[00:22:30] --- Russ has left: Disconnected
[00:38:01] --- dev-zero@jabber.org has become available
[00:38:37] --- dev-zero@jabber.org has left
[01:06:13] --- dev-zero@jabber.org has become available
[01:11:20] --- haba has left
[01:25:21] --- kaj has left
[01:26:12] --- mho has left: Lost connection
[02:10:38] --- dev-zero@jabber.org has left
[02:46:23] --- jaltman has left: Replaced by new connection
[02:46:24] --- jaltman has become available
[03:42:03] --- dev-zero@jabber.org has become available
[03:42:38] --- dev-zero@jabber.org has left
[03:52:58] --- dev-zero@jabber.org has become available
[05:14:13] <Simon Wilkinson> Is our website a git module now? Did someone throw the switch?
[05:35:17] <jaltman> the web site is maintained in the openafs-web.git repository.   the contents are published from there on grand.
[05:36:51] <Simon Wilkinson> Ah. Cool. I'd missed that happening - I thought we were still on CVS, and git was just tracking that.
[05:41:11] --- jaltman has left: Disconnected
[05:41:19] --- jaltman has become available
[05:44:12] <jhutz@jis.mit.edu/owl> Really, we should throw the switch the rest of the way, so grand isn't
involved.
[05:44:23] <jhutz@jis.mit.edu/owl> But not today.
[06:41:11] --- dev-zero@jabber.org has left
[08:01:26] --- dev-zero@jabber.org has become available
[10:31:46] --- dev-zero@jabber.org has left
[12:32:02] <kaduk@mit.edu/barnowl> Simon: it looks like the preprocessor logig in e2b65732 around
the declaration of struct ioctl_args for the XBSD case is not quite
right -- it's already declared in sysproto.h for the FBSD50 case,
so I get a build error.
[12:33:19] <shadow@gmail.com/owlBD8333E1> got a fix you can push?
[12:34:16] <kaduk@mit.edu/barnowl> I'm not sure what the fix would be, yet.
[12:34:47] <kaduk@mit.edu/barnowl> (I do have a couple of other fixes to other files that I should
be pushing, though ...)
[12:35:14] <shadow@gmail.com/owlBD8333E1> i could suggest a fix if it were clearer to me what the problem is
[12:36:53] <kaduk@mit.edu/barnowl> Mostly I'm not sure what the other BSDs should have.
[12:39:24] <kaduk@mit.edu/barnowl> I almost think it should just be elif defined(AFS_FBSD50_ENV)
instead of elif !defined(), but that's kind of a guess.
[13:21:58] <kaduk@mit.edu/barnowl> So, as I'm going through and helping the FBSD build on its way,
I expect I'll amass a number of small changes to a bunch of files.
Do I want to group these before submitting to gerrit, or submit them
individually, or just "use my judgement"?
[13:34:19] <Simon Wilkinson> kaduk: Use your judegment
[13:35:09] <Simon Wilkinson> The only thing would be that we don't really have the means to easily verify changes on the BSDs, so we'll have to either trust you, or let them back up in gerrit until we get someone to verify them as a batch
[13:35:27] <Simon Wilkinson> It's worth adding Matt Benjamin as a reviewer on BSD changes, too.
[13:36:00] <jhutz@jis.mit.edu/owl> Oh, yeah; I read that, didn't see who sent it, and so resisted saying
"use your judgement".  But yes, that's definitely what you should do.
[13:36:46] <Simon Wilkinson> 100,000 line patches = bad. Anything else, whatever you feel is right :)
[13:38:08] <shadow@gmail.com/owlBD8333E1> and we'll do the best to figure out what we should and shouldn't push
[13:38:15] --- deason has left
[13:38:18] <shadow@gmail.com/owlBD8333E1> or find someone to try it
[13:38:45] <kaduk@mit.edu/barnowl> Okay; I'll probably bias towards smaller patches since there's not
a lot of testing infrastructure.
[13:38:49] --- deason has become available
[17:51:30] <kaduk@mit.edu/barnowl> How reasonable is it for the length of the rxepoch_free queue to
be several hundred?
[17:51:59] <shadow@gmail.com/owlBD8333E1> um. probably better than them being all used up?
[17:53:52] <kaduk@mit.edu/barnowl> I am getting a kernel page fault with the rxevent_lock and glock held,
and a trace that has some seven hundred copies of
rxepoch_free() at rxepoch_free

(Also, I need to disappear, now.)
[17:55:29] <shadow@gmail.com/owlBD8333E1> that sounds less reasonable
[19:46:48] <kaduk@mit.edu/barnowl> Yeah ... my hypothesis was that the queue wasn't getting
initialized and it was following garbage into who-knows-where.
But I would expect that to die a bit faster, now that I think about it.
[21:02:11] --- Russ has become available
[21:05:08] --- jaltman has left: Replaced by new connection
[21:05:09] --- jaltman has become available
[21:08:00] --- jaltman has left: Disconnected
[21:26:36] --- jaltman has become available
[21:28:09] --- Russ has left: Disconnected
[22:06:03] --- Russ has become available
[22:52:39] --- haba has become available
[22:52:46] --- reuteras has become available
[23:02:27] --- deason has left
[23:58:48] <kaduk@mit.edu/barnowl> Hm, FBSD/osi_prototypes.h isn't being included in the rx_event.c build, so
there's no prototype for osi_fbsd_alloc, and the 32/64-bit disparity between
int and pointer is eiting me.