Home
release-team@conference.openafs.org
Thursday, June 9, 2022< ^ >
kaduk@jabber.openafs.org/barnowl has set the subject to: openafs release team
Room Configuration
Room Occupants

GMT+0
[13:51:11] Cheyenne joins the room
[15:58:58] meffie joins the room
[15:59:07] <meffie> greetings.
[15:59:41] <kaduk@jabber.openafs.org/barnowl> greetings
[15:59:47] <Cheyenne> Hello all
[16:00:15] <mvitale> howdy, just lurking - ping me if you need me for anything, otherwise my attention is elsewhere…
[16:09:05] <kaduk@jabber.openafs.org/barnowl> I have houseguests this weekend so have been a bit distracted
preparing for that.
I saw a bunch of updates land in gerrit that I want to look at; I
merged most of the bits from the "try building the linux kernel module
with clang" stack and am looking at the latest linux-5.18 now.
[16:09:16] <kaduk@jabber.openafs.org/barnowl> I know that there is also the big 1.8 stack I should look at too...
[16:10:21] <Cheyenne> yeah.. the kernel module built with clang is a bigger can of worms.. at the moment the only real way to get a build is to modify a Makefile
[16:10:51] <kaduk@jabber.openafs.org/barnowl> Yeah.
[16:10:59] <Cheyenne> but "forcing" the build with clang did expose some generic problems
[16:11:02] <kaduk@jabber.openafs.org/barnowl> But clang is still a decent static-analysis tool :)
[16:11:09] <Cheyenne> +1
[16:11:42] <meffie> ben sorry i made ticket #135302 by accident. please delete it :(
[16:13:00] <kaduk@jabber.openafs.org/barnowl> Will look, if the page ever loads...
[16:13:06] <Cheyenne> On the linux-5.19 front.. there are some required changes.. but they are all minor (the autoconf stuff is probably where most of the work will be needed)
[16:13:31] <meffie> "The Linux kernel has always traditionally been compiled with GNU toolchains such as GCC and binutils. Ongoing work has allowed for Clang and LLVM utilities to be used as viable substitutes. "
[16:13:35] wiesand joins the room
[16:14:22] <Cheyenne> Yes.. I think there are a few distros that are just now starting to support a clang built kernel
[16:14:33] <wiesand> Hello. Very sorry for being late. That railroad crossing between my office and my home is completely unpredictable…
[16:14:46] <meffie> welcome wiesand
[16:14:57] <Cheyenne> Hello Wiesand :)
[16:16:10] <wiesand> On the other hand, I don't have any topics today anyway.
[16:16:51] <Cheyenne> wiesand: I hope that the stack 1.8.x that I pushed is what you were looking for.
[16:17:20] <kaduk@jabber.openafs.org/barnowl> (Well, that stack plus a bunch of reviews of it from us folks,
presumably...)
[16:17:43] <Cheyenne> yeah.. more reviews would be helpful :)
[16:18:44] <mvitale> sorry I haven't gotten to them yet, Cheyenne, but I will
[16:18:48] <meffie> i think it would help to put in a gerrit comment why commits were cherry-picked for stable. many times it is not clear why.
[16:19:37] <Cheyenne> I tried to use the topic name (where I could edit it) for the reason.  
[16:20:22] <wiesand> Right. But thanks a lot for making that happen. While I'd like to have more review (it's not a small stack…), I think it's time I start testing it (and figuring out in which order to merge - or is that strictly by gerrit change number?)
[16:20:26] <meffie> no, not just a name, i mean, like an explanation with human words. like if it is to allow something later to cherry-pick cleanly, which ones?
[16:21:11] <Cheyenne> wiesand: the order has already been done (though I think 2 commits might need to be swapped)
[16:21:13] <meffie> the merge order is set by the git history! the parents
[16:21:51] <meffie> just be sure to checkout the "top" commit. can you make it clear in gerrit which one is the "top" commit cheyenne?
[16:22:15] <Cheyenne> 14946 is the "top" commit
[16:23:01] <kaduk@jabber.openafs.org/barnowl> I don't think the gerrit UI gives a super-great option for finding the
"top" commit directly at the moment.
[16:23:03] <Cheyenne> (just updated 14946)
[16:23:05] <kaduk@jabber.openafs.org/barnowl> So mentioning it here is good :)
[16:23:18] <Cheyenne> (comment that is)
[16:23:18] <meffie> i feel the "topic" should be something like openafs-1.8.9 or some such :)
[16:23:49] <mvitale> yes, that's a pet peeve about gerrit (there are so many)  - the order is difficult to discern
[16:24:12] <meffie> gerrit "Related Changes" works to show the stack in order, if everything is on a single linear branch, like this stack.
[16:24:22] <wiesand> Ugh. The whole "compiler-warnings/errors" topic seems to have way higher gerrit numbers?
[16:24:48] <meffie> the gerrit numbers are "random", you have to look at the git history (parents)
[16:25:32] <Cheyenne> wiesand: yes -- the current order is to avoid having to do any conflict resolving.
[16:26:04] <Cheyenne> I did note in the commit messages if there was a conflict that needed to be resolved.
[16:26:13] <wiesand> I'll have a look. Thanks.
[16:27:39] <Cheyenne> Turned out that with the stack I came up with there are only a few conflicts.  Some due to patches already being in 1.8.x and 1 or 2 because I didn't think pulling in the patches to resolve the conflicts were "proper" for 1.8.x at this time
[16:28:14] <wiesand> Good news!
[16:29:05] <wiesand> 14974 failed to be verified by buildbot? Doesn't look like a random hickup at first glance?
[16:29:57] <kaduk@jabber.openafs.org/barnowl> Tangenting off the "conflicts" topic, a thought that occured to me for
the next-stable branch, is that we could say that we'll decide when a
change is merged to master, whether or not it should go to stable, and
do the merges to stable in the same order as merges to master.  So
this meeting could include a section for "look at what was merged to
master since last time and talk about whether it goes to stable".
[16:30:46] <meffie> excellent idea.
[16:30:53] <Cheyenne> 14974 is one of the commits I think is "out of order" -- I think it just needs to be swapped around (I will test and make note)
[16:31:15] <kaduk@jabber.openafs.org/barnowl> Yeah, I remember seeing the analogous fix on master recently
[16:31:25] <kaduk@jabber.openafs.org/barnowl> so probably just needs the order swapped
[16:33:39] <wiesand> Ok, Thanks. And yes, that's a very good idea. Alas, we had a similar one a while ago, and did agree on it IIRC, and it didn't really happen. But let's try again.
[16:36:17] <kaduk@jabber.openafs.org/barnowl> So for stable we're at "Cheyenne will investigate and fix up the
commit order, and the rest of us will review"?
[16:37:32] <wiesand> I'm looking forward to next week's workshop. In particular, the FABS presentation. Could solve a real issue we're having…
[16:37:50] <meffie> neat!
[16:37:55] <wiesand> Review review review…
[16:38:22] <Cheyenne> Here are the commits that I had to resolve conflicts with: 14989, 14977, and 14970
[16:38:30] <meffie> i did review, review. i just need to review now.
[16:38:47] <kaduk@jabber.openafs.org/barnowl> meffie: :)
[16:39:12] <wiesand> Yes, Mike already was very diligent. Thank You!
[16:39:14] <meffie> thanks cheyenne. that is encouraging. just 3 commits.
[16:39:15] <Cheyenne> Yes -- I will look at the order (I think it's just swapping 2 commits).
[16:39:28] <kaduk@jabber.openafs.org/barnowl> I think I'm going to need to step out a bit early today.
I don't think I have anything else to bring up for master
[16:40:56] <Cheyenne> meffie: I tried to find the absolute minimum needed to pull in to avoid conflicts, and to get a build on a linux 5.17 with the clang and gcc compilers that are on Fedora xx (I forgot exactly which fedora it was)
[16:41:03] <wiesand> I don't have any further topics either. And I need to distribute a few cubic meters of water in the garden to prevent it from turning into a desert.
[16:41:30] <meffie> cubic meters! that sounds like a lot.
[16:41:36] <wiesand> Cheyenne: perfect strategy IMHO!
[16:41:41] <kaduk@jabber.openafs.org/barnowl> It does indeed sound like a lot!
[16:41:47] <Cheyenne> kaduk_jabberopenafsorg: let me know if you want the commit message updated for 14903
[16:42:18] <wiesand> A "full" round of watering is about 9 m^3
[16:42:38] <kaduk@jabber.openafs.org/barnowl> I'll try to take a look later today.  Thanks for the clarifications
there, it looks like I now have enough data to figure out what I want
to do (i.e., where I'm confused)
[16:43:51] <Cheyenne> and as I mentioned earlier, linux-5.19 will require some changes, but nothing huge.. most of the work will be in autoconf checks and using the right mixture of #if's
[16:44:07] <kaduk@jabber.openafs.org/barnowl> Ah, so "the usual fun"
[16:45:11] <kaduk@jabber.openafs.org/barnowl> Anything else for today?
[16:45:20] <Cheyenne> yeah.. the biggest is that a parameter was removed from one of the address space operator functions.  The other change is an AOP function was replaced (renamed/different parameter)
[16:46:26] <meffie> i have nothing. i sent mail about the workshop talk.
[16:46:43] <Cheyenne> Nothing more from me.
[16:47:20] <kaduk@jabber.openafs.org/barnowl> Motion to adjourn?
[16:47:28] <Cheyenne> 2nd
[16:47:50] <wiesand> Let's adjourn then. Thanks a lot everyone. CU at the workshop!
[16:48:13] <kaduk@jabber.openafs.org/barnowl> Thanks everyone, see you next week.
And wiesand: good luck with the desert-avoidance project!
[16:48:14] <meffie> thanks  have a good evening wiesand
[16:48:37] <Cheyenne> have a good week everyone..
[16:50:15] <meffie> smoketest of 14946 (and everything under it) passes for me.
[16:50:59] <Cheyenne> I've been running 14946 on my day-to-day system
[16:51:15] <meffie> oh, i run master. i could do that instead...
[16:51:42] <meffie> thanks all.
[16:51:45] meffie leaves the room
[18:52:21] wiesand leaves the room
[23:02:36] Cheyenne leaves the room