- Friday, February 19, 2021

release-team@conference.openafs.org

Friday, February 19, 2021< ^ >

kaduk@jabber.openafs.org/barnowl has set the subject to: openafs release team

Room Configuration

Room Occupants

GMT+0
[11:42:37] mbarbosa joins the room
[16:30:46] meffie joins the room
[16:51:34] yadayada joins the room
[16:55:05] cwills joins the room
[16:57:24] wiesand joins the room
[17:01:10] <kaduk@jabber.openafs.org/barnowl> greetings
[17:01:20] <yadayada> Hello All
[17:01:40] <cwills> Hello :)
[17:01:42] <mvita2> hi
[17:01:52] <wiesand> Hi
[17:02:06] <meffie> hi
[17:02:58] <wiesand> 1.8.x news is fairly brief today:
[17:03:12] <wiesand> merged away solaris-kstats
[17:03:31] <wiesand> pulled up fbsd changes
[17:04:21] <wiesand> the latter is almost complete (two later changes make trouble due to code skew), and everything in gerrit for 1.8.x are clean cherry-picks
[17:05:43] <yadayada> I have some update on 14082. If you remember we were seeing some crashes using Falcon LSM on centos8. With 14082 crashes are resolved.
[17:06:10] <wiesand> 14529 is a bit large, but only should remove unused code (and the commit mentioned in "Commit 123f0fb1 (config: remove support for old FreeBSD releases)
removed our support for FreeBSD releases before FreeBSD 8.") is on 1.8.x already.
[17:06:40] <cwills> (I'm at file 18 of 62)
[17:07:11] <wiesand> thanks for jumping at those :)
[17:08:13] <meffie> thanks for the report yadayada
[17:08:17] <wiesand> yadayada: good news, I hope I'll keep that in mind when writing the 1.8.8 NEWS draft
[17:08:51] <cwills> Oh.. there was a firewall problem with some of the buildbot workers that was resolved yesterday.  The symptom/problem was a ping failure against the workers
[17:10:38] <wiesand> didn't notice
[17:11:02] <kaduk@jabber.openafs.org/barnowl> I think I barely noticed but didn't actually look or get disrupted by
it
[17:11:14] <mvita2> new submissions couldn't be verified by buildbot
[17:12:40] <meffie> they were just queued for a bit.
[17:13:44] <meffie> thanks for getting the network fixed.
[17:14:04] <wiesand> there was a timeout processing 14525 (for which Cheyenne scheduled a rebuild, thanks), but that's the only glitch I noticed
[17:14:42] <cwills> that was something unrelated  that I noticed.
[17:14:53] <wiesand> ah, good ;-)
[17:15:36] <wiesand> remaining items on the 1.8.8 wish list: 10831, Linux 5.11 (if there's anything left to do), Big Sur
[17:16:30] <wiesand> My personal next action item is NEWS (and 14539 is going to help quite a bot)
[17:17:06] <wiesand> That's what I have on stable for today.
[17:17:14] <mvita2> nice.
[17:17:19] <mvita2> thank you Stephan!
[17:17:43] <meffie> thank you wiesand
[17:18:44] <kaduk@jabber.openafs.org/barnowl> This week was pretty brutal for me -- we had an IESG telechat
yesterday and maxed out the pagecount for the agenda.
[17:19:03] <kaduk@jabber.openafs.org/barnowl> I am making progress through the BigSur sockproxy change, but am not
done yet.  I really wanted to be done for today :(
[17:19:33] <meffie> thanks ben
[17:19:35] <cwills> master and 1.8.x built cleanly against the finalized Linux-5.11.  There has been no upstream activity in the linux repo since 5.11
[17:22:30] <wiesand> ok, next week we'll start talking about 5.12 then…
[17:24:05] <wiesand> anyway, my prime objective is now to get 1.8.8pre1 out (preferrably with the FBSD catchup and Big Sur support, but we shouldn't block on those IMO)
[17:25:27] <wiesand> NB what's IESG?
[17:25:53] <kaduk@jabber.openafs.org/barnowl> The Internet Engineering Steering Group, aka the thing that has been
soaking up "all my time" for the past three years
[17:27:58] <wiesand> Ah, I thought that was IETF… keep confusing those abbrevs…
[17:28:37] <kaduk@jabber.openafs.org/barnowl> The IESG is roughly the technical management board of the IETF
[17:28:51] <wiesand> thanks ;-)
[17:28:59] <kaduk@jabber.openafs.org/barnowl> So confusion is understandable :)
[17:30:27] <wiesand> So, unless anyone wants to bring up some late additional changes, I think we're done with 1.8 for today. And maybe the same for 1.9/master?
[17:31:03] <kaduk@jabber.openafs.org/barnowl> I just finalized my comments on 14431 (BigSur/sockproxy)
[17:31:11] <meffie> i'm not aware of any late changes for 1.8.8
[17:31:14] <kaduk@jabber.openafs.org/barnowl> So Marcio (&co?) will have some work to do now
[17:31:43] <meffie> hooray! thanks Dr. Kaduk
[17:32:18] <kaduk@jabber.openafs.org/barnowl> you're welcome!  Now I can feel like I can say I did something
productive...
[17:32:55] <kaduk@jabber.openafs.org/barnowl> I see we had some action on some of the static-analysis stuff and the
preprocessor indentation that's currently staged before some FBSD
changes, so hopefully I can get a chance to look at that, the 1.9.1
NEWS, etc.
[17:33:04] <cwills> There was a minor update to the LINUX_ENV cleanup (just fixing a comment).  And I got some feedback some of the static-analysis updates that I've rebased/updated that I need to go through.
[17:34:04] <meffie> thank you cwills.
[17:34:08] <kaduk@jabber.openafs.org/barnowl> Thanks for going through those, by the way.  It is often tedious but
there are some good fixes that they bring in.
[17:34:50] <kaduk@jabber.openafs.org/barnowl> I think that's all I have, though -- just my usual "please let me know
if I should prioritize a particular thing" if I didn't mention it
already
[17:34:58] <wiesand> frankly, I was wondering why a free;malloc w/o checking the success of malloc would be safer than realloc (13156)
[17:35:00] <cwills> I think I have a decent process now to run cppcheck, infer and scan-build with a utility that can compare runs
[17:36:27] <meffie> excellent. we should try to make a buildbot to do that on a regular basis
[17:36:37] <kaduk@jabber.openafs.org/barnowl> > free;malloc w/o checking
parsevnode.c does check the variable that holds the malloc result,
just a bit further down.  Was that the file you were thinking of?
[17:37:07] <kaduk@jabber.openafs.org/barnowl> +1 for trying to have buildbot do that regularly.  Fixing regressions
is easier when you hear about them quickly :)
[17:37:32] <cwills> Yes .. I need to talk with Mike on that :)
[17:38:32] <kaduk@jabber.openafs.org/barnowl> Sounds like a plan :)
[17:38:33] <wiesand> yes, parsevnode.c… but that's just more apprentice level C I'm not up to…
[17:40:09] <kaduk@jabber.openafs.org/barnowl> Don't worry too much about understanding that parsevnode.c code; it's
... not really written to modern practices.
It's effectively just using the 'symlink_buf' as a cheap cache to
mitigate malloc "being really slow", which was true 30 years ago but
isn't anymore.  Code written today would just always malloc and always
free when done.
[17:41:13] <kaduk@jabber.openafs.org/barnowl> But the key point here is that it has one path for when symlink_buf is
non-NULL and another for when it's NULL (i.e., when the allocation
failed), so it will still behave "as intended" even if the malloc
fails.
[17:41:50] <kaduk@jabber.openafs.org/barnowl> Anyway, any other topics for today?  (I'm happy to keep talking about
parsevnode if wiesand is curious, but that need not be part of the
meeting per se)
[17:42:00] <wiesand> thanks for the enlightenment :)
[17:42:20] <yadayada> just small update from my side
[17:42:21] <meffie> i have nothing else
[17:42:38] <yadayada> recently I have been working to make OpenAFS work in secure boot env
[17:43:00] <yadayada> and for that I had to sign kernel modules in our spec files.
[17:43:30] <wiesand> nasty stuff...
[17:43:34] <yadayada> Kindly let me know if those changes can be pushed to our REDHAT spec files ?
[17:43:59] <yadayada> we can make it configurable based on rpmbuild option
[17:44:06] <meffie> sounds great!
[17:44:18] <kaduk@jabber.openafs.org/barnowl> It sounds like that could just be a normal gerrit review?
[17:44:37] <kaduk@jabber.openafs.org/barnowl> (and thank you for working on that!  I am not sure if we have a way to
do that sort of thing for debian)
[17:45:13] <yadayada> for debian since DKMS is there so mainly every build we need to sign module
[17:45:54] <yadayada> but with kmod one module can work across many kernel version due to kernel abi. So for RHEL it makes sense
[17:46:07] <wiesand> I see no reason to reject such a change per se, and yes an rpmbuild option is probably the way to go.
[17:46:26] <yadayada> sure Thanks will submit gerrit this week
[17:47:05] <wiesand> KABI is cool and mostly works. It has its limitations though. Make sure you read the fine print...
[17:47:51] <yadayada> right Since we use blocklist api's so atleast within RHEL minor version I see it mostly working
[17:48:15] <meffie> yadayada, do we need to remove all the old kernel "flavors" in the spec file? are those obsolete?
[17:48:28] <meffie> (it makes for a very complex spec file)
[17:50:05] <yadayada> I need to check on old kernel flavours , but I agree we should clean up obselete stuff. Will check on it while making signing changes
[17:50:34] <wiesand> please make those separate commits though
[17:50:48] <yadayada> sure
[17:51:14] <wiesand> and speaking of it, can the "redhat-buildsys" voodoo be jettisoned then too?
[17:53:21] <meffie> i hope so :)
[17:53:53] <meffie> motion to adjourn to shovel snow?
[17:54:07] <yadayada> nothing else from my side
[17:54:36] <cwills> Nothing from me
[17:55:00] <kaduk@jabber.openafs.org/barnowl> Have fun with the snow, if you still can
[17:55:04] <wiesand> snow shoveling season seems to be over here now, but I did enough of it to sustain that motion
[17:55:36] <wiesand> Enjoy! (I love shoveling snow)
[17:55:56] <wiesand> Thanks a lot everyone!
[17:56:04] <yadayada> thanks
[17:56:08] <meffie> be safe.
[17:56:09] <kaduk@jabber.openafs.org/barnowl> Thanks everyone!
[17:56:15] <cwills> Everyone stay warm and safe
[17:56:27] wiesand leaves the room
[17:56:46] yadayada leaves the room
[18:22:23] meffie leaves the room
[18:51:11] mbarbosa leaves the room
[20:23:01] mbarbosa joins the room
[22:13:02] cwills leaves the room
[22:25:04] mbarbosa leaves the room