[00:01:58] --- Simon Wilkinson has become available
[00:23:41] --- Simon Wilkinson has left
[00:48:01] --- Simon Wilkinson has become available
[01:40:12] --- Simon Wilkinson has left
[02:10:42] --- Simon Wilkinson has become available
[03:14:00] --- lars.malinowsky has become available
[07:02:05] --- jhutz@jis.mit.edu/owl has become available
[12:49:38] <jakllsch> is parallelized make intended to work?
[12:52:02] <kaduk@mit.edu/barnowl> That's what they tell me.
[12:52:19] <kaduk@mit.edu/barnowl> But there are probably still bugs.
(Last time I tried, Simon found another one.)
[12:57:17] <jakllsch> is there a preferred make variant?
[12:57:33] <jakllsch> or is it intended to be portable to posix make?
[13:00:12] <jakllsch> anyway
[13:00:17] <kaduk@mit.edu/barnowl> I believe we try hard to work with "the system make" for appropriate
values of the same.
[13:35:24] <shadow@gmail.com/barnowlA9683AD7> >        is parallelized make intended to work?

make -j? some of the buildbot slaves certainly use it
[13:38:17] <jakllsch> yes, -j
[13:45:13] <shadow@gmail.com/barnowlA9683AD7> works for me
[14:08:51] <jakllsch> doesn't always for me
[14:08:53] <jakllsch> anyway
[14:09:11] <jakllsch> the cache manager opens 3 sockets?
[14:10:01] <kaduk@mit.edu/barnowl> Please collect build log with e.g. script(1) and send the failing
cases to openafs-devel.
[14:16:38] <jakllsch> can do, just as soon as i get this thing rebooted ... *wanders off to poke devbox*
[14:31:46] <jakllsch> apparently roken.h isn't getting generated soon enough
[14:35:51] <shadow@gmail.com/barnowlA9683AD7> soon enough for what?
[14:36:47] <shadow@gmail.com/barnowlA9683AD7> 3 sockets?
[14:37:48] <jakllsch> that's what fstat(1) (similar to lsof) shows for each afsd fork
[14:38:32] <jakllsch> ../../../openafs-lkm-mk2/src/procmgmt/procmgmt_unix.c:14:19: error: roken.h: No 
[14:38:35] <jakllsch> such file or directory
[14:38:47] <jakllsch> ../../../openafs-lkm-mk2/src/rxgen/rpc_main.c:39:19: error: roken.h: No such fil
[14:39:14] <jakllsch> more erroring due to missing roken.h follow (i did a -j8 on a single proc box)
[14:39:19] <jakllsch> (using nbmake)
[14:42:59] <shadow@gmail.com/barnowlA9683AD7> afsd is not the cache manager. afsd is afsd.
[14:54:14] <jakllsch> right.  does afsd need its own sockets?
[15:13:59] <sxw> jakllsch: If you add $(DIR_roken) to the dependencies for procmgmt and rxgen in the top level Makefile.in, and then regenerate the Makefile, does that solve your problem.
[15:39:00] --- mdionne has become available
[15:46:51] <mdionne> > is parallelized make intended to work?
I know of 2 problem spots currently.  src/libadmin/adminutil, and src/tubik
[17:09:49] --- jaltman has become available
[17:11:05] --- jaltman has left: Disconnected
[17:11:36] --- jaltman has become available
[17:11:55] --- jaltman has left: Disconnected
[17:23:56] --- Russ has become available
[18:11:44] <shadow@gmail.com/barnowlA9683AD7> >        right.  does afsd need its own sockets?

depends on what they are. it does, for example, dns.
[18:12:46] <shadow@gmail.com/barnowlA9683AD7> > jakllsch: If you add $(DIR_roken)

already in http://gerrit.openafs.org/4492
[20:29:08] --- Russ has left: Disconnected
[20:49:33] --- mdionne has left
[20:55:40] <kaduk@mit.edu/barnowl> I'm looking at forward-porting the scripts.mit.edu AFS patch (a single
global pag with daemon.scripts tokens and per-volume isolation of
their use), and came across a block in afs_AccessOK which is not
entirely straightforward. 1.6 has added a check conditional on
afs_InReadDir(avc) which allows read and lookup and nothing else, with
comment:
            /* if we are already in readdir, then they may have read and
             * lookup, and nothing else, and nevermind the real ACL.
             * Otherwise we might end up with problems trying to call
             * FetchStatus on the vnode readdir is working on, and that
             * would be a real mess.
             */
which jhutz added in 2006 as part of linux-afs-translator-xen-20060731.
Can anyone expound on what "a real mess" would be?  (The real question
I am facing is whether to put the scripts-specific check before or
after this InReadDir check.)
[21:26:20] --- Russ has become available
[23:48:36] --- Russ has left: Disconnected