] BenC: uh, I think thats exactly the error (continuously probing when nothing is there) [16:59] [322544.947660] ------------[ cut here ]------------ http://dpaste.com/82721/ [16:59] .27.4 - should I post it to lp? [17:01] CarlFK: which kernel version? [17:01] I mean, which -rc? [17:02] rtg: what is -rc? [17:03] release candidate [17:04] CarlFK: better yet, 'cat /proc/version_signature' [17:04] i installed u-server a week ago, did apt-get upgrade a few days ago [17:04] just a sec.. need to power the box back up [17:14] rtg: Ubuntu 2.6.27-4.6-server [17:15] CarlFK: update to 2.6.27-5 and see if it makes a difference. [17:16] rtg: im not really sure what caused it - guessing it had to do with powering off a mounted usb drive [17:17] sounds like there is no need to post what I have from the .4 kernel, right? [17:18] CarlFK: not unless its repeatable on a clean drive [17:19] what do you mean "clean drive" ? [17:19] CarlFK: one that has been fsck'd after being improperly shut down. [17:22] rtg: good news! [17:22] found it? [17:22] no errors with -5 [17:22] works for me :) [17:24] quest scott% dmesg | grep ata2 [17:24] [ 4.705119] ata2: SATA max UDMA/133 cmd 0x970 ctl 0xb70 bmdma 0xd808 irq 21 [17:24] [ 5.516028] ata2: SATA link down (SStatus 0 SControl 300) [17:25] interestingly, I don't see that "SATA link down" in the previous dmesg [17:25] so maybe that was the bug - sata_nv link detection was broken [18:06] whats the best application to configure and compile a costume kernel ? [18:07] freeRag: there is a wiki page on that [18:08] * ogra waves [18:08] have you read it ? [18:08] yes. [18:08] hey ogra [18:08] i have a strange prob with ath5k vs ath_hal [18:08] so you can just tell me [18:08] freeRag: there is no "application", it's manual for the most part [18:08] freeRag: no [18:08] freeRag: look at the topic for this channel [18:09] i have adevice for which ath5k doesnt work, ath_hal does, so i add a blacklist file for ath5k on this device ... during suspend/resume ath5k goets loaded despite being blacklisted [18:09] *gets loaded [18:09] ogra: rm -f ? :) [18:09] how does the unbuntu kernel differs from the linux kernel ? [18:09] BenC, from livecd-rootfs during image build ? [18:10] BenC, thats on the samsung Q1U which is our reference device for ubuntu-mobile :) [18:10] ogra: hmm, got me on that one [18:11] if i add ath_pci to SUSPEND_MODULES in pm-utils it gets loded *after* ath5k and works [18:11] shouldnt the kernel be compile specifically for each machine its going to be run from ? [18:11] freeRag: best start with the KernelTeam section of the wiki, and look through the git source repos you'll find answers to a lot of your initial questions there [18:11] freeRag: No [18:11] BenC, my thinking is that pm-utils should be taught about module blacklists [18:11] freeRag: that's why we have modules [18:11] so it doesnt try loading ath5k at all [18:11] ogra: it should be calling modprobe with -b [18:12] ogra: or it should only be loading modules that it specifically unloaded [18:12] hey guys, I'm getting some interesting looking oopses on resume on my macbook3,1 [18:12] BenC, well, the thing is that you dont need to unload ath_pci [18:12] but ive read about some duds that manage to considerably shorten boot times using especially compiled kernels with the driver built in to it [18:13] looks like http://paste.ubuntu.com/54697/ [18:13] drivers* [18:13] BenC, but ath5k gets loaded on resume nontheless [18:13] repeated for every process after resuming from suspend [18:13] so -b is it then [18:13] BenC, thats the original function :) http://paste.ubuntu.com/54700/ [18:13] so it doesnt respect blacklists it seems [18:14] freeRag: it is true that doing so may shorten boot times, but then we, as a distributor, cannot compile a kernel for every user [18:15] freeRag: if you want a custom kernel, check the wiki, but please don't ask us any more than that...you are pretty much on your own (as the wiki will warn) [18:15] thats why i say that the kernel should be compiled at install with built in drivers [18:15] automatically [18:15] sig, why oh why dos it need to use quilt [18:15] * ogra curses [18:15] freeRag: Sure, so let's add 1.5+ hours to the install to build a custom kernel [18:15] freeRag: and then 1.5+ hours to every security update to recompile it [18:16] freeRag: and 600+megs of used space for source+binaries to do so [18:16] freeRag: and 55Meg+ downloads for the updated source (as opposed to 25megs for a pre-compiled kernel) [18:17] i thought open source was all about compiling your own [18:17] freeRag: Go use gentoo if that's what you like [18:17] freeRag: you would be very wrong [18:17] freeRag: If you feel that way, compile openoffice/gcc/glibc/kernel/gnome/xorg a few times and see how much you enjoy it :) [18:18] im talking about the kernel mostly has its what makes the computer ticks [18:19] and the hardware drivers [18:19] wont such a system run faster ? [18:19] freeRag: you are making a very bad argument...we can't expect the kernel to be built at install time on every system [18:20] especially since there is never a case where a system will _always_ have all the hardware that it will ever use connected at that point [18:20] why should i have SMP if i have a single core ? [18:21] freeRag: You do know that the kernels rewrite themselves as UP code at boot for non-SMP systems, right? [18:21] i dont [18:21] im asking to find out [18:21] binary patching to nop most of the SMP related locking and such [18:21] freeRag: go read [18:23] freeRag: plenty has been written about this. best to do a bit of homework on your own before talking to others (especially after being asked to) [18:23] so what you are saying if there no real performance gains to be achieved? [18:24] freeRag: No, there may be measurable gains at a cost that we can't justify [18:25] freeRag: If you want to shave 5 seconds from boot, or get 2% better performance, feel free to do so [18:25] freeRag: I will tell you that reducing boot time is of major concern for us, and will be a big focus at our next developer summit [18:25] so the performace gains are nearlly negletable ? [18:26] what if i use the computer for very intensive graphical computations ? [18:26] * BenC *sigh* [18:26] freeRag: then you should read. === thegodfather is now known as fabbione [18:27] freeRag: or not read and use the computer for very intensive graphical computations. [18:28] notice that neither of those start with "keep asking questions here" :) [18:28] fabbione: dear godfather...tell me why we write docs for people who refuse to read them...kthxbye [18:28] you mean I shouldn't spend 20 hours recompiling my system and subjecting myself to increased chances of compiler/configuration bugs in order to save 3 minutes on a 4 day video encode? [18:28] BenC: because we have to look good :) [18:28] *gasp* [18:29] * freeRag slaps CarlFK with a 5 pounds *nix manual [18:29] :D [18:29] we need no stinking manuals [18:30] BenC: there was something i had to tell you and it's totally out of my mind now... [18:30] fabbione: you're out of your mind? [18:30] * BenC thought that was common knowledge [18:30] BenC: what's new about it? [18:30] BenC, is the -b in modprobe upstream behavor or ubuntu/debian specific ? [18:30] ogra: upstream [18:30] * ogra would like to file a bug about pm-utils upstream [18:30] ah, cool [18:31] i always use text to speak for reading the manuals ! mostly when i m having troubles sleeping [18:31] ogra: it's used in the initramfs scripts to honor the blacklists as well [18:31] ah, cool [18:32] anyway, I filed bug 279196; I'm not sure what other degugging info is relevant for oops on resume? [18:33] is ubotu dead again? [18:37] I don't think so, it's just not in this channel. [18:38] Would this be the proper place to discuss the newly-re-included e1000e driver? Or would ubuntu or ubuntu-testing be a better venue? [18:58] should i ask a question or should i let the chat die out ? [19:04] i v read kernel mode-setting is going to enable BSD on linux :D [19:04] DrDabbles: depends on what you want to discuss [19:05] thats something that should make the windows to linux converts happy [19:05] http://www.phoronix.com/scan.php?page=article&item=kernel_modesetting&num=1 [19:09] BenC: any word on this feature being available for the nest ubuntu release ? [19:09] next* [19:11] freeRag: it's terribly unstable when I tested the fedora beta [19:11] it makes the already panic-prone intel driver more prone to hanging and blowing up [19:12] i think this articles could use some glossary because im not very familiar with some of the soudbits used [19:13] I think the radeon modesetting driver is actually more stable than the intel one [19:14] is Gallium3D on ubuntu ? [19:14] TTM memory manager ? [19:15] gallium is an experimental new framework for 3d acceleration [19:15] it's not in any production distribution [19:15] it's not close to finished yet. [19:15] Multi-Pointer X - must be multi mouse support on the same desktop [19:17] would you say that linux development has accelerated exponentially in the last couple of years ? [19:17] what on earth does that even mean? [19:17] jdong: you've won buzzword bingo [19:18] I guess it depends on whether or not the ancestry is continuously differentiable in a logarithmic domain [19:18] laga_: there we go, did I top it? [19:18] awesome. [19:22] jdong: likes that supposed to mean something ?! [19:22] freeRag: what you said didn't really mean anything either. [19:22] yes it did [19:23] and I do believe that is a prerequired condition to establish proof of exponential growth :) [19:23] maybe I slept through too many math lectures though [19:23] i hate maths [19:23] freeRag: what is Linux development and how can it be numerically quantified? [19:24] and how do you find the second moment of linux development to quanitify its acceleration? :) [19:24] i hate everything that isn't in human able reading format [19:24] then maybe it'd be a better idea to stop reading ricer journalism and trying to tell kernel developers what to do? [19:25] the 1st moment would be the previous 15 years [19:25] "ricer journalism". hah. [19:25] laga_: haha you've got a better term for phoronix? :) [19:26] whats ricer journalism ? [19:26] no. [19:26] something out of asia ? [19:28] can ubuntu support 2 users on the same computer using different accounts and different usb keyboards and mice's ? [19:28] and 2 monitors [19:28] that's completely off topic to this channel [19:28] but if you try hard enough, yes. [19:30] thats a cool feature windows cant do [19:30] freeRag: google "multiseat x" [19:31] since when couldn't Windows do that? :) [19:31] can it ? [19:31] terminal services. [19:31] terminal services ? [19:32] isnt that like VNC ? [19:32] you mean terminal like in a command line terminal ? [19:33] thats not what i meant [19:33] not over network [19:34] not VNC [19:34] im meant using the same box [19:36] jdong: seems pron to miss interpret my statements [19:37] just admit windows can do that and linux can ! [19:37] just admit windows canT do that and linux can ! [19:38] i hear windows is better? [19:39] in what aspect ? [19:40] DRM implementation ? [19:40] BenC I would like to discuss an issue where my link keeps dropping. I realize this is probably a bug that should be filed and possibly pushed upstream to kernel maintainers or Intel themselves, but I wanted to see if it was local to my card or if others had the same issue. [19:41] DrDabbles: you sure it's not hardware (cable, switch)? [19:42] BenC I am using the same switch and cable now with an add-on NIC, and I had previously used both cable and switch with the onboard Intel NIC before the driver was initially disabled. [19:44] DrDabbles: hmm...well, the link dropping is generated by the phy, so it's not something that the driver would generate [19:44] DrDabbles: I would lean toward the actual port being flakey, but hard to tell [19:44] DrDabbles: best to talk to Intel [19:44] BenC I'm wondering if my EEPROM didn't get a little bit broken, and it's causing the card to reset itself repeatedly. Ah well. Thanks! [19:44] what intel gives MS takes :) [19:46] DrDabbles: e1000e might trash your kernel upgrafe :D [19:47] or is it the other way around ?!!? [19:47] time to wonder if intel isnt poluting the kernel with fake drivers ? [19:47] freeRag e1000e driver was disabled in Ubuntu when it was understood that there was a problem with the kernel driver causing the EEPROM to be corrupted or overwritten. [19:47] would anyone be to upset if someone were asked to leave? [19:48] Intel has since taken it upon themselves to dedicate internal engineers to finding the root cause of the bug (it seems to be MMIO at this point), squash it, and even help fix any bricked NICs that may exist due to this issue. [19:49] thats clarified them thanks [19:49] It is only something that people using RC kernels would have experienced, and the common idea is that only people willing to break hardware or data should be running testing code. [19:49] then* [19:50] CarlFK: i wouldn't [19:50] lots of trolls on freenode tonight it seems [19:51] CarlFK: Fiiine. I'll go. ~plays sad music~ :-) [19:52] hey... I didn't say who, or even when... :) [19:52] i use the forcedeth network card driver but it didn't killed my card :-) [19:55] freeRag: umm, terminal services is *NOT* a VNC clone [19:55] trolls ? were ? lets raid them [19:55] freeRag: it is an abstraction of the hardware into virtual systems [19:56] freeRag: using it as a remote desktop client is just one of its many capabilities [19:56] freeRag: in fact, remote desktopping with TS is attaching a different controlling display adapter and input device to the TS [19:56] TS ? [19:57] terminal services. [19:57] terminal service [19:57] http://en.wikipedia.org/wiki/Terminal_services#Architecture [19:57] im already there [19:57] but it has to much to read [19:58] and no peformance numbers or CFLAGS. [19:58] mostlly about networking [19:58] what? the section I linked has NOTHING about networking. [19:58] it's about how TS allows redirection of IO devices for multiple clients [19:59] # 4 Remote Desktop Connection [19:59] # 5 RemoteApp [19:59] # 6 Windows Desktop Sharing [19:59] that sounds like networking to me [19:59] those are the common applications of it [20:00] (Terminal Server) listens on TCP port 3389 [20:00] sounds like networking too [20:00] dude you have some reading comprehension difficulties. [20:00] I asked you to read up on the IO redirection capabilities [20:00] LOL [20:00] BenC: how do you surgically remove d-i from the debian directory since I dont want it [20:00] which is what directly pertains to your original question about multipointer/multioutput X [20:01] multi seat ? [20:02] jdong: I am going to start yelling at you, only because I think you will listen :) [20:02] if i understand all this is using the network ? [20:03] even if the keyboard and monitor a fiscally connect to the same box [20:04] freeRag: no, there is no network involved. [20:04] how come if it uses tcp ? [20:04] the multiseat packages were dropped though [20:04] freeRag: it uses TCP for remote clients that want in. [20:04] TCP doesn't mean it's automatically a remote networked service [20:05] i.e. the entire X client-server framework listens on UNIX sockets and UDP sockets [20:05] jdong: you are talking about the windows TS arent you ? [20:05] yes? [20:06] in fact I believe there is a KVM+software package for Windows that leverages this capability out of the box, without you having to use TS-magic [20:07] this ts thing is only available in windows server ? [20:08] officially and EULA-wise yes. [20:08] damn [20:08] http://en.wikipedia.org/wiki/Multiseat [20:08] im usin xp [20:09] only noticed now LOL [20:10] alright, what'd y'all developers do? The moment I reported the kernel oops I can't reproduce it again. [20:10] * jdong files Ubuntu consipiracy theory to Digg [20:14] jdong: speaking of TS, is there a linux TS Client that does cut/paste with the local X clipboard ? [20:17] use ltsp :) [21:54] is there a way to make a certain kernel be the first in grub's list? (even if new -generic kernel updates get installed lateron ?) [21:55] alex_joni: how about making a certain kernel the default ? [21:55] ah, I'm not interested in a manual - aka fix my own machine - way by editing /boot/grub/menu.lst, was wondering if there is something else [21:56] CarlFK: I'm looking for a script or soemthing I can call from a postinst or such [21:58] want a sed line that will set the default? [21:59] CarlFK: I always appreciate all I get :) [22:00] gimme a sec to boot my lappy, and hope it is still there [22:05] sed "/^default/s/^.*$/default $1/" -i menu.lst [22:06] hmm.. but I don't think that will work.. let me explain [22:06] also, you can do "default saved" and I have seen the code that will set what is saved [22:06] we have a certain kernel (2.6.24-16-rtai) [22:07] and it gets installed at a certain point of time by the user [22:07] beeing the newest kernel it's first in the list and all is well [22:07] after a while a -generic kernel gets updated [22:07] when that happens (say 2.6.24-19-generic is out), that one gets put ontop of the list, and the others shifted [22:08] ## should update-grub adjust the value of the default booted system [22:08] guessing you want to set to flip that [22:09] er... set/flip/use... study :) [22:11] updatedefaultentry ? [22:12] yup [23:17] could we get the elantech touchpad drive in intrepid please? === Ng_ is now known as Ng