[06:15] <lamont-away> *** glibc detected *** double free or corruption (!prev): 0x00021050 ***
[06:15] <lamont-away> grumble
[06:15] <lamont-away> iproute build on hpa
[06:15] <lamont-away> hppa
[06:22] <lamont-away> make[1] : Entering directory `/build/buildd/klibc-1.0.14'
[06:22] <lamont-away> MCONFIG:90: klibc/arch/parisc64/MCONFIG: No such file or directory
[06:22] <lamont-away> make[1] : *** No rule to make target `klibc/arch/parisc64/MCONFIG'.  Stop.
[06:22] <lamont-away> jbailey broke my kernel
[06:23] <fabbione> lamont-away: check if there is a parisc dir instead
[06:23] <fabbione> i had a similar issue with sparc/sparc64
[06:24] <lamont-away> yep
[06:24] <fabbione> and check the contents of the 2 :)
[06:24] <fabbione> in sparc64 there were basically empty files
[06:24] <fabbione> the sparc one was correct
[06:24] <lamont-away> heh
[12:32] <doko> infinity: ping
[12:33] <doko> any way to get the config.log from the ncurses upload for i386? I cannot reproduce it here on a real 32bit system
[01:07] <infinity> As opposed to a fake one?
[01:09] <infinity> Oh, I had that same problem with glibc.
[01:10] <infinity> doko : s/-m64/-m64 -march=x86-64 -mtune=x86-64/
[01:10] <infinity> doko : gcc-opt on the buildds sets -march/-mtune to CPUs that GCC thinks can't do -m64, so you need to reset it.
[01:11] <infinity> doko : Either that, or fix GCC to stop whining. :)
[01:19] <doko> ohh, thanks
[01:51] <doko> infinity: is this necessary for powerpc as well?
[03:29] <infinity> doko : We don't use -mtune and -march on PPC...
[03:29] <infinity> doko : PPC looks to be failing for entirely different reasons.
[03:30] <doko> yes, I see it. strange. and that's not even the -m64 build ...
[03:33] <infinity> Yeah, I'm giving it a retry to see if it's just the kernel hating me.
[03:33] <infinity> If it fails the same, I'll give it a spin on a PPC machine at home where I can debug a bit.
[03:34] <doko> infinity, lamont-away: gcc-opt should not add anything, when it sees -m64 or -m32 ...
[03:45] <infinity> Okay, NOW it failed on the configure check.  Meh.  Time to save a build tree and look at it, I guess.
[03:59] <infinity> Erm.   cc1: error: invalid option 'arch=powerpc64'
[03:59] <infinity> doko : But we don't set -mtune and -march on powerpc anyway, so you shouldn't need to try to reset them.
[04:00] <doko> ok
[04:00] <doko> that was ncurses?
[04:01] <infinity> Yes.
[04:02] <infinity> Alternately, find out what GCC sets as the "baseline" on -m64, and set it to that.
[04:02] <doko> we should fix that stupidity on the buildd ...
[04:02] <infinity> That's what I did for glibc on amd64, to make sure we got what we wanted.
[04:03] <infinity> s/fix/alter/
[04:03] <doko> lamont-away, wake up! ^^^
[04:03] <doko> :)
[04:03] <infinity> I'll make sure that we have -mtune/-march parameters for "default", "-m32" and "-m64"... But right now, we don't.
[04:14] <doko> yes, I think we only need them for a few packages now
[04:19] <doko> infinity: please could you requeue openoffice.org2 on powerpc? I didn't see that failure before ...
[06:13] <lamont-away> doko: yeah - I know
[06:14] <lamont-away> doko: we taught it to not add -march for glibc - what else?
[06:14] <lamont-away> I need to actually teach it about -m{64,32} equivalence stuff
[06:18] <doko> I think that's all. At least for the three packages I build ...
[08:07] <elmo> lamont/infinity: FYI, I'm firing up a rebuilt of main on breezy-at
[08:23] <jbailey> elmo: Will the results for that just be in the usual ~lamont/... ?
[08:23] <elmo> I don't know how the logging stuff works, sorry
[08:23] <jbailey> 'k.
[08:24] <jbailey> lamont-away, infinity: ^^