zul | 686-smp-dbg being built | 04:19 |
---|---|---|
fabbione | morning | 05:55 |
fabbione | zul: still here? | 05:56 |
fabbione | lamont: ? | 05:56 |
fabbione | infinity: ping?? | 05:59 |
=== send [~WinNT@221.218.42.179] has joined #ubuntu-kernel | ||
=== Seveas [~seveas@seveas.demon.nl] has joined #ubuntu-kernel | ||
=== doko [~doko___@dsl-082-082-206-179.arcor-ip.net] has joined #ubuntu-kernel | ||
=== JaneW [~JaneW@dumbledore.hbd.com] has joined #ubuntu-kernel | ||
infinity | 686-dbg builds right after 386-dbg, right? | 08:42 |
fabbione | yes | 08:44 |
fabbione | it should | 08:44 |
fabbione | AH AH | 08:44 |
infinity | Alright, then we're almost at the point of failure. | 08:44 |
fabbione | davis has hardware problems | 08:44 |
fabbione | it started segfaulting also on a hoary chroot | 08:44 |
infinity | And by 'hardware problems', you mean 'it came from Apple'? | 08:44 |
=== infinity goes to have a smoke and expects the kernel build to be dead or succeeded by the time he gets back. | ||
fabbione | ehhe | 08:45 |
fabbione | oh shit | 08:45 |
fabbione | i just trashed the build on concordia! | 08:45 |
fabbione | oh well it doesn't take long to redo it | 08:45 |
infinity | Building 686-dbg now.. | 08:49 |
infinity | Please fail, please fail... | 08:49 |
=== infinity joins a religion breifly so he can pray to some god or other, then unjoins. | ||
fabbione | hehe | 08:51 |
infinity | \o/ | 09:00 |
infinity | Yay. It failed the same way. | 09:00 |
infinity | And, ofr the record, the chroot is completely up to date. | 09:01 |
fabbione | oh great suckage | 09:03 |
fabbione | now it would be interesting to see if the same chroot moved to concordia will complete the build | 09:04 |
fabbione | infinity: let's try something... | 09:05 |
fabbione | go where the kernel has been unpacked | 09:06 |
fabbione | rm debian/build/build-686-dbg/stamp-build (if there is any) | 09:06 |
fabbione | and do..... | 09:06 |
infinity | (Should we try KALLSYMS_EXTRA_PASS?) | 09:07 |
fabbione | sed -i -e's/# CONFIG_KALLSYMS_EXTRA_PASS is not set/CONFIG_KALLSYMS_EXTRA_PASS=y/g' debian/build/build-686-dbg/.config | 09:07 |
infinity | Heh. | 09:07 |
fabbione | then try to complete the build | 09:08 |
infinity | Oh, for.... <sigh> | 09:10 |
=== infinity forgot the -nc switch to dpkg-buildpackage. | ||
infinity | Thankfully, this is a reasonably speedy machine. | 09:10 |
fabbione | yeah but you lose the config change | 09:10 |
infinity | I know. | 09:10 |
fabbione | you need to stop and edit debian/config/i386/686-dbg | 09:11 |
fabbione | so it will be done automatically | 09:11 |
=== infinity tries this again, with his brain engaged. | ||
fabbione | infinity: good idea :) | 09:13 |
infinity | "There is a simple workaround that is to increase the WORKING_SET define | 09:18 |
infinity | in scripts/kallsyms.c to something like 65536. This will include every | 09:18 |
infinity | symbol in the token table calculation, so that even if symbol position | 09:18 |
infinity | changes, the token table should be the same." | 09:18 |
infinity | http://seclists.org/lists/linux-kernel/2005/May/2010.html | 09:18 |
infinity | "The problem with this approach is that it takes longer to calculate the | 09:19 |
infinity | token table. (~3secs on my P4 2.8GHz, 11300 symbols)" | 09:19 |
infinity | I think we can spare 3 seconds on the buildds in the name of working kallsyms. Perhaps this is a patch/fix we should apply. :) | 09:19 |
infinity | (Or does it change the installed token table (System.map) to be huge?... That would be a runtime issue, not a build issue. Hrm) | 09:20 |
fabbione | no idea really | 09:23 |
fabbione | but than... why does it build on concordia??? | 09:23 |
fabbione | and locally? | 09:23 |
fabbione | but not in the buildd? | 09:23 |
fabbione | i don't believe the patch is the real solution to the problem, tho it can be used to workaround it | 09:23 |
infinity | And here's an explanation of the bug | 09:23 |
infinity | http://seclists.org/lists/linux-kernel/2005/May/1727.html | 09:23 |
infinity | I think the answer to "why does it only happen sometimes?" is "pure luck." | 09:24 |
fabbione | make sense | 09:25 |
infinity | (It may also have something to do with the running kernel, phase of the moon, and your favourite lunch item) | 09:25 |
fabbione | ehhee | 09:26 |
fabbione | infinity: we can just test it :) | 09:27 |
fabbione | infinity: do you still have the unpacked tree? | 09:27 |
infinity | Well, Im rebuilding with EXTRA_PASS right now. | 09:27 |
infinity | But I like Sam's patch better (expecially since he seems confident it will make mainline) | 09:27 |
fabbione | i am checking if it has been committed after rc5 | 09:28 |
fabbione | the problem is that shitnezz is in bk | 09:28 |
fabbione | and linus doesn't use bk | 09:28 |
fabbione | no it's not upstream yet | 09:29 |
fabbione | JaneW: ping? | 09:33 |
infinity | Ahh, here's why: | 09:36 |
infinity | "As noted in mail sent to you the 14th of March I have this patch queued | 09:36 |
infinity | up. But since Linus has asked for a calm down period it will wait | 09:36 |
infinity | until next kernel release opens up." | 09:36 |
infinity | So we may not see Sam's patch in mainline until 2.6.13 | 09:37 |
infinity | After this build, I'll check to see if the problem we're having is, indeed, the problem Sam's patch solves, or a differentone. | 09:38 |
infinity | (looks like kallsyms alignment has just been messy in general lately... Look at all the patches to arch-specific areas in the tree for alignment issues...) | 09:38 |
fabbione | infinity: cool thanks | 09:39 |
infinity | IOW, someone broke something recently in the name of "elegance", and everyone else is cleaning up the mess. Yay. | 09:39 |
infinity | (Is it too late to switch Ubuntu to a BSD kernel?) | 09:40 |
infinity | <cough> | 09:40 |
fabbione | infinity: ahaha | 09:40 |
fabbione | we can probably consider to create a derivate once lp is in place | 09:40 |
fabbione | the main issue is "when lp is in place" | 09:40 |
fabbione | we need a name for the security release | 09:42 |
JaneW | fabbione: pong | 09:42 |
fabbione | JaneW: right in time :) | 09:42 |
fabbione | we need 2 names lady :) | 09:42 |
JaneW | 2! | 09:42 |
fabbione | one vegetable/fruit based for the security release | 09:43 |
JaneW | don;t be greedy | 09:43 |
fabbione | and the usual nut for breezy :) | 09:43 |
fabbione | 2 releases.. 2 different targets ;) | 09:43 |
JaneW | fabbione: kernel - Laughing Lentil | 09:45 |
JaneW | fabbione: security - Prickly Pear | 09:45 |
fabbione | JaneW: rocking!! | 09:46 |
fabbione | oh rc6 is out :) | 09:47 |
infinity | JaneW : Nice to see that you've found a niche in Ubuntu. :) | 09:48 |
infinity | Yeah, -rc6 is out. That's what I was grepping for alignment changes. :) | 09:49 |
fabbione | no i was checking directly in git | 09:49 |
fabbione | and it's not there yet | 09:49 |
fabbione | there are already a bunch of commits after rc6 including build fixes | 09:49 |
fabbione | for today let's get this rc5 out with ppc64 and dbg support | 09:50 |
fabbione | rc6 can wait | 09:50 |
infinity | BUT RC6 IS SO MUCH BETTER CAUSE IT HAS A HIGHER NUMBER ON THE END!!! | 09:50 |
fabbione | even if it should take too long to switch | 09:50 |
fabbione | hahahahhaha | 09:50 |
infinity | So, do you watch the build logs, make a note of every driver that has "I can't write C to save my life" compile warnings, and make a mental note to neevr use those modules on your own system? :) | 09:51 |
fabbione | no.. i just don't use ubuntu kernels on my machine :) | 09:52 |
fabbione | ahhahha | 09:52 |
infinity | Alternately, there's the "you're using stuff that's been deprecated for 3 years, so obviously you don't maintain your dirvers" warnings. I don't like those much either. :) | 09:52 |
fabbione | yes | 09:54 |
fabbione | ops | 09:54 |
JaneW | infinity: heh yes, it's a tough job, but someone has to do it ;P | 09:54 |
=== chmj [~d3vic3@dumbledore.hbd.com] has joined #ubuntu-kernel | ||
infinity | fabbione : Builds fine with EXTRA_PASS. Will try Sam's patch instead now, for kicks. | 10:12 |
infinity | (Though EXTRA_PASS should be a good enough bandaid to get this going for now) | 10:13 |
fabbione | infinity: ok. remember to unset the EXTRA_PASS with the patch | 10:26 |
infinity | Yes, obviously. | 10:27 |
infinity | If you want to upload a -1.3 with EXTRA_PASS for now, though, it'll work on the buildds, it seems. | 10:27 |
fabbione | nah i can wait for your extra test | 10:29 |
fabbione | but hmmm | 10:29 |
fabbione | yeah | 10:29 |
fabbione | i will upload 1.3 with EXTRA_PASS | 10:29 |
fabbione | and we can figure the proper fix for 1.4 | 10:29 |
fabbione | actually there is another thing we forgot to set in the -dbg packages... | 10:30 |
fabbione | the ramdisk size that zul was talking about | 10:30 |
infinity | Oh? | 10:32 |
fabbione | infinity: a -dbg initrd is HUGE | 10:32 |
infinity | -dbg ramdisks are too big? | 10:32 |
infinity | Yeah. That makes sense. :0 | 10:32 |
fabbione | and the kernel default is not enough | 10:32 |
fabbione | so either i change the BLK_DEV_RAM_SIZE to a reasonable size | 10:32 |
fabbione | or we need to figure a hook for grub | 10:32 |
fabbione | and i guess the former > latter | 10:33 |
fabbione | now the point is that i trashed all the -dbg images to test :) | 10:33 |
fabbione | BUT WE HAVE BATTLE STAR CONCORDIA! | 10:33 |
fabbione | with ccache :) | 10:34 |
fabbione | goody.. both gcc-4.0 and gcc-3.4 are running the test suite on sparc | 10:35 |
infinity | Pfft. I'd feel your pain if I wasn't building gcc-3.3 and gcc-3.4 on m68k right now, after having just done two gcc-4.0 builds in a row. | 10:39 |
fabbione | ehhehe | 10:40 |
fabbione | [ ] linux-image-2.6.12-1-686-dbg_2.6.11.93-1.3_i386.deb 07-Jun-2005 09:57 147M | 11:08 |
fabbione | what's wrong with this???? | 11:08 |
infinity | Ouch. | 11:11 |
infinity | Also, ouch. | 11:11 |
infinity | No one's going to install that. | 11:11 |
infinity | Which makes it effectively useless. | 11:11 |
infinity | IMO. | 11:11 |
fabbione | and elmo will hang us on a cross, starts the witch dance and burns us alive | 11:12 |
fabbione | can you image over a GB of kernel push to the mirror on each update? | 11:12 |
chmj | hehehe | 11:13 |
fabbione | ok backup solution | 11:16 |
fabbione | we can provide the configs, but not build them by default | 11:17 |
fabbione | who needs it, will build it | 11:23 |
chmj | +he | 11:24 |
infinity | That's probably better. | 11:24 |
infinity | I'm on 1.5k DSL, and in most of the world, that's still considered a "really good" connection (though I think it's crap), and as a regular user, I would never download a 147 MB package to debug a bug I filed. | 11:25 |
fabbione | yeah i see no other solution | 11:25 |
infinity | And when downloading the source is less bandwidth than downloading the debug package, one starts to wonder. :) | 11:25 |
fabbione | infinity: actually.. that would be a good way to scare users in filing bugs | 11:25 |
fabbione | ahhaha | 11:25 |
fabbione | i need a big fat cigarette | 11:26 |
Mithrandir | infinity: 1.5k DSL is fairly measly. :P | 11:26 |
=== infinity heads off to go grocery shopping. | ||
infinity | Back soon. | 11:26 |
infinity | Mithrandir : Oh, I know, but it's still "decent" to most poeple. Just not geeks. | 11:26 |
Mithrandir | infinity: 1.5k? You mean 1.5M, I hope. | 11:27 |
Mithrandir | 1.5k is measly. It's slower than what you get from a 28.8 modem | 11:27 |
infinity | Err, yeah. | 11:27 |
infinity | 1.5M, obviously. | 11:27 |
chmj | I can only dream about 1M+ adsl | 11:28 |
infinity | Brain == Lame. | 11:28 |
Mithrandir | I need to get a faster networking card, since my router only has a 11Mbit WLAN card which limits my speed to ~2Mbit on my DSL. | 11:28 |
Mithrandir | apparently, I have 4Mbit now. | 11:28 |
infinity | Lucky. | 11:28 |
infinity | I'm hoping to see reasonable DSL speeds here within a year or so. | 11:28 |
chmj | that a lot off bandwidth | 11:29 |
Mithrandir | not my fault. They're upgrading the DSL for free once in a while ATM. | 11:29 |
infinity | I miss my old 7Mbps DSL in Calgary.. | 11:29 |
chmj | Mithrandir, what country u in ? | 11:29 |
Mithrandir | I ordered 2Mbit which then went to 2.2, 2.4, 2.6, then 4 | 11:29 |
Mithrandir | chmj: Norway. | 11:29 |
infinity | They claim we'll get ADSL2 and ADSL2+ (14Mbps and 28Mbps, I think) here within a year or so. | 11:29 |
Mithrandir | infinity: that's decent enough | 11:29 |
infinity | Personally, I can't wait. | 11:29 |
chmj | Mithrandir, ahh, the land off lots and lots of bandwidth | 11:29 |
infinity | Anyhow. Groceries. Back later. | 11:30 |
fabbione | you all suck | 11:30 |
Mithrandir | infinity: but they'll still steal your money by charging you 100 AUD per microbit. | 11:30 |
fabbione | but i will probably get 100Mb in a year or 2 | 11:30 |
chmj | (O.o) | 11:30 |
infinity | Mithrandir : I pay 100AUD flat rate right now for 1.5Mbps unlimited. It's not the kind of cheap I was used to in Canada, but it's reasonable for around here. | 11:30 |
Mithrandir | infinity: heh, around 20AUD more than I pay for my 4Mbit unlimited. | 11:31 |
=== Mithrandir tries to debug the acx100 problem | ||
Mithrandir | fabbione: the new driver and i386 worked, at least | 11:31 |
fabbione | Mithrandir: ah good to know | 11:32 |
Mithrandir | fabbione: nah, because then the bug is mine. :-P | 11:33 |
fabbione | that too :) | 11:33 |
fabbione | infinity: unpacked kernel is like 500MB ! | 11:33 |
fabbione | ahhaha | 11:33 |
fabbione | 41764 -rw-r--r-- 1 root root 42717184 Jun 7 11:33 initrd.img-2.6.12-1-686-dbg | 11:33 |
=== fabbione crashes on the floor | ||
mjg59 | Hahahahahaha | 11:34 |
Mithrandir | *chuckle* | 11:34 |
mjg59 | fabbione: I think something very odd has happened there | 11:34 |
Mithrandir | 42 MB initrd! Rock! | 11:34 |
Mithrandir | does it come with a free 1GB memory module too? | 11:34 |
fabbione | mjg59: with all _DEBUG configs turned on | 11:34 |
fabbione | i will try to rebuild it | 11:35 |
fabbione | time to get some food | 11:39 |
Mithrandir | oddbjoh: fabbione it still blows up with latest upstream. | 12:40 |
fabbione | Mithrandir: also on i386? | 12:45 |
fabbione | or only on amd64? | 12:45 |
Mithrandir | fabbione: I didn't try very long on i386. | 12:45 |
Mithrandir | loads and loads of warnings while compiling on amd64. | 12:46 |
fabbione | Mithrandir: no shit.. | 12:47 |
fabbione | mjg59: i did the rebuild.. it's 147MB | 01:16 |
fabbione | so no -dbg by default | 01:16 |
jbailey | g'm all. | 01:22 |
Mithrandir | hi jeff | 01:23 |
fabbione | hey jb | 01:23 |
=== doko [~doko___@dsl-084-059-046-215.arcor-ip.net] has joined #ubuntu-kernel | ||
mjg59 | Hmm. Something is very odd on this machine. | 01:30 |
mjg59 | PCI is basically broken after suspend/resume | 01:30 |
jbailey | PCI devices are overrated. | 01:31 |
jbailey | fabbione: Got time for me, or do you need to wait for a bit? | 01:36 |
fabbione | jbailey: yes in 5 minutes i will be completely your :) | 01:36 |
jbailey | Excellent. I'll afk for 5 then. | 01:36 |
infinity | fabbione : FWIW, Sam's patch also fixed the kallsyms bug. | 01:46 |
fabbione | infinity: cool | 01:47 |
fabbione | jbailey: ready to start? | 01:51 |
jbailey | fabbione: Yup, back now. | 02:00 |
fabbione | i was almost ready to go again :) | 02:00 |
fabbione | so what are we going to hack? | 02:02 |
chmj | i was gonn ask that | 02:05 |
jbailey | fabbione: making make-kpkg or whatever the piece is flexible enough to do mkinitramfs as well as mkinitrd | 02:09 |
fabbione | jbailey: ok than we are talking about kernel-package postinst files that are installed in the linux-image postinstalls | 02:09 |
jbailey | fabbione: Lovely. | 02:10 |
jbailey | I'm glad you know where all that crap is. My brain started to go square while reading make-kpkg | 02:10 |
fabbione | there are a few files that make use of mkinitrd | 02:12 |
fabbione | or call it | 02:12 |
fabbione | for what i can see we need to modify image.preinst and image.postinst | 02:13 |
jbailey | Are they all after /etc/kernel-img.conf has been sourced? | 02:13 |
fabbione | i am checking... | 02:14 |
jbailey | If yes, we can just do ramdisk = /usr/sbin/mkinitrd, and then refer to "${ramdisk}" | 02:14 |
fabbione | jbailey: for the preinstall 1) source the config and check the path to mkinitrd | 02:15 |
fabbione | i guess the call happens later on | 02:15 |
fabbione | probably only in postinst | 02:16 |
fabbione | preinstall looks more like a sanity check before messing around | 02:16 |
jbailey | MAkes sense. No sense unpacking the kernel for 30 minutes on an m68k box if it's setup poorly. | 02:16 |
fabbione | right | 02:18 |
fabbione | the postinst creates the image and does all the work of checking/unchecking | 02:18 |
fabbione | jbailey: we definetely want to make the change as smooth as possible and configurable | 02:20 |
jbailey | Right. | 02:20 |
jbailey | I can make the initramfs args the same as the initrd ones, that's easy enough. | 02:20 |
fabbione | jbailey: i think we shouldn't get crazy to make the same | 02:21 |
fabbione | but if you think that the args can be the same | 02:21 |
fabbione | it's even easier | 02:21 |
jbailey | I think it sounds less miserable than making the interface from the postinst configurable. =) | 02:21 |
jbailey | Yeah, there isn't a problem. | 02:22 |
fabbione | uh that's easy enough :) | 02:22 |
fabbione | also because we want to get these changes upstream | 02:22 |
fabbione | we don't want to blindly replace mkinitrd, do we? | 02:22 |
fabbione | or we want? | 02:22 |
jbailey | Well, that's why I'm thinking with a variable | 02:22 |
jbailey | If we call it as ${ramdisk}, where the default is /usr/sbin/mkinitrd | 02:23 |
jbailey | I will override it during my development here as /usr/sbin/mkinitramfs | 02:23 |
jbailey | And when it's time to switch, we just change the default, and what the Depends: line has. | 02:23 |
fabbione | that's more complicate than my solution :) | 02:23 |
fabbione | well actually... why do we need to touch kernel postinstall at all??? | 02:24 |
fabbione | let's make it in a very simpler way | 02:24 |
fabbione | 1) make mkramfsthingy to accept the same options as mkinitrd | 02:25 |
fabbione | 2) upload a mkinitrd package with the wrapper in place of mkinitrd | 02:26 |
fabbione | (and rename mkinitrd as mkinitrd.whatever) | 02:26 |
jbailey | And then alternatives it? | 02:26 |
fabbione | 3) upload ramfsthingy with the config file for the wrapper | 02:26 |
fabbione | no.. the wrapper will read the config file and use mkinitrd or the other one | 02:26 |
jbailey | Oh ugh. | 02:26 |
jbailey | This is simpler? | 02:26 |
fabbione | at this point all the changes are done outside the kernel | 02:27 |
fabbione | yes | 02:27 |
fabbione | becuase you change the default in the config file and you are done | 02:27 |
fabbione | when you want to kill mkinitrd, you just do it, changing the default and removing the mkinitrd.orig | 02:27 |
fabbione | and you didn't touch any of the other packages | 02:27 |
fabbione | it's all done at a lower level that the kernel doesn't give a rat ass about :))) | 02:28 |
jbailey | I think we're talking about a 2 line change to the kernel postinst, though. | 02:28 |
fabbione | jbailey: it's a 2 line changes that are objects of other variables... | 02:28 |
fabbione | first we need to change kernel-package | 02:29 |
jbailey | 3 line. | 02:29 |
jbailey | my $ramdisk = "/usr/sbin/mkinitrd"; # initrd generating program | 02:29 |
fabbione | the configs in kernel-packages can be overridden by the user and that affects also kernel postinstall | 02:29 |
fabbione | + upload the kernel with the new postinstalls | 02:29 |
jbailey | $ramdisk = "$1" if /ramdisk\s*=\s*(\S+)/ig; | 02:30 |
fabbione | and your packages to make all the above working :) | 02:30 |
jbailey | And change the line that contains "mkinitrd" to '$ramdisk' | 02:30 |
jbailey | And the solution of futzing with the postinst means that anyone can change this at a later tim. | 02:30 |
jbailey | Otherwise mkinitrd has to grow a wrapper, has to analyse a foreign config file and there's no future flexibility. | 02:31 |
jbailey | Like in case the yaird folks want to make it possible to use their setup under Debian. | 02:31 |
fabbione | hte configfile only has one entry... mkinitrd.real = true|false | 02:31 |
jbailey | Ugh You mean use an extra config file? | 02:32 |
fabbione | it doesn't need to use a foreign config file | 02:32 |
jbailey | I think that's wrong. | 02:32 |
jbailey | I think it needs to be in kernel-img.conf | 02:32 |
jbailey | Otherwise people are having to look in yet another place for this. | 02:32 |
jbailey | And there's still no flexibility | 02:32 |
fabbione | that is right, but you said you needed this for testing, didn't you? | 02:32 |
jbailey | So it would wind up having to parse it out, and forces us to keep the mkinitrd name. | 02:32 |
fabbione | or do you want to go directly for final solution? | 02:32 |
jbailey | Right, but why not make it flexible enough to do more when it appears to be a 3 line change in one file. =) | 02:33 |
fabbione | because take into account that a change to kernel-package might not go into debian :) | 02:33 |
jbailey | I'll NMU it in if I need to. | 02:33 |
jbailey | =) | 02:33 |
jbailey | initramfs is coming to town there too. | 02:33 |
jbailey | As is yaird | 02:33 |
fabbione | jbailey: that'd be Manoj's toy :) | 02:33 |
jbailey | (I will likely sponsor) | 02:33 |
jbailey | I haven't had NMU wars with someone in ages. =) | 02:33 |
fabbione | hmm ok | 02:35 |
fabbione | if you prefer that way we can do it | 02:35 |
jbailey | I'mve pinged manoj to check with him. | 02:35 |
fabbione | i really don't mind | 02:35 |
jbailey | But for 6 lines of changes (including the preinst check), it seems... excessive to add a new config file and a wrapper and all that. | 02:36 |
fabbione | sure i understand your point | 02:36 |
fabbione | i was considering the option to keep all the initrd/ramfs creation stuff outside of kernel build-deps and depends... | 02:36 |
fabbione | that will also require the ramfs to Depends: on kernel-package | 02:37 |
fabbione | that is actually only a build-dep for the kernel | 02:37 |
jbailey | hmm. | 02:38 |
jbailey | forgot about the depends: thing. | 02:38 |
fabbione | specially if you want it configurable | 02:38 |
jbailey | hmmhmmhm | 02:39 |
jbailey | Well, I suppose that could be done with provides on our side. | 02:39 |
fabbione | uh????????? | 02:40 |
fabbione | if you want /etc/kernel-img.conf you want kernel-package | 02:40 |
fabbione | that right now is only a b-d | 02:40 |
fabbione | with your change it needs to be a depends | 02:40 |
fabbione | if not for linux-image, it needs to be for the ramfs package | 02:41 |
fabbione | for linux-image is enough it's unpacked and the config there | 02:41 |
jbailey | Sorry, lagging a second from the phone.. | 02:42 |
fabbione | sure | 02:43 |
fabbione | don't worry | 02:43 |
fabbione | i am preparing 1.3 for upload | 02:44 |
fabbione | just to get ppc64 to boot and i386 to build :) | 02:44 |
fabbione | brb | 02:46 |
jbailey | Umm. | 02:47 |
jbailey | Okay, I don't think your statemenet about kernel-img.conf coming from kernel-package is true. | 02:47 |
jbailey | I think it's placed there by d-i. | 02:47 |
jbailey | If it's not there, the kernel will refuse to install, since it won't have do_initrd = Yes | 02:47 |
fabbione | dpkg -c /mirrors/ubuntu/pool/main/k/kernel-package/kernel-package_8.135ubuntu2_all.deb | 02:52 |
fabbione | -rw-r--r-- root/root 965 2004-11-17 17:33:32 ./etc/kernel-pkg.conf | 02:52 |
fabbione | oh kernel-img.conf | 02:52 |
jbailey | =) | 02:52 |
fabbione | hell i was looking at the wrong file | 02:52 |
fabbione | which package provides kernel-img.conf? | 02:53 |
jbailey | Nothing does. | 02:53 |
fabbione | kernel-package: /usr/share/doc/kernel-package/examples/sample.kernel-img.conf | 02:53 |
fabbione | kernel-package has the manpage and so on | 02:53 |
fabbione | but not the config file | 02:53 |
jbailey | asking Kamion in #u-d | 02:54 |
zul | yo | 02:59 |
fabbione | hey zul | 03:00 |
fabbione | zul: i have a bad news | 03:00 |
zul | hmm? | 03:00 |
fabbione | we need to revert the -dbg stuff | 03:00 |
fabbione | just the 686 image is like 147MB | 03:00 |
fabbione | * Do not build -dbg flavours automatically since images are huges, but still | 03:00 |
fabbione | provide configs in debian/config/debugging/$flavour for the users. | 03:00 |
fabbione | that's the best we can do :( | 03:01 |
zul | thats fine.. | 03:01 |
zul | less work for me then | 03:02 |
fabbione | btw 12rc6 is out | 03:02 |
zul | goody | 03:02 |
zul | heh...it seems like everything i do is dropped ;( | 03:03 |
fabbione | zul: it wasn't really an option to push more than a GB of data on each kernel upload | 03:04 |
fabbione | sorry.. | 03:04 |
fabbione | i didn't expect them to be so big | 03:04 |
zul | i know..i said that tongue in cheek | 03:04 |
jbailey | fabbione: rc6 should contain the module information that you told me about, yes? =) | 03:07 |
fabbione | jbailey: yes it does | 03:08 |
jbailey | w00t | 03:08 |
=== jbailey asks nicely for it. =) | ||
fabbione | jbailey: so if you want to go ahead and change kernel-package is fine for me | 03:08 |
jbailey | fabbione: What I was thinking of for the build-dep, is that I could make mkinitrd Provides: kernel-ramdisk | 03:08 |
fabbione | jbailey: that will be tomorrow's stuff | 03:08 |
jbailey | fabbione: For now if you depend on mkinitrd | kernel-ramdisk, and change it later to initramfs | kernel-ramdisk that might solve the problem. | 03:09 |
fabbione | Depends: initrd-tools (>= 0.1.78ubuntu1), coreutils | fileutils (>= 4.0), module-init-tools (>= 0.9.13) | 03:09 |
fabbione | jbailey: can't the 2 packages co-exist? | 03:10 |
fabbione | ok.. hold on.. | 03:10 |
fabbione | let's make a point here | 03:10 |
jbailey | Sure... | 03:10 |
fabbione | are we doing all this config stuff only for testing?? | 03:10 |
jbailey | It's not a conflicts, it's just to make sure you have something providing it. | 03:10 |
fabbione | or is it going to be the final solution? | 03:10 |
fabbione | because given that NOBODY owns /etc/kernel-img.conf | 03:11 |
jbailey | I'd like it to remain forever to allow some flexibility. | 03:11 |
fabbione | i am really unhappy of something mangling it | 03:11 |
jbailey | nobody owns it, but you're the only one using it. | 03:11 |
fabbione | given that we don't have the previous state of the config file | 03:11 |
fabbione | jbailey: are you sure? | 03:11 |
fabbione | i am one of the package using it | 03:11 |
fabbione | we are not sure i am the only one | 03:11 |
fabbione | also.. who is going to mangle it? | 03:12 |
fabbione | to switch default? | 03:12 |
jbailey | I haven't encountered anything else using it, and the name suggests that it's for you. | 03:12 |
jbailey | No, nothing should ever need to mangle it. | 03:12 |
jbailey | That's why everything just defaults to initrd-tools for now. | 03:12 |
fabbione | so if you make it configurable, how are we supposed to switch default? | 03:12 |
fabbione | if the default is in the config? | 03:12 |
jbailey | We just change the default when no configuration is provided. | 03:12 |
jbailey | Default doesn't go in the config. | 03:12 |
fabbione | ok | 03:12 |
fabbione | so let's summarize the steps: | 03:14 |
fabbione | 1) change kernel-package/kernel/postinst to look for that ramdisk var | 03:15 |
fabbione | 2) upload initrd-tools to provide kernel-ramdisk | 03:15 |
fabbione | 3) switch the images to Depends: initrd-tools | kernel-ramdisk | 03:16 |
fabbione | 4) test test test | 03:16 |
fabbione | 5) switch linux-image Depends: to initramfs | kernel-ramdisk | 03:16 |
fabbione | even if i really don't see the point in the kernel-ramdisk thingy | 03:17 |
fabbione | we could just switch the default and that's it | 03:17 |
fabbione | or perhaps you want to have kernel-ramdisk Provided by both initramfs and initrd-tools preferring one of them? | 03:18 |
fabbione | jbailey: are you still around? | 03:25 |
fabbione | i need to stop soon | 03:25 |
fabbione | i am tired to death | 03:25 |
jbailey | fabbione: Right. The only advantage of the kernel ram,disk stuff | 03:25 |
jbailey | is flexibility. | 03:25 |
jbailey | I'm not too worried about that. | 03:25 |
fabbione | ok | 03:25 |
jbailey | The config file on its own would make me happy. | 03:25 |
jbailey | I'll ping manoj to figure out how likely it is that he'll accept it. | 03:26 |
fabbione | do you want to start with kernel-package? | 03:26 |
jbailey | Yes. | 03:26 |
fabbione | ok | 03:26 |
jbailey | Mind if I do the hackery and upload? | 03:26 |
fabbione | not at all | 03:26 |
fabbione | go ahead :) | 03:26 |
jbailey | Lovely. | 03:26 |
jbailey | I'll try to push it into Debian sanely so that you can just sync again. | 03:26 |
fabbione | i don't care who uploads my package, until i know who is working on them and why :) | 03:27 |
fabbione | meh kernel-package is desynced anyway | 03:27 |
fabbione | i did try to talk with Manoj unsuccessfully for a few days now | 03:27 |
jbailey | Yup. Ilike to check to make sure I'm not running over other work that's being done. | 03:27 |
fabbione | well there are the changes to make powerpc64 a buildable image :) | 03:27 |
jbailey | If he doesn't reply, it's easy. I NMU in Debian because of unresponsive maintainer. | 03:27 |
fabbione | jbailey: not all ubuntu changes should go in Debian | 03:28 |
jbailey | No, just this one. | 03:28 |
jbailey | I know you're trying to keep it the same. | 03:28 |
fabbione | exactly | 03:28 |
fabbione | but for ubuntu go ahead and do it... | 03:28 |
fabbione | so we can test the solution and propose it to Manoj as "tested and it works" | 03:29 |
jbailey | Cool. I'll get that in by the end of my day. | 03:29 |
fabbione | perfect | 03:29 |
jbailey | This is supposed to be my support day, so I'll be writing specs today. | 03:29 |
jbailey | Perhaps I'll wake up the laptop and do so on the roof deck. | 03:29 |
fabbione | jbailey: poing 1 and 2 is your stuff :) | 03:29 |
fabbione | i will do point 3 in prepartion with 12rc6 | 03:29 |
fabbione | that would be my tomorrow | 03:29 |
jbailey | I'll just do point 1 for today. | 03:30 |
jbailey | I'm not fussed about the kernel-ramdisk bit for now. | 03:30 |
fabbione | ok | 03:30 |
jbailey | I haven't thought that part through completely. | 03:30 |
jbailey | I've only thought through the config file. | 03:30 |
fabbione | ok well.. there is always the option to upload an ramfs that Conflicts/Provides mkinitrd | 03:31 |
jbailey | And it solves my immediate need that all my systems should be running initramfs' now to make sure I'm testing enough. | 03:31 |
fabbione | just for the testing time | 03:31 |
jbailey | Can't, kernel-packages have a versioned depends. | 03:31 |
jbailey | The provides lose out. | 03:31 |
fabbione | ah crap | 03:31 |
fabbione | ok | 03:31 |
fabbione | well let | 03:31 |
fabbione | well let's go trough point 1/3 asap | 03:32 |
fabbione | so you get an easy working environment | 03:32 |
fabbione | we can rethink the other bits later | 03:32 |
jbailey | initrd and initramfs coexist. | 03:32 |
jbailey | So there's no fuss. I don't care for now what gets depended upon until we switch. | 03:32 |
jbailey | That's why I'm content to leave it alone for my testing. | 03:32 |
fabbione | oky | 03:32 |
fabbione | i guess i am off | 03:33 |
fabbione | i might pass by later | 03:33 |
jbailey | Thanks for staying up for me, Fabio. g'd evening! | 03:33 |
fabbione | no problem dude | 03:34 |
fabbione | it's just that today i am more tired than usual | 03:34 |
fabbione | otherwise i would have done the implementation with you | 03:34 |
jbailey | I think I've actually finished the implementation now, I just need to test. | 03:35 |
=== lamont__ [~lamont@15.238.5.47] has joined #ubuntu-kernel | ||
zul | fabbione: did you put up rc6? | 05:08 |
jbailey | zul: He said that was tomorrow's work. | 05:33 |
zul | ah ok | 05:43 |
=== Nafallo [~nafallo@nafallo.user] has joined #ubuntu-kernel | ||
=== Nafallo [~nafallo@nafallo.user] has left #ubuntu-kernel ["Read] | ||
zul | there is a cc meeting today isnt there? | 07:14 |
=== Seveas [~seveas@seveas.demon.nl] has joined #ubuntu-kernel | ||
zul | i hate dealing with recuriters | 09:47 |
fabbione | zul: i will start 12rc6 tomorrow.. | 10:31 |
fabbione | i am off to bed now | 10:31 |
zul | ok | 10:35 |
lamont__ | mjg59: ping | 11:25 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!