From: Jeff Layton <jlayton@kernel.org>
To: Alexander Viro <viro@zeniv.linux.org.uk>,
Christian Brauner <brauner@kernel.org>, Jan Kara <jack@suse.cz>,
Chuck Lever <chuck.lever@oracle.com>,
Alexander Aring <alex.aring@gmail.com>,
Trond Myklebust <trond.myklebust@hammerspace.com>,
Anna Schumaker <anna@kernel.org>,
Steve French <sfrench@samba.org>,
Paulo Alcantara <pc@manguebit.com>,
Ronnie Sahlberg <ronniesahlberg@gmail.com>,
Shyam Prasad N <sprasad@microsoft.com>,
Tom Talpey <tom@talpey.com>,
Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
"Rafael J. Wysocki" <rafael@kernel.org>,
David Howells <dhowells@redhat.com>,
Tyler Hicks <code@tyhicks.com>, Neil Brown <neilb@suse.de>,
Olga Kornievskaia <kolga@netapp.com>,
Dai Ngo <Dai.Ngo@oracle.com>, Miklos Szeredi <miklos@szeredi.hu>,
Amir Goldstein <amir73il@gmail.com>,
Namjae Jeon <linkinjeon@kernel.org>,
Sergey Senozhatsky <senozhatsky@chromium.org>,
"David S. Miller" <davem@davemloft.net>,
Eric Dumazet <edumazet@google.com>,
Jakub Kicinski <kuba@kernel.org>,
Paolo Abeni <pabeni@redhat.com>
Cc: linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org,
linux-nfs@vger.kernel.org, linux-cifs@vger.kernel.org,
samba-technical@lists.samba.org, netfs@lists.linux.dev,
ecryptfs@vger.kernel.org, linux-unionfs@vger.kernel.org,
netdev@vger.kernel.org, Jeff Layton <jlayton@kernel.org>
Subject: [PATCH RFC 23/24] nfs: optionally request a delegation on GETATTR
Date: Fri, 15 Mar 2024 12:53:14 -0400 [thread overview]
Message-ID: <20240315-dir-deleg-v1-23-a1d6209a3654@kernel.org> (raw)
In-Reply-To: <20240315-dir-deleg-v1-0-a1d6209a3654@kernel.org>
We expect directory delegations to most helpful when their children have
frequently revalidated dentries. The way we usually revalidate them is
to check that the dentry's d_time field matches the change attribute on
the parent. Most dentry revalidations that have to issue an RPC end up
as a GETATTR on the wire for the parent, if the parent hasn't changed,
then the dentry is fine.
Add a new NFS_INO_GDD_GETATTR flag to the nfsi->flags field. Whenever we
validate a dentry vs. its parent, set that flag on the parent. Then,
when we next do a GETATTR vs. the parent, test and clear the flag and do
a GDD_GETATTR if it was set. If the the GET_DIR_DELEGATION succeeds, set
the delegation in the inode.
Finally, a lot of servers don't support GET_DIR_DELEGATION, so add a
NFS_CAP_* bit to track whether the server supports it, and enable it by
default on v4.1+, and retry without a delegation if we get back an
ENOTSUPP.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
---
fs/nfs/delegation.c | 1 +
fs/nfs/dir.c | 8 ++++++++
fs/nfs/nfs4proc.c | 45 ++++++++++++++++++++++++++++++++++++++++++---
include/linux/nfs_fs.h | 1 +
include/linux/nfs_fs_sb.h | 1 +
5 files changed, 53 insertions(+), 3 deletions(-)
diff --git a/fs/nfs/delegation.c b/fs/nfs/delegation.c
index 88dbec4739d3..14c14bd03465 100644
--- a/fs/nfs/delegation.c
+++ b/fs/nfs/delegation.c
@@ -356,6 +356,7 @@ nfs_detach_delegation_locked(struct nfs_inode *nfsi,
delegation->inode = NULL;
rcu_assign_pointer(nfsi->delegation, NULL);
spin_unlock(&delegation->lock);
+ clear_bit(NFS_INO_GDD_GETATTR, &nfsi->flags);
return delegation;
}
diff --git a/fs/nfs/dir.c b/fs/nfs/dir.c
index 061648c73116..930fe7e14914 100644
--- a/fs/nfs/dir.c
+++ b/fs/nfs/dir.c
@@ -1504,12 +1504,20 @@ static int nfs_dentry_verify_change(struct inode *dir, struct dentry *dentry)
static int nfs_check_verifier(struct inode *dir, struct dentry *dentry,
int rcu_walk)
{
+ struct nfs_inode *nfsi = NFS_I(dir);
+
if (IS_ROOT(dentry))
return 1;
if (NFS_SERVER(dir)->flags & NFS_MOUNT_LOOKUP_CACHE_NONE)
return 0;
if (!nfs_dentry_verify_change(dir, dentry))
return 0;
+ /*
+ * The dentry matches the directory's change attribute, so
+ * we're likely revalidating here. Flag the dir so that we
+ * ask for a delegation on the next getattr.
+ */
+ set_bit(NFS_INO_GDD_GETATTR, &nfsi->flags);
/* Revalidate nfsi->cache_change_attribute before we declare a match */
if (nfs_mapping_need_revalidate_inode(dir)) {
if (rcu_walk)
diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index d518388bd0d6..3dbe9a18c9be 100644
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -4318,6 +4318,21 @@ static int nfs4_get_referral(struct rpc_clnt *client, struct inode *dir,
return status;
}
+static bool should_request_dir_deleg(struct inode *inode)
+{
+ if (!inode)
+ return false;
+ if (!S_ISDIR(inode->i_mode))
+ return false;
+ if (!nfs_server_capable(inode, NFS_CAP_GET_DIR_DELEG))
+ return false;
+ if (!test_and_clear_bit(NFS_INO_GDD_GETATTR, &(NFS_I(inode)->flags)))
+ return false;
+ if (nfs4_have_delegation(inode, FMODE_READ))
+ return false;
+ return true;
+}
+
static int _nfs4_proc_getattr(struct nfs_server *server, struct nfs_fh *fhandle,
struct nfs_fattr *fattr, struct inode *inode)
{
@@ -4331,11 +4346,12 @@ static int _nfs4_proc_getattr(struct nfs_server *server, struct nfs_fh *fhandle,
.server = server,
};
struct rpc_message msg = {
- .rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_GETATTR],
.rpc_argp = &args,
.rpc_resp = &res,
};
unsigned short task_flags = 0;
+ bool gdd;
+ int status;
if (nfs4_has_session(server->nfs_client))
task_flags = RPC_TASK_MOVEABLE;
@@ -4344,11 +4360,32 @@ static int _nfs4_proc_getattr(struct nfs_server *server, struct nfs_fh *fhandle,
if (inode && (server->flags & NFS_MOUNT_SOFTREVAL))
task_flags |= RPC_TASK_TIMEOUT;
+retry:
+ gdd = should_request_dir_deleg(inode);
+ if (gdd)
+ msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_GDD_GETATTR];
+ else
+ msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_GETATTR];
+
nfs4_bitmap_copy_adjust(bitmask, nfs4_bitmask(server, fattr->label), inode, 0);
nfs_fattr_init(fattr);
nfs4_init_sequence(&args.seq_args, &res.seq_res, 0, 0);
- return nfs4_do_call_sync(server->client, server, &msg,
- &args.seq_args, &res.seq_res, task_flags);
+ status = nfs4_do_call_sync(server->client, server, &msg,
+ &args.seq_args, &res.seq_res, task_flags);
+ switch (status) {
+ case 0:
+ if (gdd && res.nf_status == GDD4_OK)
+ status = nfs_inode_set_delegation(inode, current_cred(), FMODE_READ,
+ &res.deleg, 0);
+ break;
+ case -ENOTSUPP:
+ /* If the server doesn't support GET_DIR_DELEGATION, retry without it */
+ if (gdd) {
+ server->caps &= ~NFS_CAP_GET_DIR_DELEG;
+ goto retry;
+ }
+ }
+ return status;
}
int nfs4_proc_getattr(struct nfs_server *server, struct nfs_fh *fhandle,
@@ -10552,6 +10589,7 @@ static const struct nfs4_minor_version_ops nfs_v4_1_minor_ops = {
.minor_version = 1,
.init_caps = NFS_CAP_READDIRPLUS
| NFS_CAP_ATOMIC_OPEN
+ | NFS_CAP_GET_DIR_DELEG
| NFS_CAP_POSIX_LOCK
| NFS_CAP_STATEID_NFSV41
| NFS_CAP_ATOMIC_OPEN_V1
@@ -10578,6 +10616,7 @@ static const struct nfs4_minor_version_ops nfs_v4_2_minor_ops = {
.minor_version = 2,
.init_caps = NFS_CAP_READDIRPLUS
| NFS_CAP_ATOMIC_OPEN
+ | NFS_CAP_GET_DIR_DELEG
| NFS_CAP_POSIX_LOCK
| NFS_CAP_STATEID_NFSV41
| NFS_CAP_ATOMIC_OPEN_V1
diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h
index f5ce7b101146..5140e4dac98a 100644
--- a/include/linux/nfs_fs.h
+++ b/include/linux/nfs_fs.h
@@ -318,6 +318,7 @@ struct nfs4_copy_state {
#define NFS_INO_LAYOUTCOMMITTING (10) /* layoutcommit inflight */
#define NFS_INO_LAYOUTSTATS (11) /* layoutstats inflight */
#define NFS_INO_ODIRECT (12) /* I/O setting is O_DIRECT */
+#define NFS_INO_GDD_GETATTR (13) /* Ask for dir deleg on next GETATTR */
static inline struct nfs_inode *NFS_I(const struct inode *inode)
{
diff --git a/include/linux/nfs_fs_sb.h b/include/linux/nfs_fs_sb.h
index 92de074e63b9..0ab744cf52d7 100644
--- a/include/linux/nfs_fs_sb.h
+++ b/include/linux/nfs_fs_sb.h
@@ -278,6 +278,7 @@ struct nfs_server {
#define NFS_CAP_LGOPEN (1U << 5)
#define NFS_CAP_CASE_INSENSITIVE (1U << 6)
#define NFS_CAP_CASE_PRESERVING (1U << 7)
+#define NFS_CAP_GET_DIR_DELEG (1U << 13)
#define NFS_CAP_POSIX_LOCK (1U << 14)
#define NFS_CAP_UIDGID_NOMAP (1U << 15)
#define NFS_CAP_STATEID_NFSV41 (1U << 16)
--
2.44.0
next prev parent reply other threads:[~2024-03-15 16:54 UTC|newest]
Thread overview: 47+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-03-15 16:52 [PATCH RFC 00/24] vfs, nfsd, nfs: implement directory delegations Jeff Layton
2024-03-15 16:52 ` [PATCH RFC 01/24] filelock: push the S_ISREG check down to ->setlease handlers Jeff Layton
2024-03-15 16:52 ` [PATCH RFC 02/24] filelock: add a lm_set_conflict lease_manager callback Jeff Layton
2024-03-17 14:56 ` Chuck Lever
2024-03-18 11:07 ` Jeff Layton
2024-03-20 13:13 ` Christian Brauner
2024-03-20 13:17 ` Jeff Layton
2024-03-15 16:52 ` [PATCH RFC 03/24] vfs: add try_break_deleg calls for parents to vfs_{link,rename,unlink} Jeff Layton
2024-03-16 23:57 ` Al Viro
2024-03-17 14:09 ` Jeff Layton
2024-03-15 16:52 ` [PATCH RFC 04/24] vfs: allow mkdir to wait for delegation break on parent Jeff Layton
2024-03-15 16:52 ` [PATCH RFC 05/24] vfs: allow rmdir " Jeff Layton
2024-03-15 16:52 ` [PATCH RFC 06/24] vfs: break parent dir delegations in open(..., O_CREAT) codepath Jeff Layton
2024-03-17 0:19 ` Al Viro
2024-03-17 12:23 ` Jeff Layton
2024-03-18 8:25 ` Stefan Metzmacher
2024-03-18 10:21 ` Jeff Layton
2024-03-15 16:52 ` [PATCH RFC 07/24] vfs: make vfs_create break delegations on parent directory Jeff Layton
2024-03-15 16:52 ` [PATCH RFC 08/24] vfs: make vfs_mknod " Jeff Layton
2024-03-20 13:42 ` Christian Brauner
2024-03-20 20:12 ` Jeff Layton
2024-03-22 14:13 ` Christian Brauner
2024-03-15 16:53 ` [PATCH RFC 09/24] filelock: lift the ban on directory leases in generic_setlease Jeff Layton
2024-03-15 16:53 ` [PATCH RFC 10/24] nfsd: allow filecache to hold S_IFDIR files Jeff Layton
2024-03-15 16:53 ` [PATCH RFC 11/24] nfsd: allow DELEGRETURN on directories Jeff Layton
2024-03-17 15:09 ` Chuck Lever
2024-03-17 16:03 ` Trond Myklebust
2024-03-18 11:22 ` Jeff Layton
2024-03-15 16:53 ` [PATCH RFC 12/24] nfsd: encoders and decoders for GET_DIR_DELEGATION Jeff Layton
2024-03-17 15:34 ` Chuck Lever
2024-03-18 16:12 ` Jeff Layton
2024-03-18 16:19 ` Chuck Lever
2024-03-15 16:53 ` [PATCH RFC 13/24] nfsd: check for delegation conflicts vs. the same client Jeff Layton
2024-03-15 16:53 ` [PATCH RFC 14/24] nfsd: wire up GET_DIR_DELEGATION handling Jeff Layton
2024-03-17 15:42 ` Chuck Lever
2024-03-15 16:53 ` [PATCH RFC 15/24] nfs: fix nfs_stateid_hash prototype when CONFIG_CRC32 isn't set Jeff Layton
2024-03-15 16:53 ` [PATCH RFC 16/24] nfs: remove unused NFS_CALL macro Jeff Layton
2024-03-15 16:53 ` [PATCH RFC 17/24] nfs: add cache_validity to the nfs_inode_event tracepoints Jeff Layton
2024-03-15 16:53 ` [PATCH RFC 18/24] nfs: add a tracepoint to nfs_inode_detach_delegation_locked Jeff Layton
2024-03-15 16:53 ` [PATCH RFC 19/24] nfs: new tracepoint in nfs_delegation_need_return Jeff Layton
2024-03-15 16:53 ` [PATCH RFC 20/24] nfs: new tracepoint in match_stateid operation Jeff Layton
2024-03-15 16:53 ` [PATCH RFC 21/24] nfs: add a GDD_GETATTR rpc operation Jeff Layton
2024-03-15 20:50 ` Anna Schumaker
2024-03-15 21:29 ` Jeff Layton
2024-03-15 16:53 ` [PATCH RFC 22/24] nfs: skip dentry revalidation when parent dir has a delegation Jeff Layton
2024-03-15 16:53 ` Jeff Layton [this message]
2024-03-15 16:53 ` [PATCH RFC 24/24] nfs: add a module parameter to disable directory delegations 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=20240315-dir-deleg-v1-23-a1d6209a3654@kernel.org \
--to=jlayton@kernel.org \
--cc=Dai.Ngo@oracle.com \
--cc=alex.aring@gmail.com \
--cc=amir73il@gmail.com \
--cc=anna@kernel.org \
--cc=brauner@kernel.org \
--cc=chuck.lever@oracle.com \
--cc=code@tyhicks.com \
--cc=davem@davemloft.net \
--cc=dhowells@redhat.com \
--cc=ecryptfs@vger.kernel.org \
--cc=edumazet@google.com \
--cc=gregkh@linuxfoundation.org \
--cc=jack@suse.cz \
--cc=kolga@netapp.com \
--cc=kuba@kernel.org \
--cc=linkinjeon@kernel.org \
--cc=linux-cifs@vger.kernel.org \
--cc=linux-fsdevel@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-nfs@vger.kernel.org \
--cc=linux-unionfs@vger.kernel.org \
--cc=miklos@szeredi.hu \
--cc=neilb@suse.de \
--cc=netdev@vger.kernel.org \
--cc=netfs@lists.linux.dev \
--cc=pabeni@redhat.com \
--cc=pc@manguebit.com \
--cc=rafael@kernel.org \
--cc=ronniesahlberg@gmail.com \
--cc=samba-technical@lists.samba.org \
--cc=senozhatsky@chromium.org \
--cc=sfrench@samba.org \
--cc=sprasad@microsoft.com \
--cc=tom@talpey.com \
--cc=trond.myklebust@hammerspace.com \
--cc=viro@zeniv.linux.org.uk \
/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