In D54424#1243918, @asomers wrote:

This looks fine. But as the default timeout is already 300s, I suggest increasing it even more. If it were me, I would double it.

In D54016#1243870, @des wrote:

What's the point of copying the NetBSD tests if we end up having to change everything about them, and then end up adding our own test script anyway as you do in D54265?

There is really no way to tell beforehand if this timeout is enough.
Reference: https://ci.freebsd.org/view/Test/job/FreeBSD-main-aarch64-test/1867/testReport/junit/usr.sbin.makefs/makefs_zfs_tests/file_extend/

There is really no way to tell if this timeout is enough.
Reference: https://ci.freebsd.org/view/Test/job/FreeBSD-main-aarch64-test/1867/testReport/junit/usr.sbin.makefs/makefs_zfs_tests/compression/

Thank you!
I wanted as few diffs with NetBSD as possible, but using nitems() is a good suggestion; I'll probably submit it to GNATS.

A similar change has been included in the latest vendor import of mtree.
Thank you!
https://cgit.freebsd.org/src/commit/?id=f8cee1f2c2dfba6223385fd711cce9faeca76451

No longer applicable.

• Round up the used inodes percentage as well. Although not required by POSIX, it would be odd to just have one percentage rounded up.

• Keep the same NetBSD file name, we'll branch off our tests on another file.

Upstream has accepted our patch:
https://github.com/openssh/openssh-portable/commit/a22ff3c6f11edd00c19981f9cb85d3b25d305a56

Closed by commit 30aa5507f349f761770c5cf49989d6cf7eb3498d

I wonder if changing it to: tags=package=bsnmp,dev solves the issue?

NetBSD has fixed the output of their tests to match POSIX behavior (The percentage value shall be expressed as a positive integer, with any fractional result causing it to be rounded to the next highest integer) (lib/59811).

Upstream has committed a similar fix. It will land on the next vendor import. Thanks!

"... as new ways to build are rolled out" nice!