[06:15:15] --- Jeffrey Altman has become available [06:35:24] --- Jeffrey Altman has left [06:59:13] --- deason has become available [07:05:28] --- mho has become available [07:16:57] --- ksumner has left [07:19:43] --- ksumner has become available [11:13:13] Sorry. I broke the build. Works fine on x86_64 Mac. Mumble. [11:35:39] Yeah. Heimdal has a change in it that fails on 32 bit platforms. Going to need to get the fix to Love, get it pushed to Heimdal, and then import to OpenAFS. [12:35:27] --- Russ has become available [13:02:58] --- jaltman has left: Replaced by new connection [13:02:59] --- jaltman has become available [13:13:30] --- jaltman has left: Disconnected [13:22:39] --- jaltman has become available [13:52:26] --- jaltman has left: Disconnected [13:52:33] --- jaltman has become available [15:19:46] --- steven.jenkins has left [15:20:34] --- steven.jenkins has become available [16:04:16] --- deason has left [17:10:14] --- mdionne has become available [17:19:33] --- jaltman has left: Disconnected [17:26:52] --- pod has left [17:48:57] I get a deadlock in lookup with 1.5.77 (on fbsd): thread a (make) (line 534 of osi_vnodeops.c): (kgdb) p vp $14 = (struct vnode *) 0xffffff006d4bc780 (kgdb) p dvp $15 = (struct vnode *) 0xffffff006d4bc3c0 thread b (diff) (line 551 of osi_vnodeops.c): (kgdb) p vp $12 = (struct vnode *) 0xffffff006d4bc3c0 (kgdb) p dvp $13 = (struct vnode *) 0xffffff006d4bc780 [18:05:18] matt may have been wrong about what locking is wanted. i told them this at the freebsd dev summit: your vnode locking is weird. learn from apple. [18:06:46] Heh. Empirically, our current locking is not what they're expecting. [18:07:19] Unfortunately, there are options on its locking, so DEBUG_VFS_LOCKS doesn't help as much as it does elsewhere. [18:12:51] --- Brandon_CMU has left [18:12:57] --- Brandon_CMU has become available [18:13:08] commit d552426b (base64) has me a bit confused [18:14:04] src/util already had a base64.c, and the new rule in the makefile is overridden by the existing one. so the roken base64.c never gets built at all [18:17:51] lemme check. it looked fine here... [18:19:36] no, you are correct. why did this work.... [18:20:21] does anything rely on the new functions? it just compiles the old base64.c, but they don't have the same functions [18:21:59] no, i don't think anything does yet [18:22:03] if I change the Makefile to build the roken base64.c, it has trouble finding config.h. I assume that would be the hcrypto one and not ours? [18:22:18] yes [18:24:13] since there's already a base64.o we might need to rename the object file for the new one. and the hcrypto config.h might need installing so we can reference it in util... [18:52:58] change 2905 would be my attempt to straighten that out [18:58:36] --- phalenor has left [18:59:00] --- shadow@gmail.com/owl238F635B has left [19:02:09] --- phalenor has become available [19:08:03] --- shadow@gmail.com/owl4F808197 has become available [19:21:23] --- mdionne has left [19:50:01] what would be involved with providing buildbot slaves? [19:55:16] --- jaltman has become available [19:59:25] install a buildbot slave with a build environment that can run git and build openafs from source. [19:59:39] then contact Jason for a buildbot slave account and password [20:00:51] I suspect the old base64 was supposed to be removed [20:01:12] windows complains because the old base64.h is no longer present in src/util [20:15:44] --- kula has left [20:44:16] Ah. I bet that if (flags & ISDOTDOT) VOP_UNLOCK(dvp, 0, p); wants to not be in #ifndef AFS_FBSD70_ENV . [21:13:53] --- jaltman has left: Replaced by new connection [21:13:54] --- jaltman has become available [21:14:18] --- Jeffrey Altman has become available [22:02:10] --- shadow@gmail.com/owl4F808197 has left [22:05:04] --- steven.jenkins has left [22:05:46] --- steven.jenkins has become available [22:42:31] --- phalenor has left [22:46:03] --- phalenor has become available [23:16:21] --- phalenor has left [23:26:21] --- phalenor has become available [23:49:31] --- Russ has left: Disconnected