Page MenuHomeFreeBSD

14.0 relnotes: first cut
ClosedPublic

Authored by karels on Sep 20 2023, 11:57 PM.
Tags
None
Referenced Files
F101901264: D41926.id127603.diff
Tue, Nov 5, 9:51 AM
F101901261: D41926.id127609.diff
Tue, Nov 5, 9:51 AM
Unknown Object (File)
Thu, Oct 24, 2:55 AM
Unknown Object (File)
Mon, Oct 7, 6:19 PM
Unknown Object (File)
Oct 4 2024, 8:50 PM
Unknown Object (File)
Sep 12 2024, 1:14 PM
Unknown Object (File)
Sep 11 2024, 3:41 PM
Unknown Object (File)
Sep 9 2024, 2:46 AM

Details

Summary

This draft contains text for most, but not all, of the RELNOTES
entries and Relnotes: yes commits. I have questions pending with
some committers.

Diff Detail

Repository
R9 FreeBSD doc repository
Lint
Lint Not Applicable
Unit
Tests Not Applicable

Event Timeline

Starting a review now because I will be out of the office for 2 weeks starting Sunday.
I'm assuming that I should add a Doc Committers* group, not sure which.
Ed, I added you mostly to review entries for your commits, which are most of the Foundation-sponsored entries.

cperciva added inline comments.
website/content/en/releases/14.0R/relnotes.adoc
149

Should we say "experimental" here? The utility prints a warning to that effect when used. @markj

483

This might not be necessary; the original change switched from BIOS to UEFI (which meant dropping support for some instance types), but I've since switched it to "UEFI if available but BIOS if not" so we're back to the original "can boot everywhere" state (but faster if possible).

website/content/en/releases/14.0R/relnotes.adoc
483

Should this note change? I hadn't mentioned the limitation on instance types.

website/content/en/releases/14.0R/relnotes.adoc
120

we might want to say something like "general NO_SHARED option" (or possibly just not mention it at all)

149

I think that's reasonable. I don't think there's anything to suggest there are potential problems, but it makes sense to mention here if the tool prints it when run.

216–217

I expect to update to 1.13.0 still, fyi

239–240

We might want to clarify this, perhaps:

HPN option handling has been removed from OpenSSH. HPN support was deprecated long ago, but the configuration options were still accepted (and ignored) for backwards compatibility.

291

Maybe "many" rather than "most", there are still more to go.

356

we maybe don't need to mention sndbuf_dma, it's just support for ISA snd cards and doesn't really have user-facing relevance

410

I should have tagged this commit with a sponsored by: The FreeBSD Foundation

494

Most of this is now done in my copy, will push it tomorrow with anything else so far.

website/content/en/releases/14.0R/relnotes.adoc
216–217

Let me know if/when you do. I'll try to note the commit email, but may not keep up on those.

494

I have a question out to Brooks; there were 4 commits, and some were phrased in a contradictory way. I think there were several components removed, but NgATM might cover them all.

I'd also like to mention ZFS-root AMIs (code currently in the tree but images are broken due to ZFS kernel bug) and cloud-init AMIs (coming soon). No need to include them yet though (especially since the cloud-init bits haven't landed yet).

website/content/en/releases/14.0R/relnotes.adoc
149

Right now, disk images generated by makefs's ZFS support trigger FreeBSD kernel panics... that's clearly a kernel bug, but I suspect there's also a makefs bug which is causing it to create images which are broken in a way which triggers the bug. ;-)

483

Maybe change it to say "EC2 instances now boot using UEFI where available, substantially speeding the boot process." and cite bcf9147144f3499976a623f92712ecd0aca71912 as well. I marked the original commit as relnotes: yes because it was a breaking change (support for some instance types was lost) but that functionality was restored in the second commit.

website/content/en/releases/14.0R/relnotes.adoc
483

Got it. btw, should there be a more general summary item about boot speed, not just for EC2? Any suggestion on how to word it?

kbowling added inline comments.
website/content/en/releases/14.0R/relnotes.adoc
364

This had to be reverted unfortunately, maybe for 14.1 if time permits.

Versus 13.2, lem(4) learned:

  • LRO,
  • TSO4/TSO6 (disabled by default)
  • HWCSUM6 (enabled by default).

Versus 13.2, em(4) learned:

  • TSO6 (disabled by default, known issues).

Address review comments so far; add some additional items.

karels marked 11 inline comments as done.

Add doc reviewers; mark items done.

website/content/en/releases/14.0R/relnotes.adoc
364

I219 reverted. question pending on 13.2 changes.

LGTM as the initial pass, we can iterate on it in the tree

This revision is now accepted and ready to land.Sep 21 2023, 4:04 PM
website/content/en/releases/14.0R/relnotes.adoc
364

I think you can just drop this entry for now

website/content/en/releases/14.0R/relnotes.adoc
364

You mean drop the "TSO now enabled" entry above? Isn't that true now?

website/content/en/releases/14.0R/relnotes.adoc
364

Indeed

website/content/en/releases/14.0R/relnotes.adoc
364

"Indeed" I should drop the entry, or "Indeed{ the entry is true now?

website/content/en/releases/14.0R/relnotes.adoc
364

drop it

Dropping e1000 TSO entry as requested.

This revision now requires review to proceed.Sep 21 2023, 6:30 PM
This revision is now accepted and ready to land.Sep 21 2023, 6:40 PM
This revision was automatically updated to reflect the committed changes.