From: trondmy@kernel.org
To: linux-nfs@vger.kernel.org
Subject: [PATCH 00/17] Attribute revalidation updates
Date: Mon, 29 Mar 2021 20:18:18 -0400 [thread overview]
Message-ID: <20210330001835.41914-1-trondmy@kernel.org> (raw)
From: Trond Myklebust <trond.myklebust@hammerspace.com>
The following updates try to simplify some of the attribute caching, and
further split out the attribute cache tracking. In particular, we want
to make use of the fact that NFSv4.x doesn't always need to retrieve all
the attributes to optimise the way we use the GETATTR call.
Before this patch set, when I run xfstests, I see
NFSv3:: getattr: 444656, lookup: 144117, access: 112148
NFSv4.1:: getattr: 898027, lookup: 68847, access: 14766, open: 111409
NFSv4.2:: getattr: 593422, lookup: 72488, access: 16647, open: 113056
after application of the patch set:
NFSv3:: getattr: 399324, lookup: 142331, access: 112444
NFSv4.1:: getattr: 134778, lookup: 70673, access: 15059, open: 112192
NFSv4.2:: getattr: 139080, lookup: 70722, access: 16563, open: 113601
Note that these numbers are against a knfsd server that does not support
the "change_attr_type" attribute. I'd expect further improvements once
we re-introduce that server support.
Trond Myklebust (17):
NFS: Deal correctly with attribute generation counter overflow
NFS: Fix up inode cache tracing
NFS: Mask out unsupported attributes in nfs_getattr()
NFS: NFS_INO_REVAL_PAGECACHE should mark the change attribute invalid
NFS: Fix up revalidation of space used
NFS: Don't revalidate attributes that are not being asked for
NFS: Fix up statx() results
NFS: Add a cache validity flag argument to nfs_revalidate_inode()
NFS: Replace use of NFS_INO_REVAL_PAGECACHE when checking cache
validity
NFS: Don't set NFS_INO_REVAL_PAGECACHE in the inode cache validity
NFS: Separate tracking of file nlinks cache validity from the
mode/uid/gid
NFS: Fix up handling of outstanding layoutcommit in nfs_update_inode()
NFS: Remove a line of code that has no effect in nfs_update_inode()
NFS: Simplify cache consistency in nfs_check_inode_attributes()
NFSv4: Fix value of decode_fsinfo_maxsz
NFSv4: Add support for the NFSv4.2 "change_attr_type" attribute
NFS: Use information about the change attribute to optimise updates
fs/nfs/client.c | 3 +
fs/nfs/dir.c | 4 +-
fs/nfs/export.c | 6 +-
fs/nfs/file.c | 2 +-
fs/nfs/inode.c | 246 +++++++++++++++++++++++++++-----------
fs/nfs/nfs3acl.c | 2 +-
fs/nfs/nfs3xdr.c | 1 +
fs/nfs/nfs4proc.c | 45 ++++---
fs/nfs/nfs4xdr.c | 43 ++++++-
fs/nfs/nfstrace.h | 9 +-
fs/nfs/proc.c | 1 +
fs/nfs/write.c | 2 +-
include/linux/nfs4.h | 9 ++
include/linux/nfs_fs.h | 4 +-
include/linux/nfs_fs_sb.h | 3 +
include/linux/nfs_xdr.h | 2 +
16 files changed, 283 insertions(+), 99 deletions(-)
--
2.30.2
next reply other threads:[~2021-03-30 0:19 UTC|newest]
Thread overview: 18+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-03-30 0:18 trondmy [this message]
2021-03-30 0:18 ` [PATCH 01/17] NFS: Deal correctly with attribute generation counter overflow trondmy
2021-03-30 0:18 ` [PATCH 02/17] NFS: Fix up inode cache tracing trondmy
2021-03-30 0:18 ` [PATCH 03/17] NFS: Mask out unsupported attributes in nfs_getattr() trondmy
2021-03-30 0:18 ` [PATCH 04/17] NFS: NFS_INO_REVAL_PAGECACHE should mark the change attribute invalid trondmy
2021-03-30 0:18 ` [PATCH 05/17] NFS: Fix up revalidation of space used trondmy
2021-03-30 0:18 ` [PATCH 06/17] NFS: Don't revalidate attributes that are not being asked for trondmy
2021-03-30 0:18 ` [PATCH 07/17] NFS: Fix up statx() results trondmy
2021-03-30 0:18 ` [PATCH 08/17] NFS: Add a cache validity flag argument to nfs_revalidate_inode() trondmy
2021-03-30 0:18 ` [PATCH 09/17] NFS: Replace use of NFS_INO_REVAL_PAGECACHE when checking cache validity trondmy
2021-03-30 0:18 ` [PATCH 10/17] NFS: Don't set NFS_INO_REVAL_PAGECACHE in the inode " trondmy
2021-03-30 0:18 ` [PATCH 11/17] NFS: Separate tracking of file nlinks cache validity from the mode/uid/gid trondmy
2021-03-30 0:18 ` [PATCH 12/17] NFS: Fix up handling of outstanding layoutcommit in nfs_update_inode() trondmy
2021-03-30 0:18 ` [PATCH 13/17] NFS: Remove a line of code that has no effect " trondmy
2021-03-30 0:18 ` [PATCH 14/17] NFS: Simplify cache consistency in nfs_check_inode_attributes() trondmy
2021-03-30 0:18 ` [PATCH 15/17] NFSv4: Fix value of decode_fsinfo_maxsz trondmy
2021-03-30 0:18 ` [PATCH 16/17] NFSv4: Add support for the NFSv4.2 "change_attr_type" attribute trondmy
2021-03-30 0:18 ` [PATCH 17/17] NFS: Use information about the change attribute to optimise updates trondmy
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=20210330001835.41914-1-trondmy@kernel.org \
--to=trondmy@kernel.org \
--cc=linux-nfs@vger.kernel.org \
/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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.