[12:24] <BenC> kylem: The kernel ends up needing a unsigned 64-bit cmp function from libgcc
[12:24] <BenC> __ucmpdi2 to be exact
[12:25] <BenC> it's something in v4l2
[12:25] <kylem> ok
[12:26] <BenC> it'll show up in modpost
[12:26] <BenC> I don't understand why x86_64 and sparc64 didn't already show it
[12:27] <BenC> oh, well, probably because di2 isn't needed on 64-bit
[12:27] <BenC> well, it didn't show on x86 :)
[12:28] <kylem> is it ppc64?
[12:28] <BenC> nope, ppc32
[12:29] <kylem> ok.
[12:30] <kylem> could just disable that on ppc i suppose...
[12:30] <BenC> hehe, could just disable ppc :)
[12:30] <kylem> heh, but but but. :)
[01:58] <kylem> BenC, looking into the ppc failure now
[03:06] <BenC> kylem: I'm not sure if it's correct, but it may be that getting __ucmpdi2.S from gcc would be the right thing
[03:07] <kylem> i think i remember you just needing to pull something from libgcc.
[03:19] <BenC> I'm talking with benh now...hoping he has some ideas
[03:31] <kylem> ok
[03:41] <BenC> benh concludes the gcc thievery
[03:43] <BenC> kylem: you want to do the dirty work?
[03:47] <zul> heh
[03:56] <kylem> sure.
[03:59] <kylem> 00000000         *UND*  00000000 __ucmpdi2
[03:59] <kylem> thar she blows
[04:00] <BenC> kylem: Want to see something ugly?
[04:00] <BenC> git-diff-tree -p e75f9cee32827853fc2f9d1ceb6352e3edc33e9d | less
[04:00] <BenC> look for v4l2_norm_to_name
[04:01] <kylem> is this going to cause me to hate someone? :)
[04:02] <kylem> ew.
[04:03] <kylem> +               v4l2_std_id *id = arg,norm;
[04:03] <kylem> *blink*
[04:03] <BenC> that thar is the culprit of the ucmpdi2
[04:04] <BenC> kylem: Here's the cool part...v4l2_std_id is a bitmask, and that use it in a switch statement!
[04:04] <BenC> yay for programmers
[04:05] <kylem> ergh.
[04:19] <kylem> this driver is fucked.
[04:25] <kylem> doesn't look like there's a .S
[04:25] <kylem> in libgcc
[04:26] <kylem> huh kvm got merginated.
[04:30] <BenC> I've been wanting to test kvm
[04:30] <BenC> that's such a poor name though
[04:31] <kylem> fuck, i was going to test on parisc (to see how that symbol gets defined) but linus' tree is all borkinated. bleh.
[04:41] <kylem> http://people.ubuntu.com/~kyle/ppc-__ucmpdi2.diff
[04:41] <kylem> i think that ought to work, from what i recall doing on parisc.
[04:43] <kylem> void foo(void); EXPORT_SYMBOL(foo); is pretty evil, i think it will work for my atheros hack too.
[04:43] <kylem> since gcc takes care of the calling convetions for those, all we need to do is make sure it gets linked.
[05:09] <BenC> kylem: Almost line-for-line what I already started building
[05:10] <kylem> cool
[05:14] <BenC> c0115c30 T __ucmpdi2
[05:14] <BenC> sweet, worked
[05:14] <BenC> MODPOST doesn't complain either
[05:15] <kylem> unF!
[05:18] <BenC> want me to just apply that patch, or do you wan to do it and push?
[05:18] <kylem> might as well just apply it, it's pretty trivial
[05:21] <kylem> want me to mail to lkml or will you take care of that?
[06:05] <BenC> kylem: good night, and thanks
[01:27] <zul> hey
[02:52] <kylem> morning.
[02:54] <zul> morning kyle how are you
[02:56] <kylem> not bad
[02:56] <kylem> i made a western so i'm happy :)
[02:56] <zul> cool..
[02:56] <zul> its kind of warm outside today
[02:56] <kylem> yes veru
[02:56] <kylem> y
[02:57] <Mithrandir> zul: you're aware that if you want any new xen-3.0 uploads to build, you need to get libvncserver-dev in main?
[02:57] <Mithrandir> zul: you might want to write a MIR for that
[02:57] <zul> ok..
[02:57] <Mithrandir> jfyi
[03:43] <jerom1> Hi all
[03:44] <jerom1> When i add theses packages in my kickstart linux-headers-server, linux-image-server, linux-source my initrd is failed after reboot. Have you an idea ?
[03:59] <BenC> in kickstart? You mean like for an install, or what?
[03:59] <kylem> whoo
[04:00] <jerom1> benC : yes i use kickstart with PXE boot for install HP Proliant DL360 G5 Server
[04:02] <BenC> jerom1: And is it the initrd for the installer, or after you install and reboot, that is failing?
[04:02] <jerom1> it's failed after my install and reboot
[04:03] <BenC> and what's the failing message?
[04:04] <jerom1> unmoutable device ... kernel panic after reboot
[04:04] <jerom1>  and my grub conf is ok
[04:08] <derekS> are there still issues with 2.6.19-7.10 booting up? i am having them...
[04:08] <derekS> not sure if its my fault or the package
[05:22] <BenC> ppc64 is booting 2.6.20 now
[05:22] <zul> cool
[05:26] <kylem> sweet
[05:26] <_MMA_> BenC: About the spec you wanted me to get together. Should I come at it from exactly what we want and let you decide what is possible? Or should we restrain ourselves?
[05:27] <BenC> _MMA_: get it all out there
[05:27] <_MMA_> No problem.
[05:31] <derekS> hey, did you guys read my question above?
[05:31] <mjg59> BenC: Surely linking libgcc can't be the right answer?
[05:32] <kylem> mjg59, it's the right answer on a whole lot of platforms.
[05:32] <kylem> the "right" answer is shooting the v4l author.
[05:32] <mjg59> Really? Ouch.
[05:32] <BenC> mjg59: It seems to be what othr architectures do
[05:32] <BenC> mjg59: It's pretty much the same as taking the .S for that op and compiling it with the kernel
[05:32] <mjg59> Yeah
[05:32] <kylem> we /aren't/ linking libgcc.
[05:33] <kylem> it will just take the specific functions it needs from the .a
[05:33] <kylem> and throws the rest away.
[06:32] <tepsipakki> kylem: do you have an ETA for the dapper-kernel you've been working on?
[06:59] <BenC> mjg59: ping
[07:05] <kylem> sorry, was eating lunch
[07:06] <mjg59> BenC: Hi
[07:06] <BenC> mjg59: Can you look at the release/completion patch we have in some of the pcmcia controller drivers and tell me if we still need them?
[07:07] <BenC> I'm getting a softlockup in i82365 on my P4 and I think it might be related to those changes
[07:08] <mjg59> release/completition patch?
[07:10] <BenC> it adds a release() call back to the platform device, and it calls complete()...module exit does a wait_for_completion() on it
[07:10] <BenC> it's was suspend related, IIRC
[07:10] <BenC> been around since breezy I think
[07:11] <mjg59> Uh.
[07:11] <mjg59> Not actually sure if it was anything to do with me.
[07:11] <BenC> I thought I remembered you asking me to put it back in dapper because I left it out...maybe I am misremembering :)
[07:11] <BenC> if you don't know anything about it, I'm killing it
[07:12] <mjg59> Looks like it came from http://lists.infradead.org/pipermail/linux-pcmcia/2004-March/000598.html
[07:13] <BenC> that patch isn't causing the softlockup anyway
[07:15] <BenC> I think that's uneeded now...the drivers are using platform_device_alloc() which has:
[07:15] <BenC>                 pa->pdev.dev.release = platform_device_release;
[07:16] <BenC> oops, I didn't actually use the unpatched one...reverting that patch fixes it
[07:17] <kylem> hmm
[07:22] <BenC> kick ass
[07:23] <BenC> that softlockup was my last hurdle before upload
[07:25] <tepsipakki> BenC: do you plan to enable ext4 in it, if not yet then maybe later?
[07:32] <BenC> tepsipakki: When it's not marked experimental
[07:33] <tepsipakki> ok, thanks
[07:33] <BenC> If people want it, they can apt-get the source tarball and build it themselves for testing
[07:33] <tepsipakki> sure, not wise to keep experimental stuff in a supported kernel
[07:34] <tepsipakki> by default
[08:13] <kylem> hmm, still no sprint location
[08:13] <kylem> the suspense kills me. ;-P
[08:14] <kylem> BenC, i heard back from patrick, need to modify the -4572 fix before i can apply it.
[08:31] <BenC> kylem: Yeah, saw that
[09:19] <zul> BenC: ping did you get a chance to try it yet?
[09:49] <derekS> is anyone still experienceing the bug with grub+2.6.19-7?
[10:02] <BenC> derekS: "the bug"?
[10:03] <BenC> zul: Not yet, will do it for sure this evening though
[10:05] <zul> cool...its warm enough i take it? ;)
[10:05] <derekS> lemme find the number
[10:09] <derekS> BenC: https://bugs.launchpad.net/distros/ubuntu/+source/grub/+bug/73090
[10:09] <derekS> i think that is the bug i am still seeing
[10:17] <derekS> it starts grub, then nothing really happens
[10:17] <derekS> i get a few errors, then black screen
[10:36] <BenC> derekS: I don't think so
[10:36] <BenC> I mean, I don't think that's your bug
[10:36] <derekS> BenC: hmmm, ok
[10:37] <derekS> it starts grub (ie i see the 2 lines of txt) then weird stuff happens, then i get an error about nvram... then a few more warnings/errors, then something about cpu(either 0 or 1, dual cpu machine) has stalled
[10:37] <derekS> if i wait long enough the screen turns blank
[10:38] <derekS> different bug?
[10:41] <kylem> woot. there. tg3, sky2, and forcedeth work on edgy/dapper with the same code.
[10:42] <zul> neat
[10:46] <BenC> derekS: Yeah, that's nothing to do with grub, that's a softlockup
[10:46] <BenC> derekS: edit the grub command line from the grub menu, remove "quiet splash" and take a screenshot of the failure
[10:48] <derekS> BenC: i am at work and thats on my home computer :) but i can't really take a "screenshot"... i will take a photo thouhg
[10:48] <BenC> derekS: screenshot in this case means digital photo :)
[10:48] <derekS> haha ok
[10:48] <derekS> you haven't seen ab ug like that though?
[11:15] <BenC> derekS: I have, but they are all different
[11:16] <derekS> BenC: allright, i am going hom from work in a bit, i will do it, and post a link :)
[11:17] <kylem> way to go intel
[11:17] <kylem> ipw3945 is broken on 2.6.15 despite them claiming to support it.
[11:22] <BenC> derekS: Just file a bug against linux-source-2.6.19
[12:05] <kylem> there, fixed bloody ipw3945.
[12:08] <BenC> and it's not even a year old yet :)
[12:08] <kylem> heh.
[12:08] <kylem> it annoys me when vendors don't have a VCS for their drivers. :/