- Wednesday, July 23, 2014

release-team@conference.openafs.org

Wednesday, July 23, 2014< ^ >

Room Configuration

Room Occupants

GMT+0
[11:03:36] stephan.wiesand joins the room
[12:43:10] kaduk joins the room
[12:56:35] meffie joins the room
[14:00:06] Marc Dionne joins the room
[14:00:30] <stephan.wiesand> Hi
[14:00:55] <Marc Dionne> hi stephan
[14:01:31] <meffie> hello
[14:01:40] <kaduk> 'mornin'.
[14:01:48] <stephan.wiesand> Thanks for coming. Marc, how's Linux 3.16 doing?
[14:01:49] <kaduk> (kitten is finished, as I predicted.)
[14:04:04] <stephan.wiesand> I had to rebase 11203. Review would be much appreciated, since that one is supposed to go in right after 11162 (which already is).
[14:08:24] <Marc Dionne> no known problems on current 3.16
[14:08:47] <Marc Dionne> with the required patches of course
[14:09:02] <stephan.wiesand> 11308/9, right?
[14:09:11] <stephan.wiesand> (thanks for 11203)
[14:11:06] <Marc Dionne> yes, 11308/309
[14:11:37] <stephan.wiesand> 3.16 ETA is ~2 weeks?
[14:14:11] <stephan.wiesand> Merging is painful. Whenever you start looking through a stack, you soon end up with some dependency having negative (or no) review.
[14:15:08] <stephan.wiesand> Mike, do you have any comments on Andrew's concerns regarding 11252?
[14:15:46] <meffie> looking
[14:17:42] <meffie> i'll have to look over all of his comments. all this was done so long ago.
[14:17:42] Marc Dionne leaves the room
[14:18:19] <kaduk> I think they can be summarized as "gratuitous changes in the pullup that were not in the original commit"
[14:19:08] <meffie> oh.
[14:20:31] meffie leaves the room
[14:20:36] <kaduk> (But I don't really  know what the difference between afs_read and OS_READ is.)
[14:20:41] meffie joins the room
[14:22:16] Marc Dionne joins the room
[14:23:59] stephan.wiesand leaves the room
[14:24:30] stephan.wiesand joins the room
[14:25:16] <Jeffrey Altman> Hi.
There is no "afs_read".   I think you meant "read".    OS_READ() is a macro that determines whether read() or nt_read() should be used.     The 1.6 branch is broken on Windows and that is no surprise.   I don't believe 11252 should be trying to fix it.
[14:26:54] <meffie> i think i made an error in the cherry pick. i gave it a -1 and need to push a corrected commit
[14:29:11] <Jeffrey Altman> the issue on Windows is that none of the CRT functions will do the right thing for file I/O for the Windows variant of namei.   The native Win32 APIs need to be used instead.
[14:33:55] <kaduk> Does anyone remember offhand whether afs_osi_Alloc is supposed to return zero'd memory?  I think I remember Simon telling me that, but it would be hard to search for.
[14:34:12] <stephan.wiesand> Thanks for the explanation. It seems this doesn't really matter. But there are more concerns to be addressed throughout the stack (at least 11257,11258, 11263, 11266, 11268), so this could just as well be cleaned up too.
[14:38:14] <meffie> kaduk: i dont think so, but i'd have to look
[14:41:53] <kaduk> Well, our calloc stub for in-kernel hcrypto calls afs_osi_Alloc...
[14:42:57] <stephan.wiesand> That's master only?
[14:43:03] <kaduk> ... and in my debug freebsd kernel build, that backends to malloc(9) and does not request zeroed memory, so i get back space filled with 0xdeadc0de, which is not really what one wants to return from calloc
[14:43:15] <kaduk> I think that's master-only, yes.
[14:43:27] <kaduk> (Feel free to tell me to go to the other room)
[14:43:36] <stephan.wiesand> Nono.
[14:43:58] <stephan.wiesand> I just wanted to make sure that I don't have to panic :-)
[14:44:17] <kaduk> :)
[14:44:57] <meffie> 0xdeadc0de, nice
[14:50:47] stephan.wiesand leaves the room
[14:51:17] stephan.wiesand joins the room
[14:56:42] <stephan.wiesand> 'd like to merge 11158  and 11282, and then repush the whole parallel-build stack, inserting a pullup of 11160 right after 11228. Does this make sense?
[14:57:47] <kaduk> Seems to make sense from here.
[14:59:36] <meffie> yes, i think so
[14:59:51] <Jeffrey Altman> I think that is reasonable
[15:02:49] <stephan.wiesand> Thanks. Maybe we can make some progress then.
[15:21:08] <stephan.wiesand> And since the rmainder of stack-reduction needs to fixed and/or rebased now anyway, I just merged 11181/2/3. I also merged the Linux 3.16 changes - we'll need them soon.
[15:23:15] <stephan.wiesand> I have to leave now. Thanks for the chat. I still hope we can return to "normal" meetings eventually...
[15:23:24] stephan.wiesand leaves the room
[15:30:25] <kaduk> Is there any news from Simon?
[15:52:10] meffie leaves the room
[17:09:17] Marc Dionne leaves the room
[22:41:24] kaduk leaves the room