[00:55:01] --- abo has left [01:32:21] --- Russ has left: Disconnected [01:36:26] THe openafs.git in AFS should be getting updated. We should push into it in roughly the same way we push into everything else. I guess something somewhere must have broken. [03:55:57] --- haba has become available [04:12:41] --- haba has left [04:13:48] --- haba has become available [04:29:49] --- kaj has left [05:05:16] --- reuteras has left [05:41:43] --- kaj has become available [06:46:19] --- deason has become available [07:03:05] --- tharidufernando has become available [07:05:57] --- tharidufernando has left [07:09:59] --- kaj has left [07:10:05] --- kaj has become available [07:40:41] --- kaj has left [07:40:45] --- kaj has become available [07:42:48] --- mho has become available [07:48:55] --- haba has left [08:51:10] --- kaj has left [09:44:02] --- jaltman has become available [09:54:42] --- kaj has become available [10:05:44] --- jaltman has left: Disconnected [11:02:35] --- Russ has become available [12:15:20] --- jaltman has become available [12:44:13] I'm working on a win7 64-bit install. Once that's done, I'll be testing out the 64-bit afslogon.dll [12:46:39] --- haba has become available [13:21:06] --- haba has left [13:25:33] I kinda feel like I want to get rid of afs_inet_ntoa from the entire source, not just threaded stuff; is that a bad idea? [13:26:37] just so it doesn't get used in something that eventually turns into someting that must be threadsafe or whatnot [13:26:52] (there's only about 4 callers once it's gone from ubik) [13:27:02] It's probably not a bad idea. [13:27:25] In the long run, we're going to want to eliminate inet_ntoa of any variety entirely in favor of inet_ntop. [13:34:22] I can't see any downside to just killing afs_inet_ntoa [13:41:34] --- haba has become available [14:07:48] --- haba has left [15:47:38] --- deason has left [16:42:32] --- steven.jenkins has left [16:43:56] --- steven.jenkins has become available [16:45:26] deason: do you mean converting afs_inet_ntoa to afs_inet_ntoa_r or are you thinking of something else? [17:17:12] I think he means gerrit 1680 [17:22:37] --- deason has become available [17:49:19] so the answer is converting afs_inet_ntoa to afs_inet_ntoa_r. I am fine with that. [18:33:40] --- Russ has left: Disconnected [18:48:54] --- Russ has become available [19:04:19] Ah, interesting. I had not seen http://openefs.org/ before. [20:01:07] huh [20:20:26] it's fairly new [20:30:03] Clicking on the core team will show why it all sounds very familiar. [20:30:35] "Key Project Staff", that is. [20:32:17] openefs.com also exists [20:36:54] had to go down to the Spaten beer tasting, so did not yet get to test out the afslogon.dll on win7-64 [20:38:06] ick... efs folk need to learn how to make a webpage with columns [20:38:27] "had to" [20:38:39] yes... my arms were being twisted :-D [20:39:23] the 4 beers were really good [20:40:04] sure [20:40:30] 2 lagers (one light, one dark), the optimator, and a wheat beer [20:45:16] i hate you [20:47:26] to make you feel better, the dark lager (dunkel) was supposed to be on tap, but Spaten uses some odd tap connector that the bar did not have, so we had to settle for the bottle [21:10:50] Bah, I am blanking for how to turn a gerrit change into a patch. I should be able to 'git fetch origin HEAD:refs/changes/83/1683/1', I think. But where do I go to get the SHA1? [21:11:23] probably git log FETCH_HEAD [21:13:32] Hm, that seems to be giving me master only. I bet my fetch line is not quite right ... [21:15:06] --- jaltman has left: Disconnected [21:15:43] Indeed, I did not want the HEAD: prefix. [21:18:40] ah, yes [21:23:05] --- jaltman has become available [21:28:49] Interesting. If I 'aklog athena zone sipb', I get tokens for athena but not the other two, before aklog dumps core. [21:29:57] try lwp-debug build? [21:30:48] Right, I was going to do that. [21:31:23] Unfortunately, the condition variable change doesn't seem to fix my hang on read (unless I failed to properly apply the patch ...) [21:32:34] i can read files. [21:33:25] Fascinating. Maybe I will pull in a clean snapshot instead of just patching my current tree (but I don't think I have any significant local changes) [21:33:51] It does make it much farther through shutdown, though. [21:34:28] umount /afs gets to "done" but hangs [21:35:41] I am pretty sure I'm losing because my hanging cat has vnode locks held or some such [21:36:06] ah. maybe [21:37:12] (I was rebooting, instead of just unmounting /afs . All the afsds were zombies but cat and file were in afs and afsslp) [21:52:40] --- Born Fool has become available [21:54:34] configure'd with --enable-debug and --enable-debug-lwp aklog still dumps core: Assertion failed: (arena != NULL), function arena_dalloc, file /usr/src/lib/libc/stdlib/malloc.c, line 3782 [22:02:28] backtrace? [22:02:37] garbage [22:03:25] Does Mac OS X have something like valgrind? [22:03:34] this is freebsd. [22:03:38] Oh, sorry. [22:03:48] I should have known that. [22:04:08] I bet valgrind doesn't support FreeBSD. [22:04:42] \exists /usr/ports/devel/valgrind ... [22:04:53] You want that, then. [22:05:00] Debugging this kind of crap is way easier with valgrind. [22:05:10] I hate heap clobbers. [22:05:35] Yeah. But let me see if I can unmount /afs when I don't have any vnodes locked, first. [22:05:59] yeah, the vflush could be failing [22:07:35] Well, I don't have an /afs in df or mount anymore. [22:07:51] But umount appears to be hung. [22:10:48] [mntref] Hm, and gdb hangs on attach ... such fun. [22:11:06] ( I guess the latter is not actually surprising) [22:19:35] ... general protection fault in kernel mode? [22:19:47] (when starting afsd) [23:15:19] --- Born Fool has left [23:21:26] --- deason has left