Skip to contentSkip to conversation

NFSv4.1 Sessions server code for 2.6.30

PATCH Superseded Benny Halevy 7 participants 204 messages Mar 27, 2009 – Apr 3, 2009 9 messages not in the archive Page 1 of 3

Subsystems (4)

v1 of 3 Superseded Latest is v3 →

lore
Share & tools
lore
#0 lore raw

cover NFSv4.1 Sessions server code for 2.6.30

Hi Bruce,

Here's the latest server patches implementing the NFSv4.1 Sessions features.

This patchset is based over your for-2.6.30 branch and is also available from
git://linux-nfs.org/~bhalevy/linux-pnfs.git nfsd41-for-2.6.30 (tagged as nfsd41-for-2.6.30-2009-03-27)

The implementation status highlights are captured in Documentation/filesystems/nfs41-server.txt.
The gist of it is that the implementation is functional to the level of passing Connectathon tests with v2,v3,v4,v4.1 (tested with CONFIG_NFS_V4_1 defined to 'y')

The patches are listed below.

Thanks for your review and help with this project so far. Kudos to all the contributors (Authors/SOBs: Andy Adamson, Dean Hildebrand, Marc Eshel, Mike Sager, Ricardo Labiaga, and myself) This is a pretty significant milestone for us (well, the real one will be when all this ends up working flawlessly in the kernel. :)

Benny

[PATCH 01/47] nfsd: don't use the deferral service, return NFS4ERR_DELAY [PATCH 02/47] sunrpc: add cl_private field to struct rpc_clnt
[PATCH 03/47] nfsd: embed nfsd4_current_state in nfsd4_compoundres [PATCH 04/47] nfsd: add a struct nfsd4_slot pointer to struct nfsd4_compound_state [PATCH 05/47] nfs41: common protocol definitions
[PATCH 06/47] nfsd41: change NFSERR_REPLAY_ME
[PATCH 07/47] nfsd41: Add Kconfig symbols for NFSv4.1
[PATCH 08/47] nfsd41: define nfs41 error codes
[PATCH 09/47] nfsd41: sessions basic data types
[PATCH 10/47] nfsd41: introduce nfs4_client cl_sessions list
[PATCH 11/47] nfsd41: release_session when client is expired
[PATCH 12/47] nfsd41: sessionid hashing
[PATCH 13/47] nfsd41: xdr infrastructure
[PATCH 14/47] nfsd: remove nfsd4_ops array size
[PATCH 15/47] nfsd41: proc stubs
[PATCH 16/47] nfsd41: exchange_id operation
[PATCH 17/47] nfsd41: match clientid establishment method
[PATCH 18/47] nfsd41: sequence operation
[PATCH 19/47] nfsd41: enforce NFS4ERR_SEQUENCE_POS operation order rules [PATCH 20/47] nfsd41: DRC save, restore, and clear functions
[PATCH 21/47] nfsd41: hard page limit for DRC
[PATCH 22/47] nfsd41: nfsd DRC logic
[PATCH 23/47] nfsd41: clear DRC cache on free_session
[PATCH 24/47] nfsd41: create_session operation
[PATCH 25/47] nfsd41: Add a create session replay cache
[PATCH 26/47] nfsd41: non-page DRC for solo sequence responses
[PATCH 27/47] nfsd41: destroy_session operation
[PATCH 28/47] nfsd41: stateid handling
[PATCH 29/47] nfsd41: check encode size for sessions maxresponse cached [PATCH 30/47] nfsd41: clientid handling
[PATCH 31/47] nfsd41: access_valid
[PATCH 32/47] nfsd41: add OPEN4_SHARE_ACCESS_WANT nfs4_stateid bmap [PATCH 33/47] nfsd41: provide support for minor version 1 at rpc level [PATCH 34/47] nfsd: cleanup nfs4.0 callback encode routines
[PATCH 35/47] nfsd: minorversion support for the back channel
[PATCH 36/47] nfsd41: sunrpc: Added rpc server-side backchannel handling [PATCH 37/47] nfsd41: callback infrastructure
[PATCH 38/47] nfsd41: Remember the auth flavor to use for callbacks [PATCH 39/47] nfsd41: introduce cl_cb_mutex
[PATCH 40/47] nfsd41: cb_sequence callback
[PATCH 41/47] nfsd41: introduce nfs4_cb_call_sync for nfs4 and nfs41 [PATCH 42/47] nfsd41: cb_recall callback
[PATCH 43/47] nfsd41: pass writable attrs mask to nfsd4_decode_fattr [PATCH 44/47] nfsd41: support for 3-word long attribute bitmask [PATCH 45/47] nfsd41: SUPPATTR_EXCLCREAT attribute
[PATCH 46/47] nfsd41: CREATE_EXCLUSIVE4_1
[PATCH 47/47] nfsd41: Documentation/filesystems/nfs41-server.txt

#1 lore raw

PATCH 1/47 nfsd: don't use the deferral service, return NFS4ERR_DELAY

From: Andy Adamson <[email protected]>

On an NFSv4.1 server cache miss that causes an upcall, NFS4ERR_DELAY will be returned. It is up to the NFSv4.1 client to resend only the operations that have not been processed.

Initialize rq_usedeferral to 1 in svc_process(). rq_usedeferral was initialized to 0, which means that the first NFSv4.0 or NFSv4.1 rpc would be dropped due to the logic in svc_defer(). Will be turned off in nfsd4_proc_compound() only when NFSv4 Sessions are used.

Note: this isn't an adequate solution on its own. It's acceptable as a way to get some minimal 4.1 up and working, but we're going to have to find a way to avoid returning DELAY in all common cases before 4.1 can really be considered ready.

[nfsd41: reverse rq_nodeferral negative logic]

use positive rq_usedeferral logic instead.

[sunrpc: initialize rq_usedeferral]
---
 fs/nfsd/nfs4proc.c         |    8 ++++++++
 include/linux/sunrpc/svc.h |    1 +
 net/sunrpc/svc.c           |    2 ++
 net/sunrpc/svc_xprt.c      |    2 +-
 4 files changed, 12 insertions(+), 1 deletions(-)
fs/nfsd/nfs4proc.c +8 -0
index f156b85..7839654 100644
@@ -873,6 +873,8 @@ nfsd4_proc_compound(struct svc_rqst *rqstp,
 	resp->tag = args->tag;
 	resp->opcnt = 0;
 	resp->rqstp = rqstp;
+	/* Use the deferral mechanism only for NFSv4.0 compounds */
+	rqstp->rq_usedeferral = (args->minorversion == 0);
 
 	/*
 	 * According to RFC3010, this takes precedence over all other errors.
@@ -957,10 +959,16 @@ encode_op:
 
 		nfsd4_increment_op_stats(op->opnum);
 	}
+	if (!rqstp->rq_usedeferral && status == nfserr_dropit) {
+		dprintk("%s Dropit - send NFS4ERR_DELAY\n", __func__);
+		status = nfserr_jukebox;
+	}
 
 	cstate_free(cstate);
 out:
 	nfsd4_release_compoundargs(args);
+	/* Reset deferral mechanism for RPC deferrals */
+	rqstp->rq_usedeferral = 1;
 	dprintk("nfsv4 compound returned %d\n", ntohl(status));
 	return status;
 }
include/linux/sunrpc/svc.h +1 -0
index 9f9f699..815dd58 100644
@@ -230,6 +230,7 @@ struct svc_rqst {
 	struct svc_cred		rq_cred;	/* auth info */
 	void *			rq_xprt_ctxt;	/* transport specific context ptr */
 	struct svc_deferred_req*rq_deferred;	/* deferred request we are replaying */
+	int			rq_usedeferral;	/* use deferral */
 
 	size_t			rq_xprt_hlen;	/* xprt header len */
 	struct xdr_buf		rq_arg;
net/sunrpc/svc.c +2 -0
index c51fed4..6334858 100644
@@ -1023,6 +1023,8 @@ svc_process(struct svc_rqst *rqstp)
 	rqstp->rq_res.tail[0].iov_len = 0;
 	/* Will be turned off only in gss privacy case: */
 	rqstp->rq_splice_ok = 1;
+	/* Will be turned off only when NFSv4 Sessions are used */
+	rqstp->rq_usedeferral = 1;
 
 	/* Setup reply header */
 	rqstp->rq_xprt->xpt_ops->xpo_prep_reply_hdr(rqstp);
net/sunrpc/svc_xprt.c +1 -1
index 1e66f24..600d091 100644
@@ -974,7 +974,7 @@ static struct cache_deferred_req *svc_defer(struct cache_req *req)
 	struct svc_rqst *rqstp = container_of(req, struct svc_rqst, rq_chandle);
 	struct svc_deferred_req *dr;
 
-	if (rqstp->rq_arg.page_len)
+	if (rqstp->rq_arg.page_len || !rqstp->rq_usedeferral)
 		return NULL; /* if more than a page, give up FIXME */
 	if (rqstp->rq_deferred) {
 		dr = rqstp->rq_deferred;
Signature · 1 line
1.6.2.1
#4 lore raw

PATCH 2/47 sunrpc: add cl_private field to struct rpc_clnt

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

Note: the NFSv4.1 client also uses (and declares) this pointer.

---
 include/linux/sunrpc/clnt.h |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)
include/linux/sunrpc/clnt.h +1 -0
index c39a210..928c547 100644
@@ -53,6 +53,7 @@ struct rpc_clnt {
 	char			cl_pathname[30];/* Path in rpc_pipe_fs */
 	struct vfsmount *	cl_vfsmnt;
 	struct dentry *		cl_dentry;	/* inode */
+	void			*cl_private;	/* private data */
 	struct rpc_clnt *	cl_parent;	/* Points to parent of clones */
 	struct rpc_rtt		cl_rtt_default;
 	struct rpc_timeout	cl_timeout_default;
Signature · 1 line
1.6.2.1
#15 lore raw

PATCH 3/47 nfsd: embed nfsd4_current_state in nfsd4_compoundres

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

Remove the allocation of struct nfsd4_compound_state.

---
 fs/nfsd/nfs4proc.c        |   40 +++++++++-------------------------------
 include/linux/nfsd/xdr4.h |    9 +++++----
 2 files changed, 14 insertions(+), 35 deletions(-)
fs/nfsd/nfs4proc.c +9 -31
index 7839654..c8e1ad0 100644
@@ -809,29 +809,6 @@ static inline void nfsd4_increment_op_stats(u32 opnum)
 		nfsdstats.nfs4_opcount[opnum]++;
 }
 
-static void cstate_free(struct nfsd4_compound_state *cstate)
-{
-	if (cstate == NULL)
-		return;
-	fh_put(&cstate->current_fh);
-	fh_put(&cstate->save_fh);
-	BUG_ON(cstate->replay_owner);
-	kfree(cstate);
-}
-
-static struct nfsd4_compound_state *cstate_alloc(void)
-{
-	struct nfsd4_compound_state *cstate;
-
-	cstate = kmalloc(sizeof(struct nfsd4_compound_state), GFP_KERNEL);
-	if (cstate == NULL)
-		return NULL;
-	fh_init(&cstate->current_fh, NFS4_FHSIZE);
-	fh_init(&cstate->save_fh, NFS4_FHSIZE);
-	cstate->replay_owner = NULL;
-	return cstate;
-}
-
 typedef __be32(*nfsd4op_func)(struct svc_rqst *, struct nfsd4_compound_state *,
 			      void *);
 
@@ -859,12 +836,13 @@ nfsd4_proc_compound(struct svc_rqst *rqstp,
 {
 	struct nfsd4_op	*op;
 	struct nfsd4_operation *opdesc;
-	struct nfsd4_compound_state *cstate = NULL;
+	struct nfsd4_compound_state *cstate = &resp->cstate;
 	int		slack_bytes;
 	__be32		status;
 
 	resp->xbuf = &rqstp->rq_res;
-	resp->p = rqstp->rq_res.head[0].iov_base + rqstp->rq_res.head[0].iov_len;
+	resp->p = rqstp->rq_res.head[0].iov_base +
+						rqstp->rq_res.head[0].iov_len;
 	resp->tagp = resp->p;
 	/* reserve space for: taglen, tag, and opcnt */
 	resp->p += 2 + XDR_QUADLEN(args->taglen);
@@ -873,6 +851,9 @@ nfsd4_proc_compound(struct svc_rqst *rqstp,
 	resp->tag = args->tag;
 	resp->opcnt = 0;
 	resp->rqstp = rqstp;
+	resp->cstate.replay_owner = NULL;
+	fh_init(&resp->cstate.current_fh, NFS4_FHSIZE);
+	fh_init(&resp->cstate.save_fh, NFS4_FHSIZE);
 	/* Use the deferral mechanism only for NFSv4.0 compounds */
 	rqstp->rq_usedeferral = (args->minorversion == 0);
 
@@ -883,11 +864,6 @@ nfsd4_proc_compound(struct svc_rqst *rqstp,
 	if (args->minorversion > NFSD_SUPPORTED_MINOR_VERSION)
 		goto out;
 
-	status = nfserr_resource;
-	cstate = cstate_alloc();
-	if (cstate == NULL)
-		goto out;
-
 	status = nfs_ok;
 	while (!status && resp->opcnt < args->opcnt) {
 		op = &args->ops[resp->opcnt++];
@@ -964,7 +940,9 @@ encode_op:
 		status = nfserr_jukebox;
 	}
 
-	cstate_free(cstate);
+	fh_put(&resp->cstate.current_fh);
+	fh_put(&resp->cstate.save_fh);
+	BUG_ON(resp->cstate.replay_owner);
 out:
 	nfsd4_release_compoundargs(args);
 	/* Reset deferral mechanism for RPC deferrals */
include/linux/nfsd/xdr4.h +5 -4
index 27bd3e3..fd15ddc 100644
@@ -45,9 +45,9 @@
 #define XDR_LEN(n)                     (((n) + 3) & ~3)
 
 struct nfsd4_compound_state {
-	struct svc_fh current_fh;
-	struct svc_fh save_fh;
-	struct nfs4_stateowner *replay_owner;
+	struct svc_fh		current_fh;
+	struct svc_fh		save_fh;
+	struct nfs4_stateowner	*replay_owner;
 };
 
 struct nfsd4_change_info {
@@ -416,7 +416,8 @@ struct nfsd4_compoundres {
 	u32				taglen;
 	char *				tag;
 	u32				opcnt;
-	__be32 *			tagp; /* where to encode tag and  opcount */
+	__be32 *			tagp; /* tag, opcount encode location */
+	struct nfsd4_compound_state	cstate;
 };
 
 #define NFS4_SVC_XDRSIZE		sizeof(struct nfsd4_compoundargs)
Signature · 1 line
1.6.2.1
#16 lore raw

PATCH 4/47 nfsd: add a struct nfsd4_slot pointer to struct nfsd4_compound_state

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

To be used for sessions DRC replay.

---
 include/linux/nfsd/xdr4.h |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)
include/linux/nfsd/xdr4.h +2 -0
index fd15ddc..e309d55 100644
@@ -48,6 +48,8 @@ struct nfsd4_compound_state {
 	struct svc_fh		current_fh;
 	struct svc_fh		save_fh;
 	struct nfs4_stateowner	*replay_owner;
+	/* For sessions DRC */
+	struct nfsd4_slot	*slot;
 };
 
 struct nfsd4_change_info {
Signature · 1 line
1.6.2.1
#17 lore raw

PATCH 5/47 nfs41: common protocol definitions

In reply to Benny Halevy

Define all NFSv4.1 common operation and error code constants.

Note that some of the definitions are used by both the nfs41 client and the server code. This patch is duplicated in the nfs41 and nfsd41 sessions patchset.

[nfs41: add exchange id flags]
[removed server-only hunk changing NFSERR_REPLAY_ME]
[nfs41: add SEQ4_XX to nfs41-common-protocol]
[nfs41: generic error code update]
---
 include/linux/nfs4.h |  128 +++++++++++++++++++++++++++++++++++++++++++++++++-
 1 files changed, 127 insertions(+), 1 deletions(-)
include/linux/nfs4.h +127 -1
index b912311..a34683e 100644
@@ -21,6 +21,7 @@
 #define NFS4_FHSIZE		128
 #define NFS4_MAXPATHLEN		PATH_MAX
 #define NFS4_MAXNAMLEN		NAME_MAX
+#define NFS4_MAX_SESSIONID_LEN	16
 
 #define NFS4_ACCESS_READ        0x0001
 #define NFS4_ACCESS_LOOKUP      0x0002
@@ -38,6 +39,7 @@
 #define NFS4_OPEN_RESULT_CONFIRM 0x0002
 #define NFS4_OPEN_RESULT_LOCKTYPE_POSIX 0x0004
 
+#define NFS4_SHARE_ACCESS_MASK	0x000F
 #define NFS4_SHARE_ACCESS_READ	0x0001
 #define NFS4_SHARE_ACCESS_WRITE	0x0002
 #define NFS4_SHARE_ACCESS_BOTH	0x0003
@@ -45,6 +47,19 @@
 #define NFS4_SHARE_DENY_WRITE	0x0002
 #define NFS4_SHARE_DENY_BOTH	0x0003
 
+/* nfs41 */
+#define NFS4_SHARE_WANT_MASK		0xFF00
+#define NFS4_SHARE_WANT_NO_PREFERENCE	0x0000
+#define NFS4_SHARE_WANT_READ_DELEG	0x0100
+#define NFS4_SHARE_WANT_WRITE_DELEG	0x0200
+#define NFS4_SHARE_WANT_ANY_DELEG	0x0300
+#define NFS4_SHARE_WANT_NO_DELEG	0x0400
+#define NFS4_SHARE_WANT_CANCEL		0x0500
+
+#define NFS4_SHARE_WHEN_MASK		0xF0000
+#define NFS4_SHARE_SIGNAL_DELEG_WHEN_RESRC_AVAIL	0x10000
+#define NFS4_SHARE_PUSH_DELEG_WHEN_UNCONTENDED		0x20000
+
 #define NFS4_SET_TO_SERVER_TIME	0
 #define NFS4_SET_TO_CLIENT_TIME	1
 
@@ -88,6 +103,31 @@
 #define NFS4_ACE_GENERIC_EXECUTE              0x001200A0
 #define NFS4_ACE_MASK_ALL                     0x001F01FF
 
+#define EXCHGID4_FLAG_SUPP_MOVED_REFER		0x00000001
+#define EXCHGID4_FLAG_SUPP_MOVED_MIGR		0x00000002
+#define EXCHGID4_FLAG_USE_NON_PNFS		0x00010000
+#define EXCHGID4_FLAG_USE_PNFS_MDS		0x00020000
+#define EXCHGID4_FLAG_USE_PNFS_DS		0x00040000
+#define EXCHGID4_FLAG_UPD_CONFIRMED_REC_A	0x40000000
+#define EXCHGID4_FLAG_CONFIRMED_R		0x80000000
+/*
+ * Since the validity of these bits depends on whether
+ * they're set in the argument or response, have separate
+ * invalid flag masks for arg (_A) and resp (_R).
+ */
+#define EXCHGID4_INVAL_FLAG_MASK_A		0xBFF8FFFC
+#define EXCHGID4_INVAL_FLAG_MASK_R		0x7FF8FFFC
+
+#define SEQ4_STATUS_CB_PATH_DOWN		0x00000001
+#define SEQ4_STATUS_CB_GSS_CONTEXTS_EXPIRING	0x00000002
+#define SEQ4_STATUS_CB_GSS_CONTEXTS_EXPIRED	0x00000004
+#define SEQ4_STATUS_EXPIRED_ALL_STATE_REVOKED	0x00000008
+#define SEQ4_STATUS_EXPIRED_SOME_STATE_REVOKED	0x00000010
+#define SEQ4_STATUS_ADMIN_STATE_REVOKED		0x00000020
+#define SEQ4_STATUS_RECALLABLE_STATE_REVOKED	0x00000040
+#define SEQ4_STATUS_LEASE_MOVED			0x00000080
+#define SEQ4_STATUS_RESTART_RECLAIM_NEEDED	0x00000100
+
 #define NFS4_MAX_UINT64	(~(u64)0)
 
 enum nfs4_acl_whotype {
@@ -154,6 +194,28 @@ enum nfs_opnum4 {
 	OP_VERIFY = 37,
 	OP_WRITE = 38,
 	OP_RELEASE_LOCKOWNER = 39,
+
+	/* nfs41 */
+	OP_BACKCHANNEL_CTL = 40,
+	OP_BIND_CONN_TO_SESSION = 41,
+	OP_EXCHANGE_ID = 42,
+	OP_CREATE_SESSION = 43,
+	OP_DESTROY_SESSION = 44,
+	OP_FREE_STATEID = 45,
+	OP_GET_DIR_DELEGATION = 46,
+	OP_GETDEVICEINFO = 47,
+	OP_GETDEVICELIST = 48,
+	OP_LAYOUTCOMMIT = 49,
+	OP_LAYOUTGET = 50,
+	OP_LAYOUTRETURN = 51,
+	OP_SECINFO_NO_NAME = 52,
+	OP_SEQUENCE = 53,
+	OP_SET_SSV = 54,
+	OP_TEST_STATEID = 55,
+	OP_WANT_DELEGATION = 56,
+	OP_DESTROY_CLIENTID = 57,
+	OP_RECLAIM_COMPLETE = 58,
+
 	OP_ILLEGAL = 10044,
 };
 
@@ -230,7 +292,48 @@ enum nfsstat4 {
 	NFS4ERR_DEADLOCK = 10045,
 	NFS4ERR_FILE_OPEN = 10046,
 	NFS4ERR_ADMIN_REVOKED = 10047,
-	NFS4ERR_CB_PATH_DOWN = 10048
+	NFS4ERR_CB_PATH_DOWN = 10048,
+
+	/* nfs41 */
+	NFS4ERR_BADIOMODE	= 10049,
+	NFS4ERR_BADLAYOUT	= 10050,
+	NFS4ERR_BAD_SESSION_DIGEST = 10051,
+	NFS4ERR_BADSESSION	= 10052,
+	NFS4ERR_BADSLOT		= 10053,
+	NFS4ERR_COMPLETE_ALREADY = 10054,
+	NFS4ERR_CONN_NOT_BOUND_TO_SESSION = 10055,
+	NFS4ERR_DELEG_ALREADY_WANTED = 10056,
+	NFS4ERR_BACK_CHAN_BUSY	= 10057,	/* backchan reqs outstanding */
+	NFS4ERR_LAYOUTTRYLATER	= 10058,
+	NFS4ERR_LAYOUTUNAVAILABLE = 10059,
+	NFS4ERR_NOMATCHING_LAYOUT = 10060,
+	NFS4ERR_RECALLCONFLICT	= 10061,
+	NFS4ERR_UNKNOWN_LAYOUTTYPE = 10062,
+	NFS4ERR_SEQ_MISORDERED = 10063, 	/* unexpected seq.id in req */
+	NFS4ERR_SEQUENCE_POS	= 10064,	/* [CB_]SEQ. op not 1st op */
+	NFS4ERR_REQ_TOO_BIG	= 10065,	/* request too big */
+	NFS4ERR_REP_TOO_BIG	= 10066,	/* reply too big */
+	NFS4ERR_REP_TOO_BIG_TO_CACHE = 10067,	/* rep. not all cached */
+	NFS4ERR_RETRY_UNCACHED_REP = 10068,	/* retry & rep. uncached */
+	NFS4ERR_UNSAFE_COMPOUND = 10069,	/* retry/recovery too hard */
+	NFS4ERR_TOO_MANY_OPS	= 10070,	/* too many ops in [CB_]COMP */
+	NFS4ERR_OP_NOT_IN_SESSION = 10071,	/* op needs [CB_]SEQ. op */
+	NFS4ERR_HASH_ALG_UNSUPP = 10072,	/* hash alg. not supp. */
+						/* Error 10073 is unused. */
+	NFS4ERR_CLIENTID_BUSY	= 10074,	/* clientid has state */
+	NFS4ERR_PNFS_IO_HOLE	= 10075,	/* IO to _SPARSE file hole */
+	NFS4ERR_SEQ_FALSE_RETRY	= 10076,	/* retry not origional */
+	NFS4ERR_BAD_HIGH_SLOT	= 10077,	/* sequence arg bad */
+	NFS4ERR_DEADSESSION	= 10078,	/* persistent session dead */
+	NFS4ERR_ENCR_ALG_UNSUPP = 10079,	/* SSV alg mismatch */
+	NFS4ERR_PNFS_NO_LAYOUT	= 10080,	/* direct I/O with no layout */
+	NFS4ERR_NOT_ONLY_OP	= 10081,	/* bad compound */
+	NFS4ERR_WRONG_CRED	= 10082,	/* permissions:state change */
+	NFS4ERR_WRONG_TYPE	= 10083,	/* current operation mismatch */
+	NFS4ERR_DIRDELEG_UNAVAIL = 10084,	/* no directory delegation */
+	NFS4ERR_REJECT_DELEG	= 10085,	/* on callback */
+	NFS4ERR_RETURNCONFLICT	= 10086,	/* outstanding layoutreturn */
+	NFS4ERR_DELEG_REVOKED	= 10087,	/* deleg./layout revoked */
 };
 
 /*
@@ -391,6 +494,29 @@ enum {
 	NFSPROC4_CLNT_GETACL,
 	NFSPROC4_CLNT_SETACL,
 	NFSPROC4_CLNT_FS_LOCATIONS,
+
+	/* nfs41 */
+	NFSPROC4_CLNT_EXCHANGE_ID,
+	NFSPROC4_CLNT_CREATE_SESSION,
+	NFSPROC4_CLNT_DESTROY_SESSION,
+	NFSPROC4_CLNT_SEQUENCE,
+	NFSPROC4_CLNT_GET_LEASE_TIME,
+};
+
+/* nfs41 types */
+struct nfs4_sessionid {
+	unsigned char data[NFS4_MAX_SESSIONID_LEN];
+};
+
+/* Create Session Flags */
+#define SESSION4_PERSIST	 0x001
+#define SESSION4_BACK_CHAN 	 0x002
+#define SESSION4_RDMA		 0x004
+
+enum state_protect_how4 {
+	SP4_NONE	= 0,
+	SP4_MACH_CRED	= 1,
+	SP4_SSV		= 2
 };
 
 #endif
Signature · 1 line
1.6.2.1
#18 lore raw

PATCH 6/47 nfsd41: change NFSERR_REPLAY_ME

In reply to Benny Halevy

so not to overlap nfs41 error codes.

---
 include/linux/nfs.h |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)
include/linux/nfs.h +2 -1
index 54af92c..1510a64 100644
@@ -109,7 +109,8 @@
 	NFSERR_FILE_OPEN = 10046,      /*       v4 */
 	NFSERR_ADMIN_REVOKED = 10047,  /*       v4 */
 	NFSERR_CB_PATH_DOWN = 10048,   /*       v4 */
-	NFSERR_REPLAY_ME = 10049	/*       v4 */
+
+	NFSERR_REPLAY_ME = 11001,	/* linux internal */
 };
 
 /* NFSv2 file types - beware, these are not the same in NFSv3 */
Signature · 1 line
1.6.2.1
#19 lore raw

PATCH 7/47 nfsd41: Add Kconfig symbols for NFSv4.1

In reply to Benny Halevy

Added CONFIG_NFSD_V4_1 and made it depend upon NFSD_V4 and EXPERIMENTAL Indicate that CONFIG_NFS_V4_1 is for NFS developers at the moment

At the moment we're expecting folks trying out nfs41 to actively participate in the development process by helping us debug issues and ideally send patches to fix problems.

[trailing space fixed]
---
 fs/nfsd/Kconfig |    9 +++++++++
 1 files changed, 9 insertions(+), 0 deletions(-)
fs/nfsd/Kconfig +9 -0
index 503b9da..1f2f893 100644
@@ -79,3 +79,12 @@ config NFSD_V4
 	  available from http://linux-nfs.org/.
 
 	  If unsure, say N.
+
+config NFSD_V4_1
+	bool "NFS server support for NFSv4.1 (DEVELOPER ONLY)"
+	depends on NFSD_V4 && EXPERIMENTAL
+	help
+	  This option enables support for minor version 1 of the NFSv4 protocol
+	  (draft-ietf-nfsv4-minorversion1) in the kernel's NFS server.
+
+	  Unless you're an NFS developer, say N.
Signature · 1 line
1.6.2.1
#20 lore raw

PATCH 8/47 nfsd41: define nfs41 error codes

In reply to Benny Halevy

From: Marc Eshel <[email protected]>

Define all error code present in
http://tools.ietf.org/html/draft-ietf-nfsv4-minorversion1-29.

[nfsd41: clean up error code definitions]
---
 include/linux/nfs.h       |    2 --
 include/linux/nfsd/nfsd.h |   43 ++++++++++++++++++++++++++++++++++++++++++-
 2 files changed, 42 insertions(+), 3 deletions(-)
include/linux/nfs.h +0 -2
index 1510a64..214d499 100644
@@ -109,8 +109,6 @@
 	NFSERR_FILE_OPEN = 10046,      /*       v4 */
 	NFSERR_ADMIN_REVOKED = 10047,  /*       v4 */
 	NFSERR_CB_PATH_DOWN = 10048,   /*       v4 */
-
-	NFSERR_REPLAY_ME = 11001,	/* linux internal */
 };
 
 /* NFSv2 file types - beware, these are not the same in NFSv3 */
include/linux/nfsd/nfsd.h +42 -1

55 lines not on this page · raw message

Signature · 1 line
1.6.2.1
#21 lore raw

PATCH 9/47 nfsd41: sessions basic data types

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

This patch provides basic data structures representing the nfs41 sessions and slots, plus helpers for keeping a reference count on the session and freeing it.

Note that our server only support a headerpadsz of 0 and it ignores backchannel attributes at the moment.

[nfsd41: remove headerpadsz from channel attributes]
[nfsd41: embed nfsd4_channel in nfsd4_session]
[nfsd41: use bool inuse for slot state]
---
 fs/nfsd/nfs4state.c        |   22 ++++++++++++++++++++++
 include/linux/nfsd/state.h |   34 ++++++++++++++++++++++++++++++++++
 2 files changed, 56 insertions(+), 0 deletions(-)
fs/nfsd/nfs4state.c +22 -0
index 070e9e5..f4b624b 100644
@@ -382,6 +382,28 @@ static void release_openowner(struct nfs4_stateowner *sop)
 	nfs4_put_stateowner(sop);
 }
 
+#if defined(CONFIG_NFSD_V4_1)
+static void
+release_session(struct nfsd4_session *ses)
+{
+	list_del(&ses->se_hash);
+	list_del(&ses->se_perclnt);
+	nfsd4_put_session(ses);
+}
+
+void
+free_session(struct kref *kref)
+{
+	struct nfsd4_session *ses;
+
+	ses = container_of(kref, struct nfsd4_session, se_ref);
+	kfree(ses->se_slots);
+	kfree(ses);
+}
+#else /* CONFIG_NFSD_V4_1 */
+static inline void release_session(struct nfsd4_session *ses) {}
+#endif /* CONFIG_NFSD_V4_1 */
+
 static inline void
 renew_client(struct nfs4_client *clp)
 {
include/linux/nfsd/state.h +34 -0
index a6e4a00..afae103 100644
@@ -99,6 +99,40 @@ struct nfs4_callback {
 	struct rpc_clnt *       cb_client;
 };
 
+struct nfsd4_slot {
+	bool				sl_inuse;
+	struct nfsd4_session		*sl_session;
+	u32				sl_seqid;
+};
+
+struct nfsd4_session {
+	struct kref		se_ref;
+	struct list_head	se_hash;	/* hash by sessionid */
+	struct list_head	se_perclnt;
+	u32			se_flags;
+	struct nfs4_client	*se_client;	/* for expire_client */
+	struct nfs4_sessionid	se_sessionid;
+	u32			se_fmaxreq_sz;
+	u32			se_fmaxresp_sz;
+	u32			se_fmaxresp_cached;
+	u32			se_fmaxops;
+	u32			se_fnumslots;
+	struct nfsd4_slot	*se_slots;	/* forward channel slots */
+};
+
+static inline void
+nfsd4_put_session(struct nfsd4_session *ses)
+{
+	extern void free_session(struct kref *kref);
+	kref_put(&ses->se_ref, free_session);
+}
+
+static inline void
+nfsd4_get_session(struct nfsd4_session *ses)
+{
+	kref_get(&ses->se_ref);
+}
+
 #define HEXDIR_LEN     33 /* hex version of 16 byte md5 of cl_name plus '\0' */
 
 /*
Signature · 1 line
1.6.2.1
#22 lore raw

PATCH 10/47 nfsd41: introduce nfs4_client cl_sessions list

In reply to Benny Halevy

From: Marc Eshel <[email protected]>

---
 fs/nfsd/nfs4state.c        |    3 +++
 include/linux/nfsd/state.h |    3 +++
 2 files changed, 6 insertions(+), 0 deletions(-)
fs/nfsd/nfs4state.c +3 -0
index f4b624b..b71ec72 100644
@@ -532,6 +532,9 @@ static struct nfs4_client *create_client(struct xdr_netobj name, char *recdir)
 	INIT_LIST_HEAD(&clp->cl_strhash);
 	INIT_LIST_HEAD(&clp->cl_openowners);
 	INIT_LIST_HEAD(&clp->cl_delegations);
+#if defined(CONFIG_NFSD_V4_1)
+	INIT_LIST_HEAD(&clp->cl_sessions);
+#endif /* CONFIG_NFSD_V4_1 */
 	INIT_LIST_HEAD(&clp->cl_lru);
 	return clp;
 }
include/linux/nfsd/state.h +3 -0
index afae103..29624b4 100644
@@ -164,6 +164,9 @@ struct nfs4_client {
 	struct nfs4_callback	cl_callback;    /* callback info */
 	atomic_t		cl_count;	/* ref count */
 	u32			cl_firststate;	/* recovery dir creation */
+#ifdef CONFIG_NFSD_V4_1
+	struct list_head	cl_sessions;
+#endif /* CONFIG_NFSD_V4_1 */
 };
 
 /* struct nfs4_client_reset
Signature · 1 line
1.6.2.1
#23 lore raw

PATCH 11/47 nfsd41: release_session when client is expired

In reply to Benny Halevy

From: Marc Eshel <[email protected]>

[add CONFIG_NFSD_V4_1 to fix v4.0 regression bug]
---
 fs/nfsd/nfs4state.c |    8 ++++++++
 1 files changed, 8 insertions(+), 0 deletions(-)
fs/nfsd/nfs4state.c +8 -0
index b71ec72..df9d42e 100644
@@ -515,6 +515,14 @@ expire_client(struct nfs4_client *clp)
 		sop = list_entry(clp->cl_openowners.next, struct nfs4_stateowner, so_perclient);
 		release_openowner(sop);
 	}
+#ifdef CONFIG_NFSD_V4_1
+	while (!list_empty(&clp->cl_sessions)) {
+		struct nfsd4_session  *ses;
+		ses = list_entry(clp->cl_sessions.next, struct nfsd4_session,
+				 se_perclnt);
+		release_session(ses);
+	}
+#endif /* CONFIG_NFSD_V4_1 */
 	put_nfs4_client(clp);
 }
 
Signature · 1 line
1.6.2.1
#24 lore raw

PATCH 12/47 nfsd41: sessionid hashing

In reply to Benny Halevy

From: Marc Eshel <[email protected]>

Simple sessionid hashing using its monotonically increasing sequence number.

Locking considerations:
sessionid_hashtbl access is controlled by the sessionid_lock spin lock. It must be taken for insert, delete, and lookup. nfsd4_sequence looks up the session id and if the session is found, it calls nfsd4_get_session (still under the sessionid_lock). nfsd4_destroy_session calls nfsd4_put_session after unhashing it, so when the session's kref reaches zero it's going to get freed.

[we don't use a prime for sessionid hash table size]
[use sessionid_lock spin lock]
---
 fs/nfsd/nfs4state.c        |   57 +++++++++++++++++++++++++++++++++++++++++++-
 include/linux/nfsd/state.h |    7 +++++
 2 files changed, 63 insertions(+), 1 deletions(-)
fs/nfsd/nfs4state.c +56 -1
index df9d42e..ac4e8f2 100644
@@ -383,11 +383,62 @@ static void release_openowner(struct nfs4_stateowner *sop)
 }
 
 #if defined(CONFIG_NFSD_V4_1)
+static DEFINE_SPINLOCK(sessionid_lock);
+#define SESSION_HASH_SIZE	512
+static struct list_head sessionid_hashtbl[SESSION_HASH_SIZE];
+
+static inline int
+hash_sessionid(struct nfs4_sessionid *sessionid)
+{
+	struct nfsd4_sessionid *sid = (struct nfsd4_sessionid *)sessionid;
+
+	return sid->sequence % SESSION_HASH_SIZE;
+}
+
+static inline void
+dump_sessionid(const char *fn, struct nfs4_sessionid *sessionid)
+{
+	u32 *ptr = (u32 *)(&sessionid->data[0]);
+	dprintk("%s: %u:%u:%u:%u\n", fn, ptr[0], ptr[1], ptr[2], ptr[3]);
+}
+
+/* caller must hold sessionid_lock */
+static struct nfsd4_session *
+find_in_sessionid_hashtbl(struct nfs4_sessionid *sessionid)
+{
+	struct nfsd4_session *elem;
+	int idx;
+
+	dump_sessionid(__func__, sessionid);
+	idx = hash_sessionid(sessionid);
+	dprintk("%s: idx is %d\n", __func__, idx);
+	/* Search in the appropriate list */
+	list_for_each_entry(elem, &sessionid_hashtbl[idx], se_hash) {
+		dump_sessionid("list traversal", &elem->se_sessionid);
+		if (!memcmp(elem->se_sessionid.data, sessionid->data,
+			    NFS4_MAX_SESSIONID_LEN)) {
+			return elem;
+		}
+	}
+
+	dprintk("%s: session not found\n", __func__);
+	return NULL;
+}
+
+/* caller must hold sessionid_lock */
 static void
-release_session(struct nfsd4_session *ses)
+unhash_session(struct nfsd4_session *ses)
 {
 	list_del(&ses->se_hash);
 	list_del(&ses->se_perclnt);
+}
+
+static void
+release_session(struct nfsd4_session *ses)
+{
+	spin_lock(&sessionid_lock);
+	unhash_session(ses);
+	spin_unlock(&sessionid_lock);
 	nfsd4_put_session(ses);
 }
 
@@ -3213,6 +3264,10 @@ nfs4_state_init(void)
 		INIT_LIST_HEAD(&unconf_str_hashtbl[i]);
 		INIT_LIST_HEAD(&unconf_id_hashtbl[i]);
 	}
+#if defined(CONFIG_NFSD_V4_1)
+	for (i = 0; i < SESSION_HASH_SIZE; i++)
+		INIT_LIST_HEAD(&sessionid_hashtbl[i]);
+#endif /* CONFIG_NFSD_V4_1 */
 	for (i = 0; i < FILE_HASH_SIZE; i++) {
 		INIT_LIST_HEAD(&file_hashtbl[i]);
 	}
include/linux/nfsd/state.h +7 -0
index 29624b4..7592d7b 100644
@@ -133,6 +133,13 @@ nfsd4_get_session(struct nfsd4_session *ses)
 	kref_get(&ses->se_ref);
 }
 
+/* formatted contents of nfs4_sessionid */
+struct nfsd4_sessionid {
+	clientid_t	clientid;
+	u32		sequence;
+	u32		reserved;
+};
+
 #define HEXDIR_LEN     33 /* hex version of 16 byte md5 of cl_name plus '\0' */
 
 /*
Signature · 1 line
1.6.2.1
#25 lore raw

PATCH 13/47 nfsd41: xdr infrastructure

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

Define nfsd41_dec_ops vector and add it to nfsd4_minorversion for minorversion 1.

Note: nfsd4_enc_ops vector is shared for v4.0 and v4.1 since we don't need to filter out obsolete ops as this is done in the decoding phase.

exchange_id, create_session, destroy_session, and sequence ops are implemented as stubs returning nfserr_opnotsupp at this stage.

[was nfsd41: xdr stubs]
---
 fs/nfsd/nfs4xdr.c         |  156 +++++++++++++++++++++++++++++++++++++++++++++
 include/linux/nfsd/xdr4.h |   24 +++++++
 2 files changed, 180 insertions(+), 0 deletions(-)
fs/nfsd/nfs4xdr.c +156 -0
index 76a0b2a..b082d07 100644
@@ -996,6 +996,36 @@ nfsd4_decode_release_lockowner(struct nfsd4_compoundargs *argp, struct nfsd4_rel
 	DECODE_TAIL;
 }
 
+#if defined(CONFIG_NFSD_V4_1)
+static __be32
+nfsd4_decode_exchange_id(struct nfsd4_compoundargs *argp,
+			 struct nfsd4_exchange_id *clid)
+{
+	return nfserr_opnotsupp;	/* stub */
+}
+
+static __be32
+nfsd4_decode_create_session(struct nfsd4_compoundargs *argp,
+			    struct nfsd4_create_session *sess)
+{
+	return nfserr_opnotsupp;	/* stub */
+}
+
+static __be32
+nfsd4_decode_destroy_session(struct nfsd4_compoundargs *argp,
+			     struct nfsd4_destroy_session *destroy_session)
+{
+	return nfserr_opnotsupp;	/* stub */
+}
+
+static __be32
+nfsd4_decode_sequence(struct nfsd4_compoundargs *argp,
+		      struct nfsd4_sequence *seq)
+{
+	return nfserr_opnotsupp;	/* stub */
+}
+#endif /* CONFIG_NFSD_V4_1 */
+
 static __be32
 nfsd4_decode_noop(struct nfsd4_compoundargs *argp, void *p)
 {
@@ -1050,6 +1080,69 @@ static nfsd4_dec nfsd4_dec_ops[] = {
 	[OP_RELEASE_LOCKOWNER]	= (nfsd4_dec)nfsd4_decode_release_lockowner,
 };
 
+#if defined(CONFIG_NFSD_V4_1)
+static nfsd4_dec nfsd41_dec_ops[] = {
+	[OP_ACCESS]		(nfsd4_dec)nfsd4_decode_access,
+	[OP_CLOSE]		(nfsd4_dec)nfsd4_decode_close,
+	[OP_COMMIT]		(nfsd4_dec)nfsd4_decode_commit,
+	[OP_CREATE]		(nfsd4_dec)nfsd4_decode_create,
+	[OP_DELEGPURGE]		(nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_DELEGRETURN]	(nfsd4_dec)nfsd4_decode_delegreturn,
+	[OP_GETATTR]		(nfsd4_dec)nfsd4_decode_getattr,
+	[OP_GETFH]		(nfsd4_dec)nfsd4_decode_noop,
+	[OP_LINK]		(nfsd4_dec)nfsd4_decode_link,
+	[OP_LOCK]		(nfsd4_dec)nfsd4_decode_lock,
+	[OP_LOCKT]		(nfsd4_dec)nfsd4_decode_lockt,
+	[OP_LOCKU]		(nfsd4_dec)nfsd4_decode_locku,
+	[OP_LOOKUP]		(nfsd4_dec)nfsd4_decode_lookup,
+	[OP_LOOKUPP]		(nfsd4_dec)nfsd4_decode_noop,
+	[OP_NVERIFY]		(nfsd4_dec)nfsd4_decode_verify,
+	[OP_OPEN]		(nfsd4_dec)nfsd4_decode_open,
+	[OP_OPENATTR]		(nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_OPEN_CONFIRM]	(nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_OPEN_DOWNGRADE]	(nfsd4_dec)nfsd4_decode_open_downgrade,
+	[OP_PUTFH]		(nfsd4_dec)nfsd4_decode_putfh,
+	[OP_PUTPUBFH]		(nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_PUTROOTFH]		(nfsd4_dec)nfsd4_decode_noop,
+	[OP_READ]		(nfsd4_dec)nfsd4_decode_read,
+	[OP_READDIR]		(nfsd4_dec)nfsd4_decode_readdir,
+	[OP_READLINK]		(nfsd4_dec)nfsd4_decode_noop,
+	[OP_REMOVE]		(nfsd4_dec)nfsd4_decode_remove,
+	[OP_RENAME]		(nfsd4_dec)nfsd4_decode_rename,
+	[OP_RENEW]		(nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_RESTOREFH]		(nfsd4_dec)nfsd4_decode_noop,
+	[OP_SAVEFH]		(nfsd4_dec)nfsd4_decode_noop,
+	[OP_SECINFO]		(nfsd4_dec)nfsd4_decode_secinfo,
+	[OP_SETATTR]		(nfsd4_dec)nfsd4_decode_setattr,
+	[OP_SETCLIENTID]	(nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_SETCLIENTID_CONFIRM](nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_VERIFY]		(nfsd4_dec)nfsd4_decode_verify,
+	[OP_WRITE]		(nfsd4_dec)nfsd4_decode_write,
+	[OP_RELEASE_LOCKOWNER]	(nfsd4_dec)nfsd4_decode_notsupp,
+
+	/* new operations for NFSv4.1 */
+	[OP_BACKCHANNEL_CTL]	(nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_BIND_CONN_TO_SESSION](nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_EXCHANGE_ID]	(nfsd4_dec)nfsd4_decode_exchange_id,
+	[OP_CREATE_SESSION]	(nfsd4_dec)nfsd4_decode_create_session,
+	[OP_DESTROY_SESSION]	(nfsd4_dec)nfsd4_decode_destroy_session,
+	[OP_FREE_STATEID]	(nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_GET_DIR_DELEGATION]	(nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_GETDEVICEINFO]	(nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_GETDEVICELIST]	(nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_LAYOUTCOMMIT]	(nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_LAYOUTGET]		(nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_LAYOUTRETURN]	(nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_SECINFO_NO_NAME]	(nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_SEQUENCE]		(nfsd4_dec)nfsd4_decode_sequence,
+	[OP_SET_SSV]		(nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_TEST_STATEID]	(nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_WANT_DELEGATION]	(nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_DESTROY_CLIENTID]	(nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_RECLAIM_COMPLETE]	(nfsd4_dec)nfsd4_decode_notsupp,
+};
+#endif /* CONFIG_NFSD_V4_1 */
+
 struct nfsd4_minorversion_ops {
 	nfsd4_dec *decoders;
 	int nops;
@@ -1057,6 +1150,9 @@ struct nfsd4_minorversion_ops {
 
 static struct nfsd4_minorversion_ops nfsd4_minorversion[] = {
 	[0] = { nfsd4_dec_ops, ARRAY_SIZE(nfsd4_dec_ops) },
+#if defined(CONFIG_NFSD_V4_1)
+	[1] = { nfsd41_dec_ops, ARRAY_SIZE(nfsd41_dec_ops) },
+#endif /* CONFIG_NFSD_V4_1 */
 };
 
 static __be32
@@ -2571,6 +2667,40 @@ nfsd4_encode_write(struct nfsd4_compoundres *resp, __be32 nfserr, struct nfsd4_w
 	return nfserr;
 }
 
+#if defined(CONFIG_NFSD_V4_1)
+static __be32
+nfsd4_encode_exchange_id(struct nfsd4_compoundres *resp, int nfserr,
+			 struct nfsd4_exchange_id *exid)
+{
+	/* stub */
+	return nfserr;
+}
+
+static __be32
+nfsd4_encode_create_session(struct nfsd4_compoundres *resp, int nfserr,
+			    struct nfsd4_create_session *sess)
+{
+	/* stub */
+	return nfserr;
+}
+
+static __be32
+nfsd4_encode_destroy_session(struct nfsd4_compoundres *resp, int nfserr,
+			     struct nfsd4_destroy_session *destroy_session)
+{
+	/* stub */
+	return nfserr;
+}
+
+static __be32
+nfsd4_encode_sequence(struct nfsd4_compoundres *resp, int nfserr,
+		      struct nfsd4_sequence *seq)
+{
+	/* stub */
+	return nfserr;
+}
+#endif /* CONFIG_NFSD_V4_1 */
+
 static __be32
 nfsd4_encode_noop(struct nfsd4_compoundres *resp, __be32 nfserr, void *p)
 {
@@ -2579,6 +2709,11 @@ nfsd4_encode_noop(struct nfsd4_compoundres *resp, __be32 nfserr, void *p)
 
 typedef __be32(* nfsd4_enc)(struct nfsd4_compoundres *, __be32, void *);
 
+/*
+ * Note: nfsd4_enc_ops vector is shared for v4.0 and v4.1
+ * since we don't need to filter out obsolete ops as this is
+ * done in the decoding phase.
+ */
 static nfsd4_enc nfsd4_enc_ops[] = {
 	[OP_ACCESS]		= (nfsd4_enc)nfsd4_encode_access,
 	[OP_CLOSE]		= (nfsd4_enc)nfsd4_encode_close,
@@ -2617,6 +2752,27 @@ static nfsd4_enc nfsd4_enc_ops[] = {
 	[OP_VERIFY]		= (nfsd4_enc)nfsd4_encode_noop,
 	[OP_WRITE]		= (nfsd4_enc)nfsd4_encode_write,
 	[OP_RELEASE_LOCKOWNER]	= (nfsd4_enc)nfsd4_encode_noop,
+#if defined(CONFIG_NFSD_V4_1)
+	[OP_BACKCHANNEL_CTL]	= (nfsd4_enc)nfsd4_encode_noop,
+	[OP_BIND_CONN_TO_SESSION] = (nfsd4_enc)nfsd4_encode_noop,
+	[OP_EXCHANGE_ID]	= (nfsd4_enc)nfsd4_encode_exchange_id,
+	[OP_CREATE_SESSION]	= (nfsd4_enc)nfsd4_encode_create_session,
+	[OP_DESTROY_SESSION]	= (nfsd4_enc)nfsd4_encode_destroy_session,
+	[OP_FREE_STATEID]	= (nfsd4_enc)nfsd4_encode_noop,
+	[OP_GET_DIR_DELEGATION]	= (nfsd4_enc)nfsd4_encode_noop,
+	[OP_GETDEVICEINFO]	= (nfsd4_enc)nfsd4_encode_noop,
+	[OP_GETDEVICELIST]	= (nfsd4_enc)nfsd4_encode_noop,
+	[OP_LAYOUTCOMMIT]	= (nfsd4_enc)nfsd4_encode_noop,
+	[OP_LAYOUTGET]		= (nfsd4_enc)nfsd4_encode_noop,
+	[OP_LAYOUTRETURN]	= (nfsd4_enc)nfsd4_encode_noop,
+	[OP_SECINFO_NO_NAME]	= (nfsd4_enc)nfsd4_encode_noop,
+	[OP_SEQUENCE]		= (nfsd4_enc)nfsd4_encode_sequence,
+	[OP_SET_SSV]		= (nfsd4_enc)nfsd4_encode_noop,
+	[OP_TEST_STATEID]	= (nfsd4_enc)nfsd4_encode_noop,
+	[OP_WANT_DELEGATION]	= (nfsd4_enc)nfsd4_encode_noop,
+	[OP_DESTROY_CLIENTID]	= (nfsd4_enc)nfsd4_encode_noop,
+	[OP_RECLAIM_COMPLETE]	= (nfsd4_enc)nfsd4_encode_noop,
+#endif /* CONFIG_NFSD_V4_1 */
 };
 
 void
include/linux/nfsd/xdr4.h +24 -0
index e309d55..40b148e 100644
@@ -346,6 +346,24 @@ struct nfsd4_write {
 	nfs4_verifier	wr_verifier;        /* response */
 };
 
+#if defined(CONFIG_NFSD_V4_1)
+struct nfsd4_exchange_id {
+	int	foo;	/* stub */
+};
+
+struct nfsd4_create_session {
+	int	foo;	/* stub */
+};
+
+struct nfsd4_sequence {
+	int	foo;	/* stub */
+};
+
+struct nfsd4_destroy_session {
+	int	foo;	/* stub */
+};
+#endif /* CONFIG_NFSD_V4_1 */
+
 struct nfsd4_op {
 	int					opnum;
 	__be32					status;
@@ -380,6 +398,12 @@ struct nfsd4_op {
 		struct nfsd4_verify		verify;
 		struct nfsd4_write		write;
 		struct nfsd4_release_lockowner	release_lockowner;
+#if defined(CONFIG_NFSD_V4_1)
+		struct nfsd4_exchange_id	exchange_id;
+		struct nfsd4_create_session	create_session;
+		struct nfsd4_destroy_session	destroy_session;
+		struct nfsd4_sequence		sequence;
+#endif /* CONFIG_NFSD_V4_1 */
 	} u;
 	struct nfs4_replay *			replay;
 };
Signature · 1 line
1.6.2.1
#26 lore raw

PATCH 14/47 nfsd: remove nfsd4_ops array size

In reply to Benny Halevy

There's no need for it.

---
 fs/nfsd/nfs4proc.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)
fs/nfsd/nfs4proc.c +1 -1
index c8e1ad0..ded469f 100644
@@ -951,7 +951,7 @@ out:
 	return status;
 }
 
-static struct nfsd4_operation nfsd4_ops[OP_RELEASE_LOCKOWNER+1] = {
+static struct nfsd4_operation nfsd4_ops[] = {
 	[OP_ACCESS] = {
 		.op_func = (nfsd4op_func)nfsd4_access,
 		.op_name = "OP_ACCESS",
Signature · 1 line
1.6.2.1
#27 lore raw

PATCH 15/47 nfsd41: proc stubs

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

---
 fs/nfsd/nfs4proc.c        |   22 ++++++++++++++++++++++
 fs/nfsd/nfs4state.c       |   34 ++++++++++++++++++++++++++++++++++
 include/linux/nfsd/xdr4.h |   14 ++++++++++++++
 3 files changed, 70 insertions(+), 0 deletions(-)
fs/nfsd/nfs4proc.c +22 -0
index ded469f..f618e8e 100644
@@ -1101,6 +1101,28 @@ static struct nfsd4_operation nfsd4_ops[] = {
 		.op_flags = ALLOWED_WITHOUT_FH | ALLOWED_ON_ABSENT_FS,
 		.op_name = "OP_RELEASE_LOCKOWNER",
 	},
+#if defined(CONFIG_NFSD_V4_1)
+	[OP_EXCHANGE_ID] = {
+		.op_func = (nfsd4op_func)nfsd4_exchange_id,
+		.op_flags = ALLOWED_WITHOUT_FH,
+		.op_name = "OP_EXCHANGE_ID",
+	},
+	[OP_CREATE_SESSION] = {
+		.op_func = (nfsd4op_func)nfsd4_create_session,
+		.op_flags = ALLOWED_WITHOUT_FH,
+		.op_name = "OP_CREATE_SESSION",
+	},
+	[OP_DESTROY_SESSION] = {
+		.op_func = (nfsd4op_func)nfsd4_destroy_session,
+		.op_flags = ALLOWED_WITHOUT_FH,
+		.op_name = "OP_DESTROY_SESSION",
+	},
+	[OP_SEQUENCE] = {
+		.op_func = (nfsd4op_func)nfsd4_sequence,
+		.op_flags = ALLOWED_WITHOUT_FH,
+		.op_name = "OP_SEQUENCE",
+	},
+#endif /* CONFIG_NFSD_V4_1 */
 };
 
 static const char *nfsd4_op_name(unsigned opnum)
fs/nfsd/nfs4state.c +34 -0
index ac4e8f2..bbb7455 100644
@@ -840,6 +840,40 @@ out_err:
 	return;
 }
 
+#if defined(CONFIG_NFSD_V4_1)
+__be32
+nfsd4_exchange_id(struct svc_rqst *rqstp,
+		  struct nfsd4_compound_state *cstate,
+		  struct nfsd4_exchange_id *exid)
+{
+	return -1;	/* stub */
+}
+
+__be32
+nfsd4_create_session(struct svc_rqst *rqstp,
+		     struct nfsd4_compound_state *cstate,
+		     struct nfsd4_create_session *cr_ses)
+{
+	return -1;	/* stub */
+}
+
+__be32
+nfsd4_destroy_session(struct svc_rqst *r,
+		      struct nfsd4_compound_state *cstate,
+		      struct nfsd4_destroy_session *sessionid)
+{
+	return -1;	/* stub */
+}
+
+__be32
+nfsd4_sequence(struct svc_rqst *r,
+	       struct nfsd4_compound_state *cstate,
+	       struct nfsd4_sequence *seq)
+{
+	return -1;	/* stub */
+}
+#endif /* CONFIG_NFSD_V4_1 */
+
 __be32
 nfsd4_setclientid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 		  struct nfsd4_setclientid *setclid)
include/linux/nfsd/xdr4.h +14 -0
index 40b148e..0148d54 100644
@@ -475,6 +475,20 @@ extern __be32 nfsd4_setclientid(struct svc_rqst *rqstp,
 extern __be32 nfsd4_setclientid_confirm(struct svc_rqst *rqstp,
 		struct nfsd4_compound_state *,
 		struct nfsd4_setclientid_confirm *setclientid_confirm);
+#if defined(CONFIG_NFSD_V4_1)
+extern __be32 nfsd4_exchange_id(struct svc_rqst *rqstp,
+		struct nfsd4_compound_state *,
+struct nfsd4_exchange_id *);
+		extern __be32 nfsd4_create_session(struct svc_rqst *,
+		struct nfsd4_compound_state *,
+		struct nfsd4_create_session *);
+extern __be32 nfsd4_sequence(struct svc_rqst *,
+		struct nfsd4_compound_state *,
+		struct nfsd4_sequence *);
+extern __be32 nfsd4_destroy_session(struct svc_rqst *,
+		struct nfsd4_compound_state *,
+		struct nfsd4_destroy_session *);
+#endif /* CONFIG_NFSD_V4_1 */
 extern __be32 nfsd4_process_open1(struct nfsd4_open *open);
 extern __be32 nfsd4_process_open2(struct svc_rqst *rqstp,
 		struct svc_fh *current_fh, struct nfsd4_open *open);
Signature · 1 line
1.6.2.1
#28 lore raw

PATCH 16/47 nfsd41: exchange_id operation

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

Implement the exchange_id operation confoming to
http://tools.ietf.org/html/draft-ietf-nfsv4-minorversion1-28

Based on the client provided name, hash a client id. If a confirmed one is found, compare the op's creds and verifier. If the creds match and the verifier is different then expire the old client (client re-incarnated), otherwise, if both match, assume it's a replay and ignore it.

If an unconfirmed client is found, then copy the new creds and verifer if need update, otherwise assume replay.

The client is moved to a confirmed state on create_session.

In the nfs41 branch set the exchange_id flags to EXCHGID4_FLAG_USE_NON_PNFS | EXCHGID4_FLAG_SUPP_MOVED_REFER (pNFS is not supported, Referrals are supported, Migration is not.).

Address various scenarios from section 18.35 of the spec:

1. Check for EXCHGID4_FLAG_UPD_CONFIRMED_REC_A and set EXCHGID4_FLAG_CONFIRMED_R as appropriate.

2. Return error codes per 18.35.4 scenarios.

3. Update client records or generate new client ids depending on scenario.

Note: 18.35.4 case 3 probably still needs revisiting. The handling seems not quite right.

[nfsd41: use utsname for major_id (and copy to server_scope)]
[nfsd41: fix handling of various exchange id scenarios]
---
 fs/nfsd/nfs4state.c        |  138 +++++++++++++++++++++++++++++++++++++++++-
 fs/nfsd/nfs4xdr.c          |  146 +++++++++++++++++++++++++++++++++++++++++++-
 include/linux/nfsd/state.h |    2 +
 include/linux/nfsd/xdr4.h  |    8 ++-
 4 files changed, 289 insertions(+), 5 deletions(-)
fs/nfsd/nfs4state.c +137 -1
index bbb7455..09c63ff 100644
@@ -841,12 +841,148 @@ out_err:
 }
 
 #if defined(CONFIG_NFSD_V4_1)
+/*
+ * Set the exchange_id flags returned by the server.
+ */
+static void
+nfsd4_set_ex_flags(struct nfs4_client *new, struct nfsd4_exchange_id *clid)
+{
+	/* pNFS is not supported */
+	new->cl_exchange_flags |= EXCHGID4_FLAG_USE_NON_PNFS;
+
+	/* Referrals are supported, Migration is not. */
+	new->cl_exchange_flags |= EXCHGID4_FLAG_SUPP_MOVED_REFER;
+
+	/* set the wire flags to return to client. */
+	clid->flags = new->cl_exchange_flags;
+}
+
 __be32
 nfsd4_exchange_id(struct svc_rqst *rqstp,
 		  struct nfsd4_compound_state *cstate,
 		  struct nfsd4_exchange_id *exid)
 {
-	return -1;	/* stub */
+	struct nfs4_client *unconf, *conf, *new;
+	int status;
+	unsigned int		strhashval;
+	char			dname[HEXDIR_LEN];
+	nfs4_verifier		verf = exid->verifier;
+	u32			ip_addr = svc_addr_in(rqstp)->sin_addr.s_addr;
+	struct xdr_netobj clname = {
+		.len = exid->id_len,
+		.data = exid->id,
+	};
+
+	dprintk("%s rqstp=%p exid=%p clname.len=%u clname.data=%p "
+		" ip_addr=%u flags %x, spa_how %d\n",
+		__func__, rqstp, exid, clname.len, clname.data,
+		ip_addr, exid->flags, exid->spa_how);
+
+	if (!check_name(clname) || (exid->flags & EXCHGID4_INVAL_FLAG_MASK_A))
+		return nfserr_inval;
+
+	/* Currently only support SP4_NONE */
+	if (exid->spa_how != SP4_NONE)
+		return nfserr_encr_alg_unsupp;
+
+	status = nfs4_make_rec_clidname(dname, &clname);
+
+	if (status)
+		goto error;
+
+	strhashval = clientstr_hashval(dname);
+
+	nfs4_lock_state();
+	status = nfs_ok;
+
+	conf = find_confirmed_client_by_str(dname, strhashval);
+	if (conf) {
+		if (!same_verf(&verf, &conf->cl_verifier)) {
+			/* 18.35.4 case 8 */
+			if (exid->flags & EXCHGID4_FLAG_UPD_CONFIRMED_REC_A) {
+				status = nfserr_not_same;
+				goto out;
+			}
+			/* Client reboot: destroy old state */
+			expire_client(conf);
+			goto out_new;
+		}
+		if (!same_creds(&conf->cl_cred, &rqstp->rq_cred)) {
+			/* 18.35.4 case 9 */
+			if (exid->flags & EXCHGID4_FLAG_UPD_CONFIRMED_REC_A) {
+				status = nfserr_perm;
+				goto out;
+			}
+			expire_client(conf);
+			goto out_new;
+		}
+		if (ip_addr != conf->cl_addr &&
+		    !(exid->flags & EXCHGID4_FLAG_UPD_CONFIRMED_REC_A)) {
+			/* Client collision. 18.35.4 case 3 */
+			status = nfserr_clid_inuse;
+			goto out;
+		}
+		/*
+		 * Set bit when the owner id and verifier map to an already
+		 * confirmed client id (18.35.3).
+		 */
+		exid->flags |= EXCHGID4_FLAG_CONFIRMED_R;
+
+		/*
+		 * Falling into 18.35.4 case 2, possible router replay.
+		 * Leave confirmed record intact and return same result.
+		 */
+		copy_verf(conf, &verf);
+		new = conf;
+		goto out_copy;
+	} else {
+		/* 18.35.4 case 7 */
+		if (exid->flags & EXCHGID4_FLAG_UPD_CONFIRMED_REC_A) {
+			status = nfserr_noent;
+			goto out;
+		}
+	}
+
+	unconf  = find_unconfirmed_client_by_str(dname, strhashval);
+	if (unconf) {
+		/*
+		 * Possible retry or client restart.  Per 18.35.4 case 4,
+		 * a new unconfirmed record should be generated regardless
+		 * of whether any properties have changed.
+		 */
+		expire_client(unconf);
+	}
+
+out_new:
+	/* Normal case */
+	new = create_client(clname, dname);
+	if (new == NULL) {
+		status = nfserr_resource;
+		goto out;
+	}
+
+	copy_verf(new, &verf);
+	copy_cred(&new->cl_cred, &rqstp->rq_cred);
+	new->cl_addr = ip_addr;
+	gen_clid(new);
+	gen_confirm(new);
+	add_to_unconfirmed(new, strhashval);
+out_copy:
+	exid->clientid.cl_boot = new->cl_clientid.cl_boot;
+	exid->clientid.cl_id = new->cl_clientid.cl_id;
+
+	new->cl_seqid = exid->seqid = 1;
+	nfsd4_set_ex_flags(new, exid);
+
+	dprintk("nfsd4_exchange_id seqid %d flags %x\n",
+		new->cl_seqid, new->cl_exchange_flags);
+	status = nfs_ok;
+
+out:
+	nfs4_unlock_state();
+error:
+	dprintk("nfsd4_exchange_id returns %d\n", ntohl(status));
+	return status;
 }
 
 __be32
fs/nfsd/nfs4xdr.c +143 -3

165 lines not on this page · raw message

include/linux/nfsd/state.h +2 -0
index 7592d7b..5de36a7 100644
@@ -173,6 +173,8 @@ struct nfs4_client {
 	u32			cl_firststate;	/* recovery dir creation */
 #ifdef CONFIG_NFSD_V4_1
 	struct list_head	cl_sessions;
+	u32			cl_seqid;       /* seqid for create_session */
+	u32			cl_exchange_flags;
 #endif /* CONFIG_NFSD_V4_1 */
 };
 
include/linux/nfsd/xdr4.h +7 -1
index 0148d54..ea5a427 100644
@@ -348,7 +348,13 @@ struct nfsd4_write {
 
 #if defined(CONFIG_NFSD_V4_1)
 struct nfsd4_exchange_id {
-	int	foo;	/* stub */
+	nfs4_verifier	verifier;
+	u32		id_len;
+	char		*id;
+	u32		flags;
+	clientid_t	clientid;
+	u32		seqid;
+	int		spa_how;
 };
 
 struct nfsd4_create_session {
Signature · 1 line
1.6.2.1
#29 lore raw

PATCH 17/47 nfsd41: match clientid establishment method

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

We need to distinguish between client names provided by NFSv4.0 clients SETCLIENTID and those provided by NFSv4.1 via EXCHANGE_ID when looking up the clientid by string.

[nfsd41: use boolean values for use_exchange_id argument]
---
 fs/nfsd/nfs4recover.c      |    3 ++-
 fs/nfsd/nfs4state.c        |   41 ++++++++++++++++++++++++++++++-----------
 include/linux/nfsd/state.h |    2 +-
 3 files changed, 33 insertions(+), 13 deletions(-)
fs/nfsd/nfs4recover.c +2 -1
index b11cf8d..3444c00 100644
@@ -344,7 +344,8 @@ purge_old(struct dentry *parent, struct dentry *child)
 {
 	int status;
 
-	if (nfs4_has_reclaimed_state(child->d_name.name))
+	/* note: we currently use this path only for minorversion 0 */
+	if (nfs4_has_reclaimed_state(child->d_name.name, false))
 		return 0;
 
 	status = nfsd4_clear_clid_dir(parent, child);
fs/nfsd/nfs4state.c +30 -11

102 lines not on this page · raw message

include/linux/nfsd/state.h +1 -1
index 5de36a7..feab6ec 100644
@@ -331,7 +331,7 @@ extern void nfsd4_init_recdir(char *recdir_name);
 extern int nfsd4_recdir_load(void);
 extern void nfsd4_shutdown_recdir(void);
 extern int nfs4_client_to_reclaim(const char *name);
-extern int nfs4_has_reclaimed_state(const char *name);
+extern int nfs4_has_reclaimed_state(const char *name, bool use_exchange_id);
 extern void nfsd4_recdir_purge_old(void);
 extern int nfsd4_create_clid_dir(struct nfs4_client *clp);
 extern void nfsd4_remove_clid_dir(struct nfs4_client *clp);
Signature · 1 line
1.6.2.1
#30 lore raw

PATCH 18/47 nfsd41: sequence operation

In reply to Benny Halevy

Implement the sequence operation conforming to
http://tools.ietf.org/html/draft-ietf-nfsv4-minorversion1-26

Check for stale clientid (as derived from the sessionid). Enforce slotid range and exactly-once semantics using the slotid and seqid.

If everything went well renew the client lease and mark the slot INPROGRESS.

[nfsd41: rename sequence catchthis to cachethis]
[pulled some code to set cstate->slot from "nfsd DRC logic"]
[use sessionid_lock spin lock]
[nfsd41: use bool inuse for slot state]
---
 fs/nfsd/nfs4state.c       |   71 +++++++++++++++++++++++++++++++++++++++++++-
 fs/nfsd/nfs4xdr.c         |   32 +++++++++++++++++++-
 include/linux/nfsd/xdr4.h |   10 ++++++-
 3 files changed, 108 insertions(+), 5 deletions(-)
fs/nfsd/nfs4state.c +69 -2
index 0c39376..a19f292 100644
@@ -1004,6 +1004,32 @@ error:
 	return status;
 }
 
+static int
+check_slot_seqid(u32 seqid, struct nfsd4_slot *slot)
+{
+	dprintk("%s enter. seqid %d slot->sl_seqid %d\n", __func__, seqid,
+		slot->sl_seqid);
+
+	/* The slot is in use, and no response has been sent. */
+	if (slot->sl_inuse) {
+		if (seqid == slot->sl_seqid)
+			return nfserr_jukebox;
+		else
+			return nfserr_seq_misordered;
+	}
+	/* Normal */
+	if (likely(seqid == slot->sl_seqid + 1))
+		return nfs_ok;
+	/* Replay */
+	if (seqid == slot->sl_seqid)
+		return nfserr_replay_cache;
+	/* Wraparound */
+	if (seqid == 1 && (slot->sl_seqid + 1) == 0)
+		return nfs_ok;
+	/* Misordered replay or misordered new request */
+	return nfserr_seq_misordered;
+}
+
 __be32
 nfsd4_create_session(struct svc_rqst *rqstp,
 		     struct nfsd4_compound_state *cstate,
@@ -1021,11 +1047,52 @@ nfsd4_destroy_session(struct svc_rqst *r,
 }
 
 __be32
-nfsd4_sequence(struct svc_rqst *r,
+nfsd4_sequence(struct svc_rqst *rqstp,
 	       struct nfsd4_compound_state *cstate,
 	       struct nfsd4_sequence *seq)
 {
-	return -1;	/* stub */
+	struct nfsd4_session *session;
+	struct nfsd4_slot *slot;
+	int status;
+
+	spin_lock(&sessionid_lock);
+	status = nfserr_badsession;
+	session = find_in_sessionid_hashtbl(&seq->sessionid);
+	if (!session)
+		goto out;
+
+	status = nfserr_badslot;
+	if (seq->slotid >= session->se_fnumslots)
+		goto out;
+
+	slot = &session->se_slots[seq->slotid];
+	dprintk("%s: slotid %d\n", __func__, seq->slotid);
+
+	status = check_slot_seqid(seq->seqid, slot);
+	if (status == nfserr_replay_cache) {
+		cstate->slot = slot;
+		goto replay_cache;
+	}
+	if (status)
+		goto out;
+
+	/* Success! bump slot seqid */
+	slot->sl_inuse = true;
+	slot->sl_seqid = seq->seqid;
+
+	cstate->slot = slot;
+
+replay_cache:
+	/* Renew the clientid on success and on replay.
+	 * Hold a session reference until done processing the compound:
+	 * nfsd4_put_session called only if the cstate slot is set.
+	 */
+	renew_client(session->se_client);
+	nfsd4_get_session(slot->sl_session);
+out:
+	spin_unlock(&sessionid_lock);
+	dprintk("%s: return %d\n", __func__, ntohl(status));
+	return status;
 }
 #endif /* CONFIG_NFSD_V4_1 */
 
fs/nfsd/nfs4xdr.c +30 -2
index 840cf6a..c6b490e 100644
@@ -1114,7 +1114,16 @@ static __be32
 nfsd4_decode_sequence(struct nfsd4_compoundargs *argp,
 		      struct nfsd4_sequence *seq)
 {
-	return nfserr_opnotsupp;	/* stub */
+	DECODE_HEAD;
+
+	READ_BUF(NFS4_MAX_SESSIONID_LEN + 16);
+	COPYMEM(seq->sessionid.data, NFS4_MAX_SESSIONID_LEN);
+	READ32(seq->seqid);
+	READ32(seq->slotid);
+	READ32(seq->maxslots);
+	READ32(seq->cachethis);
+
+	DECODE_TAIL;
 }
 #endif /* CONFIG_NFSD_V4_1 */
 
@@ -2836,7 +2845,26 @@ static __be32
 nfsd4_encode_sequence(struct nfsd4_compoundres *resp, int nfserr,
 		      struct nfsd4_sequence *seq)
 {
-	/* stub */
+	ENCODE_HEAD;
+
+	if (nfserr)
+		goto out;
+
+	RESERVE_SPACE(NFS4_MAX_SESSIONID_LEN + 20);
+	WRITEMEM(seq->sessionid.data, NFS4_MAX_SESSIONID_LEN);
+	WRITE32(seq->seqid);
+	WRITE32(seq->slotid);
+	WRITE32(seq->maxslots);
+	/*
+	 * FIXME: for now:
+	 *   target_maxslots = maxslots
+	 *   status_flags = 0
+	 */
+	WRITE32(seq->maxslots);
+	WRITE32(0);
+
+	ADJUST_ARGS();
+out:
 	return nfserr;
 }
 #endif /* CONFIG_NFSD_V4_1 */
include/linux/nfsd/xdr4.h +9 -1
index ea5a427..9e4d8db 100644
@@ -362,7 +362,15 @@ struct nfsd4_create_session {
 };
 
 struct nfsd4_sequence {
-	int	foo;	/* stub */
+	struct nfs4_sessionid	sessionid;		/* request/response */
+	u32			seqid;			/* request/response */
+	u32			slotid;			/* request/response */
+	u32			maxslots;		/* request/response */
+	u32			cachethis;		/* request */
+#if 0
+	u32			target_maxslots;	/* response */
+	u32			status_flags;		/* response */
+#endif /* not yet */
 };
 
 struct nfsd4_destroy_session {
Signature · 1 line
1.6.2.1
#31 lore raw

PATCH 19/47 nfsd41: enforce NFS4ERR_SEQUENCE_POS operation order rules

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

---
 fs/nfsd/nfs4proc.c  |   24 ++++++++++++++++--------
 fs/nfsd/nfs4state.c |    4 ++++
 2 files changed, 20 insertions(+), 8 deletions(-)
fs/nfsd/nfs4proc.c +16 -8
index f618e8e..e703ac2 100644
@@ -811,14 +811,15 @@ static inline void nfsd4_increment_op_stats(u32 opnum)
 
 typedef __be32(*nfsd4op_func)(struct svc_rqst *, struct nfsd4_compound_state *,
 			      void *);
+enum nfsd4_op_flags {
+	ALLOWED_WITHOUT_FH = 1 << 0,	/* No current filehandle required */
+	ALLOWED_ON_ABSENT_FS = 2 << 0,	/* ops processed on absent fs */
+	ALLOWED_AS_FIRST_OP = 3 << 0,	/* ops reqired first in compound */
+};
 
 struct nfsd4_operation {
 	nfsd4op_func op_func;
 	u32 op_flags;
-/* Most ops require a valid current filehandle; a few don't: */
-#define ALLOWED_WITHOUT_FH 1
-/* GETATTR and ops not listed as returning NFS4ERR_MOVED: */
-#define ALLOWED_ON_ABSENT_FS 2
 	char *op_name;
 };
 
@@ -864,6 +865,13 @@ nfsd4_proc_compound(struct svc_rqst *rqstp,
 	if (args->minorversion > NFSD_SUPPORTED_MINOR_VERSION)
 		goto out;
 
+	op = &args->ops[0];
+	if (args->opcnt > 0 && op->status != nfserr_op_illegal &&
+	    !(nfsd4_ops[op->opnum].op_flags & ALLOWED_AS_FIRST_OP)) {
+		op->status = nfserr_sequence_pos;
+		goto encode_op;
+	}
+
 	status = nfs_ok;
 	while (!status && resp->opcnt < args->opcnt) {
 		op = &args->ops[resp->opcnt++];
@@ -1104,22 +1112,22 @@ static struct nfsd4_operation nfsd4_ops[] = {
 #if defined(CONFIG_NFSD_V4_1)
 	[OP_EXCHANGE_ID] = {
 		.op_func = (nfsd4op_func)nfsd4_exchange_id,
-		.op_flags = ALLOWED_WITHOUT_FH,
+		.op_flags = ALLOWED_WITHOUT_FH | ALLOWED_AS_FIRST_OP,
 		.op_name = "OP_EXCHANGE_ID",
 	},
 	[OP_CREATE_SESSION] = {
 		.op_func = (nfsd4op_func)nfsd4_create_session,
-		.op_flags = ALLOWED_WITHOUT_FH,
+		.op_flags = ALLOWED_WITHOUT_FH | ALLOWED_AS_FIRST_OP,
 		.op_name = "OP_CREATE_SESSION",
 	},
 	[OP_DESTROY_SESSION] = {
 		.op_func = (nfsd4op_func)nfsd4_destroy_session,
-		.op_flags = ALLOWED_WITHOUT_FH,
+		.op_flags = ALLOWED_WITHOUT_FH | ALLOWED_AS_FIRST_OP,
 		.op_name = "OP_DESTROY_SESSION",
 	},
 	[OP_SEQUENCE] = {
 		.op_func = (nfsd4op_func)nfsd4_sequence,
-		.op_flags = ALLOWED_WITHOUT_FH,
+		.op_flags = ALLOWED_WITHOUT_FH | ALLOWED_AS_FIRST_OP,
 		.op_name = "OP_SEQUENCE",
 	},
 #endif /* CONFIG_NFSD_V4_1 */
fs/nfsd/nfs4state.c +4 -0
index a19f292..10eb67b 100644
@@ -1051,10 +1051,14 @@ nfsd4_sequence(struct svc_rqst *rqstp,
 	       struct nfsd4_compound_state *cstate,
 	       struct nfsd4_sequence *seq)
 {
+	struct nfsd4_compoundres *resp = rqstp->rq_resp;
 	struct nfsd4_session *session;
 	struct nfsd4_slot *slot;
 	int status;
 
+	if (resp->opcnt != 1)
+		return nfserr_sequence_pos;
+
 	spin_lock(&sessionid_lock);
 	status = nfserr_badsession;
 	session = find_in_sessionid_hashtbl(&seq->sessionid);
Signature · 1 line
1.6.2.1
#32 lore raw

PATCH 20/47 nfsd41: DRC save, restore, and clear functions

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

Cache all the result pages, including the rpc header in rq_respages[0], for a request in the slot table cache entry.

Cache the statp pointer from nfsd_dispatch which points into rq_respages[0] just past the rpc header. When setting a cache entry, calculate and save the length of the nfs data minus the rpc header for rq_respages[0].

When replaying a cache entry, replace the cached rpc header with the replayed request rpc result header, unless there is not enough room in the cached results first page. In that case, use the cached rpc header.

The sessions fore channel maxresponse size cached is set to NFSD_PAGES_PER_SLOT
* PAGE_SIZE. For compounds we are cacheing with operations such as READDIR that use the xdr_buf->pages to hold data, we choose to cache the extra page of data rather than copying data from xdr_buf->pages into the xdr_buf->head page.

[nfsd41: limit cache to maxresponsesize_cached]
[nfsd41: mv nfsd4_set_statp under CONFIG_NFSD_V4_1]
---
 fs/nfsd/nfs4state.c        |  142 ++++++++++++++++++++++++++++++++++++++++++++
 fs/nfsd/nfssvc.c           |    4 +
 include/linux/nfsd/cache.h |    5 ++
 include/linux/nfsd/state.h |   13 ++++
 include/linux/nfsd/xdr4.h  |    4 +
 5 files changed, 168 insertions(+), 0 deletions(-)
fs/nfsd/nfs4state.c +142 -0
index 10eb67b..f0ce639 100644
@@ -860,6 +860,148 @@ out_err:
 }
 
 #if defined(CONFIG_NFSD_V4_1)
+void
+nfsd4_set_statp(struct svc_rqst *rqstp, __be32 *statp)
+{
+	struct nfsd4_compoundres *resp = rqstp->rq_resp;
+
+	resp->cstate.statp = statp;
+}
+
+/*
+ * Dereference the result pages.
+ */
+static void
+nfsd4_release_respages(struct page **respages, short resused)
+{
+	int page_no;
+
+	dprintk("--> %s\n", __func__);
+	for (page_no = 0; page_no < resused; page_no++) {
+		if (!respages[page_no])
+			continue;
+		put_page(respages[page_no]);
+		respages[page_no] = NULL;
+	}
+}
+
+static void
+nfsd4_move_pages(struct page **topages, struct page **frompages, short count)
+{
+	int page_no;
+
+	for (page_no = 0; page_no < count; page_no++) {
+		topages[page_no] = frompages[page_no];
+		if (!topages[page_no])
+			continue;
+		get_page(topages[page_no]);
+	}
+}
+
+/*
+ * Cache the reply pages up to NFSD_PAGES_PER_SLOT + 1, clearing the previous
+ * pages. We add a page to NFSD_PAGES_PER_SLOT for the case where the total
+ * length of the XDR response is less than se_fmaxresp_cached
+ * (NFSD_PAGES_PER_SLOT * PAGE_SIZE) but the xdr_buf pages is used for a
+ * of the reply (e.g. readdir).
+ *
+ * Store the base and length of the rq_req.head[0] page
+ * of the NFSv4.1 data, just past the rpc header.
+ */
+void
+nfsd4_set_cache_entry(struct nfsd4_compoundres *resp)
+{
+	struct nfsd4_cache_entry *entry = &resp->cstate.slot->sl_cache_entry;
+	struct svc_rqst *rqstp = resp->rqstp;
+	struct kvec *resv = &rqstp->rq_res.head[0];
+
+	dprintk("--> %s entry %p\n", __func__, entry);
+
+	/* Don't cache a failed OP_SEQUENCE */
+	if (resp->opcnt == 1 && resp->cstate.status)
+		return;
+	nfsd4_release_respages(entry->ce_respages, entry->ce_resused);
+	entry->ce_resused = rqstp->rq_resused;
+	if (entry->ce_resused > NFSD_PAGES_PER_SLOT + 1)
+		entry->ce_resused = NFSD_PAGES_PER_SLOT + 1;
+	nfsd4_move_pages(entry->ce_respages, rqstp->rq_respages,
+			 entry->ce_resused);
+	entry->ce_status = resp->cstate.status;
+	entry->ce_datav.iov_base = resp->cstate.statp;
+	entry->ce_datav.iov_len = resv->iov_len - ((char *)resp->cstate.statp -
+				(char *)page_address(rqstp->rq_respages[0]));
+	entry->ce_opcnt = resp->opcnt;
+	/* Current request rpc header length*/
+	entry->ce_rpchdrlen = (char *)resp->cstate.statp -
+				(char *)page_address(rqstp->rq_respages[0]);
+}
+
+/*
+ * Copy the cached NFSv4.1 reply skipping the cached rpc header into the
+ * replay result res.head[0] past the rpc header to end up with replay
+ * rpc header and cached NFSv4.1 reply.
+ */
+static int
+nfsd41_copy_replay_data(struct nfsd4_compoundres *resp,
+			struct nfsd4_cache_entry *entry)
+{
+	struct svc_rqst *rqstp = resp->rqstp;
+	struct kvec *resv = &resp->rqstp->rq_res.head[0];
+	int len;
+
+	/* Current request rpc header length*/
+	len = (char *)resp->cstate.statp -
+			(char *)page_address(rqstp->rq_respages[0]);
+	if (entry->ce_datav.iov_len + len > PAGE_SIZE) {
+		dprintk("%s v41 cached reply too large (%Zd).\n", __func__,
+			entry->ce_datav.iov_len);
+		return 0;
+	}
+	/* copy the cached reply nfsd data past the current rpc header */
+	memcpy((char *)resv->iov_base + len, entry->ce_datav.iov_base,
+		entry->ce_datav.iov_len);
+	resv->iov_len = len + entry->ce_datav.iov_len;
+	return 1;
+}
+
+/*
+ * Keep the first page of the replay. Copy the NFSv4.1 data from the first
+ * cached page.  Replace any futher replay pages from the cache.
+ */
+__be32
+nfsd4_replay_cache_entry(struct nfsd4_compoundres *resp)
+{
+	struct nfsd4_cache_entry *entry = &resp->cstate.slot->sl_cache_entry;
+	__be32 status;
+
+	dprintk("--> %s entry %p\n", __func__, entry);
+
+
+	if (!nfsd41_copy_replay_data(resp, entry)) {
+		/*
+		 * Not enough room to use the replay rpc header, send the
+		 * cached header. Release all the allocated result pages.
+		 */
+		svc_free_res_pages(resp->rqstp);
+		nfsd4_move_pages(resp->rqstp->rq_respages, entry->ce_respages,
+			entry->ce_resused);
+	} else {
+		/* Release all but the first allocated result page */
+
+		resp->rqstp->rq_resused--;
+		svc_free_res_pages(resp->rqstp);
+
+		nfsd4_move_pages(&resp->rqstp->rq_respages[1],
+				 &entry->ce_respages[1],
+				 entry->ce_resused - 1);
+	}
+
+	resp->rqstp->rq_resused = entry->ce_resused;
+	status = entry->ce_status;
+
+	return status;
+}
+
 /*
  * Set the exchange_id flags returned by the server.
  */
fs/nfsd/nfssvc.c +4 -0
index ef0a368..b5168d1 100644
@@ -515,6 +515,10 @@ nfsd_dispatch(struct svc_rqst *rqstp, __be32 *statp)
 		+ rqstp->rq_res.head[0].iov_len;
 	rqstp->rq_res.head[0].iov_len += sizeof(__be32);
 
+	/* NFSv4.1 DRC requires statp */
+	if (rqstp->rq_vers == 4)
+		nfsd4_set_statp(rqstp, statp);
+
 	/* Now call the procedure handler, and encode NFS status. */
 	nfserr = proc->pc_func(rqstp, rqstp->rq_argp, rqstp->rq_resp);
 	nfserr = map_new_errors(rqstp->rq_vers, nfserr);
include/linux/nfsd/cache.h +5 -0
index 04b355c..57a83c7 100644
@@ -75,5 +75,10 @@ int	nfsd_reply_cache_init(void);
 void	nfsd_reply_cache_shutdown(void);
 int	nfsd_cache_lookup(struct svc_rqst *, int);
 void	nfsd_cache_update(struct svc_rqst *, int, __be32 *);
+#ifdef CONFIG_NFSD_V4_1
+void	nfsd4_set_statp(struct svc_rqst *rqstp, __be32 *statp);
+#else /* CONFIG_NFSD_V4_1 */
+static inline void nfsd4_set_statp(struct svc_rqst *rqstp, __be32 *statp) {}
+#endif /* CONFIG_NFSD_V4_1 */
 
 #endif /* NFSCACHE_H */
include/linux/nfsd/state.h +13 -0
index feab6ec..8ca6a82 100644
@@ -99,10 +99,23 @@ struct nfs4_callback {
 	struct rpc_clnt *       cb_client;
 };
 
+/* Maximum number of pages per slot cache entry */
+#define NFSD_PAGES_PER_SLOT	1
+
+struct nfsd4_cache_entry {
+	__be32		ce_status;
+	struct kvec	ce_datav; /* encoded NFSv4.1 data in rq_res.head[0] */
+	struct page	*ce_respages[NFSD_PAGES_PER_SLOT + 1];
+	short		ce_resused;
+	int		ce_opcnt;
+	int		ce_rpchdrlen;
+};
+
 struct nfsd4_slot {
 	bool				sl_inuse;
 	struct nfsd4_session		*sl_session;
 	u32				sl_seqid;
+	struct nfsd4_cache_entry	sl_cache_entry;
 };
 
 struct nfsd4_session {
include/linux/nfsd/xdr4.h +4 -0
index 9e4d8db..cde8947 100644
@@ -50,6 +50,8 @@ struct nfsd4_compound_state {
 	struct nfs4_stateowner	*replay_owner;
 	/* For sessions DRC */
 	struct nfsd4_slot	*slot;
+	__be32			*statp;
+	u32			status;
 };
 
 struct nfsd4_change_info {
@@ -490,6 +492,8 @@ extern __be32 nfsd4_setclientid_confirm(struct svc_rqst *rqstp,
 		struct nfsd4_compound_state *,
 		struct nfsd4_setclientid_confirm *setclientid_confirm);
 #if defined(CONFIG_NFSD_V4_1)
+extern void nfsd4_set_cache_entry(struct nfsd4_compoundres *resp);
+extern __be32 nfsd4_replay_cache_entry(struct nfsd4_compoundres *resp);
 extern __be32 nfsd4_exchange_id(struct svc_rqst *rqstp,
 		struct nfsd4_compound_state *,
 struct nfsd4_exchange_id *);
Signature · 1 line
1.6.2.1
#33 lore raw

PATCH 21/47 nfsd41: hard page limit for DRC

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

Use no more than 1/128th of the number of free pages at nfsd startup for the v4.1 DRC.

This is an arbitrary default which should probably end up under the control of an administrator.

[moved added fields in struct svc_serv under CONFIG_NFSD_V4_1]
[fix set_max_drc calculation of sv_drc_max_pages]
---
 fs/nfsd/nfssvc.c           |   25 +++++++++++++++++++++++++
 include/linux/nfsd/nfsd.h  |    7 +++++++
 include/linux/sunrpc/svc.h |    4 ++++
 3 files changed, 36 insertions(+), 0 deletions(-)
fs/nfsd/nfssvc.c +25 -0
index b5168d1..ab60320 100644
@@ -22,6 +22,7 @@
 #include <linux/freezer.h>
 #include <linux/fs_struct.h>
 #include <linux/kthread.h>
+#include <linux/swap.h>
 
 #include <linux/sunrpc/types.h>
 #include <linux/sunrpc/stats.h>
@@ -197,6 +198,28 @@ void nfsd_reset_versions(void)
 	}
 }
 
+/*
+ * Each session guarantees a negotiated per slot memory cache for replies
+ * which in turn consumes memory beyond the v2/v3/v4.0 server. A dedicated
+ * NFSv4.1 server might want to use more memory for a DRC than a machine
+ * with mutiple services.
+ *
+ * Impose a hard limit on the number of pages for the DRC which varies
+ * according to the machines free pages. This is of course only a default.
+ *
+ * For now this is a #defined shift which could be under admin control
+ * in the future.
+ */
+static void set_max_drc(void)
+{
+#ifdef CONFIG_NFSD_V4_1
+	nfsd_serv->sv_drc_max_pages = nr_free_buffer_pages()
+						>> NFSD_DRC_SIZE_SHIFT;
+	nfsd_serv->sv_drc_pages_used = 0;
+	dprintk("%s svc_drc_max_pages %u\n", __func__,
+		nfsd_serv->sv_drc_max_pages);
+#endif /* CONFIG_NFSD_V4_1 */
+}
 
 int nfsd_create_serv(void)
 {
@@ -229,6 +252,8 @@ int nfsd_create_serv(void)
 				      nfsd_last_thread, nfsd, THIS_MODULE);
 	if (nfsd_serv == NULL)
 		err = -ENOMEM;
+	else
+		set_max_drc();
 
 	do_gettimeofday(&nfssvc_boot);		/* record boot time */
 	return err;
include/linux/nfsd/nfsd.h +7 -0
index ab9616d..3bccec3 100644
@@ -361,6 +361,13 @@ extern struct timeval	nfssvc_boot;
  | FATTR4_WORD1_TIME_DELTA   | FATTR4_WORD1_TIME_METADATA    \
  | FATTR4_WORD1_TIME_MODIFY     | FATTR4_WORD1_TIME_MODIFY_SET | FATTR4_WORD1_MOUNTED_ON_FILEID)
 
+#if defined(CONFIG_NFSD_V4_1)
+
+/* The percent of nr_free_buffer_pages used by the V4.1 server DRC */
+#define NFSD_DRC_SIZE_SHIFT	7
+
+#endif /* CONFIG_NFSD_V4_1 */
+
 /* These will return ERR_INVAL if specified in GETATTR or READDIR. */
 #define NFSD_WRITEONLY_ATTRS_WORD1							    \
 (FATTR4_WORD1_TIME_ACCESS_SET   | FATTR4_WORD1_TIME_MODIFY_SET)
include/linux/sunrpc/svc.h +4 -0
index 815dd58..c4f0e1e 100644
@@ -95,6 +95,10 @@ struct svc_serv {
 	struct module *		sv_module;	/* optional module to count when
 						 * adding threads */
 	svc_thread_fn		sv_function;	/* main function for threads */
+#if defined(CONFIG_NFSD_V4_1)
+	unsigned int		sv_drc_max_pages; /* Total pages for DRC */
+	unsigned int		sv_drc_pages_used;/* DRC pages used */
+#endif /* CONFIG_NFSD_V4_1 */
 };
 
 /*
Signature · 1 line
1.6.2.1
#34 lore raw

PATCH 22/47 nfsd41: nfsd DRC logic

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

Replay a request in nfsd4_sequence.
Add a minorversion to struct nfsd4_compound_state.

Pass the current slot to nfs4svc_encode_compound res via struct nfsd4_compoundres to set an NFSv4.1 DRC entry.

[nfsd41: use bool inuse for slot state]
---
 fs/nfsd/nfs4proc.c        |    7 +++++++
 fs/nfsd/nfs4state.c       |    6 ++++++
 fs/nfsd/nfs4xdr.c         |   13 +++++++++++++
 include/linux/nfsd/xdr4.h |    1 +
 4 files changed, 27 insertions(+), 0 deletions(-)
fs/nfsd/nfs4proc.c +7 -0
index e703ac2..bdbeb87 100644
@@ -920,6 +920,12 @@ nfsd4_proc_compound(struct svc_rqst *rqstp,
 			BUG_ON(op->status == nfs_ok);
 
 encode_op:
+		/* Only from SEQUENCE or CREATE_SESSION */
+		if (resp->cstate.status == nfserr_replay_cache) {
+			dprintk("%s NFS4.1 replay from cache\n", __func__);
+			status = op->status;
+			goto out;
+		}
 		if (op->status == nfserr_replay_me) {
 			op->replay = &cstate->replay_owner->so_replay;
 			nfsd4_encode_replay(resp, op);
@@ -948,6 +954,7 @@ encode_op:
 		status = nfserr_jukebox;
 	}
 
+	resp->cstate.status = status;
 	fh_put(&resp->cstate.current_fh);
 	fh_put(&resp->cstate.save_fh);
 	BUG_ON(resp->cstate.replay_owner);
fs/nfsd/nfs4state.c +6 -0
index f0ce639..07c869d 100644
@@ -997,6 +997,8 @@ nfsd4_replay_cache_entry(struct nfsd4_compoundres *resp)
 	}
 
 	resp->rqstp->rq_resused = entry->ce_resused;
+	resp->opcnt = entry->ce_opcnt;
+	resp->cstate.iovlen = entry->ce_datav.iov_len + entry->ce_rpchdrlen;
 	status = entry->ce_status;
 
 	return status;
@@ -1217,6 +1219,10 @@ nfsd4_sequence(struct svc_rqst *rqstp,
 	status = check_slot_seqid(seq->seqid, slot);
 	if (status == nfserr_replay_cache) {
 		cstate->slot = slot;
+		/* Return the cached reply status and set cstate->status
+		 * for nfsd4_svc_encode_compoundres processing*/
+		status = nfsd4_replay_cache_entry(resp);
+		cstate->status = nfserr_replay_cache;
 		goto replay_cache;
 	}
 	if (status)
fs/nfsd/nfs4xdr.c +13 -0
index c6b490e..57afb33 100644
@@ -3059,6 +3059,19 @@ nfs4svc_encode_compoundres(struct svc_rqst *rqstp, __be32 *p, struct nfsd4_compo
 		iov = &rqstp->rq_res.head[0];
 	iov->iov_len = ((char*)resp->p) - (char*)iov->iov_base;
 	BUG_ON(iov->iov_len > PAGE_SIZE);
+#ifdef CONFIG_NFSD_V4_1
+	if (resp->cstate.slot != NULL) {
+		if (resp->cstate.status == nfserr_replay_cache) {
+			iov->iov_len = resp->cstate.iovlen;
+		} else {
+			nfsd4_set_cache_entry(resp);
+			dprintk("%s: SET SLOT STATE TO AVAILABLE\n", __func__);
+			resp->cstate.slot->sl_inuse = 0;
+		}
+		if (resp->cstate.slot->sl_session)
+			nfsd4_put_session(resp->cstate.slot->sl_session);
+	}
+#endif /* CONFIG_NFSD_V4_1 */
 	return 1;
 }
 
include/linux/nfsd/xdr4.h +1 -0
index cde8947..5c0d376 100644
@@ -51,6 +51,7 @@ struct nfsd4_compound_state {
 	/* For sessions DRC */
 	struct nfsd4_slot	*slot;
 	__be32			*statp;
+	size_t			iovlen;
 	u32			status;
 };
 
Signature · 1 line
1.6.2.1
#35 lore raw

PATCH 23/47 nfsd41: clear DRC cache on free_session

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

---
 fs/nfsd/nfs4state.c |    7 +++++++
 1 files changed, 7 insertions(+), 0 deletions(-)
fs/nfsd/nfs4state.c +7 -0
index 07c869d..37865c9 100644
@@ -442,12 +442,19 @@ release_session(struct nfsd4_session *ses)
 	nfsd4_put_session(ses);
 }
 
+static void nfsd4_release_respages(struct page **respages, short resused);
+
 void
 free_session(struct kref *kref)
 {
 	struct nfsd4_session *ses;
+	int i;
 
 	ses = container_of(kref, struct nfsd4_session, se_ref);
+	for (i = 0; i < ses->se_fnumslots; i++) {
+		struct nfsd4_cache_entry *e = &ses->se_slots[i].sl_cache_entry;
+		nfsd4_release_respages(e->ce_respages, e->ce_resused);
+	}
 	kfree(ses->se_slots);
 	kfree(ses);
 }
Signature · 1 line
1.6.2.1
#36 lore raw

PATCH 24/47 nfsd41: create_session operation

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

Implement the create_session operation confoming to
http://tools.ietf.org/html/draft-ietf-nfsv4-minorversion1-26

Look up the client id (generated by the server on exchange_id, given by the client on create_session).
If neither a confirmed or unconfirmed client is found then the client id is stale
If a confirmed cilent is found (i.e. we already received create_session for it) then compare the sequence id to determine if it's a replay or possibly a mis-ordered rpc. If the seqid is in order, update the confirmed client seqid and procedd with updating the session parameters.

If an unconfirmed client_id is found then verify the creds and seqid. If both match move the client id to confirmed state and proceed with processing the create_session.

Currently, we do not support persistent sessions, and RDMA.

alloc_init_session generates a new sessionid and creates a session structure.

NFSD_PAGES_PER_SLOT is used for the max response cached calculation, and for the counting of DRC pages using the hard limits set in struct srv_serv.

A note on NFSD_PAGES_PER_SLOT:

Other patches in this series allow for NFSD_PAGES_PER_SLOT + 1 pages to be cached in a DRC slot when the response size is less than NFSD_PAGES_PER_SLOT * PAGE_SIZE but xdr_buf pages are used. e.g. a READDIR operation will encode a small amount of data in the xdr_buf head, and then the READDIR in the xdr_buf pages. So, the hard limit calculation use of pages by a session is underestimated by the number of cached operations using the xdr_buf pages.

Yet another patch caches no pages for the solo sequence operation, or any compound where cache_this is False. So the hard limit calculation use of pages by a session is overestimated by the number of these operations in the cache.

TODO: improve resource pre-allocation and negotiate session parameters accordingly. Respect and possibly adjust backchannel attributes.

[nfsd41: remove headerpadsz from channel attributes]

Our client and server only support a headerpadsz of 0.

[nfsd41: use DRC limits in fore channel init]
[nfsd41: do not change CREATE_SESSION back channel attrs]
[use sessionid_lock spin lock]
[nfsd41: use bool inuse for slot state]
---
 fs/nfsd/nfs4state.c        |  197 +++++++++++++++++++++++++++++++++++++++++++-
 fs/nfsd/nfs4xdr.c          |  147 ++++++++++++++++++++++++++++++++-
 include/linux/nfsd/state.h |    7 ++
 include/linux/nfsd/xdr4.h  |   21 +++++-
 4 files changed, 368 insertions(+), 4 deletions(-)
fs/nfsd/nfs4state.c +196 -1
index 37865c9..e4e2c19 100644
@@ -68,6 +68,9 @@ static u32 current_delegid = 1;
 static u32 nfs4_init;
 static stateid_t zerostateid;             /* bits all 0 */
 static stateid_t onestateid;              /* bits all 1 */
+#ifdef CONFIG_NFSD_V4_1
+static u64 current_sessionid = 1;
+#endif /* CONFIG_NFSD_V4_1 */
 
 #define ZERO_STATEID(stateid) (!memcmp((stateid), &zerostateid, sizeof(stateid_t)))
 #define ONE_STATEID(stateid)  (!memcmp((stateid), &onestateid, sizeof(stateid_t)))
@@ -402,6 +405,138 @@ dump_sessionid(const char *fn, struct nfs4_sessionid *sessionid)
 	dprintk("%s: %u:%u:%u:%u\n", fn, ptr[0], ptr[1], ptr[2], ptr[3]);
 }
 
+static void
+gen_sessionid(struct nfsd4_session *ses)
+{
+	struct nfs4_client *clp = ses->se_client;
+	struct nfsd4_sessionid *sid;
+
+	sid = (struct nfsd4_sessionid *)ses->se_sessionid.data;
+	sid->clientid = clp->cl_clientid;
+	sid->sequence = current_sessionid++;
+	sid->reserved = 0;
+}
+
+/*
+ * Give the client the number of slots it requests bound by
+ * NFSD_MAX_SLOTS_PER_SESSION and by sv_drc_max_pages.
+ *
+ * If we run out of pages (sv_drc_pages_used == sv_drc_max_pages) we
+ * should (up to a point) re-negotiate active sessions and reduce their
+ * slot usage to make rooom for new connections. For now we just fail the
+ * create session.
+ */
+static int set_forechannel_maxreqs(struct nfsd4_channel_attrs *fchan)
+{
+	int status = 0, np = fchan->maxreqs * NFSD_PAGES_PER_SLOT;
+
+	spin_lock(&nfsd_serv->sv_lock);
+	if (np + nfsd_serv->sv_drc_pages_used > nfsd_serv->sv_drc_max_pages)
+		np = nfsd_serv->sv_drc_max_pages - nfsd_serv->sv_drc_pages_used;
+	nfsd_serv->sv_drc_pages_used += np;
+	spin_unlock(&nfsd_serv->sv_lock);
+
+	if (np <= 0) {
+		status = nfserr_resource;
+		fchan->maxreqs = 0;
+	} else
+		fchan->maxreqs = np / NFSD_PAGES_PER_SLOT;
+
+	return status;
+}
+
+/*
+ * fchan holds the client values on input, and the server values on output
+ */
+static int init_forechannel_attrs(struct svc_rqst *rqstp,
+				    struct nfsd4_session *session,
+				    struct nfsd4_channel_attrs *fchan)
+{
+	int status = 0;
+	__u32   maxcount = svc_max_payload(rqstp);
+
+	/* headerpadsz set to zero in encode routine*/
+
+	/* Use the client's max request and max response size if possible */
+	if (fchan->maxreq_sz > maxcount)
+		fchan->maxreq_sz = maxcount;
+	session->se_fmaxreq_sz = fchan->maxreq_sz;
+
+	if (fchan->maxresp_sz > maxcount)
+		fchan->maxresp_sz = maxcount;
+	session->se_fmaxresp_sz = fchan->maxresp_sz;
+
+	/* Set the max response cached size our default which is
+	 * a multiple of PAGE_SIZE and small */
+	session->se_fmaxresp_cached = NFSD_PAGES_PER_SLOT * PAGE_SIZE;
+	fchan->maxresp_cached = session->se_fmaxresp_cached;
+
+	/* Use the client's maxops if possible */
+	if (fchan->maxops > NFSD_MAX_OPS_PER_COMPOUND)
+		fchan->maxops = NFSD_MAX_OPS_PER_COMPOUND;
+	session->se_fmaxops = fchan->maxops;
+
+	/* try to use the client requested number of slots */
+	if (fchan->maxreqs > NFSD_MAX_SLOTS_PER_SESSION)
+		fchan->maxreqs = NFSD_MAX_SLOTS_PER_SESSION;
+
+	/* FIXME: Error means no more DRC pages so the server should
+	 * recover pages from existing sessions. For now fail session
+	 * creation.
+	 */
+	status = set_forechannel_maxreqs(fchan);
+
+	session->se_fnumslots = fchan->maxreqs;
+	return status;
+}
+
+static int
+alloc_init_session(struct svc_rqst *rqstp, struct nfs4_client *clp,
+		   struct nfsd4_create_session *cses)
+{
+	struct nfsd4_session *new;
+	int idx, status = nfserr_resource, slotsize, i;
+
+	new = kzalloc(sizeof(*new), GFP_KERNEL);
+	if (!new)
+		goto out;
+
+	/* FIXME: For now, we just accept the client back channel attributes. */
+	status = init_forechannel_attrs(rqstp, new, &cses->fore_channel);
+	if (status)
+		goto out_free;
+
+	slotsize = new->se_fnumslots * sizeof(struct nfsd4_slot);
+	new->se_slots = kzalloc(slotsize, GFP_KERNEL);
+	if (!new->se_slots)
+		goto out_free;
+
+	for (i = 0; i < new->se_fnumslots; i++)
+		new->se_slots[i].sl_session = new;
+
+	new->se_client = clp;
+	gen_sessionid(new);
+	idx = hash_sessionid(&new->se_sessionid);
+	memcpy(clp->cl_sessionid.data, new->se_sessionid.data,
+	       NFS4_MAX_SESSIONID_LEN);
+
+	new->se_flags = cses->flags;
+	kref_init(&new->se_ref);
+	INIT_LIST_HEAD(&new->se_hash);
+	INIT_LIST_HEAD(&new->se_perclnt);
+	spin_lock(&sessionid_lock);
+	list_add(&new->se_hash, &sessionid_hashtbl[idx]);
+	list_add(&new->se_perclnt, &clp->cl_sessions);
+	spin_unlock(&sessionid_lock);
+
+	status = nfs_ok;
+out:
+	return status;
+out_free:
+	kfree(new);
+	goto out;
+}
+
 /* caller must hold sessionid_lock */
 static struct nfsd4_session *
 find_in_sessionid_hashtbl(struct nfs4_sessionid *sessionid)
@@ -1186,7 +1321,67 @@ nfsd4_create_session(struct svc_rqst *rqstp,
 		     struct nfsd4_compound_state *cstate,
 		     struct nfsd4_create_session *cr_ses)
 {
-	return -1;	/* stub */
+	u32 ip_addr = svc_addr_in(rqstp)->sin_addr.s_addr;
+	struct nfs4_client *conf, *unconf;
+	int status = 0;
+
+	nfs4_lock_state();
+	unconf = find_unconfirmed_client(&cr_ses->clientid);
+	conf = find_confirmed_client(&cr_ses->clientid);
+
+	if (conf) {
+		status = nfs_ok;
+		if (conf->cl_seqid == cr_ses->seqid) {
+			dprintk("Got a create_session replay! seqid= %d\n",
+				conf->cl_seqid);
+			goto out_replay;
+		} else if (cr_ses->seqid != conf->cl_seqid + 1) {
+			status = nfserr_seq_misordered;
+			dprintk("Sequence misordered!\n");
+			dprintk("Expected seqid= %d but got seqid= %d\n",
+				conf->cl_seqid, cr_ses->seqid);
+			goto out;
+		}
+		conf->cl_seqid++;
+	} else if (unconf) {
+		if (!same_creds(&unconf->cl_cred, &rqstp->rq_cred) ||
+		    (ip_addr != unconf->cl_addr)) {
+			status = nfserr_clid_inuse;
+			goto out;
+		}
+
+		if (unconf->cl_seqid != cr_ses->seqid) {
+			status = nfserr_seq_misordered;
+			goto out;
+		}
+
+		move_to_confirmed(unconf);
+
+		/*
+		 * We do not support RDMA or persistent sessions
+		 */
+		cr_ses->flags &= ~SESSION4_PERSIST;
+		cr_ses->flags &= ~SESSION4_RDMA;
+
+		conf = unconf;
+	} else {
+		status = nfserr_stale_clientid;
+		goto out;
+	}
+
+	status = alloc_init_session(rqstp, conf, cr_ses);
+	if (status)
+		goto out;
+
+out_replay:
+	memcpy(cr_ses->sessionid.data, conf->cl_sessionid.data,
+	       NFS4_MAX_SESSIONID_LEN);
+	cr_ses->seqid = conf->cl_seqid;
+
+out:
+	nfs4_unlock_state();
+	dprintk("%s returns %d\n", __func__, ntohl(status));
+	return status;
 }
 
 __be32
fs/nfsd/nfs4xdr.c +145 -2

159 lines not on this page · raw message

include/linux/nfsd/state.h +7 -0
index 8ca6a82..98d7b1c 100644
@@ -99,8 +99,12 @@ struct nfs4_callback {
 	struct rpc_clnt *       cb_client;
 };
 
+/* Maximum number of slots per session. 128 is useful for long haul TCP */
+#define NFSD_MAX_SLOTS_PER_SESSION	128
 /* Maximum number of pages per slot cache entry */
 #define NFSD_PAGES_PER_SLOT	1
+/* Maximum number of operations per session compound */
+#define NFSD_MAX_OPS_PER_COMPOUND	16
 
 struct nfsd4_cache_entry {
 	__be32		ce_status;
@@ -188,6 +192,9 @@ struct nfs4_client {
 	struct list_head	cl_sessions;
 	u32			cl_seqid;       /* seqid for create_session */
 	u32			cl_exchange_flags;
+	struct nfs4_sessionid	cl_sessionid;
+
+	struct svc_xprt		*cl_cb_xprt;	/* 4.1 callback transport */
 #endif /* CONFIG_NFSD_V4_1 */
 };
 
include/linux/nfsd/xdr4.h +20 -1
index 5c0d376..c7bf0a1 100644
@@ -360,8 +360,27 @@ struct nfsd4_exchange_id {
 	int		spa_how;
 };
 
+struct nfsd4_channel_attrs {
+	u32		headerpadsz;
+	u32		maxreq_sz;
+	u32		maxresp_sz;
+	u32		maxresp_cached;
+	u32		maxops;
+	u32		maxreqs;
+	u32		nr_rdma_attrs;
+	u32		rdma_attrs;
+};
+
 struct nfsd4_create_session {
-	int	foo;	/* stub */
+	clientid_t		clientid;
+	struct nfs4_sessionid	sessionid;
+	u32			seqid;
+	u32			flags;
+	struct nfsd4_channel_attrs fore_channel;
+	struct nfsd4_channel_attrs back_channel;
+	u32			callback_prog;
+	u32			uid;
+	u32			gid;
 };
 
 struct nfsd4_sequence {
Signature · 1 line
1.6.2.1
#37 lore raw

PATCH 25/47 nfsd41: Add a create session replay cache

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

Replace the nfs4_client cl_seqid field with a single struct nfs41_slot used for the create session replay cache.

The CREATE_SESSION slot sets the sl_session pointer to NULL. Otherwise, the slot and it's replay cache are used just like the session slots.

Fix unconfirmed create_session replay response by initializing the create_session slot sequence id to 0.

A future patch will set the CREATE_SESSION cache when a SEQUENCE operation preceeds the CREATE_SESSION operation. This compound is currently only cached in the session slot table.

[nfsd41: use bool inuse for slot state]
---
 fs/nfsd/nfs4state.c        |   47 ++++++++++++++++++++++++++++++-------------
 include/linux/nfsd/state.h |    2 +-
 2 files changed, 34 insertions(+), 15 deletions(-)
fs/nfsd/nfs4state.c +33 -14
index e4e2c19..61af434 100644
@@ -662,6 +662,10 @@ static inline void
 free_client(struct nfs4_client *clp)
 {
 	shutdown_callback_client(clp);
+#if defined(CONFIG_NFSD_V4_1)
+	nfsd4_release_respages(clp->cl_slot.sl_cache_entry.ce_respages,
+			     clp->cl_slot.sl_cache_entry.ce_resused);
+#endif /* CONFIG_NFSD_V4_1 */
 	if (clp->cl_cred.cr_group_info)
 		put_group_info(clp->cl_cred.cr_group_info);
 	kfree(clp->cl_principal);
@@ -1055,12 +1059,14 @@ nfsd4_set_cache_entry(struct nfsd4_compoundres *resp)
 {
 	struct nfsd4_cache_entry *entry = &resp->cstate.slot->sl_cache_entry;
 	struct svc_rqst *rqstp = resp->rqstp;
+	struct nfsd4_compoundargs *args = rqstp->rq_argp;
+	struct nfsd4_op *op = &args->ops[resp->opcnt];
 	struct kvec *resv = &rqstp->rq_res.head[0];
 
 	dprintk("--> %s entry %p\n", __func__, entry);
 
-	/* Don't cache a failed OP_SEQUENCE */
-	if (resp->opcnt == 1 && resp->cstate.status)
+	/* Don't cache a failed OP_SEQUENCE. */
+	if (resp->opcnt == 1 && op->opnum == OP_SEQUENCE && resp->cstate.status)
 		return;
 	nfsd4_release_respages(entry->ce_respages, entry->ce_resused);
 	entry->ce_resused = rqstp->rq_resused;
@@ -1276,11 +1282,12 @@ out_copy:
 	exid->clientid.cl_boot = new->cl_clientid.cl_boot;
 	exid->clientid.cl_id = new->cl_clientid.cl_id;
 
-	new->cl_seqid = exid->seqid = 1;
+	new->cl_slot.sl_seqid = 0;
+	exid->seqid = 1;
 	nfsd4_set_ex_flags(new, exid);
 
 	dprintk("nfsd4_exchange_id seqid %d flags %x\n",
-		new->cl_seqid, new->cl_exchange_flags);
+		new->cl_slot.sl_seqid, new->cl_exchange_flags);
 	status = nfs_ok;
 
 out:
@@ -1322,7 +1329,9 @@ nfsd4_create_session(struct svc_rqst *rqstp,
 		     struct nfsd4_create_session *cr_ses)
 {
 	u32 ip_addr = svc_addr_in(rqstp)->sin_addr.s_addr;
+	struct nfsd4_compoundres *resp = rqstp->rq_resp;
 	struct nfs4_client *conf, *unconf;
+	struct nfsd4_slot *slot = NULL;
 	int status = 0;
 
 	nfs4_lock_state();
@@ -1330,19 +1339,24 @@ nfsd4_create_session(struct svc_rqst *rqstp,
 	conf = find_confirmed_client(&cr_ses->clientid);
 
 	if (conf) {
-		status = nfs_ok;
-		if (conf->cl_seqid == cr_ses->seqid) {
+		slot = &conf->cl_slot;
+		status = check_slot_seqid(cr_ses->seqid, slot);
+		if (status == nfserr_replay_cache) {
 			dprintk("Got a create_session replay! seqid= %d\n",
-				conf->cl_seqid);
-			goto out_replay;
-		} else if (cr_ses->seqid != conf->cl_seqid + 1) {
+				slot->sl_seqid);
+			cstate->slot = slot;
+			cstate->status = status;
+			/* Return the cached reply status */
+			status = nfsd4_replay_cache_entry(resp);
+			goto out;
+		} else if (cr_ses->seqid != conf->cl_slot.sl_seqid + 1) {
 			status = nfserr_seq_misordered;
 			dprintk("Sequence misordered!\n");
 			dprintk("Expected seqid= %d but got seqid= %d\n",
-				conf->cl_seqid, cr_ses->seqid);
+				slot->sl_seqid, cr_ses->seqid);
 			goto out;
 		}
-		conf->cl_seqid++;
+		conf->cl_slot.sl_seqid++;
 	} else if (unconf) {
 		if (!same_creds(&unconf->cl_cred, &rqstp->rq_cred) ||
 		    (ip_addr != unconf->cl_addr)) {
@@ -1350,11 +1364,15 @@ nfsd4_create_session(struct svc_rqst *rqstp,
 			goto out;
 		}
 
-		if (unconf->cl_seqid != cr_ses->seqid) {
+		slot = &unconf->cl_slot;
+		status = check_slot_seqid(cr_ses->seqid, slot);
+		if (status) {
+			/* an unconfirmed replay returns misordered */
 			status = nfserr_seq_misordered;
 			goto out;
 		}
 
+		slot->sl_seqid++; /* from 0 to 1 */
 		move_to_confirmed(unconf);
 
 		/*
@@ -1373,11 +1391,12 @@ nfsd4_create_session(struct svc_rqst *rqstp,
 	if (status)
 		goto out;
 
-out_replay:
 	memcpy(cr_ses->sessionid.data, conf->cl_sessionid.data,
 	       NFS4_MAX_SESSIONID_LEN);
-	cr_ses->seqid = conf->cl_seqid;
+	cr_ses->seqid = slot->sl_seqid;
 
+	slot->sl_inuse = true;
+	cstate->slot = slot;
 out:
 	nfs4_unlock_state();
 	dprintk("%s returns %d\n", __func__, ntohl(status));
include/linux/nfsd/state.h +1 -1
index 98d7b1c..49d89fd 100644
@@ -190,7 +190,7 @@ struct nfs4_client {
 	u32			cl_firststate;	/* recovery dir creation */
 #ifdef CONFIG_NFSD_V4_1
 	struct list_head	cl_sessions;
-	u32			cl_seqid;       /* seqid for create_session */
+	struct nfsd4_slot	cl_slot;	/* create_session slot */
 	u32			cl_exchange_flags;
 	struct nfs4_sessionid	cl_sessionid;
 
Signature · 1 line
1.6.2.1
#38 lore raw

PATCH 26/47 nfsd41: non-page DRC for solo sequence responses

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

A session inactivity time compound (lease renewal) or a compound where the sequence operation has sa_cachethis set to FALSE do not require any pages to be held in the v4.1 DRC. This is because struct nfsd4_slot is already caching the session information.

Add logic to the nfs41 server to not cache response pages for solo sequence responses.

Return nfserr_replay_uncached_rep on the operation following the sequence operation when sa_cachethis is FALSE.

---
 fs/nfsd/nfs4proc.c         |   34 +++++++++++++++++++++++++++++-
 fs/nfsd/nfs4state.c        |   47 ++++++++++++++++++++++++++++++++++++++-----
 fs/nfsd/nfs4xdr.c          |    5 ++-
 include/linux/nfsd/state.h |    1 +
 include/linux/nfsd/xdr4.h  |   15 +++++++++++++-
 5 files changed, 91 insertions(+), 11 deletions(-)
fs/nfsd/nfs4proc.c +32 -2
index bdbeb87..a273023 100644
@@ -828,6 +828,34 @@ static struct nfsd4_operation nfsd4_ops[];
 static const char *nfsd4_op_name(unsigned opnum);
 
 /*
+ * This is a replay of a compound for which no cache entry pages
+ * were used. Encode the sequence operation, and if cachethis is FALSE
+ * encode the uncache rep error on the next operation.
+ */
+static __be32
+nfsd4_enc_no_page_replay(struct nfsd4_compoundargs *args,
+			 struct nfsd4_compoundres *resp)
+{
+	struct nfsd4_op *op;
+
+	dprintk("--> %s resp->opcnt %d ce_cachethis %u \n", __func__,
+		resp->opcnt, resp->cstate.slot->sl_cache_entry.ce_cachethis);
+
+	/* Encode the replayed sequence operation */
+	BUG_ON(resp->opcnt != 1);
+	op = &args->ops[resp->opcnt - 1];
+	nfsd4_encode_operation(resp, op);
+
+	/*return nfserr_retry_uncached_rep in next operation. */
+	if (resp->cstate.slot->sl_cache_entry.ce_cachethis == 0) {
+		op = &args->ops[resp->opcnt++];
+		op->status = nfserr_retry_uncached_rep;
+		nfsd4_encode_operation(resp, op);
+	}
+	return op->status;
+}
+
+/*
  * COMPOUND call.
  */
 static __be32
@@ -879,7 +907,6 @@ nfsd4_proc_compound(struct svc_rqst *rqstp,
 		dprintk("nfsv4 compound op #%d/%d: %d (%s)\n",
 			resp->opcnt, args->opcnt, op->opnum,
 			nfsd4_op_name(op->opnum));
-
 		/*
 		 * The XDR decode routines may have pre-set op->status;
 		 * for example, if there is a miscellaneous XDR error
@@ -923,7 +950,10 @@ encode_op:
 		/* Only from SEQUENCE or CREATE_SESSION */
 		if (resp->cstate.status == nfserr_replay_cache) {
 			dprintk("%s NFS4.1 replay from cache\n", __func__);
-			status = op->status;
+			if (nfsd4_no_page_in_cache(resp))
+				status = nfsd4_enc_no_page_replay(args, resp);
+			else
+				status = op->status;
 			goto out;
 		}
 		if (op->status == nfserr_replay_me) {
fs/nfsd/nfs4state.c +41 -6

98 lines not on this page · raw message

fs/nfsd/nfs4xdr.c +3 -2
index 60db854..a8bb04a 100644
@@ -2984,7 +2984,7 @@ nfsd4_encode_destroy_session(struct nfsd4_compoundres *resp, int nfserr,
 	return nfserr;
 }
 
-static __be32
+__be32
 nfsd4_encode_sequence(struct nfsd4_compoundres *resp, int nfserr,
 		      struct nfsd4_sequence *seq)
 {
@@ -3204,7 +3204,8 @@ nfs4svc_encode_compoundres(struct svc_rqst *rqstp, __be32 *p, struct nfsd4_compo
 	BUG_ON(iov->iov_len > PAGE_SIZE);
 #ifdef CONFIG_NFSD_V4_1
 	if (resp->cstate.slot != NULL) {
-		if (resp->cstate.status == nfserr_replay_cache) {
+		if (resp->cstate.status == nfserr_replay_cache &&
+				!nfsd4_no_page_in_cache(resp)) {
 			iov->iov_len = resp->cstate.iovlen;
 		} else {
 			nfsd4_set_cache_entry(resp);
include/linux/nfsd/state.h +1 -0
index 49d89fd..47c7836 100644
@@ -110,6 +110,7 @@ struct nfsd4_cache_entry {
 	__be32		ce_status;
 	struct kvec	ce_datav; /* encoded NFSv4.1 data in rq_res.head[0] */
 	struct page	*ce_respages[NFSD_PAGES_PER_SLOT + 1];
+	int		ce_cachethis;
 	short		ce_resused;
 	int		ce_opcnt;
 	int		ce_rpchdrlen;
include/linux/nfsd/xdr4.h +14 -1
index c7bf0a1..641e5cd 100644
@@ -482,6 +482,18 @@ struct nfsd4_compoundres {
 	struct nfsd4_compound_state	cstate;
 };
 
+static inline u32 nfsd4_is_solo_sequence(struct nfsd4_compoundres *resp)
+{
+	struct nfsd4_compoundargs *args = resp->rqstp->rq_argp;
+	return args->opcnt == 1 ? 1 : 0;
+}
+
+static inline u32 nfsd4_no_page_in_cache(struct nfsd4_compoundres *resp)
+{
+	return (resp->cstate.slot->sl_cache_entry.ce_cachethis == 0 ||
+			nfsd4_is_solo_sequence(resp));
+}
+
 #define NFS4_SVC_XDRSIZE		sizeof(struct nfsd4_compoundargs)
 
 static inline void
@@ -513,7 +525,8 @@ extern __be32 nfsd4_setclientid_confirm(struct svc_rqst *rqstp,
 		struct nfsd4_setclientid_confirm *setclientid_confirm);
 #if defined(CONFIG_NFSD_V4_1)
 extern void nfsd4_set_cache_entry(struct nfsd4_compoundres *resp);
-extern __be32 nfsd4_replay_cache_entry(struct nfsd4_compoundres *resp);
+extern __be32 nfsd4_replay_cache_entry(struct nfsd4_compoundres *resp,
+		struct nfsd4_sequence *seq);
 extern __be32 nfsd4_exchange_id(struct svc_rqst *rqstp,
 		struct nfsd4_compound_state *,
 struct nfsd4_exchange_id *);
Signature · 1 line
1.6.2.1
#39 lore raw

PATCH 27/47 nfsd41: destroy_session operation

In reply to Benny Halevy

Implement the destory_session operation confoming to
http://tools.ietf.org/html/draft-ietf-nfsv4-minorversion1-26

[use sessionid_lock spin lock]
---
 fs/nfsd/nfs4state.c       |   29 ++++++++++++++++++++++++++++-
 fs/nfsd/nfs4xdr.c         |    7 +++++--
 include/linux/nfsd/xdr4.h |    2 +-
 3 files changed, 34 insertions(+), 4 deletions(-)
fs/nfsd/nfs4state.c +28 -1
index f42cda9..9c93f96 100644
@@ -1439,7 +1439,34 @@ nfsd4_destroy_session(struct svc_rqst *r,
 		      struct nfsd4_compound_state *cstate,
 		      struct nfsd4_destroy_session *sessionid)
 {
-	return -1;	/* stub */
+	struct nfsd4_session *ses;
+	u32 status = nfserr_badsession;
+
+	/* Notes:
+	 * - The confirmed nfs4_client->cl_sessionid holds destroyed sessinid
+	 * - Should we return nfserr_back_chan_busy if waiting for
+	 *   callbacks on to-be-destroyed session?
+	 * - Do we need to clear any callback info from previous session?
+	 */
+
+	dump_sessionid(__func__, &sessionid->sessionid);
+	spin_lock(&sessionid_lock);
+	ses = find_in_sessionid_hashtbl(&sessionid->sessionid);
+	if (!ses) {
+		spin_unlock(&sessionid_lock);
+		goto out;
+	}
+
+	unhash_session(ses);
+	spin_unlock(&sessionid_lock);
+
+	/* wait for callbacks */
+	shutdown_callback_client(ses->se_client);
+	nfsd4_put_session(ses);
+	status = nfs_ok;
+out:
+	dprintk("%s returns %d\n", __func__, ntohl(status));
+	return status;
 }
 
 __be32
fs/nfsd/nfs4xdr.c +5 -2
index a8bb04a..5720aab 100644
@@ -1208,7 +1208,11 @@ static __be32
 nfsd4_decode_destroy_session(struct nfsd4_compoundargs *argp,
 			     struct nfsd4_destroy_session *destroy_session)
 {
-	return nfserr_opnotsupp;	/* stub */
+	DECODE_HEAD;
+	READ_BUF(NFS4_MAX_SESSIONID_LEN);
+	COPYMEM(destroy_session->sessionid.data, NFS4_MAX_SESSIONID_LEN);
+
+	DECODE_TAIL;
 }
 
 static __be32
@@ -2980,7 +2984,6 @@ static __be32
 nfsd4_encode_destroy_session(struct nfsd4_compoundres *resp, int nfserr,
 			     struct nfsd4_destroy_session *destroy_session)
 {
-	/* stub */
 	return nfserr;
 }
 
include/linux/nfsd/xdr4.h +1 -1
index 641e5cd..37a7c51 100644
@@ -396,7 +396,7 @@ struct nfsd4_sequence {
 };
 
 struct nfsd4_destroy_session {
-	int	foo;	/* stub */
+	struct nfs4_sessionid	sessionid;
 };
 #endif /* CONFIG_NFSD_V4_1 */
 
Signature · 1 line
1.6.2.1
#40 lore raw

PATCH 28/47 nfsd41: stateid handling

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

When sessions are used, stateful operation sequenceid and stateid handling are not used. When sessions are used, on the first open set the seqid to 1, mark state confirmed and skip seqid processing.

When sessionas are used the stateid generation number is ignored when it is zero whereas without sessions bad_stateid or stale stateid is returned.

Add flags to propagate session use to all stateful ops and down to check_stateid_generation.

[nfsd4_has_session should return a boolean, not u32]
---
 fs/nfsd/nfs4proc.c         |   17 ++++++++--
 fs/nfsd/nfs4state.c        |   70 ++++++++++++++++++++++++++++++++++---------
 fs/nfsd/nfs4xdr.c          |    2 +-
 include/linux/nfsd/state.h |    1 +
 include/linux/nfsd/xdr4.h  |    8 ++++-
 5 files changed, 77 insertions(+), 21 deletions(-)
fs/nfsd/nfs4proc.c +13 -4
index a273023..1d4b2b5 100644
@@ -179,7 +179,7 @@ nfsd4_open(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 	nfs4_lock_state();
 
 	/* check seqid for replay. set nfs4_owner */
-	status = nfsd4_process_open1(open);
+	status = nfsd4_process_open1(rqstp, open);
 	if (status == nfserr_replay_me) {
 		struct nfs4_replay *rp = &open->op_stateowner->so_replay;
 		fh_put(&cstate->current_fh);
@@ -504,6 +504,7 @@ nfsd4_read(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 	   struct nfsd4_read *read)
 {
 	__be32 status;
+	int flags = RD_STATE;
 
 	/* no need to check permission - this will be done in nfsd_read() */
 
@@ -511,11 +512,13 @@ nfsd4_read(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 	if (read->rd_offset >= OFFSET_MAX)
 		return nfserr_inval;
 
+	if (nfsd4_has_session(cstate))
+		flags |= HAS_SESSION;
 	nfs4_lock_state();
 	/* check stateid */
 	if ((status = nfs4_preprocess_stateid_op(&cstate->current_fh,
 				&read->rd_stateid,
-				RD_STATE, &read->rd_filp))) {
+				flags, &read->rd_filp))) {
 		dprintk("NFSD: nfsd4_read: couldn't process stateid!\n");
 		goto out;
 	}
@@ -643,11 +646,14 @@ nfsd4_setattr(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 	      struct nfsd4_setattr *setattr)
 {
 	__be32 status = nfs_ok;
+	int flags = WR_STATE;
 
+	if (nfsd4_has_session(cstate))
+		flags |= HAS_SESSION;
 	if (setattr->sa_iattr.ia_valid & ATTR_SIZE) {
 		nfs4_lock_state();
 		status = nfs4_preprocess_stateid_op(&cstate->current_fh,
-			&setattr->sa_stateid, WR_STATE, NULL);
+			&setattr->sa_stateid, flags, NULL);
 		nfs4_unlock_state();
 		if (status) {
 			dprintk("NFSD: nfsd4_setattr: couldn't process stateid!\n");
@@ -679,15 +685,18 @@ nfsd4_write(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 	u32 *p;
 	__be32 status = nfs_ok;
 	unsigned long cnt;
+	int flags = WR_STATE;
 
 	/* no need to check permission - this will be done in nfsd_write() */
 
 	if (write->wr_offset >= OFFSET_MAX)
 		return nfserr_inval;
 
+	if (nfsd4_has_session(cstate))
+		flags |= HAS_SESSION;
 	nfs4_lock_state();
 	status = nfs4_preprocess_stateid_op(&cstate->current_fh, stateid,
-					WR_STATE, &filp);
+					    flags, &filp);
 	if (filp)
 		get_file(filp);
 	nfs4_unlock_state();
fs/nfsd/nfs4state.c +55 -15

237 lines not on this page · raw message

fs/nfsd/nfs4xdr.c +1 -1
index 5720aab..a2682e8 100644
@@ -3206,7 +3206,7 @@ nfs4svc_encode_compoundres(struct svc_rqst *rqstp, __be32 *p, struct nfsd4_compo
 	iov->iov_len = ((char*)resp->p) - (char*)iov->iov_base;
 	BUG_ON(iov->iov_len > PAGE_SIZE);
 #ifdef CONFIG_NFSD_V4_1
-	if (resp->cstate.slot != NULL) {
+	if (nfsd4_has_session(&resp->cstate)) {
 		if (resp->cstate.status == nfserr_replay_cache &&
 				!nfsd4_no_page_in_cache(resp)) {
 			iov->iov_len = resp->cstate.iovlen;
include/linux/nfsd/state.h +1 -0
index 47c7836..302557d 100644
@@ -323,6 +323,7 @@ struct nfs4_stateid {
 };
 
 /* flags for preprocess_seqid_op() */
+#define HAS_SESSION             0x00000001
 #define CONFIRM                 0x00000002
 #define OPEN_STATE              0x00000004
 #define LOCK_STATE              0x00000008
include/linux/nfsd/xdr4.h +7 -1
index 37a7c51..aafbfdc 100644
@@ -55,6 +55,11 @@ struct nfsd4_compound_state {
 	u32			status;
 };
 
+static inline bool nfsd4_has_session(struct nfsd4_compound_state *cs)
+{
+	return cs->slot != NULL;
+}
+
 struct nfsd4_change_info {
 	u32		atomic;
 	u32		before_ctime_sec;
@@ -540,7 +545,8 @@ extern __be32 nfsd4_destroy_session(struct svc_rqst *,
 		struct nfsd4_compound_state *,
 		struct nfsd4_destroy_session *);
 #endif /* CONFIG_NFSD_V4_1 */
-extern __be32 nfsd4_process_open1(struct nfsd4_open *open);
+extern __be32 nfsd4_process_open1(struct svc_rqst *rqstp,
+		struct nfsd4_open *open);
 extern __be32 nfsd4_process_open2(struct svc_rqst *rqstp,
 		struct svc_fh *current_fh, struct nfsd4_open *open);
 extern __be32 nfsd4_open_confirm(struct svc_rqst *rqstp,
Signature · 1 line
1.6.2.1
#41 lore raw

PATCH 29/47 nfsd41: check encode size for sessions maxresponse cached

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

Calculate the space the compound response has taken after encoding the current operation.

pad: add on 8 bytes for the next operation's op_code and status so that there is room to cache a failure on the next operation.

Compare this length to the session se_fmaxresp_cached and return nfserr_rep_too_big_to_cache if the length is too large.

Our se_fmaxresp_cached will always be a multiple of PAGE_SIZE, and so will be at least a page and will therefore hold the xdr_buf head.

[nfsd41: non-page DRC for solo sequence responses]
[fixed nfsd4_check_drc_limit cosmetics]
---
 fs/nfsd/nfs4xdr.c |   58 +++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 files changed, 58 insertions(+), 0 deletions(-)
fs/nfsd/nfs4xdr.c +58 -0
index a2682e8..52ca833 100644
@@ -3089,6 +3089,61 @@ static nfsd4_enc nfsd4_enc_ops[] = {
 #endif /* CONFIG_NFSD_V4_1 */
 };
 
+#if defined(CONFIG_NFSD_V4_1)
+/*
+ * Calculate the total amount of memory that the compound response has taken
+ * after encoding the current operation.
+ *
+ * pad: add on 8 bytes for the next operation's op_code and status so that
+ * there is room to cache a failure on the next operation.
+ *
+ * Compare this length to the session se_fmaxresp_cached.
+ *
+ * Our se_fmaxresp_cached will always be a multiple of PAGE_SIZE, and so
+ * will be at least a page and will therefore hold the xdr_buf head.
+ */
+static int nfsd4_check_drc_limit(struct nfsd4_compoundres *resp)
+{
+	int status = 0;
+	struct xdr_buf *xb = &resp->rqstp->rq_res;
+	struct nfsd4_compoundargs *args = resp->rqstp->rq_argp;
+	struct nfsd4_session *session = NULL;
+	struct nfsd4_slot *slot = resp->cstate.slot;
+	u32 length, tlen = 0, pad = 8;
+
+	if (!nfsd4_has_session(&resp->cstate))
+		return status;
+
+	session = slot->sl_session;
+	if (session == NULL || slot->sl_cache_entry.ce_cachethis == 0)
+		return status;
+
+	if (resp->opcnt >= args->opcnt)
+		pad = 0; /* this is the last operation */
+
+	if (xb->page_len == 0) {
+		length = (char *)resp->p - (char *)xb->head[0].iov_base + pad;
+	} else {
+		if (xb->tail[0].iov_base && xb->tail[0].iov_len > 0)
+			tlen = (char *)resp->p - (char *)xb->tail[0].iov_base;
+
+		length = xb->head[0].iov_len + xb->page_len + tlen + pad;
+	}
+	dprintk("%s length %u, xb->page_len %u tlen %u pad %u\n", __func__,
+		length, xb->page_len, tlen, pad);
+
+	if (length <= session->se_fmaxresp_cached)
+		return status;
+	else
+		return nfserr_rep_too_big_to_cache;
+}
+#else /* CONFIG_NFSD_V4_1 */
+static inline int nfsd4_check_drc_limit(struct nfsd4_compoundres *resp)
+{
+	return 0;
+}
+#endif /* CONFIG_NFSD_V4_1 */
+
 void
 nfsd4_encode_operation(struct nfsd4_compoundres *resp, struct nfsd4_op *op)
 {
@@ -3105,6 +3160,9 @@ nfsd4_encode_operation(struct nfsd4_compoundres *resp, struct nfsd4_op *op)
 	BUG_ON(op->opnum < 0 || op->opnum >= ARRAY_SIZE(nfsd4_enc_ops) ||
 	       !nfsd4_enc_ops[op->opnum]);
 	op->status = nfsd4_enc_ops[op->opnum](resp, op->status, &op->u);
+	/* nfsd4_check_drc_limit guarantees enough room for error status */
+	if (!op->status && nfsd4_check_drc_limit(resp))
+		op->status = nfserr_rep_too_big_to_cache;
 status:
 	/*
 	 * Note: We write the status directly, instead of using WRITE32(),
Signature · 1 line
1.6.2.1
#42 lore raw

PATCH 30/47 nfsd41: clientid handling

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

Extract the clientid from sessionid to set the op_clientid on open. Verify that the clid for other stateful ops is zero for minorversion != 0 Do all other checks for stateful ops without sessions.

[fixed whitespace indent]
---
 fs/nfsd/nfs4proc.c  |   12 ++++++++++++
 fs/nfsd/nfs4state.c |   14 ++++++++------
 fs/nfsd/nfs4xdr.c   |    9 +++++++++
 3 files changed, 29 insertions(+), 6 deletions(-)
fs/nfsd/nfs4proc.c +12 -0
index 1d4b2b5..0076bb1 100644
@@ -162,6 +162,15 @@ do_open_fhandle(struct svc_rqst *rqstp, struct svc_fh *current_fh, struct nfsd4_
 	return status;
 }
 
+static void
+copy_clientid(clientid_t *clid, struct nfsd4_session *session)
+{
+	struct nfsd4_sessionid *sid =
+			(struct nfsd4_sessionid *)session->se_sessionid.data;
+
+	clid->cl_boot = sid->clientid.cl_boot;
+	clid->cl_id = sid->clientid.cl_id;
+}
 
 static __be32
 nfsd4_open(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
@@ -176,6 +185,9 @@ nfsd4_open(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 	if (open->op_create && open->op_claim_type != NFS4_OPEN_CLAIM_NULL)
 		return nfserr_inval;
 
+	if (nfsd4_has_session(cstate))
+		copy_clientid(&open->op_clientid, cstate->slot->sl_session);
+
 	nfs4_lock_state();
 
 	/* check seqid for replay. set nfs4_owner */
fs/nfsd/nfs4state.c +8 -6
index bf5b214..00ad1c2 100644
@@ -616,8 +616,8 @@ STALE_CLIENTID(clientid_t *clid)
 {
 	if (clid->cl_boot == boot_time)
 		return 0;
-	dprintk("NFSD stale clientid (%08x/%08x)\n", 
-			clid->cl_boot, clid->cl_id);
+	dprintk("NFSD stale clientid (%08x/%08x) boot_time %08lx\n",
+		clid->cl_boot, clid->cl_id, boot_time);
 	return 1;
 }
 
@@ -2968,8 +2968,9 @@ nfs4_preprocess_seqid_op(struct svc_fh *current_fh, u32 seqid, stateid_t *statei
 		if (lock->lk_is_new) {
 			if (!sop->so_is_open_owner)
 				return nfserr_bad_stateid;
-			if (!same_clid(&clp->cl_clientid, lockclid))
-			       return nfserr_bad_stateid;
+			if (!(flags & HAS_SESSION) &&
+			    !same_clid(&clp->cl_clientid, lockclid))
+				return nfserr_bad_stateid;
 			/* stp is the open stateid */
 			status = nfs4_check_openmode(stp, lkflg);
 			if (status)
@@ -3516,7 +3517,8 @@ nfsd4_lock(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 		struct nfs4_file *fp;
 		
 		status = nfserr_stale_clientid;
-		if (STALE_CLIENTID(&lock->lk_new_clientid))
+		if (!nfsd4_has_session(cstate) &&
+		    STALE_CLIENTID(&lock->lk_new_clientid))
 			goto out;
 
 		flags = OPEN_STATE;
@@ -3678,7 +3680,7 @@ nfsd4_lockt(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 	nfs4_lock_state();
 
 	status = nfserr_stale_clientid;
-	if (STALE_CLIENTID(&lockt->lt_clientid))
+	if (!nfsd4_has_session(cstate) && STALE_CLIENTID(&lockt->lt_clientid))
 		goto out;
 
 	if ((status = fh_verify(rqstp, &cstate->current_fh, S_IFREG, 0))) {
fs/nfsd/nfs4xdr.c +9 -0
index 52ca833..a1af3a3 100644
@@ -189,6 +189,11 @@ static __be32 *read_buf(struct nfsd4_compoundargs *argp, u32 nbytes)
 	return p;
 }
 
+static int zero_clientid(clientid_t *clid)
+{
+	return (clid->cl_boot == 0) && (clid->cl_id == 0);
+}
+
 static int
 defer_free(struct nfsd4_compoundargs *argp,
 		void (*release)(const void *), void *p)
@@ -584,6 +589,8 @@ nfsd4_decode_lockt(struct nfsd4_compoundargs *argp, struct nfsd4_lockt *lockt)
 	READ_BUF(lockt->lt_owner.len);
 	READMEM(lockt->lt_owner.data, lockt->lt_owner.len);
 
+	if (argp->minorversion && !zero_clientid(&lockt->lt_clientid))
+		return nfserr_inval;
 	DECODE_TAIL;
 }
 
@@ -994,6 +1001,8 @@ nfsd4_decode_release_lockowner(struct nfsd4_compoundargs *argp, struct nfsd4_rel
 	READ_BUF(rlockowner->rl_owner.len);
 	READMEM(rlockowner->rl_owner.data, rlockowner->rl_owner.len);
 
+	if (argp->minorversion && !zero_clientid(&rlockowner->rl_clientid))
+		return nfserr_inval;
 	DECODE_TAIL;
 }
 
Signature · 1 line
1.6.2.1
#43 lore raw

PATCH 31/47 nfsd41: access_valid

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

For nfs41, the open share flags are used also for delegation "wants" and "signals". Check that they are valid.

---
 fs/nfsd/nfs4proc.c        |    1 +
 fs/nfsd/nfs4state.c       |   20 +++++++++++++++-----
 include/linux/nfsd/xdr4.h |    1 +
 3 files changed, 17 insertions(+), 5 deletions(-)
fs/nfsd/nfs4proc.c +1 -0
index 0076bb1..8b4bf5e 100644
@@ -901,6 +901,7 @@ nfsd4_proc_compound(struct svc_rqst *rqstp,
 	resp->tag = args->tag;
 	resp->opcnt = 0;
 	resp->rqstp = rqstp;
+	resp->cstate.minorversion = args->minorversion;
 	resp->cstate.replay_owner = NULL;
 	fh_init(&resp->cstate.current_fh, NFS4_FHSIZE);
 	fh_init(&resp->cstate.save_fh, NFS4_FHSIZE);
fs/nfsd/nfs4state.c +15 -5
index 00ad1c2..dbf0d07 100644
@@ -1959,11 +1959,21 @@ find_file(struct inode *ino)
 	return NULL;
 }
 
-static inline int access_valid(u32 x)
+static inline int access_valid(u32 x, u32 minorversion)
 {
-	if (x < NFS4_SHARE_ACCESS_READ)
+	if ((x & NFS4_SHARE_ACCESS_MASK) < NFS4_SHARE_ACCESS_READ)
 		return 0;
-	if (x > NFS4_SHARE_ACCESS_BOTH)
+	if ((x & NFS4_SHARE_ACCESS_MASK) > NFS4_SHARE_ACCESS_BOTH)
+		return 0;
+	x &= ~NFS4_SHARE_ACCESS_MASK;
+	if (minorversion && x) {
+		if ((x & NFS4_SHARE_WANT_MASK) > NFS4_SHARE_WANT_CANCEL)
+			return 0;
+		if ((x & NFS4_SHARE_WHEN_MASK) > NFS4_SHARE_PUSH_DELEG_WHEN_UNCONTENDED)
+			return 0;
+		x &= ~(NFS4_SHARE_WANT_MASK | NFS4_SHARE_WHEN_MASK);
+	}
+	if (x)
 		return 0;
 	return 1;
 }
@@ -2511,7 +2521,7 @@ nfsd4_process_open2(struct svc_rqst *rqstp, struct svc_fh *current_fh, struct nf
 	__be32 status;
 
 	status = nfserr_inval;
-	if (!access_valid(open->op_share_access)
+	if (!access_valid(open->op_share_access, resp->cstate.minorversion)
 			|| !deny_valid(open->op_share_deny))
 		goto out;
 	/*
@@ -3108,7 +3118,7 @@ nfsd4_open_downgrade(struct svc_rqst *rqstp,
 			(int)cstate->current_fh.fh_dentry->d_name.len,
 			cstate->current_fh.fh_dentry->d_name.name);
 
-	if (!access_valid(od->od_share_access)
+	if (!access_valid(od->od_share_access, cstate->minorversion)
 			|| !deny_valid(od->od_share_deny))
 		return nfserr_inval;
 
include/linux/nfsd/xdr4.h +1 -0
index aafbfdc..881ff67 100644
@@ -52,6 +52,7 @@ struct nfsd4_compound_state {
 	struct nfsd4_slot	*slot;
 	__be32			*statp;
 	size_t			iovlen;
+	u32			minorversion;
 	u32			status;
 };
 
Signature · 1 line
1.6.2.1
#44 lore raw

PATCH 32/47 nfsd41: add OPEN4_SHARE_ACCESS_WANT nfs4_stateid bmap

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

Separate the access bits from the want bits and enable __set_bit to work correctly with st_access_bmap.

---
 fs/nfsd/nfs4state.c |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)
fs/nfsd/nfs4state.c +2 -1
index dbf0d07..8a64981 100644
@@ -1905,7 +1905,8 @@ init_stateid(struct nfs4_stateid *stp, struct nfs4_file *fp, struct nfsd4_open *
 	stp->st_stateid.si_generation = 0;
 	stp->st_access_bmap = 0;
 	stp->st_deny_bmap = 0;
-	__set_bit(open->op_share_access, &stp->st_access_bmap);
+	__set_bit(open->op_share_access & ~NFS4_SHARE_WANT_MASK,
+		  &stp->st_access_bmap);
 	__set_bit(open->op_share_deny, &stp->st_deny_bmap);
 	stp->st_openstp = NULL;
 }
Signature · 1 line
1.6.2.1
#45 lore raw

PATCH 33/47 nfsd41: provide support for minor version 1 at rpc level

In reply to Benny Halevy

From: Marc Eshel <[email protected]>

---
 include/linux/nfsd/nfsd.h |    4 ++++
 1 files changed, 4 insertions(+), 0 deletions(-)
include/linux/nfsd/nfsd.h +4 -0
index 3bccec3..d0fe9b4 100644
@@ -23,7 +23,11 @@
 /*
  * nfsd version
  */
+#if defined(CONFIG_NFSD_V4_1)
+#define NFSD_SUPPORTED_MINOR_VERSION	1
+#else /* CONFIG_NFSD_V4_1 */
 #define NFSD_SUPPORTED_MINOR_VERSION	0
+#endif /* CONFIG_NFSD_V4_1 */
 
 /*
  * Flags for nfsd_permission
Signature · 1 line
1.6.2.1
#46 lore raw

PATCH 34/47 nfsd: cleanup nfs4.0 callback encode routines

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

Mimic the client and prepare to share the back channel xdr with NFSv4.1. Bump the number of operations in each encode routine, then backfill the number of operations.

---
 fs/nfsd/nfs4callback.c |   24 ++++++++++++++++--------
 1 files changed, 16 insertions(+), 8 deletions(-)
fs/nfsd/nfs4callback.c +16 -8
index 290289b..7129b0c 100644
@@ -140,8 +140,9 @@ struct nfs4_cb_compound_hdr {
 	int		status;
 	u32		ident;
 	u32		nops;
+	__be32		*nops_p;
 	u32		taglen;
-	char *		tag;
+	char		*tag;
 };
 
 static struct {
@@ -201,7 +202,7 @@ nfs_cb_stat_to_errno(int stat)
  * XDR encode
  */
 
-static int
+static void
 encode_cb_compound_hdr(struct xdr_stream *xdr, struct nfs4_cb_compound_hdr *hdr)
 {
 	__be32 * p;
@@ -210,12 +211,18 @@ encode_cb_compound_hdr(struct xdr_stream *xdr, struct nfs4_cb_compound_hdr *hdr)
 	WRITE32(0);            /* tag length is always 0 */
 	WRITE32(NFS4_MINOR_VERSION);
 	WRITE32(hdr->ident);
+	hdr->nops_p = p;
 	WRITE32(hdr->nops);
-	return 0;
 }
 
-static int
-encode_cb_recall(struct xdr_stream *xdr, struct nfs4_cb_recall *cb_rec)
+static void encode_cb_nops(struct nfs4_cb_compound_hdr *hdr)
+{
+	*hdr->nops_p = htonl(hdr->nops);
+}
+
+static void
+encode_cb_recall(struct xdr_stream *xdr, struct nfs4_cb_recall *cb_rec,
+		 struct nfs4_cb_compound_hdr *hdr)
 {
 	__be32 *p;
 	int len = cb_rec->cbr_fh.fh_size;
@@ -227,7 +234,7 @@ encode_cb_recall(struct xdr_stream *xdr, struct nfs4_cb_recall *cb_rec)
 	WRITE32(cb_rec->cbr_trunc);
 	WRITE32(len);
 	WRITEMEM(&cb_rec->cbr_fh.fh_base, len);
-	return 0;
+	hdr->nops++;
 }
 
 static int
@@ -246,12 +253,13 @@ nfs4_xdr_enc_cb_recall(struct rpc_rqst *req, __be32 *p, struct nfs4_cb_recall *a
 	struct xdr_stream xdr;
 	struct nfs4_cb_compound_hdr hdr = {
 		.ident = args->cbr_ident,
-		.nops   = 1,
 	};
 
 	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
 	encode_cb_compound_hdr(&xdr, &hdr);
-	return (encode_cb_recall(&xdr, args));
+	encode_cb_recall(&xdr, args, &hdr);
+	encode_cb_nops(&hdr);
+	return 0;
 }
 
 
Signature · 1 line
1.6.2.1
#47 lore raw

PATCH 35/47 nfsd: minorversion support for the back channel

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

Prepare to share backchannel code with NFSv4.1.

---
 fs/nfsd/nfs4callback.c     |    7 ++++++-
 fs/nfsd/nfs4state.c        |    1 +
 include/linux/nfsd/state.h |    3 ++-
 3 files changed, 9 insertions(+), 2 deletions(-)
fs/nfsd/nfs4callback.c +6 -1
index 7129b0c..d8f2631 100644
@@ -141,6 +141,7 @@ struct nfs4_cb_compound_hdr {
 	u32		ident;
 	u32		nops;
 	__be32		*nops_p;
+	u32		minorversion;
 	u32		taglen;
 	char		*tag;
 };
@@ -209,7 +210,7 @@ encode_cb_compound_hdr(struct xdr_stream *xdr, struct nfs4_cb_compound_hdr *hdr)
 
 	RESERVE_SPACE(16);
 	WRITE32(0);            /* tag length is always 0 */
-	WRITE32(NFS4_MINOR_VERSION);
+	WRITE32(hdr->minorversion);
 	WRITE32(hdr->ident);
 	hdr->nops_p = p;
 	WRITE32(hdr->nops);
@@ -251,8 +252,11 @@ static int
 nfs4_xdr_enc_cb_recall(struct rpc_rqst *req, __be32 *p, struct nfs4_cb_recall *args)
 {
 	struct xdr_stream xdr;
+	struct nfs4_callback *cb =
+		(struct nfs4_callback *)req->rq_task->tk_client->cl_private;
 	struct nfs4_cb_compound_hdr hdr = {
 		.ident = args->cbr_ident,
+		.minorversion = cb->cb_minorversion,
 	};
 
 	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
@@ -436,6 +440,7 @@ static int do_probe_callback(void *data)
 		goto out_release_client;
 
 	cb->cb_client = client;
+	client->cl_private = cb;
 	atomic_set(&cb->cb_set, 1);
 	put_nfs4_client(clp);
 	return 0;
fs/nfsd/nfs4state.c +1 -0
index 8a64981..189f1c2 100644
@@ -994,6 +994,7 @@ gen_callback(struct nfs4_client *clp, struct nfsd4_setclientid *se)
 	if ( !(parse_ipv4(se->se_callback_addr_len, se->se_callback_addr_val,
 	                 &cb->cb_addr, &cb->cb_port)))
 		goto out_err;
+	cb->cb_minorversion = 0;
 	cb->cb_prog = se->se_callback_prog;
 	cb->cb_ident = se->se_callback_ident;
 	return;
include/linux/nfsd/state.h +2 -1
index 302557d..b78021a 100644
@@ -93,7 +93,8 @@ struct nfs4_callback {
 	u32                     cb_addr;
 	unsigned short          cb_port;
 	u32                     cb_prog;
-	u32                     cb_ident;
+	u32			cb_minorversion;
+	u32                     cb_ident;	/* minorversion 0 only */
 	/* RPC client info */
 	atomic_t		cb_set;     /* successful CB_NULL call */
 	struct rpc_clnt *       cb_client;
Signature · 1 line
1.6.2.1
#48 lore raw

PATCH 36/47 nfsd41: sunrpc: Added rpc server-side backchannel handling

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

FIXME: bhalevy: write up commit message

When the call direction is a reply, copy the xid and call direction into the req->rq_private_buf.head[0].iov_base otherwise rpc_verify_header returns rpc_garbage.

---
 include/linux/sunrpc/clnt.h    |    1 +
 include/linux/sunrpc/svcsock.h |    1 +
 include/linux/sunrpc/xprt.h    |    4 +
 net/sunrpc/clnt.c              |    1 +
 net/sunrpc/svcsock.c           |   81 +++++++++++
 net/sunrpc/xprt.c              |   41 ++++++-
 net/sunrpc/xprtsock.c          |  286 +++++++++++++++++++++++++++++++++++++++-
 7 files changed, 405 insertions(+), 10 deletions(-)
include/linux/sunrpc/clnt.h +1 -0
index 928c547..0b32b00 100644
@@ -111,6 +111,7 @@ struct rpc_create_args {
 	rpc_authflavor_t	authflavor;
 	unsigned long		flags;
 	char			*client_name;
+	struct svc_sock		*bc_sock;	/* NFSv4.1 backchannel */
 };
 
 /* Values for "flags" field */
include/linux/sunrpc/svcsock.h +1 -0
index 483e103..e8fda10 100644
@@ -28,6 +28,7 @@ struct svc_sock {
 	/* private TCP part */
 	u32			sk_reclen;	/* length of record */
 	u32			sk_tcplen;	/* current read length */
+	struct rpc_xprt	       *sk_bc_xprt;	/* NFSv4.1 backchannel xprt */
 };
 
 /*
include/linux/sunrpc/xprt.h +4 -0
index 11fc71d..e9ffee3 100644
@@ -174,6 +174,9 @@ struct rpc_xprt {
 	spinlock_t		reserve_lock;	/* lock slot table */
 	u32			xid;		/* Next XID value to use */
 	struct rpc_task *	snd_task;	/* Task blocked in send */
+#if defined(CONFIG_NFSD_V4_1)
+	struct svc_sock		*bc_sock;	/* NFSv4.1 backchannel */
+#endif /* CONFIG_NFSD_V4_1 */
 	struct list_head	recv;
 
 	struct {
@@ -197,6 +200,7 @@ struct xprt_create {
 	struct sockaddr *	srcaddr;	/* optional local address */
 	struct sockaddr *	dstaddr;	/* remote peer address */
 	size_t			addrlen;
+	struct svc_sock		*bc_sock;	/* NFSv4.1 backchannel */
 };
 
 struct xprt_class {
net/sunrpc/clnt.c +1 -0
index 836f15c..0a8f2d0 100644
@@ -266,6 +266,7 @@ struct rpc_clnt *rpc_create(struct rpc_create_args *args)
 		.srcaddr = args->saddress,
 		.dstaddr = args->address,
 		.addrlen = args->addrsize,
+		.bc_sock = args->bc_sock,
 	};
 	char servername[48];
 
net/sunrpc/svcsock.c +81 -0

117 lines not on this page · raw message

net/sunrpc/xprt.c +38 -3

59 lines not on this page · raw message

net/sunrpc/xprtsock.c +279 -7

323 lines not on this page · raw message

Signature · 1 line
1.6.2.1
#49 lore raw

PATCH 37/47 nfsd41: callback infrastructure

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

Keep the xprt used for create_session in cl_cb_xprt. Mark cl_callback.cb_minorversion = 1 and remember the client provided cl_callback.cb_prog rpc program number. Use it to probe the callback path.

Define xdr sizes and code nfs4_cb_compound header to be able to send a null callback rpc.

---
 fs/nfsd/nfs4callback.c |   35 ++++++++++++++++++++++++++++++++---
 fs/nfsd/nfs4state.c    |   10 ++++++++++
 2 files changed, 42 insertions(+), 3 deletions(-)
fs/nfsd/nfs4callback.c +32 -3
index d8f2631..6b7ef87 100644
@@ -43,6 +43,7 @@
 #include <linux/sunrpc/xdr.h>
 #include <linux/sunrpc/svc.h>
 #include <linux/sunrpc/clnt.h>
+#include <linux/sunrpc/svcsock.h>
 #include <linux/nfsd/nfsd.h>
 #include <linux/nfsd/state.h>
 #include <linux/sunrpc/sched.h>
@@ -52,16 +53,19 @@
 
 #define NFSPROC4_CB_NULL 0
 #define NFSPROC4_CB_COMPOUND 1
+#define NFS4_STATEID_SIZE 16
 
 /* Index of predefined Linux callback client operations */
 
 enum {
-        NFSPROC4_CLNT_CB_NULL = 0,
+	NFSPROC4_CLNT_CB_NULL = 0,
 	NFSPROC4_CLNT_CB_RECALL,
+	NFSPROC4_CLNT_CB_SEQUENCE,
 };
 
 enum nfs_cb_opnum4 {
 	OP_CB_RECALL            = 4,
+	OP_CB_SEQUENCE          = 11,
 };
 
 #define NFS4_MAXTAGLEN		20
@@ -70,15 +74,22 @@ enum nfs_cb_opnum4 {
 #define NFS4_dec_cb_null_sz		0
 #define cb_compound_enc_hdr_sz		4
 #define cb_compound_dec_hdr_sz		(3 + (NFS4_MAXTAGLEN >> 2))
+#define sessionid_sz			(NFS4_MAX_SESSIONID_LEN >> 2)
+#define cb_sequence_enc_sz		(sessionid_sz + 4 +             \
+					1 /* no referring calls list yet */)
+#define cb_sequence_dec_sz		(op_dec_sz + sessionid_sz + 4)
+
 #define op_enc_sz			1
 #define op_dec_sz			2
 #define enc_nfs4_fh_sz			(1 + (NFS4_FHSIZE >> 2))
 #define enc_stateid_sz			(NFS4_STATEID_SIZE >> 2)
 #define NFS4_enc_cb_recall_sz		(cb_compound_enc_hdr_sz +       \
+					cb_sequence_enc_sz +            \
 					1 + enc_stateid_sz +            \
 					enc_nfs4_fh_sz)
 
 #define NFS4_dec_cb_recall_sz		(cb_compound_dec_hdr_sz  +      \
+					cb_sequence_dec_sz +            \
 					op_dec_sz)
 
 /*
@@ -135,13 +146,19 @@ xdr_error:                                      \
 		return -EIO; \
 	} \
 } while (0)
+#define COPYMEM(x, nbytes) do {                \
+	memcpy((x), p, nbytes);                \
+	p += XDR_QUADLEN(nbytes);              \
+} while (0)
 
 struct nfs4_cb_compound_hdr {
-	int		status;
-	u32		ident;
+	/* args */
+	u32		ident;	/* minorversion 0 only */
 	u32		nops;
 	__be32		*nops_p;
 	u32		minorversion;
+	/* res */
+	int		status;
 	u32		taglen;
 	char		*tag;
 };
@@ -405,6 +422,17 @@ static struct rpc_clnt *setup_callback_client(struct nfs4_client *clp)
 	addr.sin_family = AF_INET;
 	addr.sin_port = htons(cb->cb_port);
 	addr.sin_addr.s_addr = htonl(cb->cb_addr);
+#if defined(CONFIG_NFSD_V4_1)
+	if (cb->cb_minorversion) {
+		BUG_ON(cb->cb_minorversion != 1);
+		args.bc_sock = container_of(clp->cl_cb_xprt, struct svc_sock,
+					    sk_xprt);
+	}
+#endif /* CONFIG_NFSD_V4_1 */
+
+	dprintk("%s: program %s 0x%x nrvers %u version %u minorversion %u\n",
+		__func__, args.program->name, args.prognumber,
+		args.program->nrvers, args.version, cb->cb_minorversion);
 
 	/* Create RPC client */
 	client = rpc_create(&args);
@@ -445,6 +473,7 @@ static int do_probe_callback(void *data)
 	put_nfs4_client(clp);
 	return 0;
 out_release_client:
+	dprintk("NFSD: synchronous CB_NULL failed. status=%d\n", status);
 	rpc_shutdown_client(client);
 out_err:
 	dprintk("NFSD: warning: no callback path to client %.*s: error %d\n",
fs/nfsd/nfs4state.c +10 -0
index 189f1c2..ef1e372 100644
@@ -663,6 +663,8 @@ free_client(struct nfs4_client *clp)
 {
 	shutdown_callback_client(clp);
 #if defined(CONFIG_NFSD_V4_1)
+	if (clp->cl_cb_xprt)
+		svc_xprt_put(clp->cl_cb_xprt);
 	nfsd4_release_respages(clp->cl_slot.sl_cache_entry.ce_respages,
 			     clp->cl_slot.sl_cache_entry.ce_resused);
 #endif /* CONFIG_NFSD_V4_1 */
@@ -1411,6 +1413,14 @@ nfsd4_create_session(struct svc_rqst *rqstp,
 		cr_ses->flags &= ~SESSION4_PERSIST;
 		cr_ses->flags &= ~SESSION4_RDMA;
 
+		if (cr_ses->flags & SESSION4_BACK_CHAN) {
+			unconf->cl_cb_xprt = rqstp->rq_xprt;
+			svc_xprt_get(unconf->cl_cb_xprt);
+			unconf->cl_callback.cb_minorversion = 1;
+			unconf->cl_cb_seq_nr = 0;
+			unconf->cl_callback.cb_prog = cr_ses->callback_prog;
+			nfsd4_probe_callback(unconf);
+		}
 		conf = unconf;
 	} else {
 		status = nfserr_stale_clientid;
Signature · 1 line
1.6.2.1
#50 lore raw

PATCH 38/47 nfsd41: Remember the auth flavor to use for callbacks

In reply to Benny Halevy

From: Ricardo Labiaga <[email protected]>

The callbacks will be sent using the same authentication flavor that was used during session creation. We'll add code to remember the principal in the case RPCSEC_GSS in a separate patch.

---
 fs/nfsd/nfs4state.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)
fs/nfsd/nfs4state.c +1 -0
index ef1e372..eb3d0dc 100644
@@ -1307,6 +1307,7 @@ out_new:
 	copy_verf(new, &verf);
 	copy_cred(&new->cl_cred, &rqstp->rq_cred);
 	new->cl_addr = ip_addr;
+	new->cl_flavor = rqstp->rq_flavor;
 	gen_clid(new);
 	gen_confirm(new);
 	add_to_unconfirmed(new, strhashval);
Signature · 1 line
1.6.2.1
#51 lore raw

PATCH 39/47 nfsd41: introduce cl_cb_mutex

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

To be used for serializing nfs41 callbacks.

---
 fs/nfsd/nfs4state.c        |    1 +
 include/linux/nfsd/state.h |    1 +
 2 files changed, 2 insertions(+), 0 deletions(-)
fs/nfsd/nfs4state.c +1 -0
index eb3d0dc..e5c7ddb 100644
@@ -741,6 +741,7 @@ static struct nfs4_client *create_client(struct xdr_netobj name, char *recdir)
 	INIT_LIST_HEAD(&clp->cl_delegations);
 #if defined(CONFIG_NFSD_V4_1)
 	INIT_LIST_HEAD(&clp->cl_sessions);
+	mutex_init(&clp->cl_cb_mutex);
 #endif /* CONFIG_NFSD_V4_1 */
 	INIT_LIST_HEAD(&clp->cl_lru);
 	return clp;
include/linux/nfsd/state.h +1 -0
index b78021a..f2783d4 100644
@@ -197,6 +197,7 @@ struct nfs4_client {
 	struct nfs4_sessionid	cl_sessionid;
 
 	struct svc_xprt		*cl_cb_xprt;	/* 4.1 callback transport */
+	struct mutex		cl_cb_mutex;
 #endif /* CONFIG_NFSD_V4_1 */
 };
 
Signature · 1 line
1.6.2.1
#52 lore raw

PATCH 40/47 nfsd41: cb_sequence callback

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

Implement the cb_sequence callback conforming to draft-ietf-nfsv4-minorversion1

[Rework the back channel xdr using the shared v4.0 and v4.1 framework.]
[fixed indentation]
---
 fs/nfsd/nfs4callback.c     |  118 ++++++++++++++++++++++++++++++++++++++++++++
 include/linux/nfsd/state.h |    7 ++-
 2 files changed, 124 insertions(+), 1 deletions(-)
fs/nfsd/nfs4callback.c +118 -0
index 6b7ef87..7ada6b1 100644
@@ -255,6 +255,29 @@ encode_cb_recall(struct xdr_stream *xdr, struct nfs4_cb_recall *cb_rec,
 	hdr->nops++;
 }
 
+static void
+encode_cb_sequence(struct xdr_stream *xdr, struct nfsd4_cb_sequence *args,
+		   struct nfs4_cb_compound_hdr *hdr)
+{
+	__be32 *p;
+
+	if (hdr->minorversion == 0)
+		return;
+
+	RESERVE_SPACE(1 + NFS4_MAX_SESSIONID_LEN + 20);
+
+	WRITE32(OP_CB_SEQUENCE);
+#ifdef CONFIG_NFSD_V4_1
+	WRITEMEM(args->cbs_clp->cl_sessionid.data, NFS4_MAX_SESSIONID_LEN);
+	WRITE32(args->cbs_clp->cl_cb_seq_nr);
+#endif /* CONFIG_NFSD_V4_1 */
+	WRITE32(0);		/* slotid, always 0 */
+	WRITE32(0);		/* highest slotid always 0 */
+	WRITE32(0);		/* cachethis always 0 */
+	WRITE32(0); /* FIXME: support referring_call_lists */
+	hdr->nops++;
+}
+
 static int
 nfs4_xdr_enc_cb_null(struct rpc_rqst *req, __be32 *p)
 {
@@ -319,6 +342,69 @@ decode_cb_op_hdr(struct xdr_stream *xdr, enum nfs_opnum4 expected)
 	return 0;
 }
 
+/*
+ * Our current back channel implmentation supports a single backchannel
+ * with a single slot.
+ */
+static int
+decode_cb_sequence(struct xdr_stream *xdr, struct nfsd4_cb_sequence *res,
+		   struct rpc_rqst *rqstp)
+{
+	struct nfs4_callback *cb =
+		(struct nfs4_callback *)rqstp->rq_task->tk_client->cl_private;
+	struct nfs4_sessionid id;
+	int status;
+	u32 dummy;
+	__be32 *p;
+
+	if (cb->cb_minorversion == 0)
+		return 0;
+
+	status = decode_cb_op_hdr(xdr, OP_CB_SEQUENCE);
+	if (status)
+		return status;
+
+	/*
+	 * If the server returns different values for sessionID, slotID or
+	 * sequence number, the server is looney tunes.
+	 */
+	status = -ESERVERFAULT;
+
+	READ_BUF(NFS4_MAX_SESSIONID_LEN + 16);
+	COPYMEM(id.data, NFS4_MAX_SESSIONID_LEN);
+#ifdef CONFIG_NFSD_V4_1
+	if (memcmp(id.data, res->cbs_clp->cl_sessionid.data,
+		   NFS4_MAX_SESSIONID_LEN)) {
+		dprintk("%s Invalid session id\n", __func__);
+		goto out;
+	}
+	READ32(dummy);
+	if (dummy != res->cbs_clp->cl_cb_seq_nr) {
+		dprintk("%s Invalid sequence number\n", __func__);
+		goto out;
+	}
+#endif /* CONFIG_NFSD_V4_1 */
+	READ32(dummy); 	/* slotid must be 0 */
+	if (dummy != 0) {
+		dprintk("%s Invalid slotid\n", __func__);
+		goto out;
+	}
+	READ32(dummy); 	/* highest slotid must be 0 */
+	if (dummy != 0) {
+		dprintk("%s Invalid highest slotid\n", __func__);
+		goto out;
+	}
+	READ32(dummy); 	/* target highest slotid must be 0 */
+	if (dummy != 0) {
+		dprintk("%s Invalid target highest slotid\n", __func__);
+		goto out;
+	}
+	status = 0;
+out:
+	return status;
+}
+
+
 static int
 nfs4_xdr_dec_cb_null(struct rpc_rqst *req, __be32 *p)
 {
@@ -503,6 +589,38 @@ nfsd4_probe_callback(struct nfs4_client *clp)
 	return;
 }
 
+#if defined(CONFIG_NFSD_V4_1)
+/*
+ * FIXME: cb_sequence should support referring call lists, cachethis, and
+ * multiple slots
+ */
+static int
+nfs41_cb_sequence_setup(struct nfs4_client *clp, struct nfsd4_cb_sequence *args)
+{
+	u32 *ptr = (u32 *)clp->cl_sessionid.data;
+
+	dprintk("%s: %u:%u:%u:%u\n", __func__,
+		ptr[0], ptr[1], ptr[2], ptr[3]);
+
+	mutex_lock(&clp->cl_cb_mutex);
+	args->cbs_clp = clp;
+	clp->cl_cb_seq_nr++;
+	return 0;
+}
+
+static void
+nfs41_cb_sequence_done(struct nfs4_client *clp, struct nfsd4_cb_sequence *res)
+{
+	u32 *ptr = (u32 *)clp->cl_sessionid.data;
+
+	dprintk("%s: %u:%u:%u:%u\n", __func__,
+		ptr[0], ptr[1], ptr[2], ptr[3]);
+
+	/* FIXME: support multiple callback slots */
+	mutex_unlock(&clp->cl_cb_mutex);
+}
+#endif /* CONFIG_NFSD_V4_1 */
+
 /*
  * called with dp->dl_count inc'ed.
  */
include/linux/nfsd/state.h +6 -1
index f2783d4..0ea8c89 100644
@@ -61,6 +61,10 @@ typedef struct {
 #define si_stateownerid   si_opaque.so_stateownerid
 #define si_fileid         si_opaque.so_fileid
 
+struct nfsd4_cb_sequence {
+	/* args/res */
+	struct nfs4_client	*cbs_clp;
+};
 
 struct nfs4_cb_recall {
 	u32			cbr_ident;
@@ -195,7 +199,8 @@ struct nfs4_client {
 	struct nfsd4_slot	cl_slot;	/* create_session slot */
 	u32			cl_exchange_flags;
 	struct nfs4_sessionid	cl_sessionid;
-
+	/* We currently support a single back channel with a single slot */
+	u32			cl_cb_seq_nr;
 	struct svc_xprt		*cl_cb_xprt;	/* 4.1 callback transport */
 	struct mutex		cl_cb_mutex;
 #endif /* CONFIG_NFSD_V4_1 */
Signature · 1 line
1.6.2.1
#53 lore raw

PATCH 41/47 nfsd41: introduce nfs4_cb_call_sync for nfs4 and nfs41

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

For nfs41 callbacks we need to setup the cb_sequence args and process the result. This will be called by code common to nfs4 and nfs41.

---
 fs/nfsd/nfs4callback.c |   50 ++++++++++++++++++++++++++++++++++++++++++++++++
 1 files changed, 50 insertions(+), 0 deletions(-)
fs/nfsd/nfs4callback.c +50 -0
index 7ada6b1..dd3b739 100644
@@ -92,6 +92,15 @@ enum nfs_cb_opnum4 {
 					cb_sequence_dec_sz +            \
 					op_dec_sz)
 
+struct nfs4_rpc_args {
+	void                     *args_op;
+	struct nfsd4_cb_sequence *args_seq;
+};
+
+struct nfs4_rpc_res {
+	struct nfsd4_cb_sequence *res_seq;
+};
+
 /*
 * Generic encode routines from fs/nfs/nfs4xdr.c
 */
@@ -589,6 +598,12 @@ nfsd4_probe_callback(struct nfs4_client *clp)
 	return;
 }
 
+static int _nfsd4_cb_sync(struct nfs4_client *clp,
+			  const struct rpc_message *msg, int flags)
+{
+	return rpc_call_sync(clp->cl_callback.cb_client, msg, RPC_TASK_SOFT);
+}
+
 #if defined(CONFIG_NFSD_V4_1)
 /*
  * FIXME: cb_sequence should support referring call lists, cachethis, and
@@ -619,6 +634,41 @@ nfs41_cb_sequence_done(struct nfs4_client *clp, struct nfsd4_cb_sequence *res)
 	/* FIXME: support multiple callback slots */
 	mutex_unlock(&clp->cl_cb_mutex);
 }
+
+static int _nfsd41_cb_sync(struct nfs4_client *clp,
+			   struct rpc_message *msg, int flags)
+{
+	struct nfsd4_cb_sequence seq;
+	struct nfs4_rpc_args *args;
+	struct nfs4_rpc_res res;
+	int status;
+
+	args = msg->rpc_argp;
+	args->args_seq = &seq;
+
+	res.res_seq = &seq;
+	msg->rpc_resp = &res;
+
+	nfs41_cb_sequence_setup(clp, &seq);
+	status = _nfsd4_cb_sync(clp, msg, flags);
+	nfs41_cb_sequence_done(clp, &seq);
+
+	return status;
+}
+
+static int nfsd4_cb_sync(struct nfs4_client *clp,
+			 struct rpc_message *msg, int flags)
+{
+	return clp->cl_callback.cb_minorversion ?
+		_nfsd41_cb_sync(clp, msg, flags) :
+		_nfsd4_cb_sync(clp, msg, flags);
+}
+#else  /* CONFIG_NFSD_V4_1 */
+static int nfsd4_cb_sync(struct nfs4_client *clp,
+			 struct rpc_message *msg, int flags)
+{
+	return _nfsd4_cb_sync(clp, msg, flags);
+}
 #endif /* CONFIG_NFSD_V4_1 */
 
 /*
Signature · 1 line
1.6.2.1
#54 lore raw

PATCH 42/47 nfsd41: cb_recall callback

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

Implement the cb_recall callback conforming to
http://tools.ietf.org/html/draft-ietf-nfsv4-minorversion1-26

[Share v4.0 and v4.1 back channel xdr]
---
 fs/nfsd/nfs4callback.c |   25 +++++++++++++++++++------
 1 files changed, 19 insertions(+), 6 deletions(-)
fs/nfsd/nfs4callback.c +19 -6
index dd3b739..0be75e0 100644
@@ -298,9 +298,11 @@ nfs4_xdr_enc_cb_null(struct rpc_rqst *req, __be32 *p)
 }
 
 static int
-nfs4_xdr_enc_cb_recall(struct rpc_rqst *req, __be32 *p, struct nfs4_cb_recall *args)
+nfs4_xdr_enc_cb_recall(struct rpc_rqst *req, __be32 *p,
+		       struct nfs4_rpc_args *rpc_args)
 {
 	struct xdr_stream xdr;
+	struct nfs4_cb_recall *args = rpc_args->args_op;
 	struct nfs4_callback *cb =
 		(struct nfs4_callback *)req->rq_task->tk_client->cl_private;
 	struct nfs4_cb_compound_hdr hdr = {
@@ -310,6 +312,7 @@ nfs4_xdr_enc_cb_recall(struct rpc_rqst *req, __be32 *p, struct nfs4_cb_recall *a
 
 	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
 	encode_cb_compound_hdr(&xdr, &hdr);
+	encode_cb_sequence(&xdr, rpc_args->args_seq, &hdr);
 	encode_cb_recall(&xdr, args, &hdr);
 	encode_cb_nops(&hdr);
 	return 0;
@@ -421,7 +424,8 @@ nfs4_xdr_dec_cb_null(struct rpc_rqst *req, __be32 *p)
 }
 
 static int
-nfs4_xdr_dec_cb_recall(struct rpc_rqst *rqstp, __be32 *p)
+nfs4_xdr_dec_cb_recall(struct rpc_rqst *rqstp, __be32 *p,
+		       struct nfs4_rpc_res *rpc_res)
 {
 	struct xdr_stream xdr;
 	struct nfs4_cb_compound_hdr hdr;
@@ -431,6 +435,9 @@ nfs4_xdr_dec_cb_recall(struct rpc_rqst *rqstp, __be32 *p)
 	status = decode_cb_compound_hdr(&xdr, &hdr);
 	if (status)
 		goto out;
+	status = decode_cb_sequence(&xdr, rpc_res->res_seq, rqstp);
+	if (status)
+		goto out;
 	status = decode_cb_op_hdr(&xdr, OP_CB_RECALL);
 out:
 	return status;
@@ -678,19 +685,23 @@ void
 nfsd4_cb_recall(struct nfs4_delegation *dp)
 {
 	struct nfs4_client *clp = dp->dl_client;
-	struct rpc_clnt *clnt = clp->cl_callback.cb_client;
 	struct nfs4_cb_recall *cbr = &dp->dl_recall;
+	struct nfs4_rpc_args args = {
+		.args_op = cbr,
+	};
 	struct rpc_message msg = {
 		.rpc_proc = &nfs4_cb_procedures[NFSPROC4_CLNT_CB_RECALL],
-		.rpc_argp = cbr,
+		.rpc_argp = &args,
 	};
 	int retries = 1;
 	int status = 0;
 
+	dprintk("%s: dp %p\n", __func__, dp);
+
 	cbr->cbr_trunc = 0; /* XXX need to implement truncate optimization */
 	cbr->cbr_dp = dp;
 
-	status = rpc_call_sync(clnt, &msg, RPC_TASK_SOFT);
+	status = nfsd4_cb_sync(clp, &msg, RPC_TASK_SOFT);
 	while (retries--) {
 		switch (status) {
 			case -EIO:
@@ -705,13 +716,15 @@ nfsd4_cb_recall(struct nfs4_delegation *dp)
 				goto out_put_cred;
 		}
 		ssleep(2);
-		status = rpc_call_sync(clnt, &msg, RPC_TASK_SOFT);
+		status = nfsd4_cb_sync(clp, &msg, RPC_TASK_SOFT);
 	}
 out_put_cred:
 	/*
 	 * Success or failure, now we're either waiting for lease expiration
 	 * or deleg_return.
 	 */
+	dprintk("%s: dp %p dl_flock %p dl_count %d\n",
+		__func__, dp, dp->dl_flock, atomic_read(&dp->dl_count));
 	put_nfs4_client(clp);
 	nfs4_put_delegation(dp);
 	return;
Signature · 1 line
1.6.2.1
#55 lore raw

PATCH 43/47 nfsd41: pass writable attrs mask to nfsd4_decode_fattr

In reply to Benny Halevy

In preparation for EXCLUSIVE4_1

---
 fs/nfsd/nfs4xdr.c |   21 +++++++++++++++------
 1 files changed, 15 insertions(+), 6 deletions(-)
fs/nfsd/nfs4xdr.c +15 -6
index a1af3a3..96f3eeb 100644
@@ -251,9 +251,14 @@ nfsd4_decode_bitmap(struct nfsd4_compoundargs *argp, u32 *bmval)
 	DECODE_TAIL;
 }
 
+static u32 nfsd_attrmask[] = {
+	NFSD_WRITEABLE_ATTRS_WORD0,
+	NFSD_WRITEABLE_ATTRS_WORD1
+};
+
 static __be32
-nfsd4_decode_fattr(struct nfsd4_compoundargs *argp, u32 *bmval, struct iattr *iattr,
-    struct nfs4_acl **acl)
+nfsd4_decode_fattr(struct nfsd4_compoundargs *argp, u32 *bmval, u32 *writable,
+		   struct iattr *iattr, struct nfs4_acl **acl)
 {
 	int expected_len, len = 0;
 	u32 dummy32;
@@ -271,7 +276,7 @@ nfsd4_decode_fattr(struct nfsd4_compoundargs *argp, u32 *bmval, struct iattr *ia
 	 */
 	if ((bmval[0] & ~NFSD_SUPPORTED_ATTRS_WORD0) || (bmval[1] & ~NFSD_SUPPORTED_ATTRS_WORD1))
 		return nfserr_attrnotsupp;
-	if ((bmval[0] & ~NFSD_WRITEABLE_ATTRS_WORD0) || (bmval[1] & ~NFSD_WRITEABLE_ATTRS_WORD1))
+	if ((bmval[0] & ~writable[0]) || (bmval[1] & ~writable[1]))
 		return nfserr_inval;
 
 	READ_BUF(4);
@@ -499,7 +504,9 @@ nfsd4_decode_create(struct nfsd4_compoundargs *argp, struct nfsd4_create *create
 	if ((status = check_filename(create->cr_name, create->cr_namelen, nfserr_inval)))
 		return status;
 
-	if ((status = nfsd4_decode_fattr(argp, create->cr_bmval, &create->cr_iattr, &create->cr_acl)))
+	status = nfsd4_decode_fattr(argp, create->cr_bmval, nfsd_attrmask,
+				    &create->cr_iattr, &create->cr_acl);
+	if (status)
 		goto out;
 
 	DECODE_TAIL;
@@ -660,7 +667,9 @@ nfsd4_decode_open(struct nfsd4_compoundargs *argp, struct nfsd4_open *open)
 		switch (open->op_createmode) {
 		case NFS4_CREATE_UNCHECKED:
 		case NFS4_CREATE_GUARDED:
-			if ((status = nfsd4_decode_fattr(argp, open->op_bmval, &open->op_iattr, &open->op_acl)))
+			status = nfsd4_decode_fattr(argp, open->op_bmval,
+				nfsd_attrmask, &open->op_iattr, &open->op_acl);
+			if (status)
 				goto out;
 			break;
 		case NFS4_CREATE_EXCLUSIVE:
@@ -859,7 +868,7 @@ nfsd4_decode_setattr(struct nfsd4_compoundargs *argp, struct nfsd4_setattr *seta
 	status = nfsd4_decode_stateid(argp, &setattr->sa_stateid);
 	if (status)
 		return status;
-	return nfsd4_decode_fattr(argp, setattr->sa_bmval,
+	return nfsd4_decode_fattr(argp, setattr->sa_bmval, nfsd_attrmask,
 				  &setattr->sa_iattr, &setattr->sa_acl);
 }
 
Signature · 1 line
1.6.2.1
#56 lore raw

PATCH 44/47 nfsd41: support for 3-word long attribute bitmask

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

Also, use client minorversion to generate supported attrs

---
 fs/nfsd/nfs4proc.c        |   15 +++++++----
 fs/nfsd/nfs4xdr.c         |   59 +++++++++++++++++++++++++++++++++++---------
 include/linux/nfsd/nfsd.h |   48 +++++++++++++++++++++++++++++++++++-
 include/linux/nfsd/xdr4.h |   12 ++++----
 4 files changed, 108 insertions(+), 26 deletions(-)
fs/nfsd/nfs4proc.c +9 -6
index 8b4bf5e..f3b560f 100644
@@ -460,8 +460,9 @@ nfsd4_getattr(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 	if (getattr->ga_bmval[1] & NFSD_WRITEONLY_ATTRS_WORD1)
 		return nfserr_inval;
 
-	getattr->ga_bmval[0] &= NFSD_SUPPORTED_ATTRS_WORD0;
-	getattr->ga_bmval[1] &= NFSD_SUPPORTED_ATTRS_WORD1;
+	getattr->ga_bmval[0] &= nfsd_suppattrs0(cstate->minorversion);
+	getattr->ga_bmval[1] &= nfsd_suppattrs1(cstate->minorversion);
+	getattr->ga_bmval[2] &= nfsd_suppattrs2(cstate->minorversion);
 
 	getattr->ga_fhp = &cstate->current_fh;
 	return nfs_ok;
@@ -556,8 +557,9 @@ nfsd4_readdir(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 	if (readdir->rd_bmval[1] & NFSD_WRITEONLY_ATTRS_WORD1)
 		return nfserr_inval;
 
-	readdir->rd_bmval[0] &= NFSD_SUPPORTED_ATTRS_WORD0;
-	readdir->rd_bmval[1] &= NFSD_SUPPORTED_ATTRS_WORD1;
+	readdir->rd_bmval[0] &= nfsd_suppattrs0(cstate->minorversion);
+	readdir->rd_bmval[1] &= nfsd_suppattrs1(cstate->minorversion);
+	readdir->rd_bmval[2] &= nfsd_suppattrs2(cstate->minorversion);
 
 	if ((cookie > ~(u32)0) || (cookie == 1) || (cookie == 2) ||
 	    (cookie == 0 && memcmp(readdir->rd_verf.data, zeroverf.data, NFS4_VERIFIER_SIZE)))
@@ -754,8 +756,9 @@ _nfsd4_verify(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 	if (status)
 		return status;
 
-	if ((verify->ve_bmval[0] & ~NFSD_SUPPORTED_ATTRS_WORD0)
-	    || (verify->ve_bmval[1] & ~NFSD_SUPPORTED_ATTRS_WORD1))
+	if ((verify->ve_bmval[0] & ~nfsd_suppattrs0(cstate->minorversion))
+	    || (verify->ve_bmval[1] & ~nfsd_suppattrs1(cstate->minorversion))
+	    || (verify->ve_bmval[2] & ~nfsd_suppattrs2(cstate->minorversion)))
 		return nfserr_attrnotsupp;
 	if ((verify->ve_bmval[0] & FATTR4_WORD0_RDATTR_ERROR)
 	    || (verify->ve_bmval[1] & NFSD_WRITEONLY_ATTRS_WORD1))
fs/nfsd/nfs4xdr.c +47 -12

127 lines not on this page · raw message

include/linux/nfsd/nfsd.h +46 -2

77 lines not on this page · raw message

include/linux/nfsd/xdr4.h +6 -6
index 881ff67..d7a8050 100644
@@ -101,7 +101,7 @@ struct nfsd4_create {
 			u32 specdata2;
 		} dev;    /* NF4BLK, NF4CHR */
 	} u;
-	u32		cr_bmval[2];        /* request */
+	u32		cr_bmval[3];        /* request */
 	struct iattr	cr_iattr;           /* request */
 	struct nfsd4_change_info  cr_cinfo; /* response */
 	struct nfs4_acl *cr_acl;
@@ -116,7 +116,7 @@ struct nfsd4_delegreturn {
 };
 
 struct nfsd4_getattr {
-	u32		ga_bmval[2];        /* request */
+	u32		ga_bmval[3];        /* request */
 	struct svc_fh	*ga_fhp;            /* response */
 };
 
@@ -217,7 +217,7 @@ struct nfsd4_open {
 	stateid_t       op_delegate_stateid; /* request - response */
 	u32		op_create;     	    /* request */
 	u32		op_createmode;      /* request */
-	u32		op_bmval[2];        /* request */
+	u32		op_bmval[3];        /* request */
 	union {                             /* request */
 		struct iattr	iattr;                      /* UNCHECKED4,GUARDED4 */
 		nfs4_verifier	verf;                                /* EXCLUSIVE4 */
@@ -270,7 +270,7 @@ struct nfsd4_readdir {
 	nfs4_verifier	rd_verf;            /* request */
 	u32		rd_dircount;        /* request */
 	u32		rd_maxcount;        /* request */
-	u32		rd_bmval[2];        /* request */
+	u32		rd_bmval[3];        /* request */
 	struct svc_rqst *rd_rqstp;          /* response */
 	struct svc_fh * rd_fhp;             /* response */
 
@@ -312,7 +312,7 @@ struct nfsd4_secinfo {
 
 struct nfsd4_setattr {
 	stateid_t	sa_stateid;         /* request */
-	u32		sa_bmval[2];        /* request */
+	u32		sa_bmval[3];        /* request */
 	struct iattr	sa_iattr;           /* request */
 	struct nfs4_acl *sa_acl;
 };
@@ -338,7 +338,7 @@ struct nfsd4_setclientid_confirm {
 
 /* also used for NVERIFY */
 struct nfsd4_verify {
-	u32		ve_bmval[2];        /* request */
+	u32		ve_bmval[3];        /* request */
 	u32		ve_attrlen;         /* request */
 	char *		ve_attrval;         /* request */
 };
Signature · 1 line
1.6.2.1
#57 lore raw

PATCH 45/47 nfsd41: SUPPATTR_EXCLCREAT attribute

In reply to Benny Halevy

Return bitmask for supported EXCLUSIVE4_1 create attributes.

---
 fs/nfsd/nfs4xdr.c         |    7 ++++++-
 include/linux/nfs4.h      |    2 ++
 include/linux/nfsd/nfsd.h |   14 +++++++++++++-
 3 files changed, 21 insertions(+), 2 deletions(-)
fs/nfsd/nfs4xdr.c +6 -1
index 079626c..288332c 100644
@@ -2154,7 +2154,12 @@ out_acl:
 		}
 		WRITE64(stat.ino);
 	}
-	BUG_ON(bmval2);	/* FIXME: not implemented yet */
+	if (bmval2 & FATTR4_WORD2_SUPPATTR_EXCLCREAT) {
+		WRITE32(3);
+		WRITE32(NFSD_SUPPATTR_EXCLCREAT_WORD0);
+		WRITE32(NFSD_SUPPATTR_EXCLCREAT_WORD1);
+		WRITE32(NFSD_SUPPATTR_EXCLCREAT_WORD2);
+	}
 
 	*attrlenp = htonl((char *)p - (char *)attrlenp - 4);
 	*countp = p - buffer;
include/linux/nfs4.h +2 -0
index a34683e..735c7d8 100644
@@ -404,6 +404,8 @@ enum lock_type4 {
 #define FATTR4_WORD0_UNIQUE_HANDLES     (1UL << 9)
 #define FATTR4_WORD0_LEASE_TIME         (1UL << 10)
 #define FATTR4_WORD0_RDATTR_ERROR       (1UL << 11)
+/* Mandatory in NFSv4.1 */
+#define FATTR4_WORD2_SUPPATTR_EXCLCREAT (1UL << 11)
 
 /* Recommended Attributes */
 #define FATTR4_WORD0_ACL                (1UL << 12)
include/linux/nfsd/nfsd.h +13 -1
index 3dccb76..b9e6682 100644
@@ -374,7 +374,7 @@ extern struct timeval	nfssvc_boot;
 	NFSD4_SUPPORTED_ATTRS_WORD1
 
 #define NFSD4_1_SUPPORTED_ATTRS_WORD2 \
-	NFSD4_SUPPORTED_ATTRS_WORD2
+	(NFSD4_SUPPORTED_ATTRS_WORD2 | FATTR4_WORD2_SUPPATTR_EXCLCREAT)
 
 #if defined(CONFIG_NFSD_V4_1)
 
@@ -427,6 +427,18 @@ static inline u32 nfsd_suppattrs2(u32 minorversion)
  | FATTR4_WORD1_TIME_ACCESS_SET | FATTR4_WORD1_TIME_MODIFY_SET)
 #define NFSD_WRITEABLE_ATTRS_WORD2 0
 
+#define NFSD_SUPPATTR_EXCLCREAT_WORD0 \
+	NFSD_WRITEABLE_ATTRS_WORD0
+/*
+ * we currently store the exclusive create verifier in the v_{a,m}time
+ * attributes so the client can't set these at create time using EXCLUSIVE4_1
+ */
+#define NFSD_SUPPATTR_EXCLCREAT_WORD1 \
+	(NFSD_WRITEABLE_ATTRS_WORD1 & \
+	 ~(FATTR4_WORD1_TIME_ACCESS_SET | FATTR4_WORD1_TIME_MODIFY_SET))
+#define NFSD_SUPPATTR_EXCLCREAT_WORD2 \
+	NFSD_WRITEABLE_ATTRS_WORD2
+
 #endif /* CONFIG_NFSD_V4 */
 
 #endif /* LINUX_NFSD_NFSD_H */
Signature · 1 line
1.6.2.1
#58 lore raw

PATCH 46/47 nfsd41: CREATE_EXCLUSIVE4_1

In reply to Benny Halevy

Implement the CREATE_EXCLUSIVE4_1 open mode conforming to
http://tools.ietf.org/html/draft-ietf-nfsv4-minorversion1-26

This mode allows the client to atomically create a file if it doesn't exist while setting some of its attributes.

It must be implemented if the server supports persistent reply cache and/or pnfs.

---
 fs/nfsd/nfs4proc.c        |   15 +++++++++++++++
 fs/nfsd/nfs4xdr.c         |   17 +++++++++++++++++
 include/linux/nfs4.h      |    8 +++++++-
 include/linux/nfsd/xdr4.h |   10 ++++------
 4 files changed, 43 insertions(+), 7 deletions(-)
fs/nfsd/nfs4proc.c +15 -0
index f3b560f..6d7d7c5 100644
@@ -93,6 +93,21 @@ do_open_lookup(struct svc_rqst *rqstp, struct svc_fh *current_fh, struct nfsd4_o
 	open->op_truncate = 0;
 
 	if (open->op_create) {
+		/* FIXME: check session persistence and pnfs flags.
+		 * The nfsv4.1 spec requires the following semantics:
+		 *
+		 * Persistent   | pNFS   | Server REQUIRED | Client Allowed
+		 * Reply Cache  | server |                 |
+		 * -------------+--------+-----------------+--------------------
+		 * no           | no     | EXCLUSIVE4_1    | EXCLUSIVE4_1
+		 *              |        |                 | (SHOULD)
+		 *              |        | and EXCLUSIVE4  | or EXCLUSIVE4
+		 *              |        |                 | (SHOULD NOT)
+		 * no           | yes    | EXCLUSIVE4_1    | EXCLUSIVE4_1
+		 * yes          | no     | GUARDED4        | GUARDED4
+		 * yes          | yes    | GUARDED4        | GUARDED4
+		 */
+
 		/*
 		 * Note: create modes (UNCHECKED,GUARDED...) are the same
 		 * in NFSv4 as in v3.
fs/nfsd/nfs4xdr.c +17 -0
index 288332c..0084459 100644
@@ -260,6 +260,12 @@ static u32 nfsd_attrmask[] = {
 	NFSD_WRITEABLE_ATTRS_WORD2
 };
 
+static u32 nfsd41_ex_attrmask[] = {
+	NFSD_SUPPATTR_EXCLCREAT_WORD0,
+	NFSD_SUPPATTR_EXCLCREAT_WORD1,
+	NFSD_SUPPATTR_EXCLCREAT_WORD2
+};
+
 static __be32
 nfsd4_decode_fattr(struct nfsd4_compoundargs *argp, u32 *bmval, u32 *writable,
 		   struct iattr *iattr, struct nfs4_acl **acl)
@@ -684,6 +690,17 @@ nfsd4_decode_open(struct nfsd4_compoundargs *argp, struct nfsd4_open *open)
 			READ_BUF(8);
 			COPYMEM(open->op_verf.data, 8);
 			break;
+		case NFS4_CREATE_EXCLUSIVE4_1:
+			if (argp->minorversion < 1)
+				goto xdr_error;
+			READ_BUF(8);
+			COPYMEM(open->op_verf.data, 8);
+			status = nfsd4_decode_fattr(argp, open->op_bmval,
+				nfsd41_ex_attrmask, &open->op_iattr,
+				&open->op_acl);
+			if (status)
+				goto out;
+			break;
 		default:
 			goto xdr_error;
 		}
include/linux/nfs4.h +7 -1
index 735c7d8..5756359 100644
@@ -368,7 +368,13 @@ enum opentype4 {
 enum createmode4 {
 	NFS4_CREATE_UNCHECKED = 0,
 	NFS4_CREATE_GUARDED = 1,
-	NFS4_CREATE_EXCLUSIVE = 2
+	NFS4_CREATE_EXCLUSIVE = 2,
+	/*
+	 * New to NFSv4.1. If session is persistent,
+	 * GUARDED4 MUST be used. Otherwise, use
+	 * EXCLUSIVE4_1 instead of EXCLUSIVE4.
+	 */
+	NFS4_CREATE_EXCLUSIVE4_1 = 3
 };
 
 enum limit_by4 {
include/linux/nfsd/xdr4.h +4 -6
index d7a8050..7ed9955 100644
@@ -218,10 +218,8 @@ struct nfsd4_open {
 	u32		op_create;     	    /* request */
 	u32		op_createmode;      /* request */
 	u32		op_bmval[3];        /* request */
-	union {                             /* request */
-		struct iattr	iattr;                      /* UNCHECKED4,GUARDED4 */
-		nfs4_verifier	verf;                                /* EXCLUSIVE4 */
-	} u;
+	struct iattr	iattr;              /* UNCHECKED4, GUARDED4, EXCLUSIVE4_1 */
+	nfs4_verifier	verf;               /* EXCLUSIVE4 */
 	clientid_t	op_clientid;        /* request */
 	struct xdr_netobj op_owner;           /* request */
 	u32		op_seqid;           /* request */
@@ -235,8 +233,8 @@ struct nfsd4_open {
 	struct nfs4_stateowner *op_stateowner; /* used during processing */
 	struct nfs4_acl *op_acl;
 };
-#define op_iattr	u.iattr
-#define op_verf		u.verf
+#define op_iattr	iattr
+#define op_verf		verf
 
 struct nfsd4_open_confirm {
 	stateid_t	oc_req_stateid		/* request */;
Signature · 1 line
1.6.2.1
#59 lore raw

PATCH 47/47 nfsd41: Documentation/filesystems/nfs41-server.txt

In reply to Benny Halevy

Initial nfs41 server write up describing the status of the linux server implementation.

---
 Documentation/filesystems/nfs41-server.txt |  148 ++++++++++++++++++++++++++++
 1 files changed, 148 insertions(+), 0 deletions(-)
 create mode 100644 Documentation/filesystems/nfs41-server.txt
Documentation/filesystems/nfs41-server.txt new file +148 -0
new file mode 100644
index 0000000..e62cd13
@@ -0,0 +1,148 @@
+NFSv4.1 Server Implementation
+
+The NFSv4 minorversion 1 (NFSv4.1) implementation in nfsd is based
+on the latest NFSv4.1 Internet Draft:
+http://tools.ietf.org/html/draft-ietf-nfsv4-minorversion1-29
+
+If you're a developer or just eager to test the code, set
+CONFIG_NFSD_V4_1 to 'y' in your kernel configuration file.
+This will allow the server to serve both NFSv4 and NFSv4.1
+clients concurrently.
+
+From the many new features in NFSv4.1 the current implementation
+focuses on the mandatory-to-implement NFSv4.1 Sessions, providing
+"exactly once" semantics and better control and throttling of the
+resources allocated for each client.
+
+Other NFSv4.1 features, Parallel NFS operations in particular,
+are still under development out of tree.
+See http://wiki.linux-nfs.org/wiki/index.php/PNFS_prototype_design
+for more information.
+
+The table below, taken from the NFSv4.1 document, lists
+the operations that are mandatory to implement (REQ), optional
+(OPT), and NFSv4.0 operations that are required not to implement (MNI)
+in minor version 1.  The first column indicates the operations that
+are not supported yet by the linux server implementation.
+
+The OPTIONAL features identified and their abbreviations are as follows:
+	pNFS	Parallel NFS
+	FDELG	File Delegations
+	DDELG	Directory Delegations
+
+The following abbreviations indicate the linux server implementation status.
+	I	Implemented NFSv4.1 operations.
+	NS	Not Supported.
+	NS*	unimplemented optional feature.
+	P	pNFS features implemented out of tree.
+	PNS	pNFS features that are not supported yet (out of tree).
+
+Operations
+
+   +----------------------+------------+--------------+----------------+
+   | Operation            | REQ, REC,  | Feature      | Definition     |
+   |                      | OPT, or    | (REQ, REC,   |                |
+   |                      | MNI        | or OPT)      |                |
+   +----------------------+------------+--------------+----------------+
+   | ACCESS               | REQ        |              | Section 18.1   |
+NS | BACKCHANNEL_CTL      | REQ        |              | Section 18.33  |
+NS | BIND_CONN_TO_SESSION | REQ        |              | Section 18.34  |
+   | CLOSE                | REQ        |              | Section 18.2   |
+   | COMMIT               | REQ        |              | Section 18.3   |
+   | CREATE               | REQ        |              | Section 18.4   |
+I  | CREATE_SESSION       | REQ        |              | Section 18.36  |
+NS*| DELEGPURGE           | OPT        | FDELG (REQ)  | Section 18.5   |
+   | DELEGRETURN          | OPT        | FDELG,       | Section 18.6   |
+   |                      |            | DDELG, pNFS  |                |
+   |                      |            | (REQ)        |                |
+NS | DESTROY_CLIENTID     | REQ        |              | Section 18.50  |
+I  | DESTROY_SESSION      | REQ        |              | Section 18.37  |
+I  | EXCHANGE_ID          | REQ        |              | Section 18.35  |
+NS | FREE_STATEID         | REQ        |              | Section 18.38  |
+   | GETATTR              | REQ        |              | Section 18.7   |
+P  | GETDEVICEINFO        | OPT        | pNFS (REQ)   | Section 18.40  |
+P  | GETDEVICELIST        | OPT        | pNFS (OPT)   | Section 18.41  |
+   | GETFH                | REQ        |              | Section 18.8   |
+NS*| GET_DIR_DELEGATION   | OPT        | DDELG (REQ)  | Section 18.39  |
+P  | LAYOUTCOMMIT         | OPT        | pNFS (REQ)   | Section 18.42  |
+P  | LAYOUTGET            | OPT        | pNFS (REQ)   | Section 18.43  |
+P  | LAYOUTRETURN         | OPT        | pNFS (REQ)   | Section 18.44  |
+   | LINK                 | OPT        |              | Section 18.9   |
+   | LOCK                 | REQ        |              | Section 18.10  |
+   | LOCKT                | REQ        |              | Section 18.11  |
+   | LOCKU                | REQ        |              | Section 18.12  |
+   | LOOKUP               | REQ        |              | Section 18.13  |
+   | LOOKUPP              | REQ        |              | Section 18.14  |
+   | NVERIFY              | REQ        |              | Section 18.15  |
+   | OPEN                 | REQ        |              | Section 18.16  |
+NS*| OPENATTR             | OPT        |              | Section 18.17  |
+   | OPEN_CONFIRM         | MNI        |              | N/A            |
+   | OPEN_DOWNGRADE       | REQ        |              | Section 18.18  |
+   | PUTFH                | REQ        |              | Section 18.19  |
+   | PUTPUBFH             | REQ        |              | Section 18.20  |
+   | PUTROOTFH            | REQ        |              | Section 18.21  |
+   | READ                 | REQ        |              | Section 18.22  |
+   | READDIR              | REQ        |              | Section 18.23  |
+   | READLINK             | OPT        |              | Section 18.24  |
+NS | RECLAIM_COMPLETE     | REQ        |              | Section 18.51  |
+   | RELEASE_LOCKOWNER    | MNI        |              | N/A            |
+   | REMOVE               | REQ        |              | Section 18.25  |
+   | RENAME               | REQ        |              | Section 18.26  |
+   | RENEW                | MNI        |              | N/A            |
+   | RESTOREFH            | REQ        |              | Section 18.27  |
+   | SAVEFH               | REQ        |              | Section 18.28  |
+   | SECINFO              | REQ        |              | Section 18.29  |
+NS | SECINFO_NO_NAME      | REC        | pNFS files   | Section 18.45, |
+   |                      |            | layout (REQ) | Section 13.12  |
+I  | SEQUENCE             | REQ        |              | Section 18.46  |
+   | SETATTR              | REQ        |              | Section 18.30  |
+   | SETCLIENTID          | MNI        |              | N/A            |
+   | SETCLIENTID_CONFIRM  | MNI        |              | N/A            |
+NS | SET_SSV              | REQ        |              | Section 18.47  |
+NS | TEST_STATEID         | REQ        |              | Section 18.48  |
+   | VERIFY               | REQ        |              | Section 18.31  |
+NS*| WANT_DELEGATION      | OPT        | FDELG (OPT)  | Section 18.49  |
+   | WRITE                | REQ        |              | Section 18.32  |
+
+Callback Operations
+
+   +-------------------------+-----------+-------------+---------------+
+   | Operation               | REQ, REC, | Feature     | Definition    |
+   |                         | OPT, or   | (REQ, REC,  |               |
+   |                         | MNI       | or OPT)     |               |
+   +-------------------------+-----------+-------------+---------------+
+   | CB_GETATTR              | OPT       | FDELG (REQ) | Section 20.1  |
+P  | CB_LAYOUTRECALL         | OPT       | pNFS (REQ)  | Section 20.3  |
+NS*| CB_NOTIFY               | OPT       | DDELG (REQ) | Section 20.4  |
+P  | CB_NOTIFY_DEVICEID      | OPT       | pNFS (OPT)  | Section 20.12 |
+NS*| CB_NOTIFY_LOCK          | OPT       |             | Section 20.11 |
+NS*| CB_PUSH_DELEG           | OPT       | FDELG (OPT) | Section 20.5  |
+   | CB_RECALL               | OPT       | FDELG,      | Section 20.2  |
+   |                         |           | DDELG, pNFS |               |
+   |                         |           | (REQ)       |               |
+NS*| CB_RECALL_ANY           | OPT       | FDELG,      | Section 20.6  |
+   |                         |           | DDELG, pNFS |               |
+   |                         |           | (REQ)       |               |
+NS | CB_RECALL_SLOT          | REQ       |             | Section 20.8  |
+NS*| CB_RECALLABLE_OBJ_AVAIL | OPT       | DDELG, pNFS | Section 20.7  |
+   |                         |           | (REQ)       |               |
+I  | CB_SEQUENCE             | OPT       | FDELG,      | Section 20.9  |
+   |                         |           | DDELG, pNFS |               |
+   |                         |           | (REQ)       |               |
+NS*| CB_WANTS_CANCELLED      | OPT       | FDELG,      | Section 20.10 |
+   |                         |           | DDELG, pNFS |               |
+   |                         |           | (REQ)       |               |
+   +-------------------------+-----------+-------------+---------------+
+
+Implementation notes:
+
+EXCHANGE_ID:
+* only SP4_NONE state protection supported
+* implementation ids are ignored
+
+CREATE_SESSION:
+* backchannel attributes are ignored
+* backchannel security parameters are ignored
+
+SEQUENCE:
+* no support for dynamic slot table renegotiation (optional)
Signature · 1 line
1.6.2.1
#60 lore raw
In reply to Benny Halevy
Quoted 8 lines · Benny Halevy

Hi Bruce,

Here's the latest server patches implementing the NFSv4.1 Sessions features.

This patchset is based over your for-2.6.30 branch and is also available from
git://linux-nfs.org/~bhalevy/linux-pnfs.git nfsd41-for-2.6.30

There's a few merge conflicts with my current for-2.6.30--probably my fault for not pushing that out recently enough, apologies. Would you mind updating?

--b.

Quoted 67 lines

(tagged as nfsd41-for-2.6.30-2009-03-27)

The implementation status highlights are captured in Documentation/filesystems/nfs41-server.txt.
The gist of it is that the implementation is functional to the level of passing Connectathon tests with v2,v3,v4,v4.1 (tested with CONFIG_NFS_V4_1 defined to 'y')

The patches are listed below.

Thanks for your review and help with this project so far. Kudos to all the contributors (Authors/SOBs: Andy Adamson, Dean Hildebrand, Marc Eshel, Mike Sager, Ricardo Labiaga, and myself) This is a pretty significant milestone for us (well, the real one will be when all this ends up working flawlessly in the kernel. :)

Benny

[PATCH 01/47] nfsd: don't use the deferral service, return NFS4ERR_DELAY [PATCH 02/47] sunrpc: add cl_private field to struct rpc_clnt
[PATCH 03/47] nfsd: embed nfsd4_current_state in nfsd4_compoundres [PATCH 04/47] nfsd: add a struct nfsd4_slot pointer to struct nfsd4_compound_state [PATCH 05/47] nfs41: common protocol definitions
[PATCH 06/47] nfsd41: change NFSERR_REPLAY_ME
[PATCH 07/47] nfsd41: Add Kconfig symbols for NFSv4.1
[PATCH 08/47] nfsd41: define nfs41 error codes
[PATCH 09/47] nfsd41: sessions basic data types
[PATCH 10/47] nfsd41: introduce nfs4_client cl_sessions list
[PATCH 11/47] nfsd41: release_session when client is expired
[PATCH 12/47] nfsd41: sessionid hashing
[PATCH 13/47] nfsd41: xdr infrastructure
[PATCH 14/47] nfsd: remove nfsd4_ops array size
[PATCH 15/47] nfsd41: proc stubs
[PATCH 16/47] nfsd41: exchange_id operation
[PATCH 17/47] nfsd41: match clientid establishment method
[PATCH 18/47] nfsd41: sequence operation
[PATCH 19/47] nfsd41: enforce NFS4ERR_SEQUENCE_POS operation order rules [PATCH 20/47] nfsd41: DRC save, restore, and clear functions
[PATCH 21/47] nfsd41: hard page limit for DRC
[PATCH 22/47] nfsd41: nfsd DRC logic
[PATCH 23/47] nfsd41: clear DRC cache on free_session
[PATCH 24/47] nfsd41: create_session operation
[PATCH 25/47] nfsd41: Add a create session replay cache
[PATCH 26/47] nfsd41: non-page DRC for solo sequence responses
[PATCH 27/47] nfsd41: destroy_session operation
[PATCH 28/47] nfsd41: stateid handling
[PATCH 29/47] nfsd41: check encode size for sessions maxresponse cached [PATCH 30/47] nfsd41: clientid handling
[PATCH 31/47] nfsd41: access_valid
[PATCH 32/47] nfsd41: add OPEN4_SHARE_ACCESS_WANT nfs4_stateid bmap [PATCH 33/47] nfsd41: provide support for minor version 1 at rpc level [PATCH 34/47] nfsd: cleanup nfs4.0 callback encode routines
[PATCH 35/47] nfsd: minorversion support for the back channel
[PATCH 36/47] nfsd41: sunrpc: Added rpc server-side backchannel handling [PATCH 37/47] nfsd41: callback infrastructure
[PATCH 38/47] nfsd41: Remember the auth flavor to use for callbacks [PATCH 39/47] nfsd41: introduce cl_cb_mutex
[PATCH 40/47] nfsd41: cb_sequence callback
[PATCH 41/47] nfsd41: introduce nfs4_cb_call_sync for nfs4 and nfs41 [PATCH 42/47] nfsd41: cb_recall callback
[PATCH 43/47] nfsd41: pass writable attrs mask to nfsd4_decode_fattr [PATCH 44/47] nfsd41: support for 3-word long attribute bitmask [PATCH 45/47] nfsd41: SUPPATTR_EXCLCREAT attribute
[PATCH 46/47] nfsd41: CREATE_EXCLUSIVE4_1
[PATCH 47/47] nfsd41: Documentation/filesystems/nfs41-server.txt

#2 lore raw

Re: [PATCH 01/47] nfsd: don't use the deferral service, return NFS4ERR_DELAY

In reply to Benny Halevy
Quoted 9 lines · Benny Halevy

From: Andy Adamson <[email protected]>

On an NFSv4.1 server cache miss that causes an upcall, NFS4ERR_DELAY will be returned. It is up to the NFSv4.1 client to resend only the operations that have not been processed.

Initialize rq_usedeferral to 1 in svc_process(). rq_usedeferral was initialized to 0, which means that the first NFSv4.0 or NFSv4.1 rpc would be dropped due to the logic in svc_defer().

Nit: is this a comment about a change from a previous version of the patch? It doesn't make much sense to a reader who hasn't seen that previous version.

Quoted 8 lines

Will be turned off in nfsd4_proc_compound() only
when NFSv4 Sessions are used.

Note: this isn't an adequate solution on its own. It's acceptable as a way to get some minimal 4.1 up and working, but we're going to have to find a way to avoid returning DELAY in all common cases before 4.1 can really be considered ready.

Quoted 8 lines
[nfsd41: reverse rq_nodeferral negative logic]

use positive rq_usedeferral logic instead.

[sunrpc: initialize rq_usedeferral]

I'd be inclined to just collapse that into the two signed-off-by's, and drop the patch-history information.

--b.

Quoted 78 lines
---
 fs/nfsd/nfs4proc.c         |    8 ++++++++
 include/linux/sunrpc/svc.h |    1 +
 net/sunrpc/svc.c           |    2 ++
 net/sunrpc/svc_xprt.c      |    2 +-
 4 files changed, 12 insertions(+), 1 deletions(-)
fs/nfsd/nfs4proc.c +8 -0
index f156b85..7839654 100644
@@ -873,6 +873,8 @@ nfsd4_proc_compound(struct svc_rqst *rqstp,
 	resp->tag = args->tag;
 	resp->opcnt = 0;
 	resp->rqstp = rqstp;
+	/* Use the deferral mechanism only for NFSv4.0 compounds */
+	rqstp->rq_usedeferral = (args->minorversion == 0);
 
 	/*
 	 * According to RFC3010, this takes precedence over all other errors.
@@ -957,10 +959,16 @@ encode_op:
 
 		nfsd4_increment_op_stats(op->opnum);
 	}
+	if (!rqstp->rq_usedeferral && status == nfserr_dropit) {
+		dprintk("%s Dropit - send NFS4ERR_DELAY\n", __func__);
+		status = nfserr_jukebox;
+	}
 
 	cstate_free(cstate);
 out:
 	nfsd4_release_compoundargs(args);
+	/* Reset deferral mechanism for RPC deferrals */
+	rqstp->rq_usedeferral = 1;
 	dprintk("nfsv4 compound returned %d\n", ntohl(status));
 	return status;
 }
include/linux/sunrpc/svc.h +1 -0
index 9f9f699..815dd58 100644
@@ -230,6 +230,7 @@ struct svc_rqst {
 	struct svc_cred		rq_cred;	/* auth info */
 	void *			rq_xprt_ctxt;	/* transport specific context ptr */
 	struct svc_deferred_req*rq_deferred;	/* deferred request we are replaying */
+	int			rq_usedeferral;	/* use deferral */
 
 	size_t			rq_xprt_hlen;	/* xprt header len */
 	struct xdr_buf		rq_arg;
net/sunrpc/svc.c +2 -0
index c51fed4..6334858 100644
@@ -1023,6 +1023,8 @@ svc_process(struct svc_rqst *rqstp)
 	rqstp->rq_res.tail[0].iov_len = 0;
 	/* Will be turned off only in gss privacy case: */
 	rqstp->rq_splice_ok = 1;
+	/* Will be turned off only when NFSv4 Sessions are used */
+	rqstp->rq_usedeferral = 1;
 
 	/* Setup reply header */
 	rqstp->rq_xprt->xpt_ops->xpo_prep_reply_hdr(rqstp);
net/sunrpc/svc_xprt.c +1 -1
index 1e66f24..600d091 100644
@@ -974,7 +974,7 @@ static struct cache_deferred_req *svc_defer(struct cache_req *req)
 	struct svc_rqst *rqstp = container_of(req, struct svc_rqst, rq_chandle);
 	struct svc_deferred_req *dr;
 
-	if (rqstp->rq_arg.page_len)
+	if (rqstp->rq_arg.page_len || !rqstp->rq_usedeferral)
 		return NULL; /* if more than a page, give up FIXME */
 	if (rqstp->rq_deferred) {
 		dr = rqstp->rq_deferred;
Signature · 1 line
1.6.2.1
#5 lore raw

Re: [PATCH 02/47] sunrpc: add cl_private field to struct rpc_clnt

In reply to Benny Halevy
Quoted 3 lines · Benny Halevy

From: Andy Adamson <[email protected]>

Note: the NFSv4.1 client also uses (and declares) this pointer.

OK. Ack from trond?

--b.

Quoted 21 lines
---
 include/linux/sunrpc/clnt.h |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)
include/linux/sunrpc/clnt.h +1 -0
index c39a210..928c547 100644
@@ -53,6 +53,7 @@ struct rpc_clnt {
 	char			cl_pathname[30];/* Path in rpc_pipe_fs */
 	struct vfsmount *	cl_vfsmnt;
 	struct dentry *		cl_dentry;	/* inode */
+	void			*cl_private;	/* private data */
 	struct rpc_clnt *	cl_parent;	/* Points to parent of clones */
 	struct rpc_rtt		cl_rtt_default;
 	struct rpc_timeout	cl_timeout_default;
Signature · 1 line
1.6.2.1
#6 lore raw

Re: [PATCH 02/47] sunrpc: add cl_private field to struct rpc_clnt

Quoted 6 lines · J. Bruce Fields
Quoted 3 lines · Benny Halevy

From: Andy Adamson <[email protected]>

Note: the NFSv4.1 client also uses (and declares) this pointer.

OK. Ack from trond?

First, someone would need to remind me why it is necessary, and add that justification to the changelog.

Trond

Quoted 26 lines

--b.

Quoted 22 lines

---
include/linux/sunrpc/clnt.h | 1 +

1 files changed, 1 insertions(+), 0 deletions(-)
include/linux/sunrpc/clnt.h +1 -0
index c39a210..928c547 100644
@@ -53,6 +53,7 @@ struct rpc_clnt {
   char            cl_pathname[30];/* Path in rpc_pipe_fs */
   struct vfsmount *    cl_vfsmnt;
   struct dentry *        cl_dentry;    /* inode */
+    void            *cl_private;    /* private data */
   struct rpc_clnt *    cl_parent;    /* Points to parent of clones  
 */
   struct rpc_rtt        cl_rtt_default;
   struct rpc_timeout    cl_timeout_default;
Signature · 1 line
1.6.2.1
#209 lore raw
In reply to Benny Halevy
Quoted 8 lines · Benny Halevy

Hi Bruce,

Here's the latest server patches implementing the NFSv4.1 Sessions features.

This patchset is based over your for-2.6.30 branch and is also available from
git://linux-nfs.org/~bhalevy/linux-pnfs.git nfsd41-for-2.6.30

I ran the v4.0 pynfs on that branch and got some new failures:

ACC3 st_access.testNoFh : FAILURE
value=10064 not in enum nfsstat4
CLOSE7 st_close.testNoCfh : FAILURE
value=10064 not in enum nfsstat4
CMT3 st_commit.testNoFh : FAILURE
value=10064 not in enum nfsstat4
CR8 st_create.testNoFh : FAILURE
value=10064 not in enum nfsstat4

Looks like NFS4ERR_SEQUENCE_POS is leaking out to nfsv4 code for some reason?

(Some similar results omitted.)

Also there's a bunch of stuff like this:

NVF1a st_nverify.testMandLink : FAILURE
NVerifying mandatory attributes against getattr should return NFS4ERR_SAME, instead got NFS4_OK
NVF1d st_nverify.testMandDir : FAILURE
NVerifying mandatory attributes against getattr should return NFS4ERR_SAME, instead got NFS4_OK

Probably the expansion of the attribute bitmask is exposing some preexisting bug in the 4.0 code, but let's try to figure out how to fix that.

I haven't investigated either of these any closer yet.

--b.

#3 lore raw

Re: [PATCH 01/47] nfsd: don't use the deferral service, return NFS4ERR_DELAY

In reply to J. Bruce Fields
Quoted 7 lines · J. Bruce Fields
Quoted 6 lines · Benny Halevy

From: Andy Adamson <[email protected]>

On an NFSv4.1 server cache miss that causes an upcall, NFS4ERR_DELAY will be returned. It is up to the NFSv4.1 client to resend only the operations that have not been processed.

Quoted 7 lines
Quoted 3 lines

Initialize rq_usedeferral to 1 in svc_process(). rq_usedeferral was initialized to 0, which means that the first NFSv4.0 or NFSv4.1 rpc would be dropped due to the logic in svc_defer().

Nit: is this a comment about a change from a previous version of the patch? It doesn't make much sense to a reader who hasn't seen that previous version.

Yes. I guess we can drop the second sentence in the paragraph above from the final version.

Benny

Quoted 101 lines
Quoted 9 lines

Will be turned off in nfsd4_proc_compound() only
when NFSv4 Sessions are used.

Note: this isn't an adequate solution on its own. It's acceptable as a way to get some minimal 4.1 up and working, but we're going to have to find a way to avoid returning DELAY in all common cases before 4.1 can really be considered ready.

Quoted 8 lines
[nfsd41: reverse rq_nodeferral negative logic]

use positive rq_usedeferral logic instead.

[sunrpc: initialize rq_usedeferral]

I'd be inclined to just collapse that into the two signed-off-by's, and drop the patch-history information.

--b.

Quoted 78 lines
---
 fs/nfsd/nfs4proc.c         |    8 ++++++++
 include/linux/sunrpc/svc.h |    1 +
 net/sunrpc/svc.c           |    2 ++
 net/sunrpc/svc_xprt.c      |    2 +-
 4 files changed, 12 insertions(+), 1 deletions(-)
fs/nfsd/nfs4proc.c +8 -0
index f156b85..7839654 100644
@@ -873,6 +873,8 @@ nfsd4_proc_compound(struct svc_rqst *rqstp,
 	resp->tag = args->tag;
 	resp->opcnt = 0;
 	resp->rqstp = rqstp;
+	/* Use the deferral mechanism only for NFSv4.0 compounds */
+	rqstp->rq_usedeferral = (args->minorversion == 0);
 
 	/*
 	 * According to RFC3010, this takes precedence over all other errors.
@@ -957,10 +959,16 @@ encode_op:
 
 		nfsd4_increment_op_stats(op->opnum);
 	}
+	if (!rqstp->rq_usedeferral && status == nfserr_dropit) {
+		dprintk("%s Dropit - send NFS4ERR_DELAY\n", __func__);
+		status = nfserr_jukebox;
+	}
 
 	cstate_free(cstate);
 out:
 	nfsd4_release_compoundargs(args);
+	/* Reset deferral mechanism for RPC deferrals */
+	rqstp->rq_usedeferral = 1;
 	dprintk("nfsv4 compound returned %d\n", ntohl(status));
 	return status;
 }
include/linux/sunrpc/svc.h +1 -0
index 9f9f699..815dd58 100644
@@ -230,6 +230,7 @@ struct svc_rqst {
 	struct svc_cred		rq_cred;	/* auth info */
 	void *			rq_xprt_ctxt;	/* transport specific context ptr */
 	struct svc_deferred_req*rq_deferred;	/* deferred request we are replaying */
+	int			rq_usedeferral;	/* use deferral */
 
 	size_t			rq_xprt_hlen;	/* xprt header len */
 	struct xdr_buf		rq_arg;
net/sunrpc/svc.c +2 -0
index c51fed4..6334858 100644
@@ -1023,6 +1023,8 @@ svc_process(struct svc_rqst *rqstp)
 	rqstp->rq_res.tail[0].iov_len = 0;
 	/* Will be turned off only in gss privacy case: */
 	rqstp->rq_splice_ok = 1;
+	/* Will be turned off only when NFSv4 Sessions are used */
+	rqstp->rq_usedeferral = 1;
 
 	/* Setup reply header */
 	rqstp->rq_xprt->xpt_ops->xpo_prep_reply_hdr(rqstp);
net/sunrpc/svc_xprt.c +1 -1
index 1e66f24..600d091 100644
@@ -974,7 +974,7 @@ static struct cache_deferred_req *svc_defer(struct cache_req *req)
 	struct svc_rqst *rqstp = container_of(req, struct svc_rqst, rq_chandle);
 	struct svc_deferred_req *dr;
 
-	if (rqstp->rq_arg.page_len)
+	if (rqstp->rq_arg.page_len || !rqstp->rq_usedeferral)
 		return NULL; /* if more than a page, give up FIXME */
 	if (rqstp->rq_deferred) {
 		dr = rqstp->rq_deferred;
Signature · 1 line
1.6.2.1
#7 lore raw

Re: [PATCH 02/47] sunrpc: add cl_private field to struct rpc_clnt

In reply to Myklebust, Trond
Quoted 4 lines · Trond
Quoted 1 line · J. Bruce Fields

On Fri, Mar 27, 2009 at 06:01:48AM +0300, Benny Halevy wrote:

Quoted 8 lines
Quoted 4 lines
Quoted 3 lines

From: Andy Adamson <[email protected]>

Note: the NFSv4.1 client also uses (and declares) this pointer.

OK. Ack from trond?

First, someone would need to remind me why it is necessary, and add that justification to the changelog.

First time this is used in this patchset is here:
[PATCH 35/47] nfsd: minorversion support for the back channel

The client uses cl_private to determine the minorversion (via a struct nfs_client *) to be set in the compound header, and to know when to generate a SEQUENCE op. Similarly, the server puts a struct nfs4_callback * in there for callback compounds' CB_COMPOUND and CB_SEQUENCE.

Benny

Quoted 28 lines

Trond

Quoted 26 lines

--b.

Quoted 23 lines

---
include/linux/sunrpc/clnt.h | 1 +

1 files changed, 1 insertions(+), 0 deletions(-)
include/linux/sunrpc/ renamed from include/linux/sunrpc/clnt.h +0 -0

clnt.h
index c39a210..928c547 100644

include/linux/sunrpc/clnt.h +1 -0
@@ -53,6 +53,7 @@ struct rpc_clnt {
   char            cl_pathname[30];/* Path in rpc_pipe_fs */
   struct vfsmount *    cl_vfsmnt;
   struct dentry *        cl_dentry;    /* inode */
+    void            *cl_private;    /* private data */
   struct rpc_clnt *    cl_parent;    /* Points to parent of clones  
 */
   struct rpc_rtt        cl_rtt_default;
   struct rpc_timeout    cl_timeout_default;
Signature · 1 line
1.6.2.1
#61 lore raw
In reply to J. Bruce Fields
Quoted 10 lines · J. Bruce Fields
Quoted 8 lines · Benny Halevy

Hi Bruce,

Here's the latest server patches implementing the NFSv4.1 Sessions features.

This patchset is based over your for-2.6.30 branch and is also available from
git://linux-nfs.org/~bhalevy/linux-pnfs.git nfsd41-for-2.6.30

Quoted 6 lines

There's a few merge conflicts with my current for-2.6.30--probably my fault for not pushing that out recently enough, apologies. Would you mind updating?

--b.

Sure. Here's a rebased version with two minor changes (see below)

Benny

[PATCH v2 01/47] nfsd: don't use the deferral service, return NFS4ERR_DELAY commit message edited.

[PATCH v2 02/47] nfsd: embed nfsd4_current_state in nfsd4_compoundres [PATCH v2 03/47] nfsd: add a struct nfsd4_slot pointer to struct nfsd4_compound_state [PATCH v2 04/47] nfs41: common protocol definitions
[PATCH v2 05/47] nfsd41: change NFSERR_REPLAY_ME
[PATCH v2 06/47] nfsd41: Add Kconfig symbols for NFSv4.1
[PATCH v2 07/47] nfsd41: define nfs41 error codes
[PATCH v2 08/47] nfsd41: sessions basic data types
[PATCH v2 09/47] nfsd41: introduce nfs4_client cl_sessions list
[PATCH v2 10/47] nfsd41: release_session when client is expired
[PATCH v2 11/47] nfsd41: sessionid hashing
[PATCH v2 12/47] nfsd41: xdr infrastructure
[PATCH v2 13/47] nfsd: remove nfsd4_ops array size
[PATCH v2 14/47] nfsd41: proc stubs
[PATCH v2 15/47] nfsd41: exchange_id operation
[PATCH v2 16/47] nfsd41: match clientid establishment method
[PATCH v2 17/47] nfsd41: sequence operation
[PATCH v2 18/47] nfsd41: enforce NFS4ERR_SEQUENCE_POS operation order rules [PATCH v2 19/47] nfsd41: DRC save, restore, and clear functions
[PATCH v2 20/47] nfsd41: hard page limit for DRC
[PATCH v2 21/47] nfsd41: nfsd DRC logic
[PATCH v2 22/47] nfsd41: clear DRC cache on free_session
[PATCH v2 23/47] nfsd41: create_session operation
[PATCH v2 24/47] nfsd41: Add a create session replay cache
[PATCH v2 25/47] nfsd41: non-page DRC for solo sequence responses [PATCH v2 26/47] nfsd41: destroy_session operation
[PATCH v2 27/47] nfsd41: stateid handling
[PATCH v2 28/47] nfsd41: check encode size for sessions maxresponse cached [PATCH v2 29/47] nfsd41: clientid handling
[PATCH v2 30/47] nfsd41: access_valid
[PATCH v2 31/47] nfsd41: add OPEN4_SHARE_ACCESS_WANT nfs4_stateid bmap [PATCH v2 32/47] nfsd41: provide support for minor version 1 at rpc level [PATCH v2 33/47] nfsd: cleanup nfs4.0 callback encode routines
[PATCH v2 34/47] sunrpc: add cl_private field to struct rpc_clnt
patch moved forward, right before first use.

[PATCH v2 35/47] nfsd: minorversion support for the back channel [PATCH v2 36/47] nfsd41: sunrpc: Added rpc server-side backchannel handling [PATCH v2 37/47] nfsd41: callback infrastructure
[PATCH v2 38/47] nfsd41: Remember the auth flavor to use for callbacks [PATCH v2 39/47] nfsd41: introduce cl_cb_mutex
[PATCH v2 40/47] nfsd41: cb_sequence callback
[PATCH v2 41/47] nfsd41: introduce nfs4_cb_call_sync for nfs4 and nfs41 [PATCH v2 42/47] nfsd41: cb_recall callback
[PATCH v2 43/47] nfsd41: pass writable attrs mask to nfsd4_decode_fattr [PATCH v2 44/47] nfsd41: support for 3-word long attribute bitmask [PATCH v2 45/47] nfsd41: SUPPATTR_EXCLCREAT attribute
[PATCH v2 46/47] nfsd41: CREATE_EXCLUSIVE4_1
[PATCH v2 47/47] nfsd41: Documentation/filesystems/nfs41-server.txt

#62 lore raw

PATCH v2 1/47 nfsd: don't use the deferral service, return NFS4ERR_DELAY

From: Andy Adamson <[email protected]>

On an NFSv4.1 server cache miss that causes an upcall, NFS4ERR_DELAY will be returned. It is up to the NFSv4.1 client to resend only the operations that have not been processed.

Initialize rq_usedeferral to 1 in svc_process(). It sill be turned off in nfsd4_proc_compound() only when NFSv4.1 Sessions are used.

Note: this isn't an adequate solution on its own. It's acceptable as a way to get some minimal 4.1 up and working, but we're going to have to find a way to avoid returning DELAY in all common cases before 4.1 can really be considered ready.

[nfsd41: reverse rq_nodeferral negative logic]
[sunrpc: initialize rq_usedeferral]
---
 fs/nfsd/nfs4proc.c         |    8 ++++++++
 include/linux/sunrpc/svc.h |    1 +
 net/sunrpc/svc.c           |    2 ++
 net/sunrpc/svc_xprt.c      |    2 +-
 4 files changed, 12 insertions(+), 1 deletions(-)
fs/nfsd/nfs4proc.c +8 -0
index f156b85..7839654 100644
@@ -873,6 +873,8 @@ nfsd4_proc_compound(struct svc_rqst *rqstp,
 	resp->tag = args->tag;
 	resp->opcnt = 0;
 	resp->rqstp = rqstp;
+	/* Use the deferral mechanism only for NFSv4.0 compounds */
+	rqstp->rq_usedeferral = (args->minorversion == 0);
 
 	/*
 	 * According to RFC3010, this takes precedence over all other errors.
@@ -957,10 +959,16 @@ encode_op:
 
 		nfsd4_increment_op_stats(op->opnum);
 	}
+	if (!rqstp->rq_usedeferral && status == nfserr_dropit) {
+		dprintk("%s Dropit - send NFS4ERR_DELAY\n", __func__);
+		status = nfserr_jukebox;
+	}
 
 	cstate_free(cstate);
 out:
 	nfsd4_release_compoundargs(args);
+	/* Reset deferral mechanism for RPC deferrals */
+	rqstp->rq_usedeferral = 1;
 	dprintk("nfsv4 compound returned %d\n", ntohl(status));
 	return status;
 }
include/linux/sunrpc/svc.h +1 -0
index 9f9f699..815dd58 100644
@@ -230,6 +230,7 @@ struct svc_rqst {
 	struct svc_cred		rq_cred;	/* auth info */
 	void *			rq_xprt_ctxt;	/* transport specific context ptr */
 	struct svc_deferred_req*rq_deferred;	/* deferred request we are replaying */
+	int			rq_usedeferral;	/* use deferral */
 
 	size_t			rq_xprt_hlen;	/* xprt header len */
 	struct xdr_buf		rq_arg;
net/sunrpc/svc.c +2 -0
index fff09a2..45984cb 100644
@@ -1023,6 +1023,8 @@ svc_process(struct svc_rqst *rqstp)
 	rqstp->rq_res.tail[0].iov_len = 0;
 	/* Will be turned off only in gss privacy case: */
 	rqstp->rq_splice_ok = 1;
+	/* Will be turned off only when NFSv4 Sessions are used */
+	rqstp->rq_usedeferral = 1;
 
 	/* Setup reply header */
 	rqstp->rq_xprt->xpt_ops->xpo_prep_reply_hdr(rqstp);
net/sunrpc/svc_xprt.c +1 -1
index 1e66f24..600d091 100644
@@ -974,7 +974,7 @@ static struct cache_deferred_req *svc_defer(struct cache_req *req)
 	struct svc_rqst *rqstp = container_of(req, struct svc_rqst, rq_chandle);
 	struct svc_deferred_req *dr;
 
-	if (rqstp->rq_arg.page_len)
+	if (rqstp->rq_arg.page_len || !rqstp->rq_usedeferral)
 		return NULL; /* if more than a page, give up FIXME */
 	if (rqstp->rq_deferred) {
 		dr = rqstp->rq_deferred;
Signature · 1 line
1.6.2.1
#63 lore raw

PATCH v2 2/47 nfsd: embed nfsd4_current_state in nfsd4_compoundres

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

Remove the allocation of struct nfsd4_compound_state.

---
 fs/nfsd/nfs4proc.c        |   40 +++++++++-------------------------------
 include/linux/nfsd/xdr4.h |    9 +++++----
 2 files changed, 14 insertions(+), 35 deletions(-)
fs/nfsd/nfs4proc.c +9 -31
index 7839654..c8e1ad0 100644
@@ -809,29 +809,6 @@ static inline void nfsd4_increment_op_stats(u32 opnum)
 		nfsdstats.nfs4_opcount[opnum]++;
 }
 
-static void cstate_free(struct nfsd4_compound_state *cstate)
-{
-	if (cstate == NULL)
-		return;
-	fh_put(&cstate->current_fh);
-	fh_put(&cstate->save_fh);
-	BUG_ON(cstate->replay_owner);
-	kfree(cstate);
-}
-
-static struct nfsd4_compound_state *cstate_alloc(void)
-{
-	struct nfsd4_compound_state *cstate;
-
-	cstate = kmalloc(sizeof(struct nfsd4_compound_state), GFP_KERNEL);
-	if (cstate == NULL)
-		return NULL;
-	fh_init(&cstate->current_fh, NFS4_FHSIZE);
-	fh_init(&cstate->save_fh, NFS4_FHSIZE);
-	cstate->replay_owner = NULL;
-	return cstate;
-}
-
 typedef __be32(*nfsd4op_func)(struct svc_rqst *, struct nfsd4_compound_state *,
 			      void *);
 
@@ -859,12 +836,13 @@ nfsd4_proc_compound(struct svc_rqst *rqstp,
 {
 	struct nfsd4_op	*op;
 	struct nfsd4_operation *opdesc;
-	struct nfsd4_compound_state *cstate = NULL;
+	struct nfsd4_compound_state *cstate = &resp->cstate;
 	int		slack_bytes;
 	__be32		status;
 
 	resp->xbuf = &rqstp->rq_res;
-	resp->p = rqstp->rq_res.head[0].iov_base + rqstp->rq_res.head[0].iov_len;
+	resp->p = rqstp->rq_res.head[0].iov_base +
+						rqstp->rq_res.head[0].iov_len;
 	resp->tagp = resp->p;
 	/* reserve space for: taglen, tag, and opcnt */
 	resp->p += 2 + XDR_QUADLEN(args->taglen);
@@ -873,6 +851,9 @@ nfsd4_proc_compound(struct svc_rqst *rqstp,
 	resp->tag = args->tag;
 	resp->opcnt = 0;
 	resp->rqstp = rqstp;
+	resp->cstate.replay_owner = NULL;
+	fh_init(&resp->cstate.current_fh, NFS4_FHSIZE);
+	fh_init(&resp->cstate.save_fh, NFS4_FHSIZE);
 	/* Use the deferral mechanism only for NFSv4.0 compounds */
 	rqstp->rq_usedeferral = (args->minorversion == 0);
 
@@ -883,11 +864,6 @@ nfsd4_proc_compound(struct svc_rqst *rqstp,
 	if (args->minorversion > NFSD_SUPPORTED_MINOR_VERSION)
 		goto out;
 
-	status = nfserr_resource;
-	cstate = cstate_alloc();
-	if (cstate == NULL)
-		goto out;
-
 	status = nfs_ok;
 	while (!status && resp->opcnt < args->opcnt) {
 		op = &args->ops[resp->opcnt++];
@@ -964,7 +940,9 @@ encode_op:
 		status = nfserr_jukebox;
 	}
 
-	cstate_free(cstate);
+	fh_put(&resp->cstate.current_fh);
+	fh_put(&resp->cstate.save_fh);
+	BUG_ON(resp->cstate.replay_owner);
 out:
 	nfsd4_release_compoundargs(args);
 	/* Reset deferral mechanism for RPC deferrals */
include/linux/nfsd/xdr4.h +5 -4
index 27bd3e3..fd15ddc 100644
@@ -45,9 +45,9 @@
 #define XDR_LEN(n)                     (((n) + 3) & ~3)
 
 struct nfsd4_compound_state {
-	struct svc_fh current_fh;
-	struct svc_fh save_fh;
-	struct nfs4_stateowner *replay_owner;
+	struct svc_fh		current_fh;
+	struct svc_fh		save_fh;
+	struct nfs4_stateowner	*replay_owner;
 };
 
 struct nfsd4_change_info {
@@ -416,7 +416,8 @@ struct nfsd4_compoundres {
 	u32				taglen;
 	char *				tag;
 	u32				opcnt;
-	__be32 *			tagp; /* where to encode tag and  opcount */
+	__be32 *			tagp; /* tag, opcount encode location */
+	struct nfsd4_compound_state	cstate;
 };
 
 #define NFS4_SVC_XDRSIZE		sizeof(struct nfsd4_compoundargs)
Signature · 1 line
1.6.2.1
#65 lore raw

PATCH v2 3/47 nfsd: add a struct nfsd4_slot pointer to struct nfsd4_compound_state

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

To be used for sessions DRC replay.

---
 include/linux/nfsd/xdr4.h |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)
include/linux/nfsd/xdr4.h +2 -0
index fd15ddc..e309d55 100644
@@ -48,6 +48,8 @@ struct nfsd4_compound_state {
 	struct svc_fh		current_fh;
 	struct svc_fh		save_fh;
 	struct nfs4_stateowner	*replay_owner;
+	/* For sessions DRC */
+	struct nfsd4_slot	*slot;
 };
 
 struct nfsd4_change_info {
Signature · 1 line
1.6.2.1
#68 lore raw

PATCH v2 4/47 nfs41: common protocol definitions

In reply to Benny Halevy

Define all NFSv4.1 common operation and error code constants.

Note that some of the definitions are used by both the nfs41 client and the server code. This patch is duplicated in the nfs41 and nfsd41 sessions patchset.

[nfs41: add exchange id flags]
[removed server-only hunk changing NFSERR_REPLAY_ME]
[nfs41: add SEQ4_XX to nfs41-common-protocol]
[nfs41: generic error code update]
---
 include/linux/nfs4.h |  128 +++++++++++++++++++++++++++++++++++++++++++++++++-
 1 files changed, 127 insertions(+), 1 deletions(-)
include/linux/nfs4.h +127 -1
index b912311..a34683e 100644
@@ -21,6 +21,7 @@
 #define NFS4_FHSIZE		128
 #define NFS4_MAXPATHLEN		PATH_MAX
 #define NFS4_MAXNAMLEN		NAME_MAX
+#define NFS4_MAX_SESSIONID_LEN	16
 
 #define NFS4_ACCESS_READ        0x0001
 #define NFS4_ACCESS_LOOKUP      0x0002
@@ -38,6 +39,7 @@
 #define NFS4_OPEN_RESULT_CONFIRM 0x0002
 #define NFS4_OPEN_RESULT_LOCKTYPE_POSIX 0x0004
 
+#define NFS4_SHARE_ACCESS_MASK	0x000F
 #define NFS4_SHARE_ACCESS_READ	0x0001
 #define NFS4_SHARE_ACCESS_WRITE	0x0002
 #define NFS4_SHARE_ACCESS_BOTH	0x0003
@@ -45,6 +47,19 @@
 #define NFS4_SHARE_DENY_WRITE	0x0002
 #define NFS4_SHARE_DENY_BOTH	0x0003
 
+/* nfs41 */
+#define NFS4_SHARE_WANT_MASK		0xFF00
+#define NFS4_SHARE_WANT_NO_PREFERENCE	0x0000
+#define NFS4_SHARE_WANT_READ_DELEG	0x0100
+#define NFS4_SHARE_WANT_WRITE_DELEG	0x0200
+#define NFS4_SHARE_WANT_ANY_DELEG	0x0300
+#define NFS4_SHARE_WANT_NO_DELEG	0x0400
+#define NFS4_SHARE_WANT_CANCEL		0x0500
+
+#define NFS4_SHARE_WHEN_MASK		0xF0000
+#define NFS4_SHARE_SIGNAL_DELEG_WHEN_RESRC_AVAIL	0x10000
+#define NFS4_SHARE_PUSH_DELEG_WHEN_UNCONTENDED		0x20000
+
 #define NFS4_SET_TO_SERVER_TIME	0
 #define NFS4_SET_TO_CLIENT_TIME	1
 
@@ -88,6 +103,31 @@
 #define NFS4_ACE_GENERIC_EXECUTE              0x001200A0
 #define NFS4_ACE_MASK_ALL                     0x001F01FF
 
+#define EXCHGID4_FLAG_SUPP_MOVED_REFER		0x00000001
+#define EXCHGID4_FLAG_SUPP_MOVED_MIGR		0x00000002
+#define EXCHGID4_FLAG_USE_NON_PNFS		0x00010000
+#define EXCHGID4_FLAG_USE_PNFS_MDS		0x00020000
+#define EXCHGID4_FLAG_USE_PNFS_DS		0x00040000
+#define EXCHGID4_FLAG_UPD_CONFIRMED_REC_A	0x40000000
+#define EXCHGID4_FLAG_CONFIRMED_R		0x80000000
+/*
+ * Since the validity of these bits depends on whether
+ * they're set in the argument or response, have separate
+ * invalid flag masks for arg (_A) and resp (_R).
+ */
+#define EXCHGID4_INVAL_FLAG_MASK_A		0xBFF8FFFC
+#define EXCHGID4_INVAL_FLAG_MASK_R		0x7FF8FFFC
+
+#define SEQ4_STATUS_CB_PATH_DOWN		0x00000001
+#define SEQ4_STATUS_CB_GSS_CONTEXTS_EXPIRING	0x00000002
+#define SEQ4_STATUS_CB_GSS_CONTEXTS_EXPIRED	0x00000004
+#define SEQ4_STATUS_EXPIRED_ALL_STATE_REVOKED	0x00000008
+#define SEQ4_STATUS_EXPIRED_SOME_STATE_REVOKED	0x00000010
+#define SEQ4_STATUS_ADMIN_STATE_REVOKED		0x00000020
+#define SEQ4_STATUS_RECALLABLE_STATE_REVOKED	0x00000040
+#define SEQ4_STATUS_LEASE_MOVED			0x00000080
+#define SEQ4_STATUS_RESTART_RECLAIM_NEEDED	0x00000100
+
 #define NFS4_MAX_UINT64	(~(u64)0)
 
 enum nfs4_acl_whotype {
@@ -154,6 +194,28 @@ enum nfs_opnum4 {
 	OP_VERIFY = 37,
 	OP_WRITE = 38,
 	OP_RELEASE_LOCKOWNER = 39,
+
+	/* nfs41 */
+	OP_BACKCHANNEL_CTL = 40,
+	OP_BIND_CONN_TO_SESSION = 41,
+	OP_EXCHANGE_ID = 42,
+	OP_CREATE_SESSION = 43,
+	OP_DESTROY_SESSION = 44,
+	OP_FREE_STATEID = 45,
+	OP_GET_DIR_DELEGATION = 46,
+	OP_GETDEVICEINFO = 47,
+	OP_GETDEVICELIST = 48,
+	OP_LAYOUTCOMMIT = 49,
+	OP_LAYOUTGET = 50,
+	OP_LAYOUTRETURN = 51,
+	OP_SECINFO_NO_NAME = 52,
+	OP_SEQUENCE = 53,
+	OP_SET_SSV = 54,
+	OP_TEST_STATEID = 55,
+	OP_WANT_DELEGATION = 56,
+	OP_DESTROY_CLIENTID = 57,
+	OP_RECLAIM_COMPLETE = 58,
+
 	OP_ILLEGAL = 10044,
 };
 
@@ -230,7 +292,48 @@ enum nfsstat4 {
 	NFS4ERR_DEADLOCK = 10045,
 	NFS4ERR_FILE_OPEN = 10046,
 	NFS4ERR_ADMIN_REVOKED = 10047,
-	NFS4ERR_CB_PATH_DOWN = 10048
+	NFS4ERR_CB_PATH_DOWN = 10048,
+
+	/* nfs41 */
+	NFS4ERR_BADIOMODE	= 10049,
+	NFS4ERR_BADLAYOUT	= 10050,
+	NFS4ERR_BAD_SESSION_DIGEST = 10051,
+	NFS4ERR_BADSESSION	= 10052,
+	NFS4ERR_BADSLOT		= 10053,
+	NFS4ERR_COMPLETE_ALREADY = 10054,
+	NFS4ERR_CONN_NOT_BOUND_TO_SESSION = 10055,
+	NFS4ERR_DELEG_ALREADY_WANTED = 10056,
+	NFS4ERR_BACK_CHAN_BUSY	= 10057,	/* backchan reqs outstanding */
+	NFS4ERR_LAYOUTTRYLATER	= 10058,
+	NFS4ERR_LAYOUTUNAVAILABLE = 10059,
+	NFS4ERR_NOMATCHING_LAYOUT = 10060,
+	NFS4ERR_RECALLCONFLICT	= 10061,
+	NFS4ERR_UNKNOWN_LAYOUTTYPE = 10062,
+	NFS4ERR_SEQ_MISORDERED = 10063, 	/* unexpected seq.id in req */
+	NFS4ERR_SEQUENCE_POS	= 10064,	/* [CB_]SEQ. op not 1st op */
+	NFS4ERR_REQ_TOO_BIG	= 10065,	/* request too big */
+	NFS4ERR_REP_TOO_BIG	= 10066,	/* reply too big */
+	NFS4ERR_REP_TOO_BIG_TO_CACHE = 10067,	/* rep. not all cached */
+	NFS4ERR_RETRY_UNCACHED_REP = 10068,	/* retry & rep. uncached */
+	NFS4ERR_UNSAFE_COMPOUND = 10069,	/* retry/recovery too hard */
+	NFS4ERR_TOO_MANY_OPS	= 10070,	/* too many ops in [CB_]COMP */
+	NFS4ERR_OP_NOT_IN_SESSION = 10071,	/* op needs [CB_]SEQ. op */
+	NFS4ERR_HASH_ALG_UNSUPP = 10072,	/* hash alg. not supp. */
+						/* Error 10073 is unused. */
+	NFS4ERR_CLIENTID_BUSY	= 10074,	/* clientid has state */
+	NFS4ERR_PNFS_IO_HOLE	= 10075,	/* IO to _SPARSE file hole */
+	NFS4ERR_SEQ_FALSE_RETRY	= 10076,	/* retry not origional */
+	NFS4ERR_BAD_HIGH_SLOT	= 10077,	/* sequence arg bad */
+	NFS4ERR_DEADSESSION	= 10078,	/* persistent session dead */
+	NFS4ERR_ENCR_ALG_UNSUPP = 10079,	/* SSV alg mismatch */
+	NFS4ERR_PNFS_NO_LAYOUT	= 10080,	/* direct I/O with no layout */
+	NFS4ERR_NOT_ONLY_OP	= 10081,	/* bad compound */
+	NFS4ERR_WRONG_CRED	= 10082,	/* permissions:state change */
+	NFS4ERR_WRONG_TYPE	= 10083,	/* current operation mismatch */
+	NFS4ERR_DIRDELEG_UNAVAIL = 10084,	/* no directory delegation */
+	NFS4ERR_REJECT_DELEG	= 10085,	/* on callback */
+	NFS4ERR_RETURNCONFLICT	= 10086,	/* outstanding layoutreturn */
+	NFS4ERR_DELEG_REVOKED	= 10087,	/* deleg./layout revoked */
 };
 
 /*
@@ -391,6 +494,29 @@ enum {
 	NFSPROC4_CLNT_GETACL,
 	NFSPROC4_CLNT_SETACL,
 	NFSPROC4_CLNT_FS_LOCATIONS,
+
+	/* nfs41 */
+	NFSPROC4_CLNT_EXCHANGE_ID,
+	NFSPROC4_CLNT_CREATE_SESSION,
+	NFSPROC4_CLNT_DESTROY_SESSION,
+	NFSPROC4_CLNT_SEQUENCE,
+	NFSPROC4_CLNT_GET_LEASE_TIME,
+};
+
+/* nfs41 types */
+struct nfs4_sessionid {
+	unsigned char data[NFS4_MAX_SESSIONID_LEN];
+};
+
+/* Create Session Flags */
+#define SESSION4_PERSIST	 0x001
+#define SESSION4_BACK_CHAN 	 0x002
+#define SESSION4_RDMA		 0x004
+
+enum state_protect_how4 {
+	SP4_NONE	= 0,
+	SP4_MACH_CRED	= 1,
+	SP4_SSV		= 2
 };
 
 #endif
Signature · 1 line
1.6.2.1
#71 lore raw

PATCH v2 5/47 nfsd41: change NFSERR_REPLAY_ME

In reply to Benny Halevy

so not to overlap nfs41 error codes.

---
 include/linux/nfs.h |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)
include/linux/nfs.h +2 -1
index 54af92c..1510a64 100644
@@ -109,7 +109,8 @@
 	NFSERR_FILE_OPEN = 10046,      /*       v4 */
 	NFSERR_ADMIN_REVOKED = 10047,  /*       v4 */
 	NFSERR_CB_PATH_DOWN = 10048,   /*       v4 */
-	NFSERR_REPLAY_ME = 10049	/*       v4 */
+
+	NFSERR_REPLAY_ME = 11001,	/* linux internal */
 };
 
 /* NFSv2 file types - beware, these are not the same in NFSv3 */
Signature · 1 line
1.6.2.1
#72 lore raw

PATCH v2 6/47 nfsd41: Add Kconfig symbols for NFSv4.1

In reply to Benny Halevy

Added CONFIG_NFSD_V4_1 and made it depend upon NFSD_V4 and EXPERIMENTAL Indicate that CONFIG_NFS_V4_1 is for NFS developers at the moment

At the moment we're expecting folks trying out nfs41 to actively participate in the development process by helping us debug issues and ideally send patches to fix problems.

[trailing space fixed]
---
 fs/nfsd/Kconfig |    9 +++++++++
 1 files changed, 9 insertions(+), 0 deletions(-)
fs/nfsd/Kconfig +9 -0
index 503b9da..1f2f893 100644
@@ -79,3 +79,12 @@ config NFSD_V4
 	  available from http://linux-nfs.org/.
 
 	  If unsure, say N.
+
+config NFSD_V4_1
+	bool "NFS server support for NFSv4.1 (DEVELOPER ONLY)"
+	depends on NFSD_V4 && EXPERIMENTAL
+	help
+	  This option enables support for minor version 1 of the NFSv4 protocol
+	  (draft-ietf-nfsv4-minorversion1) in the kernel's NFS server.
+
+	  Unless you're an NFS developer, say N.
Signature · 1 line
1.6.2.1
#99 lore raw

PATCH v2 7/47 nfsd41: define nfs41 error codes

In reply to Benny Halevy

From: Marc Eshel <[email protected]>

Define all error code present in
http://tools.ietf.org/html/draft-ietf-nfsv4-minorversion1-29.

[nfsd41: clean up error code definitions]
---
 include/linux/nfs.h       |    2 --
 include/linux/nfsd/nfsd.h |   43 ++++++++++++++++++++++++++++++++++++++++++-
 2 files changed, 42 insertions(+), 3 deletions(-)
include/linux/nfs.h +0 -2
index 1510a64..214d499 100644
@@ -109,8 +109,6 @@
 	NFSERR_FILE_OPEN = 10046,      /*       v4 */
 	NFSERR_ADMIN_REVOKED = 10047,  /*       v4 */
 	NFSERR_CB_PATH_DOWN = 10048,   /*       v4 */
-
-	NFSERR_REPLAY_ME = 11001,	/* linux internal */
 };
 
 /* NFSv2 file types - beware, these are not the same in NFSv3 */
include/linux/nfsd/nfsd.h +42 -1

55 lines not on this page · raw message

Signature · 1 line
1.6.2.1
#103 lore raw

PATCH v2 8/47 nfsd41: sessions basic data types

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

This patch provides basic data structures representing the nfs41 sessions and slots, plus helpers for keeping a reference count on the session and freeing it.

Note that our server only support a headerpadsz of 0 and it ignores backchannel attributes at the moment.

[nfsd41: remove headerpadsz from channel attributes]
[nfsd41: embed nfsd4_channel in nfsd4_session]
[nfsd41: use bool inuse for slot state]
---
 fs/nfsd/nfs4state.c        |   22 ++++++++++++++++++++++
 include/linux/nfsd/state.h |   34 ++++++++++++++++++++++++++++++++++
 2 files changed, 56 insertions(+), 0 deletions(-)
fs/nfsd/nfs4state.c +22 -0
index 070e9e5..f4b624b 100644
@@ -382,6 +382,28 @@ static void release_openowner(struct nfs4_stateowner *sop)
 	nfs4_put_stateowner(sop);
 }
 
+#if defined(CONFIG_NFSD_V4_1)
+static void
+release_session(struct nfsd4_session *ses)
+{
+	list_del(&ses->se_hash);
+	list_del(&ses->se_perclnt);
+	nfsd4_put_session(ses);
+}
+
+void
+free_session(struct kref *kref)
+{
+	struct nfsd4_session *ses;
+
+	ses = container_of(kref, struct nfsd4_session, se_ref);
+	kfree(ses->se_slots);
+	kfree(ses);
+}
+#else /* CONFIG_NFSD_V4_1 */
+static inline void release_session(struct nfsd4_session *ses) {}
+#endif /* CONFIG_NFSD_V4_1 */
+
 static inline void
 renew_client(struct nfs4_client *clp)
 {
include/linux/nfsd/state.h +34 -0
index a6e4a00..afae103 100644
@@ -99,6 +99,40 @@ struct nfs4_callback {
 	struct rpc_clnt *       cb_client;
 };
 
+struct nfsd4_slot {
+	bool				sl_inuse;
+	struct nfsd4_session		*sl_session;
+	u32				sl_seqid;
+};
+
+struct nfsd4_session {
+	struct kref		se_ref;
+	struct list_head	se_hash;	/* hash by sessionid */
+	struct list_head	se_perclnt;
+	u32			se_flags;
+	struct nfs4_client	*se_client;	/* for expire_client */
+	struct nfs4_sessionid	se_sessionid;
+	u32			se_fmaxreq_sz;
+	u32			se_fmaxresp_sz;
+	u32			se_fmaxresp_cached;
+	u32			se_fmaxops;
+	u32			se_fnumslots;
+	struct nfsd4_slot	*se_slots;	/* forward channel slots */
+};
+
+static inline void
+nfsd4_put_session(struct nfsd4_session *ses)
+{
+	extern void free_session(struct kref *kref);
+	kref_put(&ses->se_ref, free_session);
+}
+
+static inline void
+nfsd4_get_session(struct nfsd4_session *ses)
+{
+	kref_get(&ses->se_ref);
+}
+
 #define HEXDIR_LEN     33 /* hex version of 16 byte md5 of cl_name plus '\0' */
 
 /*
Signature · 1 line
1.6.2.1
#104 lore raw

PATCH v2 9/47 nfsd41: introduce nfs4_client cl_sessions list

In reply to Benny Halevy

From: Marc Eshel <[email protected]>

---
 fs/nfsd/nfs4state.c        |    3 +++
 include/linux/nfsd/state.h |    3 +++
 2 files changed, 6 insertions(+), 0 deletions(-)
fs/nfsd/nfs4state.c +3 -0
index f4b624b..b71ec72 100644
@@ -532,6 +532,9 @@ static struct nfs4_client *create_client(struct xdr_netobj name, char *recdir)
 	INIT_LIST_HEAD(&clp->cl_strhash);
 	INIT_LIST_HEAD(&clp->cl_openowners);
 	INIT_LIST_HEAD(&clp->cl_delegations);
+#if defined(CONFIG_NFSD_V4_1)
+	INIT_LIST_HEAD(&clp->cl_sessions);
+#endif /* CONFIG_NFSD_V4_1 */
 	INIT_LIST_HEAD(&clp->cl_lru);
 	return clp;
 }
include/linux/nfsd/state.h +3 -0
index afae103..29624b4 100644
@@ -164,6 +164,9 @@ struct nfs4_client {
 	struct nfs4_callback	cl_callback;    /* callback info */
 	atomic_t		cl_count;	/* ref count */
 	u32			cl_firststate;	/* recovery dir creation */
+#ifdef CONFIG_NFSD_V4_1
+	struct list_head	cl_sessions;
+#endif /* CONFIG_NFSD_V4_1 */
 };
 
 /* struct nfs4_client_reset
Signature · 1 line
1.6.2.1
#105 lore raw

PATCH v2 10/47 nfsd41: release_session when client is expired

In reply to Benny Halevy

From: Marc Eshel <[email protected]>

[add CONFIG_NFSD_V4_1 to fix v4.0 regression bug]
---
 fs/nfsd/nfs4state.c |    8 ++++++++
 1 files changed, 8 insertions(+), 0 deletions(-)
fs/nfsd/nfs4state.c +8 -0
index b71ec72..df9d42e 100644
@@ -515,6 +515,14 @@ expire_client(struct nfs4_client *clp)
 		sop = list_entry(clp->cl_openowners.next, struct nfs4_stateowner, so_perclient);
 		release_openowner(sop);
 	}
+#ifdef CONFIG_NFSD_V4_1
+	while (!list_empty(&clp->cl_sessions)) {
+		struct nfsd4_session  *ses;
+		ses = list_entry(clp->cl_sessions.next, struct nfsd4_session,
+				 se_perclnt);
+		release_session(ses);
+	}
+#endif /* CONFIG_NFSD_V4_1 */
 	put_nfs4_client(clp);
 }
 
Signature · 1 line
1.6.2.1
#106 lore raw

PATCH v2 11/47 nfsd41: sessionid hashing

In reply to Benny Halevy

From: Marc Eshel <[email protected]>

Simple sessionid hashing using its monotonically increasing sequence number.

Locking considerations:
sessionid_hashtbl access is controlled by the sessionid_lock spin lock. It must be taken for insert, delete, and lookup. nfsd4_sequence looks up the session id and if the session is found, it calls nfsd4_get_session (still under the sessionid_lock). nfsd4_destroy_session calls nfsd4_put_session after unhashing it, so when the session's kref reaches zero it's going to get freed.

[we don't use a prime for sessionid hash table size]
[use sessionid_lock spin lock]
---
 fs/nfsd/nfs4state.c        |   57 +++++++++++++++++++++++++++++++++++++++++++-
 include/linux/nfsd/state.h |    7 +++++
 2 files changed, 63 insertions(+), 1 deletions(-)
fs/nfsd/nfs4state.c +56 -1
index df9d42e..ac4e8f2 100644
@@ -383,11 +383,62 @@ static void release_openowner(struct nfs4_stateowner *sop)
 }
 
 #if defined(CONFIG_NFSD_V4_1)
+static DEFINE_SPINLOCK(sessionid_lock);
+#define SESSION_HASH_SIZE	512
+static struct list_head sessionid_hashtbl[SESSION_HASH_SIZE];
+
+static inline int
+hash_sessionid(struct nfs4_sessionid *sessionid)
+{
+	struct nfsd4_sessionid *sid = (struct nfsd4_sessionid *)sessionid;
+
+	return sid->sequence % SESSION_HASH_SIZE;
+}
+
+static inline void
+dump_sessionid(const char *fn, struct nfs4_sessionid *sessionid)
+{
+	u32 *ptr = (u32 *)(&sessionid->data[0]);
+	dprintk("%s: %u:%u:%u:%u\n", fn, ptr[0], ptr[1], ptr[2], ptr[3]);
+}
+
+/* caller must hold sessionid_lock */
+static struct nfsd4_session *
+find_in_sessionid_hashtbl(struct nfs4_sessionid *sessionid)
+{
+	struct nfsd4_session *elem;
+	int idx;
+
+	dump_sessionid(__func__, sessionid);
+	idx = hash_sessionid(sessionid);
+	dprintk("%s: idx is %d\n", __func__, idx);
+	/* Search in the appropriate list */
+	list_for_each_entry(elem, &sessionid_hashtbl[idx], se_hash) {
+		dump_sessionid("list traversal", &elem->se_sessionid);
+		if (!memcmp(elem->se_sessionid.data, sessionid->data,
+			    NFS4_MAX_SESSIONID_LEN)) {
+			return elem;
+		}
+	}
+
+	dprintk("%s: session not found\n", __func__);
+	return NULL;
+}
+
+/* caller must hold sessionid_lock */
 static void
-release_session(struct nfsd4_session *ses)
+unhash_session(struct nfsd4_session *ses)
 {
 	list_del(&ses->se_hash);
 	list_del(&ses->se_perclnt);
+}
+
+static void
+release_session(struct nfsd4_session *ses)
+{
+	spin_lock(&sessionid_lock);
+	unhash_session(ses);
+	spin_unlock(&sessionid_lock);
 	nfsd4_put_session(ses);
 }
 
@@ -3213,6 +3264,10 @@ nfs4_state_init(void)
 		INIT_LIST_HEAD(&unconf_str_hashtbl[i]);
 		INIT_LIST_HEAD(&unconf_id_hashtbl[i]);
 	}
+#if defined(CONFIG_NFSD_V4_1)
+	for (i = 0; i < SESSION_HASH_SIZE; i++)
+		INIT_LIST_HEAD(&sessionid_hashtbl[i]);
+#endif /* CONFIG_NFSD_V4_1 */
 	for (i = 0; i < FILE_HASH_SIZE; i++) {
 		INIT_LIST_HEAD(&file_hashtbl[i]);
 	}
include/linux/nfsd/state.h +7 -0
index 29624b4..7592d7b 100644
@@ -133,6 +133,13 @@ nfsd4_get_session(struct nfsd4_session *ses)
 	kref_get(&ses->se_ref);
 }
 
+/* formatted contents of nfs4_sessionid */
+struct nfsd4_sessionid {
+	clientid_t	clientid;
+	u32		sequence;
+	u32		reserved;
+};
+
 #define HEXDIR_LEN     33 /* hex version of 16 byte md5 of cl_name plus '\0' */
 
 /*
Signature · 1 line
1.6.2.1
#111 lore raw

PATCH v2 12/47 nfsd41: xdr infrastructure

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

Define nfsd41_dec_ops vector and add it to nfsd4_minorversion for minorversion 1.

Note: nfsd4_enc_ops vector is shared for v4.0 and v4.1 since we don't need to filter out obsolete ops as this is done in the decoding phase.

exchange_id, create_session, destroy_session, and sequence ops are implemented as stubs returning nfserr_opnotsupp at this stage.

[was nfsd41: xdr stubs]
---
 fs/nfsd/nfs4xdr.c         |  156 +++++++++++++++++++++++++++++++++++++++++++++
 include/linux/nfsd/xdr4.h |   24 +++++++
 2 files changed, 180 insertions(+), 0 deletions(-)
fs/nfsd/nfs4xdr.c +156 -0
index 76a0b2a..b082d07 100644
@@ -996,6 +996,36 @@ nfsd4_decode_release_lockowner(struct nfsd4_compoundargs *argp, struct nfsd4_rel
 	DECODE_TAIL;
 }
 
+#if defined(CONFIG_NFSD_V4_1)
+static __be32
+nfsd4_decode_exchange_id(struct nfsd4_compoundargs *argp,
+			 struct nfsd4_exchange_id *clid)
+{
+	return nfserr_opnotsupp;	/* stub */
+}
+
+static __be32
+nfsd4_decode_create_session(struct nfsd4_compoundargs *argp,
+			    struct nfsd4_create_session *sess)
+{
+	return nfserr_opnotsupp;	/* stub */
+}
+
+static __be32
+nfsd4_decode_destroy_session(struct nfsd4_compoundargs *argp,
+			     struct nfsd4_destroy_session *destroy_session)
+{
+	return nfserr_opnotsupp;	/* stub */
+}
+
+static __be32
+nfsd4_decode_sequence(struct nfsd4_compoundargs *argp,
+		      struct nfsd4_sequence *seq)
+{
+	return nfserr_opnotsupp;	/* stub */
+}
+#endif /* CONFIG_NFSD_V4_1 */
+
 static __be32
 nfsd4_decode_noop(struct nfsd4_compoundargs *argp, void *p)
 {
@@ -1050,6 +1080,69 @@ static nfsd4_dec nfsd4_dec_ops[] = {
 	[OP_RELEASE_LOCKOWNER]	= (nfsd4_dec)nfsd4_decode_release_lockowner,
 };
 
+#if defined(CONFIG_NFSD_V4_1)
+static nfsd4_dec nfsd41_dec_ops[] = {
+	[OP_ACCESS]		(nfsd4_dec)nfsd4_decode_access,
+	[OP_CLOSE]		(nfsd4_dec)nfsd4_decode_close,
+	[OP_COMMIT]		(nfsd4_dec)nfsd4_decode_commit,
+	[OP_CREATE]		(nfsd4_dec)nfsd4_decode_create,
+	[OP_DELEGPURGE]		(nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_DELEGRETURN]	(nfsd4_dec)nfsd4_decode_delegreturn,
+	[OP_GETATTR]		(nfsd4_dec)nfsd4_decode_getattr,
+	[OP_GETFH]		(nfsd4_dec)nfsd4_decode_noop,
+	[OP_LINK]		(nfsd4_dec)nfsd4_decode_link,
+	[OP_LOCK]		(nfsd4_dec)nfsd4_decode_lock,
+	[OP_LOCKT]		(nfsd4_dec)nfsd4_decode_lockt,
+	[OP_LOCKU]		(nfsd4_dec)nfsd4_decode_locku,
+	[OP_LOOKUP]		(nfsd4_dec)nfsd4_decode_lookup,
+	[OP_LOOKUPP]		(nfsd4_dec)nfsd4_decode_noop,
+	[OP_NVERIFY]		(nfsd4_dec)nfsd4_decode_verify,
+	[OP_OPEN]		(nfsd4_dec)nfsd4_decode_open,
+	[OP_OPENATTR]		(nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_OPEN_CONFIRM]	(nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_OPEN_DOWNGRADE]	(nfsd4_dec)nfsd4_decode_open_downgrade,
+	[OP_PUTFH]		(nfsd4_dec)nfsd4_decode_putfh,
+	[OP_PUTPUBFH]		(nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_PUTROOTFH]		(nfsd4_dec)nfsd4_decode_noop,
+	[OP_READ]		(nfsd4_dec)nfsd4_decode_read,
+	[OP_READDIR]		(nfsd4_dec)nfsd4_decode_readdir,
+	[OP_READLINK]		(nfsd4_dec)nfsd4_decode_noop,
+	[OP_REMOVE]		(nfsd4_dec)nfsd4_decode_remove,
+	[OP_RENAME]		(nfsd4_dec)nfsd4_decode_rename,
+	[OP_RENEW]		(nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_RESTOREFH]		(nfsd4_dec)nfsd4_decode_noop,
+	[OP_SAVEFH]		(nfsd4_dec)nfsd4_decode_noop,
+	[OP_SECINFO]		(nfsd4_dec)nfsd4_decode_secinfo,
+	[OP_SETATTR]		(nfsd4_dec)nfsd4_decode_setattr,
+	[OP_SETCLIENTID]	(nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_SETCLIENTID_CONFIRM](nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_VERIFY]		(nfsd4_dec)nfsd4_decode_verify,
+	[OP_WRITE]		(nfsd4_dec)nfsd4_decode_write,
+	[OP_RELEASE_LOCKOWNER]	(nfsd4_dec)nfsd4_decode_notsupp,
+
+	/* new operations for NFSv4.1 */
+	[OP_BACKCHANNEL_CTL]	(nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_BIND_CONN_TO_SESSION](nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_EXCHANGE_ID]	(nfsd4_dec)nfsd4_decode_exchange_id,
+	[OP_CREATE_SESSION]	(nfsd4_dec)nfsd4_decode_create_session,
+	[OP_DESTROY_SESSION]	(nfsd4_dec)nfsd4_decode_destroy_session,
+	[OP_FREE_STATEID]	(nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_GET_DIR_DELEGATION]	(nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_GETDEVICEINFO]	(nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_GETDEVICELIST]	(nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_LAYOUTCOMMIT]	(nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_LAYOUTGET]		(nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_LAYOUTRETURN]	(nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_SECINFO_NO_NAME]	(nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_SEQUENCE]		(nfsd4_dec)nfsd4_decode_sequence,
+	[OP_SET_SSV]		(nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_TEST_STATEID]	(nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_WANT_DELEGATION]	(nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_DESTROY_CLIENTID]	(nfsd4_dec)nfsd4_decode_notsupp,
+	[OP_RECLAIM_COMPLETE]	(nfsd4_dec)nfsd4_decode_notsupp,
+};
+#endif /* CONFIG_NFSD_V4_1 */
+
 struct nfsd4_minorversion_ops {
 	nfsd4_dec *decoders;
 	int nops;
@@ -1057,6 +1150,9 @@ struct nfsd4_minorversion_ops {
 
 static struct nfsd4_minorversion_ops nfsd4_minorversion[] = {
 	[0] = { nfsd4_dec_ops, ARRAY_SIZE(nfsd4_dec_ops) },
+#if defined(CONFIG_NFSD_V4_1)
+	[1] = { nfsd41_dec_ops, ARRAY_SIZE(nfsd41_dec_ops) },
+#endif /* CONFIG_NFSD_V4_1 */
 };
 
 static __be32
@@ -2571,6 +2667,40 @@ nfsd4_encode_write(struct nfsd4_compoundres *resp, __be32 nfserr, struct nfsd4_w
 	return nfserr;
 }
 
+#if defined(CONFIG_NFSD_V4_1)
+static __be32
+nfsd4_encode_exchange_id(struct nfsd4_compoundres *resp, int nfserr,
+			 struct nfsd4_exchange_id *exid)
+{
+	/* stub */
+	return nfserr;
+}
+
+static __be32
+nfsd4_encode_create_session(struct nfsd4_compoundres *resp, int nfserr,
+			    struct nfsd4_create_session *sess)
+{
+	/* stub */
+	return nfserr;
+}
+
+static __be32
+nfsd4_encode_destroy_session(struct nfsd4_compoundres *resp, int nfserr,
+			     struct nfsd4_destroy_session *destroy_session)
+{
+	/* stub */
+	return nfserr;
+}
+
+static __be32
+nfsd4_encode_sequence(struct nfsd4_compoundres *resp, int nfserr,
+		      struct nfsd4_sequence *seq)
+{
+	/* stub */
+	return nfserr;
+}
+#endif /* CONFIG_NFSD_V4_1 */
+
 static __be32
 nfsd4_encode_noop(struct nfsd4_compoundres *resp, __be32 nfserr, void *p)
 {
@@ -2579,6 +2709,11 @@ nfsd4_encode_noop(struct nfsd4_compoundres *resp, __be32 nfserr, void *p)
 
 typedef __be32(* nfsd4_enc)(struct nfsd4_compoundres *, __be32, void *);
 
+/*
+ * Note: nfsd4_enc_ops vector is shared for v4.0 and v4.1
+ * since we don't need to filter out obsolete ops as this is
+ * done in the decoding phase.
+ */
 static nfsd4_enc nfsd4_enc_ops[] = {
 	[OP_ACCESS]		= (nfsd4_enc)nfsd4_encode_access,
 	[OP_CLOSE]		= (nfsd4_enc)nfsd4_encode_close,
@@ -2617,6 +2752,27 @@ static nfsd4_enc nfsd4_enc_ops[] = {
 	[OP_VERIFY]		= (nfsd4_enc)nfsd4_encode_noop,
 	[OP_WRITE]		= (nfsd4_enc)nfsd4_encode_write,
 	[OP_RELEASE_LOCKOWNER]	= (nfsd4_enc)nfsd4_encode_noop,
+#if defined(CONFIG_NFSD_V4_1)
+	[OP_BACKCHANNEL_CTL]	= (nfsd4_enc)nfsd4_encode_noop,
+	[OP_BIND_CONN_TO_SESSION] = (nfsd4_enc)nfsd4_encode_noop,
+	[OP_EXCHANGE_ID]	= (nfsd4_enc)nfsd4_encode_exchange_id,
+	[OP_CREATE_SESSION]	= (nfsd4_enc)nfsd4_encode_create_session,
+	[OP_DESTROY_SESSION]	= (nfsd4_enc)nfsd4_encode_destroy_session,
+	[OP_FREE_STATEID]	= (nfsd4_enc)nfsd4_encode_noop,
+	[OP_GET_DIR_DELEGATION]	= (nfsd4_enc)nfsd4_encode_noop,
+	[OP_GETDEVICEINFO]	= (nfsd4_enc)nfsd4_encode_noop,
+	[OP_GETDEVICELIST]	= (nfsd4_enc)nfsd4_encode_noop,
+	[OP_LAYOUTCOMMIT]	= (nfsd4_enc)nfsd4_encode_noop,
+	[OP_LAYOUTGET]		= (nfsd4_enc)nfsd4_encode_noop,
+	[OP_LAYOUTRETURN]	= (nfsd4_enc)nfsd4_encode_noop,
+	[OP_SECINFO_NO_NAME]	= (nfsd4_enc)nfsd4_encode_noop,
+	[OP_SEQUENCE]		= (nfsd4_enc)nfsd4_encode_sequence,
+	[OP_SET_SSV]		= (nfsd4_enc)nfsd4_encode_noop,
+	[OP_TEST_STATEID]	= (nfsd4_enc)nfsd4_encode_noop,
+	[OP_WANT_DELEGATION]	= (nfsd4_enc)nfsd4_encode_noop,
+	[OP_DESTROY_CLIENTID]	= (nfsd4_enc)nfsd4_encode_noop,
+	[OP_RECLAIM_COMPLETE]	= (nfsd4_enc)nfsd4_encode_noop,
+#endif /* CONFIG_NFSD_V4_1 */
 };
 
 void
include/linux/nfsd/xdr4.h +24 -0
index e309d55..40b148e 100644
@@ -346,6 +346,24 @@ struct nfsd4_write {
 	nfs4_verifier	wr_verifier;        /* response */
 };
 
+#if defined(CONFIG_NFSD_V4_1)
+struct nfsd4_exchange_id {
+	int	foo;	/* stub */
+};
+
+struct nfsd4_create_session {
+	int	foo;	/* stub */
+};
+
+struct nfsd4_sequence {
+	int	foo;	/* stub */
+};
+
+struct nfsd4_destroy_session {
+	int	foo;	/* stub */
+};
+#endif /* CONFIG_NFSD_V4_1 */
+
 struct nfsd4_op {
 	int					opnum;
 	__be32					status;
@@ -380,6 +398,12 @@ struct nfsd4_op {
 		struct nfsd4_verify		verify;
 		struct nfsd4_write		write;
 		struct nfsd4_release_lockowner	release_lockowner;
+#if defined(CONFIG_NFSD_V4_1)
+		struct nfsd4_exchange_id	exchange_id;
+		struct nfsd4_create_session	create_session;
+		struct nfsd4_destroy_session	destroy_session;
+		struct nfsd4_sequence		sequence;
+#endif /* CONFIG_NFSD_V4_1 */
 	} u;
 	struct nfs4_replay *			replay;
 };
Signature · 1 line
1.6.2.1
#112 lore raw

PATCH v2 13/47 nfsd: remove nfsd4_ops array size

In reply to Benny Halevy

There's no need for it.

---
 fs/nfsd/nfs4proc.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)
fs/nfsd/nfs4proc.c +1 -1
index c8e1ad0..ded469f 100644
@@ -951,7 +951,7 @@ out:
 	return status;
 }
 
-static struct nfsd4_operation nfsd4_ops[OP_RELEASE_LOCKOWNER+1] = {
+static struct nfsd4_operation nfsd4_ops[] = {
 	[OP_ACCESS] = {
 		.op_func = (nfsd4op_func)nfsd4_access,
 		.op_name = "OP_ACCESS",
Signature · 1 line
1.6.2.1
#113 lore raw

PATCH v2 14/47 nfsd41: proc stubs

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

---
 fs/nfsd/nfs4proc.c        |   22 ++++++++++++++++++++++
 fs/nfsd/nfs4state.c       |   34 ++++++++++++++++++++++++++++++++++
 include/linux/nfsd/xdr4.h |   14 ++++++++++++++
 3 files changed, 70 insertions(+), 0 deletions(-)
fs/nfsd/nfs4proc.c +22 -0
index ded469f..f618e8e 100644
@@ -1101,6 +1101,28 @@ static struct nfsd4_operation nfsd4_ops[] = {
 		.op_flags = ALLOWED_WITHOUT_FH | ALLOWED_ON_ABSENT_FS,
 		.op_name = "OP_RELEASE_LOCKOWNER",
 	},
+#if defined(CONFIG_NFSD_V4_1)
+	[OP_EXCHANGE_ID] = {
+		.op_func = (nfsd4op_func)nfsd4_exchange_id,
+		.op_flags = ALLOWED_WITHOUT_FH,
+		.op_name = "OP_EXCHANGE_ID",
+	},
+	[OP_CREATE_SESSION] = {
+		.op_func = (nfsd4op_func)nfsd4_create_session,
+		.op_flags = ALLOWED_WITHOUT_FH,
+		.op_name = "OP_CREATE_SESSION",
+	},
+	[OP_DESTROY_SESSION] = {
+		.op_func = (nfsd4op_func)nfsd4_destroy_session,
+		.op_flags = ALLOWED_WITHOUT_FH,
+		.op_name = "OP_DESTROY_SESSION",
+	},
+	[OP_SEQUENCE] = {
+		.op_func = (nfsd4op_func)nfsd4_sequence,
+		.op_flags = ALLOWED_WITHOUT_FH,
+		.op_name = "OP_SEQUENCE",
+	},
+#endif /* CONFIG_NFSD_V4_1 */
 };
 
 static const char *nfsd4_op_name(unsigned opnum)
fs/nfsd/nfs4state.c +34 -0
index ac4e8f2..bbb7455 100644
@@ -840,6 +840,40 @@ out_err:
 	return;
 }
 
+#if defined(CONFIG_NFSD_V4_1)
+__be32
+nfsd4_exchange_id(struct svc_rqst *rqstp,
+		  struct nfsd4_compound_state *cstate,
+		  struct nfsd4_exchange_id *exid)
+{
+	return -1;	/* stub */
+}
+
+__be32
+nfsd4_create_session(struct svc_rqst *rqstp,
+		     struct nfsd4_compound_state *cstate,
+		     struct nfsd4_create_session *cr_ses)
+{
+	return -1;	/* stub */
+}
+
+__be32
+nfsd4_destroy_session(struct svc_rqst *r,
+		      struct nfsd4_compound_state *cstate,
+		      struct nfsd4_destroy_session *sessionid)
+{
+	return -1;	/* stub */
+}
+
+__be32
+nfsd4_sequence(struct svc_rqst *r,
+	       struct nfsd4_compound_state *cstate,
+	       struct nfsd4_sequence *seq)
+{
+	return -1;	/* stub */
+}
+#endif /* CONFIG_NFSD_V4_1 */
+
 __be32
 nfsd4_setclientid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 		  struct nfsd4_setclientid *setclid)
include/linux/nfsd/xdr4.h +14 -0
index 40b148e..0148d54 100644
@@ -475,6 +475,20 @@ extern __be32 nfsd4_setclientid(struct svc_rqst *rqstp,
 extern __be32 nfsd4_setclientid_confirm(struct svc_rqst *rqstp,
 		struct nfsd4_compound_state *,
 		struct nfsd4_setclientid_confirm *setclientid_confirm);
+#if defined(CONFIG_NFSD_V4_1)
+extern __be32 nfsd4_exchange_id(struct svc_rqst *rqstp,
+		struct nfsd4_compound_state *,
+struct nfsd4_exchange_id *);
+		extern __be32 nfsd4_create_session(struct svc_rqst *,
+		struct nfsd4_compound_state *,
+		struct nfsd4_create_session *);
+extern __be32 nfsd4_sequence(struct svc_rqst *,
+		struct nfsd4_compound_state *,
+		struct nfsd4_sequence *);
+extern __be32 nfsd4_destroy_session(struct svc_rqst *,
+		struct nfsd4_compound_state *,
+		struct nfsd4_destroy_session *);
+#endif /* CONFIG_NFSD_V4_1 */
 extern __be32 nfsd4_process_open1(struct nfsd4_open *open);
 extern __be32 nfsd4_process_open2(struct svc_rqst *rqstp,
 		struct svc_fh *current_fh, struct nfsd4_open *open);
Signature · 1 line
1.6.2.1
#114 lore raw

PATCH v2 15/47 nfsd41: exchange_id operation

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

Implement the exchange_id operation confoming to
http://tools.ietf.org/html/draft-ietf-nfsv4-minorversion1-28

Based on the client provided name, hash a client id. If a confirmed one is found, compare the op's creds and verifier. If the creds match and the verifier is different then expire the old client (client re-incarnated), otherwise, if both match, assume it's a replay and ignore it.

If an unconfirmed client is found, then copy the new creds and verifer if need update, otherwise assume replay.

The client is moved to a confirmed state on create_session.

In the nfs41 branch set the exchange_id flags to EXCHGID4_FLAG_USE_NON_PNFS | EXCHGID4_FLAG_SUPP_MOVED_REFER (pNFS is not supported, Referrals are supported, Migration is not.).

Address various scenarios from section 18.35 of the spec:

1. Check for EXCHGID4_FLAG_UPD_CONFIRMED_REC_A and set EXCHGID4_FLAG_CONFIRMED_R as appropriate.

2. Return error codes per 18.35.4 scenarios.

3. Update client records or generate new client ids depending on scenario.

Note: 18.35.4 case 3 probably still needs revisiting. The handling seems not quite right.

[nfsd41: use utsname for major_id (and copy to server_scope)]
[nfsd41: fix handling of various exchange id scenarios]
---
 fs/nfsd/nfs4state.c        |  138 +++++++++++++++++++++++++++++++++++++++++-
 fs/nfsd/nfs4xdr.c          |  146 +++++++++++++++++++++++++++++++++++++++++++-
 include/linux/nfsd/state.h |    2 +
 include/linux/nfsd/xdr4.h  |    8 ++-
 4 files changed, 289 insertions(+), 5 deletions(-)
fs/nfsd/nfs4state.c +137 -1
index bbb7455..09c63ff 100644
@@ -841,12 +841,148 @@ out_err:
 }
 
 #if defined(CONFIG_NFSD_V4_1)
+/*
+ * Set the exchange_id flags returned by the server.
+ */
+static void
+nfsd4_set_ex_flags(struct nfs4_client *new, struct nfsd4_exchange_id *clid)
+{
+	/* pNFS is not supported */
+	new->cl_exchange_flags |= EXCHGID4_FLAG_USE_NON_PNFS;
+
+	/* Referrals are supported, Migration is not. */
+	new->cl_exchange_flags |= EXCHGID4_FLAG_SUPP_MOVED_REFER;
+
+	/* set the wire flags to return to client. */
+	clid->flags = new->cl_exchange_flags;
+}
+
 __be32
 nfsd4_exchange_id(struct svc_rqst *rqstp,
 		  struct nfsd4_compound_state *cstate,
 		  struct nfsd4_exchange_id *exid)
 {
-	return -1;	/* stub */
+	struct nfs4_client *unconf, *conf, *new;
+	int status;
+	unsigned int		strhashval;
+	char			dname[HEXDIR_LEN];
+	nfs4_verifier		verf = exid->verifier;
+	u32			ip_addr = svc_addr_in(rqstp)->sin_addr.s_addr;
+	struct xdr_netobj clname = {
+		.len = exid->id_len,
+		.data = exid->id,
+	};
+
+	dprintk("%s rqstp=%p exid=%p clname.len=%u clname.data=%p "
+		" ip_addr=%u flags %x, spa_how %d\n",
+		__func__, rqstp, exid, clname.len, clname.data,
+		ip_addr, exid->flags, exid->spa_how);
+
+	if (!check_name(clname) || (exid->flags & EXCHGID4_INVAL_FLAG_MASK_A))
+		return nfserr_inval;
+
+	/* Currently only support SP4_NONE */
+	if (exid->spa_how != SP4_NONE)
+		return nfserr_encr_alg_unsupp;
+
+	status = nfs4_make_rec_clidname(dname, &clname);
+
+	if (status)
+		goto error;
+
+	strhashval = clientstr_hashval(dname);
+
+	nfs4_lock_state();
+	status = nfs_ok;
+
+	conf = find_confirmed_client_by_str(dname, strhashval);
+	if (conf) {
+		if (!same_verf(&verf, &conf->cl_verifier)) {
+			/* 18.35.4 case 8 */
+			if (exid->flags & EXCHGID4_FLAG_UPD_CONFIRMED_REC_A) {
+				status = nfserr_not_same;
+				goto out;
+			}
+			/* Client reboot: destroy old state */
+			expire_client(conf);
+			goto out_new;
+		}
+		if (!same_creds(&conf->cl_cred, &rqstp->rq_cred)) {
+			/* 18.35.4 case 9 */
+			if (exid->flags & EXCHGID4_FLAG_UPD_CONFIRMED_REC_A) {
+				status = nfserr_perm;
+				goto out;
+			}
+			expire_client(conf);
+			goto out_new;
+		}
+		if (ip_addr != conf->cl_addr &&
+		    !(exid->flags & EXCHGID4_FLAG_UPD_CONFIRMED_REC_A)) {
+			/* Client collision. 18.35.4 case 3 */
+			status = nfserr_clid_inuse;
+			goto out;
+		}
+		/*
+		 * Set bit when the owner id and verifier map to an already
+		 * confirmed client id (18.35.3).
+		 */
+		exid->flags |= EXCHGID4_FLAG_CONFIRMED_R;
+
+		/*
+		 * Falling into 18.35.4 case 2, possible router replay.
+		 * Leave confirmed record intact and return same result.
+		 */
+		copy_verf(conf, &verf);
+		new = conf;
+		goto out_copy;
+	} else {
+		/* 18.35.4 case 7 */
+		if (exid->flags & EXCHGID4_FLAG_UPD_CONFIRMED_REC_A) {
+			status = nfserr_noent;
+			goto out;
+		}
+	}
+
+	unconf  = find_unconfirmed_client_by_str(dname, strhashval);
+	if (unconf) {
+		/*
+		 * Possible retry or client restart.  Per 18.35.4 case 4,
+		 * a new unconfirmed record should be generated regardless
+		 * of whether any properties have changed.
+		 */
+		expire_client(unconf);
+	}
+
+out_new:
+	/* Normal case */
+	new = create_client(clname, dname);
+	if (new == NULL) {
+		status = nfserr_resource;
+		goto out;
+	}
+
+	copy_verf(new, &verf);
+	copy_cred(&new->cl_cred, &rqstp->rq_cred);
+	new->cl_addr = ip_addr;
+	gen_clid(new);
+	gen_confirm(new);
+	add_to_unconfirmed(new, strhashval);
+out_copy:
+	exid->clientid.cl_boot = new->cl_clientid.cl_boot;
+	exid->clientid.cl_id = new->cl_clientid.cl_id;
+
+	new->cl_seqid = exid->seqid = 1;
+	nfsd4_set_ex_flags(new, exid);
+
+	dprintk("nfsd4_exchange_id seqid %d flags %x\n",
+		new->cl_seqid, new->cl_exchange_flags);
+	status = nfs_ok;
+
+out:
+	nfs4_unlock_state();
+error:
+	dprintk("nfsd4_exchange_id returns %d\n", ntohl(status));
+	return status;
 }
 
 __be32
fs/nfsd/nfs4xdr.c +143 -3

165 lines not on this page · raw message

include/linux/nfsd/state.h +2 -0
index 7592d7b..5de36a7 100644
@@ -173,6 +173,8 @@ struct nfs4_client {
 	u32			cl_firststate;	/* recovery dir creation */
 #ifdef CONFIG_NFSD_V4_1
 	struct list_head	cl_sessions;
+	u32			cl_seqid;       /* seqid for create_session */
+	u32			cl_exchange_flags;
 #endif /* CONFIG_NFSD_V4_1 */
 };
 
include/linux/nfsd/xdr4.h +7 -1
index 0148d54..ea5a427 100644
@@ -348,7 +348,13 @@ struct nfsd4_write {
 
 #if defined(CONFIG_NFSD_V4_1)
 struct nfsd4_exchange_id {
-	int	foo;	/* stub */
+	nfs4_verifier	verifier;
+	u32		id_len;
+	char		*id;
+	u32		flags;
+	clientid_t	clientid;
+	u32		seqid;
+	int		spa_how;
 };
 
 struct nfsd4_create_session {
Signature · 1 line
1.6.2.1
#122 lore raw

PATCH v2 16/47 nfsd41: match clientid establishment method

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

We need to distinguish between client names provided by NFSv4.0 clients SETCLIENTID and those provided by NFSv4.1 via EXCHANGE_ID when looking up the clientid by string.

[nfsd41: use boolean values for use_exchange_id argument]
---
 fs/nfsd/nfs4recover.c      |    3 ++-
 fs/nfsd/nfs4state.c        |   41 ++++++++++++++++++++++++++++++-----------
 include/linux/nfsd/state.h |    2 +-
 3 files changed, 33 insertions(+), 13 deletions(-)
fs/nfsd/nfs4recover.c +2 -1
index b11cf8d..3444c00 100644
@@ -344,7 +344,8 @@ purge_old(struct dentry *parent, struct dentry *child)
 {
 	int status;
 
-	if (nfs4_has_reclaimed_state(child->d_name.name))
+	/* note: we currently use this path only for minorversion 0 */
+	if (nfs4_has_reclaimed_state(child->d_name.name, false))
 		return 0;
 
 	status = nfsd4_clear_clid_dir(parent, child);
fs/nfsd/nfs4state.c +30 -11

102 lines not on this page · raw message

include/linux/nfsd/state.h +1 -1
index 5de36a7..feab6ec 100644
@@ -331,7 +331,7 @@ extern void nfsd4_init_recdir(char *recdir_name);
 extern int nfsd4_recdir_load(void);
 extern void nfsd4_shutdown_recdir(void);
 extern int nfs4_client_to_reclaim(const char *name);
-extern int nfs4_has_reclaimed_state(const char *name);
+extern int nfs4_has_reclaimed_state(const char *name, bool use_exchange_id);
 extern void nfsd4_recdir_purge_old(void);
 extern int nfsd4_create_clid_dir(struct nfs4_client *clp);
 extern void nfsd4_remove_clid_dir(struct nfs4_client *clp);
Signature · 1 line
1.6.2.1
#133 lore raw

PATCH v2 17/47 nfsd41: sequence operation

In reply to Benny Halevy

Implement the sequence operation conforming to
http://tools.ietf.org/html/draft-ietf-nfsv4-minorversion1-26

Check for stale clientid (as derived from the sessionid). Enforce slotid range and exactly-once semantics using the slotid and seqid.

If everything went well renew the client lease and mark the slot INPROGRESS.

[nfsd41: rename sequence catchthis to cachethis]
[pulled some code to set cstate->slot from "nfsd DRC logic"]
[use sessionid_lock spin lock]
[nfsd41: use bool inuse for slot state]
---
 fs/nfsd/nfs4state.c       |   71 +++++++++++++++++++++++++++++++++++++++++++-
 fs/nfsd/nfs4xdr.c         |   32 +++++++++++++++++++-
 include/linux/nfsd/xdr4.h |   10 ++++++-
 3 files changed, 108 insertions(+), 5 deletions(-)
fs/nfsd/nfs4state.c +69 -2
index 0c39376..a19f292 100644
@@ -1004,6 +1004,32 @@ error:
 	return status;
 }
 
+static int
+check_slot_seqid(u32 seqid, struct nfsd4_slot *slot)
+{
+	dprintk("%s enter. seqid %d slot->sl_seqid %d\n", __func__, seqid,
+		slot->sl_seqid);
+
+	/* The slot is in use, and no response has been sent. */
+	if (slot->sl_inuse) {
+		if (seqid == slot->sl_seqid)
+			return nfserr_jukebox;
+		else
+			return nfserr_seq_misordered;
+	}
+	/* Normal */
+	if (likely(seqid == slot->sl_seqid + 1))
+		return nfs_ok;
+	/* Replay */
+	if (seqid == slot->sl_seqid)
+		return nfserr_replay_cache;
+	/* Wraparound */
+	if (seqid == 1 && (slot->sl_seqid + 1) == 0)
+		return nfs_ok;
+	/* Misordered replay or misordered new request */
+	return nfserr_seq_misordered;
+}
+
 __be32
 nfsd4_create_session(struct svc_rqst *rqstp,
 		     struct nfsd4_compound_state *cstate,
@@ -1021,11 +1047,52 @@ nfsd4_destroy_session(struct svc_rqst *r,
 }
 
 __be32
-nfsd4_sequence(struct svc_rqst *r,
+nfsd4_sequence(struct svc_rqst *rqstp,
 	       struct nfsd4_compound_state *cstate,
 	       struct nfsd4_sequence *seq)
 {
-	return -1;	/* stub */
+	struct nfsd4_session *session;
+	struct nfsd4_slot *slot;
+	int status;
+
+	spin_lock(&sessionid_lock);
+	status = nfserr_badsession;
+	session = find_in_sessionid_hashtbl(&seq->sessionid);
+	if (!session)
+		goto out;
+
+	status = nfserr_badslot;
+	if (seq->slotid >= session->se_fnumslots)
+		goto out;
+
+	slot = &session->se_slots[seq->slotid];
+	dprintk("%s: slotid %d\n", __func__, seq->slotid);
+
+	status = check_slot_seqid(seq->seqid, slot);
+	if (status == nfserr_replay_cache) {
+		cstate->slot = slot;
+		goto replay_cache;
+	}
+	if (status)
+		goto out;
+
+	/* Success! bump slot seqid */
+	slot->sl_inuse = true;
+	slot->sl_seqid = seq->seqid;
+
+	cstate->slot = slot;
+
+replay_cache:
+	/* Renew the clientid on success and on replay.
+	 * Hold a session reference until done processing the compound:
+	 * nfsd4_put_session called only if the cstate slot is set.
+	 */
+	renew_client(session->se_client);
+	nfsd4_get_session(slot->sl_session);
+out:
+	spin_unlock(&sessionid_lock);
+	dprintk("%s: return %d\n", __func__, ntohl(status));
+	return status;
 }
 #endif /* CONFIG_NFSD_V4_1 */
 
fs/nfsd/nfs4xdr.c +30 -2
index 840cf6a..c6b490e 100644
@@ -1114,7 +1114,16 @@ static __be32
 nfsd4_decode_sequence(struct nfsd4_compoundargs *argp,
 		      struct nfsd4_sequence *seq)
 {
-	return nfserr_opnotsupp;	/* stub */
+	DECODE_HEAD;
+
+	READ_BUF(NFS4_MAX_SESSIONID_LEN + 16);
+	COPYMEM(seq->sessionid.data, NFS4_MAX_SESSIONID_LEN);
+	READ32(seq->seqid);
+	READ32(seq->slotid);
+	READ32(seq->maxslots);
+	READ32(seq->cachethis);
+
+	DECODE_TAIL;
 }
 #endif /* CONFIG_NFSD_V4_1 */
 
@@ -2836,7 +2845,26 @@ static __be32
 nfsd4_encode_sequence(struct nfsd4_compoundres *resp, int nfserr,
 		      struct nfsd4_sequence *seq)
 {
-	/* stub */
+	ENCODE_HEAD;
+
+	if (nfserr)
+		goto out;
+
+	RESERVE_SPACE(NFS4_MAX_SESSIONID_LEN + 20);
+	WRITEMEM(seq->sessionid.data, NFS4_MAX_SESSIONID_LEN);
+	WRITE32(seq->seqid);
+	WRITE32(seq->slotid);
+	WRITE32(seq->maxslots);
+	/*
+	 * FIXME: for now:
+	 *   target_maxslots = maxslots
+	 *   status_flags = 0
+	 */
+	WRITE32(seq->maxslots);
+	WRITE32(0);
+
+	ADJUST_ARGS();
+out:
 	return nfserr;
 }
 #endif /* CONFIG_NFSD_V4_1 */
include/linux/nfsd/xdr4.h +9 -1
index ea5a427..9e4d8db 100644
@@ -362,7 +362,15 @@ struct nfsd4_create_session {
 };
 
 struct nfsd4_sequence {
-	int	foo;	/* stub */
+	struct nfs4_sessionid	sessionid;		/* request/response */
+	u32			seqid;			/* request/response */
+	u32			slotid;			/* request/response */
+	u32			maxslots;		/* request/response */
+	u32			cachethis;		/* request */
+#if 0
+	u32			target_maxslots;	/* response */
+	u32			status_flags;		/* response */
+#endif /* not yet */
 };
 
 struct nfsd4_destroy_session {
Signature · 1 line
1.6.2.1
#135 lore raw

PATCH v2 18/47 nfsd41: enforce NFS4ERR_SEQUENCE_POS operation order rules

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

---
 fs/nfsd/nfs4proc.c  |   24 ++++++++++++++++--------
 fs/nfsd/nfs4state.c |    4 ++++
 2 files changed, 20 insertions(+), 8 deletions(-)
fs/nfsd/nfs4proc.c +16 -8
index f618e8e..e703ac2 100644
@@ -811,14 +811,15 @@ static inline void nfsd4_increment_op_stats(u32 opnum)
 
 typedef __be32(*nfsd4op_func)(struct svc_rqst *, struct nfsd4_compound_state *,
 			      void *);
+enum nfsd4_op_flags {
+	ALLOWED_WITHOUT_FH = 1 << 0,	/* No current filehandle required */
+	ALLOWED_ON_ABSENT_FS = 2 << 0,	/* ops processed on absent fs */
+	ALLOWED_AS_FIRST_OP = 3 << 0,	/* ops reqired first in compound */
+};
 
 struct nfsd4_operation {
 	nfsd4op_func op_func;
 	u32 op_flags;
-/* Most ops require a valid current filehandle; a few don't: */
-#define ALLOWED_WITHOUT_FH 1
-/* GETATTR and ops not listed as returning NFS4ERR_MOVED: */
-#define ALLOWED_ON_ABSENT_FS 2
 	char *op_name;
 };
 
@@ -864,6 +865,13 @@ nfsd4_proc_compound(struct svc_rqst *rqstp,
 	if (args->minorversion > NFSD_SUPPORTED_MINOR_VERSION)
 		goto out;
 
+	op = &args->ops[0];
+	if (args->opcnt > 0 && op->status != nfserr_op_illegal &&
+	    !(nfsd4_ops[op->opnum].op_flags & ALLOWED_AS_FIRST_OP)) {
+		op->status = nfserr_sequence_pos;
+		goto encode_op;
+	}
+
 	status = nfs_ok;
 	while (!status && resp->opcnt < args->opcnt) {
 		op = &args->ops[resp->opcnt++];
@@ -1104,22 +1112,22 @@ static struct nfsd4_operation nfsd4_ops[] = {
 #if defined(CONFIG_NFSD_V4_1)
 	[OP_EXCHANGE_ID] = {
 		.op_func = (nfsd4op_func)nfsd4_exchange_id,
-		.op_flags = ALLOWED_WITHOUT_FH,
+		.op_flags = ALLOWED_WITHOUT_FH | ALLOWED_AS_FIRST_OP,
 		.op_name = "OP_EXCHANGE_ID",
 	},
 	[OP_CREATE_SESSION] = {
 		.op_func = (nfsd4op_func)nfsd4_create_session,
-		.op_flags = ALLOWED_WITHOUT_FH,
+		.op_flags = ALLOWED_WITHOUT_FH | ALLOWED_AS_FIRST_OP,
 		.op_name = "OP_CREATE_SESSION",
 	},
 	[OP_DESTROY_SESSION] = {
 		.op_func = (nfsd4op_func)nfsd4_destroy_session,
-		.op_flags = ALLOWED_WITHOUT_FH,
+		.op_flags = ALLOWED_WITHOUT_FH | ALLOWED_AS_FIRST_OP,
 		.op_name = "OP_DESTROY_SESSION",
 	},
 	[OP_SEQUENCE] = {
 		.op_func = (nfsd4op_func)nfsd4_sequence,
-		.op_flags = ALLOWED_WITHOUT_FH,
+		.op_flags = ALLOWED_WITHOUT_FH | ALLOWED_AS_FIRST_OP,
 		.op_name = "OP_SEQUENCE",
 	},
 #endif /* CONFIG_NFSD_V4_1 */
fs/nfsd/nfs4state.c +4 -0
index a19f292..10eb67b 100644
@@ -1051,10 +1051,14 @@ nfsd4_sequence(struct svc_rqst *rqstp,
 	       struct nfsd4_compound_state *cstate,
 	       struct nfsd4_sequence *seq)
 {
+	struct nfsd4_compoundres *resp = rqstp->rq_resp;
 	struct nfsd4_session *session;
 	struct nfsd4_slot *slot;
 	int status;
 
+	if (resp->opcnt != 1)
+		return nfserr_sequence_pos;
+
 	spin_lock(&sessionid_lock);
 	status = nfserr_badsession;
 	session = find_in_sessionid_hashtbl(&seq->sessionid);
Signature · 1 line
1.6.2.1
#138 lore raw

PATCH v2 19/47 nfsd41: DRC save, restore, and clear functions

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

Cache all the result pages, including the rpc header in rq_respages[0], for a request in the slot table cache entry.

Cache the statp pointer from nfsd_dispatch which points into rq_respages[0] just past the rpc header. When setting a cache entry, calculate and save the length of the nfs data minus the rpc header for rq_respages[0].

When replaying a cache entry, replace the cached rpc header with the replayed request rpc result header, unless there is not enough room in the cached results first page. In that case, use the cached rpc header.

The sessions fore channel maxresponse size cached is set to NFSD_PAGES_PER_SLOT
* PAGE_SIZE. For compounds we are cacheing with operations such as READDIR that use the xdr_buf->pages to hold data, we choose to cache the extra page of data rather than copying data from xdr_buf->pages into the xdr_buf->head page.

[nfsd41: limit cache to maxresponsesize_cached]
[nfsd41: mv nfsd4_set_statp under CONFIG_NFSD_V4_1]
---
 fs/nfsd/nfs4state.c        |  142 ++++++++++++++++++++++++++++++++++++++++++++
 fs/nfsd/nfssvc.c           |    4 +
 include/linux/nfsd/cache.h |    5 ++
 include/linux/nfsd/state.h |   13 ++++
 include/linux/nfsd/xdr4.h  |    4 +
 5 files changed, 168 insertions(+), 0 deletions(-)
fs/nfsd/nfs4state.c +142 -0
index 10eb67b..f0ce639 100644
@@ -860,6 +860,148 @@ out_err:
 }
 
 #if defined(CONFIG_NFSD_V4_1)
+void
+nfsd4_set_statp(struct svc_rqst *rqstp, __be32 *statp)
+{
+	struct nfsd4_compoundres *resp = rqstp->rq_resp;
+
+	resp->cstate.statp = statp;
+}
+
+/*
+ * Dereference the result pages.
+ */
+static void
+nfsd4_release_respages(struct page **respages, short resused)
+{
+	int page_no;
+
+	dprintk("--> %s\n", __func__);
+	for (page_no = 0; page_no < resused; page_no++) {
+		if (!respages[page_no])
+			continue;
+		put_page(respages[page_no]);
+		respages[page_no] = NULL;
+	}
+}
+
+static void
+nfsd4_move_pages(struct page **topages, struct page **frompages, short count)
+{
+	int page_no;
+
+	for (page_no = 0; page_no < count; page_no++) {
+		topages[page_no] = frompages[page_no];
+		if (!topages[page_no])
+			continue;
+		get_page(topages[page_no]);
+	}
+}
+
+/*
+ * Cache the reply pages up to NFSD_PAGES_PER_SLOT + 1, clearing the previous
+ * pages. We add a page to NFSD_PAGES_PER_SLOT for the case where the total
+ * length of the XDR response is less than se_fmaxresp_cached
+ * (NFSD_PAGES_PER_SLOT * PAGE_SIZE) but the xdr_buf pages is used for a
+ * of the reply (e.g. readdir).
+ *
+ * Store the base and length of the rq_req.head[0] page
+ * of the NFSv4.1 data, just past the rpc header.
+ */
+void
+nfsd4_set_cache_entry(struct nfsd4_compoundres *resp)
+{
+	struct nfsd4_cache_entry *entry = &resp->cstate.slot->sl_cache_entry;
+	struct svc_rqst *rqstp = resp->rqstp;
+	struct kvec *resv = &rqstp->rq_res.head[0];
+
+	dprintk("--> %s entry %p\n", __func__, entry);
+
+	/* Don't cache a failed OP_SEQUENCE */
+	if (resp->opcnt == 1 && resp->cstate.status)
+		return;
+	nfsd4_release_respages(entry->ce_respages, entry->ce_resused);
+	entry->ce_resused = rqstp->rq_resused;
+	if (entry->ce_resused > NFSD_PAGES_PER_SLOT + 1)
+		entry->ce_resused = NFSD_PAGES_PER_SLOT + 1;
+	nfsd4_move_pages(entry->ce_respages, rqstp->rq_respages,
+			 entry->ce_resused);
+	entry->ce_status = resp->cstate.status;
+	entry->ce_datav.iov_base = resp->cstate.statp;
+	entry->ce_datav.iov_len = resv->iov_len - ((char *)resp->cstate.statp -
+				(char *)page_address(rqstp->rq_respages[0]));
+	entry->ce_opcnt = resp->opcnt;
+	/* Current request rpc header length*/
+	entry->ce_rpchdrlen = (char *)resp->cstate.statp -
+				(char *)page_address(rqstp->rq_respages[0]);
+}
+
+/*
+ * Copy the cached NFSv4.1 reply skipping the cached rpc header into the
+ * replay result res.head[0] past the rpc header to end up with replay
+ * rpc header and cached NFSv4.1 reply.
+ */
+static int
+nfsd41_copy_replay_data(struct nfsd4_compoundres *resp,
+			struct nfsd4_cache_entry *entry)
+{
+	struct svc_rqst *rqstp = resp->rqstp;
+	struct kvec *resv = &resp->rqstp->rq_res.head[0];
+	int len;
+
+	/* Current request rpc header length*/
+	len = (char *)resp->cstate.statp -
+			(char *)page_address(rqstp->rq_respages[0]);
+	if (entry->ce_datav.iov_len + len > PAGE_SIZE) {
+		dprintk("%s v41 cached reply too large (%Zd).\n", __func__,
+			entry->ce_datav.iov_len);
+		return 0;
+	}
+	/* copy the cached reply nfsd data past the current rpc header */
+	memcpy((char *)resv->iov_base + len, entry->ce_datav.iov_base,
+		entry->ce_datav.iov_len);
+	resv->iov_len = len + entry->ce_datav.iov_len;
+	return 1;
+}
+
+/*
+ * Keep the first page of the replay. Copy the NFSv4.1 data from the first
+ * cached page.  Replace any futher replay pages from the cache.
+ */
+__be32
+nfsd4_replay_cache_entry(struct nfsd4_compoundres *resp)
+{
+	struct nfsd4_cache_entry *entry = &resp->cstate.slot->sl_cache_entry;
+	__be32 status;
+
+	dprintk("--> %s entry %p\n", __func__, entry);
+
+
+	if (!nfsd41_copy_replay_data(resp, entry)) {
+		/*
+		 * Not enough room to use the replay rpc header, send the
+		 * cached header. Release all the allocated result pages.
+		 */
+		svc_free_res_pages(resp->rqstp);
+		nfsd4_move_pages(resp->rqstp->rq_respages, entry->ce_respages,
+			entry->ce_resused);
+	} else {
+		/* Release all but the first allocated result page */
+
+		resp->rqstp->rq_resused--;
+		svc_free_res_pages(resp->rqstp);
+
+		nfsd4_move_pages(&resp->rqstp->rq_respages[1],
+				 &entry->ce_respages[1],
+				 entry->ce_resused - 1);
+	}
+
+	resp->rqstp->rq_resused = entry->ce_resused;
+	status = entry->ce_status;
+
+	return status;
+}
+
 /*
  * Set the exchange_id flags returned by the server.
  */
fs/nfsd/nfssvc.c +4 -0
index ef0a368..b5168d1 100644
@@ -515,6 +515,10 @@ nfsd_dispatch(struct svc_rqst *rqstp, __be32 *statp)
 		+ rqstp->rq_res.head[0].iov_len;
 	rqstp->rq_res.head[0].iov_len += sizeof(__be32);
 
+	/* NFSv4.1 DRC requires statp */
+	if (rqstp->rq_vers == 4)
+		nfsd4_set_statp(rqstp, statp);
+
 	/* Now call the procedure handler, and encode NFS status. */
 	nfserr = proc->pc_func(rqstp, rqstp->rq_argp, rqstp->rq_resp);
 	nfserr = map_new_errors(rqstp->rq_vers, nfserr);
include/linux/nfsd/cache.h +5 -0
index 04b355c..57a83c7 100644
@@ -75,5 +75,10 @@ int	nfsd_reply_cache_init(void);
 void	nfsd_reply_cache_shutdown(void);
 int	nfsd_cache_lookup(struct svc_rqst *, int);
 void	nfsd_cache_update(struct svc_rqst *, int, __be32 *);
+#ifdef CONFIG_NFSD_V4_1
+void	nfsd4_set_statp(struct svc_rqst *rqstp, __be32 *statp);
+#else /* CONFIG_NFSD_V4_1 */
+static inline void nfsd4_set_statp(struct svc_rqst *rqstp, __be32 *statp) {}
+#endif /* CONFIG_NFSD_V4_1 */
 
 #endif /* NFSCACHE_H */
include/linux/nfsd/state.h +13 -0
index feab6ec..8ca6a82 100644
@@ -99,10 +99,23 @@ struct nfs4_callback {
 	struct rpc_clnt *       cb_client;
 };
 
+/* Maximum number of pages per slot cache entry */
+#define NFSD_PAGES_PER_SLOT	1
+
+struct nfsd4_cache_entry {
+	__be32		ce_status;
+	struct kvec	ce_datav; /* encoded NFSv4.1 data in rq_res.head[0] */
+	struct page	*ce_respages[NFSD_PAGES_PER_SLOT + 1];
+	short		ce_resused;
+	int		ce_opcnt;
+	int		ce_rpchdrlen;
+};
+
 struct nfsd4_slot {
 	bool				sl_inuse;
 	struct nfsd4_session		*sl_session;
 	u32				sl_seqid;
+	struct nfsd4_cache_entry	sl_cache_entry;
 };
 
 struct nfsd4_session {
include/linux/nfsd/xdr4.h +4 -0
index 9e4d8db..cde8947 100644
@@ -50,6 +50,8 @@ struct nfsd4_compound_state {
 	struct nfs4_stateowner	*replay_owner;
 	/* For sessions DRC */
 	struct nfsd4_slot	*slot;
+	__be32			*statp;
+	u32			status;
 };
 
 struct nfsd4_change_info {
@@ -490,6 +492,8 @@ extern __be32 nfsd4_setclientid_confirm(struct svc_rqst *rqstp,
 		struct nfsd4_compound_state *,
 		struct nfsd4_setclientid_confirm *setclientid_confirm);
 #if defined(CONFIG_NFSD_V4_1)
+extern void nfsd4_set_cache_entry(struct nfsd4_compoundres *resp);
+extern __be32 nfsd4_replay_cache_entry(struct nfsd4_compoundres *resp);
 extern __be32 nfsd4_exchange_id(struct svc_rqst *rqstp,
 		struct nfsd4_compound_state *,
 struct nfsd4_exchange_id *);
Signature · 1 line
1.6.2.1
#145 lore raw

PATCH v2 20/47 nfsd41: hard page limit for DRC

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

Use no more than 1/128th of the number of free pages at nfsd startup for the v4.1 DRC.

This is an arbitrary default which should probably end up under the control of an administrator.

[moved added fields in struct svc_serv under CONFIG_NFSD_V4_1]
[fix set_max_drc calculation of sv_drc_max_pages]
---
 fs/nfsd/nfssvc.c           |   25 +++++++++++++++++++++++++
 include/linux/nfsd/nfsd.h  |    7 +++++++
 include/linux/sunrpc/svc.h |    4 ++++
 3 files changed, 36 insertions(+), 0 deletions(-)
fs/nfsd/nfssvc.c +25 -0
index b5168d1..ab60320 100644
@@ -22,6 +22,7 @@
 #include <linux/freezer.h>
 #include <linux/fs_struct.h>
 #include <linux/kthread.h>
+#include <linux/swap.h>
 
 #include <linux/sunrpc/types.h>
 #include <linux/sunrpc/stats.h>
@@ -197,6 +198,28 @@ void nfsd_reset_versions(void)
 	}
 }
 
+/*
+ * Each session guarantees a negotiated per slot memory cache for replies
+ * which in turn consumes memory beyond the v2/v3/v4.0 server. A dedicated
+ * NFSv4.1 server might want to use more memory for a DRC than a machine
+ * with mutiple services.
+ *
+ * Impose a hard limit on the number of pages for the DRC which varies
+ * according to the machines free pages. This is of course only a default.
+ *
+ * For now this is a #defined shift which could be under admin control
+ * in the future.
+ */
+static void set_max_drc(void)
+{
+#ifdef CONFIG_NFSD_V4_1
+	nfsd_serv->sv_drc_max_pages = nr_free_buffer_pages()
+						>> NFSD_DRC_SIZE_SHIFT;
+	nfsd_serv->sv_drc_pages_used = 0;
+	dprintk("%s svc_drc_max_pages %u\n", __func__,
+		nfsd_serv->sv_drc_max_pages);
+#endif /* CONFIG_NFSD_V4_1 */
+}
 
 int nfsd_create_serv(void)
 {
@@ -229,6 +252,8 @@ int nfsd_create_serv(void)
 				      nfsd_last_thread, nfsd, THIS_MODULE);
 	if (nfsd_serv == NULL)
 		err = -ENOMEM;
+	else
+		set_max_drc();
 
 	do_gettimeofday(&nfssvc_boot);		/* record boot time */
 	return err;
include/linux/nfsd/nfsd.h +7 -0
index ab9616d..3bccec3 100644
@@ -361,6 +361,13 @@ extern struct timeval	nfssvc_boot;
  | FATTR4_WORD1_TIME_DELTA   | FATTR4_WORD1_TIME_METADATA    \
  | FATTR4_WORD1_TIME_MODIFY     | FATTR4_WORD1_TIME_MODIFY_SET | FATTR4_WORD1_MOUNTED_ON_FILEID)
 
+#if defined(CONFIG_NFSD_V4_1)
+
+/* The percent of nr_free_buffer_pages used by the V4.1 server DRC */
+#define NFSD_DRC_SIZE_SHIFT	7
+
+#endif /* CONFIG_NFSD_V4_1 */
+
 /* These will return ERR_INVAL if specified in GETATTR or READDIR. */
 #define NFSD_WRITEONLY_ATTRS_WORD1							    \
 (FATTR4_WORD1_TIME_ACCESS_SET   | FATTR4_WORD1_TIME_MODIFY_SET)
include/linux/sunrpc/svc.h +4 -0
index 815dd58..c4f0e1e 100644
@@ -95,6 +95,10 @@ struct svc_serv {
 	struct module *		sv_module;	/* optional module to count when
 						 * adding threads */
 	svc_thread_fn		sv_function;	/* main function for threads */
+#if defined(CONFIG_NFSD_V4_1)
+	unsigned int		sv_drc_max_pages; /* Total pages for DRC */
+	unsigned int		sv_drc_pages_used;/* DRC pages used */
+#endif /* CONFIG_NFSD_V4_1 */
 };
 
 /*
Signature · 1 line
1.6.2.1
#146 lore raw

PATCH v2 21/47 nfsd41: nfsd DRC logic

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

Replay a request in nfsd4_sequence.
Add a minorversion to struct nfsd4_compound_state.

Pass the current slot to nfs4svc_encode_compound res via struct nfsd4_compoundres to set an NFSv4.1 DRC entry.

[nfsd41: use bool inuse for slot state]
---
 fs/nfsd/nfs4proc.c        |    7 +++++++
 fs/nfsd/nfs4state.c       |    6 ++++++
 fs/nfsd/nfs4xdr.c         |   13 +++++++++++++
 include/linux/nfsd/xdr4.h |    1 +
 4 files changed, 27 insertions(+), 0 deletions(-)
fs/nfsd/nfs4proc.c +7 -0
index e703ac2..bdbeb87 100644
@@ -920,6 +920,12 @@ nfsd4_proc_compound(struct svc_rqst *rqstp,
 			BUG_ON(op->status == nfs_ok);
 
 encode_op:
+		/* Only from SEQUENCE or CREATE_SESSION */
+		if (resp->cstate.status == nfserr_replay_cache) {
+			dprintk("%s NFS4.1 replay from cache\n", __func__);
+			status = op->status;
+			goto out;
+		}
 		if (op->status == nfserr_replay_me) {
 			op->replay = &cstate->replay_owner->so_replay;
 			nfsd4_encode_replay(resp, op);
@@ -948,6 +954,7 @@ encode_op:
 		status = nfserr_jukebox;
 	}
 
+	resp->cstate.status = status;
 	fh_put(&resp->cstate.current_fh);
 	fh_put(&resp->cstate.save_fh);
 	BUG_ON(resp->cstate.replay_owner);
fs/nfsd/nfs4state.c +6 -0
index f0ce639..07c869d 100644
@@ -997,6 +997,8 @@ nfsd4_replay_cache_entry(struct nfsd4_compoundres *resp)
 	}
 
 	resp->rqstp->rq_resused = entry->ce_resused;
+	resp->opcnt = entry->ce_opcnt;
+	resp->cstate.iovlen = entry->ce_datav.iov_len + entry->ce_rpchdrlen;
 	status = entry->ce_status;
 
 	return status;
@@ -1217,6 +1219,10 @@ nfsd4_sequence(struct svc_rqst *rqstp,
 	status = check_slot_seqid(seq->seqid, slot);
 	if (status == nfserr_replay_cache) {
 		cstate->slot = slot;
+		/* Return the cached reply status and set cstate->status
+		 * for nfsd4_svc_encode_compoundres processing*/
+		status = nfsd4_replay_cache_entry(resp);
+		cstate->status = nfserr_replay_cache;
 		goto replay_cache;
 	}
 	if (status)
fs/nfsd/nfs4xdr.c +13 -0
index c6b490e..57afb33 100644
@@ -3059,6 +3059,19 @@ nfs4svc_encode_compoundres(struct svc_rqst *rqstp, __be32 *p, struct nfsd4_compo
 		iov = &rqstp->rq_res.head[0];
 	iov->iov_len = ((char*)resp->p) - (char*)iov->iov_base;
 	BUG_ON(iov->iov_len > PAGE_SIZE);
+#ifdef CONFIG_NFSD_V4_1
+	if (resp->cstate.slot != NULL) {
+		if (resp->cstate.status == nfserr_replay_cache) {
+			iov->iov_len = resp->cstate.iovlen;
+		} else {
+			nfsd4_set_cache_entry(resp);
+			dprintk("%s: SET SLOT STATE TO AVAILABLE\n", __func__);
+			resp->cstate.slot->sl_inuse = 0;
+		}
+		if (resp->cstate.slot->sl_session)
+			nfsd4_put_session(resp->cstate.slot->sl_session);
+	}
+#endif /* CONFIG_NFSD_V4_1 */
 	return 1;
 }
 
include/linux/nfsd/xdr4.h +1 -0
index cde8947..5c0d376 100644
@@ -51,6 +51,7 @@ struct nfsd4_compound_state {
 	/* For sessions DRC */
 	struct nfsd4_slot	*slot;
 	__be32			*statp;
+	size_t			iovlen;
 	u32			status;
 };
 
Signature · 1 line
1.6.2.1
#152 lore raw

PATCH v2 22/47 nfsd41: clear DRC cache on free_session

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

---
 fs/nfsd/nfs4state.c |    7 +++++++
 1 files changed, 7 insertions(+), 0 deletions(-)
fs/nfsd/nfs4state.c +7 -0
index 07c869d..37865c9 100644
@@ -442,12 +442,19 @@ release_session(struct nfsd4_session *ses)
 	nfsd4_put_session(ses);
 }
 
+static void nfsd4_release_respages(struct page **respages, short resused);
+
 void
 free_session(struct kref *kref)
 {
 	struct nfsd4_session *ses;
+	int i;
 
 	ses = container_of(kref, struct nfsd4_session, se_ref);
+	for (i = 0; i < ses->se_fnumslots; i++) {
+		struct nfsd4_cache_entry *e = &ses->se_slots[i].sl_cache_entry;
+		nfsd4_release_respages(e->ce_respages, e->ce_resused);
+	}
 	kfree(ses->se_slots);
 	kfree(ses);
 }
Signature · 1 line
1.6.2.1
#153 lore raw

PATCH v2 23/47 nfsd41: create_session operation

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

Implement the create_session operation confoming to
http://tools.ietf.org/html/draft-ietf-nfsv4-minorversion1-26

Look up the client id (generated by the server on exchange_id, given by the client on create_session).
If neither a confirmed or unconfirmed client is found then the client id is stale
If a confirmed cilent is found (i.e. we already received create_session for it) then compare the sequence id to determine if it's a replay or possibly a mis-ordered rpc. If the seqid is in order, update the confirmed client seqid and procedd with updating the session parameters.

If an unconfirmed client_id is found then verify the creds and seqid. If both match move the client id to confirmed state and proceed with processing the create_session.

Currently, we do not support persistent sessions, and RDMA.

alloc_init_session generates a new sessionid and creates a session structure.

NFSD_PAGES_PER_SLOT is used for the max response cached calculation, and for the counting of DRC pages using the hard limits set in struct srv_serv.

A note on NFSD_PAGES_PER_SLOT:

Other patches in this series allow for NFSD_PAGES_PER_SLOT + 1 pages to be cached in a DRC slot when the response size is less than NFSD_PAGES_PER_SLOT * PAGE_SIZE but xdr_buf pages are used. e.g. a READDIR operation will encode a small amount of data in the xdr_buf head, and then the READDIR in the xdr_buf pages. So, the hard limit calculation use of pages by a session is underestimated by the number of cached operations using the xdr_buf pages.

Yet another patch caches no pages for the solo sequence operation, or any compound where cache_this is False. So the hard limit calculation use of pages by a session is overestimated by the number of these operations in the cache.

TODO: improve resource pre-allocation and negotiate session parameters accordingly. Respect and possibly adjust backchannel attributes.

[nfsd41: remove headerpadsz from channel attributes]

Our client and server only support a headerpadsz of 0.

[nfsd41: use DRC limits in fore channel init]
[nfsd41: do not change CREATE_SESSION back channel attrs]
[use sessionid_lock spin lock]
[nfsd41: use bool inuse for slot state]
---
 fs/nfsd/nfs4state.c        |  197 +++++++++++++++++++++++++++++++++++++++++++-
 fs/nfsd/nfs4xdr.c          |  147 ++++++++++++++++++++++++++++++++-
 include/linux/nfsd/state.h |    7 ++
 include/linux/nfsd/xdr4.h  |   21 +++++-
 4 files changed, 368 insertions(+), 4 deletions(-)
fs/nfsd/nfs4state.c +196 -1
index 37865c9..e4e2c19 100644
@@ -68,6 +68,9 @@ static u32 current_delegid = 1;
 static u32 nfs4_init;
 static stateid_t zerostateid;             /* bits all 0 */
 static stateid_t onestateid;              /* bits all 1 */
+#ifdef CONFIG_NFSD_V4_1
+static u64 current_sessionid = 1;
+#endif /* CONFIG_NFSD_V4_1 */
 
 #define ZERO_STATEID(stateid) (!memcmp((stateid), &zerostateid, sizeof(stateid_t)))
 #define ONE_STATEID(stateid)  (!memcmp((stateid), &onestateid, sizeof(stateid_t)))
@@ -402,6 +405,138 @@ dump_sessionid(const char *fn, struct nfs4_sessionid *sessionid)
 	dprintk("%s: %u:%u:%u:%u\n", fn, ptr[0], ptr[1], ptr[2], ptr[3]);
 }
 
+static void
+gen_sessionid(struct nfsd4_session *ses)
+{
+	struct nfs4_client *clp = ses->se_client;
+	struct nfsd4_sessionid *sid;
+
+	sid = (struct nfsd4_sessionid *)ses->se_sessionid.data;
+	sid->clientid = clp->cl_clientid;
+	sid->sequence = current_sessionid++;
+	sid->reserved = 0;
+}
+
+/*
+ * Give the client the number of slots it requests bound by
+ * NFSD_MAX_SLOTS_PER_SESSION and by sv_drc_max_pages.
+ *
+ * If we run out of pages (sv_drc_pages_used == sv_drc_max_pages) we
+ * should (up to a point) re-negotiate active sessions and reduce their
+ * slot usage to make rooom for new connections. For now we just fail the
+ * create session.
+ */
+static int set_forechannel_maxreqs(struct nfsd4_channel_attrs *fchan)
+{
+	int status = 0, np = fchan->maxreqs * NFSD_PAGES_PER_SLOT;
+
+	spin_lock(&nfsd_serv->sv_lock);
+	if (np + nfsd_serv->sv_drc_pages_used > nfsd_serv->sv_drc_max_pages)
+		np = nfsd_serv->sv_drc_max_pages - nfsd_serv->sv_drc_pages_used;
+	nfsd_serv->sv_drc_pages_used += np;
+	spin_unlock(&nfsd_serv->sv_lock);
+
+	if (np <= 0) {
+		status = nfserr_resource;
+		fchan->maxreqs = 0;
+	} else
+		fchan->maxreqs = np / NFSD_PAGES_PER_SLOT;
+
+	return status;
+}
+
+/*
+ * fchan holds the client values on input, and the server values on output
+ */
+static int init_forechannel_attrs(struct svc_rqst *rqstp,
+				    struct nfsd4_session *session,
+				    struct nfsd4_channel_attrs *fchan)
+{
+	int status = 0;
+	__u32   maxcount = svc_max_payload(rqstp);
+
+	/* headerpadsz set to zero in encode routine*/
+
+	/* Use the client's max request and max response size if possible */
+	if (fchan->maxreq_sz > maxcount)
+		fchan->maxreq_sz = maxcount;
+	session->se_fmaxreq_sz = fchan->maxreq_sz;
+
+	if (fchan->maxresp_sz > maxcount)
+		fchan->maxresp_sz = maxcount;
+	session->se_fmaxresp_sz = fchan->maxresp_sz;
+
+	/* Set the max response cached size our default which is
+	 * a multiple of PAGE_SIZE and small */
+	session->se_fmaxresp_cached = NFSD_PAGES_PER_SLOT * PAGE_SIZE;
+	fchan->maxresp_cached = session->se_fmaxresp_cached;
+
+	/* Use the client's maxops if possible */
+	if (fchan->maxops > NFSD_MAX_OPS_PER_COMPOUND)
+		fchan->maxops = NFSD_MAX_OPS_PER_COMPOUND;
+	session->se_fmaxops = fchan->maxops;
+
+	/* try to use the client requested number of slots */
+	if (fchan->maxreqs > NFSD_MAX_SLOTS_PER_SESSION)
+		fchan->maxreqs = NFSD_MAX_SLOTS_PER_SESSION;
+
+	/* FIXME: Error means no more DRC pages so the server should
+	 * recover pages from existing sessions. For now fail session
+	 * creation.
+	 */
+	status = set_forechannel_maxreqs(fchan);
+
+	session->se_fnumslots = fchan->maxreqs;
+	return status;
+}
+
+static int
+alloc_init_session(struct svc_rqst *rqstp, struct nfs4_client *clp,
+		   struct nfsd4_create_session *cses)
+{
+	struct nfsd4_session *new;
+	int idx, status = nfserr_resource, slotsize, i;
+
+	new = kzalloc(sizeof(*new), GFP_KERNEL);
+	if (!new)
+		goto out;
+
+	/* FIXME: For now, we just accept the client back channel attributes. */
+	status = init_forechannel_attrs(rqstp, new, &cses->fore_channel);
+	if (status)
+		goto out_free;
+
+	slotsize = new->se_fnumslots * sizeof(struct nfsd4_slot);
+	new->se_slots = kzalloc(slotsize, GFP_KERNEL);
+	if (!new->se_slots)
+		goto out_free;
+
+	for (i = 0; i < new->se_fnumslots; i++)
+		new->se_slots[i].sl_session = new;
+
+	new->se_client = clp;
+	gen_sessionid(new);
+	idx = hash_sessionid(&new->se_sessionid);
+	memcpy(clp->cl_sessionid.data, new->se_sessionid.data,
+	       NFS4_MAX_SESSIONID_LEN);
+
+	new->se_flags = cses->flags;
+	kref_init(&new->se_ref);
+	INIT_LIST_HEAD(&new->se_hash);
+	INIT_LIST_HEAD(&new->se_perclnt);
+	spin_lock(&sessionid_lock);
+	list_add(&new->se_hash, &sessionid_hashtbl[idx]);
+	list_add(&new->se_perclnt, &clp->cl_sessions);
+	spin_unlock(&sessionid_lock);
+
+	status = nfs_ok;
+out:
+	return status;
+out_free:
+	kfree(new);
+	goto out;
+}
+
 /* caller must hold sessionid_lock */
 static struct nfsd4_session *
 find_in_sessionid_hashtbl(struct nfs4_sessionid *sessionid)
@@ -1186,7 +1321,67 @@ nfsd4_create_session(struct svc_rqst *rqstp,
 		     struct nfsd4_compound_state *cstate,
 		     struct nfsd4_create_session *cr_ses)
 {
-	return -1;	/* stub */
+	u32 ip_addr = svc_addr_in(rqstp)->sin_addr.s_addr;
+	struct nfs4_client *conf, *unconf;
+	int status = 0;
+
+	nfs4_lock_state();
+	unconf = find_unconfirmed_client(&cr_ses->clientid);
+	conf = find_confirmed_client(&cr_ses->clientid);
+
+	if (conf) {
+		status = nfs_ok;
+		if (conf->cl_seqid == cr_ses->seqid) {
+			dprintk("Got a create_session replay! seqid= %d\n",
+				conf->cl_seqid);
+			goto out_replay;
+		} else if (cr_ses->seqid != conf->cl_seqid + 1) {
+			status = nfserr_seq_misordered;
+			dprintk("Sequence misordered!\n");
+			dprintk("Expected seqid= %d but got seqid= %d\n",
+				conf->cl_seqid, cr_ses->seqid);
+			goto out;
+		}
+		conf->cl_seqid++;
+	} else if (unconf) {
+		if (!same_creds(&unconf->cl_cred, &rqstp->rq_cred) ||
+		    (ip_addr != unconf->cl_addr)) {
+			status = nfserr_clid_inuse;
+			goto out;
+		}
+
+		if (unconf->cl_seqid != cr_ses->seqid) {
+			status = nfserr_seq_misordered;
+			goto out;
+		}
+
+		move_to_confirmed(unconf);
+
+		/*
+		 * We do not support RDMA or persistent sessions
+		 */
+		cr_ses->flags &= ~SESSION4_PERSIST;
+		cr_ses->flags &= ~SESSION4_RDMA;
+
+		conf = unconf;
+	} else {
+		status = nfserr_stale_clientid;
+		goto out;
+	}
+
+	status = alloc_init_session(rqstp, conf, cr_ses);
+	if (status)
+		goto out;
+
+out_replay:
+	memcpy(cr_ses->sessionid.data, conf->cl_sessionid.data,
+	       NFS4_MAX_SESSIONID_LEN);
+	cr_ses->seqid = conf->cl_seqid;
+
+out:
+	nfs4_unlock_state();
+	dprintk("%s returns %d\n", __func__, ntohl(status));
+	return status;
 }
 
 __be32
fs/nfsd/nfs4xdr.c +145 -2

159 lines not on this page · raw message

include/linux/nfsd/state.h +7 -0
index 8ca6a82..98d7b1c 100644
@@ -99,8 +99,12 @@ struct nfs4_callback {
 	struct rpc_clnt *       cb_client;
 };
 
+/* Maximum number of slots per session. 128 is useful for long haul TCP */
+#define NFSD_MAX_SLOTS_PER_SESSION	128
 /* Maximum number of pages per slot cache entry */
 #define NFSD_PAGES_PER_SLOT	1
+/* Maximum number of operations per session compound */
+#define NFSD_MAX_OPS_PER_COMPOUND	16
 
 struct nfsd4_cache_entry {
 	__be32		ce_status;
@@ -188,6 +192,9 @@ struct nfs4_client {
 	struct list_head	cl_sessions;
 	u32			cl_seqid;       /* seqid for create_session */
 	u32			cl_exchange_flags;
+	struct nfs4_sessionid	cl_sessionid;
+
+	struct svc_xprt		*cl_cb_xprt;	/* 4.1 callback transport */
 #endif /* CONFIG_NFSD_V4_1 */
 };
 
include/linux/nfsd/xdr4.h +20 -1
index 5c0d376..c7bf0a1 100644
@@ -360,8 +360,27 @@ struct nfsd4_exchange_id {
 	int		spa_how;
 };
 
+struct nfsd4_channel_attrs {
+	u32		headerpadsz;
+	u32		maxreq_sz;
+	u32		maxresp_sz;
+	u32		maxresp_cached;
+	u32		maxops;
+	u32		maxreqs;
+	u32		nr_rdma_attrs;
+	u32		rdma_attrs;
+};
+
 struct nfsd4_create_session {
-	int	foo;	/* stub */
+	clientid_t		clientid;
+	struct nfs4_sessionid	sessionid;
+	u32			seqid;
+	u32			flags;
+	struct nfsd4_channel_attrs fore_channel;
+	struct nfsd4_channel_attrs back_channel;
+	u32			callback_prog;
+	u32			uid;
+	u32			gid;
 };
 
 struct nfsd4_sequence {
Signature · 1 line
1.6.2.1
#156 lore raw

PATCH v2 24/47 nfsd41: Add a create session replay cache

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

Replace the nfs4_client cl_seqid field with a single struct nfs41_slot used for the create session replay cache.

The CREATE_SESSION slot sets the sl_session pointer to NULL. Otherwise, the slot and it's replay cache are used just like the session slots.

Fix unconfirmed create_session replay response by initializing the create_session slot sequence id to 0.

A future patch will set the CREATE_SESSION cache when a SEQUENCE operation preceeds the CREATE_SESSION operation. This compound is currently only cached in the session slot table.

[nfsd41: use bool inuse for slot state]
---
 fs/nfsd/nfs4state.c        |   47 ++++++++++++++++++++++++++++++-------------
 include/linux/nfsd/state.h |    2 +-
 2 files changed, 34 insertions(+), 15 deletions(-)
fs/nfsd/nfs4state.c +33 -14
index e4e2c19..61af434 100644
@@ -662,6 +662,10 @@ static inline void
 free_client(struct nfs4_client *clp)
 {
 	shutdown_callback_client(clp);
+#if defined(CONFIG_NFSD_V4_1)
+	nfsd4_release_respages(clp->cl_slot.sl_cache_entry.ce_respages,
+			     clp->cl_slot.sl_cache_entry.ce_resused);
+#endif /* CONFIG_NFSD_V4_1 */
 	if (clp->cl_cred.cr_group_info)
 		put_group_info(clp->cl_cred.cr_group_info);
 	kfree(clp->cl_principal);
@@ -1055,12 +1059,14 @@ nfsd4_set_cache_entry(struct nfsd4_compoundres *resp)
 {
 	struct nfsd4_cache_entry *entry = &resp->cstate.slot->sl_cache_entry;
 	struct svc_rqst *rqstp = resp->rqstp;
+	struct nfsd4_compoundargs *args = rqstp->rq_argp;
+	struct nfsd4_op *op = &args->ops[resp->opcnt];
 	struct kvec *resv = &rqstp->rq_res.head[0];
 
 	dprintk("--> %s entry %p\n", __func__, entry);
 
-	/* Don't cache a failed OP_SEQUENCE */
-	if (resp->opcnt == 1 && resp->cstate.status)
+	/* Don't cache a failed OP_SEQUENCE. */
+	if (resp->opcnt == 1 && op->opnum == OP_SEQUENCE && resp->cstate.status)
 		return;
 	nfsd4_release_respages(entry->ce_respages, entry->ce_resused);
 	entry->ce_resused = rqstp->rq_resused;
@@ -1276,11 +1282,12 @@ out_copy:
 	exid->clientid.cl_boot = new->cl_clientid.cl_boot;
 	exid->clientid.cl_id = new->cl_clientid.cl_id;
 
-	new->cl_seqid = exid->seqid = 1;
+	new->cl_slot.sl_seqid = 0;
+	exid->seqid = 1;
 	nfsd4_set_ex_flags(new, exid);
 
 	dprintk("nfsd4_exchange_id seqid %d flags %x\n",
-		new->cl_seqid, new->cl_exchange_flags);
+		new->cl_slot.sl_seqid, new->cl_exchange_flags);
 	status = nfs_ok;
 
 out:
@@ -1322,7 +1329,9 @@ nfsd4_create_session(struct svc_rqst *rqstp,
 		     struct nfsd4_create_session *cr_ses)
 {
 	u32 ip_addr = svc_addr_in(rqstp)->sin_addr.s_addr;
+	struct nfsd4_compoundres *resp = rqstp->rq_resp;
 	struct nfs4_client *conf, *unconf;
+	struct nfsd4_slot *slot = NULL;
 	int status = 0;
 
 	nfs4_lock_state();
@@ -1330,19 +1339,24 @@ nfsd4_create_session(struct svc_rqst *rqstp,
 	conf = find_confirmed_client(&cr_ses->clientid);
 
 	if (conf) {
-		status = nfs_ok;
-		if (conf->cl_seqid == cr_ses->seqid) {
+		slot = &conf->cl_slot;
+		status = check_slot_seqid(cr_ses->seqid, slot);
+		if (status == nfserr_replay_cache) {
 			dprintk("Got a create_session replay! seqid= %d\n",
-				conf->cl_seqid);
-			goto out_replay;
-		} else if (cr_ses->seqid != conf->cl_seqid + 1) {
+				slot->sl_seqid);
+			cstate->slot = slot;
+			cstate->status = status;
+			/* Return the cached reply status */
+			status = nfsd4_replay_cache_entry(resp);
+			goto out;
+		} else if (cr_ses->seqid != conf->cl_slot.sl_seqid + 1) {
 			status = nfserr_seq_misordered;
 			dprintk("Sequence misordered!\n");
 			dprintk("Expected seqid= %d but got seqid= %d\n",
-				conf->cl_seqid, cr_ses->seqid);
+				slot->sl_seqid, cr_ses->seqid);
 			goto out;
 		}
-		conf->cl_seqid++;
+		conf->cl_slot.sl_seqid++;
 	} else if (unconf) {
 		if (!same_creds(&unconf->cl_cred, &rqstp->rq_cred) ||
 		    (ip_addr != unconf->cl_addr)) {
@@ -1350,11 +1364,15 @@ nfsd4_create_session(struct svc_rqst *rqstp,
 			goto out;
 		}
 
-		if (unconf->cl_seqid != cr_ses->seqid) {
+		slot = &unconf->cl_slot;
+		status = check_slot_seqid(cr_ses->seqid, slot);
+		if (status) {
+			/* an unconfirmed replay returns misordered */
 			status = nfserr_seq_misordered;
 			goto out;
 		}
 
+		slot->sl_seqid++; /* from 0 to 1 */
 		move_to_confirmed(unconf);
 
 		/*
@@ -1373,11 +1391,12 @@ nfsd4_create_session(struct svc_rqst *rqstp,
 	if (status)
 		goto out;
 
-out_replay:
 	memcpy(cr_ses->sessionid.data, conf->cl_sessionid.data,
 	       NFS4_MAX_SESSIONID_LEN);
-	cr_ses->seqid = conf->cl_seqid;
+	cr_ses->seqid = slot->sl_seqid;
 
+	slot->sl_inuse = true;
+	cstate->slot = slot;
 out:
 	nfs4_unlock_state();
 	dprintk("%s returns %d\n", __func__, ntohl(status));
include/linux/nfsd/state.h +1 -1
index 98d7b1c..49d89fd 100644
@@ -190,7 +190,7 @@ struct nfs4_client {
 	u32			cl_firststate;	/* recovery dir creation */
 #ifdef CONFIG_NFSD_V4_1
 	struct list_head	cl_sessions;
-	u32			cl_seqid;       /* seqid for create_session */
+	struct nfsd4_slot	cl_slot;	/* create_session slot */
 	u32			cl_exchange_flags;
 	struct nfs4_sessionid	cl_sessionid;
 
Signature · 1 line
1.6.2.1
#158 lore raw

PATCH v2 25/47 nfsd41: non-page DRC for solo sequence responses

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

A session inactivity time compound (lease renewal) or a compound where the sequence operation has sa_cachethis set to FALSE do not require any pages to be held in the v4.1 DRC. This is because struct nfsd4_slot is already caching the session information.

Add logic to the nfs41 server to not cache response pages for solo sequence responses.

Return nfserr_replay_uncached_rep on the operation following the sequence operation when sa_cachethis is FALSE.

---
 fs/nfsd/nfs4proc.c         |   34 +++++++++++++++++++++++++++++-
 fs/nfsd/nfs4state.c        |   47 ++++++++++++++++++++++++++++++++++++++-----
 fs/nfsd/nfs4xdr.c          |    5 ++-
 include/linux/nfsd/state.h |    1 +
 include/linux/nfsd/xdr4.h  |   15 +++++++++++++-
 5 files changed, 91 insertions(+), 11 deletions(-)
fs/nfsd/nfs4proc.c +32 -2
index bdbeb87..a273023 100644
@@ -828,6 +828,34 @@ static struct nfsd4_operation nfsd4_ops[];
 static const char *nfsd4_op_name(unsigned opnum);
 
 /*
+ * This is a replay of a compound for which no cache entry pages
+ * were used. Encode the sequence operation, and if cachethis is FALSE
+ * encode the uncache rep error on the next operation.
+ */
+static __be32
+nfsd4_enc_no_page_replay(struct nfsd4_compoundargs *args,
+			 struct nfsd4_compoundres *resp)
+{
+	struct nfsd4_op *op;
+
+	dprintk("--> %s resp->opcnt %d ce_cachethis %u \n", __func__,
+		resp->opcnt, resp->cstate.slot->sl_cache_entry.ce_cachethis);
+
+	/* Encode the replayed sequence operation */
+	BUG_ON(resp->opcnt != 1);
+	op = &args->ops[resp->opcnt - 1];
+	nfsd4_encode_operation(resp, op);
+
+	/*return nfserr_retry_uncached_rep in next operation. */
+	if (resp->cstate.slot->sl_cache_entry.ce_cachethis == 0) {
+		op = &args->ops[resp->opcnt++];
+		op->status = nfserr_retry_uncached_rep;
+		nfsd4_encode_operation(resp, op);
+	}
+	return op->status;
+}
+
+/*
  * COMPOUND call.
  */
 static __be32
@@ -879,7 +907,6 @@ nfsd4_proc_compound(struct svc_rqst *rqstp,
 		dprintk("nfsv4 compound op #%d/%d: %d (%s)\n",
 			resp->opcnt, args->opcnt, op->opnum,
 			nfsd4_op_name(op->opnum));
-
 		/*
 		 * The XDR decode routines may have pre-set op->status;
 		 * for example, if there is a miscellaneous XDR error
@@ -923,7 +950,10 @@ encode_op:
 		/* Only from SEQUENCE or CREATE_SESSION */
 		if (resp->cstate.status == nfserr_replay_cache) {
 			dprintk("%s NFS4.1 replay from cache\n", __func__);
-			status = op->status;
+			if (nfsd4_no_page_in_cache(resp))
+				status = nfsd4_enc_no_page_replay(args, resp);
+			else
+				status = op->status;
 			goto out;
 		}
 		if (op->status == nfserr_replay_me) {
fs/nfsd/nfs4state.c +41 -6

98 lines not on this page · raw message

fs/nfsd/nfs4xdr.c +3 -2
index 60db854..a8bb04a 100644
@@ -2984,7 +2984,7 @@ nfsd4_encode_destroy_session(struct nfsd4_compoundres *resp, int nfserr,
 	return nfserr;
 }
 
-static __be32
+__be32
 nfsd4_encode_sequence(struct nfsd4_compoundres *resp, int nfserr,
 		      struct nfsd4_sequence *seq)
 {
@@ -3204,7 +3204,8 @@ nfs4svc_encode_compoundres(struct svc_rqst *rqstp, __be32 *p, struct nfsd4_compo
 	BUG_ON(iov->iov_len > PAGE_SIZE);
 #ifdef CONFIG_NFSD_V4_1
 	if (resp->cstate.slot != NULL) {
-		if (resp->cstate.status == nfserr_replay_cache) {
+		if (resp->cstate.status == nfserr_replay_cache &&
+				!nfsd4_no_page_in_cache(resp)) {
 			iov->iov_len = resp->cstate.iovlen;
 		} else {
 			nfsd4_set_cache_entry(resp);
include/linux/nfsd/state.h +1 -0
index 49d89fd..47c7836 100644
@@ -110,6 +110,7 @@ struct nfsd4_cache_entry {
 	__be32		ce_status;
 	struct kvec	ce_datav; /* encoded NFSv4.1 data in rq_res.head[0] */
 	struct page	*ce_respages[NFSD_PAGES_PER_SLOT + 1];
+	int		ce_cachethis;
 	short		ce_resused;
 	int		ce_opcnt;
 	int		ce_rpchdrlen;
include/linux/nfsd/xdr4.h +14 -1
index c7bf0a1..641e5cd 100644
@@ -482,6 +482,18 @@ struct nfsd4_compoundres {
 	struct nfsd4_compound_state	cstate;
 };
 
+static inline u32 nfsd4_is_solo_sequence(struct nfsd4_compoundres *resp)
+{
+	struct nfsd4_compoundargs *args = resp->rqstp->rq_argp;
+	return args->opcnt == 1 ? 1 : 0;
+}
+
+static inline u32 nfsd4_no_page_in_cache(struct nfsd4_compoundres *resp)
+{
+	return (resp->cstate.slot->sl_cache_entry.ce_cachethis == 0 ||
+			nfsd4_is_solo_sequence(resp));
+}
+
 #define NFS4_SVC_XDRSIZE		sizeof(struct nfsd4_compoundargs)
 
 static inline void
@@ -513,7 +525,8 @@ extern __be32 nfsd4_setclientid_confirm(struct svc_rqst *rqstp,
 		struct nfsd4_setclientid_confirm *setclientid_confirm);
 #if defined(CONFIG_NFSD_V4_1)
 extern void nfsd4_set_cache_entry(struct nfsd4_compoundres *resp);
-extern __be32 nfsd4_replay_cache_entry(struct nfsd4_compoundres *resp);
+extern __be32 nfsd4_replay_cache_entry(struct nfsd4_compoundres *resp,
+		struct nfsd4_sequence *seq);
 extern __be32 nfsd4_exchange_id(struct svc_rqst *rqstp,
 		struct nfsd4_compound_state *,
 struct nfsd4_exchange_id *);
Signature · 1 line
1.6.2.1
#161 lore raw

PATCH v2 26/47 nfsd41: destroy_session operation

In reply to Benny Halevy

Implement the destory_session operation confoming to
http://tools.ietf.org/html/draft-ietf-nfsv4-minorversion1-26

[use sessionid_lock spin lock]
---
 fs/nfsd/nfs4state.c       |   29 ++++++++++++++++++++++++++++-
 fs/nfsd/nfs4xdr.c         |    7 +++++--
 include/linux/nfsd/xdr4.h |    2 +-
 3 files changed, 34 insertions(+), 4 deletions(-)
fs/nfsd/nfs4state.c +28 -1
index f42cda9..9c93f96 100644
@@ -1439,7 +1439,34 @@ nfsd4_destroy_session(struct svc_rqst *r,
 		      struct nfsd4_compound_state *cstate,
 		      struct nfsd4_destroy_session *sessionid)
 {
-	return -1;	/* stub */
+	struct nfsd4_session *ses;
+	u32 status = nfserr_badsession;
+
+	/* Notes:
+	 * - The confirmed nfs4_client->cl_sessionid holds destroyed sessinid
+	 * - Should we return nfserr_back_chan_busy if waiting for
+	 *   callbacks on to-be-destroyed session?
+	 * - Do we need to clear any callback info from previous session?
+	 */
+
+	dump_sessionid(__func__, &sessionid->sessionid);
+	spin_lock(&sessionid_lock);
+	ses = find_in_sessionid_hashtbl(&sessionid->sessionid);
+	if (!ses) {
+		spin_unlock(&sessionid_lock);
+		goto out;
+	}
+
+	unhash_session(ses);
+	spin_unlock(&sessionid_lock);
+
+	/* wait for callbacks */
+	shutdown_callback_client(ses->se_client);
+	nfsd4_put_session(ses);
+	status = nfs_ok;
+out:
+	dprintk("%s returns %d\n", __func__, ntohl(status));
+	return status;
 }
 
 __be32
fs/nfsd/nfs4xdr.c +5 -2
index a8bb04a..5720aab 100644
@@ -1208,7 +1208,11 @@ static __be32
 nfsd4_decode_destroy_session(struct nfsd4_compoundargs *argp,
 			     struct nfsd4_destroy_session *destroy_session)
 {
-	return nfserr_opnotsupp;	/* stub */
+	DECODE_HEAD;
+	READ_BUF(NFS4_MAX_SESSIONID_LEN);
+	COPYMEM(destroy_session->sessionid.data, NFS4_MAX_SESSIONID_LEN);
+
+	DECODE_TAIL;
 }
 
 static __be32
@@ -2980,7 +2984,6 @@ static __be32
 nfsd4_encode_destroy_session(struct nfsd4_compoundres *resp, int nfserr,
 			     struct nfsd4_destroy_session *destroy_session)
 {
-	/* stub */
 	return nfserr;
 }
 
include/linux/nfsd/xdr4.h +1 -1
index 641e5cd..37a7c51 100644
@@ -396,7 +396,7 @@ struct nfsd4_sequence {
 };
 
 struct nfsd4_destroy_session {
-	int	foo;	/* stub */
+	struct nfs4_sessionid	sessionid;
 };
 #endif /* CONFIG_NFSD_V4_1 */
 
Signature · 1 line
1.6.2.1
#162 lore raw

PATCH v2 27/47 nfsd41: stateid handling

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

When sessions are used, stateful operation sequenceid and stateid handling are not used. When sessions are used, on the first open set the seqid to 1, mark state confirmed and skip seqid processing.

When sessionas are used the stateid generation number is ignored when it is zero whereas without sessions bad_stateid or stale stateid is returned.

Add flags to propagate session use to all stateful ops and down to check_stateid_generation.

[nfsd4_has_session should return a boolean, not u32]
---
 fs/nfsd/nfs4proc.c         |   17 ++++++++--
 fs/nfsd/nfs4state.c        |   70 ++++++++++++++++++++++++++++++++++---------
 fs/nfsd/nfs4xdr.c          |    2 +-
 include/linux/nfsd/state.h |    1 +
 include/linux/nfsd/xdr4.h  |    8 ++++-
 5 files changed, 77 insertions(+), 21 deletions(-)
fs/nfsd/nfs4proc.c +13 -4
index a273023..1d4b2b5 100644
@@ -179,7 +179,7 @@ nfsd4_open(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 	nfs4_lock_state();
 
 	/* check seqid for replay. set nfs4_owner */
-	status = nfsd4_process_open1(open);
+	status = nfsd4_process_open1(rqstp, open);
 	if (status == nfserr_replay_me) {
 		struct nfs4_replay *rp = &open->op_stateowner->so_replay;
 		fh_put(&cstate->current_fh);
@@ -504,6 +504,7 @@ nfsd4_read(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 	   struct nfsd4_read *read)
 {
 	__be32 status;
+	int flags = RD_STATE;
 
 	/* no need to check permission - this will be done in nfsd_read() */
 
@@ -511,11 +512,13 @@ nfsd4_read(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 	if (read->rd_offset >= OFFSET_MAX)
 		return nfserr_inval;
 
+	if (nfsd4_has_session(cstate))
+		flags |= HAS_SESSION;
 	nfs4_lock_state();
 	/* check stateid */
 	if ((status = nfs4_preprocess_stateid_op(&cstate->current_fh,
 				&read->rd_stateid,
-				RD_STATE, &read->rd_filp))) {
+				flags, &read->rd_filp))) {
 		dprintk("NFSD: nfsd4_read: couldn't process stateid!\n");
 		goto out;
 	}
@@ -643,11 +646,14 @@ nfsd4_setattr(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 	      struct nfsd4_setattr *setattr)
 {
 	__be32 status = nfs_ok;
+	int flags = WR_STATE;
 
+	if (nfsd4_has_session(cstate))
+		flags |= HAS_SESSION;
 	if (setattr->sa_iattr.ia_valid & ATTR_SIZE) {
 		nfs4_lock_state();
 		status = nfs4_preprocess_stateid_op(&cstate->current_fh,
-			&setattr->sa_stateid, WR_STATE, NULL);
+			&setattr->sa_stateid, flags, NULL);
 		nfs4_unlock_state();
 		if (status) {
 			dprintk("NFSD: nfsd4_setattr: couldn't process stateid!\n");
@@ -679,15 +685,18 @@ nfsd4_write(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 	u32 *p;
 	__be32 status = nfs_ok;
 	unsigned long cnt;
+	int flags = WR_STATE;
 
 	/* no need to check permission - this will be done in nfsd_write() */
 
 	if (write->wr_offset >= OFFSET_MAX)
 		return nfserr_inval;
 
+	if (nfsd4_has_session(cstate))
+		flags |= HAS_SESSION;
 	nfs4_lock_state();
 	status = nfs4_preprocess_stateid_op(&cstate->current_fh, stateid,
-					WR_STATE, &filp);
+					    flags, &filp);
 	if (filp)
 		get_file(filp);
 	nfs4_unlock_state();
fs/nfsd/nfs4state.c +55 -15

237 lines not on this page · raw message

fs/nfsd/nfs4xdr.c +1 -1
index 5720aab..a2682e8 100644
@@ -3206,7 +3206,7 @@ nfs4svc_encode_compoundres(struct svc_rqst *rqstp, __be32 *p, struct nfsd4_compo
 	iov->iov_len = ((char*)resp->p) - (char*)iov->iov_base;
 	BUG_ON(iov->iov_len > PAGE_SIZE);
 #ifdef CONFIG_NFSD_V4_1
-	if (resp->cstate.slot != NULL) {
+	if (nfsd4_has_session(&resp->cstate)) {
 		if (resp->cstate.status == nfserr_replay_cache &&
 				!nfsd4_no_page_in_cache(resp)) {
 			iov->iov_len = resp->cstate.iovlen;
include/linux/nfsd/state.h +1 -0
index 47c7836..302557d 100644
@@ -323,6 +323,7 @@ struct nfs4_stateid {
 };
 
 /* flags for preprocess_seqid_op() */
+#define HAS_SESSION             0x00000001
 #define CONFIRM                 0x00000002
 #define OPEN_STATE              0x00000004
 #define LOCK_STATE              0x00000008
include/linux/nfsd/xdr4.h +7 -1
index 37a7c51..aafbfdc 100644
@@ -55,6 +55,11 @@ struct nfsd4_compound_state {
 	u32			status;
 };
 
+static inline bool nfsd4_has_session(struct nfsd4_compound_state *cs)
+{
+	return cs->slot != NULL;
+}
+
 struct nfsd4_change_info {
 	u32		atomic;
 	u32		before_ctime_sec;
@@ -540,7 +545,8 @@ extern __be32 nfsd4_destroy_session(struct svc_rqst *,
 		struct nfsd4_compound_state *,
 		struct nfsd4_destroy_session *);
 #endif /* CONFIG_NFSD_V4_1 */
-extern __be32 nfsd4_process_open1(struct nfsd4_open *open);
+extern __be32 nfsd4_process_open1(struct svc_rqst *rqstp,
+		struct nfsd4_open *open);
 extern __be32 nfsd4_process_open2(struct svc_rqst *rqstp,
 		struct svc_fh *current_fh, struct nfsd4_open *open);
 extern __be32 nfsd4_open_confirm(struct svc_rqst *rqstp,
Signature · 1 line
1.6.2.1
#164 lore raw

PATCH v2 28/47 nfsd41: check encode size for sessions maxresponse cached

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

Calculate the space the compound response has taken after encoding the current operation.

pad: add on 8 bytes for the next operation's op_code and status so that there is room to cache a failure on the next operation.

Compare this length to the session se_fmaxresp_cached and return nfserr_rep_too_big_to_cache if the length is too large.

Our se_fmaxresp_cached will always be a multiple of PAGE_SIZE, and so will be at least a page and will therefore hold the xdr_buf head.

[nfsd41: non-page DRC for solo sequence responses]
[fixed nfsd4_check_drc_limit cosmetics]
---
 fs/nfsd/nfs4xdr.c |   58 +++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 files changed, 58 insertions(+), 0 deletions(-)
fs/nfsd/nfs4xdr.c +58 -0
index a2682e8..52ca833 100644
@@ -3089,6 +3089,61 @@ static nfsd4_enc nfsd4_enc_ops[] = {
 #endif /* CONFIG_NFSD_V4_1 */
 };
 
+#if defined(CONFIG_NFSD_V4_1)
+/*
+ * Calculate the total amount of memory that the compound response has taken
+ * after encoding the current operation.
+ *
+ * pad: add on 8 bytes for the next operation's op_code and status so that
+ * there is room to cache a failure on the next operation.
+ *
+ * Compare this length to the session se_fmaxresp_cached.
+ *
+ * Our se_fmaxresp_cached will always be a multiple of PAGE_SIZE, and so
+ * will be at least a page and will therefore hold the xdr_buf head.
+ */
+static int nfsd4_check_drc_limit(struct nfsd4_compoundres *resp)
+{
+	int status = 0;
+	struct xdr_buf *xb = &resp->rqstp->rq_res;
+	struct nfsd4_compoundargs *args = resp->rqstp->rq_argp;
+	struct nfsd4_session *session = NULL;
+	struct nfsd4_slot *slot = resp->cstate.slot;
+	u32 length, tlen = 0, pad = 8;
+
+	if (!nfsd4_has_session(&resp->cstate))
+		return status;
+
+	session = slot->sl_session;
+	if (session == NULL || slot->sl_cache_entry.ce_cachethis == 0)
+		return status;
+
+	if (resp->opcnt >= args->opcnt)
+		pad = 0; /* this is the last operation */
+
+	if (xb->page_len == 0) {
+		length = (char *)resp->p - (char *)xb->head[0].iov_base + pad;
+	} else {
+		if (xb->tail[0].iov_base && xb->tail[0].iov_len > 0)
+			tlen = (char *)resp->p - (char *)xb->tail[0].iov_base;
+
+		length = xb->head[0].iov_len + xb->page_len + tlen + pad;
+	}
+	dprintk("%s length %u, xb->page_len %u tlen %u pad %u\n", __func__,
+		length, xb->page_len, tlen, pad);
+
+	if (length <= session->se_fmaxresp_cached)
+		return status;
+	else
+		return nfserr_rep_too_big_to_cache;
+}
+#else /* CONFIG_NFSD_V4_1 */
+static inline int nfsd4_check_drc_limit(struct nfsd4_compoundres *resp)
+{
+	return 0;
+}
+#endif /* CONFIG_NFSD_V4_1 */
+
 void
 nfsd4_encode_operation(struct nfsd4_compoundres *resp, struct nfsd4_op *op)
 {
@@ -3105,6 +3160,9 @@ nfsd4_encode_operation(struct nfsd4_compoundres *resp, struct nfsd4_op *op)
 	BUG_ON(op->opnum < 0 || op->opnum >= ARRAY_SIZE(nfsd4_enc_ops) ||
 	       !nfsd4_enc_ops[op->opnum]);
 	op->status = nfsd4_enc_ops[op->opnum](resp, op->status, &op->u);
+	/* nfsd4_check_drc_limit guarantees enough room for error status */
+	if (!op->status && nfsd4_check_drc_limit(resp))
+		op->status = nfserr_rep_too_big_to_cache;
 status:
 	/*
 	 * Note: We write the status directly, instead of using WRITE32(),
Signature · 1 line
1.6.2.1
#166 lore raw

PATCH v2 29/47 nfsd41: clientid handling

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

Extract the clientid from sessionid to set the op_clientid on open. Verify that the clid for other stateful ops is zero for minorversion != 0 Do all other checks for stateful ops without sessions.

[fixed whitespace indent]
---
 fs/nfsd/nfs4proc.c  |   12 ++++++++++++
 fs/nfsd/nfs4state.c |   14 ++++++++------
 fs/nfsd/nfs4xdr.c   |    9 +++++++++
 3 files changed, 29 insertions(+), 6 deletions(-)
fs/nfsd/nfs4proc.c +12 -0
index 1d4b2b5..0076bb1 100644
@@ -162,6 +162,15 @@ do_open_fhandle(struct svc_rqst *rqstp, struct svc_fh *current_fh, struct nfsd4_
 	return status;
 }
 
+static void
+copy_clientid(clientid_t *clid, struct nfsd4_session *session)
+{
+	struct nfsd4_sessionid *sid =
+			(struct nfsd4_sessionid *)session->se_sessionid.data;
+
+	clid->cl_boot = sid->clientid.cl_boot;
+	clid->cl_id = sid->clientid.cl_id;
+}
 
 static __be32
 nfsd4_open(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
@@ -176,6 +185,9 @@ nfsd4_open(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 	if (open->op_create && open->op_claim_type != NFS4_OPEN_CLAIM_NULL)
 		return nfserr_inval;
 
+	if (nfsd4_has_session(cstate))
+		copy_clientid(&open->op_clientid, cstate->slot->sl_session);
+
 	nfs4_lock_state();
 
 	/* check seqid for replay. set nfs4_owner */
fs/nfsd/nfs4state.c +8 -6
index bf5b214..00ad1c2 100644
@@ -616,8 +616,8 @@ STALE_CLIENTID(clientid_t *clid)
 {
 	if (clid->cl_boot == boot_time)
 		return 0;
-	dprintk("NFSD stale clientid (%08x/%08x)\n", 
-			clid->cl_boot, clid->cl_id);
+	dprintk("NFSD stale clientid (%08x/%08x) boot_time %08lx\n",
+		clid->cl_boot, clid->cl_id, boot_time);
 	return 1;
 }
 
@@ -2968,8 +2968,9 @@ nfs4_preprocess_seqid_op(struct svc_fh *current_fh, u32 seqid, stateid_t *statei
 		if (lock->lk_is_new) {
 			if (!sop->so_is_open_owner)
 				return nfserr_bad_stateid;
-			if (!same_clid(&clp->cl_clientid, lockclid))
-			       return nfserr_bad_stateid;
+			if (!(flags & HAS_SESSION) &&
+			    !same_clid(&clp->cl_clientid, lockclid))
+				return nfserr_bad_stateid;
 			/* stp is the open stateid */
 			status = nfs4_check_openmode(stp, lkflg);
 			if (status)
@@ -3516,7 +3517,8 @@ nfsd4_lock(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 		struct nfs4_file *fp;
 		
 		status = nfserr_stale_clientid;
-		if (STALE_CLIENTID(&lock->lk_new_clientid))
+		if (!nfsd4_has_session(cstate) &&
+		    STALE_CLIENTID(&lock->lk_new_clientid))
 			goto out;
 
 		flags = OPEN_STATE;
@@ -3678,7 +3680,7 @@ nfsd4_lockt(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 	nfs4_lock_state();
 
 	status = nfserr_stale_clientid;
-	if (STALE_CLIENTID(&lockt->lt_clientid))
+	if (!nfsd4_has_session(cstate) && STALE_CLIENTID(&lockt->lt_clientid))
 		goto out;
 
 	if ((status = fh_verify(rqstp, &cstate->current_fh, S_IFREG, 0))) {
fs/nfsd/nfs4xdr.c +9 -0
index 52ca833..a1af3a3 100644
@@ -189,6 +189,11 @@ static __be32 *read_buf(struct nfsd4_compoundargs *argp, u32 nbytes)
 	return p;
 }
 
+static int zero_clientid(clientid_t *clid)
+{
+	return (clid->cl_boot == 0) && (clid->cl_id == 0);
+}
+
 static int
 defer_free(struct nfsd4_compoundargs *argp,
 		void (*release)(const void *), void *p)
@@ -584,6 +589,8 @@ nfsd4_decode_lockt(struct nfsd4_compoundargs *argp, struct nfsd4_lockt *lockt)
 	READ_BUF(lockt->lt_owner.len);
 	READMEM(lockt->lt_owner.data, lockt->lt_owner.len);
 
+	if (argp->minorversion && !zero_clientid(&lockt->lt_clientid))
+		return nfserr_inval;
 	DECODE_TAIL;
 }
 
@@ -994,6 +1001,8 @@ nfsd4_decode_release_lockowner(struct nfsd4_compoundargs *argp, struct nfsd4_rel
 	READ_BUF(rlockowner->rl_owner.len);
 	READMEM(rlockowner->rl_owner.data, rlockowner->rl_owner.len);
 
+	if (argp->minorversion && !zero_clientid(&rlockowner->rl_clientid))
+		return nfserr_inval;
 	DECODE_TAIL;
 }
 
Signature · 1 line
1.6.2.1
#167 lore raw

PATCH v2 30/47 nfsd41: access_valid

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

For nfs41, the open share flags are used also for delegation "wants" and "signals". Check that they are valid.

---
 fs/nfsd/nfs4proc.c        |    1 +
 fs/nfsd/nfs4state.c       |   20 +++++++++++++++-----
 include/linux/nfsd/xdr4.h |    1 +
 3 files changed, 17 insertions(+), 5 deletions(-)
fs/nfsd/nfs4proc.c +1 -0
index 0076bb1..8b4bf5e 100644
@@ -901,6 +901,7 @@ nfsd4_proc_compound(struct svc_rqst *rqstp,
 	resp->tag = args->tag;
 	resp->opcnt = 0;
 	resp->rqstp = rqstp;
+	resp->cstate.minorversion = args->minorversion;
 	resp->cstate.replay_owner = NULL;
 	fh_init(&resp->cstate.current_fh, NFS4_FHSIZE);
 	fh_init(&resp->cstate.save_fh, NFS4_FHSIZE);
fs/nfsd/nfs4state.c +15 -5
index 00ad1c2..dbf0d07 100644
@@ -1959,11 +1959,21 @@ find_file(struct inode *ino)
 	return NULL;
 }
 
-static inline int access_valid(u32 x)
+static inline int access_valid(u32 x, u32 minorversion)
 {
-	if (x < NFS4_SHARE_ACCESS_READ)
+	if ((x & NFS4_SHARE_ACCESS_MASK) < NFS4_SHARE_ACCESS_READ)
 		return 0;
-	if (x > NFS4_SHARE_ACCESS_BOTH)
+	if ((x & NFS4_SHARE_ACCESS_MASK) > NFS4_SHARE_ACCESS_BOTH)
+		return 0;
+	x &= ~NFS4_SHARE_ACCESS_MASK;
+	if (minorversion && x) {
+		if ((x & NFS4_SHARE_WANT_MASK) > NFS4_SHARE_WANT_CANCEL)
+			return 0;
+		if ((x & NFS4_SHARE_WHEN_MASK) > NFS4_SHARE_PUSH_DELEG_WHEN_UNCONTENDED)
+			return 0;
+		x &= ~(NFS4_SHARE_WANT_MASK | NFS4_SHARE_WHEN_MASK);
+	}
+	if (x)
 		return 0;
 	return 1;
 }
@@ -2511,7 +2521,7 @@ nfsd4_process_open2(struct svc_rqst *rqstp, struct svc_fh *current_fh, struct nf
 	__be32 status;
 
 	status = nfserr_inval;
-	if (!access_valid(open->op_share_access)
+	if (!access_valid(open->op_share_access, resp->cstate.minorversion)
 			|| !deny_valid(open->op_share_deny))
 		goto out;
 	/*
@@ -3108,7 +3118,7 @@ nfsd4_open_downgrade(struct svc_rqst *rqstp,
 			(int)cstate->current_fh.fh_dentry->d_name.len,
 			cstate->current_fh.fh_dentry->d_name.name);
 
-	if (!access_valid(od->od_share_access)
+	if (!access_valid(od->od_share_access, cstate->minorversion)
 			|| !deny_valid(od->od_share_deny))
 		return nfserr_inval;
 
include/linux/nfsd/xdr4.h +1 -0
index aafbfdc..881ff67 100644
@@ -52,6 +52,7 @@ struct nfsd4_compound_state {
 	struct nfsd4_slot	*slot;
 	__be32			*statp;
 	size_t			iovlen;
+	u32			minorversion;
 	u32			status;
 };
 
Signature · 1 line
1.6.2.1
#168 lore raw

PATCH v2 31/47 nfsd41: add OPEN4_SHARE_ACCESS_WANT nfs4_stateid bmap

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

Separate the access bits from the want bits and enable __set_bit to work correctly with st_access_bmap.

---
 fs/nfsd/nfs4state.c |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)
fs/nfsd/nfs4state.c +2 -1
index dbf0d07..8a64981 100644
@@ -1905,7 +1905,8 @@ init_stateid(struct nfs4_stateid *stp, struct nfs4_file *fp, struct nfsd4_open *
 	stp->st_stateid.si_generation = 0;
 	stp->st_access_bmap = 0;
 	stp->st_deny_bmap = 0;
-	__set_bit(open->op_share_access, &stp->st_access_bmap);
+	__set_bit(open->op_share_access & ~NFS4_SHARE_WANT_MASK,
+		  &stp->st_access_bmap);
 	__set_bit(open->op_share_deny, &stp->st_deny_bmap);
 	stp->st_openstp = NULL;
 }
Signature · 1 line
1.6.2.1
#169 lore raw

PATCH v2 32/47 nfsd41: provide support for minor version 1 at rpc level

In reply to Benny Halevy

From: Marc Eshel <[email protected]>

---
 include/linux/nfsd/nfsd.h |    4 ++++
 1 files changed, 4 insertions(+), 0 deletions(-)
include/linux/nfsd/nfsd.h +4 -0
index 3bccec3..d0fe9b4 100644
@@ -23,7 +23,11 @@
 /*
  * nfsd version
  */
+#if defined(CONFIG_NFSD_V4_1)
+#define NFSD_SUPPORTED_MINOR_VERSION	1
+#else /* CONFIG_NFSD_V4_1 */
 #define NFSD_SUPPORTED_MINOR_VERSION	0
+#endif /* CONFIG_NFSD_V4_1 */
 
 /*
  * Flags for nfsd_permission
Signature · 1 line
1.6.2.1
#170 lore raw

PATCH v2 33/47 nfsd: cleanup nfs4.0 callback encode routines

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

Mimic the client and prepare to share the back channel xdr with NFSv4.1. Bump the number of operations in each encode routine, then backfill the number of operations.

---
 fs/nfsd/nfs4callback.c |   24 ++++++++++++++++--------
 1 files changed, 16 insertions(+), 8 deletions(-)
fs/nfsd/nfs4callback.c +16 -8
index 290289b..7129b0c 100644
@@ -140,8 +140,9 @@ struct nfs4_cb_compound_hdr {
 	int		status;
 	u32		ident;
 	u32		nops;
+	__be32		*nops_p;
 	u32		taglen;
-	char *		tag;
+	char		*tag;
 };
 
 static struct {
@@ -201,7 +202,7 @@ nfs_cb_stat_to_errno(int stat)
  * XDR encode
  */
 
-static int
+static void
 encode_cb_compound_hdr(struct xdr_stream *xdr, struct nfs4_cb_compound_hdr *hdr)
 {
 	__be32 * p;
@@ -210,12 +211,18 @@ encode_cb_compound_hdr(struct xdr_stream *xdr, struct nfs4_cb_compound_hdr *hdr)
 	WRITE32(0);            /* tag length is always 0 */
 	WRITE32(NFS4_MINOR_VERSION);
 	WRITE32(hdr->ident);
+	hdr->nops_p = p;
 	WRITE32(hdr->nops);
-	return 0;
 }
 
-static int
-encode_cb_recall(struct xdr_stream *xdr, struct nfs4_cb_recall *cb_rec)
+static void encode_cb_nops(struct nfs4_cb_compound_hdr *hdr)
+{
+	*hdr->nops_p = htonl(hdr->nops);
+}
+
+static void
+encode_cb_recall(struct xdr_stream *xdr, struct nfs4_cb_recall *cb_rec,
+		 struct nfs4_cb_compound_hdr *hdr)
 {
 	__be32 *p;
 	int len = cb_rec->cbr_fh.fh_size;
@@ -227,7 +234,7 @@ encode_cb_recall(struct xdr_stream *xdr, struct nfs4_cb_recall *cb_rec)
 	WRITE32(cb_rec->cbr_trunc);
 	WRITE32(len);
 	WRITEMEM(&cb_rec->cbr_fh.fh_base, len);
-	return 0;
+	hdr->nops++;
 }
 
 static int
@@ -246,12 +253,13 @@ nfs4_xdr_enc_cb_recall(struct rpc_rqst *req, __be32 *p, struct nfs4_cb_recall *a
 	struct xdr_stream xdr;
 	struct nfs4_cb_compound_hdr hdr = {
 		.ident = args->cbr_ident,
-		.nops   = 1,
 	};
 
 	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
 	encode_cb_compound_hdr(&xdr, &hdr);
-	return (encode_cb_recall(&xdr, args));
+	encode_cb_recall(&xdr, args, &hdr);
+	encode_cb_nops(&hdr);
+	return 0;
 }
 
 
Signature · 1 line
1.6.2.1
#171 lore raw

PATCH v2 34/47 sunrpc: add cl_private field to struct rpc_clnt

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

Note: Both the NFSv4.1 client and server use this pointer.

The client uses cl_private to determine the minorversion (via a struct nfs_client *) to be set in the compound header, and to know when to generate a SEQUENCE op. Similarly, the server puts a struct nfs4_callback * in there for callback compounds' CB_COMPOUND and CB_SEQUENCE.

---
 include/linux/sunrpc/clnt.h |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)
include/linux/sunrpc/clnt.h +1 -0
index c39a210..928c547 100644
@@ -53,6 +53,7 @@ struct rpc_clnt {
 	char			cl_pathname[30];/* Path in rpc_pipe_fs */
 	struct vfsmount *	cl_vfsmnt;
 	struct dentry *		cl_dentry;	/* inode */
+	void			*cl_private;	/* private data */
 	struct rpc_clnt *	cl_parent;	/* Points to parent of clones */
 	struct rpc_rtt		cl_rtt_default;
 	struct rpc_timeout	cl_timeout_default;
Signature · 1 line
1.6.2.1
#172 lore raw

PATCH v2 35/47 nfsd: minorversion support for the back channel

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

Prepare to share backchannel code with NFSv4.1.

---
 fs/nfsd/nfs4callback.c     |    7 ++++++-
 fs/nfsd/nfs4state.c        |    1 +
 include/linux/nfsd/state.h |    3 ++-
 3 files changed, 9 insertions(+), 2 deletions(-)
fs/nfsd/nfs4callback.c +6 -1
index 7129b0c..d8f2631 100644
@@ -141,6 +141,7 @@ struct nfs4_cb_compound_hdr {
 	u32		ident;
 	u32		nops;
 	__be32		*nops_p;
+	u32		minorversion;
 	u32		taglen;
 	char		*tag;
 };
@@ -209,7 +210,7 @@ encode_cb_compound_hdr(struct xdr_stream *xdr, struct nfs4_cb_compound_hdr *hdr)
 
 	RESERVE_SPACE(16);
 	WRITE32(0);            /* tag length is always 0 */
-	WRITE32(NFS4_MINOR_VERSION);
+	WRITE32(hdr->minorversion);
 	WRITE32(hdr->ident);
 	hdr->nops_p = p;
 	WRITE32(hdr->nops);
@@ -251,8 +252,11 @@ static int
 nfs4_xdr_enc_cb_recall(struct rpc_rqst *req, __be32 *p, struct nfs4_cb_recall *args)
 {
 	struct xdr_stream xdr;
+	struct nfs4_callback *cb =
+		(struct nfs4_callback *)req->rq_task->tk_client->cl_private;
 	struct nfs4_cb_compound_hdr hdr = {
 		.ident = args->cbr_ident,
+		.minorversion = cb->cb_minorversion,
 	};
 
 	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
@@ -436,6 +440,7 @@ static int do_probe_callback(void *data)
 		goto out_release_client;
 
 	cb->cb_client = client;
+	client->cl_private = cb;
 	atomic_set(&cb->cb_set, 1);
 	put_nfs4_client(clp);
 	return 0;
fs/nfsd/nfs4state.c +1 -0
index 8a64981..189f1c2 100644
@@ -994,6 +994,7 @@ gen_callback(struct nfs4_client *clp, struct nfsd4_setclientid *se)
 	if ( !(parse_ipv4(se->se_callback_addr_len, se->se_callback_addr_val,
 	                 &cb->cb_addr, &cb->cb_port)))
 		goto out_err;
+	cb->cb_minorversion = 0;
 	cb->cb_prog = se->se_callback_prog;
 	cb->cb_ident = se->se_callback_ident;
 	return;
include/linux/nfsd/state.h +2 -1
index 302557d..b78021a 100644
@@ -93,7 +93,8 @@ struct nfs4_callback {
 	u32                     cb_addr;
 	unsigned short          cb_port;
 	u32                     cb_prog;
-	u32                     cb_ident;
+	u32			cb_minorversion;
+	u32                     cb_ident;	/* minorversion 0 only */
 	/* RPC client info */
 	atomic_t		cb_set;     /* successful CB_NULL call */
 	struct rpc_clnt *       cb_client;
Signature · 1 line
1.6.2.1
#173 lore raw

PATCH v2 36/47 nfsd41: sunrpc: Added rpc server-side backchannel handling

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

FIXME: bhalevy: write up commit message

When the call direction is a reply, copy the xid and call direction into the req->rq_private_buf.head[0].iov_base otherwise rpc_verify_header returns rpc_garbage.

---
 include/linux/sunrpc/clnt.h    |    1 +
 include/linux/sunrpc/svcsock.h |    1 +
 include/linux/sunrpc/xprt.h    |    4 +
 net/sunrpc/clnt.c              |    1 +
 net/sunrpc/svcsock.c           |   81 +++++++++++
 net/sunrpc/xprt.c              |   41 ++++++-
 net/sunrpc/xprtsock.c          |  286 +++++++++++++++++++++++++++++++++++++++-
 7 files changed, 405 insertions(+), 10 deletions(-)
include/linux/sunrpc/clnt.h +1 -0
index 928c547..0b32b00 100644
@@ -111,6 +111,7 @@ struct rpc_create_args {
 	rpc_authflavor_t	authflavor;
 	unsigned long		flags;
 	char			*client_name;
+	struct svc_sock		*bc_sock;	/* NFSv4.1 backchannel */
 };
 
 /* Values for "flags" field */
include/linux/sunrpc/svcsock.h +1 -0
index 483e103..e8fda10 100644
@@ -28,6 +28,7 @@ struct svc_sock {
 	/* private TCP part */
 	u32			sk_reclen;	/* length of record */
 	u32			sk_tcplen;	/* current read length */
+	struct rpc_xprt	       *sk_bc_xprt;	/* NFSv4.1 backchannel xprt */
 };
 
 /*
include/linux/sunrpc/xprt.h +4 -0
index 11fc71d..e9ffee3 100644
@@ -174,6 +174,9 @@ struct rpc_xprt {
 	spinlock_t		reserve_lock;	/* lock slot table */
 	u32			xid;		/* Next XID value to use */
 	struct rpc_task *	snd_task;	/* Task blocked in send */
+#if defined(CONFIG_NFSD_V4_1)
+	struct svc_sock		*bc_sock;	/* NFSv4.1 backchannel */
+#endif /* CONFIG_NFSD_V4_1 */
 	struct list_head	recv;
 
 	struct {
@@ -197,6 +200,7 @@ struct xprt_create {
 	struct sockaddr *	srcaddr;	/* optional local address */
 	struct sockaddr *	dstaddr;	/* remote peer address */
 	size_t			addrlen;
+	struct svc_sock		*bc_sock;	/* NFSv4.1 backchannel */
 };
 
 struct xprt_class {
net/sunrpc/clnt.c +1 -0
index 836f15c..0a8f2d0 100644
@@ -266,6 +266,7 @@ struct rpc_clnt *rpc_create(struct rpc_create_args *args)
 		.srcaddr = args->saddress,
 		.dstaddr = args->address,
 		.addrlen = args->addrsize,
+		.bc_sock = args->bc_sock,
 	};
 	char servername[48];
 
net/sunrpc/svcsock.c +81 -0

117 lines not on this page · raw message

net/sunrpc/xprt.c +38 -3

59 lines not on this page · raw message

net/sunrpc/xprtsock.c +279 -7

323 lines not on this page · raw message

Signature · 1 line
1.6.2.1
#175 lore raw

PATCH v2 37/47 nfsd41: callback infrastructure

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

Keep the xprt used for create_session in cl_cb_xprt. Mark cl_callback.cb_minorversion = 1 and remember the client provided cl_callback.cb_prog rpc program number. Use it to probe the callback path.

Define xdr sizes and code nfs4_cb_compound header to be able to send a null callback rpc.

---
 fs/nfsd/nfs4callback.c |   35 ++++++++++++++++++++++++++++++++---
 fs/nfsd/nfs4state.c    |   10 ++++++++++
 2 files changed, 42 insertions(+), 3 deletions(-)
fs/nfsd/nfs4callback.c +32 -3
index d8f2631..6b7ef87 100644
@@ -43,6 +43,7 @@
 #include <linux/sunrpc/xdr.h>
 #include <linux/sunrpc/svc.h>
 #include <linux/sunrpc/clnt.h>
+#include <linux/sunrpc/svcsock.h>
 #include <linux/nfsd/nfsd.h>
 #include <linux/nfsd/state.h>
 #include <linux/sunrpc/sched.h>
@@ -52,16 +53,19 @@
 
 #define NFSPROC4_CB_NULL 0
 #define NFSPROC4_CB_COMPOUND 1
+#define NFS4_STATEID_SIZE 16
 
 /* Index of predefined Linux callback client operations */
 
 enum {
-        NFSPROC4_CLNT_CB_NULL = 0,
+	NFSPROC4_CLNT_CB_NULL = 0,
 	NFSPROC4_CLNT_CB_RECALL,
+	NFSPROC4_CLNT_CB_SEQUENCE,
 };
 
 enum nfs_cb_opnum4 {
 	OP_CB_RECALL            = 4,
+	OP_CB_SEQUENCE          = 11,
 };
 
 #define NFS4_MAXTAGLEN		20
@@ -70,15 +74,22 @@ enum nfs_cb_opnum4 {
 #define NFS4_dec_cb_null_sz		0
 #define cb_compound_enc_hdr_sz		4
 #define cb_compound_dec_hdr_sz		(3 + (NFS4_MAXTAGLEN >> 2))
+#define sessionid_sz			(NFS4_MAX_SESSIONID_LEN >> 2)
+#define cb_sequence_enc_sz		(sessionid_sz + 4 +             \
+					1 /* no referring calls list yet */)
+#define cb_sequence_dec_sz		(op_dec_sz + sessionid_sz + 4)
+
 #define op_enc_sz			1
 #define op_dec_sz			2
 #define enc_nfs4_fh_sz			(1 + (NFS4_FHSIZE >> 2))
 #define enc_stateid_sz			(NFS4_STATEID_SIZE >> 2)
 #define NFS4_enc_cb_recall_sz		(cb_compound_enc_hdr_sz +       \
+					cb_sequence_enc_sz +            \
 					1 + enc_stateid_sz +            \
 					enc_nfs4_fh_sz)
 
 #define NFS4_dec_cb_recall_sz		(cb_compound_dec_hdr_sz  +      \
+					cb_sequence_dec_sz +            \
 					op_dec_sz)
 
 /*
@@ -135,13 +146,19 @@ xdr_error:                                      \
 		return -EIO; \
 	} \
 } while (0)
+#define COPYMEM(x, nbytes) do {                \
+	memcpy((x), p, nbytes);                \
+	p += XDR_QUADLEN(nbytes);              \
+} while (0)
 
 struct nfs4_cb_compound_hdr {
-	int		status;
-	u32		ident;
+	/* args */
+	u32		ident;	/* minorversion 0 only */
 	u32		nops;
 	__be32		*nops_p;
 	u32		minorversion;
+	/* res */
+	int		status;
 	u32		taglen;
 	char		*tag;
 };
@@ -405,6 +422,17 @@ static struct rpc_clnt *setup_callback_client(struct nfs4_client *clp)
 	addr.sin_family = AF_INET;
 	addr.sin_port = htons(cb->cb_port);
 	addr.sin_addr.s_addr = htonl(cb->cb_addr);
+#if defined(CONFIG_NFSD_V4_1)
+	if (cb->cb_minorversion) {
+		BUG_ON(cb->cb_minorversion != 1);
+		args.bc_sock = container_of(clp->cl_cb_xprt, struct svc_sock,
+					    sk_xprt);
+	}
+#endif /* CONFIG_NFSD_V4_1 */
+
+	dprintk("%s: program %s 0x%x nrvers %u version %u minorversion %u\n",
+		__func__, args.program->name, args.prognumber,
+		args.program->nrvers, args.version, cb->cb_minorversion);
 
 	/* Create RPC client */
 	client = rpc_create(&args);
@@ -445,6 +473,7 @@ static int do_probe_callback(void *data)
 	put_nfs4_client(clp);
 	return 0;
 out_release_client:
+	dprintk("NFSD: synchronous CB_NULL failed. status=%d\n", status);
 	rpc_shutdown_client(client);
 out_err:
 	dprintk("NFSD: warning: no callback path to client %.*s: error %d\n",
fs/nfsd/nfs4state.c +10 -0
index 189f1c2..ef1e372 100644
@@ -663,6 +663,8 @@ free_client(struct nfs4_client *clp)
 {
 	shutdown_callback_client(clp);
 #if defined(CONFIG_NFSD_V4_1)
+	if (clp->cl_cb_xprt)
+		svc_xprt_put(clp->cl_cb_xprt);
 	nfsd4_release_respages(clp->cl_slot.sl_cache_entry.ce_respages,
 			     clp->cl_slot.sl_cache_entry.ce_resused);
 #endif /* CONFIG_NFSD_V4_1 */
@@ -1411,6 +1413,14 @@ nfsd4_create_session(struct svc_rqst *rqstp,
 		cr_ses->flags &= ~SESSION4_PERSIST;
 		cr_ses->flags &= ~SESSION4_RDMA;
 
+		if (cr_ses->flags & SESSION4_BACK_CHAN) {
+			unconf->cl_cb_xprt = rqstp->rq_xprt;
+			svc_xprt_get(unconf->cl_cb_xprt);
+			unconf->cl_callback.cb_minorversion = 1;
+			unconf->cl_cb_seq_nr = 0;
+			unconf->cl_callback.cb_prog = cr_ses->callback_prog;
+			nfsd4_probe_callback(unconf);
+		}
 		conf = unconf;
 	} else {
 		status = nfserr_stale_clientid;
Signature · 1 line
1.6.2.1
#176 lore raw

PATCH v2 38/47 nfsd41: Remember the auth flavor to use for callbacks

In reply to Benny Halevy

From: Ricardo Labiaga <[email protected]>

The callbacks will be sent using the same authentication flavor that was used during session creation. We'll add code to remember the principal in the case RPCSEC_GSS in a separate patch.

---
 fs/nfsd/nfs4state.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)
fs/nfsd/nfs4state.c +1 -0
index ef1e372..eb3d0dc 100644
@@ -1307,6 +1307,7 @@ out_new:
 	copy_verf(new, &verf);
 	copy_cred(&new->cl_cred, &rqstp->rq_cred);
 	new->cl_addr = ip_addr;
+	new->cl_flavor = rqstp->rq_flavor;
 	gen_clid(new);
 	gen_confirm(new);
 	add_to_unconfirmed(new, strhashval);
Signature · 1 line
1.6.2.1
#177 lore raw

PATCH v2 39/47 nfsd41: introduce cl_cb_mutex

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

To be used for serializing nfs41 callbacks.

---
 fs/nfsd/nfs4state.c        |    1 +
 include/linux/nfsd/state.h |    1 +
 2 files changed, 2 insertions(+), 0 deletions(-)
fs/nfsd/nfs4state.c +1 -0
index eb3d0dc..e5c7ddb 100644
@@ -741,6 +741,7 @@ static struct nfs4_client *create_client(struct xdr_netobj name, char *recdir)
 	INIT_LIST_HEAD(&clp->cl_delegations);
 #if defined(CONFIG_NFSD_V4_1)
 	INIT_LIST_HEAD(&clp->cl_sessions);
+	mutex_init(&clp->cl_cb_mutex);
 #endif /* CONFIG_NFSD_V4_1 */
 	INIT_LIST_HEAD(&clp->cl_lru);
 	return clp;
include/linux/nfsd/state.h +1 -0
index b78021a..f2783d4 100644
@@ -197,6 +197,7 @@ struct nfs4_client {
 	struct nfs4_sessionid	cl_sessionid;
 
 	struct svc_xprt		*cl_cb_xprt;	/* 4.1 callback transport */
+	struct mutex		cl_cb_mutex;
 #endif /* CONFIG_NFSD_V4_1 */
 };
 
Signature · 1 line
1.6.2.1
#178 lore raw

PATCH v2 40/47 nfsd41: cb_sequence callback

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

Implement the cb_sequence callback conforming to draft-ietf-nfsv4-minorversion1

[Rework the back channel xdr using the shared v4.0 and v4.1 framework.]
[fixed indentation]
---
 fs/nfsd/nfs4callback.c     |  118 ++++++++++++++++++++++++++++++++++++++++++++
 include/linux/nfsd/state.h |    7 ++-
 2 files changed, 124 insertions(+), 1 deletions(-)
fs/nfsd/nfs4callback.c +118 -0
index 6b7ef87..7ada6b1 100644
@@ -255,6 +255,29 @@ encode_cb_recall(struct xdr_stream *xdr, struct nfs4_cb_recall *cb_rec,
 	hdr->nops++;
 }
 
+static void
+encode_cb_sequence(struct xdr_stream *xdr, struct nfsd4_cb_sequence *args,
+		   struct nfs4_cb_compound_hdr *hdr)
+{
+	__be32 *p;
+
+	if (hdr->minorversion == 0)
+		return;
+
+	RESERVE_SPACE(1 + NFS4_MAX_SESSIONID_LEN + 20);
+
+	WRITE32(OP_CB_SEQUENCE);
+#ifdef CONFIG_NFSD_V4_1
+	WRITEMEM(args->cbs_clp->cl_sessionid.data, NFS4_MAX_SESSIONID_LEN);
+	WRITE32(args->cbs_clp->cl_cb_seq_nr);
+#endif /* CONFIG_NFSD_V4_1 */
+	WRITE32(0);		/* slotid, always 0 */
+	WRITE32(0);		/* highest slotid always 0 */
+	WRITE32(0);		/* cachethis always 0 */
+	WRITE32(0); /* FIXME: support referring_call_lists */
+	hdr->nops++;
+}
+
 static int
 nfs4_xdr_enc_cb_null(struct rpc_rqst *req, __be32 *p)
 {
@@ -319,6 +342,69 @@ decode_cb_op_hdr(struct xdr_stream *xdr, enum nfs_opnum4 expected)
 	return 0;
 }
 
+/*
+ * Our current back channel implmentation supports a single backchannel
+ * with a single slot.
+ */
+static int
+decode_cb_sequence(struct xdr_stream *xdr, struct nfsd4_cb_sequence *res,
+		   struct rpc_rqst *rqstp)
+{
+	struct nfs4_callback *cb =
+		(struct nfs4_callback *)rqstp->rq_task->tk_client->cl_private;
+	struct nfs4_sessionid id;
+	int status;
+	u32 dummy;
+	__be32 *p;
+
+	if (cb->cb_minorversion == 0)
+		return 0;
+
+	status = decode_cb_op_hdr(xdr, OP_CB_SEQUENCE);
+	if (status)
+		return status;
+
+	/*
+	 * If the server returns different values for sessionID, slotID or
+	 * sequence number, the server is looney tunes.
+	 */
+	status = -ESERVERFAULT;
+
+	READ_BUF(NFS4_MAX_SESSIONID_LEN + 16);
+	COPYMEM(id.data, NFS4_MAX_SESSIONID_LEN);
+#ifdef CONFIG_NFSD_V4_1
+	if (memcmp(id.data, res->cbs_clp->cl_sessionid.data,
+		   NFS4_MAX_SESSIONID_LEN)) {
+		dprintk("%s Invalid session id\n", __func__);
+		goto out;
+	}
+	READ32(dummy);
+	if (dummy != res->cbs_clp->cl_cb_seq_nr) {
+		dprintk("%s Invalid sequence number\n", __func__);
+		goto out;
+	}
+#endif /* CONFIG_NFSD_V4_1 */
+	READ32(dummy); 	/* slotid must be 0 */
+	if (dummy != 0) {
+		dprintk("%s Invalid slotid\n", __func__);
+		goto out;
+	}
+	READ32(dummy); 	/* highest slotid must be 0 */
+	if (dummy != 0) {
+		dprintk("%s Invalid highest slotid\n", __func__);
+		goto out;
+	}
+	READ32(dummy); 	/* target highest slotid must be 0 */
+	if (dummy != 0) {
+		dprintk("%s Invalid target highest slotid\n", __func__);
+		goto out;
+	}
+	status = 0;
+out:
+	return status;
+}
+
+
 static int
 nfs4_xdr_dec_cb_null(struct rpc_rqst *req, __be32 *p)
 {
@@ -503,6 +589,38 @@ nfsd4_probe_callback(struct nfs4_client *clp)
 	return;
 }
 
+#if defined(CONFIG_NFSD_V4_1)
+/*
+ * FIXME: cb_sequence should support referring call lists, cachethis, and
+ * multiple slots
+ */
+static int
+nfs41_cb_sequence_setup(struct nfs4_client *clp, struct nfsd4_cb_sequence *args)
+{
+	u32 *ptr = (u32 *)clp->cl_sessionid.data;
+
+	dprintk("%s: %u:%u:%u:%u\n", __func__,
+		ptr[0], ptr[1], ptr[2], ptr[3]);
+
+	mutex_lock(&clp->cl_cb_mutex);
+	args->cbs_clp = clp;
+	clp->cl_cb_seq_nr++;
+	return 0;
+}
+
+static void
+nfs41_cb_sequence_done(struct nfs4_client *clp, struct nfsd4_cb_sequence *res)
+{
+	u32 *ptr = (u32 *)clp->cl_sessionid.data;
+
+	dprintk("%s: %u:%u:%u:%u\n", __func__,
+		ptr[0], ptr[1], ptr[2], ptr[3]);
+
+	/* FIXME: support multiple callback slots */
+	mutex_unlock(&clp->cl_cb_mutex);
+}
+#endif /* CONFIG_NFSD_V4_1 */
+
 /*
  * called with dp->dl_count inc'ed.
  */
include/linux/nfsd/state.h +6 -1
index f2783d4..0ea8c89 100644
@@ -61,6 +61,10 @@ typedef struct {
 #define si_stateownerid   si_opaque.so_stateownerid
 #define si_fileid         si_opaque.so_fileid
 
+struct nfsd4_cb_sequence {
+	/* args/res */
+	struct nfs4_client	*cbs_clp;
+};
 
 struct nfs4_cb_recall {
 	u32			cbr_ident;
@@ -195,7 +199,8 @@ struct nfs4_client {
 	struct nfsd4_slot	cl_slot;	/* create_session slot */
 	u32			cl_exchange_flags;
 	struct nfs4_sessionid	cl_sessionid;
-
+	/* We currently support a single back channel with a single slot */
+	u32			cl_cb_seq_nr;
 	struct svc_xprt		*cl_cb_xprt;	/* 4.1 callback transport */
 	struct mutex		cl_cb_mutex;
 #endif /* CONFIG_NFSD_V4_1 */
Signature · 1 line
1.6.2.1
#193 lore raw

PATCH v2 41/47 nfsd41: introduce nfs4_cb_call_sync for nfs4 and nfs41

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

For nfs41 callbacks we need to setup the cb_sequence args and process the result. This will be called by code common to nfs4 and nfs41.

---
 fs/nfsd/nfs4callback.c |   50 ++++++++++++++++++++++++++++++++++++++++++++++++
 1 files changed, 50 insertions(+), 0 deletions(-)
fs/nfsd/nfs4callback.c +50 -0
index 7ada6b1..dd3b739 100644
@@ -92,6 +92,15 @@ enum nfs_cb_opnum4 {
 					cb_sequence_dec_sz +            \
 					op_dec_sz)
 
+struct nfs4_rpc_args {
+	void                     *args_op;
+	struct nfsd4_cb_sequence *args_seq;
+};
+
+struct nfs4_rpc_res {
+	struct nfsd4_cb_sequence *res_seq;
+};
+
 /*
 * Generic encode routines from fs/nfs/nfs4xdr.c
 */
@@ -589,6 +598,12 @@ nfsd4_probe_callback(struct nfs4_client *clp)
 	return;
 }
 
+static int _nfsd4_cb_sync(struct nfs4_client *clp,
+			  const struct rpc_message *msg, int flags)
+{
+	return rpc_call_sync(clp->cl_callback.cb_client, msg, RPC_TASK_SOFT);
+}
+
 #if defined(CONFIG_NFSD_V4_1)
 /*
  * FIXME: cb_sequence should support referring call lists, cachethis, and
@@ -619,6 +634,41 @@ nfs41_cb_sequence_done(struct nfs4_client *clp, struct nfsd4_cb_sequence *res)
 	/* FIXME: support multiple callback slots */
 	mutex_unlock(&clp->cl_cb_mutex);
 }
+
+static int _nfsd41_cb_sync(struct nfs4_client *clp,
+			   struct rpc_message *msg, int flags)
+{
+	struct nfsd4_cb_sequence seq;
+	struct nfs4_rpc_args *args;
+	struct nfs4_rpc_res res;
+	int status;
+
+	args = msg->rpc_argp;
+	args->args_seq = &seq;
+
+	res.res_seq = &seq;
+	msg->rpc_resp = &res;
+
+	nfs41_cb_sequence_setup(clp, &seq);
+	status = _nfsd4_cb_sync(clp, msg, flags);
+	nfs41_cb_sequence_done(clp, &seq);
+
+	return status;
+}
+
+static int nfsd4_cb_sync(struct nfs4_client *clp,
+			 struct rpc_message *msg, int flags)
+{
+	return clp->cl_callback.cb_minorversion ?
+		_nfsd41_cb_sync(clp, msg, flags) :
+		_nfsd4_cb_sync(clp, msg, flags);
+}
+#else  /* CONFIG_NFSD_V4_1 */
+static int nfsd4_cb_sync(struct nfs4_client *clp,
+			 struct rpc_message *msg, int flags)
+{
+	return _nfsd4_cb_sync(clp, msg, flags);
+}
 #endif /* CONFIG_NFSD_V4_1 */
 
 /*
Signature · 1 line
1.6.2.1
#194 lore raw

PATCH v2 42/47 nfsd41: cb_recall callback

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

Implement the cb_recall callback conforming to
http://tools.ietf.org/html/draft-ietf-nfsv4-minorversion1-26

[Share v4.0 and v4.1 back channel xdr]
---
 fs/nfsd/nfs4callback.c |   25 +++++++++++++++++++------
 1 files changed, 19 insertions(+), 6 deletions(-)
fs/nfsd/nfs4callback.c +19 -6
index dd3b739..0be75e0 100644
@@ -298,9 +298,11 @@ nfs4_xdr_enc_cb_null(struct rpc_rqst *req, __be32 *p)
 }
 
 static int
-nfs4_xdr_enc_cb_recall(struct rpc_rqst *req, __be32 *p, struct nfs4_cb_recall *args)
+nfs4_xdr_enc_cb_recall(struct rpc_rqst *req, __be32 *p,
+		       struct nfs4_rpc_args *rpc_args)
 {
 	struct xdr_stream xdr;
+	struct nfs4_cb_recall *args = rpc_args->args_op;
 	struct nfs4_callback *cb =
 		(struct nfs4_callback *)req->rq_task->tk_client->cl_private;
 	struct nfs4_cb_compound_hdr hdr = {
@@ -310,6 +312,7 @@ nfs4_xdr_enc_cb_recall(struct rpc_rqst *req, __be32 *p, struct nfs4_cb_recall *a
 
 	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
 	encode_cb_compound_hdr(&xdr, &hdr);
+	encode_cb_sequence(&xdr, rpc_args->args_seq, &hdr);
 	encode_cb_recall(&xdr, args, &hdr);
 	encode_cb_nops(&hdr);
 	return 0;
@@ -421,7 +424,8 @@ nfs4_xdr_dec_cb_null(struct rpc_rqst *req, __be32 *p)
 }
 
 static int
-nfs4_xdr_dec_cb_recall(struct rpc_rqst *rqstp, __be32 *p)
+nfs4_xdr_dec_cb_recall(struct rpc_rqst *rqstp, __be32 *p,
+		       struct nfs4_rpc_res *rpc_res)
 {
 	struct xdr_stream xdr;
 	struct nfs4_cb_compound_hdr hdr;
@@ -431,6 +435,9 @@ nfs4_xdr_dec_cb_recall(struct rpc_rqst *rqstp, __be32 *p)
 	status = decode_cb_compound_hdr(&xdr, &hdr);
 	if (status)
 		goto out;
+	status = decode_cb_sequence(&xdr, rpc_res->res_seq, rqstp);
+	if (status)
+		goto out;
 	status = decode_cb_op_hdr(&xdr, OP_CB_RECALL);
 out:
 	return status;
@@ -678,19 +685,23 @@ void
 nfsd4_cb_recall(struct nfs4_delegation *dp)
 {
 	struct nfs4_client *clp = dp->dl_client;
-	struct rpc_clnt *clnt = clp->cl_callback.cb_client;
 	struct nfs4_cb_recall *cbr = &dp->dl_recall;
+	struct nfs4_rpc_args args = {
+		.args_op = cbr,
+	};
 	struct rpc_message msg = {
 		.rpc_proc = &nfs4_cb_procedures[NFSPROC4_CLNT_CB_RECALL],
-		.rpc_argp = cbr,
+		.rpc_argp = &args,
 	};
 	int retries = 1;
 	int status = 0;
 
+	dprintk("%s: dp %p\n", __func__, dp);
+
 	cbr->cbr_trunc = 0; /* XXX need to implement truncate optimization */
 	cbr->cbr_dp = dp;
 
-	status = rpc_call_sync(clnt, &msg, RPC_TASK_SOFT);
+	status = nfsd4_cb_sync(clp, &msg, RPC_TASK_SOFT);
 	while (retries--) {
 		switch (status) {
 			case -EIO:
@@ -705,13 +716,15 @@ nfsd4_cb_recall(struct nfs4_delegation *dp)
 				goto out_put_cred;
 		}
 		ssleep(2);
-		status = rpc_call_sync(clnt, &msg, RPC_TASK_SOFT);
+		status = nfsd4_cb_sync(clp, &msg, RPC_TASK_SOFT);
 	}
 out_put_cred:
 	/*
 	 * Success or failure, now we're either waiting for lease expiration
 	 * or deleg_return.
 	 */
+	dprintk("%s: dp %p dl_flock %p dl_count %d\n",
+		__func__, dp, dp->dl_flock, atomic_read(&dp->dl_count));
 	put_nfs4_client(clp);
 	nfs4_put_delegation(dp);
 	return;
Signature · 1 line
1.6.2.1
#195 lore raw

PATCH v2 43/47 nfsd41: pass writable attrs mask to nfsd4_decode_fattr

In reply to Benny Halevy

In preparation for EXCLUSIVE4_1

---
 fs/nfsd/nfs4xdr.c |   21 +++++++++++++++------
 1 files changed, 15 insertions(+), 6 deletions(-)
fs/nfsd/nfs4xdr.c +15 -6
index a1af3a3..96f3eeb 100644
@@ -251,9 +251,14 @@ nfsd4_decode_bitmap(struct nfsd4_compoundargs *argp, u32 *bmval)
 	DECODE_TAIL;
 }
 
+static u32 nfsd_attrmask[] = {
+	NFSD_WRITEABLE_ATTRS_WORD0,
+	NFSD_WRITEABLE_ATTRS_WORD1
+};
+
 static __be32
-nfsd4_decode_fattr(struct nfsd4_compoundargs *argp, u32 *bmval, struct iattr *iattr,
-    struct nfs4_acl **acl)
+nfsd4_decode_fattr(struct nfsd4_compoundargs *argp, u32 *bmval, u32 *writable,
+		   struct iattr *iattr, struct nfs4_acl **acl)
 {
 	int expected_len, len = 0;
 	u32 dummy32;
@@ -271,7 +276,7 @@ nfsd4_decode_fattr(struct nfsd4_compoundargs *argp, u32 *bmval, struct iattr *ia
 	 */
 	if ((bmval[0] & ~NFSD_SUPPORTED_ATTRS_WORD0) || (bmval[1] & ~NFSD_SUPPORTED_ATTRS_WORD1))
 		return nfserr_attrnotsupp;
-	if ((bmval[0] & ~NFSD_WRITEABLE_ATTRS_WORD0) || (bmval[1] & ~NFSD_WRITEABLE_ATTRS_WORD1))
+	if ((bmval[0] & ~writable[0]) || (bmval[1] & ~writable[1]))
 		return nfserr_inval;
 
 	READ_BUF(4);
@@ -499,7 +504,9 @@ nfsd4_decode_create(struct nfsd4_compoundargs *argp, struct nfsd4_create *create
 	if ((status = check_filename(create->cr_name, create->cr_namelen, nfserr_inval)))
 		return status;
 
-	if ((status = nfsd4_decode_fattr(argp, create->cr_bmval, &create->cr_iattr, &create->cr_acl)))
+	status = nfsd4_decode_fattr(argp, create->cr_bmval, nfsd_attrmask,
+				    &create->cr_iattr, &create->cr_acl);
+	if (status)
 		goto out;
 
 	DECODE_TAIL;
@@ -660,7 +667,9 @@ nfsd4_decode_open(struct nfsd4_compoundargs *argp, struct nfsd4_open *open)
 		switch (open->op_createmode) {
 		case NFS4_CREATE_UNCHECKED:
 		case NFS4_CREATE_GUARDED:
-			if ((status = nfsd4_decode_fattr(argp, open->op_bmval, &open->op_iattr, &open->op_acl)))
+			status = nfsd4_decode_fattr(argp, open->op_bmval,
+				nfsd_attrmask, &open->op_iattr, &open->op_acl);
+			if (status)
 				goto out;
 			break;
 		case NFS4_CREATE_EXCLUSIVE:
@@ -859,7 +868,7 @@ nfsd4_decode_setattr(struct nfsd4_compoundargs *argp, struct nfsd4_setattr *seta
 	status = nfsd4_decode_stateid(argp, &setattr->sa_stateid);
 	if (status)
 		return status;
-	return nfsd4_decode_fattr(argp, setattr->sa_bmval,
+	return nfsd4_decode_fattr(argp, setattr->sa_bmval, nfsd_attrmask,
 				  &setattr->sa_iattr, &setattr->sa_acl);
 }
 
Signature · 1 line
1.6.2.1
#196 lore raw

PATCH v2 44/47 nfsd41: support for 3-word long attribute bitmask

In reply to Benny Halevy

From: Andy Adamson <[email protected]>

Also, use client minorversion to generate supported attrs

---
 fs/nfsd/nfs4proc.c        |   15 +++++++----
 fs/nfsd/nfs4xdr.c         |   59 +++++++++++++++++++++++++++++++++++---------
 include/linux/nfsd/nfsd.h |   48 +++++++++++++++++++++++++++++++++++-
 include/linux/nfsd/xdr4.h |   12 ++++----
 4 files changed, 108 insertions(+), 26 deletions(-)
fs/nfsd/nfs4proc.c +9 -6
index 8b4bf5e..f3b560f 100644
@@ -460,8 +460,9 @@ nfsd4_getattr(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 	if (getattr->ga_bmval[1] & NFSD_WRITEONLY_ATTRS_WORD1)
 		return nfserr_inval;
 
-	getattr->ga_bmval[0] &= NFSD_SUPPORTED_ATTRS_WORD0;
-	getattr->ga_bmval[1] &= NFSD_SUPPORTED_ATTRS_WORD1;
+	getattr->ga_bmval[0] &= nfsd_suppattrs0(cstate->minorversion);
+	getattr->ga_bmval[1] &= nfsd_suppattrs1(cstate->minorversion);
+	getattr->ga_bmval[2] &= nfsd_suppattrs2(cstate->minorversion);
 
 	getattr->ga_fhp = &cstate->current_fh;
 	return nfs_ok;
@@ -556,8 +557,9 @@ nfsd4_readdir(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 	if (readdir->rd_bmval[1] & NFSD_WRITEONLY_ATTRS_WORD1)
 		return nfserr_inval;
 
-	readdir->rd_bmval[0] &= NFSD_SUPPORTED_ATTRS_WORD0;
-	readdir->rd_bmval[1] &= NFSD_SUPPORTED_ATTRS_WORD1;
+	readdir->rd_bmval[0] &= nfsd_suppattrs0(cstate->minorversion);
+	readdir->rd_bmval[1] &= nfsd_suppattrs1(cstate->minorversion);
+	readdir->rd_bmval[2] &= nfsd_suppattrs2(cstate->minorversion);
 
 	if ((cookie > ~(u32)0) || (cookie == 1) || (cookie == 2) ||
 	    (cookie == 0 && memcmp(readdir->rd_verf.data, zeroverf.data, NFS4_VERIFIER_SIZE)))
@@ -754,8 +756,9 @@ _nfsd4_verify(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 	if (status)
 		return status;
 
-	if ((verify->ve_bmval[0] & ~NFSD_SUPPORTED_ATTRS_WORD0)
-	    || (verify->ve_bmval[1] & ~NFSD_SUPPORTED_ATTRS_WORD1))
+	if ((verify->ve_bmval[0] & ~nfsd_suppattrs0(cstate->minorversion))
+	    || (verify->ve_bmval[1] & ~nfsd_suppattrs1(cstate->minorversion))
+	    || (verify->ve_bmval[2] & ~nfsd_suppattrs2(cstate->minorversion)))
 		return nfserr_attrnotsupp;
 	if ((verify->ve_bmval[0] & FATTR4_WORD0_RDATTR_ERROR)
 	    || (verify->ve_bmval[1] & NFSD_WRITEONLY_ATTRS_WORD1))
fs/nfsd/nfs4xdr.c +47 -12

127 lines not on this page · raw message

include/linux/nfsd/nfsd.h +46 -2

77 lines not on this page · raw message

include/linux/nfsd/xdr4.h +6 -6
index 881ff67..d7a8050 100644
@@ -101,7 +101,7 @@ struct nfsd4_create {
 			u32 specdata2;
 		} dev;    /* NF4BLK, NF4CHR */
 	} u;
-	u32		cr_bmval[2];        /* request */
+	u32		cr_bmval[3];        /* request */
 	struct iattr	cr_iattr;           /* request */
 	struct nfsd4_change_info  cr_cinfo; /* response */
 	struct nfs4_acl *cr_acl;
@@ -116,7 +116,7 @@ struct nfsd4_delegreturn {
 };
 
 struct nfsd4_getattr {
-	u32		ga_bmval[2];        /* request */
+	u32		ga_bmval[3];        /* request */
 	struct svc_fh	*ga_fhp;            /* response */
 };
 
@@ -217,7 +217,7 @@ struct nfsd4_open {
 	stateid_t       op_delegate_stateid; /* request - response */
 	u32		op_create;     	    /* request */
 	u32		op_createmode;      /* request */
-	u32		op_bmval[2];        /* request */
+	u32		op_bmval[3];        /* request */
 	union {                             /* request */
 		struct iattr	iattr;                      /* UNCHECKED4,GUARDED4 */
 		nfs4_verifier	verf;                                /* EXCLUSIVE4 */
@@ -270,7 +270,7 @@ struct nfsd4_readdir {
 	nfs4_verifier	rd_verf;            /* request */
 	u32		rd_dircount;        /* request */
 	u32		rd_maxcount;        /* request */
-	u32		rd_bmval[2];        /* request */
+	u32		rd_bmval[3];        /* request */
 	struct svc_rqst *rd_rqstp;          /* response */
 	struct svc_fh * rd_fhp;             /* response */
 
@@ -312,7 +312,7 @@ struct nfsd4_secinfo {
 
 struct nfsd4_setattr {
 	stateid_t	sa_stateid;         /* request */
-	u32		sa_bmval[2];        /* request */
+	u32		sa_bmval[3];        /* request */
 	struct iattr	sa_iattr;           /* request */
 	struct nfs4_acl *sa_acl;
 };
@@ -338,7 +338,7 @@ struct nfsd4_setclientid_confirm {
 
 /* also used for NVERIFY */
 struct nfsd4_verify {
-	u32		ve_bmval[2];        /* request */
+	u32		ve_bmval[3];        /* request */
 	u32		ve_attrlen;         /* request */
 	char *		ve_attrval;         /* request */
 };
Signature · 1 line
1.6.2.1
#197 lore raw

PATCH v2 45/47 nfsd41: SUPPATTR_EXCLCREAT attribute

In reply to Benny Halevy

Return bitmask for supported EXCLUSIVE4_1 create attributes.

---
 fs/nfsd/nfs4xdr.c         |    7 ++++++-
 include/linux/nfs4.h      |    2 ++
 include/linux/nfsd/nfsd.h |   14 +++++++++++++-
 3 files changed, 21 insertions(+), 2 deletions(-)
fs/nfsd/nfs4xdr.c +6 -1
index 079626c..288332c 100644
@@ -2154,7 +2154,12 @@ out_acl:
 		}
 		WRITE64(stat.ino);
 	}
-	BUG_ON(bmval2);	/* FIXME: not implemented yet */
+	if (bmval2 & FATTR4_WORD2_SUPPATTR_EXCLCREAT) {
+		WRITE32(3);
+		WRITE32(NFSD_SUPPATTR_EXCLCREAT_WORD0);
+		WRITE32(NFSD_SUPPATTR_EXCLCREAT_WORD1);
+		WRITE32(NFSD_SUPPATTR_EXCLCREAT_WORD2);
+	}
 
 	*attrlenp = htonl((char *)p - (char *)attrlenp - 4);
 	*countp = p - buffer;
include/linux/nfs4.h +2 -0
index a34683e..735c7d8 100644
@@ -404,6 +404,8 @@ enum lock_type4 {
 #define FATTR4_WORD0_UNIQUE_HANDLES     (1UL << 9)
 #define FATTR4_WORD0_LEASE_TIME         (1UL << 10)
 #define FATTR4_WORD0_RDATTR_ERROR       (1UL << 11)
+/* Mandatory in NFSv4.1 */
+#define FATTR4_WORD2_SUPPATTR_EXCLCREAT (1UL << 11)
 
 /* Recommended Attributes */
 #define FATTR4_WORD0_ACL                (1UL << 12)
include/linux/nfsd/nfsd.h +13 -1
index 3dccb76..b9e6682 100644
@@ -374,7 +374,7 @@ extern struct timeval	nfssvc_boot;
 	NFSD4_SUPPORTED_ATTRS_WORD1
 
 #define NFSD4_1_SUPPORTED_ATTRS_WORD2 \
-	NFSD4_SUPPORTED_ATTRS_WORD2
+	(NFSD4_SUPPORTED_ATTRS_WORD2 | FATTR4_WORD2_SUPPATTR_EXCLCREAT)
 
 #if defined(CONFIG_NFSD_V4_1)
 
@@ -427,6 +427,18 @@ static inline u32 nfsd_suppattrs2(u32 minorversion)
  | FATTR4_WORD1_TIME_ACCESS_SET | FATTR4_WORD1_TIME_MODIFY_SET)
 #define NFSD_WRITEABLE_ATTRS_WORD2 0
 
+#define NFSD_SUPPATTR_EXCLCREAT_WORD0 \
+	NFSD_WRITEABLE_ATTRS_WORD0
+/*
+ * we currently store the exclusive create verifier in the v_{a,m}time
+ * attributes so the client can't set these at create time using EXCLUSIVE4_1
+ */
+#define NFSD_SUPPATTR_EXCLCREAT_WORD1 \
+	(NFSD_WRITEABLE_ATTRS_WORD1 & \
+	 ~(FATTR4_WORD1_TIME_ACCESS_SET | FATTR4_WORD1_TIME_MODIFY_SET))
+#define NFSD_SUPPATTR_EXCLCREAT_WORD2 \
+	NFSD_WRITEABLE_ATTRS_WORD2
+
 #endif /* CONFIG_NFSD_V4 */
 
 #endif /* LINUX_NFSD_NFSD_H */
Signature · 1 line
1.6.2.1
#198 lore raw

PATCH v2 46/47 nfsd41: CREATE_EXCLUSIVE4_1

In reply to Benny Halevy

Implement the CREATE_EXCLUSIVE4_1 open mode conforming to
http://tools.ietf.org/html/draft-ietf-nfsv4-minorversion1-26

This mode allows the client to atomically create a file if it doesn't exist while setting some of its attributes.

It must be implemented if the server supports persistent reply cache and/or pnfs.

---
 fs/nfsd/nfs4proc.c        |   15 +++++++++++++++
 fs/nfsd/nfs4xdr.c         |   17 +++++++++++++++++
 include/linux/nfs4.h      |    8 +++++++-
 include/linux/nfsd/xdr4.h |   10 ++++------
 4 files changed, 43 insertions(+), 7 deletions(-)
fs/nfsd/nfs4proc.c +15 -0
index f3b560f..6d7d7c5 100644
@@ -93,6 +93,21 @@ do_open_lookup(struct svc_rqst *rqstp, struct svc_fh *current_fh, struct nfsd4_o
 	open->op_truncate = 0;
 
 	if (open->op_create) {
+		/* FIXME: check session persistence and pnfs flags.
+		 * The nfsv4.1 spec requires the following semantics:
+		 *
+		 * Persistent   | pNFS   | Server REQUIRED | Client Allowed
+		 * Reply Cache  | server |                 |
+		 * -------------+--------+-----------------+--------------------
+		 * no           | no     | EXCLUSIVE4_1    | EXCLUSIVE4_1
+		 *              |        |                 | (SHOULD)
+		 *              |        | and EXCLUSIVE4  | or EXCLUSIVE4
+		 *              |        |                 | (SHOULD NOT)
+		 * no           | yes    | EXCLUSIVE4_1    | EXCLUSIVE4_1
+		 * yes          | no     | GUARDED4        | GUARDED4
+		 * yes          | yes    | GUARDED4        | GUARDED4
+		 */
+
 		/*
 		 * Note: create modes (UNCHECKED,GUARDED...) are the same
 		 * in NFSv4 as in v3.
fs/nfsd/nfs4xdr.c +17 -0
index 288332c..0084459 100644
@@ -260,6 +260,12 @@ static u32 nfsd_attrmask[] = {
 	NFSD_WRITEABLE_ATTRS_WORD2
 };
 
+static u32 nfsd41_ex_attrmask[] = {
+	NFSD_SUPPATTR_EXCLCREAT_WORD0,
+	NFSD_SUPPATTR_EXCLCREAT_WORD1,
+	NFSD_SUPPATTR_EXCLCREAT_WORD2
+};
+
 static __be32
 nfsd4_decode_fattr(struct nfsd4_compoundargs *argp, u32 *bmval, u32 *writable,
 		   struct iattr *iattr, struct nfs4_acl **acl)
@@ -684,6 +690,17 @@ nfsd4_decode_open(struct nfsd4_compoundargs *argp, struct nfsd4_open *open)
 			READ_BUF(8);
 			COPYMEM(open->op_verf.data, 8);
 			break;
+		case NFS4_CREATE_EXCLUSIVE4_1:
+			if (argp->minorversion < 1)
+				goto xdr_error;
+			READ_BUF(8);
+			COPYMEM(open->op_verf.data, 8);
+			status = nfsd4_decode_fattr(argp, open->op_bmval,
+				nfsd41_ex_attrmask, &open->op_iattr,
+				&open->op_acl);
+			if (status)
+				goto out;
+			break;
 		default:
 			goto xdr_error;
 		}
include/linux/nfs4.h +7 -1
index 735c7d8..5756359 100644
@@ -368,7 +368,13 @@ enum opentype4 {
 enum createmode4 {
 	NFS4_CREATE_UNCHECKED = 0,
 	NFS4_CREATE_GUARDED = 1,
-	NFS4_CREATE_EXCLUSIVE = 2
+	NFS4_CREATE_EXCLUSIVE = 2,
+	/*
+	 * New to NFSv4.1. If session is persistent,
+	 * GUARDED4 MUST be used. Otherwise, use
+	 * EXCLUSIVE4_1 instead of EXCLUSIVE4.
+	 */
+	NFS4_CREATE_EXCLUSIVE4_1 = 3
 };
 
 enum limit_by4 {
include/linux/nfsd/xdr4.h +4 -6
index d7a8050..7ed9955 100644
@@ -218,10 +218,8 @@ struct nfsd4_open {
 	u32		op_create;     	    /* request */
 	u32		op_createmode;      /* request */
 	u32		op_bmval[3];        /* request */
-	union {                             /* request */
-		struct iattr	iattr;                      /* UNCHECKED4,GUARDED4 */
-		nfs4_verifier	verf;                                /* EXCLUSIVE4 */
-	} u;
+	struct iattr	iattr;              /* UNCHECKED4, GUARDED4, EXCLUSIVE4_1 */
+	nfs4_verifier	verf;               /* EXCLUSIVE4 */
 	clientid_t	op_clientid;        /* request */
 	struct xdr_netobj op_owner;           /* request */
 	u32		op_seqid;           /* request */
@@ -235,8 +233,8 @@ struct nfsd4_open {
 	struct nfs4_stateowner *op_stateowner; /* used during processing */
 	struct nfs4_acl *op_acl;
 };
-#define op_iattr	u.iattr
-#define op_verf		u.verf
+#define op_iattr	iattr
+#define op_verf		verf
 
 struct nfsd4_open_confirm {
 	stateid_t	oc_req_stateid		/* request */;
Signature · 1 line
1.6.2.1
#199 lore raw

PATCH v2 47/47 nfsd41: Documentation/filesystems/nfs41-server.txt

In reply to Benny Halevy

Initial nfs41 server write up describing the status of the linux server implementation.

---
 Documentation/filesystems/nfs41-server.txt |  148 ++++++++++++++++++++++++++++
 1 files changed, 148 insertions(+), 0 deletions(-)
 create mode 100644 Documentation/filesystems/nfs41-server.txt
Documentation/filesystems/nfs41-server.txt new file +148 -0
new file mode 100644
index 0000000..e62cd13
@@ -0,0 +1,148 @@
+NFSv4.1 Server Implementation
+
+The NFSv4 minorversion 1 (NFSv4.1) implementation in nfsd is based
+on the latest NFSv4.1 Internet Draft:
+http://tools.ietf.org/html/draft-ietf-nfsv4-minorversion1-29
+
+If you're a developer or just eager to test the code, set
+CONFIG_NFSD_V4_1 to 'y' in your kernel configuration file.
+This will allow the server to serve both NFSv4 and NFSv4.1
+clients concurrently.
+
+From the many new features in NFSv4.1 the current implementation
+focuses on the mandatory-to-implement NFSv4.1 Sessions, providing
+"exactly once" semantics and better control and throttling of the
+resources allocated for each client.
+
+Other NFSv4.1 features, Parallel NFS operations in particular,
+are still under development out of tree.
+See http://wiki.linux-nfs.org/wiki/index.php/PNFS_prototype_design
+for more information.
+
+The table below, taken from the NFSv4.1 document, lists
+the operations that are mandatory to implement (REQ), optional
+(OPT), and NFSv4.0 operations that are required not to implement (MNI)
+in minor version 1.  The first column indicates the operations that
+are not supported yet by the linux server implementation.
+
+The OPTIONAL features identified and their abbreviations are as follows:
+	pNFS	Parallel NFS
+	FDELG	File Delegations
+	DDELG	Directory Delegations
+
+The following abbreviations indicate the linux server implementation status.
+	I	Implemented NFSv4.1 operations.
+	NS	Not Supported.
+	NS*	unimplemented optional feature.
+	P	pNFS features implemented out of tree.
+	PNS	pNFS features that are not supported yet (out of tree).
+
+Operations
+
+   +----------------------+------------+--------------+----------------+
+   | Operation            | REQ, REC,  | Feature      | Definition     |
+   |                      | OPT, or    | (REQ, REC,   |                |
+   |                      | MNI        | or OPT)      |                |
+   +----------------------+------------+--------------+----------------+
+   | ACCESS               | REQ        |              | Section 18.1   |
+NS | BACKCHANNEL_CTL      | REQ        |              | Section 18.33  |
+NS | BIND_CONN_TO_SESSION | REQ        |              | Section 18.34  |
+   | CLOSE                | REQ        |              | Section 18.2   |
+   | COMMIT               | REQ        |              | Section 18.3   |
+   | CREATE               | REQ        |              | Section 18.4   |
+I  | CREATE_SESSION       | REQ        |              | Section 18.36  |
+NS*| DELEGPURGE           | OPT        | FDELG (REQ)  | Section 18.5   |
+   | DELEGRETURN          | OPT        | FDELG,       | Section 18.6   |
+   |                      |            | DDELG, pNFS  |                |
+   |                      |            | (REQ)        |                |
+NS | DESTROY_CLIENTID     | REQ        |              | Section 18.50  |
+I  | DESTROY_SESSION      | REQ        |              | Section 18.37  |
+I  | EXCHANGE_ID          | REQ        |              | Section 18.35  |
+NS | FREE_STATEID         | REQ        |              | Section 18.38  |
+   | GETATTR              | REQ        |              | Section 18.7   |
+P  | GETDEVICEINFO        | OPT        | pNFS (REQ)   | Section 18.40  |
+P  | GETDEVICELIST        | OPT        | pNFS (OPT)   | Section 18.41  |
+   | GETFH                | REQ        |              | Section 18.8   |
+NS*| GET_DIR_DELEGATION   | OPT        | DDELG (REQ)  | Section 18.39  |
+P  | LAYOUTCOMMIT         | OPT        | pNFS (REQ)   | Section 18.42  |
+P  | LAYOUTGET            | OPT        | pNFS (REQ)   | Section 18.43  |
+P  | LAYOUTRETURN         | OPT        | pNFS (REQ)   | Section 18.44  |
+   | LINK                 | OPT        |              | Section 18.9   |
+   | LOCK                 | REQ        |              | Section 18.10  |
+   | LOCKT                | REQ        |              | Section 18.11  |
+   | LOCKU                | REQ        |              | Section 18.12  |
+   | LOOKUP               | REQ        |              | Section 18.13  |
+   | LOOKUPP              | REQ        |              | Section 18.14  |
+   | NVERIFY              | REQ        |              | Section 18.15  |
+   | OPEN                 | REQ        |              | Section 18.16  |
+NS*| OPENATTR             | OPT        |              | Section 18.17  |
+   | OPEN_CONFIRM         | MNI        |              | N/A            |
+   | OPEN_DOWNGRADE       | REQ        |              | Section 18.18  |
+   | PUTFH                | REQ        |              | Section 18.19  |
+   | PUTPUBFH             | REQ        |              | Section 18.20  |
+   | PUTROOTFH            | REQ        |              | Section 18.21  |
+   | READ                 | REQ        |              | Section 18.22  |
+   | READDIR              | REQ        |              | Section 18.23  |
+   | READLINK             | OPT        |              | Section 18.24  |
+NS | RECLAIM_COMPLETE     | REQ        |              | Section 18.51  |
+   | RELEASE_LOCKOWNER    | MNI        |              | N/A            |
+   | REMOVE               | REQ        |              | Section 18.25  |
+   | RENAME               | REQ        |              | Section 18.26  |
+   | RENEW                | MNI        |              | N/A            |
+   | RESTOREFH            | REQ        |              | Section 18.27  |
+   | SAVEFH               | REQ        |              | Section 18.28  |
+   | SECINFO              | REQ        |              | Section 18.29  |
+NS | SECINFO_NO_NAME      | REC        | pNFS files   | Section 18.45, |
+   |                      |            | layout (REQ) | Section 13.12  |
+I  | SEQUENCE             | REQ        |              | Section 18.46  |
+   | SETATTR              | REQ        |              | Section 18.30  |
+   | SETCLIENTID          | MNI        |              | N/A            |
+   | SETCLIENTID_CONFIRM  | MNI        |              | N/A            |
+NS | SET_SSV              | REQ        |              | Section 18.47  |
+NS | TEST_STATEID         | REQ        |              | Section 18.48  |
+   | VERIFY               | REQ        |              | Section 18.31  |
+NS*| WANT_DELEGATION      | OPT        | FDELG (OPT)  | Section 18.49  |
+   | WRITE                | REQ        |              | Section 18.32  |
+
+Callback Operations
+
+   +-------------------------+-----------+-------------+---------------+
+   | Operation               | REQ, REC, | Feature     | Definition    |
+   |                         | OPT, or   | (REQ, REC,  |               |
+   |                         | MNI       | or OPT)     |               |
+   +-------------------------+-----------+-------------+---------------+
+   | CB_GETATTR              | OPT       | FDELG (REQ) | Section 20.1  |
+P  | CB_LAYOUTRECALL         | OPT       | pNFS (REQ)  | Section 20.3  |
+NS*| CB_NOTIFY               | OPT       | DDELG (REQ) | Section 20.4  |
+P  | CB_NOTIFY_DEVICEID      | OPT       | pNFS (OPT)  | Section 20.12 |
+NS*| CB_NOTIFY_LOCK          | OPT       |             | Section 20.11 |
+NS*| CB_PUSH_DELEG           | OPT       | FDELG (OPT) | Section 20.5  |
+   | CB_RECALL               | OPT       | FDELG,      | Section 20.2  |
+   |                         |           | DDELG, pNFS |               |
+   |                         |           | (REQ)       |               |
+NS*| CB_RECALL_ANY           | OPT       | FDELG,      | Section 20.6  |
+   |                         |           | DDELG, pNFS |               |
+   |                         |           | (REQ)       |               |
+NS | CB_RECALL_SLOT          | REQ       |             | Section 20.8  |
+NS*| CB_RECALLABLE_OBJ_AVAIL | OPT       | DDELG, pNFS | Section 20.7  |
+   |                         |           | (REQ)       |               |
+I  | CB_SEQUENCE             | OPT       | FDELG,      | Section 20.9  |
+   |                         |           | DDELG, pNFS |               |
+   |                         |           | (REQ)       |               |
+NS*| CB_WANTS_CANCELLED      | OPT       | FDELG,      | Section 20.10 |
+   |                         |           | DDELG, pNFS |               |
+   |                         |           | (REQ)       |               |
+   +-------------------------+-----------+-------------+---------------+
+
+Implementation notes:
+
+EXCHANGE_ID:
+* only SP4_NONE state protection supported
+* implementation ids are ignored
+
+CREATE_SESSION:
+* backchannel attributes are ignored
+* backchannel security parameters are ignored
+
+SEQUENCE:
+* no support for dynamic slot table renegotiation (optional)
Signature · 1 line
1.6.2.1
#200 lore raw
In reply to Benny Halevy
Quoted 11 lines · Benny Halevy
Quoted 10 lines · J. Bruce Fields
Quoted 8 lines · Benny Halevy

Hi Bruce,

Here's the latest server patches implementing the NFSv4.1 Sessions features.

This patchset is based over your for-2.6.30 branch and is also available from
git://linux-nfs.org/~bhalevy/linux-pnfs.git nfsd41-for-2.6.30

Quoted 8 lines
Quoted 6 lines

There's a few merge conflicts with my current for-2.6.30--probably my fault for not pushing that out recently enough, apologies. Would you mind updating?

--b.

Sure. Here's a rebased version with two minor changes (see below)

Got it, thanks! Reading....

--b.

#8 lore raw

Re: [PATCH 02/47] sunrpc: add cl_private field to struct rpc_clnt

In reply to Benny Halevy
Quoted 14 lines · Benny Halevy
Quoted 4 lines · Trond
Quoted 1 line · J. Bruce Fields

On Fri, Mar 27, 2009 at 06:01:48AM +0300, Benny Halevy wrote:

Quoted 8 lines
Quoted 4 lines
Quoted 3 lines

From: Andy Adamson <[email protected]>

Note: the NFSv4.1 client also uses (and declares) this pointer.

OK. Ack from trond?

First, someone would need to remind me why it is necessary, and add that justification to the changelog.

Quoted 8 lines

First time this is used in this patchset is here:
[PATCH 35/47] nfsd: minorversion support for the back channel

The client uses cl_private to determine the minorversion (via a struct nfs_client *) to be set in the compound header, and to know when to generate a SEQUENCE op. Similarly, the server puts a struct nfs4_callback * in there for callback compounds' CB_COMPOUND and CB_SEQUENCE.

Why would the rpc_client need to know and track minor versions? That is an NFS protocol specific thing.

Besides, the caller should always know what minor version it is using. It shouldn't need a back-pointer in the rpc_client...

Signature · 6 lines
Trond Myklebust
Linux NFS client maintainer

NetApp
[email protected]
www.netapp.com
#10 lore raw

Re: [PATCH 02/47] sunrpc: add cl_private field to struct rpc_clnt

In reply to a message not in the archive
Quoted 16 lines · Trond Myklebust
Quoted 15 lines · Benny Halevy
Quoted 11 lines · Trond
Quoted 5 lines · J. Bruce Fields
Quoted 3 lines · Benny Halevy

From: Andy Adamson <[email protected]>

Note: the NFSv4.1 client also uses (and declares) this pointer.

OK. Ack from trond?

First, someone would need to remind me why it is necessary, and add that justification to the changelog.

First time this is used in this patchset is here:
[PATCH 35/47] nfsd: minorversion support for the back channel

Quoted 8 lines
Quoted 5 lines

The client uses cl_private to determine the minorversion (via a struct nfs_client *) to be set in the compound header, and to know when to generate a SEQUENCE op. Similarly, the server puts a struct nfs4_callback * in there for callback compounds' CB_COMPOUND and CB_SEQUENCE.

Why would the rpc_client need to know and track minor versions? That is an NFS protocol specific thing.

Agreed. the rpc_client doesn't know anything about the minor version. This is why we store it in a void *cl_private.

Quoted 4 lines

Besides, the caller should always know what minor version it is using. It shouldn't need a back-pointer in the rpc_client...

The back pointer is used in the xdr layer like this:
struct xdr_stream xdr;
+ struct nfs_client *clp =
+ (struct nfs_client *)req->rq_task->tk_client->cl_private;
struct compound_hdr hdr = {
- .nops = 0,
+ .minorversion = clp->cl_minorversion,
};

Benny

#11 lore raw

Re: [PATCH 02/47] sunrpc: add cl_private field to struct rpc_clnt

Quoted 35 lines · Benny Halevy
Quoted 16 lines · Trond Myklebust
Quoted 15 lines · Benny Halevy
Quoted 11 lines · Trond
Quoted 5 lines · J. Bruce Fields
Quoted 3 lines · Benny Halevy

From: Andy Adamson <[email protected]>

Note: the NFSv4.1 client also uses (and declares) this pointer.

OK. Ack from trond?

First, someone would need to remind me why it is necessary, and add that justification to the changelog.

First time this is used in this patchset is here:
[PATCH 35/47] nfsd: minorversion support for the back channel

Quoted 8 lines
Quoted 5 lines

The client uses cl_private to determine the minorversion (via a struct nfs_client *) to be set in the compound header, and to know when to generate a SEQUENCE op. Similarly, the server puts a struct nfs4_callback * in there for callback compounds' CB_COMPOUND and CB_SEQUENCE.

Why would the rpc_client need to know and track minor versions? That is an NFS protocol specific thing.

Agreed. the rpc_client doesn't know anything about the minor version. This is why we store it in a void *cl_private.

Quoted 4 lines

Besides, the caller should always know what minor version it is using. It shouldn't need a back-pointer in the rpc_client...

The back pointer is used in the xdr layer like this:

Review

Quoted 7 lines
(quoted diff) +3 -1
 	struct xdr_stream xdr;
+	struct nfs_client *clp =
+		(struct nfs_client *)req->rq_task->tk_client->cl_private;
 	struct compound_hdr hdr = {
-		.nops = 0,
+		.minorversion = clp->cl_minorversion,
 	};

Just put that information in the RPC arguments where it _BELONGS_.

Signature · 6 lines
Trond Myklebust
Linux NFS client maintainer

NetApp
[email protected]
www.netapp.com
#210 lore raw
In reply to J. Bruce Fields
Quoted 21 lines · J. Bruce Fields
Quoted 8 lines · Benny Halevy

Hi Bruce,

Here's the latest server patches implementing the NFSv4.1 Sessions features.

This patchset is based over your for-2.6.30 branch and is also available from
git://linux-nfs.org/~bhalevy/linux-pnfs.git nfsd41-for-2.6.30

I ran the v4.0 pynfs on that branch and got some new failures:

ACC3 st_access.testNoFh : FAILURE
value=10064 not in enum nfsstat4
CLOSE7 st_close.testNoCfh : FAILURE
value=10064 not in enum nfsstat4
CMT3 st_commit.testNoFh : FAILURE
value=10064 not in enum nfsstat4
CR8 st_create.testNoFh : FAILURE
value=10064 not in enum nfsstat4

Quoted 2 lines

Looks like NFS4ERR_SEQUENCE_POS is leaking out to nfsv4 code for some reason?

Right, we checked ALLOWED_AS_FIRST_OP for minorversion 0 too.

Quoted 12 lines

(Some similar results omitted.)

Also there's a bunch of stuff like this:

NVF1a st_nverify.testMandLink : FAILURE
NVerifying mandatory attributes against getattr should return NFS4ERR_SAME, instead got NFS4_OK
NVF1d st_nverify.testMandDir : FAILURE
NVerifying mandatory attributes against getattr should return NFS4ERR_SAME, instead got NFS4_OK

Quoted 3 lines

Probably the expansion of the attribute bitmask is exposing some preexisting bug in the 4.0 code, but let's try to figure out how to fix that.

A change to nfsd4_encode_fattr to encode bitmaps of variable length (rather than 2) broke _nfsd4_verify.

Quoted 4 lines

I haven't investigated either of these any closer yet.

--b.

The following two patches fix these regressions:
[PATCH 1/2] SQUASHME: nfsd41: do not verify nfserr_sequence_pos for minorversion 0 [PATCH 2/2] nfsd: dynamically skip encoded fattr bitmap in _nfsd4_verify

Benny

#211 lore raw

PATCH 1/2 SQUASHME: nfsd41: do not verify nfserr_sequence_pos for minorversion 0

squash into "nfsd41: enforce NFS4ERR_SEQUENCE_POS operation order rules"

---
 fs/nfsd/nfs4proc.c |   17 ++++++++++++++---
 1 files changed, 14 insertions(+), 3 deletions(-)
fs/nfsd/nfs4proc.c +14 -3
index 6d7d7c5..20583ab 100644
@@ -894,6 +894,18 @@ nfsd4_enc_no_page_replay(struct nfsd4_compoundargs *args,
 	return op->status;
 }
 
+static bool nfs41_op_ordering_ok(struct nfsd4_compoundargs *args)
+{
+#if defined(CONFIG_NFSD_V4_1)
+	if (args->minorversion && args->opcnt > 0) {
+		struct nfsd4_op *op = &args->ops[0];
+		return (op->status == nfserr_op_illegal) ||
+		       (nfsd4_ops[op->opnum].op_flags & ALLOWED_AS_FIRST_OP);
+	}
+#endif /* CONFIG_NFSD_V4_1 */
+	return true;
+}
+
 /*
  * COMPOUND call.
  */
@@ -933,9 +945,8 @@ nfsd4_proc_compound(struct svc_rqst *rqstp,
 	if (args->minorversion > NFSD_SUPPORTED_MINOR_VERSION)
 		goto out;
 
-	op = &args->ops[0];
-	if (args->opcnt > 0 && op->status != nfserr_op_illegal &&
-	    !(nfsd4_ops[op->opnum].op_flags & ALLOWED_AS_FIRST_OP)) {
+	if (!nfs41_op_ordering_ok(args)) {
+		op = &args->ops[0];
 		op->status = nfserr_sequence_pos;
 		goto encode_op;
 	}
Signature · 1 line
1.6.2.1
#212 lore raw

PATCH 2/2 nfsd: dynamically skip encoded fattr bitmap in _nfsd4_verify

In reply to Benny Halevy

_nfsd4_verify currently skips 3 words from the encoded buffer begining. With support for 3-word attr bitmaps in nfsd41, nfsd4_encode_fattr may encode 1, 2, or 3 words, and not always 2 as it used to be, hence we need to find out where to skip using the encoded bitmap length.

Note: This patch may be applied over pre-nfsd41 nfsd.
move this patch in front of "nfsd41: support for 3-word long attribute bitmask"

---
 fs/nfsd/nfs4proc.c |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)
fs/nfsd/nfs4proc.c +2 -1
index 20583ab..a393d38 100644
@@ -801,7 +801,8 @@ _nfsd4_verify(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 	if (status)
 		goto out_kfree;
 
-	p = buf + 3;
+	/* skip bitmap */
+	p = buf + 1 + ntohl(buf[0]);
 	status = nfserr_not_same;
 	if (ntohl(*p++) != verify->ve_attrlen)
 		goto out_kfree;
Signature · 1 line
1.6.2.1
#13 lore raw

Re: [PATCH 02/47] sunrpc: add cl_private field to struct rpc_clnt

In reply to a message not in the archive
Quoted 30 lines · Trond Myklebust
Quoted 29 lines · Benny Halevy
Quoted 15 lines · Trond Myklebust
Quoted 14 lines · Benny Halevy
Quoted 10 lines · Trond
Quoted 5 lines · J. Bruce Fields
Quoted 3 lines · Benny Halevy

From: Andy Adamson <[email protected]>

Note: the NFSv4.1 client also uses (and declares) this pointer.

OK. Ack from trond?

First, someone would need to remind me why it is necessary, and add that justification to the changelog.

First time this is used in this patchset is here:
[PATCH 35/47] nfsd: minorversion support for the back channel

Quoted 7 lines
Quoted 5 lines

The client uses cl_private to determine the minorversion (via a struct nfs_client *) to be set in the compound header, and to know when to generate a SEQUENCE op. Similarly, the server puts a struct nfs4_callback * in there for callback compounds' CB_COMPOUND and CB_SEQUENCE.

Why would the rpc_client need to know and track minor versions? That is an NFS protocol specific thing.

Agreed. the rpc_client doesn't know anything about the minor version. This is why we store it in a void *cl_private.

Quoted 3 lines

Besides, the caller should always know what minor version it is using. It shouldn't need a back-pointer in the rpc_client...

Quoted 11 lines

Review

Quoted 8 lines

The back pointer is used in the xdr layer like this:

(quoted diff) +3 -1
 	struct xdr_stream xdr;
+	struct nfs_client *clp =
+		(struct nfs_client *)req->rq_task->tk_client->cl_private;
 	struct compound_hdr hdr = {
-		.nops = 0,
+		.minorversion = clp->cl_minorversion,
 	};

Just put that information in the RPC arguments where it _BELONGS_.

BTW, we are going back and forth on this... cl_private was suggested on the review we did on November '08:
http://linux-nfs.org/pipermail/pnfs/2008-November/005435.html

Anyway, the following adds a *session to the nfs4_sequence_args and res. It increases the memory footprint of all rpc args and res but I agree it's cleaner.

---
git diff --stat -p 2e00c20

 fs/nfs/client.c             |    1 -
 fs/nfs/nfs4proc.c           |    8 ++-
 fs/nfs/nfs4xdr.c            |  235 ++++++++++++++++---------------------------
 include/linux/nfs_xdr.h     |    7 ++
 include/linux/sunrpc/clnt.h |    1 -
 5 files changed, 99 insertions(+), 153 deletions(-)
fs/nfs/client.c +0 -1
index 6e3ab63..bce02da 100644
@@ -648,7 +648,6 @@ static int nfs_create_rpc_client(struct nfs_client *clp,
 	}
 
 	clp->cl_rpcclient = clnt;
-	clnt->cl_private = clp;
 	return 0;
 }
 
fs/nfs/nfs4proc.c +7 -1
index 5614cfa..e997882 100644
@@ -514,11 +514,13 @@ static int nfs41_setup_sequence(struct nfs4_session *session,
 	spin_unlock(&tbl->slot_tbl_lock);
 
 	slot = tbl->slots + slotid;
+	args->sa_session = session;
 	args->sa_slotid = slotid;
 	args->sa_cache_this = cache_reply;
 
 	dprintk("<-- %s slotid=%d seqid=%d\n", __func__, slotid, slot->seq_nr);
 
+	res->sr_session = session;
 	res->sr_slotid = slotid;
 	res->sr_renewal_time = jiffies;
 	/*
@@ -4244,6 +4246,7 @@ static int nfs4_proc_exchange_id(struct nfs_client *clp, struct rpc_cred *cred)
 {
 	nfs4_verifier verifier;
 	struct nfs41_exchange_id_args args = {
+		.client = clp,
 		.flags = clp->cl_exchange_flags,
 	};
 	struct nfs41_exchange_id_res res = {
@@ -4659,10 +4662,13 @@ static int _nfs4_proc_create_session(struct nfs_client *clp)
 		.client = clp,
 		.cb_program = NFS4_CALLBACK,
 	};
-
+	struct nfs41_create_session_res res = {
+		.client = clp,
+	};
 	struct rpc_message msg = {
 		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_CREATE_SESSION],
 		.rpc_argp = &args,
+		.rpc_resp = &res,
 	};
 	int status;
 
fs/nfs/nfs4xdr.c +85 -150

665 lines not on this page · raw message

include/linux/nfs_xdr.h +7 -0
index 517291e..4feacc6 100644
@@ -130,11 +130,13 @@ struct nfs4_slot {
 };
 
 struct nfs4_sequence_args {
+	struct nfs4_session	*sa_session;
 	u8			sa_slotid;
 	u8			sa_cache_this;
 };
 
 struct nfs4_sequence_res {
+	struct nfs4_session	*sr_session;
 	u8			sr_slotid;	/* slot used to send request */
 	unsigned long		sr_renewal_time;
 	int			sr_status;	/* sequence operation status */
@@ -870,6 +872,7 @@ struct nfs_impl_id4 {
 
 #define NFS4_EXCHANGE_ID_LEN	(48)
 struct nfs41_exchange_id_args {
+	struct nfs_client		*client;
 	nfs4_verifier			*verifier;
 	unsigned int 			id_len;
 	char 				id[NFS4_EXCHANGE_ID_LEN];
@@ -902,6 +905,10 @@ struct nfs41_create_session_args {
 	struct nfs4_channel_attrs	fc_attrs;	/* Fore Channel */
 	struct nfs4_channel_attrs	bc_attrs;	/* Back Channel */
 };
+
+struct nfs41_create_session_res {
+	struct nfs_client	       *client;
+};
 #endif /* CONFIG_NFS_V4_1 */
 
 struct nfs_page;
include/linux/sunrpc/clnt.h +0 -1
index 8ca3fef..37881f1 100644
@@ -53,7 +53,6 @@ struct rpc_clnt {
 	char			cl_pathname[30];/* Path in rpc_pipe_fs */
 	struct vfsmount *	cl_vfsmnt;
 	struct dentry *		cl_dentry;	/* inode */
-	void			*cl_private;	/* private data */
 	struct rpc_clnt *	cl_parent;	/* Points to parent of clones */
 	struct rpc_rtt		cl_rtt_default;
 	struct rpc_timeout	cl_timeout_default;
#14 lore raw

Re: [PATCH 02/47] sunrpc: add cl_private field to struct rpc_clnt

Quoted 43 lines · Benny Halevy
Quoted 30 lines · Trond Myklebust
Quoted 29 lines · Benny Halevy
Quoted 15 lines · Trond Myklebust
Quoted 14 lines · Benny Halevy
Quoted 10 lines · Trond
Quoted 5 lines · J. Bruce Fields
Quoted 3 lines · Benny Halevy

From: Andy Adamson <[email protected]>

Note: the NFSv4.1 client also uses (and declares) this pointer.

OK. Ack from trond?

First, someone would need to remind me why it is necessary, and add that justification to the changelog.

First time this is used in this patchset is here:
[PATCH 35/47] nfsd: minorversion support for the back channel

Quoted 7 lines
Quoted 5 lines

The client uses cl_private to determine the minorversion (via a struct nfs_client *) to be set in the compound header, and to know when to generate a SEQUENCE op. Similarly, the server puts a struct nfs4_callback * in there for callback compounds' CB_COMPOUND and CB_SEQUENCE.

Why would the rpc_client need to know and track minor versions? That is an NFS protocol specific thing.

Agreed. the rpc_client doesn't know anything about the minor version. This is why we store it in a void *cl_private.

Quoted 3 lines

Besides, the caller should always know what minor version it is using. It shouldn't need a back-pointer in the rpc_client...

Quoted 11 lines

Review

Quoted 8 lines

The back pointer is used in the xdr layer like this:

(quoted diff) +3 -1
 	struct xdr_stream xdr;
+	struct nfs_client *clp =
+		(struct nfs_client *)req->rq_task->tk_client->cl_private;
 	struct compound_hdr hdr = {
-		.nops = 0,
+		.minorversion = clp->cl_minorversion,
 	};

Just put that information in the RPC arguments where it _BELONGS_.

Quoted 3 lines

BTW, we are going back and forth on this... cl_private was suggested on the review we did on November '08:
http://linux-nfs.org/pipermail/pnfs/2008-November/005435.html

No. I've never acked the idea of putting RPC function arguments into private fields in the rpc_client. That's simply unacceptably ugly. If you and Andy got that impression, then one of us must have misunderstood what the other party was saying.

I do seem to remember talk about needing back-pointers to the session in order to support the NFSv4.1 backchannel, but that's an entirely different matter.

Note that it is probably worth caching the minor version inside the session pointer instead of having to dereference the nfs_client on every RPC call.
That can be a separate patch, though.

Quoted 3 lines

Anyway, the following adds a *session to the nfs4_sequence_args and res. It increases the memory footprint of all rpc args and res but I agree it's cleaner.

The session is clearly an argument to the sequence op, and so deserves to be kept in the rpc_args.

Signature · 6 lines
Trond Myklebust
Linux NFS client maintainer

NetApp
[email protected]
www.netapp.com
#64 lore raw

Re: [PATCH v2 02/47] nfsd: embed nfsd4_current_state in nfsd4_compoundres

In reply to Benny Halevy
Quoted 3 lines · Benny Halevy

From: Andy Adamson <[email protected]>

Remove the allocation of struct nfsd4_compound_state.

Thanks, applied. (Some minor fixup required as I wanted to apply this before #1.)

--b.

Quoted 121 lines
---
 fs/nfsd/nfs4proc.c        |   40 +++++++++-------------------------------
 include/linux/nfsd/xdr4.h |    9 +++++----
 2 files changed, 14 insertions(+), 35 deletions(-)
fs/nfsd/nfs4proc.c +9 -31
index 7839654..c8e1ad0 100644
@@ -809,29 +809,6 @@ static inline void nfsd4_increment_op_stats(u32 opnum)
 		nfsdstats.nfs4_opcount[opnum]++;
 }
 
-static void cstate_free(struct nfsd4_compound_state *cstate)
-{
-	if (cstate == NULL)
-		return;
-	fh_put(&cstate->current_fh);
-	fh_put(&cstate->save_fh);
-	BUG_ON(cstate->replay_owner);
-	kfree(cstate);
-}
-
-static struct nfsd4_compound_state *cstate_alloc(void)
-{
-	struct nfsd4_compound_state *cstate;
-
-	cstate = kmalloc(sizeof(struct nfsd4_compound_state), GFP_KERNEL);
-	if (cstate == NULL)
-		return NULL;
-	fh_init(&cstate->current_fh, NFS4_FHSIZE);
-	fh_init(&cstate->save_fh, NFS4_FHSIZE);
-	cstate->replay_owner = NULL;
-	return cstate;
-}
-
 typedef __be32(*nfsd4op_func)(struct svc_rqst *, struct nfsd4_compound_state *,
 			      void *);
 
@@ -859,12 +836,13 @@ nfsd4_proc_compound(struct svc_rqst *rqstp,
 {
 	struct nfsd4_op	*op;
 	struct nfsd4_operation *opdesc;
-	struct nfsd4_compound_state *cstate = NULL;
+	struct nfsd4_compound_state *cstate = &resp->cstate;
 	int		slack_bytes;
 	__be32		status;
 
 	resp->xbuf = &rqstp->rq_res;
-	resp->p = rqstp->rq_res.head[0].iov_base + rqstp->rq_res.head[0].iov_len;
+	resp->p = rqstp->rq_res.head[0].iov_base +
+						rqstp->rq_res.head[0].iov_len;
 	resp->tagp = resp->p;
 	/* reserve space for: taglen, tag, and opcnt */
 	resp->p += 2 + XDR_QUADLEN(args->taglen);
@@ -873,6 +851,9 @@ nfsd4_proc_compound(struct svc_rqst *rqstp,
 	resp->tag = args->tag;
 	resp->opcnt = 0;
 	resp->rqstp = rqstp;
+	resp->cstate.replay_owner = NULL;
+	fh_init(&resp->cstate.current_fh, NFS4_FHSIZE);
+	fh_init(&resp->cstate.save_fh, NFS4_FHSIZE);
 	/* Use the deferral mechanism only for NFSv4.0 compounds */
 	rqstp->rq_usedeferral = (args->minorversion == 0);
 
@@ -883,11 +864,6 @@ nfsd4_proc_compound(struct svc_rqst *rqstp,
 	if (args->minorversion > NFSD_SUPPORTED_MINOR_VERSION)
 		goto out;
 
-	status = nfserr_resource;
-	cstate = cstate_alloc();
-	if (cstate == NULL)
-		goto out;
-
 	status = nfs_ok;
 	while (!status && resp->opcnt < args->opcnt) {
 		op = &args->ops[resp->opcnt++];
@@ -964,7 +940,9 @@ encode_op:
 		status = nfserr_jukebox;
 	}
 
-	cstate_free(cstate);
+	fh_put(&resp->cstate.current_fh);
+	fh_put(&resp->cstate.save_fh);
+	BUG_ON(resp->cstate.replay_owner);
 out:
 	nfsd4_release_compoundargs(args);
 	/* Reset deferral mechanism for RPC deferrals */
include/linux/nfsd/xdr4.h +5 -4
index 27bd3e3..fd15ddc 100644
@@ -45,9 +45,9 @@
 #define XDR_LEN(n)                     (((n) + 3) & ~3)
 
 struct nfsd4_compound_state {
-	struct svc_fh current_fh;
-	struct svc_fh save_fh;
-	struct nfs4_stateowner *replay_owner;
+	struct svc_fh		current_fh;
+	struct svc_fh		save_fh;
+	struct nfs4_stateowner	*replay_owner;
 };
 
 struct nfsd4_change_info {
@@ -416,7 +416,8 @@ struct nfsd4_compoundres {
 	u32				taglen;
 	char *				tag;
 	u32				opcnt;
-	__be32 *			tagp; /* where to encode tag and  opcount */
+	__be32 *			tagp; /* tag, opcount encode location */
+	struct nfsd4_compound_state	cstate;
 };
 
 #define NFS4_SVC_XDRSIZE		sizeof(struct nfsd4_compoundargs)
Signature · 1 line
1.6.2.1
#66 lore raw

Re: [PATCH v2 03/47] nfsd: add a struct nfsd4_slot pointer to struct nfsd4_compound_state

In reply to Benny Halevy
Quoted 3 lines · Benny Halevy

From: Andy Adamson <[email protected]>

To be used for sessions DRC replay.

It'd be more helpful to have this combined with whatever patch first defines and/or uses nfsd4_slot.

--b.

Quoted 23 lines
---
 include/linux/nfsd/xdr4.h |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)
include/linux/nfsd/xdr4.h +2 -0
index fd15ddc..e309d55 100644
@@ -48,6 +48,8 @@ struct nfsd4_compound_state {
 	struct svc_fh		current_fh;
 	struct svc_fh		save_fh;
 	struct nfs4_stateowner	*replay_owner;
+	/* For sessions DRC */
+	struct nfsd4_slot	*slot;
 };
 
 struct nfsd4_change_info {
Signature · 1 line
1.6.2.1
#100 lore raw

Re: [PATCH v2 07/47] nfsd41: define nfs41 error codes

In reply to Benny Halevy
Quoted 18 lines · Benny Halevy

From: Marc Eshel <[email protected]>

Define all error code present in
http://tools.ietf.org/html/draft-ietf-nfsv4-minorversion1-29.

[nfsd41: clean up error code definitions]
---
 include/linux/nfs.h       |    2 --
 include/linux/nfsd/nfsd.h |   43 ++++++++++++++++++++++++++++++++++++++++++-
 2 files changed, 42 insertions(+), 3 deletions(-)
include/linux/nfs.h +0 -0
index 1510a64..214d499 100644

include/linux/nfs.h line 113

Quoted 5 lines
include/linux/nfs.h +0 -2
@@ -109,8 +109,6 @@
 	NFSERR_FILE_OPEN = 10046,      /*       v4 */
 	NFSERR_ADMIN_REVOKED = 10047,  /*       v4 */
 	NFSERR_CB_PATH_DOWN = 10048,   /*       v4 */
-
-	NFSERR_REPLAY_ME = 11001,	/* linux internal */

This renders "[PATCH v2 05/47] nfsd41: change NFSERR_REPLAY_ME" moot; just drop that previous patch.

--b.

Quoted 67 lines
include/linux/nfs.h +0 -0
 };
 
 /* NFSv2 file types - beware, these are not the same in NFSv3 */
include/linux/nfsd/nfsd.h +42 -1
index 54beda1..ab9616d 100644
@@ -249,7 +249,44 @@ void		nfsd_lockd_shutdown(void);
 #define	nfserr_cb_path_down	cpu_to_be32(NFSERR_CB_PATH_DOWN)
 #define	nfserr_locked		cpu_to_be32(NFSERR_LOCKED)
 #define	nfserr_wrongsec		cpu_to_be32(NFSERR_WRONGSEC)
-#define	nfserr_replay_me	cpu_to_be32(NFSERR_REPLAY_ME)
+#define nfserr_badiomode		cpu_to_be32(NFS4ERR_BADIOMODE)
+#define nfserr_badlayout		cpu_to_be32(NFS4ERR_BADLAYOUT)
+#define nfserr_bad_session_digest	cpu_to_be32(NFS4ERR_BAD_SESSION_DIGEST)
+#define nfserr_badsession		cpu_to_be32(NFS4ERR_BADSESSION)
+#define nfserr_badslot			cpu_to_be32(NFS4ERR_BADSLOT)
+#define nfserr_complete_already		cpu_to_be32(NFS4ERR_COMPLETE_ALREADY)
+#define nfserr_conn_not_bound_to_session cpu_to_be32(NFS4ERR_CONN_NOT_BOUND_TO_SESSION)
+#define nfserr_deleg_already_wanted	cpu_to_be32(NFS4ERR_DELEG_ALREADY_WANTED)
+#define nfserr_back_chan_busy		cpu_to_be32(NFS4ERR_BACK_CHAN_BUSY)
+#define nfserr_layouttrylater		cpu_to_be32(NFS4ERR_LAYOUTTRYLATER)
+#define nfserr_layoutunavailable	cpu_to_be32(NFS4ERR_LAYOUTUNAVAILABLE)
+#define nfserr_nomatching_layout	cpu_to_be32(NFS4ERR_NOMATCHING_LAYOUT)
+#define nfserr_recallconflict		cpu_to_be32(NFS4ERR_RECALLCONFLICT)
+#define nfserr_unknown_layouttype	cpu_to_be32(NFS4ERR_UNKNOWN_LAYOUTTYPE)
+#define nfserr_seq_misordered		cpu_to_be32(NFS4ERR_SEQ_MISORDERED)
+#define nfserr_sequence_pos		cpu_to_be32(NFS4ERR_SEQUENCE_POS)
+#define nfserr_req_too_big		cpu_to_be32(NFS4ERR_REQ_TOO_BIG)
+#define nfserr_rep_too_big		cpu_to_be32(NFS4ERR_REP_TOO_BIG)
+#define nfserr_rep_too_big_to_cache	cpu_to_be32(NFS4ERR_REP_TOO_BIG_TO_CACHE)
+#define nfserr_retry_uncached_rep	cpu_to_be32(NFS4ERR_RETRY_UNCACHED_REP)
+#define nfserr_unsafe_compound		cpu_to_be32(NFS4ERR_UNSAFE_COMPOUND)
+#define nfserr_too_many_ops		cpu_to_be32(NFS4ERR_TOO_MANY_OPS)
+#define nfserr_op_not_in_session	cpu_to_be32(NFS4ERR_OP_NOT_IN_SESSION)
+#define nfserr_hash_alg_unsupp		cpu_to_be32(NFS4ERR_HASH_ALG_UNSUPP)
+#define nfserr_clientid_busy		cpu_to_be32(NFS4ERR_CLIENTID_BUSY)
+#define nfserr_pnfs_io_hole		cpu_to_be32(NFS4ERR_PNFS_IO_HOLE)
+#define nfserr_seq_false_retry		cpu_to_be32(NFS4ERR_SEQ_FALSE_RETRY)
+#define nfserr_bad_high_slot		cpu_to_be32(NFS4ERR_BAD_HIGH_SLOT)
+#define nfserr_deadsession		cpu_to_be32(NFS4ERR_DEADSESSION)
+#define nfserr_encr_alg_unsupp		cpu_to_be32(NFS4ERR_ENCR_ALG_UNSUPP)
+#define nfserr_pnfs_no_layout		cpu_to_be32(NFS4ERR_PNFS_NO_LAYOUT)
+#define nfserr_not_only_op		cpu_to_be32(NFS4ERR_NOT_ONLY_OP)
+#define nfserr_wrong_cred		cpu_to_be32(NFS4ERR_WRONG_CRED)
+#define nfserr_wrong_type		cpu_to_be32(NFS4ERR_WRONG_TYPE)
+#define nfserr_dirdeleg_unavail		cpu_to_be32(NFS4ERR_DIRDELEG_UNAVAIL)
+#define nfserr_reject_deleg		cpu_to_be32(NFS4ERR_REJECT_DELEG)
+#define nfserr_returnconflict		cpu_to_be32(NFS4ERR_RETURNCONFLICT)
+#define nfserr_deleg_revoked		cpu_to_be32(NFS4ERR_DELEG_REVOKED)
 
 /* error codes for internal use */
 /* if a request fails due to kmalloc failure, it gets dropped.
@@ -258,6 +295,10 @@ void		nfsd_lockd_shutdown(void);
 #define	nfserr_dropit		cpu_to_be32(30000)
 /* end-of-file indicator in readdir */
 #define	nfserr_eof		cpu_to_be32(30001)
+/* replay detected */
+#define	nfserr_replay_me	cpu_to_be32(11001)
+/* nfs41 replay detected */
+#define	nfserr_replay_cache	cpu_to_be32(11002)
 
 /* Check for dir entries '.' and '..' */
 #define isdotent(n, l)	(l < 3 && n[0] == '.' && (l == 1 || n[1] == '.'))
Signature · 1 line
1.6.2.1
#201 lore raw
In reply to Benny Halevy
Quoted 11 lines · Benny Halevy
Quoted 10 lines · J. Bruce Fields
Quoted 8 lines · Benny Halevy

Hi Bruce,

Here's the latest server patches implementing the NFSv4.1 Sessions features.

This patchset is based over your for-2.6.30 branch and is also available from
git://linux-nfs.org/~bhalevy/linux-pnfs.git nfsd41-for-2.6.30

Quoted 8 lines
Quoted 6 lines

There's a few merge conflicts with my current for-2.6.30--probably my fault for not pushing that out recently enough, apologies. Would you mind updating?

--b.

Sure. Here's a rebased version with two minor changes (see below)

On the latest version of nfsd41-for-2.6.30 (a564667..) I'm getting a new NULL dereference in the callback code. Looks like it probably happened while running connectathon over NFSv4.0 with krb5p. That's all I've figured out so far.

--b.

BUG: unable to handle kernel NULL pointer dereference at (null)
IP: [<c03e4e1e>] nfs4_xdr_dec_cb_recall+0x4e/0x200
*pde = 00000000 
Oops: 0000 [#1] PREEMPT 
last sysfs file: /sys/kernel/uevent_seqnum
Modules linked in:

Pid: 3949, comm: nfs4_cb_recall Not tainted (2.6.29-rc8-00312-ga564667 #39)
EIP: 0060:[<c03e4e1e>] EFLAGS: 00010286 CPU: 0
EIP is at nfs4_xdr_dec_cb_recall+0x4e/0x200
EAX: c7bb22a0 EBX: c7bb2298 ECX: c7bb22a0 EDX: c7bb22a4
ESI: 00000000 EDI: c66e3000 EBP: c5281ea4 ESP: c5281e6c
DS: 007b ES: 007b FS: 0000 GS: 0000 SS: 0068
Process nfs4_cb_recall (pid: 3949, ti=c5280000 task=c66a6b60 task.ti=c5280000)

Stack:
 c66a6b60 00000000 c5281eb4 c086cab4 c0aa0380 00000001 c11bd4a4 c7bb22a4
 c66e302c c7bb22ac c66e302c 00000000 c6cf14e8 c66e3000 c5281ecc c08199c3
 c0819aae 00000000 c5281ee8 c5281ef0 c03e4dd0 c6cf14e8 c7bb2298 c03e4dd0
Call Trace:
 [<c086cab4>] ? schedule+0x354/0x540
 [<c08199c3>] ? rpcauth_unwrap_resp+0x63/0x90
 [<c0819aae>] ? rpcauth_checkverf+0x2e/0x70
 [<c03e4dd0>] ? nfs4_xdr_dec_cb_recall+0x0/0x200
 [<c03e4dd0>] ? nfs4_xdr_dec_cb_recall+0x0/0x200
 [<c08122ce>] ? call_decode+0x1ae/0x820
 [<c086d125>] ? out_of_line_wait_on_bit+0x65/0x80
 [<c0818720>] ? rpc_wait_bit_killable+0x0/0x40
 [<c03e4dd0>] ? nfs4_xdr_dec_cb_recall+0x0/0x200
 [<c0818e62>] ? __rpc_execute+0x92/0x290
 [<c086f4fc>] ? _spin_unlock+0x2c/0x50
 [<c08186c7>] ? rpc_set_active+0x67/0x80
 [<c081907e>] ? rpc_execute+0x1e/0x30
 [<c08129e5>] ? rpc_run_task+0x35/0x70
 [<c0812b40>] ? rpc_call_sync+0x40/0x70
 [<c03e51e0>] ? nfsd4_cb_recall+0x70/0x130
 [<c086cab4>] ? schedule+0x354/0x540
 [<c0247d1b>] ? trace_hardirqs_on+0xb/0x10
 [<c03dfd40>] ? do_recall+0x0/0x20
 [<c03dfd54>] ? do_recall+0x14/0x20
 [<c023941f>] ? kthread+0x3f/0x70
 [<c02393e0>] ? kthread+0x0/0x70
 [<c0203b67>] ? kernel_thread_helper+0x7/0x10
Code: 43 00 ba 08 00 00 00 8d 45 e4 e8 0e c1 43 00 85 c0 74 3a 8b 50 04 8d 45 e4 0f ca 83 c2 04 e8 fa c0 43 00 85 c0 0f 84 8a 00 00 00 <8b> 06 8b 00 85 c0 75 32 ba 04 00 00 00 8d 45 e4 e8 cd fe ff ff 
EIP: [<c03e4e1e>] nfs4_xdr_dec_cb_recall+0x4e/0x200 SS:ESP 0068:c5281e6c
---[ end trace 2724475d9856cb6c ]---
#202 lore raw
Quoted 19 lines · bfields
Quoted 18 lines · Benny Halevy
Quoted 16 lines · J. Bruce Fields
Quoted 8 lines · Benny Halevy

Hi Bruce,

Here's the latest server patches implementing the NFSv4.1 Sessions features.

This patchset is based over your for-2.6.30 branch and is also available from
git://linux-nfs.org/~bhalevy/linux-pnfs.git nfsd41-for-2.6.30

There's a few merge conflicts with my current for-2.6.30--probably my fault for not pushing that out recently enough, apologies. Would you mind updating?

--b.

Quoted 6 lines
Quoted 1 line

Sure. Here's a rebased version with two minor changes (see below)

On the latest version of nfsd41-for-2.6.30 (a564667..) I'm getting a new NULL dereference in the callback code. Looks like it probably happened while running connectathon over NFSv4.0 with krb5p. That's all I've figured out so far.

Nevermind about the krb5p, that can't be right. Perhaps cbs_minorversion just isn't getting set correctly, and it's trying to access an uninitialized cbs_clp?

--b.

Quoted 49 lines

--b.

BUG: unable to handle kernel NULL pointer dereference at (null)
IP: [<c03e4e1e>] nfs4_xdr_dec_cb_recall+0x4e/0x200
*pde = 00000000 
Oops: 0000 [#1] PREEMPT 
last sysfs file: /sys/kernel/uevent_seqnum
Modules linked in:

Pid: 3949, comm: nfs4_cb_recall Not tainted (2.6.29-rc8-00312-ga564667 #39)
EIP: 0060:[<c03e4e1e>] EFLAGS: 00010286 CPU: 0
EIP is at nfs4_xdr_dec_cb_recall+0x4e/0x200
EAX: c7bb22a0 EBX: c7bb2298 ECX: c7bb22a0 EDX: c7bb22a4
ESI: 00000000 EDI: c66e3000 EBP: c5281ea4 ESP: c5281e6c
DS: 007b ES: 007b FS: 0000 GS: 0000 SS: 0068
Process nfs4_cb_recall (pid: 3949, ti=c5280000 task=c66a6b60 task.ti=c5280000)

Stack:
 c66a6b60 00000000 c5281eb4 c086cab4 c0aa0380 00000001 c11bd4a4 c7bb22a4
 c66e302c c7bb22ac c66e302c 00000000 c6cf14e8 c66e3000 c5281ecc c08199c3
 c0819aae 00000000 c5281ee8 c5281ef0 c03e4dd0 c6cf14e8 c7bb2298 c03e4dd0
Call Trace:
 [<c086cab4>] ? schedule+0x354/0x540
 [<c08199c3>] ? rpcauth_unwrap_resp+0x63/0x90
 [<c0819aae>] ? rpcauth_checkverf+0x2e/0x70
 [<c03e4dd0>] ? nfs4_xdr_dec_cb_recall+0x0/0x200
 [<c03e4dd0>] ? nfs4_xdr_dec_cb_recall+0x0/0x200
 [<c08122ce>] ? call_decode+0x1ae/0x820
 [<c086d125>] ? out_of_line_wait_on_bit+0x65/0x80
 [<c0818720>] ? rpc_wait_bit_killable+0x0/0x40
 [<c03e4dd0>] ? nfs4_xdr_dec_cb_recall+0x0/0x200
 [<c0818e62>] ? __rpc_execute+0x92/0x290
 [<c086f4fc>] ? _spin_unlock+0x2c/0x50
 [<c08186c7>] ? rpc_set_active+0x67/0x80
 [<c081907e>] ? rpc_execute+0x1e/0x30
 [<c08129e5>] ? rpc_run_task+0x35/0x70
 [<c0812b40>] ? rpc_call_sync+0x40/0x70
 [<c03e51e0>] ? nfsd4_cb_recall+0x70/0x130
 [<c086cab4>] ? schedule+0x354/0x540
 [<c0247d1b>] ? trace_hardirqs_on+0xb/0x10
 [<c03dfd40>] ? do_recall+0x0/0x20
 [<c03dfd54>] ? do_recall+0x14/0x20
 [<c023941f>] ? kthread+0x3f/0x70
 [<c02393e0>] ? kthread+0x0/0x70
 [<c0203b67>] ? kernel_thread_helper+0x7/0x10
Code: 43 00 ba 08 00 00 00 8d 45 e4 e8 0e c1 43 00 85 c0 74 3a 8b 50 04 8d 45 e4 0f ca 83 c2 04 e8 fa c0 43 00 85 c0 0f 84 8a 00 00 00 <8b> 06 8b 00 85 c0 75 32 ba 04 00 00 00 8d 45 e4 e8 cd fe ff ff 
EIP: [<c03e4e1e>] nfs4_xdr_dec_cb_recall+0x4e/0x200 SS:ESP 0068:c5281e6c
---[ end trace 2724475d9856cb6c ]---
#203 lore raw
In reply to J. Bruce Fields
Quoted 15 lines · J. Bruce Fields
Quoted 14 lines · Benny Halevy
Quoted 13 lines · J. Bruce Fields
Quoted 8 lines · Benny Halevy

Hi Bruce,

Here's the latest server patches implementing the NFSv4.1 Sessions features.

This patchset is based over your for-2.6.30 branch and is also available from
git://linux-nfs.org/~bhalevy/linux-pnfs.git nfsd41-for-2.6.30

There's a few merge conflicts with my current for-2.6.30--probably my fault for not pushing that out recently enough, apologies. Would you mind updating?

Quoted 7 lines
Quoted 3 lines
Quoted 2 lines

--b.

Sure. Here's a rebased version with two minor changes (see below)

On the latest version of nfsd41-for-2.6.30 (a564667..) I'm getting a new NULL dereference in the callback code. Looks like it probably happened while running connectathon over NFSv4.0 with krb5p. That's all I've figured out so

Weird. nfs4_xdr_dec_cb_recall+0x4e doesn't seem like a valid IP for nfsd.ko @a564667. Bruce, can you please send me you .config file?

We know the current implementation is still incorrect for krb5, but I don't think it's supposed to crash either. Ricardo, have you run into this by any chance?

Benny

Quoted 50 lines

far.

--b.

BUG: unable to handle kernel NULL pointer dereference at (null)
IP: [<c03e4e1e>] nfs4_xdr_dec_cb_recall+0x4e/0x200
*pde = 00000000 
Oops: 0000 [#1] PREEMPT 
last sysfs file: /sys/kernel/uevent_seqnum
Modules linked in:

Pid: 3949, comm: nfs4_cb_recall Not tainted (2.6.29-rc8-00312-ga564667 #39)
EIP: 0060:[<c03e4e1e>] EFLAGS: 00010286 CPU: 0
EIP is at nfs4_xdr_dec_cb_recall+0x4e/0x200
EAX: c7bb22a0 EBX: c7bb2298 ECX: c7bb22a0 EDX: c7bb22a4
ESI: 00000000 EDI: c66e3000 EBP: c5281ea4 ESP: c5281e6c
DS: 007b ES: 007b FS: 0000 GS: 0000 SS: 0068
Process nfs4_cb_recall (pid: 3949, ti=c5280000 task=c66a6b60 task.ti=c5280000)

Stack:
 c66a6b60 00000000 c5281eb4 c086cab4 c0aa0380 00000001 c11bd4a4 c7bb22a4
 c66e302c c7bb22ac c66e302c 00000000 c6cf14e8 c66e3000 c5281ecc c08199c3
 c0819aae 00000000 c5281ee8 c5281ef0 c03e4dd0 c6cf14e8 c7bb2298 c03e4dd0
Call Trace:
 [<c086cab4>] ? schedule+0x354/0x540
 [<c08199c3>] ? rpcauth_unwrap_resp+0x63/0x90
 [<c0819aae>] ? rpcauth_checkverf+0x2e/0x70
 [<c03e4dd0>] ? nfs4_xdr_dec_cb_recall+0x0/0x200
 [<c03e4dd0>] ? nfs4_xdr_dec_cb_recall+0x0/0x200
 [<c08122ce>] ? call_decode+0x1ae/0x820
 [<c086d125>] ? out_of_line_wait_on_bit+0x65/0x80
 [<c0818720>] ? rpc_wait_bit_killable+0x0/0x40
 [<c03e4dd0>] ? nfs4_xdr_dec_cb_recall+0x0/0x200
 [<c0818e62>] ? __rpc_execute+0x92/0x290
 [<c086f4fc>] ? _spin_unlock+0x2c/0x50
 [<c08186c7>] ? rpc_set_active+0x67/0x80
 [<c081907e>] ? rpc_execute+0x1e/0x30
 [<c08129e5>] ? rpc_run_task+0x35/0x70
 [<c0812b40>] ? rpc_call_sync+0x40/0x70
 [<c03e51e0>] ? nfsd4_cb_recall+0x70/0x130
 [<c086cab4>] ? schedule+0x354/0x540
 [<c0247d1b>] ? trace_hardirqs_on+0xb/0x10
 [<c03dfd40>] ? do_recall+0x0/0x20
 [<c03dfd54>] ? do_recall+0x14/0x20
 [<c023941f>] ? kthread+0x3f/0x70
 [<c02393e0>] ? kthread+0x0/0x70
 [<c0203b67>] ? kernel_thread_helper+0x7/0x10
Code: 43 00 ba 08 00 00 00 8d 45 e4 e8 0e c1 43 00 85 c0 74 3a 8b 50 04 8d 45 e4 0f ca 83 c2 04 e8 fa c0 43 00 85 c0 0f 84 8a 00 00 00 <8b> 06 8b 00 85 c0 75 32 ba 04 00 00 00 8d 45 e4 e8 cd fe ff ff 
EIP: [<c03e4e1e>] nfs4_xdr_dec_cb_recall+0x4e/0x200 SS:ESP 0068:c5281e6c
---[ end trace 2724475d9856cb6c ]---