[PATCH v3 0/7] A course adjustment, maybe...

[PATCH v3 0/7] A course adjustment, maybe...

[Chuck Lever]

I'm proposing this series for v6.2 (for-next). I intend to open this
branch soon, now that the v6.1 merge window is closed.

For quite some time, we've been encouraged to disable filecache
garbage collection for NFSv4 files, and I think I found a surgical
way to do just that. That is presented in "NFSD: Add an NFSD_FILE_GC
flag to enable nfsd_file garbage collection".

The other major change in this set (previously left unmentioned)
is reworking the file_hashtbl to resize itself dynamically. This
reduces the average size of its bucket chains, greatly speeding up
hash insertion, which holds the state_lock.

Comments and opinions are welcome.

Changes since v2:
- Converted nfs4_file_rhashtbl to nfs4_file_rhltable
- Addressed most or all other review comments


Changes since RFC:
- checking nfs4_files for inode aliases is now done only on hash
 insertion
- the nfs4_file reference count is now bumped only while the RCU
 read lock is held
- comments and function names have been revised and clarified

---

Chuck Lever (7):
      NFSD: Pass the target nfsd_file to nfsd_commit()
      NFSD: Revert "NFSD: NFSv4 CLOSE should release an nfsd_file immediately"
      NFSD: Add an NFSD_FILE_GC flag to enable nfsd_file garbage collection
      NFSD: Use const pointers as parameters to fh_ helpers.
      NFSD: Use rhashtable for managing nfs4_file objects
      NFSD: Clean up nfs4_preprocess_stateid_op() call sites
      NFSD: Trace delegation revocations


 fs/nfsd/filecache.c |  79 +++++++++++-----
 fs/nfsd/filecache.h |   4 +-
 fs/nfsd/nfs3proc.c  |  10 +-
 fs/nfsd/nfs4proc.c  |  42 ++++-----
 fs/nfsd/nfs4state.c | 220 ++++++++++++++++++++++++++++++--------------
 fs/nfsd/nfsfh.h     |  10 +-
 fs/nfsd/state.h     |   5 +-
 fs/nfsd/trace.h     |  58 +++++++++++-
 fs/nfsd/vfs.c       |  19 ++--
 fs/nfsd/vfs.h       |   3 +-
 10 files changed, 304 insertions(+), 146 deletions(-)

--
Chuck Lever

[PATCH v3 1/7] NFSD: Pass the target nfsd_file to nfsd_commit()

[Chuck Lever]

In a moment I'm going to introduce separate nfsd_file types, one of
which is garbage-collected; the other, not. The garbage-collected
variety is to be used by NFSv2 and v3, and the non-garbage-collected
variety is to be used by NFSv4.

nfsd_commit() is invoked by both NFSv3 and NFSv4 consumers. We want
nfsd_commit() to find and use the correct variety of cached
nfsd_file object for the NFS version that is in use.

This is a refactoring change. No behavior change is expected.

Tested-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
---
 fs/nfsd/nfs3proc.c |   10 +++++++++-
 fs/nfsd/nfs4proc.c |   11 ++++++++++-
 fs/nfsd/vfs.c      |   15 ++++-----------
 fs/nfsd/vfs.h      |    3 ++-
 4 files changed, 25 insertions(+), 14 deletions(-)

diff --git a/fs/nfsd/nfs3proc.c b/fs/nfsd/nfs3proc.c
index a41cca619338..d12823371857 100644
--- a/fs/nfsd/nfs3proc.c
+++ b/fs/nfsd/nfs3proc.c
@@ -13,6 +13,7 @@
 #include "cache.h"
 #include "xdr3.h"
 #include "vfs.h"
+#include "filecache.h"
 
 #define NFSDDBG_FACILITY		NFSDDBG_PROC
 
@@ -770,6 +771,7 @@ nfsd3_proc_commit(struct svc_rqst *rqstp)
 {
 	struct nfsd3_commitargs *argp = rqstp->rq_argp;
 	struct nfsd3_commitres *resp = rqstp->rq_resp;
+	struct nfsd_file *nf;
 
 	dprintk("nfsd: COMMIT(3)   %s %u@%Lu\n",
 				SVCFH_fmt(&argp->fh),
@@ -777,8 +779,14 @@ nfsd3_proc_commit(struct svc_rqst *rqstp)
 				(unsigned long long) argp->offset);
 
 	fh_copy(&resp->fh, &argp->fh);
-	resp->status = nfsd_commit(rqstp, &resp->fh, argp->offset,
+	resp->status = nfsd_file_acquire(rqstp, &resp->fh, NFSD_MAY_WRITE |
+					 NFSD_MAY_NOT_BREAK_LEASE, &nf);
+	if (resp->status)
+		goto out;
+	resp->status = nfsd_commit(rqstp, &resp->fh, nf, argp->offset,
 				   argp->count, resp->verf);
+	nfsd_file_put(nf);
+out:
 	return rpc_success;
 }
 
diff --git a/fs/nfsd/nfs4proc.c b/fs/nfsd/nfs4proc.c
index a72ab97f77ef..d0d976f847ca 100644
--- a/fs/nfsd/nfs4proc.c
+++ b/fs/nfsd/nfs4proc.c
@@ -739,10 +739,19 @@ nfsd4_commit(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 	     union nfsd4_op_u *u)
 {
 	struct nfsd4_commit *commit = &u->commit;
+	struct nfsd_file *nf;
+	__be32 status;
 
-	return nfsd_commit(rqstp, &cstate->current_fh, commit->co_offset,
+	status = nfsd_file_acquire(rqstp, &cstate->current_fh, NFSD_MAY_WRITE |
+				   NFSD_MAY_NOT_BREAK_LEASE, &nf);
+	if (status != nfs_ok)
+		return status;
+
+	status = nfsd_commit(rqstp, &cstate->current_fh, nf, commit->co_offset,
 			     commit->co_count,
 			     (__be32 *)commit->co_verf.data);
+	nfsd_file_put(nf);
+	return status;
 }
 
 static __be32
diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c
index fc17b0ac8729..44f210ba17cc 100644
--- a/fs/nfsd/vfs.c
+++ b/fs/nfsd/vfs.c
@@ -1108,6 +1108,7 @@ nfsd_write(struct svc_rqst *rqstp, struct svc_fh *fhp, loff_t offset,
  * nfsd_commit - Commit pending writes to stable storage
  * @rqstp: RPC request being processed
  * @fhp: NFS filehandle
+ * @nf: target file
  * @offset: raw offset from beginning of file
  * @count: raw count of bytes to sync
  * @verf: filled in with the server's current write verifier
@@ -1124,19 +1125,13 @@ nfsd_write(struct svc_rqst *rqstp, struct svc_fh *fhp, loff_t offset,
  *   An nfsstat value in network byte order.
  */
 __be32
-nfsd_commit(struct svc_rqst *rqstp, struct svc_fh *fhp, u64 offset,
-	    u32 count, __be32 *verf)
+nfsd_commit(struct svc_rqst *rqstp, struct svc_fh *fhp, struct nfsd_file *nf,
+	    u64 offset, u32 count, __be32 *verf)
 {
+	__be32			err = nfs_ok;
 	u64			maxbytes;
 	loff_t			start, end;
 	struct nfsd_net		*nn;
-	struct nfsd_file	*nf;
-	__be32			err;
-
-	err = nfsd_file_acquire(rqstp, fhp,
-			NFSD_MAY_WRITE|NFSD_MAY_NOT_BREAK_LEASE, &nf);
-	if (err)
-		goto out;
 
 	/*
 	 * Convert the client-provided (offset, count) range to a
@@ -1177,8 +1172,6 @@ nfsd_commit(struct svc_rqst *rqstp, struct svc_fh *fhp, u64 offset,
 	} else
 		nfsd_copy_write_verifier(verf, nn);
 
-	nfsd_file_put(nf);
-out:
 	return err;
 }
 
diff --git a/fs/nfsd/vfs.h b/fs/nfsd/vfs.h
index c95cd414b4bb..04dfd80570f0 100644
--- a/fs/nfsd/vfs.h
+++ b/fs/nfsd/vfs.h
@@ -88,7 +88,8 @@ __be32		nfsd_access(struct svc_rqst *, struct svc_fh *, u32 *, u32 *);
 __be32		nfsd_create_setattr(struct svc_rqst *rqstp, struct svc_fh *fhp,
 				struct svc_fh *resfhp, struct nfsd_attrs *iap);
 __be32		nfsd_commit(struct svc_rqst *rqst, struct svc_fh *fhp,
-				u64 offset, u32 count, __be32 *verf);
+				struct nfsd_file *nf, u64 offset, u32 count,
+				__be32 *verf);
 #ifdef CONFIG_NFSD_V4
 __be32		nfsd_getxattr(struct svc_rqst *rqstp, struct svc_fh *fhp,
 			    char *name, void **bufp, int *lenp);

[PATCH v3 2/7] NFSD: Revert "NFSD: NFSv4 CLOSE should release an nfsd_file immediately"

[Chuck Lever]

This reverts commit 5e138c4a750dc140d881dab4a8804b094bbc08d2.

That commit attempted to make files available to other users as soon
as all NFSv4 clients were done with them, rather than waiting until
the filecache LRU had garbage collected them.

It gets the reference counting wrong, for one thing.

But it also misses that DELEGRETURN should release a file in the
same fashion. In fact, any nfsd_file_put() on an file held open
by an NFSv4 client needs potentially to release the file
immediately...

Clear the way for implementing that idea.

Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
---
 fs/nfsd/filecache.c |   18 ------------------
 fs/nfsd/filecache.h |    1 -
 fs/nfsd/nfs4state.c |    4 ++--
 3 files changed, 2 insertions(+), 21 deletions(-)

diff --git a/fs/nfsd/filecache.c b/fs/nfsd/filecache.c
index a2adfc247648..b7aa523c2010 100644
--- a/fs/nfsd/filecache.c
+++ b/fs/nfsd/filecache.c
@@ -444,24 +444,6 @@ nfsd_file_put(struct nfsd_file *nf)
 		nfsd_file_put_noref(nf);
 }
 
-/**
- * nfsd_file_close - Close an nfsd_file
- * @nf: nfsd_file to close
- *
- * If this is the final reference for @nf, free it immediately.
- * This reflects an on-the-wire CLOSE or DELEGRETURN into the
- * VFS and exported filesystem.
- */
-void nfsd_file_close(struct nfsd_file *nf)
-{
-	nfsd_file_put(nf);
-	if (refcount_dec_if_one(&nf->nf_ref)) {
-		nfsd_file_unhash(nf);
-		nfsd_file_lru_remove(nf);
-		nfsd_file_free(nf);
-	}
-}
-
 struct nfsd_file *
 nfsd_file_get(struct nfsd_file *nf)
 {
diff --git a/fs/nfsd/filecache.h b/fs/nfsd/filecache.h
index 8e8c0c47d67d..f81c198f4ed6 100644
--- a/fs/nfsd/filecache.h
+++ b/fs/nfsd/filecache.h
@@ -52,7 +52,6 @@ void nfsd_file_cache_shutdown(void);
 int nfsd_file_cache_start_net(struct net *net);
 void nfsd_file_cache_shutdown_net(struct net *net);
 void nfsd_file_put(struct nfsd_file *nf);
-void nfsd_file_close(struct nfsd_file *nf);
 struct nfsd_file *nfsd_file_get(struct nfsd_file *nf);
 void nfsd_file_close_inode_sync(struct inode *inode);
 bool nfsd_file_is_cached(struct inode *inode);
diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c
index c5d199d7e6b4..2b850de288cf 100644
--- a/fs/nfsd/nfs4state.c
+++ b/fs/nfsd/nfs4state.c
@@ -820,9 +820,9 @@ static void __nfs4_file_put_access(struct nfs4_file *fp, int oflag)
 			swap(f2, fp->fi_fds[O_RDWR]);
 		spin_unlock(&fp->fi_lock);
 		if (f1)
-			nfsd_file_close(f1);
+			nfsd_file_put(f1);
 		if (f2)
-			nfsd_file_close(f2);
+			nfsd_file_put(f2);
 	}
 }
 

[PATCH v3 3/7] NFSD: Add an NFSD_FILE_GC flag to enable nfsd_file garbage collection

[Chuck Lever]

NFSv4 operations manage the lifetime of nfsd_file items they use by
means of NFSv4 OPEN and CLOSE. Hence there's no need for them to be
garbage collected.

Introduce a mechanism to enable garbage collection for nfsd_file
items used only by NFSv2/3 callers.

Note that the change in nfsd_file_put() ensures that both CLOSE and
DELEGRETURN will actually close out and free an nfsd_file on last
reference of a non-garbage-collected file.

Link: https://bugzilla.linux-nfs.org/show_bug.cgi?id=394
Suggested-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Tested-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
---
 fs/nfsd/filecache.c |   61 +++++++++++++++++++++++++++++++++++++++++++++------
 fs/nfsd/filecache.h |    3 +++
 fs/nfsd/nfs3proc.c  |    4 ++-
 fs/nfsd/trace.h     |    3 ++-
 fs/nfsd/vfs.c       |    4 ++-
 5 files changed, 63 insertions(+), 12 deletions(-)

diff --git a/fs/nfsd/filecache.c b/fs/nfsd/filecache.c
index b7aa523c2010..87fce5c95726 100644
--- a/fs/nfsd/filecache.c
+++ b/fs/nfsd/filecache.c
@@ -63,6 +63,7 @@ struct nfsd_file_lookup_key {
 	struct net			*net;
 	const struct cred		*cred;
 	unsigned char			need;
+	unsigned char			gc:1;
 	enum nfsd_file_lookup_type	type;
 };
 
@@ -162,6 +163,8 @@ static int nfsd_file_obj_cmpfn(struct rhashtable_compare_arg *arg,
 			return 1;
 		if (!nfsd_match_cred(nf->nf_cred, key->cred))
 			return 1;
+		if (test_bit(NFSD_FILE_GC, &nf->nf_flags) != key->gc)
+			return 1;
 		if (test_bit(NFSD_FILE_HASHED, &nf->nf_flags) == 0)
 			return 1;
 		break;
@@ -297,6 +300,8 @@ nfsd_file_alloc(struct nfsd_file_lookup_key *key, unsigned int may)
 		nf->nf_flags = 0;
 		__set_bit(NFSD_FILE_HASHED, &nf->nf_flags);
 		__set_bit(NFSD_FILE_PENDING, &nf->nf_flags);
+		if (key->gc)
+			__set_bit(NFSD_FILE_GC, &nf->nf_flags);
 		nf->nf_inode = key->inode;
 		/* nf_ref is pre-incremented for hash table */
 		refcount_set(&nf->nf_ref, 2);
@@ -428,16 +433,27 @@ nfsd_file_put_noref(struct nfsd_file *nf)
 	}
 }
 
+static void
+nfsd_file_unhash_and_put(struct nfsd_file *nf)
+{
+	if (nfsd_file_unhash(nf))
+		nfsd_file_put_noref(nf);
+}
+
 void
 nfsd_file_put(struct nfsd_file *nf)
 {
 	might_sleep();
 
-	nfsd_file_lru_add(nf);
+	if (test_bit(NFSD_FILE_GC, &nf->nf_flags) == 1)
+		nfsd_file_lru_add(nf);
+	else if (refcount_read(&nf->nf_ref) == 2)
+		nfsd_file_unhash_and_put(nf);
+
 	if (test_bit(NFSD_FILE_HASHED, &nf->nf_flags) == 0) {
 		nfsd_file_flush(nf);
 		nfsd_file_put_noref(nf);
-	} else if (nf->nf_file) {
+	} else if (nf->nf_file && test_bit(NFSD_FILE_GC, &nf->nf_flags) == 1) {
 		nfsd_file_put_noref(nf);
 		nfsd_file_schedule_laundrette();
 	} else
@@ -1017,12 +1033,14 @@ nfsd_file_is_cached(struct inode *inode)
 
 static __be32
 nfsd_file_do_acquire(struct svc_rqst *rqstp, struct svc_fh *fhp,
-		     unsigned int may_flags, struct nfsd_file **pnf, bool open)
+		     unsigned int may_flags, struct nfsd_file **pnf,
+		     bool open, int want_gc)
 {
 	struct nfsd_file_lookup_key key = {
 		.type	= NFSD_FILE_KEY_FULL,
 		.need	= may_flags & NFSD_FILE_MAY_MASK,
 		.net	= SVC_NET(rqstp),
+		.gc	= want_gc,
 	};
 	bool open_retry = true;
 	struct nfsd_file *nf;
@@ -1117,14 +1135,35 @@ nfsd_file_do_acquire(struct svc_rqst *rqstp, struct svc_fh *fhp,
 	 * then unhash.
 	 */
 	if (status != nfs_ok || key.inode->i_nlink == 0)
-		if (nfsd_file_unhash(nf))
-			nfsd_file_put_noref(nf);
+		nfsd_file_unhash_and_put(nf);
 	clear_bit_unlock(NFSD_FILE_PENDING, &nf->nf_flags);
 	smp_mb__after_atomic();
 	wake_up_bit(&nf->nf_flags, NFSD_FILE_PENDING);
 	goto out;
 }
 
+/**
+ * nfsd_file_acquire_gc - Get a struct nfsd_file with an open file
+ * @rqstp: the RPC transaction being executed
+ * @fhp: the NFS filehandle of the file to be opened
+ * @may_flags: NFSD_MAY_ settings for the file
+ * @pnf: OUT: new or found "struct nfsd_file" object
+ *
+ * The nfsd_file object returned by this API is reference-counted
+ * and garbage-collected. The object is retained for a few
+ * seconds after the final nfsd_file_put() in case the caller
+ * wants to re-use it.
+ *
+ * Returns nfs_ok and sets @pnf on success; otherwise an nfsstat in
+ * network byte order is returned.
+ */
+__be32
+nfsd_file_acquire_gc(struct svc_rqst *rqstp, struct svc_fh *fhp,
+		     unsigned int may_flags, struct nfsd_file **pnf)
+{
+	return nfsd_file_do_acquire(rqstp, fhp, may_flags, pnf, true, 1);
+}
+
 /**
  * nfsd_file_acquire - Get a struct nfsd_file with an open file
  * @rqstp: the RPC transaction being executed
@@ -1132,6 +1171,10 @@ nfsd_file_do_acquire(struct svc_rqst *rqstp, struct svc_fh *fhp,
  * @may_flags: NFSD_MAY_ settings for the file
  * @pnf: OUT: new or found "struct nfsd_file" object
  *
+ * The nfsd_file_object returned by this API is reference-counted
+ * but not garbage-collected. The object is unhashed after the
+ * final nfsd_file_put().
+ *
  * Returns nfs_ok and sets @pnf on success; otherwise an nfsstat in
  * network byte order is returned.
  */
@@ -1139,7 +1182,7 @@ __be32
 nfsd_file_acquire(struct svc_rqst *rqstp, struct svc_fh *fhp,
 		  unsigned int may_flags, struct nfsd_file **pnf)
 {
-	return nfsd_file_do_acquire(rqstp, fhp, may_flags, pnf, true);
+	return nfsd_file_do_acquire(rqstp, fhp, may_flags, pnf, true, 0);
 }
 
 /**
@@ -1149,6 +1192,10 @@ nfsd_file_acquire(struct svc_rqst *rqstp, struct svc_fh *fhp,
  * @may_flags: NFSD_MAY_ settings for the file
  * @pnf: OUT: new or found "struct nfsd_file" object
  *
+ * The nfsd_file_object returned by this API is reference-counted
+ * but not garbage-collected. The object is released immediately
+ * one RCU grace period after the final nfsd_file_put().
+ *
  * Returns nfs_ok and sets @pnf on success; otherwise an nfsstat in
  * network byte order is returned.
  */
@@ -1156,7 +1203,7 @@ __be32
 nfsd_file_create(struct svc_rqst *rqstp, struct svc_fh *fhp,
 		 unsigned int may_flags, struct nfsd_file **pnf)
 {
-	return nfsd_file_do_acquire(rqstp, fhp, may_flags, pnf, false);
+	return nfsd_file_do_acquire(rqstp, fhp, may_flags, pnf, false, 0);
 }
 
 /*
diff --git a/fs/nfsd/filecache.h b/fs/nfsd/filecache.h
index f81c198f4ed6..0f6546bcd3e0 100644
--- a/fs/nfsd/filecache.h
+++ b/fs/nfsd/filecache.h
@@ -38,6 +38,7 @@ struct nfsd_file {
 #define NFSD_FILE_HASHED	(0)
 #define NFSD_FILE_PENDING	(1)
 #define NFSD_FILE_REFERENCED	(2)
+#define NFSD_FILE_GC		(3)
 	unsigned long		nf_flags;
 	struct inode		*nf_inode;	/* don't deref */
 	refcount_t		nf_ref;
@@ -55,6 +56,8 @@ void nfsd_file_put(struct nfsd_file *nf);
 struct nfsd_file *nfsd_file_get(struct nfsd_file *nf);
 void nfsd_file_close_inode_sync(struct inode *inode);
 bool nfsd_file_is_cached(struct inode *inode);
+__be32 nfsd_file_acquire_gc(struct svc_rqst *rqstp, struct svc_fh *fhp,
+		  unsigned int may_flags, struct nfsd_file **nfp);
 __be32 nfsd_file_acquire(struct svc_rqst *rqstp, struct svc_fh *fhp,
 		  unsigned int may_flags, struct nfsd_file **nfp);
 __be32 nfsd_file_create(struct svc_rqst *rqstp, struct svc_fh *fhp,
diff --git a/fs/nfsd/nfs3proc.c b/fs/nfsd/nfs3proc.c
index d12823371857..6a5ad6c91d8e 100644
--- a/fs/nfsd/nfs3proc.c
+++ b/fs/nfsd/nfs3proc.c
@@ -779,8 +779,8 @@ nfsd3_proc_commit(struct svc_rqst *rqstp)
 				(unsigned long long) argp->offset);
 
 	fh_copy(&resp->fh, &argp->fh);
-	resp->status = nfsd_file_acquire(rqstp, &resp->fh, NFSD_MAY_WRITE |
-					 NFSD_MAY_NOT_BREAK_LEASE, &nf);
+	resp->status = nfsd_file_acquire_gc(rqstp, &resp->fh, NFSD_MAY_WRITE |
+					    NFSD_MAY_NOT_BREAK_LEASE, &nf);
 	if (resp->status)
 		goto out;
 	resp->status = nfsd_commit(rqstp, &resp->fh, nf, argp->offset,
diff --git a/fs/nfsd/trace.h b/fs/nfsd/trace.h
index 9ebd67d461f9..4921144880d3 100644
--- a/fs/nfsd/trace.h
+++ b/fs/nfsd/trace.h
@@ -742,7 +742,8 @@ DEFINE_CLID_EVENT(confirmed_r);
 	__print_flags(val, "|",						\
 		{ 1 << NFSD_FILE_HASHED,	"HASHED" },		\
 		{ 1 << NFSD_FILE_PENDING,	"PENDING" },		\
-		{ 1 << NFSD_FILE_REFERENCED,	"REFERENCED"})
+		{ 1 << NFSD_FILE_REFERENCED,	"REFERENCED"},		\
+		{ 1 << NFSD_FILE_GC,		"GC"})
 
 DECLARE_EVENT_CLASS(nfsd_file_class,
 	TP_PROTO(struct nfsd_file *nf),
diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c
index 44f210ba17cc..89d682a56fc4 100644
--- a/fs/nfsd/vfs.c
+++ b/fs/nfsd/vfs.c
@@ -1060,7 +1060,7 @@ __be32 nfsd_read(struct svc_rqst *rqstp, struct svc_fh *fhp,
 	__be32 err;
 
 	trace_nfsd_read_start(rqstp, fhp, offset, *count);
-	err = nfsd_file_acquire(rqstp, fhp, NFSD_MAY_READ, &nf);
+	err = nfsd_file_acquire_gc(rqstp, fhp, NFSD_MAY_READ, &nf);
 	if (err)
 		return err;
 
@@ -1092,7 +1092,7 @@ nfsd_write(struct svc_rqst *rqstp, struct svc_fh *fhp, loff_t offset,
 
 	trace_nfsd_write_start(rqstp, fhp, offset, *cnt);
 
-	err = nfsd_file_acquire(rqstp, fhp, NFSD_MAY_WRITE, &nf);
+	err = nfsd_file_acquire_gc(rqstp, fhp, NFSD_MAY_WRITE, &nf);
 	if (err)
 		goto out;
 

[PATCH v3 4/7] NFSD: Use const pointers as parameters to fh_ helpers.

[Chuck Lever]

Enable callers to use const pointers where they are able to.

Reviewed-by: Jeff Layton <jlayton@kernel.org>
Tested-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
---
 fs/nfsd/nfsfh.h |   10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)

diff --git a/fs/nfsd/nfsfh.h b/fs/nfsd/nfsfh.h
index c3ae6414fc5c..513e028b0bbe 100644
--- a/fs/nfsd/nfsfh.h
+++ b/fs/nfsd/nfsfh.h
@@ -220,7 +220,7 @@ __be32	fh_update(struct svc_fh *);
 void	fh_put(struct svc_fh *);
 
 static __inline__ struct svc_fh *
-fh_copy(struct svc_fh *dst, struct svc_fh *src)
+fh_copy(struct svc_fh *dst, const struct svc_fh *src)
 {
 	WARN_ON(src->fh_dentry);
 
@@ -229,7 +229,7 @@ fh_copy(struct svc_fh *dst, struct svc_fh *src)
 }
 
 static inline void
-fh_copy_shallow(struct knfsd_fh *dst, struct knfsd_fh *src)
+fh_copy_shallow(struct knfsd_fh *dst, const struct knfsd_fh *src)
 {
 	dst->fh_size = src->fh_size;
 	memcpy(&dst->fh_raw, &src->fh_raw, src->fh_size);
@@ -243,7 +243,8 @@ fh_init(struct svc_fh *fhp, int maxsize)
 	return fhp;
 }
 
-static inline bool fh_match(struct knfsd_fh *fh1, struct knfsd_fh *fh2)
+static inline bool fh_match(const struct knfsd_fh *fh1,
+			    const struct knfsd_fh *fh2)
 {
 	if (fh1->fh_size != fh2->fh_size)
 		return false;
@@ -252,7 +253,8 @@ static inline bool fh_match(struct knfsd_fh *fh1, struct knfsd_fh *fh2)
 	return true;
 }
 
-static inline bool fh_fsid_match(struct knfsd_fh *fh1, struct knfsd_fh *fh2)
+static inline bool fh_fsid_match(const struct knfsd_fh *fh1,
+				 const struct knfsd_fh *fh2)
 {
 	if (fh1->fh_fsid_type != fh2->fh_fsid_type)
 		return false;

[PATCH v3 5/7] NFSD: Use rhashtable for managing nfs4_file objects

[Chuck Lever]

fh_match() is costly, especially when filehandles are large (as is
the case for NFSv4). It needs to be used sparingly when searching
data structures. Unfortunately, With common workloads, I see
multiple thousands of objects stored in file_hashtbl[], which always
has only 256 buckets, which makes the hash chains quite lengthy.

Walking long hash chains with the state_lock held blocks other
activity that needs that lock.

To help mitigate the cost of searching with fh_match(), replace the
nfs4_file hash table with an rhashtable, which can dynamically
resize its bucket array to minimize hash chain length. The ideal for
this use case is one bucket per inode.

The result is an improvement in the latency of NFSv4 operations
and the reduction of nfsd CPU utilization due to the cost of
fh_match() and the CPU cache misses incurred while walking long
hash chains in the nfs4_file hash table.

Note that hash removal is no longer protected by the state_lock.
This is because insert_nfs4_file() takes the RCU read lock then the
state_lock. rhltable_remove() takes the RCU read lock internally;
if remove_nfs4_file() took the state_lock before calling
rhltable_remove(), that would result in a lock inversion.

Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
---
 fs/nfsd/nfs4state.c |  214 +++++++++++++++++++++++++++++++++++----------------
 fs/nfsd/state.h     |    5 -
 2 files changed, 146 insertions(+), 73 deletions(-)

diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c
index 2b850de288cf..9ebf99498647 100644
--- a/fs/nfsd/nfs4state.c
+++ b/fs/nfsd/nfs4state.c
@@ -44,7 +44,9 @@
 #include <linux/jhash.h>
 #include <linux/string_helpers.h>
 #include <linux/fsnotify.h>
+#include <linux/rhashtable.h>
 #include <linux/nfs_ssc.h>
+
 #include "xdr4.h"
 #include "xdr4cb.h"
 #include "vfs.h"
@@ -84,6 +86,7 @@ static bool check_for_locks(struct nfs4_file *fp, struct nfs4_lockowner *lowner)
 static void nfs4_free_ol_stateid(struct nfs4_stid *stid);
 void nfsd4_end_grace(struct nfsd_net *nn);
 static void _free_cpntf_state_locked(struct nfsd_net *nn, struct nfs4_cpntf_state *cps);
+static void remove_nfs4_file(struct nfs4_file *fi);
 
 /* Locking: */
 
@@ -577,11 +580,8 @@ static void nfsd4_free_file_rcu(struct rcu_head *rcu)
 void
 put_nfs4_file(struct nfs4_file *fi)
 {
-	might_lock(&state_lock);
-
-	if (refcount_dec_and_lock(&fi->fi_ref, &state_lock)) {
-		hlist_del_rcu(&fi->fi_hash);
-		spin_unlock(&state_lock);
+	if (refcount_dec_and_test(&fi->fi_ref)) {
+		remove_nfs4_file(fi);
 		WARN_ON_ONCE(!list_empty(&fi->fi_clnt_odstate));
 		WARN_ON_ONCE(!list_empty(&fi->fi_delegations));
 		call_rcu(&fi->fi_rcu, nfsd4_free_file_rcu);
@@ -695,19 +695,82 @@ static unsigned int ownerstr_hashval(struct xdr_netobj *ownername)
 	return ret & OWNER_HASH_MASK;
 }
 
-/* hash table for nfs4_file */
-#define FILE_HASH_BITS                   8
-#define FILE_HASH_SIZE                  (1 << FILE_HASH_BITS)
+static struct rhltable nfs4_file_rhltable ____cacheline_aligned_in_smp;
+
+/*
+ * The returned hash value is based solely on the address of an in-code
+ * inode, a pointer to a slab-allocated object. The entropy in such a
+ * pointer is concentrated in its middle bits.
+ */
+static u32 nfs4_file_inode_hash(const struct inode *inode, u32 seed)
+{
+	u32 k;
+
+	k = ((unsigned long)inode) >> L1_CACHE_SHIFT;
+	return jhash2(&k, 1, seed);
+}
 
-static unsigned int file_hashval(struct svc_fh *fh)
+/**
+ * nfs4_file_key_hashfn - Compute the hash value of a lookup key
+ * @data: key on which to compute the hash value
+ * @len: rhash table's key_len parameter (unused)
+ * @seed: rhash table's random seed of the day
+ *
+ * Return value:
+ *   Computed 32-bit hash value
+ */
+static u32 nfs4_file_key_hashfn(const void *data, u32 len, u32 seed)
 {
-	struct inode *inode = d_inode(fh->fh_dentry);
+	const struct svc_fh *fhp = data;
 
-	/* XXX: why not (here & in file cache) use inode? */
-	return (unsigned int)hash_long(inode->i_ino, FILE_HASH_BITS);
+	return nfs4_file_inode_hash(d_inode(fhp->fh_dentry), seed);
 }
 
-static struct hlist_head file_hashtbl[FILE_HASH_SIZE];
+/**
+ * nfs4_file_obj_hashfn - Compute the hash value of an nfs4_file object
+ * @data: object on which to compute the hash value
+ * @len: rhash table's key_len parameter (unused)
+ * @seed: rhash table's random seed of the day
+ *
+ * Return value:
+ *   Computed 32-bit hash value
+ */
+static u32 nfs4_file_obj_hashfn(const void *data, u32 len, u32 seed)
+{
+	const struct nfs4_file *fi = data;
+
+	return nfs4_file_inode_hash(fi->fi_inode, seed);
+}
+
+/**
+ * nfs4_file_obj_cmpfn - Match a cache item against search criteria
+ * @arg: search criteria
+ * @ptr: cache item to check
+ *
+ * Return values:
+ *   %0 - Success; item matches search criteria
+ */
+static int nfs4_file_obj_cmpfn(struct rhashtable_compare_arg *arg,
+			       const void *ptr)
+{
+	const struct svc_fh *fhp = arg->key;
+	const struct nfs4_file *fi = ptr;
+
+	return fh_match(&fi->fi_fhandle, &fhp->fh_handle) ? 0 : 1;
+}
+
+static const struct rhashtable_params nfs4_file_rhash_params = {
+	.key_len		= sizeof_field(struct nfs4_file, fi_inode),
+	.key_offset		= offsetof(struct nfs4_file, fi_inode),
+	.head_offset		= offsetof(struct nfs4_file, fi_rhash),
+	.hashfn			= nfs4_file_key_hashfn,
+	.obj_hashfn		= nfs4_file_obj_hashfn,
+	.obj_cmpfn		= nfs4_file_obj_cmpfn,
+
+	/* Reduce resizing churn on light workloads */
+	.min_size		= 512,		/* buckets */
+	.automatic_shrinking	= true,
+};
 
 /*
  * Check if courtesy clients have conflicting access and resolve it if possible
@@ -4251,11 +4314,8 @@ static struct nfs4_file *nfsd4_alloc_file(void)
 }
 
 /* OPEN Share state helper functions */
-static void nfsd4_init_file(struct svc_fh *fh, unsigned int hashval,
-				struct nfs4_file *fp)
+static void init_nfs4_file(const struct svc_fh *fh, struct nfs4_file *fp)
 {
-	lockdep_assert_held(&state_lock);
-
 	refcount_set(&fp->fi_ref, 1);
 	spin_lock_init(&fp->fi_lock);
 	INIT_LIST_HEAD(&fp->fi_stateids);
@@ -4273,7 +4333,6 @@ static void nfsd4_init_file(struct svc_fh *fh, unsigned int hashval,
 	INIT_LIST_HEAD(&fp->fi_lo_states);
 	atomic_set(&fp->fi_lo_recalls, 0);
 #endif
-	hlist_add_head_rcu(&fp->fi_hash, &file_hashtbl[hashval]);
 }
 
 void
@@ -4626,71 +4685,74 @@ move_to_close_lru(struct nfs4_ol_stateid *s, struct net *net)
 		nfs4_put_stid(&last->st_stid);
 }
 
-/* search file_hashtbl[] for file */
-static struct nfs4_file *
-find_file_locked(struct svc_fh *fh, unsigned int hashval)
+static struct nfs4_file *find_nfs4_file(const struct svc_fh *fhp)
 {
-	struct nfs4_file *fp;
+	struct rhlist_head *tmp, *list;
+	struct nfs4_file *fi = NULL;
 
-	hlist_for_each_entry_rcu(fp, &file_hashtbl[hashval], fi_hash,
-				lockdep_is_held(&state_lock)) {
-		if (fh_match(&fp->fi_fhandle, &fh->fh_handle)) {
-			if (refcount_inc_not_zero(&fp->fi_ref))
-				return fp;
+	list = rhltable_lookup(&nfs4_file_rhltable, fhp,
+			       nfs4_file_rhash_params);
+	rhl_for_each_entry_rcu(fi, tmp, list, fi_rhash)
+		if (fh_match(&fi->fi_fhandle, &fhp->fh_handle)) {
+			if (!refcount_inc_not_zero(&fi->fi_ref))
+				fi = NULL;
+			break;
 		}
-	}
-	return NULL;
+	return fi;
 }
 
-static struct nfs4_file *insert_file(struct nfs4_file *new, struct svc_fh *fh,
-				     unsigned int hashval)
+/*
+ * On hash insertion, identify entries with the same inode but
+ * distinct filehandles. They will all be in the same hash bucket
+ * because we hash by the address of the nfs4_file's struct inode.
+ */
+static struct nfs4_file *insert_file(struct nfs4_file *new,
+				     const struct svc_fh *fhp)
 {
-	struct nfs4_file *fp;
-	struct nfs4_file *ret = NULL;
-	bool alias_found = false;
+	struct rhlist_head *tmp, *list;
+	struct nfs4_file *fi;
+	int err;
 
 	spin_lock(&state_lock);
-	hlist_for_each_entry_rcu(fp, &file_hashtbl[hashval], fi_hash,
-				 lockdep_is_held(&state_lock)) {
-		if (fh_match(&fp->fi_fhandle, &fh->fh_handle)) {
-			if (refcount_inc_not_zero(&fp->fi_ref))
-				ret = fp;
-		} else if (d_inode(fh->fh_dentry) == fp->fi_inode)
-			fp->fi_aliased = alias_found = true;
-	}
-	if (likely(ret == NULL)) {
-		nfsd4_init_file(fh, hashval, new);
-		new->fi_aliased = alias_found;
-		ret = new;
+
+	err = rhltable_insert_key(&nfs4_file_rhltable, fhp,
+				  &new->fi_rhash,
+				  nfs4_file_rhash_params);
+	if (err) {
+		spin_unlock(&state_lock);
+		return NULL;
 	}
+
+	list = rhltable_lookup(&nfs4_file_rhltable, fhp,
+			       nfs4_file_rhash_params);
+	rhl_for_each_entry_rcu(fi, tmp, list, fi_rhash)
+		if (!fh_match(&fi->fi_fhandle, &fhp->fh_handle)) {
+			fi->fi_aliased = true;
+			new->fi_aliased = true;
+		}
+
 	spin_unlock(&state_lock);
-	return ret;
+	return new;
 }
 
-static struct nfs4_file * find_file(struct svc_fh *fh)
+static noinline struct nfs4_file *
+insert_nfs4_file(struct nfs4_file *new, const struct svc_fh *fhp)
 {
-	struct nfs4_file *fp;
-	unsigned int hashval = file_hashval(fh);
+	struct nfs4_file *fi;
 
-	rcu_read_lock();
-	fp = find_file_locked(fh, hashval);
-	rcu_read_unlock();
-	return fp;
+	/* Do not hash the same filehandle more than once */
+	fi = find_nfs4_file(fhp);
+	if (unlikely(fi))
+		return fi;
+
+	init_nfs4_file(fhp, new);
+	return insert_file(new, fhp);
 }
 
-static struct nfs4_file *
-find_or_add_file(struct nfs4_file *new, struct svc_fh *fh)
+static noinline void remove_nfs4_file(struct nfs4_file *fi)
 {
-	struct nfs4_file *fp;
-	unsigned int hashval = file_hashval(fh);
-
-	rcu_read_lock();
-	fp = find_file_locked(fh, hashval);
-	rcu_read_unlock();
-	if (fp)
-		return fp;
-
-	return insert_file(new, fh, hashval);
+	rhltable_remove(&nfs4_file_rhltable, &fi->fi_rhash,
+			nfs4_file_rhash_params);
 }
 
 /*
@@ -4703,9 +4765,12 @@ nfs4_share_conflict(struct svc_fh *current_fh, unsigned int deny_type)
 	struct nfs4_file *fp;
 	__be32 ret = nfs_ok;
 
-	fp = find_file(current_fh);
+	rcu_read_lock();
+	fp = find_nfs4_file(current_fh);
+	rcu_read_unlock();
 	if (!fp)
 		return ret;
+
 	/* Check for conflicting share reservations */
 	spin_lock(&fp->fi_lock);
 	if (fp->fi_share_deny & deny_type)
@@ -5548,7 +5613,11 @@ nfsd4_process_open2(struct svc_rqst *rqstp, struct svc_fh *current_fh, struct nf
 	 * and check for delegations in the process of being recalled.
 	 * If not found, create the nfs4_file struct
 	 */
-	fp = find_or_add_file(open->op_file, current_fh);
+	rcu_read_lock();
+	fp = insert_nfs4_file(open->op_file, current_fh);
+	rcu_read_unlock();
+	if (unlikely(!fp))
+		return nfserr_jukebox;
 	if (fp != open->op_file) {
 		status = nfs4_check_deleg(cl, open, &dp);
 		if (status)
@@ -7905,10 +7974,16 @@ nfs4_state_start(void)
 {
 	int ret;
 
-	ret = nfsd4_create_callback_queue();
+	ret = rhltable_init(&nfs4_file_rhltable, &nfs4_file_rhash_params);
 	if (ret)
 		return ret;
 
+	ret = nfsd4_create_callback_queue();
+	if (ret) {
+		rhltable_destroy(&nfs4_file_rhltable);
+		return ret;
+	}
+
 	set_max_delegations();
 	return 0;
 }
@@ -7939,6 +8014,7 @@ nfs4_state_shutdown_net(struct net *net)
 
 	nfsd4_client_tracking_exit(net);
 	nfs4_state_destroy_net(net);
+	rhltable_destroy(&nfs4_file_rhltable);
 #ifdef CONFIG_NFSD_V4_2_INTER_SSC
 	nfsd4_ssc_shutdown_umount(nn);
 #endif
diff --git a/fs/nfsd/state.h b/fs/nfsd/state.h
index ae596dbf8667..ad20467c9dee 100644
--- a/fs/nfsd/state.h
+++ b/fs/nfsd/state.h
@@ -536,16 +536,13 @@ struct nfs4_clnt_odstate {
  * inode can have multiple filehandles associated with it, so there is
  * (potentially) a many to one relationship between this struct and struct
  * inode.
- *
- * These are hashed by filehandle in the file_hashtbl, which is protected by
- * the global state_lock spinlock.
  */
 struct nfs4_file {
 	refcount_t		fi_ref;
 	struct inode *		fi_inode;
 	bool			fi_aliased;
 	spinlock_t		fi_lock;
-	struct hlist_node       fi_hash;	/* hash on fi_fhandle */
+	struct rhlist_head	fi_rhash;
 	struct list_head        fi_stateids;
 	union {
 		struct list_head	fi_delegations;

[PATCH v3 6/7] NFSD: Clean up nfs4_preprocess_stateid_op() call sites

[Chuck Lever]

Remove the lame-duck dprintk()s around nfs4_preprocess_stateid_op()
call sites.

Reviewed-by: Jeff Layton <jlayton@kernel.org>
Tested-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
---
 fs/nfsd/nfs4proc.c |   31 +++++++------------------------
 1 file changed, 7 insertions(+), 24 deletions(-)

diff --git a/fs/nfsd/nfs4proc.c b/fs/nfsd/nfs4proc.c
index d0d976f847ca..42b81e88ea14 100644
--- a/fs/nfsd/nfs4proc.c
+++ b/fs/nfsd/nfs4proc.c
@@ -951,12 +951,7 @@ nfsd4_read(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 	status = nfs4_preprocess_stateid_op(rqstp, cstate, &cstate->current_fh,
 					&read->rd_stateid, RD_STATE,
 					&read->rd_nf, NULL);
-	if (status) {
-		dprintk("NFSD: nfsd4_read: couldn't process stateid!\n");
-		goto out;
-	}
-	status = nfs_ok;
-out:
+
 	read->rd_rqstp = rqstp;
 	read->rd_fhp = &cstate->current_fh;
 	return status;
@@ -1125,10 +1120,8 @@ nfsd4_setattr(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 		status = nfs4_preprocess_stateid_op(rqstp, cstate,
 				&cstate->current_fh, &setattr->sa_stateid,
 				WR_STATE, NULL, NULL);
-		if (status) {
-			dprintk("NFSD: nfsd4_setattr: couldn't process stateid!\n");
+		if (status)
 			return status;
-		}
 	}
 	err = fh_want_write(&cstate->current_fh);
 	if (err)
@@ -1176,10 +1169,8 @@ nfsd4_write(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 			       write->wr_offset, cnt);
 	status = nfs4_preprocess_stateid_op(rqstp, cstate, &cstate->current_fh,
 						stateid, WR_STATE, &nf, NULL);
-	if (status) {
-		dprintk("NFSD: nfsd4_write: couldn't process stateid!\n");
+	if (status)
 		return status;
-	}
 
 	write->wr_how_written = write->wr_stable_how;
 
@@ -1210,17 +1201,13 @@ nfsd4_verify_copy(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 
 	status = nfs4_preprocess_stateid_op(rqstp, cstate, &cstate->save_fh,
 					    src_stateid, RD_STATE, src, NULL);
-	if (status) {
-		dprintk("NFSD: %s: couldn't process src stateid!\n", __func__);
+	if (status)
 		goto out;
-	}
 
 	status = nfs4_preprocess_stateid_op(rqstp, cstate, &cstate->current_fh,
 					    dst_stateid, WR_STATE, dst, NULL);
-	if (status) {
-		dprintk("NFSD: %s: couldn't process dst stateid!\n", __func__);
+	if (status)
 		goto out_put_src;
-	}
 
 	/* fix up for NFS-specific error code */
 	if (!S_ISREG(file_inode((*src)->nf_file)->i_mode) ||
@@ -1955,10 +1942,8 @@ nfsd4_fallocate(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 	status = nfs4_preprocess_stateid_op(rqstp, cstate, &cstate->current_fh,
 					    &fallocate->falloc_stateid,
 					    WR_STATE, &nf, NULL);
-	if (status != nfs_ok) {
-		dprintk("NFSD: nfsd4_fallocate: couldn't process stateid!\n");
+	if (status != nfs_ok)
 		return status;
-	}
 
 	status = nfsd4_vfs_fallocate(rqstp, &cstate->current_fh, nf->nf_file,
 				     fallocate->falloc_offset,
@@ -2014,10 +1999,8 @@ nfsd4_seek(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 	status = nfs4_preprocess_stateid_op(rqstp, cstate, &cstate->current_fh,
 					    &seek->seek_stateid,
 					    RD_STATE, &nf, NULL);
-	if (status) {
-		dprintk("NFSD: nfsd4_seek: couldn't process stateid!\n");
+	if (status)
 		return status;
-	}
 
 	switch (seek->seek_whence) {
 	case NFS4_CONTENT_DATA:

[PATCH v3 7/7] NFSD: Trace delegation revocations

[Chuck Lever]

Delegation revocation is an exceptional event that is not otherwise
visible externally (eg, no network traffic is emitted). Generate a
trace record when it occurs so that revocation can be observed or
other activity can be triggered. Example:

nfsd-1104  [005]  1912.002544: nfsd_stid_revoke:        client 633c9343:4e82788d stateid 00000003:00000001 ref=2 type=DELEG

Trace infrastructure is provided for subsequent additional tracing
related to nfs4_stid activity.

Reviewed-by: Jeff Layton <jlayton@kernel.org>
Tested-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
---
 fs/nfsd/nfs4state.c |    2 ++
 fs/nfsd/trace.h     |   55 +++++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 57 insertions(+)

diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c
index 9ebf99498647..aaae2e460039 100644
--- a/fs/nfsd/nfs4state.c
+++ b/fs/nfsd/nfs4state.c
@@ -1401,6 +1401,8 @@ static void revoke_delegation(struct nfs4_delegation *dp)
 
 	WARN_ON(!list_empty(&dp->dl_recall_lru));
 
+	trace_nfsd_stid_revoke(&dp->dl_stid);
+
 	if (clp->cl_minorversion) {
 		dp->dl_stid.sc_type = NFS4_REVOKED_DELEG_STID;
 		refcount_inc(&dp->dl_stid.sc_count);
diff --git a/fs/nfsd/trace.h b/fs/nfsd/trace.h
index 4921144880d3..23fb39c957af 100644
--- a/fs/nfsd/trace.h
+++ b/fs/nfsd/trace.h
@@ -561,6 +561,61 @@ DEFINE_EVENT(nfsd_stateseqid_class, nfsd_##name, \
 DEFINE_STATESEQID_EVENT(preprocess);
 DEFINE_STATESEQID_EVENT(open_confirm);
 
+TRACE_DEFINE_ENUM(NFS4_OPEN_STID);
+TRACE_DEFINE_ENUM(NFS4_LOCK_STID);
+TRACE_DEFINE_ENUM(NFS4_DELEG_STID);
+TRACE_DEFINE_ENUM(NFS4_CLOSED_STID);
+TRACE_DEFINE_ENUM(NFS4_REVOKED_DELEG_STID);
+TRACE_DEFINE_ENUM(NFS4_CLOSED_DELEG_STID);
+TRACE_DEFINE_ENUM(NFS4_LAYOUT_STID);
+
+#define show_stid_type(x)						\
+	__print_flags(x, "|",						\
+		{ NFS4_OPEN_STID,		"OPEN" },		\
+		{ NFS4_LOCK_STID,		"LOCK" },		\
+		{ NFS4_DELEG_STID,		"DELEG" },		\
+		{ NFS4_CLOSED_STID,		"CLOSED" },		\
+		{ NFS4_REVOKED_DELEG_STID,	"REVOKED" },		\
+		{ NFS4_CLOSED_DELEG_STID,	"CLOSED_DELEG" },	\
+		{ NFS4_LAYOUT_STID,		"LAYOUT" })
+
+DECLARE_EVENT_CLASS(nfsd_stid_class,
+	TP_PROTO(
+		const struct nfs4_stid *stid
+	),
+	TP_ARGS(stid),
+	TP_STRUCT__entry(
+		__field(unsigned long, sc_type)
+		__field(int, sc_count)
+		__field(u32, cl_boot)
+		__field(u32, cl_id)
+		__field(u32, si_id)
+		__field(u32, si_generation)
+	),
+	TP_fast_assign(
+		const stateid_t *stp = &stid->sc_stateid;
+
+		__entry->sc_type = stid->sc_type;
+		__entry->sc_count = refcount_read(&stid->sc_count);
+		__entry->cl_boot = stp->si_opaque.so_clid.cl_boot;
+		__entry->cl_id = stp->si_opaque.so_clid.cl_id;
+		__entry->si_id = stp->si_opaque.so_id;
+		__entry->si_generation = stp->si_generation;
+	),
+	TP_printk("client %08x:%08x stateid %08x:%08x ref=%d type=%s",
+		__entry->cl_boot, __entry->cl_id,
+		__entry->si_id, __entry->si_generation,
+		__entry->sc_count, show_stid_type(__entry->sc_type)
+	)
+);
+
+#define DEFINE_STID_EVENT(name)					\
+DEFINE_EVENT(nfsd_stid_class, nfsd_stid_##name,			\
+	TP_PROTO(const struct nfs4_stid *stid),			\
+	TP_ARGS(stid))
+
+DEFINE_STID_EVENT(revoke);
+
 DECLARE_EVENT_CLASS(nfsd_clientid_class,
 	TP_PROTO(const clientid_t *clid),
 	TP_ARGS(clid),