[12:01] (I have a hand-hacked xorg.conf, so dexconf wouldn't have rewritten it automatically) [12:01] Though my cursor seems to have gone back to the old-style one [12:02] Uh. It just started x-session-manager [12:04] mjg59: you're the lead on laptop stuff, right? [12:04] do a dpkg-reconfigure [12:04] schweeb: Yeah [12:04] jdub: On what? [12:04] or switch to the right fonts dirs [12:04] xserver-xorg [12:04] jdub: I just added the symlink [12:04] It breaks upgrades at the moment [12:04] mjg59: heard of anyone with an IBM X41 yet? just got one like 2 days ago, about to start testing breezy on it [12:05] schweeb : Is that the convertible tablet [12:05] ? [12:05] no, just the laptop [12:05] schweeb: Not yet, but as far as I know it's just an X40 with i915 [12:05] k [12:05] infinity: There's a tablet version, but not all of them are tablets [12:05] ogra: About? [12:05] tablet was finally released like the day after I ordered this one [12:06] my sound doesn't work yet, but I'm assuming that's the same IRQ problem I had with my Dell [12:06] mjg59, yep [12:06] Uh. On Breezy? [12:06] There should be no IRQ issues [12:06] no, hoary at the moment [12:06] Is snd_intel8x0 getting loaded? [12:06] yep [12:06] ogra: We need to tidy up the power preferences [12:06] schweeb: 2.6.10? [12:07] yep, latest ubuntu kernel pkg [12:07] mjg59, lets wait til next week, havent seen the new preferences window yet [12:07] ogra: Ok, no problem [12:07] ogra: Oh, can you get it added to the default session? [12:07] (Check if it's installed, and if so run it) [12:07] mjg59: on the Dell, the parport and intel8x0 were trying to unsuccessfully use the same IRQ... I always had to blacklist the parport [12:07] schweeb: That shouldn't have happened post warty [12:08] i think it should work like update-manager, say the user has to run it once... [12:08] Hmph [12:08] hrm [12:08] ogra: Hmm. I'm inclined to think it should always be running (much like the networkmanager back end) [12:08] Otherwise people won't get low battery notifications by default [12:09] /proc/interrupts doesn't show that the parport or the snd driver even have an interrupt... === schweeb investigates further [12:09] schweeb: Check dmesg for signs that the sound driver actually bound to the device at all [12:09] schweeb: The other possibility is that we're missing PCI IDs [12:10] yea, this does have some odd hardware for a laptop [12:10] SATA, PCI Express [12:10] mjg59, i'll try to get the backend started by default... [12:10] haven't tried bluetooth yet either [12:10] ogra: Cool [12:10] schweeb: Oh, it's SATA? Right, you're not going to get suspend/resume [12:10] but my wifi works great, that's all that matters at this exact moment :p [12:10] mjg59, but network manager's backend is a system service [12:10] mjg59: oh, boo [12:11] ogra: Yeah, but that doesn't matter to the user :) === r0bby [~wakawaka@cpe-204-210-133-115.hvc.res.rr.com] has joined #ubuntu-devel [12:11] mjg59, nope, but for the implementation ;) [12:11] thom: networkmanager hasn't crashed on me yet! [12:11] mjg59: that makes me weep, wish I would have known that one (hell, I didn't know it was SATA until I installed ubuntu) [12:12] sata lappy? nice! (mostly) [12:12] so, that's a SATA wide issue? [12:13] jdub, sata isn't useful, really. :P [12:13] Only thing its good for is reducing cable clutter. [12:13] i don't care about useful [12:13] i care about SPEED [12:13] HWolf: especially considering this has a 1.8" 4200 RPM drive :( === jdub revs his very old athlon. [12:14] but it's so damn light [12:14] and his server with no lid on it. [12:14] lol [12:14] both of which his laptop outperforms. :| [12:14] ide wouldn't limit even a decent desktop drive, let alone a laptop drive [12:15] mjg59: I could get you lspci output if you'd like, just so you guys can get PCI IDs [12:15] it'll show no increase in speed whatsoever, even in a synthetic benchmark. [12:15] I already submitted to hwdb === infinity wonders if his new T43 will be SATA.... [12:15] The specs claimes ATA133, but the specs often lie. [12:15] Guess we'll see when it gets here. [12:15] But now I'm off, before I get kicked out for being OT, or I drop my sleepy head on my keyboard. :) [12:16] schweeb: Yeah, that'd be good [12:16] any specific options? -v, or anything further than that? [12:16] schweeb: SATA support for suspend/resume ought to appear in Breezy soon [12:16] lspci -n [12:17] email or pm? [12:17] anyone here have sip love? === Seveas [~seveas@seveas.demon.nl] has joined #ubuntu-devel [12:18] i'm bored of calling echo tests and random red hat pstn numbers [12:18] jdub: whiprush would probably be more than happy to be a test subject [12:18] jdub, call anthony [12:19] ogra: it's 8am on sunday morning [12:19] ogra: i don't believe he's seen one of those since he got up to watch church on tv in his younger days [12:19] hehe, great, the right time.... he will love you :) [12:19] jdub: go to sleep then! [12:19] i just got up! === scorpix_ [~scorpix@as17-166.qualitynet.net] has joined #ubuntu-devel [12:19] why [12:20] i'm now on EXTREME .au time [12:20] which means early to bed, early to rise [12:20] you have like 6 more hours of sleep ahead ofy ou [12:22] mjg59: what's your email, or should I email them to ubuntu-devel list [12:24] schweeb: mjg59@srcf.ucam.org [12:26] sent === Burgundavia [~corey@S0106000000cc07fc.gv.shawcable.net] has joined #ubuntu-devel [12:38] mjg59: heh, cool [12:40] thom, have you heard any reports of network manager writing "nameserver 127.0.0.1" to a /etc/resolv.conf ? [12:40] Burgundavia: uh, that's what it's supposed to do? [12:40] so, yes, i've heard that it does that :P [12:40] thom, it also broke networking on my machine, when rebooted into recovery mode [12:42] how? [12:42] the machine cannot contact the outsite world [12:42] through DNS that it [12:42] s/it/is [12:42] thom: no name resolution... [12:42] no bind [12:43] ok, so that's _not_ breaking networking [12:43] please don't confuse the two [12:43] seems that something else is looking for /etc/resolv.conf to be a simlink [12:43] jdub: nod, working on it [12:43] symlink even [12:43] Burgundavia: resolvconf, yes [12:43] thom, for me dhcp doesnt work [12:43] thom: so will /e/n/interfaces basically become irrelevant? [12:44] jdub: well, no [12:44] jdub: NM reads it and uses it in the case of static config [12:44] ahr [12:44] long term we'll do more integration [12:45] ogra: doesn't work is a big and unhelpful phrase. does it work normally? is this on your wifi card? does the rest of the networking stuff get set up correctly? [12:45] its a orinoco silver pcmcia card, if i ifdown/up the interface, it works normally... [12:45] thom: have you looked at the zeroconf package? [12:46] what i need to do is figure out how to make resolvconf and NM play in such a way as to not break in the case where NM isn't running [12:46] jdub: NM does zeroconf [12:46] or it should [12:46] oh, ok [12:46] ogra: is this the one that hal doesn't know is a wifi card? [12:46] yeps [12:46] you're doomed then; kernel bug [12:46] argh [12:47] Well, lack of kernel functionality [12:47] (it needs to export the correct info in sysfs) [12:47] mjg59: well, old driver, yeah [12:47] thom: The main problem is that pcmcia cards don't seem to play nicely with the sysfs stuff [12:47] cardbus is fine [12:47] thom: what do you get out of using resolvconf? [12:47] oh, just old school pcmcia [12:48] jdub: i'm hoping that i can get it to the point of being able to get NM to use resolvconf to generate a resolv.conf, so if NM isn't running you still get a reasonable resolv.conf === Goshawk [~Goshawk@host42-45.pool8251.interbusiness.it] has joined #ubuntu-devel [12:49] hi [12:49] thom: the server case, basically? [12:50] and the recovery mode [12:50] that is where nm bit me [12:50] Burgundavia: that's avoidable in other ways (ie. don't even try) [12:51] jdub: server case isn't really interesting, you won't ever have NM installed [12:51] thom: have you drawn little workflow diagrams of this? [12:51] jdub: but being able to play nice with the system is definitely a feature [12:51] jdub: no [12:51] not yet, anyway [12:51] my brain is scrawling little workflows on the inside of my skull [12:52] this may be a gnome bug: afther you have created a parent folder and many child folders, with files inside, and folders without any file, and click the right mouse button on the parent folder and choose "Create Archive" in format tar.gz, the archive is created but all the empty folders are missed... should it be a bug? should i report it to reportbug? [12:52] Goshawk: bugzilla.gnome.org is probably a better place (and yeah, sounds like a good bug to me) [12:53] jdub, thanks :D [12:53] my ears are ringing too loud from U2 for my brain to work right now ;-) [12:53] thom: so in the n-m exists case, what happens when you invoke ifup or ifdown? [12:53] jdub: at the moment, shitfight central [12:53] awesome [12:54] do you want to make it so they send appropriate d-bus messages to n-m, and ignore everything else? [12:55] hmm [12:55] jdub: i'm unsure as yet. Thomas Hood is interested in working on this too, need to talk to him at somepoint again [12:55] thom: man i love the moment after n-m realizes it doesnt have total control over a network device [12:55] jtdhood? [12:55] jdub: yeah [12:55] cool [12:56] tseng: heh [12:56] man, this is hard stuff [12:56] tseng: i thought "shitfight" covered it quite accurately :-) [12:56] yes [12:56] i wonder if i put nm-applet in my session if ill get properly associated at login [12:56] (brb) === eazel7 [~ezequiel@host152.201-252-82.telecom.net.ar] has joined #ubuntu-devel [12:56] hi ppl === chuck_ [~chuck@CPE0006258ec6c2-CM000a73655d0e.cpe.net.cable.rogers.com] has joined #ubuntu-devel [12:57] thom: what if n-m became a type of /e/n/i configurations? [12:57] so you had managed interfaces and unmanaged interfaces === bskahan [~bskahan@pool-70-19-41-225.ny325.east.verizon.net] has joined #ubuntu-devel [12:58] jdub: hrrrrm === thom writes tomboy notes [12:58] then ifup on a managed interface would tell n-m to do its thing (via dbus, i imagine) [12:58] ifup on an unmanaged interface would just do the right thing [12:58] as per configuration === eazel7 [~ezequiel@host152.201-252-82.telecom.net.ar] has left #ubuntu-devel ["Abandonando"] [12:58] and n-m would *never* try to manipulate an unmanaged interface === eazel7 [~ezequiel@host152.201-252-82.telecom.net.ar] has joined #ubuntu-devel [12:58] like, it would completely ignore them [12:58] hrm === eazel7 [~ezequiel@host152.201-252-82.telecom.net.ar] has left #ubuntu-devel ["Abandonando"] [12:58] though you'd want n-m to own anything that wasn't specified [12:59] yeah [12:59] jdub: would if* work properly in the absense of nm? [12:59] i wonder if you could use a hotplug mapping script to do that [12:59] tseng: you mean, would a managed interface work in the absence of nm? [12:59] jdub: sure [12:59] hmm [01:00] there's no reason for it not to, is there? [01:00] thom: what would it do? [01:00] it's like [01:00] if you told people they need dbus and nm on a server, then you would find out what shitfit central is [01:00] ifup eth3 when you don't have an eth3 definition in interfaces [01:00] tseng: yes [01:00] tseng: well, in that case, you just have all unmanaged interfaces as per normal [01:00] hm right [01:00] what i'm thinking of is this (which thom groks, i think): [01:00] i see how you mean [01:01] iface eth1 inet managed [01:01] ^ in /etc/network/interfaces [01:01] oh [01:01] jdub: not really; the interface is defined, all you're really doing is notifying NM that you want it to be available; there's no reason that you couldn't also have fallback config i think [01:01] heh, you could even set mandatory (unchangeable) settings in the stanza :) [01:01] that too [01:02] thom: so if i did ifup eth1 when nm wasn't running, what do you see happening? (when the only configuration is that one line) === womble [~mpalmer@eth1859.nsw.adsl.internode.on.net] has joined #ubuntu-devel [01:02] jdub: more like: "iface eth1 inet dhcp \n mode managed"; don't forget that NM can use static configs because it can do the ifplugd case of bringing an iface up when the it gets a connection [01:03] but then you've got a mandatory setting of dhcp, right? :) [01:04] hrm [01:04] yes, and that's fine [01:04] what if you have a managed interface without a dhcp or static mandatory setting? [01:04] in the scheme i just proposed, you can't [01:05] however, in the scheme i proposed, you couldn't have a mandatory setting for that at all [01:05] unless you had another stanza item [01:05] i can't see how you'd usefully use that, unless it just zeroconfs the iface [01:05] NM still has to know how it should bring the iface up [01:06] it works that out itself, doesn't it? [01:06] no? it defaults to dhcp but parses /e/n/i for any clues [01:06] that static/dhcp bit is 'how to do it' and nm is a 'how to do it' [01:07] no; think of /e/n/i as a policy statement, NM just implements that policy [01:07] so static configurations are not stored by NM? === thom thinks he's characterising this right [01:07] not afaik [01:07] could be wrong though [01:08] aha, so [01:08] then you could use the same thing [01:08] iface eth1 inet managed [01:08] and all of the fields are optional [01:08] *if* you set address, it doesn't dhcp [01:09] if you set any of the others, they're overrides or something [01:09] does NM interpret maps in interfaces? [01:09] hrm, see what you mean [01:10] jdub: Unless you manually override in the interface at some point [01:10] mjg59: hrm? [01:11] jdub: I ought to be able to override anything in /e/n/i through the UI at runtime [01:11] mjg59: yeah, though we were just talking about mandatory settings too [01:11] mjg59: perhaps there should just be another field 'mandatory yes' or something to do that [01:12] Yes, it needs some amount of lock-down support [01:12] but you're right, that should be optional === jdub nudges fascist hat further away [01:13] does NM start really early? [01:13] it starts with dbus [01:13] oh yeah [01:14] 20 in rc2 [01:14] not, actually, early enough [01:14] mmm [01:14] i think we should get dbus up ASAP [01:14] like, in the initrd or so... ;-) [01:15] pondering if ifup eth1 would invoke NM if it wasn't running, or just talk to it (and NM would kick all the managed interfaces off when it starts) [01:15] haha === Duck_away is now known as Duck_busy [01:17] thom: in en_GB, 'tomboy' should be 'thomboy' ;-) [01:18] piss off :-) === jdub writes a letter to orph. [01:18] jdub: it's not like it's a usual spelling here, either [01:19] jdub: beat him senseless until he replaces the shithouse icons [01:19] jdub: i mentioned i replaced it and he started cursing [01:21] thom: Eh? [01:21] jbailey: initrd is a highlight these days too? [01:22] thom: If you're serious, please look at the initramfs hooks. There won't be glibc in there, so I'll need to be able to compile it with klibc or uclibc. [01:22] thom: Yeah. =) [01:22] jbailey: no, i'm not serious. but it does need to be early [01:22] (i'm not sure it'd build with klibc, anyway) [01:22] thom: Oh good. =) I was growing this deep fear that perhaps usplash might need it for something. [01:23] heh [01:23] usplash is not me, but it might well [01:23] tseng: seriously? i thought they changed upstream too? [01:23] jdub: no way [01:23] we could have udbus ;-) [01:23] the stuff I did to date is just statically built and striped. comes to < 30kB [01:24] jdub: its still the dumbass comic guy from 1940 [01:24] sladen: So, when are we getting more of it? [01:24] don't call tin tin a dumbass [01:24] sladen: Sounds lovely. Did you get a chance to look at the intiramfs hooks? [01:24] you'll offend seb [01:24] ha ha [01:24] jdub: we should send orph a profile shot of seb [01:24] YES [01:24] sladen: 0.11 is quite well baked for raid1, hd boot and nfsroot. [01:24] it'd be *so* cool [01:24] mjg59: ah, yes. urm. [01:24] thom: a seb hackergotchi, which i don't have... [01:24] that should be april fools next year [01:24] tomboy -> seb [01:25] the ubuntu hackergotchi icon set? [01:25] sladen: You should be able to install initramfs-tools, add one line to your kernel-img.conf: "ramdisk = /usr/sbin/mkinitramfs" and reinstall your kernel and have it Just Work(tm) [01:25] that'd be terrifying! [01:25] thom: yes! [01:25] replace tseng with a Tub of Lard [01:25] Phear the April 1 Ubuntugotchi set. [01:25] (are you calling me fat?) [01:26] jbailey: now? [01:26] The following extra packages will be installed: [01:26] busybox-cvs-initramfs klibc-utils libklibc [01:26] ouch ;) [01:26] jdub: No, April 1 was a little while ago. [01:26] tseng: it's the clasic thing that Have I Got News FFor You do when one of their guests does a no-show [01:26] jbailey: initramfs, silly :) [01:26] Oh, I see. =) [01:26] Need to get 518kB of archives. [01:26] After unpacking 1372kB of additional disk space will be used. [01:26] sladen: ah. [01:26] boh ;) [01:27] Yes. mdz is using it already and I'm booting one of my systems on it all the time. [01:27] jdub: Doesn't do lvm or evms. === jdub does it [01:27] Tested on i386 and ppc64 [01:27] nice [01:27] Don't mind the 9mb initrd, that will get smaller. =) [01:28] Right now you have a pile of static binaries, glibc, klibc and about 30 megs of drivers packed into there. =) [01:28] ouch [01:28] is there a nice initramfs boot process doc around? === sladen grabs initramfs [01:28] I think that needs to be about 6 megs worth of drivers, and about half a meg worth of code. [01:28] jdub: You're talking kernel side or my package? [01:29] both [01:29] 30MB initrd?! [01:29] sladen: Nonono, 9mb, but that's compressed. [01:29] Much need for shrinkage. [01:29] My smallest test case was 250k [01:29] sladen: dude you need the whole kernel source tree in there for good luck [01:29] And that could still be reduced further. [01:29] jdub: No, for my package, but I can give you the door through it in about 2 minutes. [01:30] I've been waiting for usplash to write docs, since I think that's the next thing that will probably cause the design to stretch a little. [01:30] jbailey: you should add support for sticking GCC + binutils in there so that you can compile the driver on the fly. the Gentoo converts would *love* that [01:31] jdub: http://lwn.net/Articles/14776/ is a reasonable place to start. [01:31] sladen: If you have the backing store, honestly nothing stops it. [01:31] sladen: And the assembly model that we're working towards would make it pretty trivial to do, sadly. [01:32] sladen: It's worth noting that the initramfs includes *all* of the hardware autodetection. You can take this initramfs and copy it to another machine and expect it to Just Work. [01:32] yeah, reading it already ;) [01:32] first google hit [01:33] i've read it before but i never remember why it rocks === Jimbob [~jcape@c-24-14-115-101.hsd1.il.comcast.net] has joined #ubuntu-devel [01:34] jdub: Less crap in kernel, basically. [01:34] jdub: There's a bunch of things that otherwise have to stay resident, like dhcp code or raid detection code that can be jettisoned. [01:34] jbailey: so in future, you imagine that the basic initramfs image will be built into the kernel, so not requiring much at all to be built on kernel install? [01:34] Having that in userspace also means that if there's a bug, it's not risking taking out the whole box. [01:35] ie. i won't have to install klibc or busy box on my machine? :) [01:35] Hopefully, yes. [01:35] nice! [01:35] klibc is intended to go into the kernel source tree. [01:35] so modules will be stacked on top and so on [01:35] I wound up adding busybox when doing the raid1 stuff. [01:36] Right, your bootloader will be able to take various cpio images and mash them together. [01:36] So whatever pieces you need can be made available. [01:36] If you're not network booting, there's no reason to include the network drivers. [01:36] So I can imagine: scsi-2.6.12-1-powerpc64 ide-2.6.12-1-powerpc64 net-2.6.12-1-powerpc64 etc.. [01:36] Your bootloader would just put the interesting ones together. [01:37] Plus a "base" initramfs of some sort that's all nice and pre-assembled. [01:37] and we can do dsdt stuff more sanely :) [01:37] Hopefully. I haven't finished the DSDT bits yet. === D1 [~damian@pool-71-104-112-195.lsanca.dsl-w.verizon.net] has joined #ubuntu-devel [01:38] I know with the initrd it's just mashed onto the end with a magic number. I'm guessing it's the same in this case. [01:38] So again, adding it becomes a matter of bootloader configuration, or having the helper program assemble them together at kernel install time. [01:39] reboot time ;) [01:39] Rather than building based on what it thinks you might need. [01:39] Enjoy, and good luck. =) [01:39] will be rad for thin client foo [01:40] haha [01:41] oooookay [01:41] Is that a good laugh or a bad one? [01:41] fatal error inserting ipw2200, unknown symbol in module [01:41] jbailey: so what exactly does 'cpio --dereference' do if it doesn't do what it says on the tin? [01:41] Oh that's ad. [01:41] no such file /conf/modules [01:41] sad. [01:41] no such file /proc/mdstat [01:41] hda1 doesn't exist [01:42] dropping to shell [01:42] Oh shit. IDE case. [01:42] sh can't access tty; job control turned off [01:42] You probably want to modprobe ide-generic. I haven't tested on a pure IDE system yet. [01:42] Forgot about ide-generic madness. [01:42] oh, what've you been using? [01:42] SATA [01:43] jbailey: and could that hardcoded stuff be done in a more generic way with 'find'? [01:43] ok, ide-generic loaded [01:43] sladen: Which hardcoded stuff? [01:43] jdub: rerun udevstart to create the device files. [01:43] sladen: I donj't have find available in the initramfs right now. I'm trying to reduce the bits that I need so I can get rid of busybox. [01:44] jbailey: i ran udevstart, still no hda* [01:44] jbailey: all the 'lkn -s's in mkinitramfs [01:45] jdub: modprobe ide-disk ? [01:46] jdub: I'll cook up the ide bus walking stuff and put it in. I keep forgetting that ide is a bit weird. [01:46] jbailey: ended up doing ide-core + ide-disk and it was fine [01:46] sladen: No, those are likely to be more specific. [01:46] sladen: I want to include an absolute minimum of stuff in there. Right now I'm already pulling in crap like gzip. [01:47] The reason I'm doing ln -s's is to avoid copying it, then making another copy with cpio. ln -s is nice and quick. [01:47] jbailey: next step? :) [01:48] modprobe ext3 [01:48] Or for amusement, you can type fstype < /dev/hdaN [01:48] (replace N) [01:49] And then modprobe whatever FSTYPE comes out as. [01:49] nice :) [01:49] mount -r -t ${FSTYPE} ${ROOT} /root [01:49] umount /sys [01:50] umount /proc === henriquemaia [~henriquem@cb-217-129-170-55.netvisao.pt] has joined #ubuntu-devel [01:50] exec run-init /root /sbin/init [01:50] --- [01:50] And that should chain to your working system. [01:50] jdub: If you feel brave enough to try again, add ide-disk and ide-generic to /etc/initramfs/modules [01:50] Rebuild and try again. [01:51] blammo! [01:51] good blammo [01:51] :) [01:51] ok [01:51] I think ide-core should get pulled in them as a sideeffect. [01:52] oh MAAAN [01:52] my keyboard isn't working again === jbailey quiets the warning about /conf/modules === jbailey blinks [01:53] anyone run a laptop that has a smart battery and was able to get it to work in ubuntu? [01:53] jbailey: unrelated [01:53]