Brian Behlendorf

Right, the `receive-o-x_props_override` failure is unrelated. @asomers I believe it used to be intermittent, but now we're hitting it reliably on FreeBSD main (and only there). If you be chance...

> I've opened openzfs/zfs-buildbot#232 to install rsync on all the test platforms. I'm a little surprised we weren't already using rsync, it's a completely reasonable addition so I've merged the...

> We should have the new TX type, but before we introduce it, figure out an easy-to-use abstraction for 'ZIL replay of this dataset will require the following zpool features'....

> To be honest we might just want to make a single TX_RENAMEAT2 so we don't need to make any new TX types In the end, we opted to add...

> I guess we should return -EINVAL if you have flags != 0 with renameat2 but without the new feature being enabled in the pool? Right, that seems to fit...

> I was hoping that I could unify the implementation of both features as much as possible Yes, I agree that's definitely the way to go. The feature can be...

> One issue I had is that all of the ZFS features are defined in zfeature_common.c which means that FreeBSD will think it supports this feature even though it doesn't....

> Yeah I've just wrapped it in #ifdef __linux__. It would be non-trivial to implement. Sorry about the tardy reply. That should work fine, and makes good sense. If we...

@GregorKopka indeed, this does look like a possible deadlock on a dbuf. It would be useful to know if this can be reproduced in master since some of the relevant...

> I made a PR for building for a mac on GitHub https://github.com/openzfsonosx/openzfs-fork/pull/3 That's great. My preference would be to leverage the GitHub hosted runners for testing if possible, and...