aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorTrond Myklebust <Trond.Myklebust@netapp.com>2008-07-15 18:34:16 -0400
committerTrond Myklebust <Trond.Myklebust@netapp.com>2008-07-15 18:34:16 -0400
commite89e896d31b11a51a54ddcd3f72a76bd286dd86c (patch)
tree105a9a93a4021caeeba76e479778ad2dd2b87b57
parent3a628b0fd42f7eaf9d052447784d48ceae9ffb8e (diff)
parentd67d1c7bf948341fd8678c8e337ec27f4b46b206 (diff)
Merge branch 'devel' into next
Conflicts: fs/nfs/file.c Fix up the conflict with Jon Corbet's bkl-removal tree
-rw-r--r--fs/Kconfig136
-rw-r--r--fs/lockd/clntproc.c2
-rw-r--r--fs/nfs/callback.c34
-rw-r--r--fs/nfs/client.c13
-rw-r--r--fs/nfs/dir.c26
-rw-r--r--fs/nfs/direct.c4
-rw-r--r--fs/nfs/file.c151
-rw-r--r--fs/nfs/inode.c8
-rw-r--r--fs/nfs/internal.h1
-rw-r--r--fs/nfs/iostat.h119
-rw-r--r--fs/nfs/nfs3acl.c9
-rw-r--r--fs/nfs/nfs3proc.c275
-rw-r--r--fs/nfs/nfs4proc.c263
-rw-r--r--fs/nfs/nfsroot.c10
-rw-r--r--fs/nfs/proc.c28
-rw-r--r--fs/nfs/super.c878
-rw-r--r--fs/nfs/write.c307
-rw-r--r--fs/nfsd/nfs4callback.c2
-rw-r--r--include/linux/inet.h7
-rw-r--r--include/linux/nfs_fs.h10
-rw-r--r--include/linux/nfs_iostat.h119
-rw-r--r--include/linux/nfs_page.h9
-rw-r--r--include/linux/nfs_xdr.h3
-rw-r--r--include/linux/sunrpc/clnt.h7
-rw-r--r--include/linux/sunrpc/sched.h1
-rw-r--r--net/sunrpc/auth_gss/auth_gss.c27
-rw-r--r--net/sunrpc/auth_gss/gss_krb5_mech.c4
-rw-r--r--net/sunrpc/auth_gss/gss_spkm3_mech.c4
-rw-r--r--net/sunrpc/auth_gss/gss_spkm3_token.c2
-rw-r--r--net/sunrpc/auth_unix.c2
-rw-r--r--net/sunrpc/clnt.c161
-rw-r--r--net/sunrpc/rpcb_clnt.c356
-rw-r--r--net/sunrpc/sched.c14
-rw-r--r--net/sunrpc/xprt.c9
-rw-r--r--net/sunrpc/xprtsock.c2
35 files changed, 1844 insertions, 1159 deletions
diff --git a/fs/Kconfig b/fs/Kconfig
index 313b2e06ded5..84ab76a206a0 100644
--- a/fs/Kconfig
+++ b/fs/Kconfig
@@ -1544,10 +1544,6 @@ config UFS_FS
1544 The recently released UFS2 variant (used in FreeBSD 5.x) is 1544 The recently released UFS2 variant (used in FreeBSD 5.x) is
1545 READ-ONLY supported. 1545 READ-ONLY supported.
1546 1546
1547 If you only intend to mount files from some other Unix over the
1548 network using NFS, you don't need the UFS file system support (but
1549 you need NFS file system support obviously).
1550
1551 Note that this option is generally not needed for floppies, since a 1547 Note that this option is generally not needed for floppies, since a
1552 good portable way to transport files and directories between unixes 1548 good portable way to transport files and directories between unixes
1553 (and even other operating systems) is given by the tar program ("man 1549 (and even other operating systems) is given by the tar program ("man
@@ -1587,6 +1583,7 @@ menuconfig NETWORK_FILESYSTEMS
1587 Say Y here to get to see options for network filesystems and 1583 Say Y here to get to see options for network filesystems and
1588 filesystem-related networking code, such as NFS daemon and 1584 filesystem-related networking code, such as NFS daemon and
1589 RPCSEC security modules. 1585 RPCSEC security modules.
1586
1590 This option alone does not add any kernel code. 1587 This option alone does not add any kernel code.
1591 1588
1592 If you say N, all options in this submenu will be skipped and 1589 If you say N, all options in this submenu will be skipped and
@@ -1595,76 +1592,92 @@ menuconfig NETWORK_FILESYSTEMS
1595if NETWORK_FILESYSTEMS 1592if NETWORK_FILESYSTEMS
1596 1593
1597config NFS_FS 1594config NFS_FS
1598 tristate "NFS file system support" 1595 tristate "NFS client support"
1599 depends on INET 1596 depends on INET
1600 select LOCKD 1597 select LOCKD
1601 select SUNRPC 1598 select SUNRPC
1602 select NFS_ACL_SUPPORT if NFS_V3_ACL 1599 select NFS_ACL_SUPPORT if NFS_V3_ACL
1603 help 1600 help
1604 If you are connected to some other (usually local) Unix computer 1601 Choose Y here if you want to access files residing on other
1605 (using SLIP, PLIP, PPP or Ethernet) and want to mount files residing 1602 computers using Sun's Network File System protocol. To compile
1606 on that computer (the NFS server) using the Network File Sharing 1603 this file system support as a module, choose M here: the module
1607 protocol, say Y. "Mounting files" means that the client can access 1604 will be called nfs.
1608 the files with usual UNIX commands as if they were sitting on the
1609 client's hard disk. For this to work, the server must run the
1610 programs nfsd and mountd (but does not need to have NFS file system
1611 support enabled in its kernel). NFS is explained in the Network
1612 Administrator's Guide, available from
1613 <http://www.tldp.org/docs.html#guide>, on its man page: "man
1614 nfs", and in the NFS-HOWTO.
1615
1616 A superior but less widely used alternative to NFS is provided by
1617 the Coda file system; see "Coda file system support" below.
1618 1605
1619 If you say Y here, you should have said Y to TCP/IP networking also. 1606 To mount file systems exported by NFS servers, you also need to
1620 This option would enlarge your kernel by about 27 KB. 1607 install the user space mount.nfs command which can be found in
1608 the Linux nfs-utils package, available from http://linux-nfs.org/.
1609 Information about using the mount command is available in the
1610 mount(8) man page. More detail about the Linux NFS client
1611 implementation is available via the nfs(5) man page.
1621 1612
1622 To compile this file system support as a module, choose M here: the 1613 Below you can choose which versions of the NFS protocol are
1623 module will be called nfs. 1614 available in the kernel to mount NFS servers. Support for NFS
1615 version 2 (RFC 1094) is always available when NFS_FS is selected.
1624 1616
1625 If you are configuring a diskless machine which will mount its root 1617 To configure a system which mounts its root file system via NFS
1626 file system over NFS at boot time, say Y here and to "Kernel 1618 at boot time, say Y here, select "Kernel level IP
1627 level IP autoconfiguration" above and to "Root file system on NFS" 1619 autoconfiguration" in the NETWORK menu, and select "Root file
1628 below. You cannot compile this driver as a module in this case. 1620 system on NFS" below. You cannot compile this file system as a
1629 There are two packages designed for booting diskless machines over 1621 module in this case.
1630 the net: netboot, available from
1631 <http://ftp1.sourceforge.net/netboot/>, and Etherboot,
1632 available from <http://ftp1.sourceforge.net/etherboot/>.
1633 1622
1634 If you don't know what all this is about, say N. 1623 If unsure, say N.
1635 1624
1636config NFS_V3 1625config NFS_V3
1637 bool "Provide NFSv3 client support" 1626 bool "NFS client support for NFS version 3"
1638 depends on NFS_FS 1627 depends on NFS_FS
1639 help 1628 help
1640 Say Y here if you want your NFS client to be able to speak version 1629 This option enables support for version 3 of the NFS protocol
1641 3 of the NFS protocol. 1630 (RFC 1813) in the kernel's NFS client.
1642 1631
1643 If unsure, say Y. 1632 If unsure, say Y.
1644 1633
1645config NFS_V3_ACL 1634config NFS_V3_ACL
1646 bool "Provide client support for the NFSv3 ACL protocol extension" 1635 bool "NFS client support for the NFSv3 ACL protocol extension"
1647 depends on NFS_V3 1636 depends on NFS_V3
1648 help 1637 help
1649 Implement the NFSv3 ACL protocol extension for manipulating POSIX 1638 Some NFS servers support an auxiliary NFSv3 ACL protocol that
1650 Access Control Lists. The server should also be compiled with 1639 Sun added to Solaris but never became an official part of the
1651 the NFSv3 ACL protocol extension; see the CONFIG_NFSD_V3_ACL option. 1640 NFS version 3 protocol. This protocol extension allows
1641 applications on NFS clients to manipulate POSIX Access Control
1642 Lists on files residing on NFS servers. NFS servers enforce
1643 ACLs on local files whether this protocol is available or not.
1644
1645 Choose Y here if your NFS server supports the Solaris NFSv3 ACL
1646 protocol extension and you want your NFS client to allow
1647 applications to access and modify ACLs on files on the server.
1648
1649 Most NFS servers don't support the Solaris NFSv3 ACL protocol
1650 extension. You can choose N here or specify the "noacl" mount
1651 option to prevent your NFS client from trying to use the NFSv3
1652 ACL protocol.
1652 1653
1653 If unsure, say N. 1654 If unsure, say N.
1654 1655
1655config NFS_V4 1656config NFS_V4
1656 bool "Provide NFSv4 client support (EXPERIMENTAL)" 1657 bool "NFS client support for NFS version 4 (EXPERIMENTAL)"
1657 depends on NFS_FS && EXPERIMENTAL 1658 depends on NFS_FS && EXPERIMENTAL
1658 select RPCSEC_GSS_KRB5 1659 select RPCSEC_GSS_KRB5
1659 help 1660 help
1660 Say Y here if you want your NFS client to be able to speak the newer 1661 This option enables support for version 4 of the NFS protocol
1661 version 4 of the NFS protocol. 1662 (RFC 3530) in the kernel's NFS client.
1662 1663
1663 Note: Requires auxiliary userspace daemons which may be found on 1664 To mount NFS servers using NFSv4, you also need to install user
1664 http://www.citi.umich.edu/projects/nfsv4/ 1665 space programs which can be found in the Linux nfs-utils package,
1666 available from http://linux-nfs.org/.
1665 1667
1666 If unsure, say N. 1668 If unsure, say N.
1667 1669
1670config ROOT_NFS
1671 bool "Root file system on NFS"
1672 depends on NFS_FS=y && IP_PNP
1673 help
1674 If you want your system to mount its root file system via NFS,
1675 choose Y here. This is common practice for managing systems
1676 without local permanent storage. For details, read
1677 <file:Documentation/filesystems/nfsroot.txt>.
1678
1679 Most people say N here.
1680
1668config NFSD 1681config NFSD
1669 tristate "NFS server support" 1682 tristate "NFS server support"
1670 depends on INET 1683 depends on INET
@@ -1746,20 +1759,6 @@ config NFSD_V4
1746 1759
1747 If unsure, say N. 1760 If unsure, say N.
1748 1761
1749config ROOT_NFS
1750 bool "Root file system on NFS"
1751 depends on NFS_FS=y && IP_PNP
1752 help
1753 If you want your Linux box to mount its whole root file system (the
1754 one containing the directory /) from some other computer over the
1755 net via NFS (presumably because your box doesn't have a hard disk),
1756 say Y. Read <file:Documentation/filesystems/nfsroot.txt> for
1757 details. It is likely that in this case, you also want to say Y to
1758 "Kernel level IP autoconfiguration" so that your box can discover
1759 its network address at boot time.
1760
1761 Most people say N here.
1762
1763config LOCKD 1762config LOCKD
1764 tristate 1763 tristate
1765 1764
@@ -1800,27 +1799,6 @@ config SUNRPC_XPRT_RDMA
1800 1799
1801 If unsure, say N. 1800 If unsure, say N.
1802 1801
1803config SUNRPC_BIND34
1804 bool "Support for rpcbind versions 3 & 4 (EXPERIMENTAL)"
1805 depends on SUNRPC && EXPERIMENTAL
1806 default n
1807 help
1808 RPC requests over IPv6 networks require support for larger
1809 addresses when performing an RPC bind. Sun added support for
1810 IPv6 addressing by creating two new versions of the rpcbind
1811 protocol (RFC 1833).
1812
1813 This option enables support in the kernel RPC client for
1814 querying rpcbind servers via versions 3 and 4 of the rpcbind
1815 protocol. The kernel automatically falls back to version 2
1816 if a remote rpcbind service does not support versions 3 or 4.
1817 By themselves, these new versions do not provide support for
1818 RPC over IPv6, but the new protocol versions are necessary to
1819 support it.
1820
1821 If unsure, say N to get traditional behavior (version 2 rpcbind
1822 requests only).
1823
1824config RPCSEC_GSS_KRB5 1802config RPCSEC_GSS_KRB5
1825 tristate "Secure RPC: Kerberos V mechanism (EXPERIMENTAL)" 1803 tristate "Secure RPC: Kerberos V mechanism (EXPERIMENTAL)"
1826 depends on SUNRPC && EXPERIMENTAL 1804 depends on SUNRPC && EXPERIMENTAL
diff --git a/fs/lockd/clntproc.c b/fs/lockd/clntproc.c
index 5df517b81f3f..fd7d4669776e 100644
--- a/fs/lockd/clntproc.c
+++ b/fs/lockd/clntproc.c
@@ -430,7 +430,7 @@ nlmclnt_test(struct nlm_rqst *req, struct file_lock *fl)
430 * Report the conflicting lock back to the application. 430 * Report the conflicting lock back to the application.
431 */ 431 */
432 fl->fl_start = req->a_res.lock.fl.fl_start; 432 fl->fl_start = req->a_res.lock.fl.fl_start;
433 fl->fl_end = req->a_res.lock.fl.fl_start; 433 fl->fl_end = req->a_res.lock.fl.fl_end;
434 fl->fl_type = req->a_res.lock.fl.fl_type; 434 fl->fl_type = req->a_res.lock.fl.fl_type;
435 fl->fl_pid = 0; 435 fl->fl_pid = 0;
436 break; 436 break;
diff --git a/fs/nfs/callback.c b/fs/nfs/callback.c
index c1e7c8300629..f447f4b4476c 100644
--- a/fs/nfs/callback.c
+++ b/fs/nfs/callback.c
@@ -27,7 +27,7 @@
27 27
28struct nfs_callback_data { 28struct nfs_callback_data {
29 unsigned int users; 29 unsigned int users;
30 struct svc_serv *serv; 30 struct svc_rqst *rqst;
31 struct task_struct *task; 31 struct task_struct *task;
32}; 32};
33 33
@@ -91,21 +91,17 @@ nfs_callback_svc(void *vrqstp)
91 svc_process(rqstp); 91 svc_process(rqstp);
92 } 92 }
93 unlock_kernel(); 93 unlock_kernel();
94 nfs_callback_info.task = NULL;
95 svc_exit_thread(rqstp);
96 return 0; 94 return 0;
97} 95}
98 96
99/* 97/*
100 * Bring up the server process if it is not already up. 98 * Bring up the callback thread if it is not already up.
101 */ 99 */
102int nfs_callback_up(void) 100int nfs_callback_up(void)
103{ 101{
104 struct svc_serv *serv = NULL; 102 struct svc_serv *serv = NULL;
105 struct svc_rqst *rqstp;
106 int ret = 0; 103 int ret = 0;
107 104
108 lock_kernel();
109 mutex_lock(&nfs_callback_mutex); 105 mutex_lock(&nfs_callback_mutex);
110 if (nfs_callback_info.users++ || nfs_callback_info.task != NULL) 106 if (nfs_callback_info.users++ || nfs_callback_info.task != NULL)
111 goto out; 107 goto out;
@@ -121,22 +117,23 @@ int nfs_callback_up(void)
121 nfs_callback_tcpport = ret; 117 nfs_callback_tcpport = ret;
122 dprintk("Callback port = 0x%x\n", nfs_callback_tcpport); 118 dprintk("Callback port = 0x%x\n", nfs_callback_tcpport);
123 119
124 rqstp = svc_prepare_thread(serv, &serv->sv_pools[0]); 120 nfs_callback_info.rqst = svc_prepare_thread(serv, &serv->sv_pools[0]);
125 if (IS_ERR(rqstp)) { 121 if (IS_ERR(nfs_callback_info.rqst)) {
126 ret = PTR_ERR(rqstp); 122 ret = PTR_ERR(nfs_callback_info.rqst);
123 nfs_callback_info.rqst = NULL;
127 goto out_err; 124 goto out_err;
128 } 125 }
129 126
130 svc_sock_update_bufs(serv); 127 svc_sock_update_bufs(serv);
131 nfs_callback_info.serv = serv;
132 128
133 nfs_callback_info.task = kthread_run(nfs_callback_svc, rqstp, 129 nfs_callback_info.task = kthread_run(nfs_callback_svc,
130 nfs_callback_info.rqst,
134 "nfsv4-svc"); 131 "nfsv4-svc");
135 if (IS_ERR(nfs_callback_info.task)) { 132 if (IS_ERR(nfs_callback_info.task)) {
136 ret = PTR_ERR(nfs_callback_info.task); 133 ret = PTR_ERR(nfs_callback_info.task);
137 nfs_callback_info.serv = NULL; 134 svc_exit_thread(nfs_callback_info.rqst);
135 nfs_callback_info.rqst = NULL;
138 nfs_callback_info.task = NULL; 136 nfs_callback_info.task = NULL;
139 svc_exit_thread(rqstp);
140 goto out_err; 137 goto out_err;
141 } 138 }
142out: 139out:
@@ -149,7 +146,6 @@ out:
149 if (serv) 146 if (serv)
150 svc_destroy(serv); 147 svc_destroy(serv);
151 mutex_unlock(&nfs_callback_mutex); 148 mutex_unlock(&nfs_callback_mutex);
152 unlock_kernel();
153 return ret; 149 return ret;
154out_err: 150out_err:
155 dprintk("Couldn't create callback socket or server thread; err = %d\n", 151 dprintk("Couldn't create callback socket or server thread; err = %d\n",
@@ -159,17 +155,19 @@ out_err:
159} 155}
160 156
161/* 157/*
162 * Kill the server process if it is not already down. 158 * Kill the callback thread if it's no longer being used.
163 */ 159 */
164void nfs_callback_down(void) 160void nfs_callback_down(void)
165{ 161{
166 lock_kernel();
167 mutex_lock(&nfs_callback_mutex); 162 mutex_lock(&nfs_callback_mutex);
168 nfs_callback_info.users--; 163 nfs_callback_info.users--;
169 if (nfs_callback_info.users == 0 && nfs_callback_info.task != NULL) 164 if (nfs_callback_info.users == 0 && nfs_callback_info.task != NULL) {
170 kthread_stop(nfs_callback_info.task); 165 kthread_stop(nfs_callback_info.task);
166 svc_exit_thread(nfs_callback_info.rqst);
167 nfs_callback_info.rqst = NULL;
168 nfs_callback_info.task = NULL;
169 }
171 mutex_unlock(&nfs_callback_mutex); 170 mutex_unlock(&nfs_callback_mutex);
172 unlock_kernel();
173} 171}
174 172
175static int nfs_callback_authenticate(struct svc_rqst *rqstp) 173static int nfs_callback_authenticate(struct svc_rqst *rqstp)
diff --git a/fs/nfs/client.c b/fs/nfs/client.c
index f2a092ca69b5..5ee23e7058b3 100644
--- a/fs/nfs/client.c
+++ b/fs/nfs/client.c
@@ -431,14 +431,14 @@ static void nfs_init_timeout_values(struct rpc_timeout *to, int proto,
431{ 431{
432 to->to_initval = timeo * HZ / 10; 432 to->to_initval = timeo * HZ / 10;
433 to->to_retries = retrans; 433 to->to_retries = retrans;
434 if (!to->to_retries)
435 to->to_retries = 2;
436 434
437 switch (proto) { 435 switch (proto) {
438 case XPRT_TRANSPORT_TCP: 436 case XPRT_TRANSPORT_TCP:
439 case XPRT_TRANSPORT_RDMA: 437 case XPRT_TRANSPORT_RDMA:
438 if (to->to_retries == 0)
439 to->to_retries = NFS_DEF_TCP_RETRANS;
440 if (to->to_initval == 0) 440 if (to->to_initval == 0)
441 to->to_initval = 60 * HZ; 441 to->to_initval = NFS_DEF_TCP_TIMEO * HZ / 10;
442 if (to->to_initval > NFS_MAX_TCP_TIMEOUT) 442 if (to->to_initval > NFS_MAX_TCP_TIMEOUT)
443 to->to_initval = NFS_MAX_TCP_TIMEOUT; 443 to->to_initval = NFS_MAX_TCP_TIMEOUT;
444 to->to_increment = to->to_initval; 444 to->to_increment = to->to_initval;
@@ -450,14 +450,17 @@ static void nfs_init_timeout_values(struct rpc_timeout *to, int proto,
450 to->to_exponential = 0; 450 to->to_exponential = 0;
451 break; 451 break;
452 case XPRT_TRANSPORT_UDP: 452 case XPRT_TRANSPORT_UDP:
453 default: 453 if (to->to_retries == 0)
454 to->to_retries = NFS_DEF_UDP_RETRANS;
454 if (!to->to_initval) 455 if (!to->to_initval)
455 to->to_initval = 11 * HZ / 10; 456 to->to_initval = NFS_DEF_UDP_TIMEO * HZ / 10;
456 if (to->to_initval > NFS_MAX_UDP_TIMEOUT) 457 if (to->to_initval > NFS_MAX_UDP_TIMEOUT)
457 to->to_initval = NFS_MAX_UDP_TIMEOUT; 458 to->to_initval = NFS_MAX_UDP_TIMEOUT;
458 to->to_maxval = NFS_MAX_UDP_TIMEOUT; 459 to->to_maxval = NFS_MAX_UDP_TIMEOUT;
459 to->to_exponential = 1; 460 to->to_exponential = 1;
460 break; 461 break;
462 default:
463 BUG();
461 } 464 }
462} 465}
463 466
diff --git a/fs/nfs/dir.c b/fs/nfs/dir.c
index 982a2064fe4c..b1940660502f 100644
--- a/fs/nfs/dir.c
+++ b/fs/nfs/dir.c
@@ -133,8 +133,11 @@ nfs_opendir(struct inode *inode, struct file *filp)
133{ 133{
134 int res; 134 int res;
135 135
136 dfprintk(VFS, "NFS: opendir(%s/%ld)\n", 136 dfprintk(FILE, "NFS: open dir(%s/%s)\n",
137 inode->i_sb->s_id, inode->i_ino); 137 filp->f_path.dentry->d_parent->d_name.name,
138 filp->f_path.dentry->d_name.name);
139
140 nfs_inc_stats(inode, NFSIOS_VFSOPEN);
138 141
139 lock_kernel(); 142 lock_kernel();
140 /* Call generic open code in order to cache credentials */ 143 /* Call generic open code in order to cache credentials */
@@ -528,7 +531,7 @@ static int nfs_readdir(struct file *filp, void *dirent, filldir_t filldir)
528 struct nfs_fattr fattr; 531 struct nfs_fattr fattr;
529 long res; 532 long res;
530 533
531 dfprintk(VFS, "NFS: readdir(%s/%s) starting at cookie %Lu\n", 534 dfprintk(FILE, "NFS: readdir(%s/%s) starting at cookie %llu\n",
532 dentry->d_parent->d_name.name, dentry->d_name.name, 535 dentry->d_parent->d_name.name, dentry->d_name.name,
533 (long long)filp->f_pos); 536 (long long)filp->f_pos);
534 nfs_inc_stats(inode, NFSIOS_VFSGETDENTS); 537 nfs_inc_stats(inode, NFSIOS_VFSGETDENTS);
@@ -595,7 +598,7 @@ out:
595 unlock_kernel(); 598 unlock_kernel();
596 if (res > 0) 599 if (res > 0)
597 res = 0; 600 res = 0;
598 dfprintk(VFS, "NFS: readdir(%s/%s) returns %ld\n", 601 dfprintk(FILE, "NFS: readdir(%s/%s) returns %ld\n",
599 dentry->d_parent->d_name.name, dentry->d_name.name, 602 dentry->d_parent->d_name.name, dentry->d_name.name,
600 res); 603 res);
601 return res; 604 return res;
@@ -603,7 +606,15 @@ out:
603 606
604static loff_t nfs_llseek_dir(struct file *filp, loff_t offset, int origin) 607static loff_t nfs_llseek_dir(struct file *filp, loff_t offset, int origin)
605{ 608{
606 mutex_lock(&filp->f_path.dentry->d_inode->i_mutex); 609 struct dentry *dentry = filp->f_path.dentry;
610 struct inode *inode = dentry->d_inode;
611
612 dfprintk(FILE, "NFS: llseek dir(%s/%s, %lld, %d)\n",
613 dentry->d_parent->d_name.name,
614 dentry->d_name.name,
615 offset, origin);
616
617 mutex_lock(&inode->i_mutex);
607 switch (origin) { 618 switch (origin) {
608 case 1: 619 case 1:
609 offset += filp->f_pos; 620 offset += filp->f_pos;
@@ -619,7 +630,7 @@ static loff_t nfs_llseek_dir(struct file *filp, loff_t offset, int origin)
619 nfs_file_open_context(filp)->dir_cookie = 0; 630 nfs_file_open_context(filp)->dir_cookie = 0;
620 } 631 }
621out: 632out:
622 mutex_unlock(&filp->f_path.dentry->d_inode->i_mutex); 633 mutex_unlock(&inode->i_mutex);
623 return offset; 634 return offset;
624} 635}
625 636
@@ -629,10 +640,11 @@ out:
629 */ 640 */
630static int nfs_fsync_dir(struct file *filp, struct dentry *dentry, int datasync) 641static int nfs_fsync_dir(struct file *filp, struct dentry *dentry, int datasync)
631{ 642{
632 dfprintk(VFS, "NFS: fsync_dir(%s/%s) datasync %d\n", 643 dfprintk(FILE, "NFS: fsync dir(%s/%s) datasync %d\n",
633 dentry->d_parent->d_name.name, dentry->d_name.name, 644 dentry->d_parent->d_name.name, dentry->d_name.name,
634 datasync); 645 datasync);
635 646
647 nfs_inc_stats(dentry->d_inode, NFSIOS_VFSFSYNC);
636 return 0; 648 return 0;
637} 649}
638 650
diff --git a/fs/nfs/direct.c b/fs/nfs/direct.c
index 4757a2b326a1..08f6b040d289 100644
--- a/fs/nfs/direct.c
+++ b/fs/nfs/direct.c
@@ -890,7 +890,7 @@ ssize_t nfs_file_direct_read(struct kiocb *iocb, const struct iovec *iov,
890 count = iov_length(iov, nr_segs); 890 count = iov_length(iov, nr_segs);
891 nfs_add_stats(mapping->host, NFSIOS_DIRECTREADBYTES, count); 891 nfs_add_stats(mapping->host, NFSIOS_DIRECTREADBYTES, count);
892 892
893 dprintk("nfs: direct read(%s/%s, %zd@%Ld)\n", 893 dfprintk(FILE, "NFS: direct read(%s/%s, %zd@%Ld)\n",
894 file->f_path.dentry->d_parent->d_name.name, 894 file->f_path.dentry->d_parent->d_name.name,
895 file->f_path.dentry->d_name.name, 895 file->f_path.dentry->d_name.name,
896 count, (long long) pos); 896 count, (long long) pos);
@@ -947,7 +947,7 @@ ssize_t nfs_file_direct_write(struct kiocb *iocb, const struct iovec *iov,
947 count = iov_length(iov, nr_segs); 947 count = iov_length(iov, nr_segs);
948 nfs_add_stats(mapping->host, NFSIOS_DIRECTWRITTENBYTES, count); 948 nfs_add_stats(mapping->host, NFSIOS_DIRECTWRITTENBYTES, count);
949 949
950 dfprintk(VFS, "nfs: direct write(%s/%s, %zd@%Ld)\n", 950 dfprintk(FILE, "NFS: direct write(%s/%s, %zd@%Ld)\n",
951 file->f_path.dentry->d_parent->d_name.name, 951 file->f_path.dentry->d_parent->d_name.name,
952 file->f_path.dentry->d_name.name, 952 file->f_path.dentry->d_name.name,
953 count, (long long) pos); 953 count, (long long) pos);
diff --git a/fs/nfs/file.c b/fs/nfs/file.c
index 4e98a56a1777..43164fe86069 100644
--- a/fs/nfs/file.c
+++ b/fs/nfs/file.c
@@ -50,7 +50,7 @@ static ssize_t nfs_file_read(struct kiocb *, const struct iovec *iov,
50static ssize_t nfs_file_write(struct kiocb *, const struct iovec *iov, 50static ssize_t nfs_file_write(struct kiocb *, const struct iovec *iov,
51 unsigned long nr_segs, loff_t pos); 51 unsigned long nr_segs, loff_t pos);
52static int nfs_file_flush(struct file *, fl_owner_t id); 52static int nfs_file_flush(struct file *, fl_owner_t id);
53static int nfs_fsync(struct file *, struct dentry *dentry, int datasync); 53static int nfs_file_fsync(struct file *, struct dentry *dentry, int datasync);
54static int nfs_check_flags(int flags); 54static int nfs_check_flags(int flags);
55static int nfs_lock(struct file *filp, int cmd, struct file_lock *fl); 55static int nfs_lock(struct file *filp, int cmd, struct file_lock *fl);
56static int nfs_flock(struct file *filp, int cmd, struct file_lock *fl); 56static int nfs_flock(struct file *filp, int cmd, struct file_lock *fl);
@@ -72,7 +72,7 @@ const struct file_operations nfs_file_operations = {
72 .open = nfs_file_open, 72 .open = nfs_file_open,
73 .flush = nfs_file_flush, 73 .flush = nfs_file_flush,
74 .release = nfs_file_release, 74 .release = nfs_file_release,
75 .fsync = nfs_fsync, 75 .fsync = nfs_file_fsync,
76 .lock = nfs_lock, 76 .lock = nfs_lock,
77 .flock = nfs_flock, 77 .flock = nfs_flock,
78 .splice_read = nfs_file_splice_read, 78 .splice_read = nfs_file_splice_read,
@@ -119,13 +119,17 @@ nfs_file_open(struct inode *inode, struct file *filp)
119{ 119{
120 int res; 120 int res;
121 121
122 dprintk("NFS: open file(%s/%s)\n",
123 filp->f_path.dentry->d_parent->d_name.name,
124 filp->f_path.dentry->d_name.name);
125
122 res = nfs_check_flags(filp->f_flags); 126 res = nfs_check_flags(filp->f_flags);
123 if (res) 127 if (res)
124 return res; 128 return res;
125 129
126 nfs_inc_stats(inode, NFSIOS_VFSOPEN); 130 nfs_inc_stats(inode, NFSIOS_VFSOPEN);
127 lock_kernel(); 131 lock_kernel();
128 res = NFS_PROTO(inode)->file_open(inode, filp); 132 res = nfs_open(inode, filp);
129 unlock_kernel(); 133 unlock_kernel();
130 return res; 134 return res;
131} 135}
@@ -133,11 +137,17 @@ nfs_file_open(struct inode *inode, struct file *filp)
133static int 137static int
134nfs_file_release(struct inode *inode, struct file *filp) 138nfs_file_release(struct inode *inode, struct file *filp)
135{ 139{
140 struct dentry *dentry = filp->f_path.dentry;
141
142 dprintk("NFS: release(%s/%s)\n",
143 dentry->d_parent->d_name.name,
144 dentry->d_name.name);
145
136 /* Ensure that dirty pages are flushed out with the right creds */ 146 /* Ensure that dirty pages are flushed out with the right creds */
137 if (filp->f_mode & FMODE_WRITE) 147 if (filp->f_mode & FMODE_WRITE)
138 nfs_wb_all(filp->f_path.dentry->d_inode); 148 nfs_wb_all(dentry->d_inode);
139 nfs_inc_stats(inode, NFSIOS_VFSRELEASE); 149 nfs_inc_stats(inode, NFSIOS_VFSRELEASE);
140 return NFS_PROTO(inode)->file_release(inode, filp); 150 return nfs_release(inode, filp);
141} 151}
142 152
143/** 153/**
@@ -171,6 +181,12 @@ force_reval:
171static loff_t nfs_file_llseek(struct file *filp, loff_t offset, int origin) 181static loff_t nfs_file_llseek(struct file *filp, loff_t offset, int origin)
172{ 182{
173 loff_t loff; 183 loff_t loff;
184
185 dprintk("NFS: llseek file(%s/%s, %lld, %d)\n",
186 filp->f_path.dentry->d_parent->d_name.name,
187 filp->f_path.dentry->d_name.name,
188 offset, origin);
189
174 /* origin == SEEK_END => we must revalidate the cached file length */ 190 /* origin == SEEK_END => we must revalidate the cached file length */
175 if (origin == SEEK_END) { 191 if (origin == SEEK_END) {
176 struct inode *inode = filp->f_mapping->host; 192 struct inode *inode = filp->f_mapping->host;
@@ -185,7 +201,7 @@ static loff_t nfs_file_llseek(struct file *filp, loff_t offset, int origin)
185} 201}
186 202
187/* 203/*
188 * Helper for nfs_file_flush() and nfs_fsync() 204 * Helper for nfs_file_flush() and nfs_file_fsync()
189 * 205 *
190 * Notice that it clears the NFS_CONTEXT_ERROR_WRITE before synching to 206 * Notice that it clears the NFS_CONTEXT_ERROR_WRITE before synching to
191 * disk, but it retrieves and clears ctx->error after synching, despite 207 * disk, but it retrieves and clears ctx->error after synching, despite
@@ -211,16 +227,18 @@ static int nfs_do_fsync(struct nfs_open_context *ctx, struct inode *inode)
211 227
212/* 228/*
213 * Flush all dirty pages, and check for write errors. 229 * Flush all dirty pages, and check for write errors.
214 *
215 */ 230 */
216static int 231static int
217nfs_file_flush(struct file *file, fl_owner_t id) 232nfs_file_flush(struct file *file, fl_owner_t id)
218{ 233{
219 struct nfs_open_context *ctx = nfs_file_open_context(file); 234 struct nfs_open_context *ctx = nfs_file_open_context(file);
220 struct inode *inode = file->f_path.dentry->d_inode; 235 struct dentry *dentry = file->f_path.dentry;
236 struct inode *inode = dentry->d_inode;
221 int status; 237 int status;
222 238
223 dfprintk(VFS, "nfs: flush(%s/%ld)\n", inode->i_sb->s_id, inode->i_ino); 239 dprintk("NFS: flush(%s/%s)\n",
240 dentry->d_parent->d_name.name,
241 dentry->d_name.name);
224 242
225 if ((file->f_mode & FMODE_WRITE) == 0) 243 if ((file->f_mode & FMODE_WRITE) == 0)
226 return 0; 244 return 0;
@@ -245,7 +263,7 @@ nfs_file_read(struct kiocb *iocb, const struct iovec *iov,
245 if (iocb->ki_filp->f_flags & O_DIRECT) 263 if (iocb->ki_filp->f_flags & O_DIRECT)
246 return nfs_file_direct_read(iocb, iov, nr_segs, pos); 264 return nfs_file_direct_read(iocb, iov, nr_segs, pos);
247 265
248 dfprintk(VFS, "nfs: read(%s/%s, %lu@%lu)\n", 266 dprintk("NFS: read(%s/%s, %lu@%lu)\n",
249 dentry->d_parent->d_name.name, dentry->d_name.name, 267 dentry->d_parent->d_name.name, dentry->d_name.name,
250 (unsigned long) count, (unsigned long) pos); 268 (unsigned long) count, (unsigned long) pos);
251 269
@@ -265,7 +283,7 @@ nfs_file_splice_read(struct file *filp, loff_t *ppos,
265 struct inode *inode = dentry->d_inode; 283 struct inode *inode = dentry->d_inode;
266 ssize_t res; 284 ssize_t res;
267 285
268 dfprintk(VFS, "nfs: splice_read(%s/%s, %lu@%Lu)\n", 286 dprintk("NFS: splice_read(%s/%s, %lu@%Lu)\n",
269 dentry->d_parent->d_name.name, dentry->d_name.name, 287 dentry->d_parent->d_name.name, dentry->d_name.name,
270 (unsigned long) count, (unsigned long long) *ppos); 288 (unsigned long) count, (unsigned long long) *ppos);
271 289
@@ -282,7 +300,7 @@ nfs_file_mmap(struct file * file, struct vm_area_struct * vma)
282 struct inode *inode = dentry->d_inode; 300 struct inode *inode = dentry->d_inode;
283 int status; 301 int status;
284 302
285 dfprintk(VFS, "nfs: mmap(%s/%s)\n", 303 dprintk("NFS: mmap(%s/%s)\n",
286 dentry->d_parent->d_name.name, dentry->d_name.name); 304 dentry->d_parent->d_name.name, dentry->d_name.name);
287 305
288 status = nfs_revalidate_mapping(inode, file->f_mapping); 306 status = nfs_revalidate_mapping(inode, file->f_mapping);
@@ -300,12 +318,14 @@ nfs_file_mmap(struct file * file, struct vm_area_struct * vma)
300 * whether any write errors occurred for this process. 318 * whether any write errors occurred for this process.
301 */ 319 */
302static int 320static int
303nfs_fsync(struct file *file, struct dentry *dentry, int datasync) 321nfs_file_fsync(struct file *file, struct dentry *dentry, int datasync)
304{ 322{
305 struct nfs_open_context *ctx = nfs_file_open_context(file); 323 struct nfs_open_context *ctx = nfs_file_open_context(file);
306 struct inode *inode = dentry->d_inode; 324 struct inode *inode = dentry->d_inode;
307 325
308 dfprintk(VFS, "nfs: fsync(%s/%ld)\n", inode->i_sb->s_id, inode->i_ino); 326 dprintk("NFS: fsync file(%s/%s) datasync %d\n",
327 dentry->d_parent->d_name.name, dentry->d_name.name,
328 datasync);
309 329
310 nfs_inc_stats(inode, NFSIOS_VFSFSYNC); 330 nfs_inc_stats(inode, NFSIOS_VFSFSYNC);
311 return nfs_do_fsync(ctx, inode); 331 return nfs_do_fsync(ctx, inode);
@@ -328,6 +348,11 @@ static int nfs_write_begin(struct file *file, struct address_space *mapping,
328 struct page *page; 348 struct page *page;
329 index = pos >> PAGE_CACHE_SHIFT; 349 index = pos >> PAGE_CACHE_SHIFT;
330 350
351 dfprintk(PAGECACHE, "NFS: write_begin(%s/%s(%ld), %u@%lld)\n",
352 file->f_path.dentry->d_parent->d_name.name,
353 file->f_path.dentry->d_name.name,
354 mapping->host->i_ino, len, (long long) pos);
355
331 page = __grab_cache_page(mapping, index); 356 page = __grab_cache_page(mapping, index);
332 if (!page) 357 if (!page)
333 return -ENOMEM; 358 return -ENOMEM;
@@ -348,6 +373,31 @@ static int nfs_write_end(struct file *file, struct address_space *mapping,
348 unsigned offset = pos & (PAGE_CACHE_SIZE - 1); 373 unsigned offset = pos & (PAGE_CACHE_SIZE - 1);
349 int status; 374 int status;
350 375
376 dfprintk(PAGECACHE, "NFS: write_end(%s/%s(%ld), %u@%lld)\n",
377 file->f_path.dentry->d_parent->d_name.name,
378 file->f_path.dentry->d_name.name,
379 mapping->host->i_ino, len, (long long) pos);
380
381 /*
382 * Zero any uninitialised parts of the page, and then mark the page
383 * as up to date if it turns out that we're extending the file.
384 */
385 if (!PageUptodate(page)) {
386 unsigned pglen = nfs_page_length(page);
387 unsigned end = offset + len;
388
389 if (pglen == 0) {
390 zero_user_segments(page, 0, offset,
391 end, PAGE_CACHE_SIZE);
392 SetPageUptodate(page);
393 } else if (end >= pglen) {
394 zero_user_segment(page, end, PAGE_CACHE_SIZE);
395 if (offset == 0)
396 SetPageUptodate(page);
397 } else
398 zero_user_segment(page, pglen, PAGE_CACHE_SIZE);
399 }
400
351 lock_kernel(); 401 lock_kernel();
352 status = nfs_updatepage(file, page, offset, copied); 402 status = nfs_updatepage(file, page, offset, copied);
353 unlock_kernel(); 403 unlock_kernel();
@@ -362,6 +412,8 @@ static int nfs_write_end(struct file *file, struct address_space *mapping,
362 412
363static void nfs_invalidate_page(struct page *page, unsigned long offset) 413static void nfs_invalidate_page(struct page *page, unsigned long offset)
364{ 414{
415 dfprintk(PAGECACHE, "NFS: invalidate_page(%p, %lu)\n", page, offset);
416
365 if (offset != 0) 417 if (offset != 0)
366 return; 418 return;
367 /* Cancel any unstarted writes on this page */ 419 /* Cancel any unstarted writes on this page */
@@ -370,13 +422,20 @@ static void nfs_invalidate_page(struct page *page, unsigned long offset)
370 422
371static int nfs_release_page(struct page *page, gfp_t gfp) 423static int nfs_release_page(struct page *page, gfp_t gfp)
372{ 424{
425 dfprintk(PAGECACHE, "NFS: release_page(%p)\n", page);
426
373 /* If PagePrivate() is set, then the page is not freeable */ 427 /* If PagePrivate() is set, then the page is not freeable */
374 return 0; 428 return 0;
375} 429}
376 430
377static int nfs_launder_page(struct page *page) 431static int nfs_launder_page(struct page *page)
378{ 432{
379 return nfs_wb_page(page->mapping->host, page); 433 struct inode *inode = page->mapping->host;
434
435 dfprintk(PAGECACHE, "NFS: launder_page(%ld, %llu)\n",
436 inode->i_ino, (long long)page_offset(page));
437
438 return nfs_wb_page(inode, page);
380} 439}
381 440
382const struct address_space_operations nfs_file_aops = { 441const struct address_space_operations nfs_file_aops = {
@@ -396,13 +455,19 @@ const struct address_space_operations nfs_file_aops = {
396static int nfs_vm_page_mkwrite(struct vm_area_struct *vma, struct page *page) 455static int nfs_vm_page_mkwrite(struct vm_area_struct *vma, struct page *page)
397{ 456{
398 struct file *filp = vma->vm_file; 457 struct file *filp = vma->vm_file;
458 struct dentry *dentry = filp->f_path.dentry;
399 unsigned pagelen; 459 unsigned pagelen;
400 int ret = -EINVAL; 460 int ret = -EINVAL;
401 struct address_space *mapping; 461 struct address_space *mapping;
402 462
463 dfprintk(PAGECACHE, "NFS: vm_page_mkwrite(%s/%s(%ld), offset %lld)\n",
464 dentry->d_parent->d_name.name, dentry->d_name.name,
465 filp->f_mapping->host->i_ino,
466 (long long)page_offset(page));
467
403 lock_page(page); 468 lock_page(page);
404 mapping = page->mapping; 469 mapping = page->mapping;
405 if (mapping != vma->vm_file->f_path.dentry->d_inode->i_mapping) 470 if (mapping != dentry->d_inode->i_mapping)
406 goto out_unlock; 471 goto out_unlock;
407 472
408 ret = 0; 473 ret = 0;
@@ -450,9 +515,9 @@ static ssize_t nfs_file_write(struct kiocb *iocb, const struct iovec *iov,
450 if (iocb->ki_filp->f_flags & O_DIRECT) 515 if (iocb->ki_filp->f_flags & O_DIRECT)
451 return nfs_file_direct_write(iocb, iov, nr_segs, pos); 516 return nfs_file_direct_write(iocb, iov, nr_segs, pos);
452 517
453 dfprintk(VFS, "nfs: write(%s/%s(%ld), %lu@%Ld)\n", 518 dprintk("NFS: write(%s/%s, %lu@%Ld)\n",
454 dentry->d_parent->d_name.name, dentry->d_name.name, 519 dentry->d_parent->d_name.name, dentry->d_name.name,
455 inode->i_ino, (unsigned long) count, (long long) pos); 520 (unsigned long) count, (long long) pos);
456 521
457 result = -EBUSY; 522 result = -EBUSY;
458 if (IS_SWAPFILE(inode)) 523 if (IS_SWAPFILE(inode))
@@ -586,7 +651,8 @@ static int do_setlk(struct file *filp, int cmd, struct file_lock *fl)
586 * This makes locking act as a cache coherency point. 651 * This makes locking act as a cache coherency point.
587 */ 652 */
588 nfs_sync_mapping(filp->f_mapping); 653 nfs_sync_mapping(filp->f_mapping);
589 nfs_zap_caches(inode); 654 if (!nfs_have_delegation(inode, FMODE_READ))
655 nfs_zap_caches(inode);
590out: 656out:
591 return status; 657 return status;
592} 658}
@@ -596,23 +662,35 @@ out:
596 */ 662 */
597static int nfs_lock(struct file *filp, int cmd, struct file_lock *fl) 663static int nfs_lock(struct file *filp, int cmd, struct file_lock *fl)
598{ 664{
599 struct inode * inode = filp->f_mapping->host; 665 struct inode *inode = filp->f_mapping->host;
666 int ret = -ENOLCK;
600 667
601 dprintk("NFS: nfs_lock(f=%s/%ld, t=%x, fl=%x, r=%Ld:%Ld)\n", 668 dprintk("NFS: lock(%s/%s, t=%x, fl=%x, r=%lld:%lld)\n",
602 inode->i_sb->s_id, inode->i_ino, 669 filp->f_path.dentry->d_parent->d_name.name,
670 filp->f_path.dentry->d_name.name,
603 fl->fl_type, fl->fl_flags, 671 fl->fl_type, fl->fl_flags,
604 (long long)fl->fl_start, (long long)fl->fl_end); 672 (long long)fl->fl_start, (long long)fl->fl_end);
673
605 nfs_inc_stats(inode, NFSIOS_VFSLOCK); 674 nfs_inc_stats(inode, NFSIOS_VFSLOCK);
606 675
607 /* No mandatory locks over NFS */ 676 /* No mandatory locks over NFS */
608 if (__mandatory_lock(inode) && fl->fl_type != F_UNLCK) 677 if (__mandatory_lock(inode) && fl->fl_type != F_UNLCK)
609 return -ENOLCK; 678 goto out_err;
679
680 if (NFS_PROTO(inode)->lock_check_bounds != NULL) {
681 ret = NFS_PROTO(inode)->lock_check_bounds(fl);
682 if (ret < 0)
683 goto out_err;
684 }
610 685
611 if (IS_GETLK(cmd)) 686 if (IS_GETLK(cmd))
612 return do_getlk(filp, cmd, fl); 687 ret = do_getlk(filp, cmd, fl);
613 if (fl->fl_type == F_UNLCK) 688 else if (fl->fl_type == F_UNLCK)
614 return do_unlk(filp, cmd, fl); 689 ret = do_unlk(filp, cmd, fl);
615 return do_setlk(filp, cmd, fl); 690 else
691 ret = do_setlk(filp, cmd, fl);
692out_err:
693 return ret;
616} 694}
617 695
618/* 696/*
@@ -620,9 +698,9 @@ static int nfs_lock(struct file *filp, int cmd, struct file_lock *fl)
620 */ 698 */
621static int nfs_flock(struct file *filp, int cmd, struct file_lock *fl) 699static int nfs_flock(struct file *filp, int cmd, struct file_lock *fl)
622{ 700{
623 dprintk("NFS: nfs_flock(f=%s/%ld, t=%x, fl=%x)\n", 701 dprintk("NFS: flock(%s/%s, t=%x, fl=%x)\n",
624 filp->f_path.dentry->d_inode->i_sb->s_id, 702 filp->f_path.dentry->d_parent->d_name.name,
625 filp->f_path.dentry->d_inode->i_ino, 703 filp->f_path.dentry->d_name.name,
626 fl->fl_type, fl->fl_flags); 704 fl->fl_type, fl->fl_flags);
627 705
628 /* 706 /*
@@ -645,12 +723,15 @@ static int nfs_flock(struct file *filp, int cmd, struct file_lock *fl)
645 return do_setlk(filp, cmd, fl); 723 return do_setlk(filp, cmd, fl);
646} 724}
647 725
726/*
727 * There is no protocol support for leases, so we have no way to implement
728 * them correctly in the face of opens by other clients.
729 */
648static int nfs_setlease(struct file *file, long arg, struct file_lock **fl) 730static int nfs_setlease(struct file *file, long arg, struct file_lock **fl)
649{ 731{
650 /* 732 dprintk("NFS: setlease(%s/%s, arg=%ld)\n",
651 * There is no protocol support for leases, so we have no way 733 file->f_path.dentry->d_parent->d_name.name,
652 * to implement them correctly in the face of opens by other 734 file->f_path.dentry->d_name.name, arg);
653 * clients. 735
654 */
655 return -EINVAL; 736 return -EINVAL;
656} 737}
diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c
index 596c5d8e86f4..2c23d067e2a6 100644
--- a/fs/nfs/inode.c
+++ b/fs/nfs/inode.c
@@ -57,8 +57,6 @@ static int enable_ino64 = NFS_64_BIT_INODE_NUMBERS_ENABLED;
57static void nfs_invalidate_inode(struct inode *); 57static void nfs_invalidate_inode(struct inode *);
58static int nfs_update_inode(struct inode *, struct nfs_fattr *); 58static int nfs_update_inode(struct inode *, struct nfs_fattr *);
59 59
60static void nfs_zap_acl_cache(struct inode *);
61
62static struct kmem_cache * nfs_inode_cachep; 60static struct kmem_cache * nfs_inode_cachep;
63 61
64static inline unsigned long 62static inline unsigned long
@@ -167,7 +165,7 @@ void nfs_zap_mapping(struct inode *inode, struct address_space *mapping)
167 } 165 }
168} 166}
169 167
170static void nfs_zap_acl_cache(struct inode *inode) 168void nfs_zap_acl_cache(struct inode *inode)
171{ 169{
172 void (*clear_acl_cache)(struct inode *); 170 void (*clear_acl_cache)(struct inode *);
173 171
@@ -347,7 +345,7 @@ out_no_inode:
347 goto out; 345 goto out;
348} 346}
349 347
350#define NFS_VALID_ATTRS (ATTR_MODE|ATTR_UID|ATTR_GID|ATTR_SIZE|ATTR_ATIME|ATTR_ATIME_SET|ATTR_MTIME|ATTR_MTIME_SET) 348#define NFS_VALID_ATTRS (ATTR_MODE|ATTR_UID|ATTR_GID|ATTR_SIZE|ATTR_ATIME|ATTR_ATIME_SET|ATTR_MTIME|ATTR_MTIME_SET|ATTR_FILE)
351 349
352int 350int
353nfs_setattr(struct dentry *dentry, struct iattr *attr) 351nfs_setattr(struct dentry *dentry, struct iattr *attr)
@@ -369,7 +367,7 @@ nfs_setattr(struct dentry *dentry, struct iattr *attr)
369 367
370 /* Optimization: if the end result is no change, don't RPC */ 368 /* Optimization: if the end result is no change, don't RPC */
371 attr->ia_valid &= NFS_VALID_ATTRS; 369 attr->ia_valid &= NFS_VALID_ATTRS;
372 if (attr->ia_valid == 0) 370 if ((attr->ia_valid & ~ATTR_FILE) == 0)
373 return 0; 371 return 0;
374 372
375 lock_kernel(); 373 lock_kernel();
diff --git a/fs/nfs/internal.h b/fs/nfs/internal.h
index 04ae867dddba..24241fcbb98d 100644
--- a/fs/nfs/internal.h
+++ b/fs/nfs/internal.h
@@ -150,6 +150,7 @@ extern void nfs_clear_inode(struct inode *);
150#ifdef CONFIG_NFS_V4 150#ifdef CONFIG_NFS_V4
151extern void nfs4_clear_inode(struct inode *); 151extern void nfs4_clear_inode(struct inode *);
152#endif 152#endif
153void nfs_zap_acl_cache(struct inode *inode);
153 154
154/* super.c */ 155/* super.c */
155extern struct file_system_type nfs_xdev_fs_type; 156extern struct file_system_type nfs_xdev_fs_type;
diff --git a/fs/nfs/iostat.h b/fs/nfs/iostat.h
index 6350ecbde589..a36952810032 100644
--- a/fs/nfs/iostat.h
+++ b/fs/nfs/iostat.h
@@ -5,135 +5,41 @@
5 * 5 *
6 * Copyright (C) 2005, 2006 Chuck Lever <cel@netapp.com> 6 * Copyright (C) 2005, 2006 Chuck Lever <cel@netapp.com>
7 * 7 *
8 * NFS client per-mount statistics provide information about the health of
9 * the NFS client and the health of each NFS mount point. Generally these
10 * are not for detailed problem diagnosis, but simply to indicate that there
11 * is a problem.
12 *
13 * These counters are not meant to be human-readable, but are meant to be
14 * integrated into system monitoring tools such as "sar" and "iostat". As
15 * such, the counters are sampled by the tools over time, and are never
16 * zeroed after a file system is mounted. Moving averages can be computed
17 * by the tools by taking the difference between two instantaneous samples
18 * and dividing that by the time between the samples.
19 */ 8 */
20 9
21#ifndef _NFS_IOSTAT 10#ifndef _NFS_IOSTAT
22#define _NFS_IOSTAT 11#define _NFS_IOSTAT
23 12
24#define NFS_IOSTAT_VERS "1.0"
25
26/*
27 * NFS byte counters
28 *
29 * 1. SERVER - the number of payload bytes read from or written to the
30 * server by the NFS client via an NFS READ or WRITE request.
31 *
32 * 2. NORMAL - the number of bytes read or written by applications via
33 * the read(2) and write(2) system call interfaces.
34 *
35 * 3. DIRECT - the number of bytes read or written from files opened
36 * with the O_DIRECT flag.
37 *
38 * These counters give a view of the data throughput into and out of the NFS
39 * client. Comparing the number of bytes requested by an application with the
40 * number of bytes the client requests from the server can provide an
41 * indication of client efficiency (per-op, cache hits, etc).
42 *
43 * These counters can also help characterize which access methods are in
44 * use. DIRECT by itself shows whether there is any O_DIRECT traffic.
45 * NORMAL + DIRECT shows how much data is going through the system call
46 * interface. A large amount of SERVER traffic without much NORMAL or
47 * DIRECT traffic shows that applications are using mapped files.
48 *
49 * NFS page counters
50 *
51 * These count the number of pages read or written via nfs_readpage(),
52 * nfs_readpages(), or their write equivalents.
53 */
54enum nfs_stat_bytecounters {
55 NFSIOS_NORMALREADBYTES = 0,
56 NFSIOS_NORMALWRITTENBYTES,
57 NFSIOS_DIRECTREADBYTES,
58 NFSIOS_DIRECTWRITTENBYTES,
59 NFSIOS_SERVERREADBYTES,
60 NFSIOS_SERVERWRITTENBYTES,
61 NFSIOS_READPAGES,
62 NFSIOS_WRITEPAGES,
63 __NFSIOS_BYTESMAX,
64};
65
66/*
67 * NFS event counters
68 *
69 * These counters provide a low-overhead way of monitoring client activity
70 * without enabling NFS trace debugging. The counters show the rate at
71 * which VFS requests are made, and how often the client invalidates its
72 * data and attribute caches. This allows system administrators to monitor
73 * such things as how close-to-open is working, and answer questions such
74 * as "why are there so many GETATTR requests on the wire?"
75 *
76 * They also count anamolous events such as short reads and writes, silly
77 * renames due to close-after-delete, and operations that change the size
78 * of a file (such operations can often be the source of data corruption
79 * if applications aren't using file locking properly).
80 */
81enum nfs_stat_eventcounters {
82 NFSIOS_INODEREVALIDATE = 0,
83 NFSIOS_DENTRYREVALIDATE,
84 NFSIOS_DATAINVALIDATE,
85 NFSIOS_ATTRINVALIDATE,
86 NFSIOS_VFSOPEN,
87 NFSIOS_VFSLOOKUP,
88 NFSIOS_VFSACCESS,
89 NFSIOS_VFSUPDATEPAGE,
90 NFSIOS_VFSREADPAGE,
91 NFSIOS_VFSREADPAGES,
92 NFSIOS_VFSWRITEPAGE,
93 NFSIOS_VFSWRITEPAGES,
94 NFSIOS_VFSGETDENTS,
95 NFSIOS_VFSSETATTR,
96 NFSIOS_VFSFLUSH,
97 NFSIOS_VFSFSYNC,
98 NFSIOS_VFSLOCK,
99 NFSIOS_VFSRELEASE,
100 NFSIOS_CONGESTIONWAIT,
101 NFSIOS_SETATTRTRUNC,
102 NFSIOS_EXTENDWRITE,
103 NFSIOS_SILLYRENAME,
104 NFSIOS_SHORTREAD,
105 NFSIOS_SHORTWRITE,
106 NFSIOS_DELAY,
107 __NFSIOS_COUNTSMAX,
108};
109
110#ifdef __KERNEL__
111
112#include <linux/percpu.h> 13#include <linux/percpu.h>
113#include <linux/cache.h> 14#include <linux/cache.h>
15#include <linux/nfs_iostat.h>
114 16
115struct nfs_iostats { 17struct nfs_iostats {
116 unsigned long long bytes[__NFSIOS_BYTESMAX]; 18 unsigned long long bytes[__NFSIOS_BYTESMAX];
117 unsigned long events[__NFSIOS_COUNTSMAX]; 19 unsigned long events[__NFSIOS_COUNTSMAX];
118} ____cacheline_aligned; 20} ____cacheline_aligned;
119 21
120static inline void nfs_inc_server_stats(struct nfs_server *server, enum nfs_stat_eventcounters stat) 22static inline void nfs_inc_server_stats(const struct nfs_server *server,
23 enum nfs_stat_eventcounters stat)
121{ 24{
122 struct nfs_iostats *iostats; 25 struct nfs_iostats *iostats;
123 int cpu; 26 int cpu;
124 27
125 cpu = get_cpu(); 28 cpu = get_cpu();
126 iostats = per_cpu_ptr(server->io_stats, cpu); 29 iostats = per_cpu_ptr(server->io_stats, cpu);
127 iostats->events[stat] ++; 30 iostats->events[stat]++;
128 put_cpu_no_resched(); 31 put_cpu_no_resched();
129} 32}
130 33
131static inline void nfs_inc_stats(struct inode *inode, enum nfs_stat_eventcounters stat) 34static inline void nfs_inc_stats(const struct inode *inode,
35 enum nfs_stat_eventcounters stat)
132{ 36{
133 nfs_inc_server_stats(NFS_SERVER(inode), stat); 37 nfs_inc_server_stats(NFS_SERVER(inode), stat);
134} 38}
135 39
136static inline void nfs_add_server_stats(struct nfs_server *server, enum nfs_stat_bytecounters stat, unsigned long addend) 40static inline void nfs_add_server_stats(const struct nfs_server *server,
41 enum nfs_stat_bytecounters stat,
42 unsigned long addend)
137{ 43{
138 struct nfs_iostats *iostats; 44 struct nfs_iostats *iostats;
139 int cpu; 45 int cpu;
@@ -144,7 +50,9 @@ static inline void nfs_add_server_stats(struct nfs_server *server, enum nfs_stat
144 put_cpu_no_resched(); 50 put_cpu_no_resched();
145} 51}
146 52
147static inline void nfs_add_stats(struct inode *inode, enum nfs_stat_bytecounters stat, unsigned long addend) 53static inline void nfs_add_stats(const struct inode *inode,
54 enum nfs_stat_bytecounters stat,
55 unsigned long addend)
148{ 56{
149 nfs_add_server_stats(NFS_SERVER(inode), stat, addend); 57 nfs_add_server_stats(NFS_SERVER(inode), stat, addend);
150} 58}
@@ -160,5 +68,4 @@ static inline void nfs_free_iostats(struct nfs_iostats *stats)
160 free_percpu(stats); 68 free_percpu(stats);
161} 69}
162 70
163#endif 71#endif /* _NFS_IOSTAT */
164#endif
diff --git a/fs/nfs/nfs3acl.c b/fs/nfs/nfs3acl.c
index 9b7362565c0c..423842f51ac9 100644
--- a/fs/nfs/nfs3acl.c
+++ b/fs/nfs/nfs3acl.c
@@ -5,6 +5,8 @@
5#include <linux/posix_acl_xattr.h> 5#include <linux/posix_acl_xattr.h>
6#include <linux/nfsacl.h> 6#include <linux/nfsacl.h>
7 7
8#include "internal.h"
9
8#define NFSDBG_FACILITY NFSDBG_PROC 10#define NFSDBG_FACILITY NFSDBG_PROC
9 11
10ssize_t nfs3_listxattr(struct dentry *dentry, char *buffer, size_t size) 12ssize_t nfs3_listxattr(struct dentry *dentry, char *buffer, size_t size)
@@ -205,6 +207,8 @@ struct posix_acl *nfs3_proc_getacl(struct inode *inode, int type)
205 status = nfs_revalidate_inode(server, inode); 207 status = nfs_revalidate_inode(server, inode);
206 if (status < 0) 208 if (status < 0)
207 return ERR_PTR(status); 209 return ERR_PTR(status);
210 if (NFS_I(inode)->cache_validity & NFS_INO_INVALID_ACL)
211 nfs_zap_acl_cache(inode);
208 acl = nfs3_get_cached_acl(inode, type); 212 acl = nfs3_get_cached_acl(inode, type);
209 if (acl != ERR_PTR(-EAGAIN)) 213 if (acl != ERR_PTR(-EAGAIN))
210 return acl; 214 return acl;
@@ -319,9 +323,8 @@ static int nfs3_proc_setacls(struct inode *inode, struct posix_acl *acl,
319 dprintk("NFS call setacl\n"); 323 dprintk("NFS call setacl\n");
320 msg.rpc_proc = &server->client_acl->cl_procinfo[ACLPROC3_SETACL]; 324 msg.rpc_proc = &server->client_acl->cl_procinfo[ACLPROC3_SETACL];
321 status = rpc_call_sync(server->client_acl, &msg, 0); 325 status = rpc_call_sync(server->client_acl, &msg, 0);
322 spin_lock(&inode->i_lock); 326 nfs_access_zap_cache(inode);
323 NFS_I(inode)->cache_validity |= NFS_INO_INVALID_ACCESS; 327 nfs_zap_acl_cache(inode);
324 spin_unlock(&inode->i_lock);
325 dprintk("NFS reply setacl: %d\n", status); 328 dprintk("NFS reply setacl: %d\n", status);
326 329
327 /* pages may have been allocated at the xdr layer. */ 330 /* pages may have been allocated at the xdr layer. */
diff --git a/fs/nfs/nfs3proc.c b/fs/nfs/nfs3proc.c
index c3523ad03ed1..1e750e4574a9 100644
--- a/fs/nfs/nfs3proc.c
+++ b/fs/nfs/nfs3proc.c
@@ -129,6 +129,8 @@ nfs3_proc_setattr(struct dentry *dentry, struct nfs_fattr *fattr,
129 int status; 129 int status;
130 130
131 dprintk("NFS call setattr\n"); 131 dprintk("NFS call setattr\n");
132 if (sattr->ia_valid & ATTR_FILE)
133 msg.rpc_cred = nfs_file_cred(sattr->ia_file);
132 nfs_fattr_init(fattr); 134 nfs_fattr_init(fattr);
133 status = rpc_call_sync(NFS_CLIENT(inode), &msg, 0); 135 status = rpc_call_sync(NFS_CLIENT(inode), &msg, 0);
134 if (status == 0) 136 if (status == 0)
@@ -248,6 +250,53 @@ static int nfs3_proc_readlink(struct inode *inode, struct page *page,
248 return status; 250 return status;
249} 251}
250 252
253struct nfs3_createdata {
254 struct rpc_message msg;
255 union {
256 struct nfs3_createargs create;
257 struct nfs3_mkdirargs mkdir;
258 struct nfs3_symlinkargs symlink;
259 struct nfs3_mknodargs mknod;
260 } arg;
261 struct nfs3_diropres res;
262 struct nfs_fh fh;
263 struct nfs_fattr fattr;
264 struct nfs_fattr dir_attr;
265};
266
267static struct nfs3_createdata *nfs3_alloc_createdata(void)
268{
269 struct nfs3_createdata *data;
270
271 data = kzalloc(sizeof(*data), GFP_KERNEL);
272 if (data != NULL) {
273 data->msg.rpc_argp = &data->arg;
274 data->msg.rpc_resp = &data->res;
275 data->res.fh = &data->fh;
276 data->res.fattr = &data->fattr;
277 data->res.dir_attr = &data->dir_attr;
278 nfs_fattr_init(data->res.fattr);
279 nfs_fattr_init(data->res.dir_attr);
280 }
281 return data;
282}
283
284static int nfs3_do_create(struct inode *dir, struct dentry *dentry, struct nfs3_createdata *data)
285{
286 int status;
287
288 status = rpc_call_sync(NFS_CLIENT(dir), &data->msg, 0);
289 nfs_post_op_update_inode(dir, data->res.dir_attr);
290 if (status == 0)
291 status = nfs_instantiate(dentry, data->res.fh, data->res.fattr);
292 return status;
293}
294
295static void nfs3_free_createdata(struct nfs3_createdata *data)
296{
297 kfree(data);
298}
299
251/* 300/*
252 * Create a regular file. 301 * Create a regular file.
253 * For now, we don't implement O_EXCL. 302 * For now, we don't implement O_EXCL.
@@ -256,70 +305,60 @@ static int
256nfs3_proc_create(struct inode *dir, struct dentry *dentry, struct iattr *sattr, 305nfs3_proc_create(struct inode *dir, struct dentry *dentry, struct iattr *sattr,
257 int flags, struct nameidata *nd) 306 int flags, struct nameidata *nd)
258{ 307{
259 struct nfs_fh fhandle; 308 struct nfs3_createdata *data;
260 struct nfs_fattr fattr;
261 struct nfs_fattr dir_attr;
262 struct nfs3_createargs arg = {
263 .fh = NFS_FH(dir),
264 .name = dentry->d_name.name,
265 .len = dentry->d_name.len,
266 .sattr = sattr,
267 };
268 struct nfs3_diropres res = {
269 .dir_attr = &dir_attr,
270 .fh = &fhandle,
271 .fattr = &fattr
272 };
273 struct rpc_message msg = {
274 .rpc_proc = &nfs3_procedures[NFS3PROC_CREATE],
275 .rpc_argp = &arg,
276 .rpc_resp = &res,
277 };
278 mode_t mode = sattr->ia_mode; 309 mode_t mode = sattr->ia_mode;
279 int status; 310 int status = -ENOMEM;
280 311
281 dprintk("NFS call create %s\n", dentry->d_name.name); 312 dprintk("NFS call create %s\n", dentry->d_name.name);
282 arg.createmode = NFS3_CREATE_UNCHECKED; 313
314 data = nfs3_alloc_createdata();
315 if (data == NULL)
316 goto out;
317
318 data->msg.rpc_proc = &nfs3_procedures[NFS3PROC_CREATE];
319 data->arg.create.fh = NFS_FH(dir);
320 data->arg.create.name = dentry->d_name.name;
321 data->arg.create.len = dentry->d_name.len;
322 data->arg.create.sattr = sattr;
323
324 data->arg.create.createmode = NFS3_CREATE_UNCHECKED;
283 if (flags & O_EXCL) { 325 if (flags & O_EXCL) {
284 arg.createmode = NFS3_CREATE_EXCLUSIVE; 326 data->arg.create.createmode = NFS3_CREATE_EXCLUSIVE;
285 arg.verifier[0] = jiffies; 327 data->arg.create.verifier[0] = jiffies;
286 arg.verifier[1] = current->pid; 328 data->arg.create.verifier[1] = current->pid;
287 } 329 }
288 330
289 sattr->ia_mode &= ~current->fs->umask; 331 sattr->ia_mode &= ~current->fs->umask;
290 332
291again: 333 for (;;) {
292 nfs_fattr_init(&dir_attr); 334 status = nfs3_do_create(dir, dentry, data);
293 nfs_fattr_init(&fattr);
294 status = rpc_call_sync(NFS_CLIENT(dir), &msg, 0);
295 nfs_refresh_inode(dir, &dir_attr);
296 335
297 /* If the server doesn't support the exclusive creation semantics, 336 if (status != -ENOTSUPP)
298 * try again with simple 'guarded' mode. */ 337 break;
299 if (status == -ENOTSUPP) { 338 /* If the server doesn't support the exclusive creation
300 switch (arg.createmode) { 339 * semantics, try again with simple 'guarded' mode. */
340 switch (data->arg.create.createmode) {
301 case NFS3_CREATE_EXCLUSIVE: 341 case NFS3_CREATE_EXCLUSIVE:
302 arg.createmode = NFS3_CREATE_GUARDED; 342 data->arg.create.createmode = NFS3_CREATE_GUARDED;
303 break; 343 break;
304 344
305 case NFS3_CREATE_GUARDED: 345 case NFS3_CREATE_GUARDED:
306 arg.createmode = NFS3_CREATE_UNCHECKED; 346 data->arg.create.createmode = NFS3_CREATE_UNCHECKED;
307 break; 347 break;
308 348
309 case NFS3_CREATE_UNCHECKED: 349 case NFS3_CREATE_UNCHECKED:
310 goto out; 350 goto out;
311 } 351 }
312 goto again; 352 nfs_fattr_init(data->res.dir_attr);
353 nfs_fattr_init(data->res.fattr);
313 } 354 }
314 355
315 if (status == 0)
316 status = nfs_instantiate(dentry, &fhandle, &fattr);
317 if (status != 0) 356 if (status != 0)
318 goto out; 357 goto out;
319 358
320 /* When we created the file with exclusive semantics, make 359 /* When we created the file with exclusive semantics, make
321 * sure we set the attributes afterwards. */ 360 * sure we set the attributes afterwards. */
322 if (arg.createmode == NFS3_CREATE_EXCLUSIVE) { 361 if (data->arg.create.createmode == NFS3_CREATE_EXCLUSIVE) {
323 dprintk("NFS call setattr (post-create)\n"); 362 dprintk("NFS call setattr (post-create)\n");
324 363
325 if (!(sattr->ia_valid & ATTR_ATIME_SET)) 364 if (!(sattr->ia_valid & ATTR_ATIME_SET))
@@ -330,14 +369,15 @@ again:
330 /* Note: we could use a guarded setattr here, but I'm 369 /* Note: we could use a guarded setattr here, but I'm
331 * not sure this buys us anything (and I'd have 370 * not sure this buys us anything (and I'd have
332 * to revamp the NFSv3 XDR code) */ 371 * to revamp the NFSv3 XDR code) */
333 status = nfs3_proc_setattr(dentry, &fattr, sattr); 372 status = nfs3_proc_setattr(dentry, data->res.fattr, sattr);
334 nfs_post_op_update_inode(dentry->d_inode, &fattr); 373 nfs_post_op_update_inode(dentry->d_inode, data->res.fattr);
335 dprintk("NFS reply setattr (post-create): %d\n", status); 374 dprintk("NFS reply setattr (post-create): %d\n", status);
375 if (status != 0)
376 goto out;
336 } 377 }
337 if (status != 0)
338 goto out;
339 status = nfs3_proc_set_default_acl(dir, dentry->d_inode, mode); 378 status = nfs3_proc_set_default_acl(dir, dentry->d_inode, mode);
340out: 379out:
380 nfs3_free_createdata(data);
341 dprintk("NFS reply create: %d\n", status); 381 dprintk("NFS reply create: %d\n", status);
342 return status; 382 return status;
343} 383}
@@ -452,40 +492,28 @@ static int
452nfs3_proc_symlink(struct inode *dir, struct dentry *dentry, struct page *page, 492nfs3_proc_symlink(struct inode *dir, struct dentry *dentry, struct page *page,
453 unsigned int len, struct iattr *sattr) 493 unsigned int len, struct iattr *sattr)
454{ 494{
455 struct nfs_fh fhandle; 495 struct nfs3_createdata *data;
456 struct nfs_fattr fattr, dir_attr; 496 int status = -ENOMEM;
457 struct nfs3_symlinkargs arg = {
458 .fromfh = NFS_FH(dir),
459 .fromname = dentry->d_name.name,
460 .fromlen = dentry->d_name.len,
461 .pages = &page,
462 .pathlen = len,
463 .sattr = sattr
464 };
465 struct nfs3_diropres res = {
466 .dir_attr = &dir_attr,
467 .fh = &fhandle,
468 .fattr = &fattr
469 };
470 struct rpc_message msg = {
471 .rpc_proc = &nfs3_procedures[NFS3PROC_SYMLINK],
472 .rpc_argp = &arg,
473 .rpc_resp = &res,
474 };
475 int status;
476 497
477 if (len > NFS3_MAXPATHLEN) 498 if (len > NFS3_MAXPATHLEN)
478 return -ENAMETOOLONG; 499 return -ENAMETOOLONG;
479 500
480 dprintk("NFS call symlink %s\n", dentry->d_name.name); 501 dprintk("NFS call symlink %s\n", dentry->d_name.name);
481 502
482 nfs_fattr_init(&dir_attr); 503 data = nfs3_alloc_createdata();
483 nfs_fattr_init(&fattr); 504 if (data == NULL)
484 status = rpc_call_sync(NFS_CLIENT(dir), &msg, 0);
485 nfs_post_op_update_inode(dir, &dir_attr);
486 if (status != 0)
487 goto out; 505 goto out;
488 status = nfs_instantiate(dentry, &fhandle, &fattr); 506 data->msg.rpc_proc = &nfs3_procedures[NFS3PROC_SYMLINK];
507 data->arg.symlink.fromfh = NFS_FH(dir);
508 data->arg.symlink.fromname = dentry->d_name.name;
509 data->arg.symlink.fromlen = dentry->d_name.len;
510 data->arg.symlink.pages = &page;
511 data->arg.symlink.pathlen = len;
512 data->arg.symlink.sattr = sattr;
513
514 status = nfs3_do_create(dir, dentry, data);
515
516 nfs3_free_createdata(data);
489out: 517out:
490 dprintk("NFS reply symlink: %d\n", status); 518 dprintk("NFS reply symlink: %d\n", status);
491 return status; 519 return status;
@@ -494,42 +522,31 @@ out:
494static int 522static int
495nfs3_proc_mkdir(struct inode *dir, struct dentry *dentry, struct iattr *sattr) 523nfs3_proc_mkdir(struct inode *dir, struct dentry *dentry, struct iattr *sattr)
496{ 524{
497 struct nfs_fh fhandle; 525 struct nfs3_createdata *data;
498 struct nfs_fattr fattr, dir_attr;
499 struct nfs3_mkdirargs arg = {
500 .fh = NFS_FH(dir),
501 .name = dentry->d_name.name,
502 .len = dentry->d_name.len,
503 .sattr = sattr
504 };
505 struct nfs3_diropres res = {
506 .dir_attr = &dir_attr,
507 .fh = &fhandle,
508 .fattr = &fattr
509 };
510 struct rpc_message msg = {
511 .rpc_proc = &nfs3_procedures[NFS3PROC_MKDIR],
512 .rpc_argp = &arg,
513 .rpc_resp = &res,
514 };
515 int mode = sattr->ia_mode; 526 int mode = sattr->ia_mode;
516 int status; 527 int status = -ENOMEM;
517 528
518 dprintk("NFS call mkdir %s\n", dentry->d_name.name); 529 dprintk("NFS call mkdir %s\n", dentry->d_name.name);
519 530
520 sattr->ia_mode &= ~current->fs->umask; 531 sattr->ia_mode &= ~current->fs->umask;
521 532
522 nfs_fattr_init(&dir_attr); 533 data = nfs3_alloc_createdata();
523 nfs_fattr_init(&fattr); 534 if (data == NULL)
524 status = rpc_call_sync(NFS_CLIENT(dir), &msg, 0);
525 nfs_post_op_update_inode(dir, &dir_attr);
526 if (status != 0)
527 goto out; 535 goto out;
528 status = nfs_instantiate(dentry, &fhandle, &fattr); 536
537 data->msg.rpc_proc = &nfs3_procedures[NFS3PROC_MKDIR];
538 data->arg.mkdir.fh = NFS_FH(dir);
539 data->arg.mkdir.name = dentry->d_name.name;
540 data->arg.mkdir.len = dentry->d_name.len;
541 data->arg.mkdir.sattr = sattr;
542
543 status = nfs3_do_create(dir, dentry, data);
529 if (status != 0) 544 if (status != 0)
530 goto out; 545 goto out;
546
531 status = nfs3_proc_set_default_acl(dir, dentry->d_inode, mode); 547 status = nfs3_proc_set_default_acl(dir, dentry->d_inode, mode);
532out: 548out:
549 nfs3_free_createdata(data);
533 dprintk("NFS reply mkdir: %d\n", status); 550 dprintk("NFS reply mkdir: %d\n", status);
534 return status; 551 return status;
535} 552}
@@ -615,52 +632,50 @@ static int
615nfs3_proc_mknod(struct inode *dir, struct dentry *dentry, struct iattr *sattr, 632nfs3_proc_mknod(struct inode *dir, struct dentry *dentry, struct iattr *sattr,
616 dev_t rdev) 633 dev_t rdev)
617{ 634{
618 struct nfs_fh fh; 635 struct nfs3_createdata *data;
619 struct nfs_fattr fattr, dir_attr;
620 struct nfs3_mknodargs arg = {
621 .fh = NFS_FH(dir),
622 .name = dentry->d_name.name,
623 .len = dentry->d_name.len,
624 .sattr = sattr,
625 .rdev = rdev
626 };
627 struct nfs3_diropres res = {
628 .dir_attr = &dir_attr,
629 .fh = &fh,
630 .fattr = &fattr
631 };
632 struct rpc_message msg = {
633 .rpc_proc = &nfs3_procedures[NFS3PROC_MKNOD],
634 .rpc_argp = &arg,
635 .rpc_resp = &res,
636 };
637 mode_t mode = sattr->ia_mode; 636 mode_t mode = sattr->ia_mode;
638 int status; 637 int status = -ENOMEM;
639
640 switch (sattr->ia_mode & S_IFMT) {
641 case S_IFBLK: arg.type = NF3BLK; break;
642 case S_IFCHR: arg.type = NF3CHR; break;
643 case S_IFIFO: arg.type = NF3FIFO; break;
644 case S_IFSOCK: arg.type = NF3SOCK; break;
645 default: return -EINVAL;
646 }
647 638
648 dprintk("NFS call mknod %s %u:%u\n", dentry->d_name.name, 639 dprintk("NFS call mknod %s %u:%u\n", dentry->d_name.name,
649 MAJOR(rdev), MINOR(rdev)); 640 MAJOR(rdev), MINOR(rdev));
650 641
651 sattr->ia_mode &= ~current->fs->umask; 642 sattr->ia_mode &= ~current->fs->umask;
652 643
653 nfs_fattr_init(&dir_attr); 644 data = nfs3_alloc_createdata();
654 nfs_fattr_init(&fattr); 645 if (data == NULL)
655 status = rpc_call_sync(NFS_CLIENT(dir), &msg, 0);
656 nfs_post_op_update_inode(dir, &dir_attr);
657 if (status != 0)
658 goto out; 646 goto out;
659 status = nfs_instantiate(dentry, &fh, &fattr); 647
648 data->msg.rpc_proc = &nfs3_procedures[NFS3PROC_MKNOD];
649 data->arg.mknod.fh = NFS_FH(dir);
650 data->arg.mknod.name = dentry->d_name.name;
651 data->arg.mknod.len = dentry->d_name.len;
652 data->arg.mknod.sattr = sattr;
653 data->arg.mknod.rdev = rdev;
654
655 switch (sattr->ia_mode & S_IFMT) {
656 case S_IFBLK:
657 data->arg.mknod.type = NF3BLK;
658 break;
659 case S_IFCHR:
660 data->arg.mknod.type = NF3CHR;
661 break;
662 case S_IFIFO:
663 data->arg.mknod.type = NF3FIFO;
664 break;
665 case S_IFSOCK:
666 data->arg.mknod.type = NF3SOCK;
667 break;
668 default:
669 status = -EINVAL;
670 goto out;
671 }
672
673 status = nfs3_do_create(dir, dentry, data);
660 if (status != 0) 674 if (status != 0)
661 goto out; 675 goto out;
662 status = nfs3_proc_set_default_acl(dir, dentry->d_inode, mode); 676 status = nfs3_proc_set_default_acl(dir, dentry->d_inode, mode);
663out: 677out:
678 nfs3_free_createdata(data);
664 dprintk("NFS reply mknod: %d\n", status); 679 dprintk("NFS reply mknod: %d\n", status);
665 return status; 680 return status;
666} 681}
@@ -801,8 +816,6 @@ const struct nfs_rpc_ops nfs_v3_clientops = {
801 .write_done = nfs3_write_done, 816 .write_done = nfs3_write_done,
802 .commit_setup = nfs3_proc_commit_setup, 817 .commit_setup = nfs3_proc_commit_setup,
803 .commit_done = nfs3_commit_done, 818 .commit_done = nfs3_commit_done,
804 .file_open = nfs_open,
805 .file_release = nfs_release,
806 .lock = nfs3_proc_lock, 819 .lock = nfs3_proc_lock,
807 .clear_acl_cache = nfs3_forget_cached_acls, 820 .clear_acl_cache = nfs3_forget_cached_acls,
808}; 821};
diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index 1293e0acd82b..4451287a81d1 100644
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -1139,8 +1139,9 @@ static struct nfs4_state *nfs4_do_open(struct inode *dir, struct path *path, int
1139 return res; 1139 return res;
1140} 1140}
1141 1141
1142static int _nfs4_do_setattr(struct inode *inode, struct nfs_fattr *fattr, 1142static int _nfs4_do_setattr(struct inode *inode, struct rpc_cred *cred,
1143 struct iattr *sattr, struct nfs4_state *state) 1143 struct nfs_fattr *fattr, struct iattr *sattr,
1144 struct nfs4_state *state)
1144{ 1145{
1145 struct nfs_server *server = NFS_SERVER(inode); 1146 struct nfs_server *server = NFS_SERVER(inode);
1146 struct nfs_setattrargs arg = { 1147 struct nfs_setattrargs arg = {
@@ -1154,9 +1155,10 @@ static int _nfs4_do_setattr(struct inode *inode, struct nfs_fattr *fattr,
1154 .server = server, 1155 .server = server,
1155 }; 1156 };
1156 struct rpc_message msg = { 1157 struct rpc_message msg = {
1157 .rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SETATTR], 1158 .rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SETATTR],
1158 .rpc_argp = &arg, 1159 .rpc_argp = &arg,
1159 .rpc_resp = &res, 1160 .rpc_resp = &res,
1161 .rpc_cred = cred,
1160 }; 1162 };
1161 unsigned long timestamp = jiffies; 1163 unsigned long timestamp = jiffies;
1162 int status; 1164 int status;
@@ -1166,7 +1168,6 @@ static int _nfs4_do_setattr(struct inode *inode, struct nfs_fattr *fattr,
1166 if (nfs4_copy_delegation_stateid(&arg.stateid, inode)) { 1168 if (nfs4_copy_delegation_stateid(&arg.stateid, inode)) {
1167 /* Use that stateid */ 1169 /* Use that stateid */
1168 } else if (state != NULL) { 1170 } else if (state != NULL) {
1169 msg.rpc_cred = state->owner->so_cred;
1170 nfs4_copy_stateid(&arg.stateid, state, current->files); 1171 nfs4_copy_stateid(&arg.stateid, state, current->files);
1171 } else 1172 } else
1172 memcpy(&arg.stateid, &zero_stateid, sizeof(arg.stateid)); 1173 memcpy(&arg.stateid, &zero_stateid, sizeof(arg.stateid));
@@ -1177,15 +1178,16 @@ static int _nfs4_do_setattr(struct inode *inode, struct nfs_fattr *fattr,
1177 return status; 1178 return status;
1178} 1179}
1179 1180
1180static int nfs4_do_setattr(struct inode *inode, struct nfs_fattr *fattr, 1181static int nfs4_do_setattr(struct inode *inode, struct rpc_cred *cred,
1181 struct iattr *sattr, struct nfs4_state *state) 1182 struct nfs_fattr *fattr, struct iattr *sattr,
1183 struct nfs4_state *state)
1182{ 1184{
1183 struct nfs_server *server = NFS_SERVER(inode); 1185 struct nfs_server *server = NFS_SERVER(inode);
1184 struct nfs4_exception exception = { }; 1186 struct nfs4_exception exception = { };
1185 int err; 1187 int err;
1186 do { 1188 do {
1187 err = nfs4_handle_exception(server, 1189 err = nfs4_handle_exception(server,
1188 _nfs4_do_setattr(inode, fattr, sattr, state), 1190 _nfs4_do_setattr(inode, cred, fattr, sattr, state),
1189 &exception); 1191 &exception);
1190 } while (exception.retry); 1192 } while (exception.retry);
1191 return err; 1193 return err;
@@ -1647,29 +1649,25 @@ static int
1647nfs4_proc_setattr(struct dentry *dentry, struct nfs_fattr *fattr, 1649nfs4_proc_setattr(struct dentry *dentry, struct nfs_fattr *fattr,
1648 struct iattr *sattr) 1650 struct iattr *sattr)
1649{ 1651{
1650 struct rpc_cred *cred;
1651 struct inode *inode = dentry->d_inode; 1652 struct inode *inode = dentry->d_inode;
1652 struct nfs_open_context *ctx; 1653 struct rpc_cred *cred = NULL;
1653 struct nfs4_state *state = NULL; 1654 struct nfs4_state *state = NULL;
1654 int status; 1655 int status;
1655 1656
1656 nfs_fattr_init(fattr); 1657 nfs_fattr_init(fattr);
1657 1658
1658 cred = rpc_lookup_cred();
1659 if (IS_ERR(cred))
1660 return PTR_ERR(cred);
1661
1662 /* Search for an existing open(O_WRITE) file */ 1659 /* Search for an existing open(O_WRITE) file */
1663 ctx = nfs_find_open_context(inode, cred, FMODE_WRITE); 1660 if (sattr->ia_valid & ATTR_FILE) {
1664 if (ctx != NULL) 1661 struct nfs_open_context *ctx;
1662
1663 ctx = nfs_file_open_context(sattr->ia_file);
1664 cred = ctx->cred;
1665 state = ctx->state; 1665 state = ctx->state;
1666 }
1666 1667
1667 status = nfs4_do_setattr(inode, fattr, sattr, state); 1668 status = nfs4_do_setattr(inode, cred, fattr, sattr, state);
1668 if (status == 0) 1669 if (status == 0)
1669 nfs_setattr_update_inode(inode, sattr); 1670 nfs_setattr_update_inode(inode, sattr);
1670 if (ctx != NULL)
1671 put_nfs_open_context(ctx);
1672 put_rpccred(cred);
1673 return status; 1671 return status;
1674} 1672}
1675 1673
@@ -1897,17 +1895,16 @@ nfs4_proc_create(struct inode *dir, struct dentry *dentry, struct iattr *sattr,
1897 goto out; 1895 goto out;
1898 } 1896 }
1899 state = nfs4_do_open(dir, &path, flags, sattr, cred); 1897 state = nfs4_do_open(dir, &path, flags, sattr, cred);
1900 put_rpccred(cred);
1901 d_drop(dentry); 1898 d_drop(dentry);
1902 if (IS_ERR(state)) { 1899 if (IS_ERR(state)) {
1903 status = PTR_ERR(state); 1900 status = PTR_ERR(state);
1904 goto out; 1901 goto out_putcred;
1905 } 1902 }
1906 d_add(dentry, igrab(state->inode)); 1903 d_add(dentry, igrab(state->inode));
1907 nfs_set_verifier(dentry, nfs_save_change_attribute(dir)); 1904 nfs_set_verifier(dentry, nfs_save_change_attribute(dir));
1908 if (flags & O_EXCL) { 1905 if (flags & O_EXCL) {
1909 struct nfs_fattr fattr; 1906 struct nfs_fattr fattr;
1910 status = nfs4_do_setattr(state->inode, &fattr, sattr, state); 1907 status = nfs4_do_setattr(state->inode, cred, &fattr, sattr, state);
1911 if (status == 0) 1908 if (status == 0)
1912 nfs_setattr_update_inode(state->inode, sattr); 1909 nfs_setattr_update_inode(state->inode, sattr);
1913 nfs_post_op_update_inode(state->inode, &fattr); 1910 nfs_post_op_update_inode(state->inode, &fattr);
@@ -1916,6 +1913,8 @@ nfs4_proc_create(struct inode *dir, struct dentry *dentry, struct iattr *sattr,
1916 status = nfs4_intent_set_file(nd, &path, state); 1913 status = nfs4_intent_set_file(nd, &path, state);
1917 else 1914 else
1918 nfs4_close_sync(&path, state, flags); 1915 nfs4_close_sync(&path, state, flags);
1916out_putcred:
1917 put_rpccred(cred);
1919out: 1918out:
1920 return status; 1919 return status;
1921} 1920}
@@ -2079,47 +2078,81 @@ static int nfs4_proc_link(struct inode *inode, struct inode *dir, struct qstr *n
2079 return err; 2078 return err;
2080} 2079}
2081 2080
2081struct nfs4_createdata {
2082 struct rpc_message msg;
2083 struct nfs4_create_arg arg;
2084 struct nfs4_create_res res;
2085 struct nfs_fh fh;
2086 struct nfs_fattr fattr;
2087 struct nfs_fattr dir_fattr;
2088};
2089
2090static struct nfs4_createdata *nfs4_alloc_createdata(struct inode *dir,
2091 struct qstr *name, struct iattr *sattr, u32 ftype)
2092{
2093 struct nfs4_createdata *data;
2094
2095 data = kzalloc(sizeof(*data), GFP_KERNEL);
2096 if (data != NULL) {
2097 struct nfs_server *server = NFS_SERVER(dir);
2098
2099 data->msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_CREATE];
2100 data->msg.rpc_argp = &data->arg;
2101 data->msg.rpc_resp = &data->res;
2102 data->arg.dir_fh = NFS_FH(dir);
2103 data->arg.server = server;
2104 data->arg.name = name;
2105 data->arg.attrs = sattr;
2106 data->arg.ftype = ftype;
2107 data->arg.bitmask = server->attr_bitmask;
2108 data->res.server = server;
2109 data->res.fh = &data->fh;
2110 data->res.fattr = &data->fattr;
2111 data->res.dir_fattr = &data->dir_fattr;
2112 nfs_fattr_init(data->res.fattr);
2113 nfs_fattr_init(data->res.dir_fattr);
2114 }
2115 return data;
2116}
2117
2118static int nfs4_do_create(struct inode *dir, struct dentry *dentry, struct nfs4_createdata *data)
2119{
2120 int status = rpc_call_sync(NFS_CLIENT(dir), &data->msg, 0);
2121 if (status == 0) {
2122 update_changeattr(dir, &data->res.dir_cinfo);
2123 nfs_post_op_update_inode(dir, data->res.dir_fattr);
2124 status = nfs_instantiate(dentry, data->res.fh, data->res.fattr);
2125 }
2126 return status;
2127}
2128
2129static void nfs4_free_createdata(struct nfs4_createdata *data)
2130{
2131 kfree(data);
2132}
2133
2082static int _nfs4_proc_symlink(struct inode *dir, struct dentry *dentry, 2134static int _nfs4_proc_symlink(struct inode *dir, struct dentry *dentry,
2083 struct page *page, unsigned int len, struct iattr *sattr) 2135 struct page *page, unsigned int len, struct iattr *sattr)
2084{ 2136{
2085 struct nfs_server *server = NFS_SERVER(dir); 2137 struct nfs4_createdata *data;
2086 struct nfs_fh fhandle; 2138 int status = -ENAMETOOLONG;
2087 struct nfs_fattr fattr, dir_fattr;
2088 struct nfs4_create_arg arg = {
2089 .dir_fh = NFS_FH(dir),
2090 .server = server,
2091 .name = &dentry->d_name,
2092 .attrs = sattr,
2093 .ftype = NF4LNK,
2094 .bitmask = server->attr_bitmask,
2095 };
2096 struct nfs4_create_res res = {
2097 .server = server,
2098 .fh = &fhandle,
2099 .fattr = &fattr,
2100 .dir_fattr = &dir_fattr,
2101 };
2102 struct rpc_message msg = {
2103 .rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SYMLINK],
2104 .rpc_argp = &arg,
2105 .rpc_resp = &res,
2106 };
2107 int status;
2108 2139
2109 if (len > NFS4_MAXPATHLEN) 2140 if (len > NFS4_MAXPATHLEN)
2110 return -ENAMETOOLONG; 2141 goto out;
2111 2142
2112 arg.u.symlink.pages = &page; 2143 status = -ENOMEM;
2113 arg.u.symlink.len = len; 2144 data = nfs4_alloc_createdata(dir, &dentry->d_name, sattr, NF4LNK);
2114 nfs_fattr_init(&fattr); 2145 if (data == NULL)
2115 nfs_fattr_init(&dir_fattr); 2146 goto out;
2147
2148 data->msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SYMLINK];
2149 data->arg.u.symlink.pages = &page;
2150 data->arg.u.symlink.len = len;
2116 2151
2117 status = rpc_call_sync(NFS_CLIENT(dir), &msg, 0); 2152 status = nfs4_do_create(dir, dentry, data);
2118 if (!status) { 2153
2119 update_changeattr(dir, &res.dir_cinfo); 2154 nfs4_free_createdata(data);
2120 nfs_post_op_update_inode(dir, res.dir_fattr); 2155out:
2121 status = nfs_instantiate(dentry, &fhandle, &fattr);
2122 }
2123 return status; 2156 return status;
2124} 2157}
2125 2158
@@ -2140,39 +2173,17 @@ static int nfs4_proc_symlink(struct inode *dir, struct dentry *dentry,
2140static int _nfs4_proc_mkdir(struct inode *dir, struct dentry *dentry, 2173static int _nfs4_proc_mkdir(struct inode *dir, struct dentry *dentry,
2141 struct iattr *sattr) 2174 struct iattr *sattr)
2142{ 2175{
2143 struct nfs_server *server = NFS_SERVER(dir); 2176 struct nfs4_createdata *data;
2144 struct nfs_fh fhandle; 2177 int status = -ENOMEM;
2145 struct nfs_fattr fattr, dir_fattr;
2146 struct nfs4_create_arg arg = {
2147 .dir_fh = NFS_FH(dir),
2148 .server = server,
2149 .name = &dentry->d_name,
2150 .attrs = sattr,
2151 .ftype = NF4DIR,
2152 .bitmask = server->attr_bitmask,
2153 };
2154 struct nfs4_create_res res = {
2155 .server = server,
2156 .fh = &fhandle,
2157 .fattr = &fattr,
2158 .dir_fattr = &dir_fattr,
2159 };
2160 struct rpc_message msg = {
2161 .rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_CREATE],
2162 .rpc_argp = &arg,
2163 .rpc_resp = &res,
2164 };
2165 int status;
2166 2178
2167 nfs_fattr_init(&fattr); 2179 data = nfs4_alloc_createdata(dir, &dentry->d_name, sattr, NF4DIR);
2168 nfs_fattr_init(&dir_fattr); 2180 if (data == NULL)
2169 2181 goto out;
2170 status = rpc_call_sync(NFS_CLIENT(dir), &msg, 0); 2182
2171 if (!status) { 2183 status = nfs4_do_create(dir, dentry, data);
2172 update_changeattr(dir, &res.dir_cinfo); 2184
2173 nfs_post_op_update_inode(dir, res.dir_fattr); 2185 nfs4_free_createdata(data);
2174 status = nfs_instantiate(dentry, &fhandle, &fattr); 2186out:
2175 }
2176 return status; 2187 return status;
2177} 2188}
2178 2189
@@ -2242,56 +2253,34 @@ static int nfs4_proc_readdir(struct dentry *dentry, struct rpc_cred *cred,
2242static int _nfs4_proc_mknod(struct inode *dir, struct dentry *dentry, 2253static int _nfs4_proc_mknod(struct inode *dir, struct dentry *dentry,
2243 struct iattr *sattr, dev_t rdev) 2254 struct iattr *sattr, dev_t rdev)
2244{ 2255{
2245 struct nfs_server *server = NFS_SERVER(dir); 2256 struct nfs4_createdata *data;
2246 struct nfs_fh fh; 2257 int mode = sattr->ia_mode;
2247 struct nfs_fattr fattr, dir_fattr; 2258 int status = -ENOMEM;
2248 struct nfs4_create_arg arg = {
2249 .dir_fh = NFS_FH(dir),
2250 .server = server,
2251 .name = &dentry->d_name,
2252 .attrs = sattr,
2253 .bitmask = server->attr_bitmask,
2254 };
2255 struct nfs4_create_res res = {
2256 .server = server,
2257 .fh = &fh,
2258 .fattr = &fattr,
2259 .dir_fattr = &dir_fattr,
2260 };
2261 struct rpc_message msg = {
2262 .rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_CREATE],
2263 .rpc_argp = &arg,
2264 .rpc_resp = &res,
2265 };
2266 int status;
2267 int mode = sattr->ia_mode;
2268
2269 nfs_fattr_init(&fattr);
2270 nfs_fattr_init(&dir_fattr);
2271 2259
2272 BUG_ON(!(sattr->ia_valid & ATTR_MODE)); 2260 BUG_ON(!(sattr->ia_valid & ATTR_MODE));
2273 BUG_ON(!S_ISFIFO(mode) && !S_ISBLK(mode) && !S_ISCHR(mode) && !S_ISSOCK(mode)); 2261 BUG_ON(!S_ISFIFO(mode) && !S_ISBLK(mode) && !S_ISCHR(mode) && !S_ISSOCK(mode));
2262
2263 data = nfs4_alloc_createdata(dir, &dentry->d_name, sattr, NF4SOCK);
2264 if (data == NULL)
2265 goto out;
2266
2274 if (S_ISFIFO(mode)) 2267 if (S_ISFIFO(mode))
2275 arg.ftype = NF4FIFO; 2268 data->arg.ftype = NF4FIFO;
2276 else if (S_ISBLK(mode)) { 2269 else if (S_ISBLK(mode)) {
2277 arg.ftype = NF4BLK; 2270 data->arg.ftype = NF4BLK;
2278 arg.u.device.specdata1 = MAJOR(rdev); 2271 data->arg.u.device.specdata1 = MAJOR(rdev);
2279 arg.u.device.specdata2 = MINOR(rdev); 2272 data->arg.u.device.specdata2 = MINOR(rdev);
2280 } 2273 }
2281 else if (S_ISCHR(mode)) { 2274 else if (S_ISCHR(mode)) {
2282 arg.ftype = NF4CHR; 2275 data->arg.ftype = NF4CHR;
2283 arg.u.device.specdata1 = MAJOR(rdev); 2276 data->arg.u.device.specdata1 = MAJOR(rdev);
2284 arg.u.device.specdata2 = MINOR(rdev); 2277 data->arg.u.device.specdata2 = MINOR(rdev);
2285 } 2278 }
2286 else
2287 arg.ftype = NF4SOCK;
2288 2279
2289 status = rpc_call_sync(NFS_CLIENT(dir), &msg, 0); 2280 status = nfs4_do_create(dir, dentry, data);
2290 if (status == 0) { 2281
2291 update_changeattr(dir, &res.dir_cinfo); 2282 nfs4_free_createdata(data);
2292 nfs_post_op_update_inode(dir, res.dir_fattr); 2283out:
2293 status = nfs_instantiate(dentry, &fh, &fattr);
2294 }
2295 return status; 2284 return status;
2296} 2285}
2297 2286
@@ -2706,6 +2695,8 @@ static ssize_t nfs4_proc_get_acl(struct inode *inode, void *buf, size_t buflen)
2706 ret = nfs_revalidate_inode(server, inode); 2695 ret = nfs_revalidate_inode(server, inode);
2707 if (ret < 0) 2696 if (ret < 0)
2708 return ret; 2697 return ret;
2698 if (NFS_I(inode)->cache_validity & NFS_INO_INVALID_ACL)
2699 nfs_zap_acl_cache(inode);
2709 ret = nfs4_read_cached_acl(inode, buf, buflen); 2700 ret = nfs4_read_cached_acl(inode, buf, buflen);
2710 if (ret != -ENOENT) 2701 if (ret != -ENOENT)
2711 return ret; 2702 return ret;
@@ -2733,7 +2724,8 @@ static int __nfs4_proc_set_acl(struct inode *inode, const void *buf, size_t bufl
2733 nfs_inode_return_delegation(inode); 2724 nfs_inode_return_delegation(inode);
2734 buf_to_pages(buf, buflen, arg.acl_pages, &arg.acl_pgbase); 2725 buf_to_pages(buf, buflen, arg.acl_pages, &arg.acl_pgbase);
2735 ret = rpc_call_sync(NFS_CLIENT(inode), &msg, 0); 2726 ret = rpc_call_sync(NFS_CLIENT(inode), &msg, 0);
2736 nfs_zap_caches(inode); 2727 nfs_access_zap_cache(inode);
2728 nfs_zap_acl_cache(inode);
2737 return ret; 2729 return ret;
2738} 2730}
2739 2731
@@ -2767,8 +2759,7 @@ nfs4_async_handle_error(struct rpc_task *task, const struct nfs_server *server)
2767 task->tk_status = 0; 2759 task->tk_status = 0;
2768 return -EAGAIN; 2760 return -EAGAIN;
2769 case -NFS4ERR_DELAY: 2761 case -NFS4ERR_DELAY:
2770 nfs_inc_server_stats((struct nfs_server *) server, 2762 nfs_inc_server_stats(server, NFSIOS_DELAY);
2771 NFSIOS_DELAY);
2772 case -NFS4ERR_GRACE: 2763 case -NFS4ERR_GRACE:
2773 rpc_delay(task, NFS4_POLL_RETRY_MAX); 2764 rpc_delay(task, NFS4_POLL_RETRY_MAX);
2774 task->tk_status = 0; 2765 task->tk_status = 0;
@@ -2933,7 +2924,7 @@ static int _nfs4_proc_setclientid_confirm(struct nfs_client *clp, struct rpc_cre
2933 2924
2934int nfs4_proc_setclientid_confirm(struct nfs_client *clp, struct rpc_cred *cred) 2925int nfs4_proc_setclientid_confirm(struct nfs_client *clp, struct rpc_cred *cred)
2935{ 2926{
2936 long timeout; 2927 long timeout = 0;
2937 int err; 2928 int err;
2938 do { 2929 do {
2939 err = _nfs4_proc_setclientid_confirm(clp, cred); 2930 err = _nfs4_proc_setclientid_confirm(clp, cred);
@@ -3725,8 +3716,6 @@ const struct nfs_rpc_ops nfs_v4_clientops = {
3725 .write_done = nfs4_write_done, 3716 .write_done = nfs4_write_done,
3726 .commit_setup = nfs4_proc_commit_setup, 3717 .commit_setup = nfs4_proc_commit_setup,
3727 .commit_done = nfs4_commit_done, 3718 .commit_done = nfs4_commit_done,
3728 .file_open = nfs_open,
3729 .file_release = nfs_release,
3730 .lock = nfs4_proc_lock, 3719 .lock = nfs4_proc_lock,
3731 .clear_acl_cache = nfs4_zap_acl_attr, 3720 .clear_acl_cache = nfs4_zap_acl_attr,
3732}; 3721};
diff --git a/fs/nfs/nfsroot.c b/fs/nfs/nfsroot.c
index 531379d36823..46763d1cd397 100644
--- a/fs/nfs/nfsroot.c
+++ b/fs/nfs/nfsroot.c
@@ -1,6 +1,4 @@
1/* 1/*
2 * $Id: nfsroot.c,v 1.45 1998/03/07 10:44:46 mj Exp $
3 *
4 * Copyright (C) 1995, 1996 Gero Kuhlmann <gero@gkminix.han.de> 2 * Copyright (C) 1995, 1996 Gero Kuhlmann <gero@gkminix.han.de>
5 * 3 *
6 * Allow an NFS filesystem to be mounted as root. The way this works is: 4 * Allow an NFS filesystem to be mounted as root. The way this works is:
@@ -297,10 +295,10 @@ static int __init root_nfs_name(char *name)
297 nfs_data.flags = NFS_MOUNT_NONLM; /* No lockd in nfs root yet */ 295 nfs_data.flags = NFS_MOUNT_NONLM; /* No lockd in nfs root yet */
298 nfs_data.rsize = NFS_DEF_FILE_IO_SIZE; 296 nfs_data.rsize = NFS_DEF_FILE_IO_SIZE;
299 nfs_data.wsize = NFS_DEF_FILE_IO_SIZE; 297 nfs_data.wsize = NFS_DEF_FILE_IO_SIZE;
300 nfs_data.acregmin = 3; 298 nfs_data.acregmin = NFS_DEF_ACREGMIN;
301 nfs_data.acregmax = 60; 299 nfs_data.acregmax = NFS_DEF_ACREGMAX;
302 nfs_data.acdirmin = 30; 300 nfs_data.acdirmin = NFS_DEF_ACDIRMIN;
303 nfs_data.acdirmax = 60; 301 nfs_data.acdirmax = NFS_DEF_ACDIRMAX;
304 strcpy(buf, NFS_ROOT); 302 strcpy(buf, NFS_ROOT);
305 303
306 /* Process options received from the remote server */ 304 /* Process options received from the remote server */
diff --git a/fs/nfs/proc.c b/fs/nfs/proc.c
index 03599bfe81cf..4dbb84df1b68 100644
--- a/fs/nfs/proc.c
+++ b/fs/nfs/proc.c
@@ -129,6 +129,8 @@ nfs_proc_setattr(struct dentry *dentry, struct nfs_fattr *fattr,
129 sattr->ia_mode &= S_IALLUGO; 129 sattr->ia_mode &= S_IALLUGO;
130 130
131 dprintk("NFS call setattr\n"); 131 dprintk("NFS call setattr\n");
132 if (sattr->ia_valid & ATTR_FILE)
133 msg.rpc_cred = nfs_file_cred(sattr->ia_file);
132 nfs_fattr_init(fattr); 134 nfs_fattr_init(fattr);
133 status = rpc_call_sync(NFS_CLIENT(inode), &msg, 0); 135 status = rpc_call_sync(NFS_CLIENT(inode), &msg, 0);
134 if (status == 0) 136 if (status == 0)
@@ -598,6 +600,29 @@ nfs_proc_lock(struct file *filp, int cmd, struct file_lock *fl)
598 return nlmclnt_proc(NFS_SERVER(inode)->nlm_host, cmd, fl); 600 return nlmclnt_proc(NFS_SERVER(inode)->nlm_host, cmd, fl);
599} 601}
600 602
603/* Helper functions for NFS lock bounds checking */
604#define NFS_LOCK32_OFFSET_MAX ((__s32)0x7fffffffUL)
605static int nfs_lock_check_bounds(const struct file_lock *fl)
606{
607 __s32 start, end;
608
609 start = (__s32)fl->fl_start;
610 if ((loff_t)start != fl->fl_start)
611 goto out_einval;
612
613 if (fl->fl_end != OFFSET_MAX) {
614 end = (__s32)fl->fl_end;
615 if ((loff_t)end != fl->fl_end)
616 goto out_einval;
617 } else
618 end = NFS_LOCK32_OFFSET_MAX;
619
620 if (start < 0 || start > end)
621 goto out_einval;
622 return 0;
623out_einval:
624 return -EINVAL;
625}
601 626
602const struct nfs_rpc_ops nfs_v2_clientops = { 627const struct nfs_rpc_ops nfs_v2_clientops = {
603 .version = 2, /* protocol version */ 628 .version = 2, /* protocol version */
@@ -630,7 +655,6 @@ const struct nfs_rpc_ops nfs_v2_clientops = {
630 .write_setup = nfs_proc_write_setup, 655 .write_setup = nfs_proc_write_setup,
631 .write_done = nfs_write_done, 656 .write_done = nfs_write_done,
632 .commit_setup = nfs_proc_commit_setup, 657 .commit_setup = nfs_proc_commit_setup,
633 .file_open = nfs_open,
634 .file_release = nfs_release,
635 .lock = nfs_proc_lock, 658 .lock = nfs_proc_lock,
659 .lock_check_bounds = nfs_lock_check_bounds,
636}; 660};
diff --git a/fs/nfs/super.c b/fs/nfs/super.c
index 614efeed5437..47cf83e917be 100644
--- a/fs/nfs/super.c
+++ b/fs/nfs/super.c
@@ -47,6 +47,7 @@
47#include <linux/inet.h> 47#include <linux/inet.h>
48#include <linux/in6.h> 48#include <linux/in6.h>
49#include <net/ipv6.h> 49#include <net/ipv6.h>
50#include <linux/netdevice.h>
50#include <linux/nfs_xdr.h> 51#include <linux/nfs_xdr.h>
51#include <linux/magic.h> 52#include <linux/magic.h>
52#include <linux/parser.h> 53#include <linux/parser.h>
@@ -65,7 +66,6 @@
65enum { 66enum {
66 /* Mount options that take no arguments */ 67 /* Mount options that take no arguments */
67 Opt_soft, Opt_hard, 68 Opt_soft, Opt_hard,
68 Opt_intr, Opt_nointr,
69 Opt_posix, Opt_noposix, 69 Opt_posix, Opt_noposix,
70 Opt_cto, Opt_nocto, 70 Opt_cto, Opt_nocto,
71 Opt_ac, Opt_noac, 71 Opt_ac, Opt_noac,
@@ -92,8 +92,8 @@ enum {
92 Opt_sec, Opt_proto, Opt_mountproto, Opt_mounthost, 92 Opt_sec, Opt_proto, Opt_mountproto, Opt_mounthost,
93 Opt_addr, Opt_mountaddr, Opt_clientaddr, 93 Opt_addr, Opt_mountaddr, Opt_clientaddr,
94 94
95 /* Mount options that are ignored */ 95 /* Special mount options */
96 Opt_userspace, Opt_deprecated, 96 Opt_userspace, Opt_deprecated, Opt_sloppy,
97 97
98 Opt_err 98 Opt_err
99}; 99};
@@ -101,10 +101,14 @@ enum {
101static match_table_t nfs_mount_option_tokens = { 101static match_table_t nfs_mount_option_tokens = {
102 { Opt_userspace, "bg" }, 102 { Opt_userspace, "bg" },
103 { Opt_userspace, "fg" }, 103 { Opt_userspace, "fg" },
104 { Opt_userspace, "retry=%s" },
105
106 { Opt_sloppy, "sloppy" },
107
104 { Opt_soft, "soft" }, 108 { Opt_soft, "soft" },
105 { Opt_hard, "hard" }, 109 { Opt_hard, "hard" },
106 { Opt_intr, "intr" }, 110 { Opt_deprecated, "intr" },
107 { Opt_nointr, "nointr" }, 111 { Opt_deprecated, "nointr" },
108 { Opt_posix, "posix" }, 112 { Opt_posix, "posix" },
109 { Opt_noposix, "noposix" }, 113 { Opt_noposix, "noposix" },
110 { Opt_cto, "cto" }, 114 { Opt_cto, "cto" },
@@ -136,7 +140,6 @@ static match_table_t nfs_mount_option_tokens = {
136 { Opt_acdirmin, "acdirmin=%u" }, 140 { Opt_acdirmin, "acdirmin=%u" },
137 { Opt_acdirmax, "acdirmax=%u" }, 141 { Opt_acdirmax, "acdirmax=%u" },
138 { Opt_actimeo, "actimeo=%u" }, 142 { Opt_actimeo, "actimeo=%u" },
139 { Opt_userspace, "retry=%u" },
140 { Opt_namelen, "namlen=%u" }, 143 { Opt_namelen, "namlen=%u" },
141 { Opt_mountport, "mountport=%u" }, 144 { Opt_mountport, "mountport=%u" },
142 { Opt_mountvers, "mountvers=%u" }, 145 { Opt_mountvers, "mountvers=%u" },
@@ -207,6 +210,7 @@ static int nfs_xdev_get_sb(struct file_system_type *fs_type,
207 int flags, const char *dev_name, void *raw_data, struct vfsmount *mnt); 210 int flags, const char *dev_name, void *raw_data, struct vfsmount *mnt);
208static void nfs_kill_super(struct super_block *); 211static void nfs_kill_super(struct super_block *);
209static void nfs_put_super(struct super_block *); 212static void nfs_put_super(struct super_block *);
213static int nfs_remount(struct super_block *sb, int *flags, char *raw_data);
210 214
211static struct file_system_type nfs_fs_type = { 215static struct file_system_type nfs_fs_type = {
212 .owner = THIS_MODULE, 216 .owner = THIS_MODULE,
@@ -234,6 +238,7 @@ static const struct super_operations nfs_sops = {
234 .umount_begin = nfs_umount_begin, 238 .umount_begin = nfs_umount_begin,
235 .show_options = nfs_show_options, 239 .show_options = nfs_show_options,
236 .show_stats = nfs_show_stats, 240 .show_stats = nfs_show_stats,
241 .remount_fs = nfs_remount,
237}; 242};
238 243
239#ifdef CONFIG_NFS_V4 244#ifdef CONFIG_NFS_V4
@@ -278,6 +283,7 @@ static const struct super_operations nfs4_sops = {
278 .umount_begin = nfs_umount_begin, 283 .umount_begin = nfs_umount_begin,
279 .show_options = nfs_show_options, 284 .show_options = nfs_show_options,
280 .show_stats = nfs_show_stats, 285 .show_stats = nfs_show_stats,
286 .remount_fs = nfs_remount,
281}; 287};
282#endif 288#endif
283 289
@@ -514,13 +520,13 @@ static void nfs_show_mount_options(struct seq_file *m, struct nfs_server *nfss,
514 if (nfss->bsize != 0) 520 if (nfss->bsize != 0)
515 seq_printf(m, ",bsize=%u", nfss->bsize); 521 seq_printf(m, ",bsize=%u", nfss->bsize);
516 seq_printf(m, ",namlen=%u", nfss->namelen); 522 seq_printf(m, ",namlen=%u", nfss->namelen);
517 if (nfss->acregmin != 3*HZ || showdefaults) 523 if (nfss->acregmin != NFS_DEF_ACREGMIN*HZ || showdefaults)
518 seq_printf(m, ",acregmin=%u", nfss->acregmin/HZ); 524 seq_printf(m, ",acregmin=%u", nfss->acregmin/HZ);
519 if (nfss->acregmax != 60*HZ || showdefaults) 525 if (nfss->acregmax != NFS_DEF_ACREGMAX*HZ || showdefaults)
520 seq_printf(m, ",acregmax=%u", nfss->acregmax/HZ); 526 seq_printf(m, ",acregmax=%u", nfss->acregmax/HZ);
521 if (nfss->acdirmin != 30*HZ || showdefaults) 527 if (nfss->acdirmin != NFS_DEF_ACDIRMIN*HZ || showdefaults)
522 seq_printf(m, ",acdirmin=%u", nfss->acdirmin/HZ); 528 seq_printf(m, ",acdirmin=%u", nfss->acdirmin/HZ);
523 if (nfss->acdirmax != 60*HZ || showdefaults) 529 if (nfss->acdirmax != NFS_DEF_ACDIRMAX*HZ || showdefaults)
524 seq_printf(m, ",acdirmax=%u", nfss->acdirmax/HZ); 530 seq_printf(m, ",acdirmax=%u", nfss->acdirmax/HZ);
525 for (nfs_infop = nfs_info; nfs_infop->flag; nfs_infop++) { 531 for (nfs_infop = nfs_info; nfs_infop->flag; nfs_infop++) {
526 if (nfss->flags & nfs_infop->flag) 532 if (nfss->flags & nfs_infop->flag)
@@ -702,49 +708,233 @@ static int nfs_verify_server_address(struct sockaddr *addr)
702 return 0; 708 return 0;
703} 709}
704 710
711static void nfs_parse_ipv4_address(char *string, size_t str_len,
712 struct sockaddr *sap, size_t *addr_len)
713{
714 struct sockaddr_in *sin = (struct sockaddr_in *)sap;
715 u8 *addr = (u8 *)&sin->sin_addr.s_addr;
716
717 if (str_len <= INET_ADDRSTRLEN) {
718 dfprintk(MOUNT, "NFS: parsing IPv4 address %*s\n",
719 (int)str_len, string);
720
721 sin->sin_family = AF_INET;
722 *addr_len = sizeof(*sin);
723 if (in4_pton(string, str_len, addr, '\0', NULL))
724 return;
725 }
726
727 sap->sa_family = AF_UNSPEC;
728 *addr_len = 0;
729}
730
731#define IPV6_SCOPE_DELIMITER '%'
732
733#if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
734static void nfs_parse_ipv6_scope_id(const char *string, const size_t str_len,
735 const char *delim,
736 struct sockaddr_in6 *sin6)
737{
738 char *p;
739 size_t len;
740
741 if (!(ipv6_addr_type(&sin6->sin6_addr) & IPV6_ADDR_LINKLOCAL))
742 return ;
743 if (*delim != IPV6_SCOPE_DELIMITER)
744 return;
745
746 len = (string + str_len) - delim - 1;
747 p = kstrndup(delim + 1, len, GFP_KERNEL);
748 if (p) {
749 unsigned long scope_id = 0;
750 struct net_device *dev;
751
752 dev = dev_get_by_name(&init_net, p);
753 if (dev != NULL) {
754 scope_id = dev->ifindex;
755 dev_put(dev);
756 } else {
757 /* scope_id is set to zero on error */
758 strict_strtoul(p, 10, &scope_id);
759 }
760
761 kfree(p);
762 sin6->sin6_scope_id = scope_id;
763 dfprintk(MOUNT, "NFS: IPv6 scope ID = %lu\n", scope_id);
764 }
765}
766
767static void nfs_parse_ipv6_address(char *string, size_t str_len,
768 struct sockaddr *sap, size_t *addr_len)
769{
770 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)sap;
771 u8 *addr = (u8 *)&sin6->sin6_addr.in6_u;
772 const char *delim;
773
774 if (str_len <= INET6_ADDRSTRLEN) {
775 dfprintk(MOUNT, "NFS: parsing IPv6 address %*s\n",
776 (int)str_len, string);
777
778 sin6->sin6_family = AF_INET6;
779 *addr_len = sizeof(*sin6);
780 if (in6_pton(string, str_len, addr, IPV6_SCOPE_DELIMITER, &delim)) {
781 nfs_parse_ipv6_scope_id(string, str_len, delim, sin6);
782 return;
783 }
784 }
785
786 sap->sa_family = AF_UNSPEC;
787 *addr_len = 0;
788}
789#else
790static void nfs_parse_ipv6_address(char *string, size_t str_len,
791 struct sockaddr *sap, size_t *addr_len)
792{
793 sap->sa_family = AF_UNSPEC;
794 *addr_len = 0;
795}
796#endif
797
705/* 798/*
706 * Parse string addresses passed in via a mount option, 799 * Construct a sockaddr based on the contents of a string that contains
707 * and construct a sockaddr based on the result. 800 * an IP address in presentation format.
708 * 801 *
709 * If address parsing fails, set the sockaddr's address 802 * If there is a problem constructing the new sockaddr, set the address
710 * family to AF_UNSPEC to force nfs_verify_server_address() 803 * family to AF_UNSPEC.
711 * to punt the mount.
712 */ 804 */
713static void nfs_parse_server_address(char *value, 805static void nfs_parse_ip_address(char *string, size_t str_len,
714 struct sockaddr *sap, 806 struct sockaddr *sap, size_t *addr_len)
715 size_t *len)
716{ 807{
717 if (strchr(value, ':')) { 808 unsigned int i, colons;
718 struct sockaddr_in6 *ap = (struct sockaddr_in6 *)sap;
719 u8 *addr = (u8 *)&ap->sin6_addr.in6_u;
720 809
721 ap->sin6_family = AF_INET6; 810 colons = 0;
722 *len = sizeof(*ap); 811 for (i = 0; i < str_len; i++)
723 if (in6_pton(value, -1, addr, '\0', NULL)) 812 if (string[i] == ':')
724 return; 813 colons++;
725 } else {
726 struct sockaddr_in *ap = (struct sockaddr_in *)sap;
727 u8 *addr = (u8 *)&ap->sin_addr.s_addr;
728 814
729 ap->sin_family = AF_INET; 815 if (colons >= 2)
730 *len = sizeof(*ap); 816 nfs_parse_ipv6_address(string, str_len, sap, addr_len);
731 if (in4_pton(value, -1, addr, '\0', NULL)) 817 else
818 nfs_parse_ipv4_address(string, str_len, sap, addr_len);
819}
820
821/*
822 * Sanity check the NFS transport protocol.
823 *
824 */
825static void nfs_validate_transport_protocol(struct nfs_parsed_mount_data *mnt)
826{
827 switch (mnt->nfs_server.protocol) {
828 case XPRT_TRANSPORT_UDP:
829 case XPRT_TRANSPORT_TCP:
830 case XPRT_TRANSPORT_RDMA:
831 break;
832 default:
833 mnt->nfs_server.protocol = XPRT_TRANSPORT_TCP;
834 }
835}
836
837/*
838 * For text based NFSv2/v3 mounts, the mount protocol transport default
839 * settings should depend upon the specified NFS transport.
840 */
841static void nfs_set_mount_transport_protocol(struct nfs_parsed_mount_data *mnt)
842{
843 nfs_validate_transport_protocol(mnt);
844
845 if (mnt->mount_server.protocol == XPRT_TRANSPORT_UDP ||
846 mnt->mount_server.protocol == XPRT_TRANSPORT_TCP)
732 return; 847 return;
848 switch (mnt->nfs_server.protocol) {
849 case XPRT_TRANSPORT_UDP:
850 mnt->mount_server.protocol = XPRT_TRANSPORT_UDP;
851 break;
852 case XPRT_TRANSPORT_TCP:
853 case XPRT_TRANSPORT_RDMA:
854 mnt->mount_server.protocol = XPRT_TRANSPORT_TCP;
733 } 855 }
856}
734 857
735 sap->sa_family = AF_UNSPEC; 858/*
736 *len = 0; 859 * Parse the value of the 'sec=' option.
860 *
861 * The flavor_len setting is for v4 mounts.
862 */
863static int nfs_parse_security_flavors(char *value,
864 struct nfs_parsed_mount_data *mnt)
865{
866 substring_t args[MAX_OPT_ARGS];
867
868 dfprintk(MOUNT, "NFS: parsing sec=%s option\n", value);
869
870 switch (match_token(value, nfs_secflavor_tokens, args)) {
871 case Opt_sec_none:
872 mnt->auth_flavor_len = 0;
873 mnt->auth_flavors[0] = RPC_AUTH_NULL;
874 break;
875 case Opt_sec_sys:
876 mnt->auth_flavor_len = 0;
877 mnt->auth_flavors[0] = RPC_AUTH_UNIX;
878 break;
879 case Opt_sec_krb5:
880 mnt->auth_flavor_len = 1;
881 mnt->auth_flavors[0] = RPC_AUTH_GSS_KRB5;
882 break;
883 case Opt_sec_krb5i:
884 mnt->auth_flavor_len = 1;
885 mnt->auth_flavors[0] = RPC_AUTH_GSS_KRB5I;
886 break;
887 case Opt_sec_krb5p:
888 mnt->auth_flavor_len = 1;
889 mnt->auth_flavors[0] = RPC_AUTH_GSS_KRB5P;
890 break;
891 case Opt_sec_lkey:
892 mnt->auth_flavor_len = 1;
893 mnt->auth_flavors[0] = RPC_AUTH_GSS_LKEY;
894 break;
895 case Opt_sec_lkeyi:
896 mnt->auth_flavor_len = 1;
897 mnt->auth_flavors[0] = RPC_AUTH_GSS_LKEYI;
898 break;
899 case Opt_sec_lkeyp:
900 mnt->auth_flavor_len = 1;
901 mnt->auth_flavors[0] = RPC_AUTH_GSS_LKEYP;
902 break;
903 case Opt_sec_spkm:
904 mnt->auth_flavor_len = 1;
905 mnt->auth_flavors[0] = RPC_AUTH_GSS_SPKM;
906 break;
907 case Opt_sec_spkmi:
908 mnt->auth_flavor_len = 1;
909 mnt->auth_flavors[0] = RPC_AUTH_GSS_SPKMI;
910 break;
911 case Opt_sec_spkmp:
912 mnt->auth_flavor_len = 1;
913 mnt->auth_flavors[0] = RPC_AUTH_GSS_SPKMP;
914 break;
915 default:
916 return 0;
917 }
918
919 return 1;
920}
921
922static void nfs_parse_invalid_value(const char *option)
923{
924 dfprintk(MOUNT, "NFS: bad value specified for %s option\n", option);
737} 925}
738 926
739/* 927/*
740 * Error-check and convert a string of mount options from user space into 928 * Error-check and convert a string of mount options from user space into
741 * a data structure 929 * a data structure. The whole mount string is processed; bad options are
930 * skipped as they are encountered. If there were no errors, return 1;
931 * otherwise return 0 (zero).
742 */ 932 */
743static int nfs_parse_mount_options(char *raw, 933static int nfs_parse_mount_options(char *raw,
744 struct nfs_parsed_mount_data *mnt) 934 struct nfs_parsed_mount_data *mnt)
745{ 935{
746 char *p, *string, *secdata; 936 char *p, *string, *secdata;
747 int rc; 937 int rc, sloppy = 0, errors = 0;
748 938
749 if (!raw) { 939 if (!raw) {
750 dfprintk(MOUNT, "NFS: mount options string was NULL.\n"); 940 dfprintk(MOUNT, "NFS: mount options string was NULL.\n");
@@ -777,15 +967,16 @@ static int nfs_parse_mount_options(char *raw,
777 967
778 token = match_token(p, nfs_mount_option_tokens, args); 968 token = match_token(p, nfs_mount_option_tokens, args);
779 switch (token) { 969 switch (token) {
970
971 /*
972 * boolean options: foo/nofoo
973 */
780 case Opt_soft: 974 case Opt_soft:
781 mnt->flags |= NFS_MOUNT_SOFT; 975 mnt->flags |= NFS_MOUNT_SOFT;
782 break; 976 break;
783 case Opt_hard: 977 case Opt_hard:
784 mnt->flags &= ~NFS_MOUNT_SOFT; 978 mnt->flags &= ~NFS_MOUNT_SOFT;
785 break; 979 break;
786 case Opt_intr:
787 case Opt_nointr:
788 break;
789 case Opt_posix: 980 case Opt_posix:
790 mnt->flags |= NFS_MOUNT_POSIX; 981 mnt->flags |= NFS_MOUNT_POSIX;
791 break; 982 break;
@@ -819,20 +1010,14 @@ static int nfs_parse_mount_options(char *raw,
819 case Opt_udp: 1010 case Opt_udp:
820 mnt->flags &= ~NFS_MOUNT_TCP; 1011 mnt->flags &= ~NFS_MOUNT_TCP;
821 mnt->nfs_server.protocol = XPRT_TRANSPORT_UDP; 1012 mnt->nfs_server.protocol = XPRT_TRANSPORT_UDP;
822 mnt->timeo = 7;
823 mnt->retrans = 5;
824 break; 1013 break;
825 case Opt_tcp: 1014 case Opt_tcp:
826 mnt->flags |= NFS_MOUNT_TCP; 1015 mnt->flags |= NFS_MOUNT_TCP;
827 mnt->nfs_server.protocol = XPRT_TRANSPORT_TCP; 1016 mnt->nfs_server.protocol = XPRT_TRANSPORT_TCP;
828 mnt->timeo = 600;
829 mnt->retrans = 2;
830 break; 1017 break;
831 case Opt_rdma: 1018 case Opt_rdma:
832 mnt->flags |= NFS_MOUNT_TCP; /* for side protocols */ 1019 mnt->flags |= NFS_MOUNT_TCP; /* for side protocols */
833 mnt->nfs_server.protocol = XPRT_TRANSPORT_RDMA; 1020 mnt->nfs_server.protocol = XPRT_TRANSPORT_RDMA;
834 mnt->timeo = 600;
835 mnt->retrans = 2;
836 break; 1021 break;
837 case Opt_acl: 1022 case Opt_acl:
838 mnt->flags &= ~NFS_MOUNT_NOACL; 1023 mnt->flags &= ~NFS_MOUNT_NOACL;
@@ -853,165 +1038,144 @@ static int nfs_parse_mount_options(char *raw,
853 mnt->flags |= NFS_MOUNT_UNSHARED; 1038 mnt->flags |= NFS_MOUNT_UNSHARED;
854 break; 1039 break;
855 1040
1041 /*
1042 * options that take numeric values
1043 */
856 case Opt_port: 1044 case Opt_port:
857 if (match_int(args, &option)) 1045 if (match_int(args, &option) ||
858 return 0; 1046 option < 0 || option > USHORT_MAX) {
859 if (option < 0 || option > 65535) 1047 errors++;
860 return 0; 1048 nfs_parse_invalid_value("port");
861 mnt->nfs_server.port = option; 1049 } else
1050 mnt->nfs_server.port = option;
862 break; 1051 break;
863 case Opt_rsize: 1052 case Opt_rsize:
864 if (match_int(args, &mnt->rsize)) 1053 if (match_int(args, &option) || option < 0) {
865 return 0; 1054 errors++;
1055 nfs_parse_invalid_value("rsize");
1056 } else
1057 mnt->rsize = option;
866 break; 1058 break;
867 case Opt_wsize: 1059 case Opt_wsize:
868 if (match_int(args, &mnt->wsize)) 1060 if (match_int(args, &option) || option < 0) {
869 return 0; 1061 errors++;
1062 nfs_parse_invalid_value("wsize");
1063 } else
1064 mnt->wsize = option;
870 break; 1065 break;
871 case Opt_bsize: 1066 case Opt_bsize:
872 if (match_int(args, &option)) 1067 if (match_int(args, &option) || option < 0) {
873 return 0; 1068 errors++;
874 if (option < 0) 1069 nfs_parse_invalid_value("bsize");
875 return 0; 1070 } else
876 mnt->bsize = option; 1071 mnt->bsize = option;
877 break; 1072 break;
878 case Opt_timeo: 1073 case Opt_timeo:
879 if (match_int(args, &mnt->timeo)) 1074 if (match_int(args, &option) || option <= 0) {
880 return 0; 1075 errors++;
1076 nfs_parse_invalid_value("timeo");
1077 } else
1078 mnt->timeo = option;
881 break; 1079 break;
882 case Opt_retrans: 1080 case Opt_retrans:
883 if (match_int(args, &mnt->retrans)) 1081 if (match_int(args, &option) || option <= 0) {
884 return 0; 1082 errors++;
1083 nfs_parse_invalid_value("retrans");
1084 } else
1085 mnt->retrans = option;
885 break; 1086 break;
886 case Opt_acregmin: 1087 case Opt_acregmin:
887 if (match_int(args, &mnt->acregmin)) 1088 if (match_int(args, &option) || option < 0) {
888 return 0; 1089 errors++;
1090 nfs_parse_invalid_value("acregmin");
1091 } else
1092 mnt->acregmin = option;
889 break; 1093 break;
890 case Opt_acregmax: 1094 case Opt_acregmax:
891 if (match_int(args, &mnt->acregmax)) 1095 if (match_int(args, &option) || option < 0) {
892 return 0; 1096 errors++;
1097 nfs_parse_invalid_value("acregmax");
1098 } else
1099 mnt->acregmax = option;
893 break; 1100 break;
894 case Opt_acdirmin: 1101 case Opt_acdirmin:
895 if (match_int(args, &mnt->acdirmin)) 1102 if (match_int(args, &option) || option < 0) {
896 return 0; 1103 errors++;
1104 nfs_parse_invalid_value("acdirmin");
1105 } else
1106 mnt->acdirmin = option;
897 break; 1107 break;
898 case Opt_acdirmax: 1108 case Opt_acdirmax:
899 if (match_int(args, &mnt->acdirmax)) 1109 if (match_int(args, &option) || option < 0) {
900 return 0; 1110 errors++;
1111 nfs_parse_invalid_value("acdirmax");
1112 } else
1113 mnt->acdirmax = option;
901 break; 1114 break;
902 case Opt_actimeo: 1115 case Opt_actimeo:
903 if (match_int(args, &option)) 1116 if (match_int(args, &option) || option < 0) {
904 return 0; 1117 errors++;
905 if (option < 0) 1118 nfs_parse_invalid_value("actimeo");
906 return 0; 1119 } else
907 mnt->acregmin = 1120 mnt->acregmin = mnt->acregmax =
908 mnt->acregmax = 1121 mnt->acdirmin = mnt->acdirmax = option;
909 mnt->acdirmin =
910 mnt->acdirmax = option;
911 break; 1122 break;
912 case Opt_namelen: 1123 case Opt_namelen:
913 if (match_int(args, &mnt->namlen)) 1124 if (match_int(args, &option) || option < 0) {
914 return 0; 1125 errors++;
1126 nfs_parse_invalid_value("namlen");
1127 } else
1128 mnt->namlen = option;
915 break; 1129 break;
916 case Opt_mountport: 1130 case Opt_mountport:
917 if (match_int(args, &option)) 1131 if (match_int(args, &option) ||
918 return 0; 1132 option < 0 || option > USHORT_MAX) {
919 if (option < 0 || option > 65535) 1133 errors++;
920 return 0; 1134 nfs_parse_invalid_value("mountport");
921 mnt->mount_server.port = option; 1135 } else
1136 mnt->mount_server.port = option;
922 break; 1137 break;
923 case Opt_mountvers: 1138 case Opt_mountvers:
924 if (match_int(args, &option)) 1139 if (match_int(args, &option) ||
925 return 0; 1140 option < NFS_MNT_VERSION ||
926 if (option < 0) 1141 option > NFS_MNT3_VERSION) {
927 return 0; 1142 errors++;
928 mnt->mount_server.version = option; 1143 nfs_parse_invalid_value("mountvers");
1144 } else
1145 mnt->mount_server.version = option;
929 break; 1146 break;
930 case Opt_nfsvers: 1147 case Opt_nfsvers:
931 if (match_int(args, &option)) 1148 if (match_int(args, &option)) {
932 return 0; 1149 errors++;
1150 nfs_parse_invalid_value("nfsvers");
1151 break;
1152 }
933 switch (option) { 1153 switch (option) {
934 case 2: 1154 case NFS2_VERSION:
935 mnt->flags &= ~NFS_MOUNT_VER3; 1155 mnt->flags &= ~NFS_MOUNT_VER3;
936 break; 1156 break;
937 case 3: 1157 case NFS3_VERSION:
938 mnt->flags |= NFS_MOUNT_VER3; 1158 mnt->flags |= NFS_MOUNT_VER3;
939 break; 1159 break;
940 default: 1160 default:
941 goto out_unrec_vers; 1161 errors++;
1162 nfs_parse_invalid_value("nfsvers");
942 } 1163 }
943 break; 1164 break;
944 1165
1166 /*
1167 * options that take text values
1168 */
945 case Opt_sec: 1169 case Opt_sec:
946 string = match_strdup(args); 1170 string = match_strdup(args);
947 if (string == NULL) 1171 if (string == NULL)
948 goto out_nomem; 1172 goto out_nomem;
949 token = match_token(string, nfs_secflavor_tokens, args); 1173 rc = nfs_parse_security_flavors(string, mnt);
950 kfree(string); 1174 kfree(string);
951 1175 if (!rc) {
952 /* 1176 errors++;
953 * The flags setting is for v2/v3. The flavor_len 1177 dfprintk(MOUNT, "NFS: unrecognized "
954 * setting is for v4. v2/v3 also need to know the 1178 "security flavor\n");
955 * difference between NULL and UNIX.
956 */
957 switch (token) {
958 case Opt_sec_none:
959 mnt->flags &= ~NFS_MOUNT_SECFLAVOUR;
960 mnt->auth_flavor_len = 0;
961 mnt->auth_flavors[0] = RPC_AUTH_NULL;
962 break;
963 case Opt_sec_sys:
964 mnt->flags &= ~NFS_MOUNT_SECFLAVOUR;
965 mnt->auth_flavor_len = 0;
966 mnt->auth_flavors[0] = RPC_AUTH_UNIX;
967 break;
968 case Opt_sec_krb5:
969 mnt->flags |= NFS_MOUNT_SECFLAVOUR;
970 mnt->auth_flavor_len = 1;
971 mnt->auth_flavors[0] = RPC_AUTH_GSS_KRB5;
972 break;
973 case Opt_sec_krb5i:
974 mnt->flags |= NFS_MOUNT_SECFLAVOUR;
975 mnt->auth_flavor_len = 1;
976 mnt->auth_flavors[0] = RPC_AUTH_GSS_KRB5I;
977 break;
978 case Opt_sec_krb5p:
979 mnt->flags |= NFS_MOUNT_SECFLAVOUR;
980 mnt->auth_flavor_len = 1;
981 mnt->auth_flavors[0] = RPC_AUTH_GSS_KRB5P;
982 break;
983 case Opt_sec_lkey:
984 mnt->flags |= NFS_MOUNT_SECFLAVOUR;
985 mnt->auth_flavor_len = 1;
986 mnt->auth_flavors[0] = RPC_AUTH_GSS_LKEY;
987 break;
988 case Opt_sec_lkeyi:
989 mnt->flags |= NFS_MOUNT_SECFLAVOUR;
990 mnt->auth_flavor_len = 1;
991 mnt->auth_flavors[0] = RPC_AUTH_GSS_LKEYI;
992 break;
993 case Opt_sec_lkeyp:
994 mnt->flags |= NFS_MOUNT_SECFLAVOUR;
995 mnt->auth_flavor_len = 1;
996 mnt->auth_flavors[0] = RPC_AUTH_GSS_LKEYP;
997 break;
998 case Opt_sec_spkm:
999 mnt->flags |= NFS_MOUNT_SECFLAVOUR;
1000 mnt->auth_flavor_len = 1;
1001 mnt->auth_flavors[0] = RPC_AUTH_GSS_SPKM;
1002 break;
1003 case Opt_sec_spkmi:
1004 mnt->flags |= NFS_MOUNT_SECFLAVOUR;
1005 mnt->auth_flavor_len = 1;
1006 mnt->auth_flavors[0] = RPC_AUTH_GSS_SPKMI;
1007 break;
1008 case Opt_sec_spkmp:
1009 mnt->flags |= NFS_MOUNT_SECFLAVOUR;
1010 mnt->auth_flavor_len = 1;
1011 mnt->auth_flavors[0] = RPC_AUTH_GSS_SPKMP;
1012 break;
1013 default:
1014 goto out_unrec_sec;
1015 } 1179 }
1016 break; 1180 break;
1017 case Opt_proto: 1181 case Opt_proto:
@@ -1026,24 +1190,20 @@ static int nfs_parse_mount_options(char *raw,
1026 case Opt_xprt_udp: 1190 case Opt_xprt_udp:
1027 mnt->flags &= ~NFS_MOUNT_TCP; 1191 mnt->flags &= ~NFS_MOUNT_TCP;
1028 mnt->nfs_server.protocol = XPRT_TRANSPORT_UDP; 1192 mnt->nfs_server.protocol = XPRT_TRANSPORT_UDP;
1029 mnt->timeo = 7;
1030 mnt->retrans = 5;
1031 break; 1193 break;
1032 case Opt_xprt_tcp: 1194 case Opt_xprt_tcp:
1033 mnt->flags |= NFS_MOUNT_TCP; 1195 mnt->flags |= NFS_MOUNT_TCP;
1034 mnt->nfs_server.protocol = XPRT_TRANSPORT_TCP; 1196 mnt->nfs_server.protocol = XPRT_TRANSPORT_TCP;
1035 mnt->timeo = 600;
1036 mnt->retrans = 2;
1037 break; 1197 break;
1038 case Opt_xprt_rdma: 1198 case Opt_xprt_rdma:
1039 /* vector side protocols to TCP */ 1199 /* vector side protocols to TCP */
1040 mnt->flags |= NFS_MOUNT_TCP; 1200 mnt->flags |= NFS_MOUNT_TCP;
1041 mnt->nfs_server.protocol = XPRT_TRANSPORT_RDMA; 1201 mnt->nfs_server.protocol = XPRT_TRANSPORT_RDMA;
1042 mnt->timeo = 600;
1043 mnt->retrans = 2;
1044 break; 1202 break;
1045 default: 1203 default:
1046 goto out_unrec_xprt; 1204 errors++;
1205 dfprintk(MOUNT, "NFS: unrecognized "
1206 "transport protocol\n");
1047 } 1207 }
1048 break; 1208 break;
1049 case Opt_mountproto: 1209 case Opt_mountproto:
@@ -1063,16 +1223,19 @@ static int nfs_parse_mount_options(char *raw,
1063 break; 1223 break;
1064 case Opt_xprt_rdma: /* not used for side protocols */ 1224 case Opt_xprt_rdma: /* not used for side protocols */
1065 default: 1225 default:
1066 goto out_unrec_xprt; 1226 errors++;
1227 dfprintk(MOUNT, "NFS: unrecognized "
1228 "transport protocol\n");
1067 } 1229 }
1068 break; 1230 break;
1069 case Opt_addr: 1231 case Opt_addr:
1070 string = match_strdup(args); 1232 string = match_strdup(args);
1071 if (string == NULL) 1233 if (string == NULL)
1072 goto out_nomem; 1234 goto out_nomem;
1073 nfs_parse_server_address(string, (struct sockaddr *) 1235 nfs_parse_ip_address(string, strlen(string),
1074 &mnt->nfs_server.address, 1236 (struct sockaddr *)
1075 &mnt->nfs_server.addrlen); 1237 &mnt->nfs_server.address,
1238 &mnt->nfs_server.addrlen);
1076 kfree(string); 1239 kfree(string);
1077 break; 1240 break;
1078 case Opt_clientaddr: 1241 case Opt_clientaddr:
@@ -1093,24 +1256,33 @@ static int nfs_parse_mount_options(char *raw,
1093 string = match_strdup(args); 1256 string = match_strdup(args);
1094 if (string == NULL) 1257 if (string == NULL)
1095 goto out_nomem; 1258 goto out_nomem;
1096 nfs_parse_server_address(string, (struct sockaddr *) 1259 nfs_parse_ip_address(string, strlen(string),
1097 &mnt->mount_server.address, 1260 (struct sockaddr *)
1098 &mnt->mount_server.addrlen); 1261 &mnt->mount_server.address,
1262 &mnt->mount_server.addrlen);
1099 kfree(string); 1263 kfree(string);
1100 break; 1264 break;
1101 1265
1266 /*
1267 * Special options
1268 */
1269 case Opt_sloppy:
1270 sloppy = 1;
1271 dfprintk(MOUNT, "NFS: relaxing parsing rules\n");
1272 break;
1102 case Opt_userspace: 1273 case Opt_userspace:
1103 case Opt_deprecated: 1274 case Opt_deprecated:
1275 dfprintk(MOUNT, "NFS: ignoring mount option "
1276 "'%s'\n", p);
1104 break; 1277 break;
1105 1278
1106 default: 1279 default:
1107 goto out_unknown; 1280 errors++;
1281 dfprintk(MOUNT, "NFS: unrecognized mount option "
1282 "'%s'\n", p);
1108 } 1283 }
1109 } 1284 }
1110 1285
1111 nfs_set_port((struct sockaddr *)&mnt->nfs_server.address,
1112 mnt->nfs_server.port);
1113
1114 return 1; 1286 return 1;
1115 1287
1116out_nomem: 1288out_nomem:
@@ -1120,21 +1292,6 @@ out_security_failure:
1120 free_secdata(secdata); 1292 free_secdata(secdata);
1121 printk(KERN_INFO "NFS: security options invalid: %d\n", rc); 1293 printk(KERN_INFO "NFS: security options invalid: %d\n", rc);
1122 return 0; 1294 return 0;
1123out_unrec_vers:
1124 printk(KERN_INFO "NFS: unrecognized NFS version number\n");
1125 return 0;
1126
1127out_unrec_xprt:
1128 printk(KERN_INFO "NFS: unrecognized transport protocol\n");
1129 return 0;
1130
1131out_unrec_sec:
1132 printk(KERN_INFO "NFS: unrecognized security flavor\n");
1133 return 0;
1134
1135out_unknown:
1136 printk(KERN_INFO "NFS: unknown mount option: %s\n", p);
1137 return 0;
1138} 1295}
1139 1296
1140/* 1297/*
@@ -1188,11 +1345,146 @@ static int nfs_try_mount(struct nfs_parsed_mount_data *args,
1188 if (status == 0) 1345 if (status == 0)
1189 return 0; 1346 return 0;
1190 1347
1191 dfprintk(MOUNT, "NFS: unable to mount server %s, error %d", 1348 dfprintk(MOUNT, "NFS: unable to mount server %s, error %d\n",
1192 hostname, status); 1349 hostname, status);
1193 return status; 1350 return status;
1194} 1351}
1195 1352
1353static int nfs_parse_simple_hostname(const char *dev_name,
1354 char **hostname, size_t maxnamlen,
1355 char **export_path, size_t maxpathlen)
1356{
1357 size_t len;
1358 char *colon, *comma;
1359
1360 colon = strchr(dev_name, ':');
1361 if (colon == NULL)
1362 goto out_bad_devname;
1363
1364 len = colon - dev_name;
1365 if (len > maxnamlen)
1366 goto out_hostname;
1367
1368 /* N.B. caller will free nfs_server.hostname in all cases */
1369 *hostname = kstrndup(dev_name, len, GFP_KERNEL);
1370 if (!*hostname)
1371 goto out_nomem;
1372
1373 /* kill possible hostname list: not supported */
1374 comma = strchr(*hostname, ',');
1375 if (comma != NULL) {
1376 if (comma == *hostname)
1377 goto out_bad_devname;
1378 *comma = '\0';
1379 }
1380
1381 colon++;
1382 len = strlen(colon);
1383 if (len > maxpathlen)
1384 goto out_path;
1385 *export_path = kstrndup(colon, len, GFP_KERNEL);
1386 if (!*export_path)
1387 goto out_nomem;
1388
1389 dfprintk(MOUNT, "NFS: MNTPATH: '%s'\n", *export_path);
1390 return 0;
1391
1392out_bad_devname:
1393 dfprintk(MOUNT, "NFS: device name not in host:path format\n");
1394 return -EINVAL;
1395
1396out_nomem:
1397 dfprintk(MOUNT, "NFS: not enough memory to parse device name\n");
1398 return -ENOMEM;
1399
1400out_hostname:
1401 dfprintk(MOUNT, "NFS: server hostname too long\n");
1402 return -ENAMETOOLONG;
1403
1404out_path:
1405 dfprintk(MOUNT, "NFS: export pathname too long\n");
1406 return -ENAMETOOLONG;
1407}
1408
1409/*
1410 * Hostname has square brackets around it because it contains one or
1411 * more colons. We look for the first closing square bracket, and a
1412 * colon must follow it.
1413 */
1414static int nfs_parse_protected_hostname(const char *dev_name,
1415 char **hostname, size_t maxnamlen,
1416 char **export_path, size_t maxpathlen)
1417{
1418 size_t len;
1419 char *start, *end;
1420
1421 start = (char *)(dev_name + 1);
1422
1423 end = strchr(start, ']');
1424 if (end == NULL)
1425 goto out_bad_devname;
1426 if (*(end + 1) != ':')
1427 goto out_bad_devname;
1428
1429 len = end - start;
1430 if (len > maxnamlen)
1431 goto out_hostname;
1432
1433 /* N.B. caller will free nfs_server.hostname in all cases */
1434 *hostname = kstrndup(start, len, GFP_KERNEL);
1435 if (*hostname == NULL)
1436 goto out_nomem;
1437
1438 end += 2;
1439 len = strlen(end);
1440 if (len > maxpathlen)
1441 goto out_path;
1442 *export_path = kstrndup(end, len, GFP_KERNEL);
1443 if (!*export_path)
1444 goto out_nomem;
1445
1446 return 0;
1447
1448out_bad_devname:
1449 dfprintk(MOUNT, "NFS: device name not in host:path format\n");
1450 return -EINVAL;
1451
1452out_nomem:
1453 dfprintk(MOUNT, "NFS: not enough memory to parse device name\n");
1454 return -ENOMEM;
1455
1456out_hostname:
1457 dfprintk(MOUNT, "NFS: server hostname too long\n");
1458 return -ENAMETOOLONG;
1459
1460out_path:
1461 dfprintk(MOUNT, "NFS: export pathname too long\n");
1462 return -ENAMETOOLONG;
1463}
1464
1465/*
1466 * Split "dev_name" into "hostname:export_path".
1467 *
1468 * The leftmost colon demarks the split between the server's hostname
1469 * and the export path. If the hostname starts with a left square
1470 * bracket, then it may contain colons.
1471 *
1472 * Note: caller frees hostname and export path, even on error.
1473 */
1474static int nfs_parse_devname(const char *dev_name,
1475 char **hostname, size_t maxnamlen,
1476 char **export_path, size_t maxpathlen)
1477{
1478 if (*dev_name == '[')
1479 return nfs_parse_protected_hostname(dev_name,
1480 hostname, maxnamlen,
1481 export_path, maxpathlen);
1482
1483 return nfs_parse_simple_hostname(dev_name,
1484 hostname, maxnamlen,
1485 export_path, maxpathlen);
1486}
1487
1196/* 1488/*
1197 * Validate the NFS2/NFS3 mount data 1489 * Validate the NFS2/NFS3 mount data
1198 * - fills in the mount root filehandle 1490 * - fills in the mount root filehandle
@@ -1222,16 +1514,14 @@ static int nfs_validate_mount_data(void *options,
1222 args->flags = (NFS_MOUNT_VER3 | NFS_MOUNT_TCP); 1514 args->flags = (NFS_MOUNT_VER3 | NFS_MOUNT_TCP);
1223 args->rsize = NFS_MAX_FILE_IO_SIZE; 1515 args->rsize = NFS_MAX_FILE_IO_SIZE;
1224 args->wsize = NFS_MAX_FILE_IO_SIZE; 1516 args->wsize = NFS_MAX_FILE_IO_SIZE;
1225 args->timeo = 600; 1517 args->acregmin = NFS_DEF_ACREGMIN;
1226 args->retrans = 2; 1518 args->acregmax = NFS_DEF_ACREGMAX;
1227 args->acregmin = 3; 1519 args->acdirmin = NFS_DEF_ACDIRMIN;
1228 args->acregmax = 60; 1520 args->acdirmax = NFS_DEF_ACDIRMAX;
1229 args->acdirmin = 30;
1230 args->acdirmax = 60;
1231 args->mount_server.port = 0; /* autobind unless user sets port */ 1521 args->mount_server.port = 0; /* autobind unless user sets port */
1232 args->mount_server.protocol = XPRT_TRANSPORT_UDP;
1233 args->nfs_server.port = 0; /* autobind unless user sets port */ 1522 args->nfs_server.port = 0; /* autobind unless user sets port */
1234 args->nfs_server.protocol = XPRT_TRANSPORT_TCP; 1523 args->nfs_server.protocol = XPRT_TRANSPORT_TCP;
1524 args->auth_flavors[0] = RPC_AUTH_UNIX;
1235 1525
1236 switch (data->version) { 1526 switch (data->version) {
1237 case 1: 1527 case 1:
@@ -1289,7 +1579,9 @@ static int nfs_validate_mount_data(void *options,
1289 args->nfs_server.hostname = kstrdup(data->hostname, GFP_KERNEL); 1579 args->nfs_server.hostname = kstrdup(data->hostname, GFP_KERNEL);
1290 args->namlen = data->namlen; 1580 args->namlen = data->namlen;
1291 args->bsize = data->bsize; 1581 args->bsize = data->bsize;
1292 args->auth_flavors[0] = data->pseudoflavor; 1582
1583 if (data->flags & NFS_MOUNT_SECFLAVOUR)
1584 args->auth_flavors[0] = data->pseudoflavor;
1293 if (!args->nfs_server.hostname) 1585 if (!args->nfs_server.hostname)
1294 goto out_nomem; 1586 goto out_nomem;
1295 1587
@@ -1321,8 +1613,6 @@ static int nfs_validate_mount_data(void *options,
1321 1613
1322 break; 1614 break;
1323 default: { 1615 default: {
1324 unsigned int len;
1325 char *c;
1326 int status; 1616 int status;
1327 1617
1328 if (nfs_parse_mount_options((char *)options, args) == 0) 1618 if (nfs_parse_mount_options((char *)options, args) == 0)
@@ -1332,21 +1622,22 @@ static int nfs_validate_mount_data(void *options,
1332 &args->nfs_server.address)) 1622 &args->nfs_server.address))
1333 goto out_no_address; 1623 goto out_no_address;
1334 1624
1335 c = strchr(dev_name, ':'); 1625 nfs_set_port((struct sockaddr *)&args->nfs_server.address,
1336 if (c == NULL) 1626 args->nfs_server.port);
1337 return -EINVAL;
1338 len = c - dev_name;
1339 /* N.B. caller will free nfs_server.hostname in all cases */
1340 args->nfs_server.hostname = kstrndup(dev_name, len, GFP_KERNEL);
1341 if (!args->nfs_server.hostname)
1342 goto out_nomem;
1343 1627
1344 c++; 1628 nfs_set_mount_transport_protocol(args);
1345 if (strlen(c) > NFS_MAXPATHLEN) 1629
1346 return -ENAMETOOLONG; 1630 status = nfs_parse_devname(dev_name,
1347 args->nfs_server.export_path = c; 1631 &args->nfs_server.hostname,
1632 PAGE_SIZE,
1633 &args->nfs_server.export_path,
1634 NFS_MAXPATHLEN);
1635 if (!status)
1636 status = nfs_try_mount(args, mntfh);
1637
1638 kfree(args->nfs_server.export_path);
1639 args->nfs_server.export_path = NULL;
1348 1640
1349 status = nfs_try_mount(args, mntfh);
1350 if (status) 1641 if (status)
1351 return status; 1642 return status;
1352 1643
@@ -1354,9 +1645,6 @@ static int nfs_validate_mount_data(void *options,
1354 } 1645 }
1355 } 1646 }
1356 1647
1357 if (!(args->flags & NFS_MOUNT_SECFLAVOUR))
1358 args->auth_flavors[0] = RPC_AUTH_UNIX;
1359
1360#ifndef CONFIG_NFS_V3 1648#ifndef CONFIG_NFS_V3
1361 if (args->flags & NFS_MOUNT_VER3) 1649 if (args->flags & NFS_MOUNT_VER3)
1362 goto out_v3_not_compiled; 1650 goto out_v3_not_compiled;
@@ -1396,6 +1684,80 @@ out_invalid_fh:
1396 return -EINVAL; 1684 return -EINVAL;
1397} 1685}
1398 1686
1687static int
1688nfs_compare_remount_data(struct nfs_server *nfss,
1689 struct nfs_parsed_mount_data *data)
1690{
1691 if (data->flags != nfss->flags ||
1692 data->rsize != nfss->rsize ||
1693 data->wsize != nfss->wsize ||
1694 data->retrans != nfss->client->cl_timeout->to_retries ||
1695 data->auth_flavors[0] != nfss->client->cl_auth->au_flavor ||
1696 data->acregmin != nfss->acregmin / HZ ||
1697 data->acregmax != nfss->acregmax / HZ ||
1698 data->acdirmin != nfss->acdirmin / HZ ||
1699 data->acdirmax != nfss->acdirmax / HZ ||
1700 data->timeo != (10U * nfss->client->cl_timeout->to_initval / HZ) ||
1701 data->nfs_server.addrlen != nfss->nfs_client->cl_addrlen ||
1702 memcmp(&data->nfs_server.address, &nfss->nfs_client->cl_addr,
1703 data->nfs_server.addrlen) != 0)
1704 return -EINVAL;
1705
1706 return 0;
1707}
1708
1709static int
1710nfs_remount(struct super_block *sb, int *flags, char *raw_data)
1711{
1712 int error;
1713 struct nfs_server *nfss = sb->s_fs_info;
1714 struct nfs_parsed_mount_data *data;
1715 struct nfs_mount_data *options = (struct nfs_mount_data *)raw_data;
1716 struct nfs4_mount_data *options4 = (struct nfs4_mount_data *)raw_data;
1717 u32 nfsvers = nfss->nfs_client->rpc_ops->version;
1718
1719 /*
1720 * Userspace mount programs that send binary options generally send
1721 * them populated with default values. We have no way to know which
1722 * ones were explicitly specified. Fall back to legacy behavior and
1723 * just return success.
1724 */
1725 if ((nfsvers == 4 && options4->version == 1) ||
1726 (nfsvers <= 3 && options->version >= 1 &&
1727 options->version <= 6))
1728 return 0;
1729
1730 data = kzalloc(sizeof(*data), GFP_KERNEL);
1731 if (data == NULL)
1732 return -ENOMEM;
1733
1734 /* fill out struct with values from existing mount */
1735 data->flags = nfss->flags;
1736 data->rsize = nfss->rsize;
1737 data->wsize = nfss->wsize;
1738 data->retrans = nfss->client->cl_timeout->to_retries;
1739 data->auth_flavors[0] = nfss->client->cl_auth->au_flavor;
1740 data->acregmin = nfss->acregmin / HZ;
1741 data->acregmax = nfss->acregmax / HZ;
1742 data->acdirmin = nfss->acdirmin / HZ;
1743 data->acdirmax = nfss->acdirmax / HZ;
1744 data->timeo = 10U * nfss->client->cl_timeout->to_initval / HZ;
1745 data->nfs_server.addrlen = nfss->nfs_client->cl_addrlen;
1746 memcpy(&data->nfs_server.address, &nfss->nfs_client->cl_addr,
1747 data->nfs_server.addrlen);
1748
1749 /* overwrite those values with any that were specified */
1750 error = nfs_parse_mount_options((char *)options, data);
1751 if (error < 0)
1752 goto out;
1753
1754 /* compare new mount options with old ones */
1755 error = nfs_compare_remount_data(nfss, data);
1756out:
1757 kfree(data);
1758 return error;
1759}
1760
1399/* 1761/*
1400 * Initialise the common bits of the superblock 1762 * Initialise the common bits of the superblock
1401 */ 1763 */
@@ -1811,14 +2173,13 @@ static int nfs4_validate_mount_data(void *options,
1811 2173
1812 args->rsize = NFS_MAX_FILE_IO_SIZE; 2174 args->rsize = NFS_MAX_FILE_IO_SIZE;
1813 args->wsize = NFS_MAX_FILE_IO_SIZE; 2175 args->wsize = NFS_MAX_FILE_IO_SIZE;
1814 args->timeo = 600; 2176 args->acregmin = NFS_DEF_ACREGMIN;
1815 args->retrans = 2; 2177 args->acregmax = NFS_DEF_ACREGMAX;
1816 args->acregmin = 3; 2178 args->acdirmin = NFS_DEF_ACDIRMIN;
1817 args->acregmax = 60; 2179 args->acdirmax = NFS_DEF_ACDIRMAX;
1818 args->acdirmin = 30;
1819 args->acdirmax = 60;
1820 args->nfs_server.port = NFS_PORT; /* 2049 unless user set port= */ 2180 args->nfs_server.port = NFS_PORT; /* 2049 unless user set port= */
1821 args->nfs_server.protocol = XPRT_TRANSPORT_TCP; 2181 args->auth_flavors[0] = RPC_AUTH_UNIX;
2182 args->auth_flavor_len = 0;
1822 2183
1823 switch (data->version) { 2184 switch (data->version) {
1824 case 1: 2185 case 1:
@@ -1834,18 +2195,13 @@ static int nfs4_validate_mount_data(void *options,
1834 &args->nfs_server.address)) 2195 &args->nfs_server.address))
1835 goto out_no_address; 2196 goto out_no_address;
1836 2197
1837 switch (data->auth_flavourlen) { 2198 if (data->auth_flavourlen) {
1838 case 0: 2199 if (data->auth_flavourlen > 1)
1839 args->auth_flavors[0] = RPC_AUTH_UNIX; 2200 goto out_inval_auth;
1840 break;
1841 case 1:
1842 if (copy_from_user(&args->auth_flavors[0], 2201 if (copy_from_user(&args->auth_flavors[0],
1843 data->auth_flavours, 2202 data->auth_flavours,
1844 sizeof(args->auth_flavors[0]))) 2203 sizeof(args->auth_flavors[0])))
1845 return -EFAULT; 2204 return -EFAULT;
1846 break;
1847 default:
1848 goto out_inval_auth;
1849 } 2205 }
1850 2206
1851 c = strndup_user(data->hostname.data, NFS4_MAXNAMLEN); 2207 c = strndup_user(data->hostname.data, NFS4_MAXNAMLEN);
@@ -1879,10 +2235,11 @@ static int nfs4_validate_mount_data(void *options,
1879 args->acdirmin = data->acdirmin; 2235 args->acdirmin = data->acdirmin;
1880 args->acdirmax = data->acdirmax; 2236 args->acdirmax = data->acdirmax;
1881 args->nfs_server.protocol = data->proto; 2237 args->nfs_server.protocol = data->proto;
2238 nfs_validate_transport_protocol(args);
1882 2239
1883 break; 2240 break;
1884 default: { 2241 default: {
1885 unsigned int len; 2242 int status;
1886 2243
1887 if (nfs_parse_mount_options((char *)options, args) == 0) 2244 if (nfs_parse_mount_options((char *)options, args) == 0)
1888 return -EINVAL; 2245 return -EINVAL;
@@ -1891,44 +2248,25 @@ static int nfs4_validate_mount_data(void *options,
1891 &args->nfs_server.address)) 2248 &args->nfs_server.address))
1892 return -EINVAL; 2249 return -EINVAL;
1893 2250
1894 switch (args->auth_flavor_len) { 2251 nfs_set_port((struct sockaddr *)&args->nfs_server.address,
1895 case 0: 2252 args->nfs_server.port);
1896 args->auth_flavors[0] = RPC_AUTH_UNIX;
1897 break;
1898 case 1:
1899 break;
1900 default:
1901 goto out_inval_auth;
1902 }
1903 2253
1904 /* 2254 nfs_validate_transport_protocol(args);
1905 * Split "dev_name" into "hostname:mntpath".
1906 */
1907 c = strchr(dev_name, ':');
1908 if (c == NULL)
1909 return -EINVAL;
1910 /* while calculating len, pretend ':' is '\0' */
1911 len = c - dev_name;
1912 if (len > NFS4_MAXNAMLEN)
1913 return -ENAMETOOLONG;
1914 /* N.B. caller will free nfs_server.hostname in all cases */
1915 args->nfs_server.hostname = kstrndup(dev_name, len, GFP_KERNEL);
1916 if (!args->nfs_server.hostname)
1917 goto out_nomem;
1918
1919 c++; /* step over the ':' */
1920 len = strlen(c);
1921 if (len > NFS4_MAXPATHLEN)
1922 return -ENAMETOOLONG;
1923 args->nfs_server.export_path = kstrndup(c, len, GFP_KERNEL);
1924 if (!args->nfs_server.export_path)
1925 goto out_nomem;
1926 2255
1927 dprintk("NFS: MNTPATH: '%s'\n", args->nfs_server.export_path); 2256 if (args->auth_flavor_len > 1)
2257 goto out_inval_auth;
1928 2258
1929 if (args->client_address == NULL) 2259 if (args->client_address == NULL)
1930 goto out_no_client_address; 2260 goto out_no_client_address;
1931 2261
2262 status = nfs_parse_devname(dev_name,
2263 &args->nfs_server.hostname,
2264 NFS4_MAXNAMLEN,
2265 &args->nfs_server.export_path,
2266 NFS4_MAXPATHLEN);
2267 if (status < 0)
2268 return status;
2269
1932 break; 2270 break;
1933 } 2271 }
1934 } 2272 }
@@ -1944,10 +2282,6 @@ out_inval_auth:
1944 data->auth_flavourlen); 2282 data->auth_flavourlen);
1945 return -EINVAL; 2283 return -EINVAL;
1946 2284
1947out_nomem:
1948 dfprintk(MOUNT, "NFS4: not enough memory to handle mount options\n");
1949 return -ENOMEM;
1950
1951out_no_address: 2285out_no_address:
1952 dfprintk(MOUNT, "NFS4: mount program didn't pass remote address\n"); 2286 dfprintk(MOUNT, "NFS4: mount program didn't pass remote address\n");
1953 return -EINVAL; 2287 return -EINVAL;
diff --git a/fs/nfs/write.c b/fs/nfs/write.c
index f333848fd3be..feca8c648766 100644
--- a/fs/nfs/write.c
+++ b/fs/nfs/write.c
@@ -34,9 +34,6 @@
34/* 34/*
35 * Local function declarations 35 * Local function declarations
36 */ 36 */
37static struct nfs_page * nfs_update_request(struct nfs_open_context*,
38 struct page *,
39 unsigned int, unsigned int);
40static void nfs_pageio_init_write(struct nfs_pageio_descriptor *desc, 37static void nfs_pageio_init_write(struct nfs_pageio_descriptor *desc,
41 struct inode *inode, int ioflags); 38 struct inode *inode, int ioflags);
42static void nfs_redirty_request(struct nfs_page *req); 39static void nfs_redirty_request(struct nfs_page *req);
@@ -169,29 +166,6 @@ static void nfs_mark_uptodate(struct page *page, unsigned int base, unsigned int
169 SetPageUptodate(page); 166 SetPageUptodate(page);
170} 167}
171 168
172static int nfs_writepage_setup(struct nfs_open_context *ctx, struct page *page,
173 unsigned int offset, unsigned int count)
174{
175 struct nfs_page *req;
176 int ret;
177
178 for (;;) {
179 req = nfs_update_request(ctx, page, offset, count);
180 if (!IS_ERR(req))
181 break;
182 ret = PTR_ERR(req);
183 if (ret != -EBUSY)
184 return ret;
185 ret = nfs_wb_page(page->mapping->host, page);
186 if (ret != 0)
187 return ret;
188 }
189 /* Update file length */
190 nfs_grow_file(page, offset, count);
191 nfs_clear_page_tag_locked(req);
192 return 0;
193}
194
195static int wb_priority(struct writeback_control *wbc) 169static int wb_priority(struct writeback_control *wbc)
196{ 170{
197 if (wbc->for_reclaim) 171 if (wbc->for_reclaim)
@@ -268,12 +242,9 @@ static int nfs_page_async_flush(struct nfs_pageio_descriptor *pgio,
268 return ret; 242 return ret;
269 spin_lock(&inode->i_lock); 243 spin_lock(&inode->i_lock);
270 } 244 }
271 if (test_bit(PG_NEED_COMMIT, &req->wb_flags)) { 245 if (test_bit(PG_CLEAN, &req->wb_flags)) {
272 /* This request is marked for commit */
273 spin_unlock(&inode->i_lock); 246 spin_unlock(&inode->i_lock);
274 nfs_clear_page_tag_locked(req); 247 BUG();
275 nfs_pageio_complete(pgio);
276 return 0;
277 } 248 }
278 if (nfs_set_page_writeback(page) != 0) { 249 if (nfs_set_page_writeback(page) != 0) {
279 spin_unlock(&inode->i_lock); 250 spin_unlock(&inode->i_lock);
@@ -355,11 +326,19 @@ int nfs_writepages(struct address_space *mapping, struct writeback_control *wbc)
355/* 326/*
356 * Insert a write request into an inode 327 * Insert a write request into an inode
357 */ 328 */
358static void nfs_inode_add_request(struct inode *inode, struct nfs_page *req) 329static int nfs_inode_add_request(struct inode *inode, struct nfs_page *req)
359{ 330{
360 struct nfs_inode *nfsi = NFS_I(inode); 331 struct nfs_inode *nfsi = NFS_I(inode);
361 int error; 332 int error;
362 333
334 error = radix_tree_preload(GFP_NOFS);
335 if (error != 0)
336 goto out;
337
338 /* Lock the request! */
339 nfs_lock_request_dontget(req);
340
341 spin_lock(&inode->i_lock);
363 error = radix_tree_insert(&nfsi->nfs_page_tree, req->wb_index, req); 342 error = radix_tree_insert(&nfsi->nfs_page_tree, req->wb_index, req);
364 BUG_ON(error); 343 BUG_ON(error);
365 if (!nfsi->npages) { 344 if (!nfsi->npages) {
@@ -373,6 +352,10 @@ static void nfs_inode_add_request(struct inode *inode, struct nfs_page *req)
373 kref_get(&req->wb_kref); 352 kref_get(&req->wb_kref);
374 radix_tree_tag_set(&nfsi->nfs_page_tree, req->wb_index, 353 radix_tree_tag_set(&nfsi->nfs_page_tree, req->wb_index,
375 NFS_PAGE_TAG_LOCKED); 354 NFS_PAGE_TAG_LOCKED);
355 spin_unlock(&inode->i_lock);
356 radix_tree_preload_end();
357out:
358 return error;
376} 359}
377 360
378/* 361/*
@@ -405,19 +388,6 @@ nfs_mark_request_dirty(struct nfs_page *req)
405 __set_page_dirty_nobuffers(req->wb_page); 388 __set_page_dirty_nobuffers(req->wb_page);
406} 389}
407 390
408/*
409 * Check if a request is dirty
410 */
411static inline int
412nfs_dirty_request(struct nfs_page *req)
413{
414 struct page *page = req->wb_page;
415
416 if (page == NULL || test_bit(PG_NEED_COMMIT, &req->wb_flags))
417 return 0;
418 return !PageWriteback(page);
419}
420
421#if defined(CONFIG_NFS_V3) || defined(CONFIG_NFS_V4) 391#if defined(CONFIG_NFS_V3) || defined(CONFIG_NFS_V4)
422/* 392/*
423 * Add a request to the inode's commit list. 393 * Add a request to the inode's commit list.
@@ -430,7 +400,7 @@ nfs_mark_request_commit(struct nfs_page *req)
430 400
431 spin_lock(&inode->i_lock); 401 spin_lock(&inode->i_lock);
432 nfsi->ncommit++; 402 nfsi->ncommit++;
433 set_bit(PG_NEED_COMMIT, &(req)->wb_flags); 403 set_bit(PG_CLEAN, &(req)->wb_flags);
434 radix_tree_tag_set(&nfsi->nfs_page_tree, 404 radix_tree_tag_set(&nfsi->nfs_page_tree,
435 req->wb_index, 405 req->wb_index,
436 NFS_PAGE_TAG_COMMIT); 406 NFS_PAGE_TAG_COMMIT);
@@ -440,6 +410,19 @@ nfs_mark_request_commit(struct nfs_page *req)
440 __mark_inode_dirty(inode, I_DIRTY_DATASYNC); 410 __mark_inode_dirty(inode, I_DIRTY_DATASYNC);
441} 411}
442 412
413static int
414nfs_clear_request_commit(struct nfs_page *req)
415{
416 struct page *page = req->wb_page;
417
418 if (test_and_clear_bit(PG_CLEAN, &(req)->wb_flags)) {
419 dec_zone_page_state(page, NR_UNSTABLE_NFS);
420 dec_bdi_stat(page->mapping->backing_dev_info, BDI_RECLAIMABLE);
421 return 1;
422 }
423 return 0;
424}
425
443static inline 426static inline
444int nfs_write_need_commit(struct nfs_write_data *data) 427int nfs_write_need_commit(struct nfs_write_data *data)
445{ 428{
@@ -449,7 +432,7 @@ int nfs_write_need_commit(struct nfs_write_data *data)
449static inline 432static inline
450int nfs_reschedule_unstable_write(struct nfs_page *req) 433int nfs_reschedule_unstable_write(struct nfs_page *req)
451{ 434{
452 if (test_bit(PG_NEED_COMMIT, &req->wb_flags)) { 435 if (test_and_clear_bit(PG_NEED_COMMIT, &req->wb_flags)) {
453 nfs_mark_request_commit(req); 436 nfs_mark_request_commit(req);
454 return 1; 437 return 1;
455 } 438 }
@@ -465,6 +448,12 @@ nfs_mark_request_commit(struct nfs_page *req)
465{ 448{
466} 449}
467 450
451static inline int
452nfs_clear_request_commit(struct nfs_page *req)
453{
454 return 0;
455}
456
468static inline 457static inline
469int nfs_write_need_commit(struct nfs_write_data *data) 458int nfs_write_need_commit(struct nfs_write_data *data)
470{ 459{
@@ -522,11 +511,8 @@ static void nfs_cancel_commit_list(struct list_head *head)
522 511
523 while(!list_empty(head)) { 512 while(!list_empty(head)) {
524 req = nfs_list_entry(head->next); 513 req = nfs_list_entry(head->next);
525 dec_zone_page_state(req->wb_page, NR_UNSTABLE_NFS);
526 dec_bdi_stat(req->wb_page->mapping->backing_dev_info,
527 BDI_RECLAIMABLE);
528 nfs_list_remove_request(req); 514 nfs_list_remove_request(req);
529 clear_bit(PG_NEED_COMMIT, &(req)->wb_flags); 515 nfs_clear_request_commit(req);
530 nfs_inode_remove_request(req); 516 nfs_inode_remove_request(req);
531 nfs_unlock_request(req); 517 nfs_unlock_request(req);
532 } 518 }
@@ -564,110 +550,124 @@ static inline int nfs_scan_commit(struct inode *inode, struct list_head *dst, pg
564#endif 550#endif
565 551
566/* 552/*
567 * Try to update any existing write request, or create one if there is none. 553 * Search for an existing write request, and attempt to update
568 * In order to match, the request's credentials must match those of 554 * it to reflect a new dirty region on a given page.
569 * the calling process.
570 * 555 *
571 * Note: Should always be called with the Page Lock held! 556 * If the attempt fails, then the existing request is flushed out
557 * to disk.
572 */ 558 */
573static struct nfs_page * nfs_update_request(struct nfs_open_context* ctx, 559static struct nfs_page *nfs_try_to_update_request(struct inode *inode,
574 struct page *page, unsigned int offset, unsigned int bytes) 560 struct page *page,
561 unsigned int offset,
562 unsigned int bytes)
575{ 563{
576 struct address_space *mapping = page->mapping; 564 struct nfs_page *req;
577 struct inode *inode = mapping->host; 565 unsigned int rqend;
578 struct nfs_page *req, *new = NULL; 566 unsigned int end;
579 pgoff_t rqend, end; 567 int error;
568
569 if (!PagePrivate(page))
570 return NULL;
580 571
581 end = offset + bytes; 572 end = offset + bytes;
573 spin_lock(&inode->i_lock);
582 574
583 for (;;) { 575 for (;;) {
584 /* Loop over all inode entries and see if we find 576 req = nfs_page_find_request_locked(page);
585 * A request for the page we wish to update 577 if (req == NULL)
578 goto out_unlock;
579
580 rqend = req->wb_offset + req->wb_bytes;
581 /*
582 * Tell the caller to flush out the request if
583 * the offsets are non-contiguous.
584 * Note: nfs_flush_incompatible() will already
585 * have flushed out requests having wrong owners.
586 */ 586 */
587 if (new) { 587 if (offset > rqend
588 if (radix_tree_preload(GFP_NOFS)) { 588 || end < req->wb_offset)
589 nfs_release_request(new); 589 goto out_flushme;
590 return ERR_PTR(-ENOMEM);
591 }
592 }
593 590
594 spin_lock(&inode->i_lock); 591 if (nfs_set_page_tag_locked(req))
595 req = nfs_page_find_request_locked(page);
596 if (req) {
597 if (!nfs_set_page_tag_locked(req)) {
598 int error;
599
600 spin_unlock(&inode->i_lock);
601 error = nfs_wait_on_request(req);
602 nfs_release_request(req);
603 if (error < 0) {
604 if (new) {
605 radix_tree_preload_end();
606 nfs_release_request(new);
607 }
608 return ERR_PTR(error);
609 }
610 continue;
611 }
612 spin_unlock(&inode->i_lock);
613 if (new) {
614 radix_tree_preload_end();
615 nfs_release_request(new);
616 }
617 break; 592 break;
618 }
619 593
620 if (new) { 594 /* The request is locked, so wait and then retry */
621 nfs_lock_request_dontget(new);
622 nfs_inode_add_request(inode, new);
623 spin_unlock(&inode->i_lock);
624 radix_tree_preload_end();
625 req = new;
626 goto zero_page;
627 }
628 spin_unlock(&inode->i_lock); 595 spin_unlock(&inode->i_lock);
629 596 error = nfs_wait_on_request(req);
630 new = nfs_create_request(ctx, inode, page, offset, bytes); 597 nfs_release_request(req);
631 if (IS_ERR(new)) 598 if (error != 0)
632 return new; 599 goto out_err;
600 spin_lock(&inode->i_lock);
633 } 601 }
634 602
635 /* We have a request for our page. 603 if (nfs_clear_request_commit(req))
636 * If the creds don't match, or the 604 radix_tree_tag_clear(&NFS_I(inode)->nfs_page_tree,
637 * page addresses don't match, 605 req->wb_index, NFS_PAGE_TAG_COMMIT);
638 * tell the caller to wait on the conflicting
639 * request.
640 */
641 rqend = req->wb_offset + req->wb_bytes;
642 if (req->wb_context != ctx
643 || req->wb_page != page
644 || !nfs_dirty_request(req)
645 || offset > rqend || end < req->wb_offset) {
646 nfs_clear_page_tag_locked(req);
647 return ERR_PTR(-EBUSY);
648 }
649 606
650 /* Okay, the request matches. Update the region */ 607 /* Okay, the request matches. Update the region */
651 if (offset < req->wb_offset) { 608 if (offset < req->wb_offset) {
652 req->wb_offset = offset; 609 req->wb_offset = offset;
653 req->wb_pgbase = offset; 610 req->wb_pgbase = offset;
654 req->wb_bytes = max(end, rqend) - req->wb_offset;
655 goto zero_page;
656 } 611 }
657
658 if (end > rqend) 612 if (end > rqend)
659 req->wb_bytes = end - req->wb_offset; 613 req->wb_bytes = end - req->wb_offset;
660 614 else
615 req->wb_bytes = rqend - req->wb_offset;
616out_unlock:
617 spin_unlock(&inode->i_lock);
661 return req; 618 return req;
662zero_page: 619out_flushme:
663 /* If this page might potentially be marked as up to date, 620 spin_unlock(&inode->i_lock);
664 * then we need to zero any uninitalised data. */ 621 nfs_release_request(req);
665 if (req->wb_pgbase == 0 && req->wb_bytes != PAGE_CACHE_SIZE 622 error = nfs_wb_page(inode, page);
666 && !PageUptodate(req->wb_page)) 623out_err:
667 zero_user_segment(req->wb_page, req->wb_bytes, PAGE_CACHE_SIZE); 624 return ERR_PTR(error);
625}
626
627/*
628 * Try to update an existing write request, or create one if there is none.
629 *
630 * Note: Should always be called with the Page Lock held to prevent races
631 * if we have to add a new request. Also assumes that the caller has
632 * already called nfs_flush_incompatible() if necessary.
633 */
634static struct nfs_page * nfs_setup_write_request(struct nfs_open_context* ctx,
635 struct page *page, unsigned int offset, unsigned int bytes)
636{
637 struct inode *inode = page->mapping->host;
638 struct nfs_page *req;
639 int error;
640
641 req = nfs_try_to_update_request(inode, page, offset, bytes);
642 if (req != NULL)
643 goto out;
644 req = nfs_create_request(ctx, inode, page, offset, bytes);
645 if (IS_ERR(req))
646 goto out;
647 error = nfs_inode_add_request(inode, req);
648 if (error != 0) {
649 nfs_release_request(req);
650 req = ERR_PTR(error);
651 }
652out:
668 return req; 653 return req;
669} 654}
670 655
656static int nfs_writepage_setup(struct nfs_open_context *ctx, struct page *page,
657 unsigned int offset, unsigned int count)
658{
659 struct nfs_page *req;
660
661 req = nfs_setup_write_request(ctx, page, offset, count);
662 if (IS_ERR(req))
663 return PTR_ERR(req);
664 /* Update file length */
665 nfs_grow_file(page, offset, count);
666 nfs_mark_uptodate(page, req->wb_pgbase, req->wb_bytes);
667 nfs_clear_page_tag_locked(req);
668 return 0;
669}
670
671int nfs_flush_incompatible(struct file *file, struct page *page) 671int nfs_flush_incompatible(struct file *file, struct page *page)
672{ 672{
673 struct nfs_open_context *ctx = nfs_file_open_context(file); 673 struct nfs_open_context *ctx = nfs_file_open_context(file);
@@ -685,8 +685,7 @@ int nfs_flush_incompatible(struct file *file, struct page *page)
685 req = nfs_page_find_request(page); 685 req = nfs_page_find_request(page);
686 if (req == NULL) 686 if (req == NULL)
687 return 0; 687 return 0;
688 do_flush = req->wb_page != page || req->wb_context != ctx 688 do_flush = req->wb_page != page || req->wb_context != ctx;
689 || !nfs_dirty_request(req);
690 nfs_release_request(req); 689 nfs_release_request(req);
691 if (!do_flush) 690 if (!do_flush)
692 return 0; 691 return 0;
@@ -721,10 +720,10 @@ int nfs_updatepage(struct file *file, struct page *page,
721 720
722 nfs_inc_stats(inode, NFSIOS_VFSUPDATEPAGE); 721 nfs_inc_stats(inode, NFSIOS_VFSUPDATEPAGE);
723 722
724 dprintk("NFS: nfs_updatepage(%s/%s %d@%Ld)\n", 723 dprintk("NFS: nfs_updatepage(%s/%s %d@%lld)\n",
725 file->f_path.dentry->d_parent->d_name.name, 724 file->f_path.dentry->d_parent->d_name.name,
726 file->f_path.dentry->d_name.name, count, 725 file->f_path.dentry->d_name.name, count,
727 (long long)(page_offset(page) +offset)); 726 (long long)(page_offset(page) + offset));
728 727
729 /* If we're not using byte range locks, and we know the page 728 /* If we're not using byte range locks, and we know the page
730 * is up to date, it may be more efficient to extend the write 729 * is up to date, it may be more efficient to extend the write
@@ -744,7 +743,7 @@ int nfs_updatepage(struct file *file, struct page *page,
744 else 743 else
745 __set_page_dirty_nobuffers(page); 744 __set_page_dirty_nobuffers(page);
746 745
747 dprintk("NFS: nfs_updatepage returns %d (isize %Ld)\n", 746 dprintk("NFS: nfs_updatepage returns %d (isize %lld)\n",
748 status, (long long)i_size_read(inode)); 747 status, (long long)i_size_read(inode));
749 return status; 748 return status;
750} 749}
@@ -752,12 +751,7 @@ int nfs_updatepage(struct file *file, struct page *page,
752static void nfs_writepage_release(struct nfs_page *req) 751static void nfs_writepage_release(struct nfs_page *req)
753{ 752{
754 753
755 if (PageError(req->wb_page)) { 754 if (PageError(req->wb_page) || !nfs_reschedule_unstable_write(req)) {
756 nfs_end_page_writeback(req->wb_page);
757 nfs_inode_remove_request(req);
758 } else if (!nfs_reschedule_unstable_write(req)) {
759 /* Set the PG_uptodate flag */
760 nfs_mark_uptodate(req->wb_page, req->wb_pgbase, req->wb_bytes);
761 nfs_end_page_writeback(req->wb_page); 755 nfs_end_page_writeback(req->wb_page);
762 nfs_inode_remove_request(req); 756 nfs_inode_remove_request(req);
763 } else 757 } else
@@ -834,7 +828,7 @@ static int nfs_write_rpcsetup(struct nfs_page *req,
834 NFS_PROTO(inode)->write_setup(data, &msg); 828 NFS_PROTO(inode)->write_setup(data, &msg);
835 829
836 dprintk("NFS: %5u initiated write call " 830 dprintk("NFS: %5u initiated write call "
837 "(req %s/%Ld, %u bytes @ offset %Lu)\n", 831 "(req %s/%lld, %u bytes @ offset %llu)\n",
838 data->task.tk_pid, 832 data->task.tk_pid,
839 inode->i_sb->s_id, 833 inode->i_sb->s_id,
840 (long long)NFS_FILEID(inode), 834 (long long)NFS_FILEID(inode),
@@ -978,13 +972,13 @@ static void nfs_pageio_init_write(struct nfs_pageio_descriptor *pgio,
978static void nfs_writeback_done_partial(struct rpc_task *task, void *calldata) 972static void nfs_writeback_done_partial(struct rpc_task *task, void *calldata)
979{ 973{
980 struct nfs_write_data *data = calldata; 974 struct nfs_write_data *data = calldata;
981 struct nfs_page *req = data->req;
982 975
983 dprintk("NFS: write (%s/%Ld %d@%Ld)", 976 dprintk("NFS: %5u write(%s/%lld %d@%lld)",
984 req->wb_context->path.dentry->d_inode->i_sb->s_id, 977 task->tk_pid,
985 (long long)NFS_FILEID(req->wb_context->path.dentry->d_inode), 978 data->req->wb_context->path.dentry->d_inode->i_sb->s_id,
986 req->wb_bytes, 979 (long long)
987 (long long)req_offset(req)); 980 NFS_FILEID(data->req->wb_context->path.dentry->d_inode),
981 data->req->wb_bytes, (long long)req_offset(data->req));
988 982
989 nfs_writeback_done(task, data); 983 nfs_writeback_done(task, data);
990} 984}
@@ -1058,7 +1052,8 @@ static void nfs_writeback_release_full(void *calldata)
1058 1052
1059 nfs_list_remove_request(req); 1053 nfs_list_remove_request(req);
1060 1054
1061 dprintk("NFS: write (%s/%Ld %d@%Ld)", 1055 dprintk("NFS: %5u write (%s/%lld %d@%lld)",
1056 data->task.tk_pid,
1062 req->wb_context->path.dentry->d_inode->i_sb->s_id, 1057 req->wb_context->path.dentry->d_inode->i_sb->s_id,
1063 (long long)NFS_FILEID(req->wb_context->path.dentry->d_inode), 1058 (long long)NFS_FILEID(req->wb_context->path.dentry->d_inode),
1064 req->wb_bytes, 1059 req->wb_bytes,
@@ -1078,8 +1073,6 @@ static void nfs_writeback_release_full(void *calldata)
1078 dprintk(" marked for commit\n"); 1073 dprintk(" marked for commit\n");
1079 goto next; 1074 goto next;
1080 } 1075 }
1081 /* Set the PG_uptodate flag? */
1082 nfs_mark_uptodate(page, req->wb_pgbase, req->wb_bytes);
1083 dprintk(" OK\n"); 1076 dprintk(" OK\n");
1084remove_request: 1077remove_request:
1085 nfs_end_page_writeback(page); 1078 nfs_end_page_writeback(page);
@@ -1133,7 +1126,7 @@ int nfs_writeback_done(struct rpc_task *task, struct nfs_write_data *data)
1133 static unsigned long complain; 1126 static unsigned long complain;
1134 1127
1135 if (time_before(complain, jiffies)) { 1128 if (time_before(complain, jiffies)) {
1136 dprintk("NFS: faulty NFS server %s:" 1129 dprintk("NFS: faulty NFS server %s:"
1137 " (committed = %d) != (stable = %d)\n", 1130 " (committed = %d) != (stable = %d)\n",
1138 NFS_SERVER(data->inode)->nfs_client->cl_hostname, 1131 NFS_SERVER(data->inode)->nfs_client->cl_hostname,
1139 resp->verf->committed, argp->stable); 1132 resp->verf->committed, argp->stable);
@@ -1297,12 +1290,9 @@ static void nfs_commit_release(void *calldata)
1297 while (!list_empty(&data->pages)) { 1290 while (!list_empty(&data->pages)) {
1298 req = nfs_list_entry(data->pages.next); 1291 req = nfs_list_entry(data->pages.next);
1299 nfs_list_remove_request(req); 1292 nfs_list_remove_request(req);
1300 clear_bit(PG_NEED_COMMIT, &(req)->wb_flags); 1293 nfs_clear_request_commit(req);
1301 dec_zone_page_state(req->wb_page, NR_UNSTABLE_NFS);
1302 dec_bdi_stat(req->wb_page->mapping->backing_dev_info,
1303 BDI_RECLAIMABLE);
1304 1294
1305 dprintk("NFS: commit (%s/%Ld %d@%Ld)", 1295 dprintk("NFS: commit (%s/%lld %d@%lld)",
1306 req->wb_context->path.dentry->d_inode->i_sb->s_id, 1296 req->wb_context->path.dentry->d_inode->i_sb->s_id,
1307 (long long)NFS_FILEID(req->wb_context->path.dentry->d_inode), 1297 (long long)NFS_FILEID(req->wb_context->path.dentry->d_inode),
1308 req->wb_bytes, 1298 req->wb_bytes,
@@ -1318,9 +1308,6 @@ static void nfs_commit_release(void *calldata)
1318 * returned by the server against all stored verfs. */ 1308 * returned by the server against all stored verfs. */
1319 if (!memcmp(req->wb_verf.verifier, data->verf.verifier, sizeof(data->verf.verifier))) { 1309 if (!memcmp(req->wb_verf.verifier, data->verf.verifier, sizeof(data->verf.verifier))) {
1320 /* We have a match */ 1310 /* We have a match */
1321 /* Set the PG_uptodate flag */
1322 nfs_mark_uptodate(req->wb_page, req->wb_pgbase,
1323 req->wb_bytes);
1324 nfs_inode_remove_request(req); 1311 nfs_inode_remove_request(req);
1325 dprintk(" OK\n"); 1312 dprintk(" OK\n");
1326 goto next; 1313 goto next;
@@ -1479,7 +1466,7 @@ int nfs_wb_page_cancel(struct inode *inode, struct page *page)
1479 req = nfs_page_find_request(page); 1466 req = nfs_page_find_request(page);
1480 if (req == NULL) 1467 if (req == NULL)
1481 goto out; 1468 goto out;
1482 if (test_bit(PG_NEED_COMMIT, &req->wb_flags)) { 1469 if (test_bit(PG_CLEAN, &req->wb_flags)) {
1483 nfs_release_request(req); 1470 nfs_release_request(req);
1484 break; 1471 break;
1485 } 1472 }
diff --git a/fs/nfsd/nfs4callback.c b/fs/nfsd/nfs4callback.c
index 4d4760e687c3..702fa577aa6e 100644
--- a/fs/nfsd/nfs4callback.c
+++ b/fs/nfsd/nfs4callback.c
@@ -381,7 +381,7 @@ static int do_probe_callback(void *data)
381 .program = &cb_program, 381 .program = &cb_program,
382 .version = nfs_cb_version[1]->number, 382 .version = nfs_cb_version[1]->number,
383 .authflavor = RPC_AUTH_UNIX, /* XXX: need AUTH_GSS... */ 383 .authflavor = RPC_AUTH_UNIX, /* XXX: need AUTH_GSS... */
384 .flags = (RPC_CLNT_CREATE_NOPING), 384 .flags = (RPC_CLNT_CREATE_NOPING | RPC_CLNT_CREATE_QUIET),
385 }; 385 };
386 struct rpc_message msg = { 386 struct rpc_message msg = {
387 .rpc_proc = &nfs4_cb_procedures[NFSPROC4_CLNT_CB_NULL], 387 .rpc_proc = &nfs4_cb_procedures[NFSPROC4_CLNT_CB_NULL],
diff --git a/include/linux/inet.h b/include/linux/inet.h
index 1354080cf8cf..4cca05c9678e 100644
--- a/include/linux/inet.h
+++ b/include/linux/inet.h
@@ -44,6 +44,13 @@
44 44
45#include <linux/types.h> 45#include <linux/types.h>
46 46
47/*
48 * These mimic similar macros defined in user-space for inet_ntop(3).
49 * See /usr/include/netinet/in.h .
50 */
51#define INET_ADDRSTRLEN (16)
52#define INET6_ADDRSTRLEN (48)
53
47extern __be32 in_aton(const char *str); 54extern __be32 in_aton(const char *str);
48extern int in4_pton(const char *src, int srclen, u8 *dst, int delim, const char **end); 55extern int in4_pton(const char *src, int srclen, u8 *dst, int delim, const char **end);
49extern int in6_pton(const char *src, int srclen, u8 *dst, int delim, const char **end); 56extern int in6_pton(const char *src, int srclen, u8 *dst, int delim, const char **end);
diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h
index 27d6a8d98cef..29d261918734 100644
--- a/include/linux/nfs_fs.h
+++ b/include/linux/nfs_fs.h
@@ -12,9 +12,19 @@
12#include <linux/magic.h> 12#include <linux/magic.h>
13 13
14/* Default timeout values */ 14/* Default timeout values */
15#define NFS_DEF_UDP_TIMEO (11)
16#define NFS_DEF_UDP_RETRANS (3)
17#define NFS_DEF_TCP_TIMEO (600)
18#define NFS_DEF_TCP_RETRANS (2)
19
15#define NFS_MAX_UDP_TIMEOUT (60*HZ) 20#define NFS_MAX_UDP_TIMEOUT (60*HZ)
16#define NFS_MAX_TCP_TIMEOUT (600*HZ) 21#define NFS_MAX_TCP_TIMEOUT (600*HZ)
17 22
23#define NFS_DEF_ACREGMIN (3)
24#define NFS_DEF_ACREGMAX (60)
25#define NFS_DEF_ACDIRMIN (30)
26#define NFS_DEF_ACDIRMAX (60)
27
18/* 28/*
19 * When flushing a cluster of dirty pages, there can be different 29 * When flushing a cluster of dirty pages, there can be different
20 * strategies: 30 * strategies:
diff --git a/include/linux/nfs_iostat.h b/include/linux/nfs_iostat.h
new file mode 100644
index 000000000000..1cb9a3fed2b3
--- /dev/null
+++ b/include/linux/nfs_iostat.h
@@ -0,0 +1,119 @@
1/*
2 * User-space visible declarations for NFS client per-mount
3 * point statistics
4 *
5 * Copyright (C) 2005, 2006 Chuck Lever <cel@netapp.com>
6 *
7 * NFS client per-mount statistics provide information about the
8 * health of the NFS client and the health of each NFS mount point.
9 * Generally these are not for detailed problem diagnosis, but
10 * simply to indicate that there is a problem.
11 *
12 * These counters are not meant to be human-readable, but are meant
13 * to be integrated into system monitoring tools such as "sar" and
14 * "iostat". As such, the counters are sampled by the tools over
15 * time, and are never zeroed after a file system is mounted.
16 * Moving averages can be computed by the tools by taking the
17 * difference between two instantaneous samples and dividing that
18 * by the time between the samples.
19 */
20
21#ifndef _LINUX_NFS_IOSTAT
22#define _LINUX_NFS_IOSTAT
23
24#define NFS_IOSTAT_VERS "1.0"
25
26/*
27 * NFS byte counters
28 *
29 * 1. SERVER - the number of payload bytes read from or written
30 * to the server by the NFS client via an NFS READ or WRITE
31 * request.
32 *
33 * 2. NORMAL - the number of bytes read or written by applications
34 * via the read(2) and write(2) system call interfaces.
35 *
36 * 3. DIRECT - the number of bytes read or written from files
37 * opened with the O_DIRECT flag.
38 *
39 * These counters give a view of the data throughput into and out
40 * of the NFS client. Comparing the number of bytes requested by
41 * an application with the number of bytes the client requests from
42 * the server can provide an indication of client efficiency
43 * (per-op, cache hits, etc).
44 *
45 * These counters can also help characterize which access methods
46 * are in use. DIRECT by itself shows whether there is any O_DIRECT
47 * traffic. NORMAL + DIRECT shows how much data is going through
48 * the system call interface. A large amount of SERVER traffic
49 * without much NORMAL or DIRECT traffic shows that applications
50 * are using mapped files.
51 *
52 * NFS page counters
53 *
54 * These count the number of pages read or written via nfs_readpage(),
55 * nfs_readpages(), or their write equivalents.
56 *
57 * NB: When adding new byte counters, please include the measured
58 * units in the name of each byte counter to help users of this
59 * interface determine what exactly is being counted.
60 */
61enum nfs_stat_bytecounters {
62 NFSIOS_NORMALREADBYTES = 0,
63 NFSIOS_NORMALWRITTENBYTES,
64 NFSIOS_DIRECTREADBYTES,
65 NFSIOS_DIRECTWRITTENBYTES,
66 NFSIOS_SERVERREADBYTES,
67 NFSIOS_SERVERWRITTENBYTES,
68 NFSIOS_READPAGES,
69 NFSIOS_WRITEPAGES,
70 __NFSIOS_BYTESMAX,
71};
72
73/*
74 * NFS event counters
75 *
76 * These counters provide a low-overhead way of monitoring client
77 * activity without enabling NFS trace debugging. The counters
78 * show the rate at which VFS requests are made, and how often the
79 * client invalidates its data and attribute caches. This allows
80 * system administrators to monitor such things as how close-to-open
81 * is working, and answer questions such as "why are there so many
82 * GETATTR requests on the wire?"
83 *
84 * They also count anamolous events such as short reads and writes,
85 * silly renames due to close-after-delete, and operations that
86 * change the size of a file (such operations can often be the
87 * source of data corruption if applications aren't using file
88 * locking properly).
89 */
90enum nfs_stat_eventcounters {
91 NFSIOS_INODEREVALIDATE = 0,
92 NFSIOS_DENTRYREVALIDATE,
93 NFSIOS_DATAINVALIDATE,
94 NFSIOS_ATTRINVALIDATE,
95 NFSIOS_VFSOPEN,
96 NFSIOS_VFSLOOKUP,
97 NFSIOS_VFSACCESS,
98 NFSIOS_VFSUPDATEPAGE,
99 NFSIOS_VFSREADPAGE,
100 NFSIOS_VFSREADPAGES,
101 NFSIOS_VFSWRITEPAGE,
102 NFSIOS_VFSWRITEPAGES,
103 NFSIOS_VFSGETDENTS,
104 NFSIOS_VFSSETATTR,
105 NFSIOS_VFSFLUSH,
106 NFSIOS_VFSFSYNC,
107 NFSIOS_VFSLOCK,
108 NFSIOS_VFSRELEASE,
109 NFSIOS_CONGESTIONWAIT,
110 NFSIOS_SETATTRTRUNC,
111 NFSIOS_EXTENDWRITE,
112 NFSIOS_SILLYRENAME,
113 NFSIOS_SHORTREAD,
114 NFSIOS_SHORTWRITE,
115 NFSIOS_DELAY,
116 __NFSIOS_COUNTSMAX,
117};
118
119#endif /* _LINUX_NFS_IOSTAT */
diff --git a/include/linux/nfs_page.h b/include/linux/nfs_page.h
index a1676e19e491..3c60685d972b 100644
--- a/include/linux/nfs_page.h
+++ b/include/linux/nfs_page.h
@@ -27,9 +27,12 @@
27/* 27/*
28 * Valid flags for a dirty buffer 28 * Valid flags for a dirty buffer
29 */ 29 */
30#define PG_BUSY 0 30enum {
31#define PG_NEED_COMMIT 1 31 PG_BUSY = 0,
32#define PG_NEED_RESCHED 2 32 PG_CLEAN,
33 PG_NEED_COMMIT,
34 PG_NEED_RESCHED,
35};
33 36
34struct nfs_inode; 37struct nfs_inode;
35struct nfs_page { 38struct nfs_page {
diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h
index 24263bb8e0be..8c77c11224d1 100644
--- a/include/linux/nfs_xdr.h
+++ b/include/linux/nfs_xdr.h
@@ -829,9 +829,8 @@ struct nfs_rpc_ops {
829 int (*write_done) (struct rpc_task *, struct nfs_write_data *); 829 int (*write_done) (struct rpc_task *, struct nfs_write_data *);
830 void (*commit_setup) (struct nfs_write_data *, struct rpc_message *); 830 void (*commit_setup) (struct nfs_write_data *, struct rpc_message *);
831 int (*commit_done) (struct rpc_task *, struct nfs_write_data *); 831 int (*commit_done) (struct rpc_task *, struct nfs_write_data *);
832 int (*file_open) (struct inode *, struct file *);
833 int (*file_release) (struct inode *, struct file *);
834 int (*lock)(struct file *, int, struct file_lock *); 832 int (*lock)(struct file *, int, struct file_lock *);
833 int (*lock_check_bounds)(const struct file_lock *);
835 void (*clear_acl_cache)(struct inode *); 834 void (*clear_acl_cache)(struct inode *);
836}; 835};
837 836
diff --git a/include/linux/sunrpc/clnt.h b/include/linux/sunrpc/clnt.h
index 6fff7f82ef12..e5bfe01ee305 100644
--- a/include/linux/sunrpc/clnt.h
+++ b/include/linux/sunrpc/clnt.h
@@ -42,7 +42,8 @@ struct rpc_clnt {
42 42
43 unsigned int cl_softrtry : 1,/* soft timeouts */ 43 unsigned int cl_softrtry : 1,/* soft timeouts */
44 cl_discrtry : 1,/* disconnect before retry */ 44 cl_discrtry : 1,/* disconnect before retry */
45 cl_autobind : 1;/* use getport() */ 45 cl_autobind : 1,/* use getport() */
46 cl_chatty : 1;/* be verbose */
46 47
47 struct rpc_rtt * cl_rtt; /* RTO estimator data */ 48 struct rpc_rtt * cl_rtt; /* RTO estimator data */
48 const struct rpc_timeout *cl_timeout; /* Timeout strategy */ 49 const struct rpc_timeout *cl_timeout; /* Timeout strategy */
@@ -114,6 +115,7 @@ struct rpc_create_args {
114#define RPC_CLNT_CREATE_NONPRIVPORT (1UL << 3) 115#define RPC_CLNT_CREATE_NONPRIVPORT (1UL << 3)
115#define RPC_CLNT_CREATE_NOPING (1UL << 4) 116#define RPC_CLNT_CREATE_NOPING (1UL << 4)
116#define RPC_CLNT_CREATE_DISCRTRY (1UL << 5) 117#define RPC_CLNT_CREATE_DISCRTRY (1UL << 5)
118#define RPC_CLNT_CREATE_QUIET (1UL << 6)
117 119
118struct rpc_clnt *rpc_create(struct rpc_create_args *args); 120struct rpc_clnt *rpc_create(struct rpc_create_args *args);
119struct rpc_clnt *rpc_bind_new_program(struct rpc_clnt *, 121struct rpc_clnt *rpc_bind_new_program(struct rpc_clnt *,
@@ -123,6 +125,9 @@ void rpc_shutdown_client(struct rpc_clnt *);
123void rpc_release_client(struct rpc_clnt *); 125void rpc_release_client(struct rpc_clnt *);
124 126
125int rpcb_register(u32, u32, int, unsigned short, int *); 127int rpcb_register(u32, u32, int, unsigned short, int *);
128int rpcb_v4_register(const u32 program, const u32 version,
129 const struct sockaddr *address,
130 const char *netid, int *result);
126int rpcb_getport_sync(struct sockaddr_in *, u32, u32, int); 131int rpcb_getport_sync(struct sockaddr_in *, u32, u32, int);
127void rpcb_getport_async(struct rpc_task *); 132void rpcb_getport_async(struct rpc_task *);
128 133
diff --git a/include/linux/sunrpc/sched.h b/include/linux/sunrpc/sched.h
index d1a5c8c1a0f1..64981a2f1cae 100644
--- a/include/linux/sunrpc/sched.h
+++ b/include/linux/sunrpc/sched.h
@@ -135,7 +135,6 @@ struct rpc_task_setup {
135#define RPC_IS_SWAPPER(t) ((t)->tk_flags & RPC_TASK_SWAPPER) 135#define RPC_IS_SWAPPER(t) ((t)->tk_flags & RPC_TASK_SWAPPER)
136#define RPC_DO_ROOTOVERRIDE(t) ((t)->tk_flags & RPC_TASK_ROOTCREDS) 136#define RPC_DO_ROOTOVERRIDE(t) ((t)->tk_flags & RPC_TASK_ROOTCREDS)
137#define RPC_ASSASSINATED(t) ((t)->tk_flags & RPC_TASK_KILLED) 137#define RPC_ASSASSINATED(t) ((t)->tk_flags & RPC_TASK_KILLED)
138#define RPC_DO_CALLBACK(t) ((t)->tk_callback != NULL)
139#define RPC_IS_SOFT(t) ((t)->tk_flags & RPC_TASK_SOFT) 138#define RPC_IS_SOFT(t) ((t)->tk_flags & RPC_TASK_SOFT)
140 139
141#define RPC_TASK_RUNNING 0 140#define RPC_TASK_RUNNING 0
diff --git a/net/sunrpc/auth_gss/auth_gss.c b/net/sunrpc/auth_gss/auth_gss.c
index cc12d5f5d5da..834a83199bdf 100644
--- a/net/sunrpc/auth_gss/auth_gss.c
+++ b/net/sunrpc/auth_gss/auth_gss.c
@@ -63,22 +63,11 @@ static const struct rpc_credops gss_nullops;
63# define RPCDBG_FACILITY RPCDBG_AUTH 63# define RPCDBG_FACILITY RPCDBG_AUTH
64#endif 64#endif
65 65
66#define NFS_NGROUPS 16 66#define GSS_CRED_SLACK 1024
67
68#define GSS_CRED_SLACK 1024 /* XXX: unused */
69/* length of a krb5 verifier (48), plus data added before arguments when 67/* length of a krb5 verifier (48), plus data added before arguments when
70 * using integrity (two 4-byte integers): */ 68 * using integrity (two 4-byte integers): */
71#define GSS_VERF_SLACK 100 69#define GSS_VERF_SLACK 100
72 70
73/* XXX this define must match the gssd define
74* as it is passed to gssd to signal the use of
75* machine creds should be part of the shared rpc interface */
76
77#define CA_RUN_AS_MACHINE 0x00000200
78
79/* dump the buffer in `emacs-hexl' style */
80#define isprint(c) ((c > 0x1f) && (c < 0x7f))
81
82struct gss_auth { 71struct gss_auth {
83 struct kref kref; 72 struct kref kref;
84 struct rpc_auth rpc_auth; 73 struct rpc_auth rpc_auth;
@@ -146,7 +135,7 @@ simple_get_netobj(const void *p, const void *end, struct xdr_netobj *dest)
146 q = (const void *)((const char *)p + len); 135 q = (const void *)((const char *)p + len);
147 if (unlikely(q > end || q < p)) 136 if (unlikely(q > end || q < p))
148 return ERR_PTR(-EFAULT); 137 return ERR_PTR(-EFAULT);
149 dest->data = kmemdup(p, len, GFP_KERNEL); 138 dest->data = kmemdup(p, len, GFP_NOFS);
150 if (unlikely(dest->data == NULL)) 139 if (unlikely(dest->data == NULL))
151 return ERR_PTR(-ENOMEM); 140 return ERR_PTR(-ENOMEM);
152 dest->len = len; 141 dest->len = len;
@@ -171,7 +160,7 @@ gss_alloc_context(void)
171{ 160{
172 struct gss_cl_ctx *ctx; 161 struct gss_cl_ctx *ctx;
173 162
174 ctx = kzalloc(sizeof(*ctx), GFP_KERNEL); 163 ctx = kzalloc(sizeof(*ctx), GFP_NOFS);
175 if (ctx != NULL) { 164 if (ctx != NULL) {
176 ctx->gc_proc = RPC_GSS_PROC_DATA; 165 ctx->gc_proc = RPC_GSS_PROC_DATA;
177 ctx->gc_seq = 1; /* NetApp 6.4R1 doesn't accept seq. no. 0 */ 166 ctx->gc_seq = 1; /* NetApp 6.4R1 doesn't accept seq. no. 0 */
@@ -272,7 +261,7 @@ __gss_find_upcall(struct rpc_inode *rpci, uid_t uid)
272 return NULL; 261 return NULL;
273} 262}
274 263
275/* Try to add a upcall to the pipefs queue. 264/* Try to add an upcall to the pipefs queue.
276 * If an upcall owned by our uid already exists, then we return a reference 265 * If an upcall owned by our uid already exists, then we return a reference
277 * to that upcall instead of adding the new upcall. 266 * to that upcall instead of adding the new upcall.
278 */ 267 */
@@ -341,7 +330,7 @@ gss_alloc_msg(struct gss_auth *gss_auth, uid_t uid)
341{ 330{
342 struct gss_upcall_msg *gss_msg; 331 struct gss_upcall_msg *gss_msg;
343 332
344 gss_msg = kzalloc(sizeof(*gss_msg), GFP_KERNEL); 333 gss_msg = kzalloc(sizeof(*gss_msg), GFP_NOFS);
345 if (gss_msg != NULL) { 334 if (gss_msg != NULL) {
346 INIT_LIST_HEAD(&gss_msg->list); 335 INIT_LIST_HEAD(&gss_msg->list);
347 rpc_init_wait_queue(&gss_msg->rpc_waitqueue, "RPCSEC_GSS upcall waitq"); 336 rpc_init_wait_queue(&gss_msg->rpc_waitqueue, "RPCSEC_GSS upcall waitq");
@@ -493,7 +482,6 @@ gss_pipe_downcall(struct file *filp, const char __user *src, size_t mlen)
493{ 482{
494 const void *p, *end; 483 const void *p, *end;
495 void *buf; 484 void *buf;
496 struct rpc_clnt *clnt;
497 struct gss_upcall_msg *gss_msg; 485 struct gss_upcall_msg *gss_msg;
498 struct inode *inode = filp->f_path.dentry->d_inode; 486 struct inode *inode = filp->f_path.dentry->d_inode;
499 struct gss_cl_ctx *ctx; 487 struct gss_cl_ctx *ctx;
@@ -503,11 +491,10 @@ gss_pipe_downcall(struct file *filp, const char __user *src, size_t mlen)
503 if (mlen > MSG_BUF_MAXSIZE) 491 if (mlen > MSG_BUF_MAXSIZE)
504 goto out; 492 goto out;
505 err = -ENOMEM; 493 err = -ENOMEM;
506 buf = kmalloc(mlen, GFP_KERNEL); 494 buf = kmalloc(mlen, GFP_NOFS);
507 if (!buf) 495 if (!buf)
508 goto out; 496 goto out;
509 497
510 clnt = RPC_I(inode)->private;
511 err = -EFAULT; 498 err = -EFAULT;
512 if (copy_from_user(buf, src, mlen)) 499 if (copy_from_user(buf, src, mlen))
513 goto err; 500 goto err;
@@ -806,7 +793,7 @@ gss_create_cred(struct rpc_auth *auth, struct auth_cred *acred, int flags)
806 dprintk("RPC: gss_create_cred for uid %d, flavor %d\n", 793 dprintk("RPC: gss_create_cred for uid %d, flavor %d\n",
807 acred->uid, auth->au_flavor); 794 acred->uid, auth->au_flavor);
808 795
809 if (!(cred = kzalloc(sizeof(*cred), GFP_KERNEL))) 796 if (!(cred = kzalloc(sizeof(*cred), GFP_NOFS)))
810 goto out_err; 797 goto out_err;
811 798
812 rpcauth_init_cred(&cred->gc_base, acred, auth, &gss_credops); 799 rpcauth_init_cred(&cred->gc_base, acred, auth, &gss_credops);
diff --git a/net/sunrpc/auth_gss/gss_krb5_mech.c b/net/sunrpc/auth_gss/gss_krb5_mech.c
index 60c3dba545d7..ef45eba22485 100644
--- a/net/sunrpc/auth_gss/gss_krb5_mech.c
+++ b/net/sunrpc/auth_gss/gss_krb5_mech.c
@@ -70,7 +70,7 @@ simple_get_netobj(const void *p, const void *end, struct xdr_netobj *res)
70 q = (const void *)((const char *)p + len); 70 q = (const void *)((const char *)p + len);
71 if (unlikely(q > end || q < p)) 71 if (unlikely(q > end || q < p))
72 return ERR_PTR(-EFAULT); 72 return ERR_PTR(-EFAULT);
73 res->data = kmemdup(p, len, GFP_KERNEL); 73 res->data = kmemdup(p, len, GFP_NOFS);
74 if (unlikely(res->data == NULL)) 74 if (unlikely(res->data == NULL))
75 return ERR_PTR(-ENOMEM); 75 return ERR_PTR(-ENOMEM);
76 res->len = len; 76 res->len = len;
@@ -131,7 +131,7 @@ gss_import_sec_context_kerberos(const void *p,
131 struct krb5_ctx *ctx; 131 struct krb5_ctx *ctx;
132 int tmp; 132 int tmp;
133 133
134 if (!(ctx = kzalloc(sizeof(*ctx), GFP_KERNEL))) 134 if (!(ctx = kzalloc(sizeof(*ctx), GFP_NOFS)))
135 goto out_err; 135 goto out_err;
136 136
137 p = simple_get_bytes(p, end, &ctx->initiate, sizeof(ctx->initiate)); 137 p = simple_get_bytes(p, end, &ctx->initiate, sizeof(ctx->initiate));
diff --git a/net/sunrpc/auth_gss/gss_spkm3_mech.c b/net/sunrpc/auth_gss/gss_spkm3_mech.c
index 5deb4b6e4514..035e1dd6af1b 100644
--- a/net/sunrpc/auth_gss/gss_spkm3_mech.c
+++ b/net/sunrpc/auth_gss/gss_spkm3_mech.c
@@ -76,7 +76,7 @@ simple_get_netobj(const void *p, const void *end, struct xdr_netobj *res)
76 q = (const void *)((const char *)p + len); 76 q = (const void *)((const char *)p + len);
77 if (unlikely(q > end || q < p)) 77 if (unlikely(q > end || q < p))
78 return ERR_PTR(-EFAULT); 78 return ERR_PTR(-EFAULT);
79 res->data = kmemdup(p, len, GFP_KERNEL); 79 res->data = kmemdup(p, len, GFP_NOFS);
80 if (unlikely(res->data == NULL)) 80 if (unlikely(res->data == NULL))
81 return ERR_PTR(-ENOMEM); 81 return ERR_PTR(-ENOMEM);
82 return q; 82 return q;
@@ -90,7 +90,7 @@ gss_import_sec_context_spkm3(const void *p, size_t len,
90 struct spkm3_ctx *ctx; 90 struct spkm3_ctx *ctx;
91 int version; 91 int version;
92 92
93 if (!(ctx = kzalloc(sizeof(*ctx), GFP_KERNEL))) 93 if (!(ctx = kzalloc(sizeof(*ctx), GFP_NOFS)))
94 goto out_err; 94 goto out_err;
95 95
96 p = simple_get_bytes(p, end, &version, sizeof(version)); 96 p = simple_get_bytes(p, end, &version, sizeof(version));
diff --git a/net/sunrpc/auth_gss/gss_spkm3_token.c b/net/sunrpc/auth_gss/gss_spkm3_token.c
index 6cdd241ad267..3308157436d2 100644
--- a/net/sunrpc/auth_gss/gss_spkm3_token.c
+++ b/net/sunrpc/auth_gss/gss_spkm3_token.c
@@ -90,7 +90,7 @@ asn1_bitstring_len(struct xdr_netobj *in, int *enclen, int *zerobits)
90int 90int
91decode_asn1_bitstring(struct xdr_netobj *out, char *in, int enclen, int explen) 91decode_asn1_bitstring(struct xdr_netobj *out, char *in, int enclen, int explen)
92{ 92{
93 if (!(out->data = kzalloc(explen,GFP_KERNEL))) 93 if (!(out->data = kzalloc(explen,GFP_NOFS)))
94 return 0; 94 return 0;
95 out->len = explen; 95 out->len = explen;
96 memcpy(out->data, in, enclen); 96 memcpy(out->data, in, enclen);
diff --git a/net/sunrpc/auth_unix.c b/net/sunrpc/auth_unix.c
index 44920b90bdc4..46b2647c5bd2 100644
--- a/net/sunrpc/auth_unix.c
+++ b/net/sunrpc/auth_unix.c
@@ -66,7 +66,7 @@ unx_create_cred(struct rpc_auth *auth, struct auth_cred *acred, int flags)
66 dprintk("RPC: allocating UNIX cred for uid %d gid %d\n", 66 dprintk("RPC: allocating UNIX cred for uid %d gid %d\n",
67 acred->uid, acred->gid); 67 acred->uid, acred->gid);
68 68
69 if (!(cred = kmalloc(sizeof(*cred), GFP_KERNEL))) 69 if (!(cred = kmalloc(sizeof(*cred), GFP_NOFS)))
70 return ERR_PTR(-ENOMEM); 70 return ERR_PTR(-ENOMEM);
71 71
72 rpcauth_init_cred(&cred->uc_base, acred, auth, &unix_credops); 72 rpcauth_init_cred(&cred->uc_base, acred, auth, &unix_credops);
diff --git a/net/sunrpc/clnt.c b/net/sunrpc/clnt.c
index 8945307556ec..76739e928d0d 100644
--- a/net/sunrpc/clnt.c
+++ b/net/sunrpc/clnt.c
@@ -25,6 +25,7 @@
25 25
26#include <linux/module.h> 26#include <linux/module.h>
27#include <linux/types.h> 27#include <linux/types.h>
28#include <linux/kallsyms.h>
28#include <linux/mm.h> 29#include <linux/mm.h>
29#include <linux/slab.h> 30#include <linux/slab.h>
30#include <linux/smp_lock.h> 31#include <linux/smp_lock.h>
@@ -58,7 +59,6 @@ static void call_start(struct rpc_task *task);
58static void call_reserve(struct rpc_task *task); 59static void call_reserve(struct rpc_task *task);
59static void call_reserveresult(struct rpc_task *task); 60static void call_reserveresult(struct rpc_task *task);
60static void call_allocate(struct rpc_task *task); 61static void call_allocate(struct rpc_task *task);
61static void call_encode(struct rpc_task *task);
62static void call_decode(struct rpc_task *task); 62static void call_decode(struct rpc_task *task);
63static void call_bind(struct rpc_task *task); 63static void call_bind(struct rpc_task *task);
64static void call_bind_status(struct rpc_task *task); 64static void call_bind_status(struct rpc_task *task);
@@ -70,9 +70,9 @@ static void call_refreshresult(struct rpc_task *task);
70static void call_timeout(struct rpc_task *task); 70static void call_timeout(struct rpc_task *task);
71static void call_connect(struct rpc_task *task); 71static void call_connect(struct rpc_task *task);
72static void call_connect_status(struct rpc_task *task); 72static void call_connect_status(struct rpc_task *task);
73static __be32 * call_header(struct rpc_task *task);
74static __be32 * call_verify(struct rpc_task *task);
75 73
74static __be32 *rpc_encode_header(struct rpc_task *task);
75static __be32 *rpc_verify_header(struct rpc_task *task);
76static int rpc_ping(struct rpc_clnt *clnt, int flags); 76static int rpc_ping(struct rpc_clnt *clnt, int flags);
77 77
78static void rpc_register_client(struct rpc_clnt *clnt) 78static void rpc_register_client(struct rpc_clnt *clnt)
@@ -324,6 +324,8 @@ struct rpc_clnt *rpc_create(struct rpc_create_args *args)
324 clnt->cl_autobind = 1; 324 clnt->cl_autobind = 1;
325 if (args->flags & RPC_CLNT_CREATE_DISCRTRY) 325 if (args->flags & RPC_CLNT_CREATE_DISCRTRY)
326 clnt->cl_discrtry = 1; 326 clnt->cl_discrtry = 1;
327 if (!(args->flags & RPC_CLNT_CREATE_QUIET))
328 clnt->cl_chatty = 1;
327 329
328 return clnt; 330 return clnt;
329} 331}
@@ -690,6 +692,21 @@ rpc_restart_call(struct rpc_task *task)
690} 692}
691EXPORT_SYMBOL_GPL(rpc_restart_call); 693EXPORT_SYMBOL_GPL(rpc_restart_call);
692 694
695#ifdef RPC_DEBUG
696static const char *rpc_proc_name(const struct rpc_task *task)
697{
698 const struct rpc_procinfo *proc = task->tk_msg.rpc_proc;
699
700 if (proc) {
701 if (proc->p_name)
702 return proc->p_name;
703 else
704 return "NULL";
705 } else
706 return "no proc";
707}
708#endif
709
693/* 710/*
694 * 0. Initial state 711 * 0. Initial state
695 * 712 *
@@ -701,9 +718,9 @@ call_start(struct rpc_task *task)
701{ 718{
702 struct rpc_clnt *clnt = task->tk_client; 719 struct rpc_clnt *clnt = task->tk_client;
703 720
704 dprintk("RPC: %5u call_start %s%d proc %d (%s)\n", task->tk_pid, 721 dprintk("RPC: %5u call_start %s%d proc %s (%s)\n", task->tk_pid,
705 clnt->cl_protname, clnt->cl_vers, 722 clnt->cl_protname, clnt->cl_vers,
706 task->tk_msg.rpc_proc->p_proc, 723 rpc_proc_name(task),
707 (RPC_IS_ASYNC(task) ? "async" : "sync")); 724 (RPC_IS_ASYNC(task) ? "async" : "sync"));
708 725
709 /* Increment call count */ 726 /* Increment call count */
@@ -861,7 +878,7 @@ rpc_xdr_buf_init(struct xdr_buf *buf, void *start, size_t len)
861 * 3. Encode arguments of an RPC call 878 * 3. Encode arguments of an RPC call
862 */ 879 */
863static void 880static void
864call_encode(struct rpc_task *task) 881rpc_xdr_encode(struct rpc_task *task)
865{ 882{
866 struct rpc_rqst *req = task->tk_rqstp; 883 struct rpc_rqst *req = task->tk_rqstp;
867 kxdrproc_t encode; 884 kxdrproc_t encode;
@@ -876,23 +893,19 @@ call_encode(struct rpc_task *task)
876 (char *)req->rq_buffer + req->rq_callsize, 893 (char *)req->rq_buffer + req->rq_callsize,
877 req->rq_rcvsize); 894 req->rq_rcvsize);
878 895
879 /* Encode header and provided arguments */ 896 p = rpc_encode_header(task);
880 encode = task->tk_msg.rpc_proc->p_encode; 897 if (p == NULL) {
881 if (!(p = call_header(task))) { 898 printk(KERN_INFO "RPC: couldn't encode RPC header, exit EIO\n");
882 printk(KERN_INFO "RPC: call_header failed, exit EIO\n");
883 rpc_exit(task, -EIO); 899 rpc_exit(task, -EIO);
884 return; 900 return;
885 } 901 }
902
903 encode = task->tk_msg.rpc_proc->p_encode;
886 if (encode == NULL) 904 if (encode == NULL)
887 return; 905 return;
888 906
889 task->tk_status = rpcauth_wrap_req(task, encode, req, p, 907 task->tk_status = rpcauth_wrap_req(task, encode, req, p,
890 task->tk_msg.rpc_argp); 908 task->tk_msg.rpc_argp);
891 if (task->tk_status == -ENOMEM) {
892 /* XXX: Is this sane? */
893 rpc_delay(task, 3*HZ);
894 task->tk_status = -EAGAIN;
895 }
896} 909}
897 910
898/* 911/*
@@ -929,11 +942,9 @@ call_bind_status(struct rpc_task *task)
929 } 942 }
930 943
931 switch (task->tk_status) { 944 switch (task->tk_status) {
932 case -EAGAIN: 945 case -ENOMEM:
933 dprintk("RPC: %5u rpcbind waiting for another request " 946 dprintk("RPC: %5u rpcbind out of memory\n", task->tk_pid);
934 "to finish\n", task->tk_pid); 947 rpc_delay(task, HZ >> 2);
935 /* avoid busy-waiting here -- could be a network outage. */
936 rpc_delay(task, 5*HZ);
937 goto retry_timeout; 948 goto retry_timeout;
938 case -EACCES: 949 case -EACCES:
939 dprintk("RPC: %5u remote rpcbind: RPC program/version " 950 dprintk("RPC: %5u remote rpcbind: RPC program/version "
@@ -1046,10 +1057,16 @@ call_transmit(struct rpc_task *task)
1046 /* Encode here so that rpcsec_gss can use correct sequence number. */ 1057 /* Encode here so that rpcsec_gss can use correct sequence number. */
1047 if (rpc_task_need_encode(task)) { 1058 if (rpc_task_need_encode(task)) {
1048 BUG_ON(task->tk_rqstp->rq_bytes_sent != 0); 1059 BUG_ON(task->tk_rqstp->rq_bytes_sent != 0);
1049 call_encode(task); 1060 rpc_xdr_encode(task);
1050 /* Did the encode result in an error condition? */ 1061 /* Did the encode result in an error condition? */
1051 if (task->tk_status != 0) 1062 if (task->tk_status != 0) {
1063 /* Was the error nonfatal? */
1064 if (task->tk_status == -EAGAIN)
1065 rpc_delay(task, HZ >> 4);
1066 else
1067 rpc_exit(task, task->tk_status);
1052 return; 1068 return;
1069 }
1053 } 1070 }
1054 xprt_transmit(task); 1071 xprt_transmit(task);
1055 if (task->tk_status < 0) 1072 if (task->tk_status < 0)
@@ -1132,7 +1149,8 @@ call_status(struct rpc_task *task)
1132 rpc_exit(task, status); 1149 rpc_exit(task, status);
1133 break; 1150 break;
1134 default: 1151 default:
1135 printk("%s: RPC call returned error %d\n", 1152 if (clnt->cl_chatty)
1153 printk("%s: RPC call returned error %d\n",
1136 clnt->cl_protname, -status); 1154 clnt->cl_protname, -status);
1137 rpc_exit(task, status); 1155 rpc_exit(task, status);
1138 } 1156 }
@@ -1157,7 +1175,8 @@ call_timeout(struct rpc_task *task)
1157 task->tk_timeouts++; 1175 task->tk_timeouts++;
1158 1176
1159 if (RPC_IS_SOFT(task)) { 1177 if (RPC_IS_SOFT(task)) {
1160 printk(KERN_NOTICE "%s: server %s not responding, timed out\n", 1178 if (clnt->cl_chatty)
1179 printk(KERN_NOTICE "%s: server %s not responding, timed out\n",
1161 clnt->cl_protname, clnt->cl_server); 1180 clnt->cl_protname, clnt->cl_server);
1162 rpc_exit(task, -EIO); 1181 rpc_exit(task, -EIO);
1163 return; 1182 return;
@@ -1165,7 +1184,8 @@ call_timeout(struct rpc_task *task)
1165 1184
1166 if (!(task->tk_flags & RPC_CALL_MAJORSEEN)) { 1185 if (!(task->tk_flags & RPC_CALL_MAJORSEEN)) {
1167 task->tk_flags |= RPC_CALL_MAJORSEEN; 1186 task->tk_flags |= RPC_CALL_MAJORSEEN;
1168 printk(KERN_NOTICE "%s: server %s not responding, still trying\n", 1187 if (clnt->cl_chatty)
1188 printk(KERN_NOTICE "%s: server %s not responding, still trying\n",
1169 clnt->cl_protname, clnt->cl_server); 1189 clnt->cl_protname, clnt->cl_server);
1170 } 1190 }
1171 rpc_force_rebind(clnt); 1191 rpc_force_rebind(clnt);
@@ -1196,8 +1216,9 @@ call_decode(struct rpc_task *task)
1196 task->tk_pid, task->tk_status); 1216 task->tk_pid, task->tk_status);
1197 1217
1198 if (task->tk_flags & RPC_CALL_MAJORSEEN) { 1218 if (task->tk_flags & RPC_CALL_MAJORSEEN) {
1199 printk(KERN_NOTICE "%s: server %s OK\n", 1219 if (clnt->cl_chatty)
1200 clnt->cl_protname, clnt->cl_server); 1220 printk(KERN_NOTICE "%s: server %s OK\n",
1221 clnt->cl_protname, clnt->cl_server);
1201 task->tk_flags &= ~RPC_CALL_MAJORSEEN; 1222 task->tk_flags &= ~RPC_CALL_MAJORSEEN;
1202 } 1223 }
1203 1224
@@ -1224,8 +1245,7 @@ call_decode(struct rpc_task *task)
1224 goto out_retry; 1245 goto out_retry;
1225 } 1246 }
1226 1247
1227 /* Verify the RPC header */ 1248 p = rpc_verify_header(task);
1228 p = call_verify(task);
1229 if (IS_ERR(p)) { 1249 if (IS_ERR(p)) {
1230 if (p == ERR_PTR(-EAGAIN)) 1250 if (p == ERR_PTR(-EAGAIN))
1231 goto out_retry; 1251 goto out_retry;
@@ -1243,7 +1263,7 @@ call_decode(struct rpc_task *task)
1243 return; 1263 return;
1244out_retry: 1264out_retry:
1245 task->tk_status = 0; 1265 task->tk_status = 0;
1246 /* Note: call_verify() may have freed the RPC slot */ 1266 /* Note: rpc_verify_header() may have freed the RPC slot */
1247 if (task->tk_rqstp == req) { 1267 if (task->tk_rqstp == req) {
1248 req->rq_received = req->rq_rcv_buf.len = 0; 1268 req->rq_received = req->rq_rcv_buf.len = 0;
1249 if (task->tk_client->cl_discrtry) 1269 if (task->tk_client->cl_discrtry)
@@ -1290,11 +1310,8 @@ call_refreshresult(struct rpc_task *task)
1290 return; 1310 return;
1291} 1311}
1292 1312
1293/*
1294 * Call header serialization
1295 */
1296static __be32 * 1313static __be32 *
1297call_header(struct rpc_task *task) 1314rpc_encode_header(struct rpc_task *task)
1298{ 1315{
1299 struct rpc_clnt *clnt = task->tk_client; 1316 struct rpc_clnt *clnt = task->tk_client;
1300 struct rpc_rqst *req = task->tk_rqstp; 1317 struct rpc_rqst *req = task->tk_rqstp;
@@ -1314,11 +1331,8 @@ call_header(struct rpc_task *task)
1314 return p; 1331 return p;
1315} 1332}
1316 1333
1317/*
1318 * Reply header verification
1319 */
1320static __be32 * 1334static __be32 *
1321call_verify(struct rpc_task *task) 1335rpc_verify_header(struct rpc_task *task)
1322{ 1336{
1323 struct kvec *iov = &task->tk_rqstp->rq_rcv_buf.head[0]; 1337 struct kvec *iov = &task->tk_rqstp->rq_rcv_buf.head[0];
1324 int len = task->tk_rqstp->rq_rcv_buf.len >> 2; 1338 int len = task->tk_rqstp->rq_rcv_buf.len >> 2;
@@ -1392,7 +1406,7 @@ call_verify(struct rpc_task *task)
1392 task->tk_action = call_bind; 1406 task->tk_action = call_bind;
1393 goto out_retry; 1407 goto out_retry;
1394 case RPC_AUTH_TOOWEAK: 1408 case RPC_AUTH_TOOWEAK:
1395 printk(KERN_NOTICE "call_verify: server %s requires stronger " 1409 printk(KERN_NOTICE "RPC: server %s requires stronger "
1396 "authentication.\n", task->tk_client->cl_server); 1410 "authentication.\n", task->tk_client->cl_server);
1397 break; 1411 break;
1398 default: 1412 default:
@@ -1431,10 +1445,10 @@ call_verify(struct rpc_task *task)
1431 error = -EPROTONOSUPPORT; 1445 error = -EPROTONOSUPPORT;
1432 goto out_err; 1446 goto out_err;
1433 case RPC_PROC_UNAVAIL: 1447 case RPC_PROC_UNAVAIL:
1434 dprintk("RPC: %5u %s: proc %p unsupported by program %u, " 1448 dprintk("RPC: %5u %s: proc %s unsupported by program %u, "
1435 "version %u on server %s\n", 1449 "version %u on server %s\n",
1436 task->tk_pid, __func__, 1450 task->tk_pid, __func__,
1437 task->tk_msg.rpc_proc, 1451 rpc_proc_name(task),
1438 task->tk_client->cl_prog, 1452 task->tk_client->cl_prog,
1439 task->tk_client->cl_vers, 1453 task->tk_client->cl_vers,
1440 task->tk_client->cl_server); 1454 task->tk_client->cl_server);
@@ -1517,44 +1531,53 @@ struct rpc_task *rpc_call_null(struct rpc_clnt *clnt, struct rpc_cred *cred, int
1517EXPORT_SYMBOL_GPL(rpc_call_null); 1531EXPORT_SYMBOL_GPL(rpc_call_null);
1518 1532
1519#ifdef RPC_DEBUG 1533#ifdef RPC_DEBUG
1534static void rpc_show_header(void)
1535{
1536 printk(KERN_INFO "-pid- flgs status -client- --rqstp- "
1537 "-timeout ---ops--\n");
1538}
1539
1540static void rpc_show_task(const struct rpc_clnt *clnt,
1541 const struct rpc_task *task)
1542{
1543 const char *rpc_waitq = "none";
1544 char *p, action[KSYM_SYMBOL_LEN];
1545
1546 if (RPC_IS_QUEUED(task))
1547 rpc_waitq = rpc_qname(task->tk_waitqueue);
1548
1549 /* map tk_action pointer to a function name; then trim off
1550 * the "+0x0 [sunrpc]" */
1551 sprint_symbol(action, (unsigned long)task->tk_action);
1552 p = strchr(action, '+');
1553 if (p)
1554 *p = '\0';
1555
1556 printk(KERN_INFO "%5u %04x %6d %8p %8p %8ld %8p %sv%u %s a:%s q:%s\n",
1557 task->tk_pid, task->tk_flags, task->tk_status,
1558 clnt, task->tk_rqstp, task->tk_timeout, task->tk_ops,
1559 clnt->cl_protname, clnt->cl_vers, rpc_proc_name(task),
1560 action, rpc_waitq);
1561}
1562
1520void rpc_show_tasks(void) 1563void rpc_show_tasks(void)
1521{ 1564{
1522 struct rpc_clnt *clnt; 1565 struct rpc_clnt *clnt;
1523 struct rpc_task *t; 1566 struct rpc_task *task;
1567 int header = 0;
1524 1568
1525 spin_lock(&rpc_client_lock); 1569 spin_lock(&rpc_client_lock);
1526 if (list_empty(&all_clients))
1527 goto out;
1528 printk("-pid- proc flgs status -client- -prog- --rqstp- -timeout "
1529 "-rpcwait -action- ---ops--\n");
1530 list_for_each_entry(clnt, &all_clients, cl_clients) { 1570 list_for_each_entry(clnt, &all_clients, cl_clients) {
1531 if (list_empty(&clnt->cl_tasks))
1532 continue;
1533 spin_lock(&clnt->cl_lock); 1571 spin_lock(&clnt->cl_lock);
1534 list_for_each_entry(t, &clnt->cl_tasks, tk_task) { 1572 list_for_each_entry(task, &clnt->cl_tasks, tk_task) {
1535 const char *rpc_waitq = "none"; 1573 if (!header) {
1536 int proc; 1574 rpc_show_header();
1537 1575 header++;
1538 if (t->tk_msg.rpc_proc) 1576 }
1539 proc = t->tk_msg.rpc_proc->p_proc; 1577 rpc_show_task(clnt, task);
1540 else
1541 proc = -1;
1542
1543 if (RPC_IS_QUEUED(t))
1544 rpc_waitq = rpc_qname(t->tk_waitqueue);
1545
1546 printk("%5u %04d %04x %6d %8p %6d %8p %8ld %8s %8p %8p\n",
1547 t->tk_pid, proc,
1548 t->tk_flags, t->tk_status,
1549 t->tk_client,
1550 (t->tk_client ? t->tk_client->cl_prog : 0),
1551 t->tk_rqstp, t->tk_timeout,
1552 rpc_waitq,
1553 t->tk_action, t->tk_ops);
1554 } 1578 }
1555 spin_unlock(&clnt->cl_lock); 1579 spin_unlock(&clnt->cl_lock);
1556 } 1580 }
1557out:
1558 spin_unlock(&rpc_client_lock); 1581 spin_unlock(&rpc_client_lock);
1559} 1582}
1560#endif 1583#endif
diff --git a/net/sunrpc/rpcb_clnt.c b/net/sunrpc/rpcb_clnt.c
index e6fb21b19b86..24db2b4d12d3 100644
--- a/net/sunrpc/rpcb_clnt.c
+++ b/net/sunrpc/rpcb_clnt.c
@@ -32,6 +32,10 @@
32#define RPCBIND_PROGRAM (100000u) 32#define RPCBIND_PROGRAM (100000u)
33#define RPCBIND_PORT (111u) 33#define RPCBIND_PORT (111u)
34 34
35#define RPCBVERS_2 (2u)
36#define RPCBVERS_3 (3u)
37#define RPCBVERS_4 (4u)
38
35enum { 39enum {
36 RPCBPROC_NULL, 40 RPCBPROC_NULL,
37 RPCBPROC_SET, 41 RPCBPROC_SET,
@@ -64,6 +68,7 @@ enum {
64#define RPCB_MAXOWNERLEN sizeof(RPCB_OWNER_STRING) 68#define RPCB_MAXOWNERLEN sizeof(RPCB_OWNER_STRING)
65 69
66static void rpcb_getport_done(struct rpc_task *, void *); 70static void rpcb_getport_done(struct rpc_task *, void *);
71static void rpcb_map_release(void *data);
67static struct rpc_program rpcb_program; 72static struct rpc_program rpcb_program;
68 73
69struct rpcbind_args { 74struct rpcbind_args {
@@ -76,41 +81,73 @@ struct rpcbind_args {
76 const char * r_netid; 81 const char * r_netid;
77 const char * r_addr; 82 const char * r_addr;
78 const char * r_owner; 83 const char * r_owner;
84
85 int r_status;
79}; 86};
80 87
81static struct rpc_procinfo rpcb_procedures2[]; 88static struct rpc_procinfo rpcb_procedures2[];
82static struct rpc_procinfo rpcb_procedures3[]; 89static struct rpc_procinfo rpcb_procedures3[];
90static struct rpc_procinfo rpcb_procedures4[];
83 91
84struct rpcb_info { 92struct rpcb_info {
85 int rpc_vers; 93 u32 rpc_vers;
86 struct rpc_procinfo * rpc_proc; 94 struct rpc_procinfo * rpc_proc;
87}; 95};
88 96
89static struct rpcb_info rpcb_next_version[]; 97static struct rpcb_info rpcb_next_version[];
90static struct rpcb_info rpcb_next_version6[]; 98static struct rpcb_info rpcb_next_version6[];
91 99
100static const struct rpc_call_ops rpcb_getport_ops = {
101 .rpc_call_done = rpcb_getport_done,
102 .rpc_release = rpcb_map_release,
103};
104
105static void rpcb_wake_rpcbind_waiters(struct rpc_xprt *xprt, int status)
106{
107 xprt_clear_binding(xprt);
108 rpc_wake_up_status(&xprt->binding, status);
109}
110
92static void rpcb_map_release(void *data) 111static void rpcb_map_release(void *data)
93{ 112{
94 struct rpcbind_args *map = data; 113 struct rpcbind_args *map = data;
95 114
115 rpcb_wake_rpcbind_waiters(map->r_xprt, map->r_status);
96 xprt_put(map->r_xprt); 116 xprt_put(map->r_xprt);
97 kfree(map); 117 kfree(map);
98} 118}
99 119
100static const struct rpc_call_ops rpcb_getport_ops = { 120static const struct sockaddr_in rpcb_inaddr_loopback = {
101 .rpc_call_done = rpcb_getport_done, 121 .sin_family = AF_INET,
102 .rpc_release = rpcb_map_release, 122 .sin_addr.s_addr = htonl(INADDR_LOOPBACK),
123 .sin_port = htons(RPCBIND_PORT),
103}; 124};
104 125
105static void rpcb_wake_rpcbind_waiters(struct rpc_xprt *xprt, int status) 126static const struct sockaddr_in6 rpcb_in6addr_loopback = {
127 .sin6_family = AF_INET6,
128 .sin6_addr = IN6ADDR_LOOPBACK_INIT,
129 .sin6_port = htons(RPCBIND_PORT),
130};
131
132static struct rpc_clnt *rpcb_create_local(struct sockaddr *addr,
133 size_t addrlen, u32 version)
106{ 134{
107 xprt_clear_binding(xprt); 135 struct rpc_create_args args = {
108 rpc_wake_up_status(&xprt->binding, status); 136 .protocol = XPRT_TRANSPORT_UDP,
137 .address = addr,
138 .addrsize = addrlen,
139 .servername = "localhost",
140 .program = &rpcb_program,
141 .version = version,
142 .authflavor = RPC_AUTH_UNIX,
143 .flags = RPC_CLNT_CREATE_NOPING,
144 };
145
146 return rpc_create(&args);
109} 147}
110 148
111static struct rpc_clnt *rpcb_create(char *hostname, struct sockaddr *srvaddr, 149static struct rpc_clnt *rpcb_create(char *hostname, struct sockaddr *srvaddr,
112 size_t salen, int proto, u32 version, 150 size_t salen, int proto, u32 version)
113 int privileged)
114{ 151{
115 struct rpc_create_args args = { 152 struct rpc_create_args args = {
116 .protocol = proto, 153 .protocol = proto,
@@ -120,7 +157,8 @@ static struct rpc_clnt *rpcb_create(char *hostname, struct sockaddr *srvaddr,
120 .program = &rpcb_program, 157 .program = &rpcb_program,
121 .version = version, 158 .version = version,
122 .authflavor = RPC_AUTH_UNIX, 159 .authflavor = RPC_AUTH_UNIX,
123 .flags = RPC_CLNT_CREATE_NOPING, 160 .flags = (RPC_CLNT_CREATE_NOPING |
161 RPC_CLNT_CREATE_NONPRIVPORT),
124 }; 162 };
125 163
126 switch (srvaddr->sa_family) { 164 switch (srvaddr->sa_family) {
@@ -134,29 +172,72 @@ static struct rpc_clnt *rpcb_create(char *hostname, struct sockaddr *srvaddr,
134 return NULL; 172 return NULL;
135 } 173 }
136 174
137 if (!privileged)
138 args.flags |= RPC_CLNT_CREATE_NONPRIVPORT;
139 return rpc_create(&args); 175 return rpc_create(&args);
140} 176}
141 177
178static int rpcb_register_call(struct sockaddr *addr, size_t addrlen,
179 u32 version, struct rpc_message *msg,
180 int *result)
181{
182 struct rpc_clnt *rpcb_clnt;
183 int error = 0;
184
185 *result = 0;
186
187 rpcb_clnt = rpcb_create_local(addr, addrlen, version);
188 if (!IS_ERR(rpcb_clnt)) {
189 error = rpc_call_sync(rpcb_clnt, msg, 0);
190 rpc_shutdown_client(rpcb_clnt);
191 } else
192 error = PTR_ERR(rpcb_clnt);
193
194 if (error < 0)
195 printk(KERN_WARNING "RPC: failed to contact local rpcbind "
196 "server (errno %d).\n", -error);
197 dprintk("RPC: registration status %d/%d\n", error, *result);
198
199 return error;
200}
201
142/** 202/**
143 * rpcb_register - set or unset a port registration with the local rpcbind svc 203 * rpcb_register - set or unset a port registration with the local rpcbind svc
144 * @prog: RPC program number to bind 204 * @prog: RPC program number to bind
145 * @vers: RPC version number to bind 205 * @vers: RPC version number to bind
146 * @prot: transport protocol to use to make this request 206 * @prot: transport protocol to register
147 * @port: port value to register 207 * @port: port value to register
148 * @okay: result code 208 * @okay: OUT: result code
209 *
210 * RPC services invoke this function to advertise their contact
211 * information via the system's rpcbind daemon. RPC services
212 * invoke this function once for each [program, version, transport]
213 * tuple they wish to advertise.
214 *
215 * Callers may also unregister RPC services that are no longer
216 * available by setting the passed-in port to zero. This removes
217 * all registered transports for [program, version] from the local
218 * rpcbind database.
219 *
220 * Returns zero if the registration request was dispatched
221 * successfully and a reply was received. The rpcbind daemon's
222 * boolean result code is stored in *okay.
223 *
224 * Returns an errno value and sets *result to zero if there was
225 * some problem that prevented the rpcbind request from being
226 * dispatched, or if the rpcbind daemon did not respond within
227 * the timeout.
149 * 228 *
150 * port == 0 means unregister, port != 0 means register. 229 * This function uses rpcbind protocol version 2 to contact the
230 * local rpcbind daemon.
151 * 231 *
152 * This routine supports only rpcbind version 2. 232 * Registration works over both AF_INET and AF_INET6, and services
233 * registered via this function are advertised as available for any
234 * address. If the local rpcbind daemon is listening on AF_INET6,
235 * services registered via this function will be advertised on
236 * IN6ADDR_ANY (ie available for all AF_INET and AF_INET6
237 * addresses).
153 */ 238 */
154int rpcb_register(u32 prog, u32 vers, int prot, unsigned short port, int *okay) 239int rpcb_register(u32 prog, u32 vers, int prot, unsigned short port, int *okay)
155{ 240{
156 struct sockaddr_in sin = {
157 .sin_family = AF_INET,
158 .sin_addr.s_addr = htonl(INADDR_LOOPBACK),
159 };
160 struct rpcbind_args map = { 241 struct rpcbind_args map = {
161 .r_prog = prog, 242 .r_prog = prog,
162 .r_vers = vers, 243 .r_vers = vers,
@@ -164,32 +245,159 @@ int rpcb_register(u32 prog, u32 vers, int prot, unsigned short port, int *okay)
164 .r_port = port, 245 .r_port = port,
165 }; 246 };
166 struct rpc_message msg = { 247 struct rpc_message msg = {
167 .rpc_proc = &rpcb_procedures2[port ?
168 RPCBPROC_SET : RPCBPROC_UNSET],
169 .rpc_argp = &map, 248 .rpc_argp = &map,
170 .rpc_resp = okay, 249 .rpc_resp = okay,
171 }; 250 };
172 struct rpc_clnt *rpcb_clnt;
173 int error = 0;
174 251
175 dprintk("RPC: %sregistering (%u, %u, %d, %u) with local " 252 dprintk("RPC: %sregistering (%u, %u, %d, %u) with local "
176 "rpcbind\n", (port ? "" : "un"), 253 "rpcbind\n", (port ? "" : "un"),
177 prog, vers, prot, port); 254 prog, vers, prot, port);
178 255
179 rpcb_clnt = rpcb_create("localhost", (struct sockaddr *) &sin, 256 msg.rpc_proc = &rpcb_procedures2[RPCBPROC_UNSET];
180 sizeof(sin), XPRT_TRANSPORT_UDP, 2, 1); 257 if (port)
181 if (IS_ERR(rpcb_clnt)) 258 msg.rpc_proc = &rpcb_procedures2[RPCBPROC_SET];
182 return PTR_ERR(rpcb_clnt);
183 259
184 error = rpc_call_sync(rpcb_clnt, &msg, 0); 260 return rpcb_register_call((struct sockaddr *)&rpcb_inaddr_loopback,
261 sizeof(rpcb_inaddr_loopback),
262 RPCBVERS_2, &msg, okay);
263}
185 264
186 rpc_shutdown_client(rpcb_clnt); 265/*
187 if (error < 0) 266 * Fill in AF_INET family-specific arguments to register
188 printk(KERN_WARNING "RPC: failed to contact local rpcbind " 267 */
189 "server (errno %d).\n", -error); 268static int rpcb_register_netid4(struct sockaddr_in *address_to_register,
190 dprintk("RPC: registration status %d/%d\n", error, *okay); 269 struct rpc_message *msg)
270{
271 struct rpcbind_args *map = msg->rpc_argp;
272 unsigned short port = ntohs(address_to_register->sin_port);
273 char buf[32];
274
275 /* Construct AF_INET universal address */
276 snprintf(buf, sizeof(buf),
277 NIPQUAD_FMT".%u.%u",
278 NIPQUAD(address_to_register->sin_addr.s_addr),
279 port >> 8, port & 0xff);
280 map->r_addr = buf;
281
282 dprintk("RPC: %sregistering [%u, %u, %s, '%s'] with "
283 "local rpcbind\n", (port ? "" : "un"),
284 map->r_prog, map->r_vers,
285 map->r_addr, map->r_netid);
286
287 msg->rpc_proc = &rpcb_procedures4[RPCBPROC_UNSET];
288 if (port)
289 msg->rpc_proc = &rpcb_procedures4[RPCBPROC_SET];
290
291 return rpcb_register_call((struct sockaddr *)&rpcb_inaddr_loopback,
292 sizeof(rpcb_inaddr_loopback),
293 RPCBVERS_4, msg, msg->rpc_resp);
294}
191 295
192 return error; 296/*
297 * Fill in AF_INET6 family-specific arguments to register
298 */
299static int rpcb_register_netid6(struct sockaddr_in6 *address_to_register,
300 struct rpc_message *msg)
301{
302 struct rpcbind_args *map = msg->rpc_argp;
303 unsigned short port = ntohs(address_to_register->sin6_port);
304 char buf[64];
305
306 /* Construct AF_INET6 universal address */
307 snprintf(buf, sizeof(buf),
308 NIP6_FMT".%u.%u",
309 NIP6(address_to_register->sin6_addr),
310 port >> 8, port & 0xff);
311 map->r_addr = buf;
312
313 dprintk("RPC: %sregistering [%u, %u, %s, '%s'] with "
314 "local rpcbind\n", (port ? "" : "un"),
315 map->r_prog, map->r_vers,
316 map->r_addr, map->r_netid);
317
318 msg->rpc_proc = &rpcb_procedures4[RPCBPROC_UNSET];
319 if (port)
320 msg->rpc_proc = &rpcb_procedures4[RPCBPROC_SET];
321
322 return rpcb_register_call((struct sockaddr *)&rpcb_in6addr_loopback,
323 sizeof(rpcb_in6addr_loopback),
324 RPCBVERS_4, msg, msg->rpc_resp);
325}
326
327/**
328 * rpcb_v4_register - set or unset a port registration with the local rpcbind
329 * @program: RPC program number of service to (un)register
330 * @version: RPC version number of service to (un)register
331 * @address: address family, IP address, and port to (un)register
332 * @netid: netid of transport protocol to (un)register
333 * @result: result code from rpcbind RPC call
334 *
335 * RPC services invoke this function to advertise their contact
336 * information via the system's rpcbind daemon. RPC services
337 * invoke this function once for each [program, version, address,
338 * netid] tuple they wish to advertise.
339 *
340 * Callers may also unregister RPC services that are no longer
341 * available by setting the port number in the passed-in address
342 * to zero. Callers pass a netid of "" to unregister all
343 * transport netids associated with [program, version, address].
344 *
345 * Returns zero if the registration request was dispatched
346 * successfully and a reply was received. The rpcbind daemon's
347 * result code is stored in *result.
348 *
349 * Returns an errno value and sets *result to zero if there was
350 * some problem that prevented the rpcbind request from being
351 * dispatched, or if the rpcbind daemon did not respond within
352 * the timeout.
353 *
354 * This function uses rpcbind protocol version 4 to contact the
355 * local rpcbind daemon. The local rpcbind daemon must support
356 * version 4 of the rpcbind protocol in order for these functions
357 * to register a service successfully.
358 *
359 * Supported netids include "udp" and "tcp" for UDP and TCP over
360 * IPv4, and "udp6" and "tcp6" for UDP and TCP over IPv6,
361 * respectively.
362 *
363 * The contents of @address determine the address family and the
364 * port to be registered. The usual practice is to pass INADDR_ANY
365 * as the raw address, but specifying a non-zero address is also
366 * supported by this API if the caller wishes to advertise an RPC
367 * service on a specific network interface.
368 *
369 * Note that passing in INADDR_ANY does not create the same service
370 * registration as IN6ADDR_ANY. The former advertises an RPC
371 * service on any IPv4 address, but not on IPv6. The latter
372 * advertises the service on all IPv4 and IPv6 addresses.
373 */
374int rpcb_v4_register(const u32 program, const u32 version,
375 const struct sockaddr *address, const char *netid,
376 int *result)
377{
378 struct rpcbind_args map = {
379 .r_prog = program,
380 .r_vers = version,
381 .r_netid = netid,
382 .r_owner = RPCB_OWNER_STRING,
383 };
384 struct rpc_message msg = {
385 .rpc_argp = &map,
386 .rpc_resp = result,
387 };
388
389 *result = 0;
390
391 switch (address->sa_family) {
392 case AF_INET:
393 return rpcb_register_netid4((struct sockaddr_in *)address,
394 &msg);
395 case AF_INET6:
396 return rpcb_register_netid6((struct sockaddr_in6 *)address,
397 &msg);
398 }
399
400 return -EAFNOSUPPORT;
193} 401}
194 402
195/** 403/**
@@ -227,7 +435,7 @@ int rpcb_getport_sync(struct sockaddr_in *sin, u32 prog, u32 vers, int prot)
227 __func__, NIPQUAD(sin->sin_addr.s_addr), prog, vers, prot); 435 __func__, NIPQUAD(sin->sin_addr.s_addr), prog, vers, prot);
228 436
229 rpcb_clnt = rpcb_create(NULL, (struct sockaddr *)sin, 437 rpcb_clnt = rpcb_create(NULL, (struct sockaddr *)sin,
230 sizeof(*sin), prot, 2, 0); 438 sizeof(*sin), prot, RPCBVERS_2);
231 if (IS_ERR(rpcb_clnt)) 439 if (IS_ERR(rpcb_clnt))
232 return PTR_ERR(rpcb_clnt); 440 return PTR_ERR(rpcb_clnt);
233 441
@@ -289,17 +497,16 @@ void rpcb_getport_async(struct rpc_task *task)
289 /* Autobind on cloned rpc clients is discouraged */ 497 /* Autobind on cloned rpc clients is discouraged */
290 BUG_ON(clnt->cl_parent != clnt); 498 BUG_ON(clnt->cl_parent != clnt);
291 499
500 /* Put self on the wait queue to ensure we get notified if
501 * some other task is already attempting to bind the port */
502 rpc_sleep_on(&xprt->binding, task, NULL);
503
292 if (xprt_test_and_set_binding(xprt)) { 504 if (xprt_test_and_set_binding(xprt)) {
293 status = -EAGAIN; /* tell caller to check again */
294 dprintk("RPC: %5u %s: waiting for another binder\n", 505 dprintk("RPC: %5u %s: waiting for another binder\n",
295 task->tk_pid, __func__); 506 task->tk_pid, __func__);
296 goto bailout_nowake; 507 return;
297 } 508 }
298 509
299 /* Put self on queue before sending rpcbind request, in case
300 * rpcb_getport_done completes before we return from rpc_run_task */
301 rpc_sleep_on(&xprt->binding, task, NULL);
302
303 /* Someone else may have bound if we slept */ 510 /* Someone else may have bound if we slept */
304 if (xprt_bound(xprt)) { 511 if (xprt_bound(xprt)) {
305 status = 0; 512 status = 0;
@@ -338,7 +545,7 @@ void rpcb_getport_async(struct rpc_task *task)
338 task->tk_pid, __func__, bind_version); 545 task->tk_pid, __func__, bind_version);
339 546
340 rpcb_clnt = rpcb_create(clnt->cl_server, sap, salen, xprt->prot, 547 rpcb_clnt = rpcb_create(clnt->cl_server, sap, salen, xprt->prot,
341 bind_version, 0); 548 bind_version);
342 if (IS_ERR(rpcb_clnt)) { 549 if (IS_ERR(rpcb_clnt)) {
343 status = PTR_ERR(rpcb_clnt); 550 status = PTR_ERR(rpcb_clnt);
344 dprintk("RPC: %5u %s: rpcb_create failed, error %ld\n", 551 dprintk("RPC: %5u %s: rpcb_create failed, error %ld\n",
@@ -361,15 +568,15 @@ void rpcb_getport_async(struct rpc_task *task)
361 map->r_netid = rpc_peeraddr2str(clnt, RPC_DISPLAY_NETID); 568 map->r_netid = rpc_peeraddr2str(clnt, RPC_DISPLAY_NETID);
362 map->r_addr = rpc_peeraddr2str(rpcb_clnt, RPC_DISPLAY_UNIVERSAL_ADDR); 569 map->r_addr = rpc_peeraddr2str(rpcb_clnt, RPC_DISPLAY_UNIVERSAL_ADDR);
363 map->r_owner = RPCB_OWNER_STRING; /* ignored for GETADDR */ 570 map->r_owner = RPCB_OWNER_STRING; /* ignored for GETADDR */
571 map->r_status = -EIO;
364 572
365 child = rpcb_call_async(rpcb_clnt, map, proc); 573 child = rpcb_call_async(rpcb_clnt, map, proc);
366 rpc_release_client(rpcb_clnt); 574 rpc_release_client(rpcb_clnt);
367 if (IS_ERR(child)) { 575 if (IS_ERR(child)) {
368 status = -EIO;
369 /* rpcb_map_release() has freed the arguments */ 576 /* rpcb_map_release() has freed the arguments */
370 dprintk("RPC: %5u %s: rpc_run_task failed\n", 577 dprintk("RPC: %5u %s: rpc_run_task failed\n",
371 task->tk_pid, __func__); 578 task->tk_pid, __func__);
372 goto bailout_nofree; 579 return;
373 } 580 }
374 rpc_put_task(child); 581 rpc_put_task(child);
375 582
@@ -378,7 +585,6 @@ void rpcb_getport_async(struct rpc_task *task)
378 585
379bailout_nofree: 586bailout_nofree:
380 rpcb_wake_rpcbind_waiters(xprt, status); 587 rpcb_wake_rpcbind_waiters(xprt, status);
381bailout_nowake:
382 task->tk_status = status; 588 task->tk_status = status;
383} 589}
384EXPORT_SYMBOL_GPL(rpcb_getport_async); 590EXPORT_SYMBOL_GPL(rpcb_getport_async);
@@ -417,9 +623,13 @@ static void rpcb_getport_done(struct rpc_task *child, void *data)
417 dprintk("RPC: %5u rpcb_getport_done(status %d, port %u)\n", 623 dprintk("RPC: %5u rpcb_getport_done(status %d, port %u)\n",
418 child->tk_pid, status, map->r_port); 624 child->tk_pid, status, map->r_port);
419 625
420 rpcb_wake_rpcbind_waiters(xprt, status); 626 map->r_status = status;
421} 627}
422 628
629/*
630 * XDR functions for rpcbind
631 */
632
423static int rpcb_encode_mapping(struct rpc_rqst *req, __be32 *p, 633static int rpcb_encode_mapping(struct rpc_rqst *req, __be32 *p,
424 struct rpcbind_args *rpcb) 634 struct rpcbind_args *rpcb)
425{ 635{
@@ -438,7 +648,7 @@ static int rpcb_decode_getport(struct rpc_rqst *req, __be32 *p,
438 unsigned short *portp) 648 unsigned short *portp)
439{ 649{
440 *portp = (unsigned short) ntohl(*p++); 650 *portp = (unsigned short) ntohl(*p++);
441 dprintk("RPC: rpcb_decode_getport result %u\n", 651 dprintk("RPC: rpcb_decode_getport result %u\n",
442 *portp); 652 *portp);
443 return 0; 653 return 0;
444} 654}
@@ -447,8 +657,8 @@ static int rpcb_decode_set(struct rpc_rqst *req, __be32 *p,
447 unsigned int *boolp) 657 unsigned int *boolp)
448{ 658{
449 *boolp = (unsigned int) ntohl(*p++); 659 *boolp = (unsigned int) ntohl(*p++);
450 dprintk("RPC: rpcb_decode_set result %u\n", 660 dprintk("RPC: rpcb_decode_set: call %s\n",
451 *boolp); 661 (*boolp ? "succeeded" : "failed"));
452 return 0; 662 return 0;
453} 663}
454 664
@@ -571,52 +781,60 @@ out_err:
571static struct rpc_procinfo rpcb_procedures2[] = { 781static struct rpc_procinfo rpcb_procedures2[] = {
572 PROC(SET, mapping, set), 782 PROC(SET, mapping, set),
573 PROC(UNSET, mapping, set), 783 PROC(UNSET, mapping, set),
574 PROC(GETADDR, mapping, getport), 784 PROC(GETPORT, mapping, getport),
575}; 785};
576 786
577static struct rpc_procinfo rpcb_procedures3[] = { 787static struct rpc_procinfo rpcb_procedures3[] = {
578 PROC(SET, mapping, set), 788 PROC(SET, getaddr, set),
579 PROC(UNSET, mapping, set), 789 PROC(UNSET, getaddr, set),
580 PROC(GETADDR, getaddr, getaddr), 790 PROC(GETADDR, getaddr, getaddr),
581}; 791};
582 792
583static struct rpc_procinfo rpcb_procedures4[] = { 793static struct rpc_procinfo rpcb_procedures4[] = {
584 PROC(SET, mapping, set), 794 PROC(SET, getaddr, set),
585 PROC(UNSET, mapping, set), 795 PROC(UNSET, getaddr, set),
796 PROC(GETADDR, getaddr, getaddr),
586 PROC(GETVERSADDR, getaddr, getaddr), 797 PROC(GETVERSADDR, getaddr, getaddr),
587}; 798};
588 799
589static struct rpcb_info rpcb_next_version[] = { 800static struct rpcb_info rpcb_next_version[] = {
590#ifdef CONFIG_SUNRPC_BIND34 801 {
591 { 4, &rpcb_procedures4[RPCBPROC_GETVERSADDR] }, 802 .rpc_vers = RPCBVERS_2,
592 { 3, &rpcb_procedures3[RPCBPROC_GETADDR] }, 803 .rpc_proc = &rpcb_procedures2[RPCBPROC_GETPORT],
593#endif 804 },
594 { 2, &rpcb_procedures2[RPCBPROC_GETPORT] }, 805 {
595 { 0, NULL }, 806 .rpc_proc = NULL,
807 },
596}; 808};
597 809
598static struct rpcb_info rpcb_next_version6[] = { 810static struct rpcb_info rpcb_next_version6[] = {
599#ifdef CONFIG_SUNRPC_BIND34 811 {
600 { 4, &rpcb_procedures4[RPCBPROC_GETVERSADDR] }, 812 .rpc_vers = RPCBVERS_4,
601 { 3, &rpcb_procedures3[RPCBPROC_GETADDR] }, 813 .rpc_proc = &rpcb_procedures4[RPCBPROC_GETADDR],
602#endif 814 },
603 { 0, NULL }, 815 {
816 .rpc_vers = RPCBVERS_3,
817 .rpc_proc = &rpcb_procedures3[RPCBPROC_GETADDR],
818 },
819 {
820 .rpc_proc = NULL,
821 },
604}; 822};
605 823
606static struct rpc_version rpcb_version2 = { 824static struct rpc_version rpcb_version2 = {
607 .number = 2, 825 .number = RPCBVERS_2,
608 .nrprocs = RPCB_HIGHPROC_2, 826 .nrprocs = RPCB_HIGHPROC_2,
609 .procs = rpcb_procedures2 827 .procs = rpcb_procedures2
610}; 828};
611 829
612static struct rpc_version rpcb_version3 = { 830static struct rpc_version rpcb_version3 = {
613 .number = 3, 831 .number = RPCBVERS_3,
614 .nrprocs = RPCB_HIGHPROC_3, 832 .nrprocs = RPCB_HIGHPROC_3,
615 .procs = rpcb_procedures3 833 .procs = rpcb_procedures3
616}; 834};
617 835
618static struct rpc_version rpcb_version4 = { 836static struct rpc_version rpcb_version4 = {
619 .number = 4, 837 .number = RPCBVERS_4,
620 .nrprocs = RPCB_HIGHPROC_4, 838 .nrprocs = RPCB_HIGHPROC_4,
621 .procs = rpcb_procedures4 839 .procs = rpcb_procedures4
622}; 840};
diff --git a/net/sunrpc/sched.c b/net/sunrpc/sched.c
index 6eab9bf94baf..6288af05c20f 100644
--- a/net/sunrpc/sched.c
+++ b/net/sunrpc/sched.c
@@ -626,19 +626,15 @@ static void __rpc_execute(struct rpc_task *task)
626 /* 626 /*
627 * Execute any pending callback. 627 * Execute any pending callback.
628 */ 628 */
629 if (RPC_DO_CALLBACK(task)) { 629 if (task->tk_callback) {
630 /* Define a callback save pointer */
631 void (*save_callback)(struct rpc_task *); 630 void (*save_callback)(struct rpc_task *);
632 631
633 /* 632 /*
634 * If a callback exists, save it, reset it, 633 * We set tk_callback to NULL before calling it,
635 * call it. 634 * in case it sets the tk_callback field itself:
636 * The save is needed to stop from resetting
637 * another callback set within the callback handler
638 * - Dave
639 */ 635 */
640 save_callback=task->tk_callback; 636 save_callback = task->tk_callback;
641 task->tk_callback=NULL; 637 task->tk_callback = NULL;
642 save_callback(task); 638 save_callback(task);
643 } 639 }
644 640
diff --git a/net/sunrpc/xprt.c b/net/sunrpc/xprt.c
index e1770f7ba0b3..99a52aabe332 100644
--- a/net/sunrpc/xprt.c
+++ b/net/sunrpc/xprt.c
@@ -690,7 +690,7 @@ static void xprt_connect_status(struct rpc_task *task)
690{ 690{
691 struct rpc_xprt *xprt = task->tk_xprt; 691 struct rpc_xprt *xprt = task->tk_xprt;
692 692
693 if (task->tk_status >= 0) { 693 if (task->tk_status == 0) {
694 xprt->stat.connect_count++; 694 xprt->stat.connect_count++;
695 xprt->stat.connect_time += (long)jiffies - xprt->stat.connect_start; 695 xprt->stat.connect_time += (long)jiffies - xprt->stat.connect_start;
696 dprintk("RPC: %5u xprt_connect_status: connection established\n", 696 dprintk("RPC: %5u xprt_connect_status: connection established\n",
@@ -699,12 +699,6 @@ static void xprt_connect_status(struct rpc_task *task)
699 } 699 }
700 700
701 switch (task->tk_status) { 701 switch (task->tk_status) {
702 case -ECONNREFUSED:
703 case -ECONNRESET:
704 dprintk("RPC: %5u xprt_connect_status: server %s refused "
705 "connection\n", task->tk_pid,
706 task->tk_client->cl_server);
707 break;
708 case -ENOTCONN: 702 case -ENOTCONN:
709 dprintk("RPC: %5u xprt_connect_status: connection broken\n", 703 dprintk("RPC: %5u xprt_connect_status: connection broken\n",
710 task->tk_pid); 704 task->tk_pid);
@@ -878,6 +872,7 @@ void xprt_transmit(struct rpc_task *task)
878 return; 872 return;
879 873
880 req->rq_connect_cookie = xprt->connect_cookie; 874 req->rq_connect_cookie = xprt->connect_cookie;
875 req->rq_xtime = jiffies;
881 status = xprt->ops->send_request(task); 876 status = xprt->ops->send_request(task);
882 if (status == 0) { 877 if (status == 0) {
883 dprintk("RPC: %5u xmit complete\n", task->tk_pid); 878 dprintk("RPC: %5u xmit complete\n", task->tk_pid);
diff --git a/net/sunrpc/xprtsock.c b/net/sunrpc/xprtsock.c
index ddbe981ab516..4486c59c3aca 100644
--- a/net/sunrpc/xprtsock.c
+++ b/net/sunrpc/xprtsock.c
@@ -579,7 +579,6 @@ static int xs_udp_send_request(struct rpc_task *task)
579 req->rq_svec->iov_base, 579 req->rq_svec->iov_base,
580 req->rq_svec->iov_len); 580 req->rq_svec->iov_len);
581 581
582 req->rq_xtime = jiffies;
583 status = xs_sendpages(transport->sock, 582 status = xs_sendpages(transport->sock,
584 xs_addr(xprt), 583 xs_addr(xprt),
585 xprt->addrlen, xdr, 584 xprt->addrlen, xdr,
@@ -671,7 +670,6 @@ static int xs_tcp_send_request(struct rpc_task *task)
671 * to cope with writespace callbacks arriving _after_ we have 670 * to cope with writespace callbacks arriving _after_ we have
672 * called sendmsg(). */ 671 * called sendmsg(). */
673 while (1) { 672 while (1) {
674 req->rq_xtime = jiffies;
675 status = xs_sendpages(transport->sock, 673 status = xs_sendpages(transport->sock,
676 NULL, 0, xdr, req->rq_bytes_sent); 674 NULL, 0, xdr, req->rq_bytes_sent);
677 675