From: Jeff Layton <jlayton@kernel.org>
To: NeilBrown <neilb@suse.de>
Cc: tytso@mit.edu, adilger.kernel@dilger.ca, djwong@kernel.org,
david@fromorbit.com, trondmy@hammerspace.com,
viro@zeniv.linux.org.uk, zohar@linux.ibm.com, xiubli@redhat.com,
chuck.lever@oracle.com, lczerner@redhat.com, jack@suse.cz,
bfields@fieldses.org, brauner@kernel.org, fweimer@redhat.com,
linux-btrfs@vger.kernel.org, linux-fsdevel@vger.kernel.org,
linux-kernel@vger.kernel.org, ceph-devel@vger.kernel.org,
linux-ext4@vger.kernel.org, linux-nfs@vger.kernel.org,
linux-xfs@vger.kernel.org
Subject: Re: [PATCH v6 4/9] nfs: report the inode version in getattr if requested
Date: Tue, 04 Oct 2022 05:43:50 -0400 [thread overview]
Message-ID: <822ce678d47be0767464fc580d04981c24ccd28e.camel@kernel.org> (raw)
In-Reply-To: <166483977325.14457.7085950126736913468@noble.neil.brown.name>
On Tue, 2022-10-04 at 10:29 +1100, NeilBrown wrote:
> On Fri, 30 Sep 2022, Jeff Layton wrote:
> > Allow NFS to report the i_version in getattr requests. Since the cost to
> > fetch it is relatively cheap, do it unconditionally and just set the
> > flag if it looks like it's valid. Also, conditionally enable the
> > MONOTONIC flag when the server reports its change attr type as such.
> >
> > Signed-off-by: Jeff Layton <jlayton@kernel.org>
> > ---
> > fs/nfs/inode.c | 10 ++++++++--
> > 1 file changed, 8 insertions(+), 2 deletions(-)
> >
> > diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c
> > index bea7c005119c..5cb7017e5089 100644
> > --- a/fs/nfs/inode.c
> > +++ b/fs/nfs/inode.c
> > @@ -830,6 +830,8 @@ static u32 nfs_get_valid_attrmask(struct inode *inode)
> > reply_mask |= STATX_UID | STATX_GID;
> > if (!(cache_validity & NFS_INO_INVALID_BLOCKS))
> > reply_mask |= STATX_BLOCKS;
> > + if (!(cache_validity & NFS_INO_INVALID_CHANGE))
> > + reply_mask |= STATX_VERSION;
> > return reply_mask;
> > }
> >
> > @@ -848,7 +850,7 @@ int nfs_getattr(struct user_namespace *mnt_userns, const struct path *path,
> >
> > request_mask &= STATX_TYPE | STATX_MODE | STATX_NLINK | STATX_UID |
> > STATX_GID | STATX_ATIME | STATX_MTIME | STATX_CTIME |
> > - STATX_INO | STATX_SIZE | STATX_BLOCKS;
> > + STATX_INO | STATX_SIZE | STATX_BLOCKS | STATX_VERSION;
> >
> > if ((query_flags & AT_STATX_DONT_SYNC) && !force_sync) {
> > if (readdirplus_enabled)
> > @@ -877,7 +879,7 @@ int nfs_getattr(struct user_namespace *mnt_userns, const struct path *path,
> > /* Is the user requesting attributes that might need revalidation? */
> > if (!(request_mask & (STATX_MODE|STATX_NLINK|STATX_ATIME|STATX_CTIME|
> > STATX_MTIME|STATX_UID|STATX_GID|
> > - STATX_SIZE|STATX_BLOCKS)))
> > + STATX_SIZE|STATX_BLOCKS|STATX_VERSION)))
> > goto out_no_revalidate;
> >
> > /* Check whether the cached attributes are stale */
> > @@ -915,6 +917,10 @@ int nfs_getattr(struct user_namespace *mnt_userns, const struct path *path,
> >
> > generic_fillattr(&init_user_ns, inode, stat);
> > stat->ino = nfs_compat_user_ino64(NFS_FILEID(inode));
> > + stat->version = inode_peek_iversion_raw(inode);
>
> This looks wrong.
> 1/ it includes the I_VERSION_QUERIED bit, which should be hidden.
> 2/ it doesn't set that bit.
>
> I understand that the bit was already set when the generic code called
> inode_query_iversion(), but it might have changed if we needed to
> refresh the attrs.
>
> I'm beginning to think I shouldn't have approved the 3/9 patch. The
> stat->version shouldn't be set in vfs_getattr_nosec() - maybe in
> generic_fillattr(), but not a lot of point.
>
NFS (and Ceph), do not set the SB_I_VERSION flag and they don't use the
QUERIED bit. These are "server managed" implementations of i_version.
The server is responsible for incrementing the value, and we just store
the result in the i_version field and present it when needed. That's why
the patch for NFS is using the "raw" API.
> > + stat->attributes_mask |= STATX_ATTR_VERSION_MONOTONIC;
> > + if (server->change_attr_type != NFS4_CHANGE_TYPE_IS_UNDEFINED)
> > + stat->attributes |= STATX_ATTR_VERSION_MONOTONIC;
>
> So if the server tells us that the change attrs is based on time
> metadata, we accept that it will be monotonic (and RFC7862 encourages
> this), even though we seem to worry about timestamps going backwards
> (which we know that can)... Interesting.
>
>
I followed suit from nfs_inode_attrs_cmp(). It seems to treat any value
that isn't UNDEFINED as MONOTONIC, though it does use a less strict
comparator for NFS4_CHANGE_TYPE_IS_TIME_METADATA. It may make sense to
carve that out as an exception.
This is probably an indicator that we need a more strict definition for
STATX_ATTR_VERSION_MONOTONIC.
>
> > if (S_ISDIR(inode->i_mode))
> > stat->blksize = NFS_SERVER(inode)->dtsize;
> > out:
> > --
> > 2.37.3
> >
> >
--
Jeff Layton <jlayton@kernel.org>
next prev parent reply other threads:[~2022-10-04 9:47 UTC|newest]
Thread overview: 37+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-09-30 11:18 [PATCH v6 0/9] vfs/nfsd: clean up handling of i_version counter Jeff Layton
2022-09-30 11:18 ` [PATCH v6 1/9] iversion: move inode_query_iversion to libfs.c Jeff Layton
2022-10-03 23:05 ` NeilBrown
2022-09-30 11:18 ` [PATCH v6 2/9] iversion: clarify when the i_version counter must be updated Jeff Layton
2022-10-03 23:10 ` NeilBrown
2022-10-04 9:53 ` Jeff Layton
2022-09-30 11:18 ` [PATCH v6 3/9] vfs: plumb i_version handling into struct kstat Jeff Layton
2022-10-03 23:14 ` NeilBrown
2022-09-30 11:18 ` [PATCH v6 4/9] nfs: report the inode version in getattr if requested Jeff Layton
2022-10-03 23:29 ` NeilBrown
2022-10-04 9:43 ` Jeff Layton [this message]
2022-10-04 22:27 ` NeilBrown
2022-09-30 11:18 ` [PATCH v6 5/9] ceph: " Jeff Layton
2022-09-30 11:18 ` [PATCH v6 6/9] nfsd: use the getattr operation to fetch i_version Jeff Layton
2022-09-30 14:34 ` Chuck Lever III
2022-09-30 22:32 ` Dave Chinner
2022-10-03 23:39 ` NeilBrown
2022-10-05 10:06 ` Jeff Layton
2022-10-05 13:33 ` Chuck Lever III
2022-10-05 13:34 ` Trond Myklebust
2022-10-05 13:57 ` Jeff Layton
2022-10-05 21:14 ` NeilBrown
2022-10-06 11:15 ` Jeff Layton
2022-10-06 21:17 ` Dave Chinner
2022-09-30 11:18 ` [PATCH v6 7/9] vfs: expose STATX_VERSION to userland Jeff Layton
2022-10-03 23:42 ` NeilBrown
2022-10-05 10:08 ` Jeff Layton
2022-09-30 11:18 ` [PATCH v6 8/9] vfs: update times after copying data in __generic_file_write_iter Jeff Layton
2022-10-01 20:39 ` kernel test robot
2022-10-01 21:00 ` kernel test robot
2022-10-02 7:08 ` Amir Goldstein
2022-10-03 13:01 ` Jeff Layton
2022-10-03 13:52 ` Amir Goldstein
2022-10-03 22:56 ` NeilBrown
2022-10-05 16:40 ` Jeff Layton
2022-10-05 21:40 ` NeilBrown
2022-09-30 11:18 ` [PATCH v6 9/9] ext4: update times after I/O in write codepaths Jeff Layton
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=822ce678d47be0767464fc580d04981c24ccd28e.camel@kernel.org \
--to=jlayton@kernel.org \
--cc=adilger.kernel@dilger.ca \
--cc=bfields@fieldses.org \
--cc=brauner@kernel.org \
--cc=ceph-devel@vger.kernel.org \
--cc=chuck.lever@oracle.com \
--cc=david@fromorbit.com \
--cc=djwong@kernel.org \
--cc=fweimer@redhat.com \
--cc=jack@suse.cz \
--cc=lczerner@redhat.com \
--cc=linux-btrfs@vger.kernel.org \
--cc=linux-ext4@vger.kernel.org \
--cc=linux-fsdevel@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-nfs@vger.kernel.org \
--cc=linux-xfs@vger.kernel.org \
--cc=neilb@suse.de \
--cc=trondmy@hammerspace.com \
--cc=tytso@mit.edu \
--cc=viro@zeniv.linux.org.uk \
--cc=xiubli@redhat.com \
--cc=zohar@linux.ibm.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).