Re: git: 2a58b312b62f - main - zfs: merge openzfs/zfs@431083f75

From: Cy Schubert <Cy.Schubert_at_cschubert.com>
Date: Tue, 04 Apr 2023 05:07:40 UTC
In message <CAM5tNy4p5UbAkujksyTR=J0tGuDf=Y6v2LuctK4P_p-H3PZpPg@mail.gmail.c
om>
, Rick Macklem writes:
> On Mon, Apr 3, 2023 at 9:00=E2=80=AFPM Cy Schubert <Cy.Schubert@cschubert.c=
> om> wrote:
> >
> > In message <202304031513.333FD6qw014903@gitrepo.freebsd.org>, Martin
> > Matuska wr
> > ites:
> > > The branch main has been updated by mm:
> > >
> > > URL: https://cgit.FreeBSD.org/src/commit/?id=3D2a58b312b62f908ec92311d1=
> bd8536db
> > > aeb8e55b
> > >
> > > commit 2a58b312b62f908ec92311d1bd8536dbaeb8e55b
> > > Merge: b98fbf3781df 431083f75bdd
> > > Author:     Martin Matuska <mm@FreeBSD.org>
> > > AuthorDate: 2023-04-03 14:49:30 +0000
> > > Commit:     Martin Matuska <mm@FreeBSD.org>
> > > CommitDate: 2023-04-03 14:49:30 +0000
> > >
> > >     zfs: merge openzfs/zfs@431083f75
> > >
> > >     Notable upstream pull request merges:
> > >       #12194 Fix short-lived txg caused by autotrim
> > >       #13368 ZFS_IOC_COUNT_FILLED does unnecessary txg_wait_synced()
> > >       #13392 Implementation of block cloning for ZFS
> > >       #13741 SHA2 reworking and API for iterating over multiple impleme=
> ntatio
> > > ns
> > >       #14282 Sync thread should avoid holding the spa config write lock
> > >              when possible
> > >       #14283 txg_sync should handle write errors in ZIL
> > >       #14359 More adaptive ARC eviction
> > >       #14469 Fix NULL pointer dereference in zio_ready()
> > >       #14479 zfs redact fails when dnodesize=3Dauto
> > >       #14496 improve error message of zfs redact
> > >       #14500 Skip memory allocation when compressing holes
> > >       #14501 FreeBSD: don't verify recycled vnode for zfs control direc=
> tory
> > >       #14502 partially revert PR 14304 (eee9362a7)
> > >       #14509 Fix per-jail zfs.mount_snapshot setting
> > >       #14514 Fix data race between zil_commit() and zil_suspend()
> > >       #14516 System-wide speculative prefetch limit
> > >       #14517 Use rw_tryupgrade() in dmu_bonus_hold_by_dnode()
> > >       #14519 Do not hold spa_config in ZIL while blocked on IO
> > >       #14523 Move dmu_buf_rele() after dsl_dataset_sync_done()
> > >       #14524 Ignore too large stack in case of dsl_deadlist_merge
> > >       #14526 Use .section .rodata instead of .rodata on FreeBSD
> > >       #14528 ICP: AES-GCM: Refactor gcm_clear_ctx()
> > >       #14529 ICP: AES-GCM: Unify gcm_init_ctx() and gmac_init_ctx()
> > >       #14532 Handle unexpected errors in zil_lwb_commit() without ASSER=
> T()
> > >       #14544 icp: Prevent compilers from optimizing away memset()
> > >              in gcm_clear_ctx()
> > >       #14546 Revert zfeature_active() to static
> > >       #14556 Remove bad kmem_free() oversight from previous zfsdev_stat=
> e_list
> > >              patch
> > >       #14563 Optimize the is_l2cacheable functions
> > >       #14565 FreeBSD: zfs_znode_alloc: lock the vnode earlier
> > >       #14566 FreeBSD: fix false assert in cache_vop_rmdir when replayin=
> g ZIL
> > >       #14567 spl: Add cmn_err_once() to log a message only on the first=
>  call
> > >       #14568 Fix incremental receive silently failing for recursive sen=
> ds
> > >       #14569 Restore ASMABI and other Unify work
> > >       #14576 Fix detection of IBM Power8 machines (ISA 2.07)
> > >       #14577 Better handling for future crypto parameters
> > >       #14600 zcommon: Refactor FPU state handling in fletcher4
> > >       #14603 Fix prefetching of indirect blocks while destroying
> > >       #14633 Fixes in persistent error log
> > >       #14639 FreeBSD: Remove extra arc_reduce_target_size() call
> > >       #14641 Additional limits on hole reporting
> > >       #14649 Drop lying to the compiler in the fletcher4 code
> > >       #14652 panic loop when removing slog device
> > >       #14653 Update vdev state for spare vdev
> > >       #14655 Fix cloning into already dirty dbufs
> > >       #14678 Revert "Do not hold spa_config in ZIL while blocked on IO"
> > >
> > >     Obtained from:  OpenZFS
> > >     OpenZFS commit: 431083f75bdd3efaee992bdd672625ec7240d252
> >
> > Another problem related to copy_file_range() is the following example.
> >
> > slippy$ df -h build/make/Makefile .
> > Filesystem        Size    Used   Avail Capacity  Mounted on
> > t/wrkdir/amd64     52G     53M     52G     0%    /export/wrkdir/amd64
> > t/wrkdir/amd64     52G     53M     52G     0%    /export/wrkdir/amd64
> > slippy$ cp build/make/Makefile .
> > cp: build/make/Makefile: Cross-device link
> > slippy$
> >
> > And,
> >
> > slippy$ cp y4menc.h foobar
> > cp: y4menc.h: Cross-device link
> > slippy$
> >
> > But the following works because /tmp is mfs and /var/tmp is ufs.
> >
> > slippy$ cp y4menc.h /tmp
> > slippy$
> > slippy$ cp y4menc.h /var/tmp
> > slippy$
> >
> > Copying files from one zpool to a dataset in another zpool also works as
> > does copying files from different datasets on the same zpool. Only copyin=
> g
> > files from/to the same dataset results in cross device link.
> >
> zfs_copy_file_range() will only be called if the source and destination
> both exist on the same mount point. (I'm guessing that's what you mean
> by "same dataset".)

Correct.

The long term solution to the problem (for users) would be to enable 
block_cloning. One can test this by testing, checkpoint the pool, enable 
block_cloning, problem goes away, export and import the pool with 
--rewind-to-checkpoint to look at the regression again.

As to why we see a Cross-device link,

dtrace: script './dtrace.d' matched 1 probe
CPU     ID                    FUNCTION:NAME
  3  26104        vn_copy_file_range:return int64_t 0x12

0x12 == EXDEV.

As to why, zfs_clone_range() is called unconditionally at 
zfs/zfs_vnops_os.c:6292. If the feature is not enabled zfs_clone_range() 
sets the return code to EXDEV at line 1083 of zfs_vnops.c. Therefore 
block_cloning MUST be enabled.

To reiterate, the long term solution would be to enable block_cloning. For 
those who either wish or need to defer enabling of block_cloning, because 
the zpool is occasionally imported by older versions of freebsd requiring 
write acess, or other reason.


-- 
Cheers,
Cy Schubert <Cy.Schubert@cschubert.com>
FreeBSD UNIX:  <cy@FreeBSD.org>   Web:  https://FreeBSD.org
NTP:           <cy@nwtime.org>    Web:  https://nwtime.org

			e^(i*pi)+1=0