[01:55:06] --- Simon Wilkinson has left [05:24:05] --- Simon Wilkinson has become available [05:33:20] --- Simon Wilkinson has left [05:40:54] --- squinney has become available [06:07:19] buildbot master failed overnight. It has been restarted. [06:30:19] --- Simon Wilkinson has become available [06:53:00] So, am I right that 1.6.2 does not support the latest Fedora 18 kernel? [06:56:25] Is Fedora 18 3.8 ? [06:58:27] 3.8.1-201.fc18 [06:59:47] Yeah, Ken posted a list of those patches to openafs-devel recently. [07:00:26] What we've talked about in the past is doing point releases for Linux versions - so a 1.6.2.1 that's Linux only, and just carries kernel compatibility stuff. [07:00:40] Might be worth bringing that up in release-team [07:01:16] yes, the timing is a little unfortunate [07:01:30] 1.6.2.1 might well be appropriate [07:02:05] It's always a little unfortunate. The problem is that it's not always clear what's going to be in the latest Linux kernel until it's release (whilst some stuff goes through -next, not everything does) [07:02:28] And Fedora, in particular, can be very aggressive about pushing out new kernels. [07:04:25] yes, it always surprises how quickly they appear [07:45:11] --- Stephan Wiesand has become available [08:10:26] --- deason/gmail has become available [08:38:07] --- Stephan Wiesand has left [08:42:30] --- Ken Dreyer has become available [08:48:07] speaking of things, any idea if this is fixed in a newer afs rpm? [08:48:45] file /usr/lib/debug/usr/bin/kpasswd.debug from install of openafs-debuginfo-1.6.1-1.el6.x86_64 conflicts with file from package krb5-debuginfo-1.10.3-10.el6.x86_64 [08:48:58] Hmmm. We were going to fix that. [08:49:09] fixing for 1.6.2 would be... useful [08:49:15] "Too late" [08:49:25] You tested the prereleases, right? [08:49:53] on my fedora box, yes... tho, I don't have the debug packages installed there [08:50:05] I put the --disable-kauth configure argument on master and cherry-picked it for the freebsd packaging, but I think we did not want it to go to 1.6.x (?) [08:50:24] Yeah, I don't think we can start shipping without kauth midway through 1.6 [08:51:05] Right. It's less clear that we can't make the configure knob available, though. [08:51:24] Sure. But it wouldn't help in this case. [08:51:44] I wanna say openafs-kpasswd still conflicted with /usr/bin/kpasswd from krb5-workstation [08:51:57] Yeah, that's deliberate. Pick which one you want. [08:52:18] you can't... krb5-workstation is necessary if you want to use kerberos stuff [08:54:16] If you're using kapasswd, you don't want to use kerberos stuff... [08:54:41] the binary isn't called kapasswd, however [08:54:48] (or at least that was the original logic. It makes migrations interesting) [08:55:05] So, if you're an AFS user from way back, you expect kpasswd to change your AFS password. [08:55:15] if you want kerberos stuff, you use krb5-workstation; if you want ka stuff, you install the ka stuff [08:55:23] the places that want both at the same time are very small [08:55:28] hm, ok [08:55:40] so, back to debuginfo :) [08:56:27] So, what should be happening is that AFS's kpasswd is installed as kapasswd, and has a debuginfo for kapasswd. The openafs-kpasswd RPM then just installs a symlink from kpasswd->kapasswd. [08:56:36] So it's not clear to me what's going on with your debuginfo. [08:56:58] But I suspect that the best way to get it fixed is going to be to supply a patch. [09:00:13] IIRC the RPM does the rename itself, so it might be just that you need to change the ordering of when that rename happens and when the debuginfo is compiled [09:09:29] --- squinney has left [09:16:02] --- Simon Wilkinson has left [09:46:42] --- stephan.wiesand has become available [10:53:58] --- Simon Wilkinson has become available [11:26:51] --- Simon Wilkinson has left [11:49:34] --- Simon Wilkinson has become available [11:52:45] --- Simon Wilkinson has left [12:11:40] --- stephan.wiesand has left [12:18:24] --- meffie has become available [12:20:38] --- ballbery has become available [12:21:55] --- deason/gmail has left [12:22:21] --- deason has become available [13:29:04] --- mdionne has become available [14:04:46] --- ballbery has left [14:40:51] Linux 3.8 support didn’t make it into 1.6.2? Did I miss something? [14:45:41] --- Simon Wilkinson has become available [15:13:37] --- Simon Wilkinson has left [15:55:27] --- meffie has left [16:05:29] that is correct. linux 3.8 is not supported by openafs 1.6.2. It wasn't out when the change cutoff for 1.6.2 was reached. support for 3.8 will be in 1.6.3. [16:09:50] --- deason has left [19:26:09] --- mdionne has left [21:06:43] Hmm. 1.6.2 _userspace_ fails to build in raring too. In fact, so does 1.6.2pre3. I know 1.6.2pre3 used to work, so something changed on Ubuntu’s end. make[3]: Entering directory `/…/src/afsd' gcc -O -I/…/src/config -I/…/include -I. -I. -D_LARGEFILE64_SOURCE -I/…/src/afs -I/…/src/afs/UKERNEL -I/…/src/rx/UKERNEL -DUKERNEL -D_FILE_OFFSET_BITS=64 -I/usr/include/fuse -o afsd.fuse afsd_fuse.o /…/lib/libuafs.a /…/lib/libdes.a /…/lib/libafsutil.a /…/lib/libcmd.a /…/lib/libafsutil.a -lresolv -lfuse /usr/bin/ld.bfd.real: /…/lib/libuafs.a(afs_usrops.o): undefined reference to symbol 'pthread_setspecific@@GLIBC_2.2.5' /usr/bin/ld.bfd.real: note: 'pthread_setspecific@@GLIBC_2.2.5' is defined in DSO /lib/x86_64-linux-gnu/libpthread.so.0 so try adding it to the linker command line /lib/x86_64-linux-gnu/libpthread.so.0: could not read symbols: Invalid operation collect2: error: ld returned 1 exit status make[3]: *** [afsd.fuse] Error 1 make[3]: Leaving directory `/…/src/afsd' [22:04:13] Works with binutils 2.23.1-0ubuntu6, fails with binutils 2.23.1-0ubuntu8. [22:34:36] Bisected to http://sourceware.org/git/?p=binutils.git;a=commit;h=1f3ed420ff362153347d179cfbd3410507870d22 (which I guess means there’s some more work to do). [23:06:15] Rebisected on binutils master to http://sourceware.org/git/?p=binutils.git;a=commit;h=63f3bee644cc30f1e6d049ca54d307a7c63b8186 [23:12:25] whose author claims: http://sourceware-org.1504.n7.nabble.com/Spurious-undefined-reference-error-td217132.html#a218275 “I'll probably be blamed for breaking a number of projects with this change, but of course all I'm doing here is making http://sourceware.org/ml/binutils/2011-08/msg00129.html effective.” [23:13:45] So I guess this is just a standard instance of --no-copy-dt-needed breakage: -lfuse implicitly pulled in -lpthread, but doesn’t anymore.