[12:20] <mjg59> BenC: I've just sent you a couple of ndiswrapper patches
[12:22] <mjg59> BenC: So in the last couple of days I've sent you:
[12:24] <mjg59> BenC: alps suspend, ACPI hotkey, hp sound support, sk98lin, PCMCIA bridge fixes, PCMCIA hibernate fix, HP reboot patch, ATI 2x timer, Appletouch driver, ndiswrapper patches
[12:24] <mjg59> BenC: When you wake up, could you confirm that you've got all of those? :)
[12:24] <mjg59> BenC: Oh, also ipw2200 device shutdown
[01:22] <BenC> mjg59: sure thing
[01:22] <BenC> mjg59: and thanks
[01:24] <mjg59> BenC: No problem
[01:40] <mjg59> BenC: Wurgh. Sorry, one more.
[01:40] <mjg59> Low priority
[01:40] <mjg59> tc1100-wmi
[05:20] <zul> heylo
[05:21] <zul> BenC, i just commited to my arch sk98lin is there sky2 isnt im all set.
[05:23] <fabbione> morning guys
[05:23] <zul> morning fabbione
[05:24] <fabbione> hey zul
[05:25] <zul> right im off my laptop is dying so im off
[05:25] <fabbione> eheh good night
[05:25] <zul> later..
[05:31] <fabbione> BenC: ping?
[05:36] <TheMuso> c
[05:46] <fabbione> BenC: when you need to bump the ABI, please use the target bumpapi: in debian/rules
[05:46] <fabbione> BenC: it will do the right thing for you
[05:47] <fabbione> it will save you from little issues like killing the build system, building empty pkgs and destroy hppa build :)
[07:16] <BenC> fabbione: ah, ok
[07:17] <fabbione> BenC: anyway.. i did fix it..
[07:17] <fabbione> and committed
[07:17] <BenC> saw that, thanks
[07:20] <fabbione> no problem
[07:21] <fabbione> BenC: did you have any time to look at klibc?
[07:21] <fabbione> it start to be of a big concern for me
[07:21] <fabbione> because we are getting very close to release and it would really really suck to miss breezy for one package
[07:22] <fabbione> if you need access to a sparc, i can let you use mine
[01:09] <ivoks> hi
[01:10] <ivoks> i don't know if this is known problem, but after recent update i don't have /dev/psaux anymore
[01:13] <ivoks> oh, udev... sorry about this...
[02:12] <zul> hey
[02:15] <mjg59> zul: Hi
[02:15] <mjg59> You were looking for me last night?
[02:24] <zul> mjg59: yeah i just wanted to let you know i commited sk98lin to my arch
[02:25] <mjg59> Ok, cool
[02:27] <BenC> zul: hey
[02:28] <mjg59> BenC: Have you got all the patches I sent you?
[02:28] <BenC> yeah
[02:29] <mjg59> Cool
[02:31] <zul> BenC: heylo
[02:31] <BenC> zul: ready for me to pull yet?
[02:32] <zul> yep..i just have sk98lin patch and the ndiswrapper amd64 stuff in my tree, i didnt commit the buslogic because i was having problems with it getting it to compile
[02:34] <BenC> zul: ok, can you tell me your repo URL again?
[02:35] <BenC> I lost it with the rest of my system last week
[02:35] <BenC> and can you email me the buslogic diff you have? I can see about getting it to compile
[02:35] <zul> http://zulinux.homelinux.net/arch/zulcss@gmail.com--2005
[02:35] <zul> yeah i can do that tonight when i get home
[02:36] <zul> i have a meeting in 30 minutes for 2 hours and then another after that...fun fun
[02:38] <BenC> good luck...bring a pillow :)
[02:45] <BenC> zul?
[02:54] <zul^> damn it
[03:50] <fabbione> hey guys
[04:09] <BenC> hey
[04:10] <fabbione> BenC: *cough*klibc*cough* ;)
[04:14] <BenC> I tried building it, but for some reason it isn't finding the kernel headers
[04:14] <fabbione> doh!
[04:14] <fabbione> stupid question.. did you install them?
[04:14] <fabbione> it doesn't use the one from the running kernel
[04:15] <fabbione> it wants the generic ones
[04:22] <BenC> generic as in /usr/include/{linux,asm}?
[04:35] <BenC> fabbione: I'm thinking that klibc and everything that uses it may need to be compiled 64-bit
[04:36] <BenC> I'll let you know though
[04:36] <fabbione> BenC: no as in /usr/src/linux-kernel-headers-2.6.12-8
[04:44] <BenC> ok, compiled for me
[04:44] <BenC> had to prepend the build command like "sparc32 debian/rules"
[04:48] <BenC> the static binaries work (gzip, etc)
[04:48] <BenC> ash works too
[04:50] <fabbione> so it does build with our headers..
[04:50] <fabbione> and in 32 bit...
[04:50] <fabbione> but i guess you are using debian based chroot.. aren't you?
[04:50] <BenC> yeah
[04:50] <fabbione> ok
[04:50] <BenC> the key may be the sparc32 command
[04:50] <fabbione> we need to understand why it fails in a breezy chroot
[04:50] <BenC> otherwise it seems to go by uname
[04:50] <fabbione> we only use sparc32 on buildd
[04:51] <fabbione> like debian does
[04:51] <fabbione> but it fails
[04:51] <fabbione> so it might be a toolchain issue
[04:51] <BenC> is there a real ubuntu machine I can get to?
[04:52] <fabbione> yes
[04:52] <fabbione> just send me your ssh public key
[04:53] <BenC> sent
[04:58] <fabbione> added
[04:59] <lamont> Sep 21 08:55:36 localhost kernel: [4300057.793000]  Process iwlist (pid: 23067, threadinfo=e94d6000 task=c3fd8a60)
[04:59] <lamont> Sep 21 08:55:36 localhost kernel: [4300057.793000]  Call Trace:
[04:59] <lamont> Sep 21 08:55:36 localhost kernel: [4300057.793000]   [pg0+944840526/1069872128]  ide_do_request+0x1f3/0x367 [ide_core] 
[04:59] <lamont> Sep 21 08:55:36 localhost kernel: [4300057.793000]   [io_schedule+14/22]  io_schedule+0xe/0x16
[04:59] <lamont> Sep 21 08:55:36 localhost kernel: [4300057.793000]   [__wait_on_bit_lock+87/97]  __wait_on_bit_lock+0x57/0x61
[04:59] <lamont> Sep 21 08:55:36 localhost kernel: [4300057.793000]   [prep_new_page+115/120]  prep_new_page+0x73/0x78
[04:59] <lamont> Sep 21 08:55:36 localhost kernel: [4300057.793000]   [buffered_rmqueue+326/447]  buffered_rmqueue+0x146/0x1bf
[04:59] <lamont> Sep 21 08:55:36 localhost kernel: [4300057.793000]   [__alloc_pages+374/998]  __alloc_pages+0x176/0x3e6
[04:59] <lamont> Sep 21 08:55:36 localhost kernel: [4300057.793000]   [pg0+948702806/1069872128]  ieee80211_iterate_nodes+0x3b/0x66 [wlan] 
[04:59] <lamont> Sep 21 08:55:36 localhost kernel: [4300057.793000]   [pg0+948729857/1069872128]  ieee80211_ioctl_giwscan+0x5c/0xa8 [wlan] 
[04:59] <lamont> Sep 21 08:55:36 localhost kernel: [4300057.793000]   [pg0+948722528/1069872128]  read_ap_result+0x0/0x5b8 [wlan] 
[04:59] <lamont> Sep 21 08:55:36 localhost kernel: [4300057.793000]   [wireless_process_ioctl+1472/2023]  wireless_process_ioctl+0x5c0/0x7e7
[04:59] <lamont> bad ath0
[05:00] <lamont> and then ifconfig hangs
[05:00] <fabbione> lamont: known bug
[05:00] <fabbione> patch welcome :)
[05:00] <lamont> ah, okl
[05:10] <zul> morning
[05:11] <BenC> zul, I've merged your stuff, thanks
[05:13] <fabbione> hm let me know when it's time to build on sparc :)
[05:14] <BenC> : undefined reference to `.umul'
[05:14] <BenC> /usr/lib/gcc/sparc-linux-gnu/4.0.2/libgcc.a(_muldi3.o): In function `__muldi3':
[05:15] <zul> BenC: cool thanks
[05:15] <BenC> libgcc is broken
[05:15] <BenC> or...
[05:16] <fabbione> or?
[05:16] <fabbione> can we build it with like gcc-3.4?
[05:16] <fabbione> i don't think i did try that
[05:17] <BenC> I'm thinking it is compiling for v7, when we can do with v8
[05:17] <BenC> is there no editor in this chroot? :)
[05:18] <fabbione> nope.. let me install it for you
[05:18] <fabbione> i can see you already did
[05:19] <fabbione> BenC: take into account we did drop all the sparc32 support.. there should be nothing left
[05:19] <BenC> -m32 may be setting -mcpu=v7
[05:23] <fabbione> there is vi installed
[05:23] <fabbione> dunno if you noticed
[05:24] <fabbione> or do you need something more fanc
[05:24] <fabbione> +y
[05:25] <BenC> vi: command not found
[05:25] <BenC> I had to install vim
[05:25] <fabbione> yeah when i did try to install, it was already installed :)
[05:27] <jbailey> fabbione: Well, dropped pre-ultrasparc support.  It's not like we opted for a 64bit usersapce.
[05:27] <fabbione> yeah..
[05:27] <fabbione> i know
[05:27] <fabbione> we could consider to switch to full 64 bit
[05:27] <fabbione> but i guess if we won't make breezy, my sparc will stop being a buildd
[05:28] <fabbione> and become my firewall/router
[05:28] <jbailey> What will serve as the buildd then?
[05:28] <jbailey> Or are you just not going to bother?
[05:29] <BenC> full 64-bit userspace is not suggested
[05:30] <BenC> some programs are suggested to be 64-bit though (like anything messing in /proc, fs tools, etc)
[05:30] <BenC> odd, there's a umul.S for sparc in klibc
[05:30] <jbailey> Right, it's confusing that it doesn't pull it in.
[05:30] <jbailey> In fact, it's confusing that it's using the system libgcc at all when it has a bundled one.
[05:30] <BenC> yeah, maybe I need to find out why
[05:30] <BenC> oh, really?
[05:31] <jbailey> They all seem to by the top MCONFIG file.
[05:31] <fabbione> jbailey: i might give dapper a shot.. failed that i will stop..
[05:31] <BenC> fabbione: FYI, I mistook my build before, it was on a debian system, I forgot to chroot
[05:31] <BenC> so it does in fact build
[05:31] <BenC> just not in breezy userspace
[05:32] <fabbione> that sounds more like a gcc fucks up
[05:32] <fabbione> what gcc are you using on Debian?
[05:32] <jbailey> fabbione: What was the blocker for Hoary?
[05:32] <jbailey> Also keep in mind that the sparc glibc is totally different on Ubuntu than it is on Debian.
[05:32] <fabbione> jbailey: gcc wrong B-D
[05:33] <fabbione> jbailey: how do i force gcc with cdbs?
[05:33] <fabbione> DEB_MAKE_CC ??
[05:34] <jbailey> CC=
[05:34] <BenC> umul is getting compiled
[05:34] <jbailey> BenC: You've got as far as I did before my connection to Fabio's sparc stopped being useful. =)
[05:35] <jbailey> *sigh*
[05:35] <fabbione> jbailey: you are still logged in according to vultus5
[05:35] <jbailey> fabbione: Nope.
[05:35] <jbailey> fabbione: The firewalls like to drop connections.
[05:35] <jbailey> And I hit a huge lag while I was debugging.
[05:36] <fabbione> jbailey: try to force the connection to 195.22.207.161 than
[05:36] <BenC> .umul is in libc.a too
[05:36] <fabbione> it's not natted
[05:36] <jbailey> fabbione: I go through two nat boxes here to get out.
[05:37] <fabbione> ah ok
[05:37] <jbailey> fabbione: with klibc?  Setting CC is probably not enough.
[05:37] <jbailey> You'll need to touch MCONFIG, I suspect.
[05:38] <jbailey> klibc doesn't use autotools or much in the way of standardish things.
[05:38] <BenC> ok, this is an ordering problem of some kind
[05:38] <fabbione> it was enough to add CC=gcc-3.3 at make invocation inside debian/rules
[05:39] <jbailey> Ah, cool.
[05:44] <BenC> looks like gcc-3.3 did the trick
[05:45] <fabbione> yeah
[05:45] <fabbione> it builds
[05:47] <fabbione> i guess there is only one way to know if it works :)
[05:47] <BenC> ash doesn't work
[05:48] <jbailey> Make sure you're in the static directory.
[05:49] <BenC> zcat does
[05:49] <jbailey> I don't use ash for initramfs FWIW
[05:49] <BenC> $ file sh
[05:49] <BenC> sh: ELF 32-bit MSB executable, SPARC, version 1 (SYSV), statically linked (uses shared libs), stripped
[05:49] <BenC> ./sh
[05:49] <BenC> -bash: ./sh: No such file or directory
[05:50] <jbailey> What directory are you in, though? =)
[05:50] <jbailey> file won't show it as dynamically linked, it's not an ELF file.
[05:50] <BenC> how is it "statically linked" AND "uses shared libs"
[05:50] <jbailey> klibc is thpethial.
[05:50] <BenC> it shows it is ELF
[05:50] <jbailey> It's sort of lying.
[05:50] <BenC> ok
[05:50] <jbailey> It's ELF enough for the kernel to load it.
[05:50] <BenC> sh.g works
[05:50] <BenC> sh must be linked to shared
[05:51] <BenC> so we're good
[05:51] <fabbione> ok..
[05:51] <jbailey> fabbione: Want me to smack it with a GCC-3.3 for sparc, and send her off?
[05:51] <fabbione> i guess we need to try to boot a kernel with a initramfs
[05:51] <fabbione> jbailey: let see if it works first
[05:51] <BenC> ./static/uname -a
[05:51] <BenC> Linux vultus5 2.6.12-8-sparc64 #1 Wed Aug 31 06:10:19 UTC 2005 sparc64 sparc64
[05:52] <jbailey> Right.  Let's see how well initrmafs works on sparc in the first place. =)
[05:52] <jbailey> fabbione: Add the word 'break' to the kernel command line to get dropped into a shell in the initramfs
[05:52] <jbailey> BenC: The reason it's not pure ELF is to do away with semantics that don't need to be true for early userspace.
[05:53] <fabbione> jbailey: i hope it's not going to dump me like last time
[05:53] <jbailey> BenC: There's no ongoing binary compataibility, symbol versioning, etc.
[05:53] <BenC> gotcha
[05:53] <jbailey> BenC: That's why the klibc.so file name is so weird - it's a sha1 hash of the symbol list. =)
[05:53] <BenC> so it's like aout, but in elf format :)
[05:53] <jbailey> Any change, and you can't use it. =)
[05:53] <jbailey> Right.
[05:53] <jbailey> The side effect is that it can be *really* small. =)
[05:54] <BenC> and still shared libs
[05:54] <fabbione> Setting up libklibc (1.0.14-1ubuntu2) ...
[05:54] <fabbione> Setting up klibc-utils (1.0.14-1ubuntu2) ...
[05:54] <fabbione> jbailey: these 2 are enough.. right?
[05:54] <jbailey> fabbione: busybox-cvs-initramfs
[05:55] <fabbione> oh ok
[05:58] <fabbione> that's already installed..
[05:59] <jbailey> initramfs-tools depends on everything it needs, so it ought to be fine.
[05:59] <fabbione> yup it is
[05:59] <jbailey> HEy, once it's through the NEW queue, it'll even be in Debian. =)
[05:59] <fabbione> i am distupgrading the box ...
[06:01] <infinity> jbailey : Have you talked with dilinger and friends about moving Debian to initramfs, once our guinea pigs^W^Wusers have helped iron out the bigger bugs?
[06:02] <jbailey> infinity: Not in detail, no.
[06:02] <jbailey> infinity: Or rather, the last time we all talked about it was a kernel team meeting almost a year ago.
[06:03] <infinity> It irritates me slightly less than mkinitrd (though, I still prefer no initrd, but that's me), so I wouldn't mind it becoming the Debian standard.
[06:03] <jbailey> Hard to do no initrd in a distribution kernel.
[06:03] <infinity> yes, exactly.
[06:03] <fabbione> infinity: klibc is not portable yet
[06:03] <fabbione> not everywhere
[06:03] <infinity> Hence "the standard".
[06:03] <infinity> The Adam Standard is no initrd and monolithic kernel builds.
[06:04] <jbailey> fabbione: klibc is generally portable.  It uses kernel header files for the missing bits.
[06:04] <jbailey> So generally if you have a working kernel for an arch, klibc ought to work/
[06:04] <infinity> fabbione : How many arches are we missing, and how evil is the code?
[06:04] <fabbione> jbailey: but it has not been tested on many arches
[06:04] <infinity> jbailey : Ahh, you answered my question.
[06:04] <fabbione> infinity: no idea.. i did read the doc a while ago
[06:04] <infinity> Getting it testing on m68k is just a matter of throwing it at smarenka (our d-i hacker and reboot bitch) and saying "test this, yo"
[06:05] <jbailey> fabbione: Eh?  It's been tested on a surprising amount of them.
[06:05] <fabbione> jbailey: some of Debian arches arent''
[06:05] <jbailey> infinity: It's even easier than that.  Build it and do utils/static/uname -a
[06:05] <fabbione> including sparc64
[06:05] <fabbione> we are the guinea pigs..
[06:05] <jbailey> fabbione: sparc64 works fine.  It's sparc32 that has the problem, remember? =)
[06:05] <fabbione> (vultus5 rebooting now)
[06:06] <jbailey> infinity: If that works, the rest of it generally ought to since it's just syscalls.
[06:06] <infinity> jbailey : Oh, that's all it takes to declare it working?
[06:06] <jbailey> infinity: Unless there's some ELF magic that's special on m68k
[06:06] <jbailey> But even then we could just pull the static versions for m68k for now.
[06:06] <infinity> Nothing particularly magic compared to other arches.
[06:06] <fabbione> YEAH!
[06:06] <jbailey> Hmm, I should add that test to the klibc build.
[06:06] <fabbione> jbailey: you are good to go
[06:06] <fabbione> upload
[06:06] <jbailey> fabbione: Ist g?
[06:06] <fabbione> like..
[06:07] <fabbione> yesterday
[06:07] <jbailey> fabbione: So if the arch is sparc, force gcc-3.3?
[06:07] <jbailey> No other patches needed?
[06:07] <fabbione> nope.. i did only force gcc-3.3
[06:07] <infinity> jbailey : A very simple testsuite (even if it's just running the one binary and checking the exit code) would be very appreciated.
[06:07] <fabbione> with the code that's there
[06:07] <jbailey> fabbione: FWIW, other sparc distros seem to have the same problem (acc to Google), so it's probably generally gcc-4 breakage.
[06:08] <infinity> jbailey : I wish more packages did something that simple.
[06:08] <jbailey> fabbione: I can try to talk to some of the sparc gcc folks.
[06:08] <fabbione> jbailey: ok...
[06:08] <jbailey> infinity: YEah.  I'd like to see Debian policy changed such that if a package has a testsuite it must be run.
[06:08] <jbailey> Awwww.
[06:08] <jbailey> infinity: Even that would be a start.
[06:08] <jbailey> Then get an army of ants to go write testsuites. =)
[06:09] <infinity> jbailey : I dunno, some testsuites are insane (check out the X test suite sometime), but ANYTHING that can tell me "yeah, this probably won't segfault when you call --help" is so far from where we are today, it would be great.
[06:10] <jbailey> infinity: Well, it might discourage Daniel from uploading daily. ;)
[06:10] <infinity> It could block Ubuntu's buildds for a week.
[06:10] <infinity> Lord only knows what it would do to m68k or arm.
[06:11] <infinity> I'd probably just conditionally disable the suite on slower arches, which completely defeats the purpose.
[06:11] <dilinger> infinity: aiui, we will release 2.6.13 with initramfs
[06:12] <infinity> What most packages need is a reasonably small set of testcases that can cover the "more common usages that really ought to work", so we can see if gcc/binutils is providing us with completely dud binaries.
[06:12] <infinity> dilinger : Rock.
[06:12] <dilinger> initramfs just entered the archive; no noe is really interested in keeping devfs around, and i haven't seen anyone get yaird working w/ it
[06:13] <infinity> initramfs actually appears to work, so I'm all for using it. ;)
[06:13] <dilinger> pfft
[06:13] <dilinger> whether it works or not is irrelevant ;)
[06:14] <infinity> If Etch leapfrogs breezy/dapper (very likely, given dapper's more conservative release goals), I may just switch this laptop to Debian again.
[06:14] <infinity> (But I didn't say that in an #ubuntu- channel, no sir)
[06:15] <jbailey> leapfrogs how? =)
[06:15] <infinity> Spiffy-feature, laptop-lovin', new-software-havin'-wise.
[06:16] <infinity> If I have anything to say about it, Ubuntu won't even attempt to be a feature leader for Dapper's development, cause I want rock-hard stability.
[06:16] <infinity> But I'm sure this argument will come up over and over (and over) again at UBZ.
[06:16] <infinity> I just hope I win.
[06:16] <jbailey> lamont: What?  Do you finally have lvm2? =)
[06:16] <lamont> infinity: dapper willl wind up with whatever gnome is latest/greatest, you can count on that.
[06:17] <lamont> jbailey: /lvm2_2.01.04-5ubuntu1: Installed [:] 
[06:17] <jbailey> \o/
[06:17] <infinity> lamont : Yeah, tailing the GNOME release schedule is much less harmful than some other crack people will try to foist on us, though.
[06:18] <lamont> built on sept 1
[06:18] <lamont> true
[06:18] <lamont> jbailey: last time I tried it, ISTR klibc was the only holdout (well, actually, its rbuild-deps...)
[06:19] <BenC> anyone else think I should start with 2.6.14-rc's for post breezy rather than messing with 2.6.13?
[06:19] <jbailey> Yup
[06:20] <jbailey> I do't think it's worth the hassle of porting the patches twice.
[06:20] <fabbione> BenC: skip .13 
[06:20] <fabbione> we can do like we did for breezy and started on .12rc directly
[06:20] <BenC> sounds like a plan then
[06:20] <fabbione> BenC: what we did was:
[06:20] <fabbione> 2.6.11.90 (12rc1)
[06:20] <fabbione> no ABI changes tracking
[06:21] <fabbione> lots of uploads
[06:21] <fabbione> we did start tracking the ABI as soon as .12 was final
[06:24] <BenC> fabbione: anytime between now and tomorrow is a good time for a sparc test build
[06:24] <fabbione> BenC: do you have any pending commit to baz?
[06:25] <BenC> hold on
[06:25] <BenC> I have a failing patch
[06:25] <fabbione> ok
[06:25] <BenC> only thing I have left is a 5-liner for x86
[06:25] <BenC> let me fix this patch
[06:26] <fabbione> ok
[06:29] <jbailey> BenC: When can I book your for kernel postinst love? =)
[06:30] <jbailey> BenC: I really want it to use update-initramfs instead of mkinitramfs
[06:30] <jbailey> That will make it so that I can just do "update-initramfs -u" from usplash
[06:30] <fabbione> jbailey: than you should probably do it by now
[06:30] <jbailey> fabbione: I've been asking for help for a week now..
[06:30] <jbailey> I don't know perl well enoyugh to hack that file safely.
[06:31] <fabbione> jbailey: we need to hack the postinst.. right?
[06:31] <jbailey> and postrm, yes.
[06:31] <fabbione> hm ok
[06:31] <fabbione> BenC: when is your upload planned?
[06:32] <fabbione> jbailey: if you can tell me exactly what i should replace with what, i can do it tomorrow
[06:33] <jbailey> fabbione: Yup.  Right now it makes a call to mkinitramfs -o /boot/initrd.img-$(uname -r) $(uname -r)
[06:34] <jbailey> fabbione: It should instead do /usr/sbin/update-initramfs -k $(uname -r) -u -t
[06:34] <jbailey> fabbione: At pruge time, it currently just rm's the initramfs.
[06:34] <jbailey> fabbione: It should instead do /sbin/update-initramfs -k $(uname -r) -d
[06:35] <fabbione>   my $ret = system("$ramdisk " .
[06:35] <fabbione>                    ($mkimage ? "-m '$mkimage' " : "") .
[06:35] <fabbione>                    "-o $initrd_path.new $modules_base/$version");
[06:35] <fabbione> this is postinst
[06:35] <fabbione> ramdisk = mkinitramfs
[06:35] <BenC> tomorrow
[06:35] <BenC> jbailey: if you explain to me how to use it properly, I'll do it this time
[06:36] <fabbione> BenC: at what time + o -?
[06:36] <fabbione> because infinity is planning a l-r-m upload to fix some bugs
[06:36] <BenC> fabbione: probably late, 17:00 EST
[06:36] <fabbione> ok
[06:36] <jbailey> fabbione: What appends the kernel version to that call?
[06:36] <jbailey> Oh, duh, sorry.
[06:36] <jbailey> Last line.
[06:37] <fabbione> hmmm
[06:38] <jbailey> fabbione: So I guess something like my $ret = system("/usr/sbin/update-initramfs -u -t -k " . "$version");
[06:38] <fabbione> the problem is that mkimage and module_base
[06:38] <fabbione> yes.. i got to that already...
[06:38] <jbailey> That mkimage is just ignored on initramfs anyway.
[06:38] <fabbione> but it will remove 2 bits from that code
[06:38] <jbailey> There's no userful value that -m can have.
[06:38] <jbailey> It used to be possible to choose cramfs of ext2.
[06:38] <jbailey> With initramfs it's always cpio
[06:38] <fabbione> yes, but who says the user is not using ramdisk=/path/to/mkinitrd ?
[06:39] <fabbione> what if he hardcoded initramfs
[06:39] <jbailey> fabbione: Yeah, the compatability case sould probably be kept in case someone needs the old initrd stuff or whatever.
[06:39] <fabbione> in /etc/kernel.img or whatever crappy conf file we use
[06:39] <fabbione> jbailey: how badly is this change needed?
[06:40] <jbailey> fabbione: Without it usplash and friends can't update on their own.
[06:40] <infinity> Will the new kernel upload involve an ABI bump?
[06:40] <jbailey> fabbione: I don't think anyone would really notice if it weren't there, honest, although for a while Matt had marked it as preview-critical.
[06:40] <fabbione> infinity: apparently yes
[06:40] <jbailey> I had update-initramfs in time for that, but then he said not to worry about it for preview.
[06:41] <jbailey> and hasn't asked since.
[06:41] <infinity> Oh, kay.  Well, I can hold off on l-r-m, or I can just upload twice.  No big deal.
[06:41] <fabbione> infinity: i would prefer 2 uploads
[06:41] <fabbione> infinity: unleash the bug fixes
[06:41] <fabbione> bump the abi is easy and BenC can do it
[06:41] <infinity> Yeah, s'why I'm still up. ;)
[06:41] <fabbione> jbailey: i am thiniking..
[06:42] <jbailey> fabbione: Either way.  But given that I didn't get it right last time without any logic changes, I'm not comfortable doing the change myself.
[06:42] <infinity> Letting BenC do the ABI bump removes me from the "you touched it last" maintenance assignment, which is a bonus.
[06:42] <fabbione> jbailey: last time it was only a typo dude!
[06:43] <jbailey> fabbione: You made it sound like I had fubard your baz setup... =)
[06:43] <fabbione> jbailey: eh no...
[06:43] <fabbione> kernel-package (8.135ubuntu4) breezy; urgency=low
[06:43] <fabbione>   * Add a simple cosmetic change to the kernel postinst.
[06:43] <fabbione>  -- Fabio M. Di Nitto <fabbione@ubuntu.com>  Wed, 08 Jun 2005 13:53:44 +0200
[06:43] <fabbione> kernel-package (8.135ubuntu3) breezy; urgency=low
[06:43] <fabbione>   * New variable in /etc/kernel-img.conf: ramdisk
[06:43] <fabbione>     Allows the admin to choose a different program than
[06:43] <fabbione>     /usr/sbin/mkinitrd for initrd generation.
[06:43] <fabbione>     Note: Must take the same arguments as mkinitrd.
[06:43] <fabbione>  -- Jeff Bailey <jbailey@ubuntu.com>  Wed,  8 Jun 2005 01:47:00 +0000
[06:43] <fabbione> 8.135ubuntu3 <- your upload
[06:44] <fabbione> 8.135ubuntu4 <- my upload to cover your error :)
[06:44] <fabbione> how can that be the "world is falling down because you suck" changelog :)
[06:44] <jbailey> You go easy on me so often ;)
[06:45] <fabbione> ;)
[06:45] <jbailey> *g*
[06:46] <fabbione> ok i think that we need to retain full backwards compatibility
[06:46] <jbailey> The !@#$ing author should write some damned docs one day...
[06:46] <fabbione> meaning that i need to check if we are going for the default update-initramfs
[06:46] <fabbione> or calling something else
[06:46] <jbailey> fabbione: I also don't know how this intersects with custom kernels.
[06:47] <fabbione> HMMMM
[06:47] <jbailey> fabbione: I've seen a few comments that people doing that have run into initramfs not supporting the -m option.
[06:47] <jbailey> fabbione: So far I"ve been able to get away with just telling them "Don't do that." *g*
[06:47] <fabbione> kthxbye
[06:47] <jbailey> Suits me fine.
[06:47] <fabbione> perfect
[06:47] <jbailey> Hmm
[06:48] <jbailey> Just have to make sure that usplash does it's love before the kernel is updated.
[06:48] <jbailey> That way upgrades to breezy actually get it.
[06:48] <fabbione> jbailey: it does now...
[06:48] <fabbione> i did test hoary -> breezy yesterday..
[06:48] <jbailey> Does it?  How is that promised?
[06:48] <fabbione> but i can retest it again
[06:48] <fabbione> dunno.. i didn't really pay attention to it
[06:48] <jbailey> I know Colin did an installer hack.
[06:48] <fabbione> DEB_MAKE_INVOKE = make -C $(DEB_BUILDDIR) $(DEB_MAKE_ENVVARS)
[06:48] <fabbione>  ->
[06:49] <fabbione> DEB_MAKE_INVOKE = make CC=gcc-3.3 -C $(DEB_BUILDDIR) $(DEB_MAKE_ENVVARS)
[06:49] <fabbione> and don't forget the B-D :)
[06:49] <fabbione> jbailey: you can just change that to look like:
[06:49] <fabbione> DEB_MAKE_INVOKE = make CC=$(CC) -C $(DEB_BUILDDIR) $(DEB_MAKE_ENVVARS)
[06:49] <fabbione> and set CC=gcc by dafault
[06:49] <fabbione> for sparc you set CC=gcc-3.3
[06:50] <fabbione> that should work
[06:50] <jbailey> CC is already gcc by default, I think.
[06:50] <fabbione> and you can test it locally
[06:50] <fabbione> just add it.. never be too sure
[06:50] <fabbione> it's one redundant line in the worst case
[06:50] <jbailey> It was the ifeq ($(DEB_HOST_ARCH),sparc) that I was having trouble remembering.
[06:50] <fabbione> jbailey: if you want i can do it
[06:50] <jbailey> fabbione: nah, I have a parisc patch for it to that I Was holding for the sparc upload.
[06:51] <fabbione> DEB_HOST_GNU_CPU := $(shell dpkg-architecture -qDEB_HOST_GNU_CPU)
[06:51] <fabbione> ifeq "$(DEB_HOST_GNU_CPU)" "sparc"
[06:51] <fabbione> CC = gcc-3.3
[06:51] <fabbione> endif
[06:51] <jbailey> I prefer to do it by arch rather than CPU.
[06:51] <jbailey> CPU types change too easily.
[06:51] <fabbione> ok
[06:51] <jbailey> ifeq($(DEB_HOST_ARCH),sparc)
[06:51] <jbailey> CC=gcc-3.3
[06:51] <jbailey> endif
[06:51] <fabbione> well it's frigging late
[06:52] <fabbione> BenC: please be sure to commit everything in baz so tomorrow morning i can test build
[06:53] <fabbione> ok i am off for a while
[06:53] <fabbione> i might pass by later
[06:56] <lamont> BenC: have you had any time to give ia64 some love?
[06:57] <lamont> not DEB_BUILD_ARCH???
[06:57] <lamont> jbailey: ^^^
[06:57] <jbailey> lamont: Have I got them backwards?
[06:57] <jbailey> I don't care what arch is building it if it's cross compiling...
[06:57] <lamont> build is what you're building, host is where you're running
[06:57] <jbailey> Really?
[06:57] <BenC> lamont: none :(
[06:58] <jbailey> That's backwards to how ./configure does it. =(
[06:58] <lamont> if I build hppa binaries on sparc, BUILD==hppa, HOST=sparc
[06:58] <lamont> really
[06:58] <lamont> I could be completely wrong, mind you
[06:58] <jbailey> Yeah, build is the machine you're building on.  host is where it will run, and target is what it should produce binaries for (for gcc and a canadian cross)
[06:59] <jbailey>        DEB_BUILD_ARCH
[06:59] <jbailey>            The Debian architecture of the build machine.
[06:59] <jbailey> Useful
[06:59] <lamont> yeah.  and _HOST_ == 'of the host machine'
[07:00] <lamont> so don't mind me
[07:00] <lamont> jbailey: did you make time to pester #parisc about linux32 love?
[07:00] <lamont> (yet, that is.)
[07:05] <jbailey> lamont: I talked about it in there
[07:05] <jbailey> (Why, are you watching my talk with bdale in #debian-boot ? *g*)
[07:06] <jbailey> lamont: Right now the personality stuff has been used for a bunch of signal handling bits.
[07:06] <jbailey> lamont: ppc and sparc used a thread identifer for that instead.
[07:06] <jbailey> That leaves the personality syscall free for munging the uname output
[07:07] <jbailey> If I were to do the uame munging output right now, all 32 bit binaries would return parisc, and all 64bit binaries would return parisc64, not ideal. =)
[07:07] <lamont> nah - was just wondering if you'd made headway
[07:08] <lamont> ah, that's what the thread-specific crap was, eh?
[07:08] <jbailey> Basically some.   I know what needs to be done, and I eiter need to fire up my a500 now to try it or find a remote serial console/reboot setup.
[07:08] <jbailey> bdale's j5k has such a setup, and I'm basically the only one who uses it.
[07:12] <BenC> how do I get back a file that I did baz rm on?
[07:13] <jbailey> In bzr, I just 'bzr revert FILE', dunno if baz has the same.
[07:15] <BenC> doesn't seem to
[07:18] <jbailey> Hmm, jblack is 10 hours idle.  I usually poke him if I'm forced to use baz and can't remember a command. =)
[07:18] <BenC> there's baz undo
[07:19] <BenC> but that complains about the file not existing (which is my problem :)
[07:20] <jbailey> Try touching it and doing it?
[07:23] <BenC> baz undo works
[07:23] <BenC> if I undo the whole tree, and then re-apply the changes
[07:26] <BenC> a little clunky, but it works
[07:26] <infinity> A little clunky?
[07:27] <infinity> You're so terribly polite. :)
[07:27] <BenC> hehe
[07:39] <BenC> lamont?
[07:40] <lamont> BenC: you were mumbling something about maybe playing with your ia64 box...
[07:41] <lamont> I was wondering if you had any more luck in actually doing that than I have.
[07:46] <zul> hmm...what did i miss?
[07:46] <lamont> btw, I wish that the aironet driver didn't come up with 'tsunami' as the SSID by default...
[07:48] <mjg59> Hahahaha
[07:50] <lamont> mjg59: feh
[07:56] <BenC> lamont: nah, I I'm in the middle of moving, so I've no time atm
[07:56] <BenC> lamont: is there an ia64 I can login to though?
[07:56] <BenC> mjg59: all but that last patch is in baz right now, I'll put that one in later after I make sure things are building ok as-is
[07:57] <mjg59> BenC: Which is the last one? The HP laptop thing?
[08:30] <BenC> mjg59: yeah
[08:58] <BenC> is there a quick way to take a kernel and make a bootable (net inst?) cd image?
[09:03] <fabbione> BenC: not exactly quickly but it can be done
[09:04] <fabbione> you need to store the kernel udebs in a local repo
[09:04] <fabbione> and build d-i telling it to use that repo too
[09:04] <fabbione> that will give you the image
[09:06] <BenC> probably too much trouble to test this one patch
[09:06] <fabbione> hmm mostlikely
[09:06] <fabbione> did you fix the patch in the repo?
[09:07] <BenC> yes, the repo is buildable now
[09:07] <BenC> i386 has almost completed
[09:08] <BenC> the last two patches I have don't touch anything sparc related, so a test now would be fine
[09:08] <fabbione> perfect
[09:11] <zul> this is the last 2.6.12 isnt it?
[09:11] <fabbione> zul: mostlikely yes
[09:12] <zul> wohoo...
[09:12] <fabbione> BenC: did you already coordinate with elmo/Kamion/mdz to bump the abi?
[09:17] <BenC> zul?
[09:17] <BenC> fabbione: not yet, right now I'm making sure this will actually happen, which I should know within the next hour
[09:18] <BenC> emailed kernel-team, which I'm sure mdz is on, but not sure about elmo or Kamion
[09:18] <fabbione> BenC: no they are not..  better you CC them
[09:19] <fabbione> it's important that they are around (at least one of them) to assist you with NEW love on katie
[09:19] <zul> BenC: hmm?
[09:19] <BenC> zul: I got the buslogic MODULE_DEVICE_TABLE taken care of
[09:19] <zul> oh you did? damn..
[09:20] <BenC> I didn't bother rewriting the probe stuff, since it doesn't really need to support hotplug
[09:20] <zul> me either...
[09:20] <BenC> just the device table to hotplug scripts will load the module
[09:20] <zul> where did you place the struct?
[09:20] <BenC> for instance, you wont be able to plug in a pcmcia buslogic card (if there is such a thing)
[09:20] <BenC> near the init function
[09:21] <BenC> doesn't matter, the MODULE_DEVICE_TABLE() exports it no matter where it is
[09:22] <BenC> zul: do you have a vmware 4.5 install to test with?
[09:22] <zul> will have one tonight though
[09:22] <BenC> cool, patch is commited, so if you can test, I'd be grateful
[09:23] <zul> okie dokie
[09:23] <fabbione> BenC: updating the changelog to remove some obsolete entries my side
[09:24] <BenC> ok
[09:30] <BenC> ok, all my patches are in now
[09:30] <BenC> I still need to do one more thing, and that's get the ABI file in the .deb's
[09:31] <BenC> it may be tomorrow morning when I get this done fabbione
[09:31] <BenC> the upload that is
[09:31] <BenC> so you may want to do the sparc build now and let me know when you get up if it went ok
[09:32] <fabbione> BenC: ok..
[09:32] <fabbione> i might not be around when you start
[09:32] <BenC> what's your time now?
[09:32] <fabbione> i have school from 14:00 UTC till 19:00 UTC
[09:32] <fabbione> it's 21:30
[09:32] <BenC> well, if you email me when you get up in the morning, that will be well before I get up
[09:32] <fabbione> sure
[09:33] <fabbione> i usually wake up around 04:00 UTC
[09:33] <BenC> how long does your ultra take to do a build?
[09:33] <fabbione> sometimes earlier than that
[09:33] <fabbione> a couple of hours usually
[09:33] <BenC> I need to get my e3k running, 6 cpu's and 6gigs, it zips through builds
[09:33] <fabbione> but i am planning to start the build now, if i get baz to checkout :)
[09:33] <BenC> I wouldn't mind making it a buildd either
[09:34] <fabbione> BenC: that would rock so much
[09:34] <mjg59> BenC: If you have an x86 test build, I'll happily test it on all the machines I have here
[09:34] <BenC> mjg59: ok, it's almost done I think
[09:34] <mjg59> BenC: Cool
[09:38] <BenC> building k8 now, so just about there
[09:39] <mjg59> Excellent
[09:41] <zul> mjg59: the ndiswrapper stuff for amd64 is there as well
[09:41] <BenC> yeah, I got that in
[09:42] <zul> so we would need the user space tools for amd64
[09:43] <mjg59> I talked to elmo about those - they ought to have been sorted
[09:55] <zul> goodie
[09:58] <zul> later folks
[10:08] <fabbione> applying patch arch-i386-kernel-hp-laptop-reboot-fix to ./ ... failed.
[10:08] <fabbione> make[1] : *** [debian/monolith/patch-2.6.12-9.14]  Error 1
[10:08] <fabbione> make[1] : Leaving directory `/home/sparcbuildd/linux-source-2.6.12-2.6.12'
[10:08] <fabbione> BenC: the patches still fails to apply
[10:08] <fabbione> i really need to get some sleep
[10:09] <fabbione> BenC: if you feel lucky.. there is a kernel chroot on vultus5 with everything ready :)
[10:10] <fabbione> good night
[10:17] <BenC> it's fixed
[10:17] <BenC> update again
[10:32] <Mithrandir> so, is it known that the current kernel fucks up ipw2200?
[10:37] <BenC> mjg59: deb's are in my ~/ on concordia
[10:37] <BenC> Mithrandir: should be fixed in 9.14
[10:46] <mjg59> BenC: I don't have access to concordia
[10:46] <BenC> let me put it on ppl
[10:46] <mjg59> Ta
[11:29] <BenC> mjg59: kernels are going to be at http://people.ubuntu.com/~bcollins/kernels/
[11:29] <BenC> just keep an eye out in the next 10-15 minutes
[11:31] <mjg59> Ok, ta
[12:01] <BenC> done