From 286d7d6a0cb38d3d4316a1dfea9b0c0fc5a6455b Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Tue, 3 Jan 2006 09:55:26 +0100 Subject: NFSv4: Remove requirement for machine creds for the "setclientid" operation Use a cred from the nfs4_client->cl_state_owners list. Signed-off-by: Trond Myklebust --- fs/nfs/inode.c | 14 -------------- 1 file changed, 14 deletions(-) (limited to 'fs/nfs/inode.c') diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c index 432f41cd75e6..740d3cd40246 100644 --- a/fs/nfs/inode.c +++ b/fs/nfs/inode.c @@ -1822,23 +1822,9 @@ static int nfs4_fill_super(struct super_block *sb, struct nfs4_mount_data *data, clnt->cl_softrtry = 1; clnt->cl_chatty = 1; clp->cl_rpcclient = clnt; - clp->cl_cred = rpcauth_lookupcred(clnt->cl_auth, 0); - if (IS_ERR(clp->cl_cred)) { - up_write(&clp->cl_sem); - err = PTR_ERR(clp->cl_cred); - clp->cl_cred = NULL; - goto out_fail; - } memcpy(clp->cl_ipaddr, server->ip_addr, sizeof(clp->cl_ipaddr)); nfs_idmap_new(clp); } - if (list_empty(&clp->cl_superblocks)) { - err = nfs4_init_client(clp); - if (err != 0) { - up_write(&clp->cl_sem); - goto out_fail; - } - } list_add_tail(&server->nfs4_siblings, &clp->cl_superblocks); clnt = rpc_clone_client(clp->cl_rpcclient); if (!IS_ERR(clnt)) -- cgit v1.2.2 From 6b59a75460eb9527145d7bd55068e5d32bee8a44 Mon Sep 17 00:00:00 2001 From: Chuck Lever Date: Wed, 30 Nov 2005 18:08:19 -0500 Subject: NFS: Fix error recovery code in fs/nfs/inode.c:__init_nfs() Red Hat found a problem in the error recovery logic in __init_nfs. Signed-off-by: Chuck Lever Signed-off-by: Trond Myklebust --- fs/nfs/inode.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'fs/nfs/inode.c') diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c index 740d3cd40246..da58207b366b 100644 --- a/fs/nfs/inode.c +++ b/fs/nfs/inode.c @@ -2152,11 +2152,11 @@ out: #ifdef CONFIG_PROC_FS rpc_proc_unregister("nfs"); #endif - nfs_destroy_writepagecache(); #ifdef CONFIG_NFS_DIRECTIO -out0: nfs_destroy_directcache(); +out0: #endif + nfs_destroy_writepagecache(); out1: nfs_destroy_readpagecache(); out2: -- cgit v1.2.2 From dc20f803904dbf30f834dcc43c14701dfce32491 Mon Sep 17 00:00:00 2001 From: Chuck Lever Date: Wed, 30 Nov 2005 18:08:57 -0500 Subject: NFS: get rid of useless kernel log message nfs_statfs() generates a log message when GETATTR returns an error. This is usually a useless message. Make it a dprintk. Test plan: None Signed-off-by: Chuck Lever Signed-off-by: Trond Myklebust --- fs/nfs/inode.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'fs/nfs/inode.c') diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c index da58207b366b..cc753928e411 100644 --- a/fs/nfs/inode.c +++ b/fs/nfs/inode.c @@ -575,11 +575,10 @@ nfs_statfs(struct super_block *sb, struct kstatfs *buf) buf->f_namelen = server->namelen; out: unlock_kernel(); - return 0; out_err: - printk(KERN_WARNING "nfs_statfs: statfs error = %d\n", -error); + dprintk("%s: statfs error = %d\n", __FUNCTION__, -error); buf->f_bsize = buf->f_blocks = buf->f_bfree = buf->f_bavail = -1; goto out; -- cgit v1.2.2 From 325cfed9ae901320e9234b18c21434b783dbe342 Mon Sep 17 00:00:00 2001 From: Chuck Lever Date: Wed, 30 Nov 2005 18:08:55 -0500 Subject: NFS: make "inode number mismatch" message more useful To help NFS users and server developers, make the "inode number mismatch" message display more useful information. Test-plan: None. Signed-off-by: Chuck Lever Signed-off-by: Trond Myklebust --- fs/nfs/inode.c | 17 +++++++++-------- 1 file changed, 9 insertions(+), 8 deletions(-) (limited to 'fs/nfs/inode.c') diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c index cc753928e411..4e6558df54b8 100644 --- a/fs/nfs/inode.c +++ b/fs/nfs/inode.c @@ -1409,14 +1409,8 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr) if ((fattr->valid & NFS_ATTR_FATTR) == 0) return 0; - if (nfsi->fileid != fattr->fileid) { - printk(KERN_ERR "%s: inode number mismatch\n" - "expected (%s/0x%Lx), got (%s/0x%Lx)\n", - __FUNCTION__, - inode->i_sb->s_id, (long long)nfsi->fileid, - inode->i_sb->s_id, (long long)fattr->fileid); - goto out_err; - } + if (nfsi->fileid != fattr->fileid) + goto out_fileid; /* * Make sure the inode's type hasn't changed. @@ -1538,6 +1532,13 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr) */ nfs_invalidate_inode(inode); return -ESTALE; + + out_fileid: + printk(KERN_ERR "NFS: server %s error: fileid changed\n" + "fsid %s: expected fileid 0x%Lx, got 0x%Lx\n", + NFS_SERVER(inode)->hostname, inode->i_sb->s_id, + (long long)nfsi->fileid, (long long)fattr->fileid); + goto out_err; } /* -- cgit v1.2.2 From 40859d7ee64ed6bfad8a4e93f9bb5c1074afadff Mon Sep 17 00:00:00 2001 From: Chuck Lever Date: Wed, 30 Nov 2005 18:09:02 -0500 Subject: NFS: support large reads and writes on the wire Most NFS server implementations allow up to 64KB reads and writes on the wire. The Solaris NFS server allows up to a megabyte, for instance. Now the Linux NFS client supports transfer sizes up to 1MB, too. This will help reduce protocol and context switch overhead on read/write intensive NFS workloads, and support larger atomic read and write operations on servers that support them. Test-plan: Connectathon and iozone on mount point with wsize=rsize>32768 over TCP. Tests with NFS over UDP to verify the maximum RPC payload size cap. Signed-off-by: Chuck Lever Signed-off-by: Trond Myklebust --- fs/nfs/inode.c | 25 ++++++++++--------------- 1 file changed, 10 insertions(+), 15 deletions(-) (limited to 'fs/nfs/inode.c') diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c index 4e6558df54b8..acde2c5725bf 100644 --- a/fs/nfs/inode.c +++ b/fs/nfs/inode.c @@ -221,10 +221,10 @@ nfs_calc_block_size(u64 tsize) static inline unsigned long nfs_block_size(unsigned long bsize, unsigned char *nrbitsp) { - if (bsize < 1024) - bsize = NFS_DEF_FILE_IO_BUFFER_SIZE; - else if (bsize >= NFS_MAX_FILE_IO_BUFFER_SIZE) - bsize = NFS_MAX_FILE_IO_BUFFER_SIZE; + if (bsize < NFS_MIN_FILE_IO_SIZE) + bsize = NFS_DEF_FILE_IO_SIZE; + else if (bsize >= NFS_MAX_FILE_IO_SIZE) + bsize = NFS_MAX_FILE_IO_SIZE; return nfs_block_bits(bsize, nrbitsp); } @@ -307,20 +307,15 @@ nfs_sb_init(struct super_block *sb, rpc_authflavor_t authflavor) max_rpc_payload = nfs_block_size(rpc_max_payload(server->client), NULL); if (server->rsize > max_rpc_payload) server->rsize = max_rpc_payload; - if (server->wsize > max_rpc_payload) - server->wsize = max_rpc_payload; - + if (server->rsize > NFS_MAX_FILE_IO_SIZE) + server->rsize = NFS_MAX_FILE_IO_SIZE; server->rpages = (server->rsize + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT; - if (server->rpages > NFS_READ_MAXIOV) { - server->rpages = NFS_READ_MAXIOV; - server->rsize = server->rpages << PAGE_CACHE_SHIFT; - } + if (server->wsize > max_rpc_payload) + server->wsize = max_rpc_payload; + if (server->wsize > NFS_MAX_FILE_IO_SIZE) + server->wsize = NFS_MAX_FILE_IO_SIZE; server->wpages = (server->wsize + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT; - if (server->wpages > NFS_WRITE_MAXIOV) { - server->wpages = NFS_WRITE_MAXIOV; - server->wsize = server->wpages << PAGE_CACHE_SHIFT; - } if (sb->s_blocksize == 0) sb->s_blocksize = nfs_block_bits(server->wsize, -- cgit v1.2.2 From 70b9ecbdb9c5fdc731f8780bffd45d9519020c4a Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Tue, 3 Jan 2006 09:55:34 +0100 Subject: NFS: Make stat() return updated mtimes after a write() The SuS states that a call to write() will cause mtime to be updated on the file. In order to satisfy that requirement, we need to flush out any cached writes in nfs_getattr(). Speed things up slightly by not committing the writes. Signed-off-by: Trond Myklebust --- fs/nfs/inode.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'fs/nfs/inode.c') diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c index acde2c5725bf..2c7f8aac1dec 100644 --- a/fs/nfs/inode.c +++ b/fs/nfs/inode.c @@ -952,6 +952,8 @@ int nfs_getattr(struct vfsmount *mnt, struct dentry *dentry, struct kstat *stat) int need_atime = NFS_I(inode)->cache_validity & NFS_INO_INVALID_ATIME; int err; + /* Flush out writes to the server in order to update c/mtime */ + nfs_sync_inode(inode, 0, 0, FLUSH_WAIT|FLUSH_NOCOMMIT); if (__IS_FLG(inode, MS_NOATIME)) need_atime = 0; else if (__IS_FLG(inode, MS_NODIRATIME) && S_ISDIR(inode->i_mode)) -- cgit v1.2.2 From a895b4a198dd06f8353328867e4f6cfd28b63081 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Tue, 3 Jan 2006 09:55:40 +0100 Subject: NFS: Clean up weak cache consistency code ...and ensure that nfs_update_inode() respects wcc Signed-off-by: Trond Myklebust --- fs/nfs/inode.c | 60 ++++++++++++++++++++++++++++++++++++++-------------------- 1 file changed, 40 insertions(+), 20 deletions(-) (limited to 'fs/nfs/inode.c') diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c index 2c7f8aac1dec..7270b1d73d30 100644 --- a/fs/nfs/inode.c +++ b/fs/nfs/inode.c @@ -1248,6 +1248,33 @@ void nfs_end_data_update(struct inode *inode) atomic_dec(&nfsi->data_updates); } +static void nfs_wcc_update_inode(struct inode *inode, struct nfs_fattr *fattr) +{ + struct nfs_inode *nfsi = NFS_I(inode); + + if ((fattr->valid & NFS_ATTR_PRE_CHANGE) != 0 + && nfsi->change_attr == fattr->pre_change_attr) { + nfsi->change_attr = fattr->change_attr; + nfsi->cache_change_attribute = jiffies; + } + + /* If we have atomic WCC data, we may update some attributes */ + if ((fattr->valid & NFS_ATTR_WCC) != 0) { + if (timespec_equal(&inode->i_ctime, &fattr->pre_ctime)) { + memcpy(&inode->i_ctime, &fattr->ctime, sizeof(inode->i_ctime)); + nfsi->cache_change_attribute = jiffies; + } + if (timespec_equal(&inode->i_mtime, &fattr->pre_mtime)) { + memcpy(&inode->i_mtime, &fattr->mtime, sizeof(inode->i_mtime)); + nfsi->cache_change_attribute = jiffies; + } + if (inode->i_size == fattr->pre_size && nfsi->npages == 0) { + inode->i_size = fattr->size; + nfsi->cache_change_attribute = jiffies; + } + } +} + /** * nfs_check_inode_attributes - verify consistency of the inode attribute cache * @inode - pointer to inode @@ -1264,22 +1291,20 @@ static int nfs_check_inode_attributes(struct inode *inode, struct nfs_fattr *fat int data_unstable; + if ((fattr->valid & NFS_ATTR_FATTR) == 0) + return 0; + /* Are we in the process of updating data on the server? */ data_unstable = nfs_caches_unstable(inode); - if (fattr->valid & NFS_ATTR_FATTR_V4) { - if ((fattr->valid & NFS_ATTR_PRE_CHANGE) != 0 - && nfsi->change_attr == fattr->pre_change_attr) - nfsi->change_attr = fattr->change_attr; - if (nfsi->change_attr != fattr->change_attr) { - nfsi->cache_validity |= NFS_INO_INVALID_ATTR; - if (!data_unstable) - nfsi->cache_validity |= NFS_INO_REVAL_PAGECACHE; - } - } + /* Do atomic weak cache consistency updates */ + nfs_wcc_update_inode(inode, fattr); - if ((fattr->valid & NFS_ATTR_FATTR) == 0) { - return 0; + if ((fattr->valid & NFS_ATTR_FATTR_V4) != 0 && + nfsi->change_attr != fattr->change_attr) { + nfsi->cache_validity |= NFS_INO_INVALID_ATTR; + if (!data_unstable) + nfsi->cache_validity |= NFS_INO_REVAL_PAGECACHE; } /* Has the inode gone and changed behind our back? */ @@ -1291,14 +1316,6 @@ static int nfs_check_inode_attributes(struct inode *inode, struct nfs_fattr *fat cur_size = i_size_read(inode); new_isize = nfs_size_to_loff_t(fattr->size); - /* If we have atomic WCC data, we may update some attributes */ - if ((fattr->valid & NFS_ATTR_WCC) != 0) { - if (timespec_equal(&inode->i_ctime, &fattr->pre_ctime)) - memcpy(&inode->i_ctime, &fattr->ctime, sizeof(inode->i_ctime)); - if (timespec_equal(&inode->i_mtime, &fattr->pre_mtime)) - memcpy(&inode->i_mtime, &fattr->mtime, sizeof(inode->i_mtime)); - } - /* Verify a few of the more important attributes */ if (!timespec_equal(&inode->i_mtime, &fattr->mtime)) { nfsi->cache_validity |= NFS_INO_INVALID_ATTR; @@ -1426,6 +1443,9 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr) if (data_stable) nfsi->cache_validity &= ~(NFS_INO_INVALID_ATTR|NFS_INO_INVALID_ATIME); + /* Do atomic weak cache consistency updates */ + nfs_wcc_update_inode(inode, fattr); + /* Check if our cached file size is stale */ new_isize = nfs_size_to_loff_t(fattr->size); cur_isize = i_size_read(inode); -- cgit v1.2.2 From a72b44222d222749d54b3e370d825094352e389f Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Tue, 3 Jan 2006 09:55:41 +0100 Subject: NFSv4: Allow user to set the port used by the NFSv4 callback channel Signed-off-by: Trond Myklebust --- fs/nfs/inode.c | 37 +++++++++++++++++++++++++++++++++++-- 1 file changed, 35 insertions(+), 2 deletions(-) (limited to 'fs/nfs/inode.c') diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c index 7270b1d73d30..648cb1aef3b1 100644 --- a/fs/nfs/inode.c +++ b/fs/nfs/inode.c @@ -40,6 +40,7 @@ #include #include "nfs4_fs.h" +#include "callback.h" #include "delegation.h" #define NFSDBG_FACILITY NFSDBG_VFS @@ -2036,6 +2037,21 @@ static struct file_system_type nfs4_fs_type = { .fs_flags = FS_ODD_RENAME|FS_REVAL_DOT|FS_BINARY_MOUNTDATA, }; +static const int nfs_set_port_min = 0; +static const int nfs_set_port_max = 65535; +static int param_set_port(const char *val, struct kernel_param *kp) +{ + char *endp; + int num = simple_strtol(val, &endp, 0); + if (endp == val || *endp || num < nfs_set_port_min || num > nfs_set_port_max) + return -EINVAL; + *((int *)kp->arg) = num; + return 0; +} + +module_param_call(callback_tcpport, param_set_port, param_get_int, + &nfs_callback_set_tcpport, 0644); + #define nfs4_init_once(nfsi) \ do { \ INIT_LIST_HEAD(&(nfsi)->open_states); \ @@ -2043,8 +2059,25 @@ static struct file_system_type nfs4_fs_type = { nfsi->delegation_state = 0; \ init_rwsem(&nfsi->rwsem); \ } while(0) -#define register_nfs4fs() register_filesystem(&nfs4_fs_type) -#define unregister_nfs4fs() unregister_filesystem(&nfs4_fs_type) + +static inline int register_nfs4fs(void) +{ + int ret; + + ret = nfs_register_sysctl(); + if (ret != 0) + return ret; + ret = register_filesystem(&nfs4_fs_type); + if (ret != 0) + nfs_unregister_sysctl(); + return ret; +} + +static inline void unregister_nfs4fs(void) +{ + unregister_filesystem(&nfs4_fs_type); + nfs_unregister_sysctl(); +} #else #define nfs4_init_once(nfsi) \ do { } while (0) -- cgit v1.2.2 From f518e35aec984036903c1003e867f833747a9d79 Mon Sep 17 00:00:00 2001 From: Chuck Lever Date: Tue, 3 Jan 2006 09:55:52 +0100 Subject: SUNRPC: get rid of cl_chatty Clean up: Every ULP that uses the in-kernel RPC client, except the NLM client, sets cl_chatty. There's no reason why NLM shouldn't set it, so just get rid of cl_chatty and always be verbose. Test-plan: Compile with CONFIG_NFS enabled. Signed-off-by: Chuck Lever Signed-off-by: Trond Myklebust --- fs/nfs/inode.c | 2 -- 1 file changed, 2 deletions(-) (limited to 'fs/nfs/inode.c') diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c index 648cb1aef3b1..4625479a6b62 100644 --- a/fs/nfs/inode.c +++ b/fs/nfs/inode.c @@ -413,7 +413,6 @@ nfs_create_client(struct nfs_server *server, const struct nfs_mount_data *data) clnt->cl_intr = 1; clnt->cl_softrtry = 1; - clnt->cl_chatty = 1; return clnt; @@ -1838,7 +1837,6 @@ static int nfs4_fill_super(struct super_block *sb, struct nfs4_mount_data *data, } clnt->cl_intr = 1; clnt->cl_softrtry = 1; - clnt->cl_chatty = 1; clp->cl_rpcclient = clnt; memcpy(clp->cl_ipaddr, server->ip_addr, sizeof(clp->cl_ipaddr)); nfs_idmap_new(clp); -- cgit v1.2.2 From 58df095b732529ade8f4051b41d7c29731afecd6 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Tue, 3 Jan 2006 09:55:57 +0100 Subject: NFSv4: Allow entries in the idmap cache to expire If someone changes the uid/gid mapping in userland, then we do eventually want those changes to be propagated to the kernel. Currently the kernel assumes that it may cache entries forever. Add an expiration time + garbage collector for idmap entries. Signed-off-by: Trond Myklebust --- fs/nfs/inode.c | 14 ++++++++++++++ 1 file changed, 14 insertions(+) (limited to 'fs/nfs/inode.c') diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c index 4625479a6b62..e7bd0d92600f 100644 --- a/fs/nfs/inode.c +++ b/fs/nfs/inode.c @@ -2050,6 +2050,20 @@ static int param_set_port(const char *val, struct kernel_param *kp) module_param_call(callback_tcpport, param_set_port, param_get_int, &nfs_callback_set_tcpport, 0644); +static int param_set_idmap_timeout(const char *val, struct kernel_param *kp) +{ + char *endp; + int num = simple_strtol(val, &endp, 0); + int jif = num * HZ; + if (endp == val || *endp || num < 0 || jif < num) + return -EINVAL; + *((int *)kp->arg) = jif; + return 0; +} + +module_param_call(idmap_cache_timeout, param_set_idmap_timeout, param_get_int, + &nfs_idmap_cache_timeout, 0644); + #define nfs4_init_once(nfsi) \ do { \ INIT_LIST_HEAD(&(nfsi)->open_states); \ -- cgit v1.2.2