Page MenuHomeFreeBSD

openzfs: Revert to last known good state.
AbandonedPublic

Authored by des on Aug 16 2023, 6:31 PM.
Tags
None
Referenced Files
F96174282: D41483.id.diff
Mon, Sep 23, 11:24 PM
Unknown Object (File)
Sat, Sep 21, 4:53 AM
Unknown Object (File)
Fri, Sep 20, 3:47 PM
Unknown Object (File)
Fri, Sep 20, 1:48 AM
Unknown Object (File)
Tue, Sep 17, 1:14 PM
Unknown Object (File)
Tue, Sep 17, 10:56 AM
Unknown Object (File)
Tue, Sep 17, 6:03 AM
Unknown Object (File)
Wed, Sep 11, 9:27 PM

Details

Reviewers
mm
mav
Summary

Revert "zfs: cherry-pick fix from openzfs"

This reverts commit 28d2e3b5dedf24938e5ec0806084941510621f13.

Revert "zfs: cherry-pick fix from openzfs"

This reverts commit cd25b0f740f8c846562fd66e7380437eb898c875.

Revert "zfs: set autotrim default to 'off'"

This reverts commit b36f469a15ecf508a2fdc2b58a76f0f2b4a0b88d.

Revert "zfs: merge openzfs/zfs@009d3288d"

This reverts commit 0a97523d467443fa54171ef7a399b1c9043eac75, reversing
changes made to cd9da8d072e4dff4c1eda7307547c6a5a1e5bd35.

Revert "Revert "openzfs: use IN_BASE instead of IN_FREEBSD_BASE in spa.h""

This reverts commit af2939f227db4a80c0ee463adbcdd6608731dacb.

Revert "openzfs: use IN_BASE instead of IN_FREEBSD_BASE in spa.h"

This reverts commit d2a45e9e817ad68f3e163d13404744b8ea9c675b.

Revert "zfs: merge openzfs/zfs@a9d6b0690"

This reverts commit 7b5e687355afe772894661db63d1d4466751d806, reversing
changes made to 27cebb4eac3d03b81defa2785eae350f6caca2c7.

Revert "openzfs: Remove broken symlinks"

This reverts commit 2d38d9839b275a0499cdde08320b9b156846847d.

Revert "zfs: merge openzfs/zfs@8e8acabdc"

This reverts commit f190c36b5d45cbfadc922a69d79628c43cdda22f, reversing
changes made to 229d643c4dd5c1a1695e5de48fab0524132f5deb.

Revert "zfs: merge openzfs/zfs@10e36e176"

This reverts commit b7198dcfc03967cba191a373d99df47ee52d6e2a, reversing
changes made to 2c7279bae7767e9b10a73f44817adb168b69b522.

Revert "zfs: merge openzfs/zfs@feff9dfed"

This reverts commit 4e8d558c9d1cf3e7e424e3fb123b01979c3d57f2, reversing
changes made to 5ca7f02946940ff698741d4da4a30b554cb0f592.

Diff Detail

Repository
rG FreeBSD src repository
Lint
Lint Skipped
Unit
Tests Skipped
Build Status
Buildable 53137
Build 50028: arc lint + arc unit

Event Timeline

des requested review of this revision.Aug 16 2023, 6:31 PM

This makes ZFS usable for poudriere again. It's been broken for two months and there is no indication that anyone is working on a fix.

This makes ZFS usable for poudriere again. It's been broken for two months and there is no indication that anyone is working on a fix.

jrtc27 added inline comments.
sys/cddl/compat/opensolaris/sys/atomic.h
160 ↗(On Diff #126088)

Please keep this change, the code on the right was always totally bogus

When I understand this correctly, 4e8d558c9d1cf3e7e424e3fb123b01979c3d57f2 has broken poudriere? Is there an easy way how to reproduce the bug? I would like to narrow it down to a specific commit.

In D41483#944957, @mm wrote:

When I understand this correctly, 4e8d558c9d1cf3e7e424e3fb123b01979c3d57f2 has broken poudriere? Is there an easy way how to reproduce the bug? I would like to narrow it down to a specific commit.

https://lists.freebsd.org/archives/freebsd-current/2023-August/004162.html

https://bugs.freebsd.org/271945

des marked an inline comment as done.Aug 16 2023, 9:49 PM

I am not a fan of blind reverts like this. I haven't heard about any problems, mentioned to be there for two months, I haven't seen it to be reported to OpenZFS, despite being subscribed there, neither anybody pinged me about the FreeBSD PR. I only noticed it few days ago on current, but haven't looked yet. At this point we are already behind upstream OpenZFS approaching 2.2 release due to 14 branching, and I'd hope us to move forward updating to stable/14 to ZFS 2.2 release and main to ZFS master. I don't think it is reasonable to do extra movements without even trying to do something productive first. Let me at least to take a look, while somebody who has reproduction could try to narrow down the scope to specific OpenZFS commit -- it would a huge help.

In D41483#945037, @mav wrote:

I am not a fan of blind reverts like this. I haven't heard about any problems [...]

See my reply to @mm above.