git: a5e284038edc - main - open(2): Add O_DSYNC flag.

[Kyle Evans]

On Thu, Jan 7, 2021 at 6:17 PM Thomas Munro <tmunro at freebsd.org> wrote:
>
> The branch main has been updated by tmunro:
>
> URL: https://cgit.FreeBSD.org/src/commit/?id=a5e284038edc36b0447f1e6337419a3c0ea1788d
>
> commit a5e284038edc36b0447f1e6337419a3c0ea1788d
> Author:     Thomas Munro <tmunro at FreeBSD.org>
> AuthorDate: 2021-01-07 10:46:51 +0000
> Commit:     Thomas Munro <tmunro at FreeBSD.org>
> CommitDate: 2021-01-08 00:15:56 +0000
>
>     open(2): Add O_DSYNC flag.
>
>     POSIX O_DSYNC means that writes include an implicit fdatasync(2), just
>     as O_SYNC implies fsync(2).
>
>     VOP_WRITE() functions that understand the new IO_DATASYNC flag can act
>     accordingly, but we'll still pass down IO_SYNC so that file systems that
>     don't understand it will continue to provide the stronger O_SYNC
>     behaviour.
>
>     Flag also applies to fcntl(2).
>
>     Reviewed by: kib, delphij
>     Differential Revision: https://reviews.freebsd.org/D25090
> ---
>  lib/libc/sys/fcntl.2                                 | 19 +++++++++++++++++--
>  lib/libc/sys/open.2                                  | 20 +++++++++++++-------
>  .../openzfs/include/os/freebsd/spl/sys/ccompile.h    |  1 -
>  .../openzfs/include/os/freebsd/spl/sys/vnode.h       |  2 ++
>  sys/kern/vfs_vnops.c                                 |  7 +++++++
>  sys/sys/fcntl.h                                      | 15 +++++++++------
>  sys/sys/vnode.h                                      |  1 +
>  7 files changed, 49 insertions(+), 16 deletions(-)
>
> [... snip ...]
> diff --git a/sys/sys/fcntl.h b/sys/sys/fcntl.h
> index 2f424d173949..e2597726c53b 100644
> --- a/sys/sys/fcntl.h
> +++ b/sys/sys/fcntl.h
> @@ -141,8 +141,10 @@ typedef    __pid_t         pid_t;
>                                            return back */
>  #endif
>
> +#define        O_DSYNC         0x00800000      /* POSIX data sync */
> +
>  /*
> - * XXX missing O_DSYNC, O_RSYNC.
> + * XXX missing O_RSYNC.
>   */
>
>  #ifdef _KERNEL

I think the value of O_DSYNC got resolved incorrectly in a merge
conflict; O_RESOLVE_BENEATH in the just prior __BSD_VISIBLE block took
0x00800000, so I believe this should be 0x01000000.

Thanks,

Kyle Evans