fs/locks.c | 51 - fs/nfs/Makefile | 3 fs/nfs/callback.c | 325 +++++++++++ fs/nfs/callback.h | 70 ++ fs/nfs/callback_proc.c | 85 ++ fs/nfs/callback_xdr.c | 481 ++++++++++++++++ fs/nfs/delegation.c | 320 +++++++++++ fs/nfs/delegation.h | 56 + fs/nfs/dir.c | 121 ++-- fs/nfs/direct.c | 47 - fs/nfs/file.c | 149 +++-- fs/nfs/inode.c | 202 +++++-- fs/nfs/nfs3proc.c | 74 -- fs/nfs/nfs4proc.c | 1235 +++++++++++++++++++++++++++++++------------ fs/nfs/nfs4state.c | 392 +++++++------ fs/nfs/nfs4xdr.c | 432 +++++++++++---- fs/nfs/pagelist.c | 47 - fs/nfs/proc.c | 48 - fs/nfs/read.c | 62 +- fs/nfs/unlink.c | 3 fs/nfs/write.c | 102 +-- fs/open.c | 2 include/linux/fs.h | 2 include/linux/nfs4.h | 5 include/linux/nfs_fs.h | 135 ++-- include/linux/nfs_page.h | 29 - include/linux/nfs_xdr.h | 61 +- include/linux/sunrpc/sched.h | 55 + include/linux/sunrpc/svc.h | 10 net/sunrpc/clnt.c | 4 net/sunrpc/sched.c | 510 +++++------------ net/sunrpc/sunrpc_syms.c | 1 net/sunrpc/svc.c | 9 net/sunrpc/xprt.c | 2 34 files changed, 3631 insertions(+), 1499 deletions(-) diff -u --recursive --new-file --show-c-function linux-2.6.7/fs/locks.c linux-2.6.7-23-rpc_queue_lock/fs/locks.c --- linux-2.6.7/fs/locks.c 2004-06-16 14:58:27.000000000 -0400 +++ linux-2.6.7-23-rpc_queue_lock/fs/locks.c 2004-06-22 16:48:19.000000000 -0400 @@ -317,7 +317,7 @@ static int flock_to_posix_lock(struct fi if (l->l_len == 0) fl->fl_end = OFFSET_MAX; - fl->fl_owner = current->files; + fl->fl_owner = 0; fl->fl_pid = current->tgid; fl->fl_file = filp; fl->fl_flags = FL_POSIX; @@ -357,7 +357,7 @@ static int flock64_to_posix_lock(struct if (l->l_len == 0) fl->fl_end = OFFSET_MAX; - fl->fl_owner = current->files; + fl->fl_owner = 0; fl->fl_pid = current->tgid; fl->fl_file = filp; fl->fl_flags = FL_POSIX; @@ -920,7 +920,7 @@ int posix_lock_file(struct file *filp, s */ int locks_mandatory_locked(struct inode *inode) { - fl_owner_t owner = current->files; + unsigned int pid = current->tgid; struct file_lock *fl; /* @@ -930,7 +930,9 @@ int locks_mandatory_locked(struct inode for (fl = inode->i_flock; fl != NULL; fl = fl->fl_next) { if (!IS_POSIX(fl)) continue; - if (fl->fl_owner != owner) + if (fl->fl_owner != 0) + break; + if (fl->fl_pid != pid) break; } unlock_kernel(); @@ -958,7 +960,7 @@ int locks_mandatory_area(int read_write, int error; locks_init_lock(&fl); - fl.fl_owner = current->files; + fl.fl_owner = 0; fl.fl_pid = current->tgid; fl.fl_file = filp; fl.fl_flags = FL_POSIX | FL_ACCESS; @@ -1684,7 +1686,7 @@ void locks_remove_posix(struct file *fil lock_kernel(); while (*before != NULL) { struct file_lock *fl = *before; - if (IS_POSIX(fl) && (fl->fl_owner == owner)) { + if (IS_POSIX(fl) && posix_same_owner(fl, &lock)) { locks_delete_lock(before); continue; } @@ -1982,18 +1984,6 @@ int lock_may_write(struct inode *inode, EXPORT_SYMBOL(lock_may_write); -static inline void __steal_locks(struct file *file, fl_owner_t from) -{ - struct inode *inode = file->f_dentry->d_inode; - struct file_lock *fl = inode->i_flock; - - while (fl) { - if (fl->fl_file == file && fl->fl_owner == from) - fl->fl_owner = current->files; - fl = fl->fl_next; - } -} - /* When getting ready for executing a binary, we make sure that current * has a files_struct on its own. Before dropping the old files_struct, * we take over ownership of all locks for all file descriptors we own. @@ -2002,31 +1992,6 @@ static inline void __steal_locks(struct */ void steal_locks(fl_owner_t from) { - struct files_struct *files = current->files; - int i, j; - - if (from == files) - return; - - lock_kernel(); - j = 0; - for (;;) { - unsigned long set; - i = j * __NFDBITS; - if (i >= files->max_fdset || i >= files->max_fds) - break; - set = files->open_fds->fds_bits[j++]; - while (set) { - if (set & 1) { - struct file *file = files->fd[i]; - if (file) - __steal_locks(file, from); - } - i++; - set >>= 1; - } - } - unlock_kernel(); } EXPORT_SYMBOL(steal_locks); diff -u --recursive --new-file --show-c-function linux-2.6.7/fs/nfs/callback.c linux-2.6.7-23-rpc_queue_lock/fs/nfs/callback.c --- linux-2.6.7/fs/nfs/callback.c 1969-12-31 19:00:00.000000000 -0500 +++ linux-2.6.7-23-rpc_queue_lock/fs/nfs/callback.c 2004-06-22 16:49:06.000000000 -0400 @@ -0,0 +1,325 @@ +/* + * linux/fs/nfs/callback.c + * + * Copyright (C) 2004 Trond Myklebust + * + * NFSv4 callback handling + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include "callback.h" + +#define NFSDBG_FACILITY NFSDBG_CALLBACK + +struct nfs_callback_data { + unsigned int users; + struct svc_serv *serv; + pid_t pid; + struct completion started; + struct completion stopped; +}; + +static struct nfs_callback_data nfs_callback_info; +static DECLARE_MUTEX(nfs_callback_sema); +static struct svc_program nfs4_callback_program; + +unsigned short nfs_callback_tcpport; + +/* + * This is the callback kernel thread. + */ +static void nfs_callback_svc(struct svc_rqst *rqstp) +{ + struct svc_serv *serv = rqstp->rq_server; + int err; + + __module_get(THIS_MODULE); + lock_kernel(); + + nfs_callback_info.pid = current->pid; + daemonize("nfsv4-svc"); + /* Process request with signals blocked, but allow SIGKILL. */ + allow_signal(SIGKILL); + + complete(&nfs_callback_info.started); + + while (nfs_callback_info.users != 0 || !signalled()) { + /* + * Listen for a request on the socket + */ + err = svc_recv(serv, rqstp, MAX_SCHEDULE_TIMEOUT); + if (err == -EAGAIN || err == -EINTR) + continue; + if (err < 0) { + printk(KERN_WARNING + "%s: terminating on error %d\n", + __FUNCTION__, -err); + break; + } + dprintk("%s: request from %u.%u.%u.%u\n", __FUNCTION__, + NIPQUAD(rqstp->rq_addr.sin_addr.s_addr)); + svc_process(serv, rqstp); + } + + nfs_callback_info.pid = 0; + complete(&nfs_callback_info.stopped); + unlock_kernel(); + module_put_and_exit(0); +} + +/* + * Bring up the server process if it is not already up. + */ +int nfs_callback_up(void) +{ + struct svc_serv *serv; + struct svc_sock *svsk; + int ret = 0; + + lock_kernel(); + down(&nfs_callback_sema); + if (nfs_callback_info.users++ || nfs_callback_info.pid != 0) + goto out; + init_completion(&nfs_callback_info.started); + init_completion(&nfs_callback_info.stopped); + serv = svc_create(&nfs4_callback_program, NFS4_CALLBACK_BUFSIZE); + ret = -ENOMEM; + if (!serv) + goto out_err; + /* FIXME: We don't want to register this socket with the portmapper */ + ret = svc_makesock(serv, IPPROTO_TCP, 0); + if (ret < 0) + goto out_destroy; + if (!list_empty(&serv->sv_permsocks)) { + svsk = list_entry(serv->sv_permsocks.next, + struct svc_sock, sk_list); + nfs_callback_tcpport = ntohs(inet_sk(svsk->sk_sk)->sport); + dprintk ("Callback port = 0x%x\n", nfs_callback_tcpport); + } else + BUG(); + ret = svc_create_thread(nfs_callback_svc, serv); + if (ret < 0) + goto out_destroy; + nfs_callback_info.serv = serv; + wait_for_completion(&nfs_callback_info.started); +out: + up(&nfs_callback_sema); + unlock_kernel(); + return ret; +out_destroy: + svc_destroy(serv); +out_err: + nfs_callback_info.users--; + goto out; +} + +/* + * Kill the server process if it is not already up. + */ +int nfs_callback_down(void) +{ + int ret = 0; + + lock_kernel(); + down(&nfs_callback_sema); + if (--nfs_callback_info.users || nfs_callback_info.pid == 0) + goto out; + kill_proc(nfs_callback_info.pid, SIGKILL, 1); + wait_for_completion(&nfs_callback_info.stopped); +out: + up(&nfs_callback_sema); + unlock_kernel(); + return ret; +} + +/* + * AUTH_NULL authentication + */ +static int nfs_callback_null_accept(struct svc_rqst *rqstp, u32 *authp) +{ + struct iovec *argv = &rqstp->rq_arg.head[0]; + struct iovec *resv = &rqstp->rq_res.head[0]; + + if (argv->iov_len < 3*4) + return SVC_GARBAGE; + + if (svc_getu32(argv) != 0) { + dprintk("svc: bad null cred\n"); + *authp = rpc_autherr_badcred; + return SVC_DENIED; + } + if (svc_getu32(argv) != RPC_AUTH_NULL || svc_getu32(argv) != 0) { + dprintk("svc: bad null verf\n"); + *authp = rpc_autherr_badverf; + return SVC_DENIED; + } + + /* Signal that mapping to nobody uid/gid is required */ + rqstp->rq_cred.cr_uid = (uid_t) -1; + rqstp->rq_cred.cr_gid = (gid_t) -1; + rqstp->rq_cred.cr_group_info = groups_alloc(0); + if (rqstp->rq_cred.cr_group_info == NULL) + return SVC_DROP; /* kmalloc failure - client must retry */ + + /* Put NULL verifier */ + svc_putu32(resv, RPC_AUTH_NULL); + svc_putu32(resv, 0); + dprintk("%s: success, returning %d!\n", __FUNCTION__, SVC_OK); + return SVC_OK; +} + +static int nfs_callback_null_release(struct svc_rqst *rqstp) +{ + if (rqstp->rq_cred.cr_group_info) + put_group_info(rqstp->rq_cred.cr_group_info); + rqstp->rq_cred.cr_group_info = NULL; + return 0; /* don't drop */ +} + +static struct auth_ops nfs_callback_auth_null = { + .name = "null", + .flavour = RPC_AUTH_NULL, + .accept = nfs_callback_null_accept, + .release = nfs_callback_null_release, +}; + +/* + * AUTH_SYS authentication + */ +static int nfs_callback_unix_accept(struct svc_rqst *rqstp, u32 *authp) +{ + struct iovec *argv = &rqstp->rq_arg.head[0]; + struct iovec *resv = &rqstp->rq_res.head[0]; + struct svc_cred *cred = &rqstp->rq_cred; + u32 slen, i; + int len = argv->iov_len; + + dprintk("%s: start\n", __FUNCTION__); + cred->cr_group_info = NULL; + rqstp->rq_client = NULL; + if ((len -= 3*4) < 0) + return SVC_GARBAGE; + + /* Get length, time stamp and machine name */ + svc_getu32(argv); + svc_getu32(argv); + slen = XDR_QUADLEN(ntohl(svc_getu32(argv))); + if (slen > 64 || (len -= (slen + 3)*4) < 0) + goto badcred; + argv->iov_base = (void*)((u32*)argv->iov_base + slen); + argv->iov_len -= slen*4; + + cred->cr_uid = ntohl(svc_getu32(argv)); + cred->cr_gid = ntohl(svc_getu32(argv)); + slen = ntohl(svc_getu32(argv)); + if (slen > 16 || (len -= (slen + 2)*4) < 0) + goto badcred; + cred->cr_group_info = groups_alloc(slen); + if (cred->cr_group_info == NULL) + return SVC_DROP; + for (i = 0; i < slen; i++) + GROUP_AT(cred->cr_group_info, i) = ntohl(svc_getu32(argv)); + + if (svc_getu32(argv) != RPC_AUTH_NULL || svc_getu32(argv) != 0) { + *authp = rpc_autherr_badverf; + return SVC_DENIED; + } + /* Put NULL verifier */ + svc_putu32(resv, RPC_AUTH_NULL); + svc_putu32(resv, 0); + dprintk("%s: success, returning %d!\n", __FUNCTION__, SVC_OK); + return SVC_OK; +badcred: + *authp = rpc_autherr_badcred; + return SVC_DENIED; +} + +static int nfs_callback_unix_release(struct svc_rqst *rqstp) +{ + if (rqstp->rq_cred.cr_group_info) + put_group_info(rqstp->rq_cred.cr_group_info); + rqstp->rq_cred.cr_group_info = NULL; + return 0; +} + +static struct auth_ops nfs_callback_auth_unix = { + .name = "unix", + .flavour = RPC_AUTH_UNIX, + .accept = nfs_callback_unix_accept, + .release = nfs_callback_unix_release, +}; + +/* + * Hook the authentication protocol + */ +static int nfs_callback_auth(struct svc_rqst *rqstp, u32 *authp) +{ + struct in_addr *addr = &rqstp->rq_addr.sin_addr; + struct nfs4_client *clp; + struct iovec *argv = &rqstp->rq_arg.head[0]; + int flavour; + int retval; + + /* Don't talk to strangers */ + clp = nfs4_find_client(addr); + if (clp == NULL) + return SVC_DROP; + dprintk("%s: %u.%u.%u.%u NFSv4 callback!\n", __FUNCTION__, NIPQUAD(addr)); + nfs4_put_client(clp); + flavour = ntohl(svc_getu32(argv)); + switch(flavour) { + case RPC_AUTH_NULL: + if (rqstp->rq_proc != CB_NULL) { + *authp = rpc_autherr_tooweak; + retval = SVC_DENIED; + break; + } + rqstp->rq_authop = &nfs_callback_auth_null; + retval = nfs_callback_null_accept(rqstp, authp); + break; + case RPC_AUTH_UNIX: + /* Eat the authentication flavour */ + rqstp->rq_authop = &nfs_callback_auth_unix; + retval = nfs_callback_unix_accept(rqstp, authp); + break; + default: + /* FIXME: need to add RPCSEC_GSS upcalls */ +#if 0 + svc_ungetu32(argv); + retval = svc_authenticate(rqstp, authp); +#else + *authp = rpc_autherr_rejectedcred; + retval = SVC_DENIED; +#endif + } + dprintk("%s: flavour %d returning error %d\n", __FUNCTION__, flavour, retval); + return retval; +} + +/* + * Define NFS4 callback program + */ +extern struct svc_version nfs4_callback_version1; + +static struct svc_version *nfs4_callback_version[] = { + [1] = &nfs4_callback_version1, +}; + +static struct svc_stat nfs4_callback_stats; + +static struct svc_program nfs4_callback_program = { + .pg_prog = NFS4_CALLBACK, /* RPC service number */ + .pg_nvers = ARRAY_SIZE(nfs4_callback_version), /* Number of entries */ + .pg_vers = nfs4_callback_version, /* version table */ + .pg_name = "NFSv4 callback", /* service name */ + .pg_class = "nfs", /* authentication class */ + .pg_stats = &nfs4_callback_stats, + .pg_authenticate = nfs_callback_auth, +}; diff -u --recursive --new-file --show-c-function linux-2.6.7/fs/nfs/callback.h linux-2.6.7-23-rpc_queue_lock/fs/nfs/callback.h --- linux-2.6.7/fs/nfs/callback.h 1969-12-31 19:00:00.000000000 -0500 +++ linux-2.6.7-23-rpc_queue_lock/fs/nfs/callback.h 2004-06-22 16:49:06.000000000 -0400 @@ -0,0 +1,70 @@ +/* + * linux/fs/nfs/callback.h + * + * Copyright (C) 2004 Trond Myklebust + * + * NFSv4 callback definitions + */ +#ifndef __LINUX_FS_NFS_CALLBACK_H +#define __LINUX_FS_NFS_CALLBACK_H + +#define NFS4_CALLBACK 0x40000000 +#define NFS4_CALLBACK_XDRSIZE 2048 +#define NFS4_CALLBACK_BUFSIZE (1024 + NFS4_CALLBACK_XDRSIZE) + +enum nfs4_callback_procnum { + CB_NULL = 0, + CB_COMPOUND = 1, +}; + +enum nfs4_callback_opnum { + OP_CB_GETATTR = 3, + OP_CB_RECALL = 4, + OP_CB_ILLEGAL = 10044, +}; + +struct cb_compound_hdr_arg { + int taglen; + const char *tag; + unsigned int callback_ident; + unsigned nops; +}; + +struct cb_compound_hdr_res { + uint32_t *status; + int taglen; + const char *tag; + uint32_t *nops; +}; + +struct cb_getattrargs { + struct sockaddr_in *addr; + struct nfs_fh fh; + uint32_t bitmap[2]; +}; + +struct cb_getattrres { + uint32_t status; + uint32_t bitmap[2]; + uint64_t size; + uint64_t change_attr; + struct timespec ctime; + struct timespec mtime; +}; + +struct cb_recallargs { + struct sockaddr_in *addr; + struct nfs_fh fh; + nfs4_stateid stateid; + uint32_t truncate; +}; + +extern unsigned nfs4_callback_getattr(struct cb_getattrargs *args, struct cb_getattrres *res); +extern unsigned nfs4_callback_recall(struct cb_recallargs *args, void *dummy); + +extern int nfs_callback_up(void); +extern int nfs_callback_down(void); + +extern unsigned short nfs_callback_tcpport; + +#endif /* __LINUX_FS_NFS_CALLBACK_H */ diff -u --recursive --new-file --show-c-function linux-2.6.7/fs/nfs/callback_proc.c linux-2.6.7-23-rpc_queue_lock/fs/nfs/callback_proc.c --- linux-2.6.7/fs/nfs/callback_proc.c 1969-12-31 19:00:00.000000000 -0500 +++ linux-2.6.7-23-rpc_queue_lock/fs/nfs/callback_proc.c 2004-06-22 16:49:06.000000000 -0400 @@ -0,0 +1,85 @@ +/* + * linux/fs/nfs/callback_proc.c + * + * Copyright (C) 2004 Trond Myklebust + * + * NFSv4 callback procedures + */ +#include +#include +#include +#include "callback.h" +#include "delegation.h" + +#define NFSDBG_FACILITY NFSDBG_CALLBACK + +unsigned nfs4_callback_getattr(struct cb_getattrargs *args, struct cb_getattrres *res) +{ + struct nfs4_client *clp; + struct nfs_delegation *delegation; + struct nfs_inode *nfsi; + struct inode *inode; + + res->bitmap[0] = res->bitmap[1] = 0; + res->status = htonl(NFS4ERR_BADHANDLE); + clp = nfs4_find_client(&args->addr->sin_addr); + if (clp == NULL) + goto out; + inode = nfs_delegation_find_inode(clp, &args->fh); + if (inode == NULL) + goto out_putclient; + nfsi = NFS_I(inode); + down_read(&nfsi->rwsem); + delegation = nfsi->delegation; + if (delegation == NULL || (delegation->type & FMODE_WRITE) == 0) + goto out_iput; + res->size = i_size_read(inode); + res->change_attr = NFS_CHANGE_ATTR(inode); + res->ctime = inode->i_ctime; + res->mtime = inode->i_mtime; + res->bitmap[0] = (FATTR4_WORD0_CHANGE|FATTR4_WORD0_SIZE) & + args->bitmap[0]; + res->bitmap[1] = (FATTR4_WORD1_TIME_METADATA|FATTR4_WORD1_TIME_MODIFY) & + args->bitmap[1]; + res->status = 0; +out_iput: + up_read(&nfsi->rwsem); + iput(inode); +out_putclient: + nfs4_put_client(clp); +out: + dprintk("%s: exit with status = %d\n", __FUNCTION__, ntohl(res->status)); + return res->status; +} + +unsigned nfs4_callback_recall(struct cb_recallargs *args, void *dummy) +{ + struct nfs4_client *clp; + struct inode *inode; + unsigned res; + + res = htonl(NFS4ERR_BADHANDLE); + clp = nfs4_find_client(&args->addr->sin_addr); + if (clp == NULL) + goto out; + inode = nfs_delegation_find_inode(clp, &args->fh); + if (inode == NULL) + goto out_putclient; + /* Set up a helper thread to actually return the delegation */ + switch(nfs_async_inode_return_delegation(inode, &args->stateid)) { + case 0: + res = 0; + break; + case -ENOENT: + res = htonl(NFS4ERR_BAD_STATEID); + break; + default: + res = htonl(NFS4ERR_RESOURCE); + } + iput(inode); +out_putclient: + nfs4_put_client(clp); +out: + dprintk("%s: exit with status = %d\n", __FUNCTION__, ntohl(res)); + return res; +} diff -u --recursive --new-file --show-c-function linux-2.6.7/fs/nfs/callback_xdr.c linux-2.6.7-23-rpc_queue_lock/fs/nfs/callback_xdr.c --- linux-2.6.7/fs/nfs/callback_xdr.c 1969-12-31 19:00:00.000000000 -0500 +++ linux-2.6.7-23-rpc_queue_lock/fs/nfs/callback_xdr.c 2004-06-22 16:49:06.000000000 -0400 @@ -0,0 +1,481 @@ +/* + * linux/fs/nfs/callback_xdr.c + * + * Copyright (C) 2004 Trond Myklebust + * + * NFSv4 callback encode/decode procedures + */ +#include +#include +#include +#include +#include +#include "callback.h" + +#define CB_OP_TAGLEN_MAXSZ (512) +#define CB_OP_HDR_RES_MAXSZ (2 + CB_OP_TAGLEN_MAXSZ) +#define CB_OP_GETATTR_BITMAP_MAXSZ (4) +#define CB_OP_GETATTR_RES_MAXSZ (CB_OP_HDR_RES_MAXSZ + \ + CB_OP_GETATTR_BITMAP_MAXSZ + \ + 2 + 2 + 3 + 3) +#define CB_OP_RECALL_RES_MAXSZ (CB_OP_HDR_RES_MAXSZ) + +#define NFSDBG_FACILITY NFSDBG_CALLBACK + +typedef unsigned (*callback_process_op_t)(void *, void *); +typedef unsigned (*callback_decode_arg_t)(struct svc_rqst *, struct xdr_stream *, void *); +typedef unsigned (*callback_encode_res_t)(struct svc_rqst *, struct xdr_stream *, void *); + + +struct callback_op { + callback_process_op_t process_op; + callback_decode_arg_t decode_args; + callback_encode_res_t encode_res; + long res_maxsize; +}; + +static struct callback_op callback_ops[]; + +static int nfs4_callback_null(struct svc_rqst *rqstp, void *argp, void *resp) +{ + return htonl(NFS4_OK); +} + +static int nfs4_decode_void(struct svc_rqst *rqstp, uint32_t *p, void *dummy) +{ + return xdr_argsize_check(rqstp, p); +} + +static int nfs4_encode_void(struct svc_rqst *rqstp, uint32_t *p, void *dummy) +{ + return xdr_ressize_check(rqstp, p); +} + +static uint32_t *read_buf(struct xdr_stream *xdr, int nbytes) +{ + uint32_t *p; + + p = xdr_inline_decode(xdr, nbytes); + if (unlikely(p == NULL)) + printk(KERN_WARNING "NFSv4 callback reply buffer overflowed!\n"); + return p; +} + +static unsigned decode_string(struct xdr_stream *xdr, unsigned int *len, const char **str) +{ + uint32_t *p; + + p = read_buf(xdr, 4); + if (unlikely(p == NULL)) + return htonl(NFS4ERR_RESOURCE); + *len = ntohl(*p); + + if (*len != 0) { + p = read_buf(xdr, *len); + if (unlikely(p == NULL)) + return htonl(NFS4ERR_RESOURCE); + *str = (const char *)p; + } else + *str = NULL; + + return 0; +} + +static unsigned decode_fh(struct xdr_stream *xdr, struct nfs_fh *fh) +{ + uint32_t *p; + + p = read_buf(xdr, 4); + if (unlikely(p == NULL)) + return htonl(NFS4ERR_RESOURCE); + fh->size = ntohl(*p); + if (fh->size > NFS4_FHSIZE) + return htonl(NFS4ERR_BADHANDLE); + p = read_buf(xdr, fh->size); + if (unlikely(p == NULL)) + return htonl(NFS4ERR_RESOURCE); + memcpy(&fh->data[0], p, fh->size); + memset(&fh->data[fh->size], 0, sizeof(fh->data) - fh->size); + return 0; +} + +static unsigned decode_bitmap(struct xdr_stream *xdr, uint32_t *bitmap) +{ + uint32_t *p; + unsigned int attrlen; + + p = read_buf(xdr, 4); + if (unlikely(p == NULL)) + return htonl(NFS4ERR_RESOURCE); + attrlen = ntohl(*p); + p = read_buf(xdr, attrlen << 2); + if (unlikely(p == NULL)) + return htonl(NFS4ERR_RESOURCE); + if (likely(attrlen > 0)) + bitmap[0] = ntohl(*p++); + if (attrlen > 1) + bitmap[1] = ntohl(*p); + return 0; +} + +static unsigned decode_stateid(struct xdr_stream *xdr, nfs4_stateid *stateid) +{ + uint32_t *p; + + p = read_buf(xdr, 16); + if (unlikely(p == NULL)) + return htonl(NFS4ERR_RESOURCE); + memcpy(stateid->data, p, 16); + return 0; +} + +static unsigned decode_compound_hdr_arg(struct xdr_stream *xdr, struct cb_compound_hdr_arg *hdr) +{ + uint32_t *p; + unsigned int minor_version; + unsigned status; + + status = decode_string(xdr, &hdr->taglen, &hdr->tag); + if (unlikely(status != 0)) + return status; + /* We do not like overly long tags! */ + if (hdr->taglen > CB_OP_TAGLEN_MAXSZ-12 || hdr->taglen < 0) { + printk("NFSv4 CALLBACK %s: client sent tag of length %u\n", + __FUNCTION__, hdr->taglen); + return htonl(NFS4ERR_RESOURCE); + } + p = read_buf(xdr, 12); + if (unlikely(p == NULL)) + return htonl(NFS4ERR_RESOURCE); + minor_version = ntohl(*p++); + /* Check minor version is zero. */ + if (minor_version != 0) { + printk(KERN_WARNING "%s: NFSv4 server callback with illegal minor version %u!\n", + __FUNCTION__, minor_version); + return htonl(NFS4ERR_MINOR_VERS_MISMATCH); + } + hdr->callback_ident = ntohl(*p++); + hdr->nops = ntohl(*p); + return 0; +} + +static unsigned decode_op_hdr(struct xdr_stream *xdr, unsigned int *op) +{ + uint32_t *p; + p = read_buf(xdr, 4); + if (unlikely(p == NULL)) + return htonl(NFS4ERR_RESOURCE); + *op = ntohl(*p); + return 0; +} + +static unsigned decode_getattr_args(struct svc_rqst *rqstp, struct xdr_stream *xdr, struct cb_getattrargs *args) +{ + unsigned status; + + status = decode_fh(xdr, &args->fh); + if (unlikely(status != 0)) + goto out; + args->addr = &rqstp->rq_addr; + status = decode_bitmap(xdr, args->bitmap); +out: + dprintk("%s: exit with status = %d\n", __FUNCTION__, status); + return status; +} + +static unsigned decode_recall_args(struct svc_rqst *rqstp, struct xdr_stream *xdr, struct cb_recallargs *args) +{ + uint32_t *p; + unsigned status; + + args->addr = &rqstp->rq_addr; + status = decode_stateid(xdr, &args->stateid); + if (unlikely(status != 0)) + goto out; + p = read_buf(xdr, 4); + if (unlikely(p == NULL)) { + status = htonl(NFS4ERR_RESOURCE); + goto out; + } + args->truncate = ntohl(*p); + status = decode_fh(xdr, &args->fh); +out: + dprintk("%s: exit with status = %d\n", __FUNCTION__, status); + return 0; +} + +static unsigned encode_string(struct xdr_stream *xdr, unsigned int len, const char *str) +{ + uint32_t *p; + + p = xdr_reserve_space(xdr, 4 + len); + if (unlikely(p == NULL)) + return htonl(NFS4ERR_RESOURCE); + xdr_encode_opaque(p, str, len); + return 0; +} + +#define CB_SUPPORTED_ATTR0 (FATTR4_WORD0_CHANGE|FATTR4_WORD0_SIZE) +#define CB_SUPPORTED_ATTR1 (FATTR4_WORD1_TIME_METADATA|FATTR4_WORD1_TIME_MODIFY) +static unsigned encode_attr_bitmap(struct xdr_stream *xdr, const uint32_t *bitmap, uint32_t **savep) +{ + uint32_t bm[2]; + uint32_t *p; + + bm[0] = htonl(bitmap[0] & CB_SUPPORTED_ATTR0); + bm[1] = htonl(bitmap[1] & CB_SUPPORTED_ATTR1); + if (bm[1] != 0) { + p = xdr_reserve_space(xdr, 16); + if (unlikely(p == NULL)) + return htonl(NFS4ERR_RESOURCE); + *p++ = htonl(2); + *p++ = bm[0]; + *p++ = bm[1]; + } else if (bm[0] != 0) { + p = xdr_reserve_space(xdr, 12); + if (unlikely(p == NULL)) + return htonl(NFS4ERR_RESOURCE); + *p++ = htonl(1); + *p++ = bm[0]; + } else { + p = xdr_reserve_space(xdr, 8); + if (unlikely(p == NULL)) + return htonl(NFS4ERR_RESOURCE); + *p++ = htonl(0); + } + *savep = p; + return 0; +} + +static unsigned encode_attr_change(struct xdr_stream *xdr, const uint32_t *bitmap, uint64_t change) +{ + uint32_t *p; + + if (!(bitmap[0] & FATTR4_WORD0_CHANGE)) + return 0; + p = xdr_reserve_space(xdr, 8); + if (unlikely(p == 0)) + return htonl(NFS4ERR_RESOURCE); + p = xdr_encode_hyper(p, change); + return 0; +} + +static unsigned encode_attr_size(struct xdr_stream *xdr, const uint32_t *bitmap, uint64_t size) +{ + uint32_t *p; + + if (!(bitmap[0] & FATTR4_WORD0_SIZE)) + return 0; + p = xdr_reserve_space(xdr, 8); + if (unlikely(p == 0)) + return htonl(NFS4ERR_RESOURCE); + p = xdr_encode_hyper(p, size); + return 0; +} + +static unsigned encode_attr_time(struct xdr_stream *xdr, const struct timespec *time) +{ + uint32_t *p; + + p = xdr_reserve_space(xdr, 12); + if (unlikely(p == 0)) + return htonl(NFS4ERR_RESOURCE); + p = xdr_encode_hyper(p, time->tv_sec); + *p = htonl(time->tv_nsec); + return 0; +} + +static unsigned encode_attr_ctime(struct xdr_stream *xdr, const uint32_t *bitmap, const struct timespec *time) +{ + if (!(bitmap[1] & FATTR4_WORD1_TIME_METADATA)) + return 0; + return encode_attr_time(xdr,time); +} + +static unsigned encode_attr_mtime(struct xdr_stream *xdr, const uint32_t *bitmap, const struct timespec *time) +{ + if (!(bitmap[1] & FATTR4_WORD1_TIME_MODIFY)) + return 0; + return encode_attr_time(xdr,time); +} + +static unsigned encode_compound_hdr_res(struct xdr_stream *xdr, struct cb_compound_hdr_res *hdr) +{ + unsigned status; + + hdr->status = xdr_reserve_space(xdr, 4); + if (unlikely(hdr->status == NULL)) + return htonl(NFS4ERR_RESOURCE); + status = encode_string(xdr, hdr->taglen, hdr->tag); + if (unlikely(status != 0)) + return status; + hdr->nops = xdr_reserve_space(xdr, 4); + if (unlikely(hdr->nops == NULL)) + return htonl(NFS4ERR_RESOURCE); + return 0; +} + +static unsigned encode_op_hdr(struct xdr_stream *xdr, uint32_t op, uint32_t res) +{ + uint32_t *p; + + p = xdr_reserve_space(xdr, 8); + if (unlikely(p == NULL)) + return htonl(NFS4ERR_RESOURCE); + *p++ = htonl(op); + *p = htonl(res); + return 0; +} + +static unsigned encode_getattr_res(struct svc_rqst *rqstp, struct xdr_stream *xdr, const struct cb_getattrres *res) +{ + uint32_t *savep; + unsigned status = res->status; + + if (unlikely(status != 0)) + goto out; + status = encode_attr_bitmap(xdr, res->bitmap, &savep); + if (unlikely(status != 0)) + goto out; + status = encode_attr_change(xdr, res->bitmap, res->change_attr); + if (unlikely(status != 0)) + goto out; + status = encode_attr_size(xdr, res->bitmap, res->size); + if (unlikely(status != 0)) + goto out; + status = encode_attr_ctime(xdr, res->bitmap, &res->ctime); + if (unlikely(status != 0)) + goto out; + status = encode_attr_mtime(xdr, res->bitmap, &res->mtime); + *savep = htonl((unsigned int)((char *)xdr->p - (char *)(savep+1))); +out: + dprintk("%s: exit with status = %d\n", __FUNCTION__, status); + return status; +} + +static unsigned process_op(struct svc_rqst *rqstp, + struct xdr_stream *xdr_in, void *argp, + struct xdr_stream *xdr_out, void *resp) +{ + struct callback_op *op; + unsigned int op_nr; + unsigned int status = 0; + long maxlen; + unsigned res; + + dprintk("%s: start\n", __FUNCTION__); + status = decode_op_hdr(xdr_in, &op_nr); + if (unlikely(status != 0)) { + op_nr = OP_CB_ILLEGAL; + op = &callback_ops[0]; + } else if (unlikely(op_nr != OP_CB_GETATTR && op_nr != OP_CB_RECALL)) { + op_nr = OP_CB_ILLEGAL; + op = &callback_ops[0]; + status = htonl(NFS4ERR_OP_ILLEGAL); + } else + op = &callback_ops[op_nr]; + + maxlen = xdr_out->end - xdr_out->p; + if (maxlen > 0 && maxlen < PAGE_SIZE) { + if (likely(status == 0 && op->decode_args != NULL)) + status = op->decode_args(rqstp, xdr_in, argp); + if (likely(status == 0 && op->process_op != NULL)) + status = op->process_op(argp, resp); + } else + status = htonl(NFS4ERR_RESOURCE); + + res = encode_op_hdr(xdr_out, op_nr, status); + if (status == 0) + status = res; + if (op->encode_res != NULL && status == 0) + status = op->encode_res(rqstp, xdr_out, resp); + dprintk("%s: done, status = %d\n", __FUNCTION__, status); + return status; +} + +/* + * Decode, process and encode a COMPOUND + */ +static int nfs4_callback_compound(struct svc_rqst *rqstp, void *argp, void *resp) +{ + struct cb_compound_hdr_arg hdr_arg; + struct cb_compound_hdr_res hdr_res; + struct xdr_stream xdr_in, xdr_out; + uint32_t *p; + unsigned int status; + unsigned int nops = 1; + + dprintk("%s: start\n", __FUNCTION__); + + xdr_init_decode(&xdr_in, &rqstp->rq_arg, rqstp->rq_arg.head[0].iov_base); + + p = (uint32_t*)((char *)rqstp->rq_res.head[0].iov_base + rqstp->rq_res.head[0].iov_len); + rqstp->rq_res.head[0].iov_len = PAGE_SIZE; + xdr_init_encode(&xdr_out, &rqstp->rq_res, p); + + decode_compound_hdr_arg(&xdr_in, &hdr_arg); + hdr_res.taglen = hdr_arg.taglen; + hdr_res.tag = hdr_arg.tag; + encode_compound_hdr_res(&xdr_out, &hdr_res); + + for (;;) { + status = process_op(rqstp, &xdr_in, argp, &xdr_out, resp); + if (status != 0) + break; + if (nops == hdr_arg.nops) + break; + nops++; + } + *hdr_res.status = status; + *hdr_res.nops = htonl(nops); + dprintk("%s: done, status = %u\n", __FUNCTION__, status); + return rpc_success; +} + +/* + * Define NFS4 callback COMPOUND ops. + */ +static struct callback_op callback_ops[] = { + [0] = { + .res_maxsize = CB_OP_HDR_RES_MAXSZ, + }, + [OP_CB_GETATTR] = { + .process_op = (callback_process_op_t)nfs4_callback_getattr, + .decode_args = (callback_decode_arg_t)decode_getattr_args, + .encode_res = (callback_encode_res_t)encode_getattr_res, + .res_maxsize = CB_OP_GETATTR_RES_MAXSZ, + }, + [OP_CB_RECALL] = { + .process_op = (callback_process_op_t)nfs4_callback_recall, + .decode_args = (callback_decode_arg_t)decode_recall_args, + .res_maxsize = CB_OP_RECALL_RES_MAXSZ, + } +}; + +/* + * Define NFS4 callback procedures + */ +static struct svc_procedure nfs4_callback_procedures1[] = { + [CB_NULL] = { + .pc_func = nfs4_callback_null, + .pc_decode = (kxdrproc_t)nfs4_decode_void, + .pc_encode = (kxdrproc_t)nfs4_encode_void, + .pc_xdrressize = 1, + }, + [CB_COMPOUND] = { + .pc_func = nfs4_callback_compound, + .pc_encode = (kxdrproc_t)nfs4_encode_void, + .pc_argsize = 256, + .pc_ressize = 256, + .pc_xdrressize = NFS4_CALLBACK_BUFSIZE, + } +}; + +struct svc_version nfs4_callback_version1 = { + .vs_vers = 1, + .vs_nproc = ARRAY_SIZE(nfs4_callback_procedures1), + .vs_proc = nfs4_callback_procedures1, + .vs_xdrsize = NFS4_CALLBACK_XDRSIZE, + .vs_dispatch = NULL, +}; + diff -u --recursive --new-file --show-c-function linux-2.6.7/fs/nfs/delegation.c linux-2.6.7-23-rpc_queue_lock/fs/nfs/delegation.c --- linux-2.6.7/fs/nfs/delegation.c 1969-12-31 19:00:00.000000000 -0500 +++ linux-2.6.7-23-rpc_queue_lock/fs/nfs/delegation.c 2004-06-22 16:49:41.000000000 -0400 @@ -0,0 +1,320 @@ +/* + * linux/fs/nfs/delegation.c + * + * Copyright (C) 2004 Trond Myklebust + * + * NFS file delegation management + * + */ +#include +#include +#include +#include +#include + +#include +#include +#include + +#include "delegation.h" + +static struct nfs_delegation *nfs_alloc_delegation(void) +{ + return (struct nfs_delegation *)kmalloc(sizeof(struct nfs_delegation), GFP_KERNEL); +} + +static void nfs_free_delegation(struct nfs_delegation *delegation) +{ + if (delegation->cred) + put_rpccred(delegation->cred); + kfree(delegation); +} + +static void nfs_delegation_claim_opens(struct inode *inode) +{ + struct nfs_inode *nfsi = NFS_I(inode); + struct nfs_open_context *ctx; + struct nfs4_state *state; + +again: + spin_lock(&inode->i_lock); + list_for_each_entry(ctx, &nfsi->open_files, list) { + state = ctx->state; + if (state == NULL) + continue; + if (!test_bit(NFS_DELEGATED_STATE, &state->flags)) + continue; + get_nfs_open_context(ctx); + spin_unlock(&inode->i_lock); + if (nfs4_open_delegation_recall(ctx->dentry, state) < 0) + return; + put_nfs_open_context(ctx); + goto again; + } + spin_unlock(&inode->i_lock); +} + +/* + * Set up a delegation on an inode + */ +void nfs_inode_reclaim_delegation(struct inode *inode, struct rpc_cred *cred, struct nfs_openres *res) +{ + struct nfs_delegation *delegation = NFS_I(inode)->delegation; + + if (delegation == NULL) + return; + memcpy(delegation->stateid.data, res->delegation.data, + sizeof(delegation->stateid.data)); + delegation->type = res->delegation_type; + delegation->maxsize = res->maxsize; + put_rpccred(cred); + delegation->cred = get_rpccred(cred); + delegation->flags &= ~NFS_DELEGATION_NEED_RECLAIM; + NFS_I(inode)->delegation_state = delegation->type; + wmb(); +} + +/* + * Set up a delegation on an inode + */ +int nfs_inode_set_delegation(struct inode *inode, struct rpc_cred *cred, struct nfs_openres *res) +{ + struct nfs4_client *clp = NFS_SERVER(inode)->nfs4_state; + struct nfs_inode *nfsi = NFS_I(inode); + struct nfs_delegation *delegation; + int status = 0; + + delegation = nfs_alloc_delegation(); + if (delegation == NULL) + return -ENOMEM; + memcpy(delegation->stateid.data, res->delegation.data, + sizeof(delegation->stateid.data)); + delegation->type = res->delegation_type; + delegation->maxsize = res->maxsize; + delegation->cred = get_rpccred(cred); + delegation->inode = inode; + + spin_lock(&clp->cl_lock); + if (nfsi->delegation == NULL) { + list_add(&delegation->super_list, &clp->cl_delegations); + nfsi->delegation = delegation; + nfsi->delegation_state = delegation->type; + delegation = NULL; + } else { + if (memcmp(&delegation->stateid, &nfsi->delegation->stateid, + sizeof(delegation->stateid)) != 0 || + delegation->type != nfsi->delegation->type) { + printk("%s: server %u.%u.%u.%u, handed out a duplicate delegation!\n", + __FUNCTION__, NIPQUAD(clp->cl_addr)); + status = -EIO; + } + } + spin_unlock(&clp->cl_lock); + if (delegation != NULL) + kfree(delegation); + return status; +} + +static int nfs_do_return_delegation(struct inode *inode, struct nfs_delegation *delegation) +{ + int res = 0; + + __nfs_revalidate_inode(NFS_SERVER(inode), inode); + + res = nfs4_proc_delegreturn(inode, delegation->cred, &delegation->stateid); + nfs_free_delegation(delegation); + return res; +} + +/* Sync all data to disk upon delegation return */ +static void nfs_msync_inode(struct inode *inode) +{ + down(&inode->i_sem); + filemap_fdatawrite(inode->i_mapping); + nfs_wb_all(inode); + filemap_fdatawait(inode->i_mapping); + up(&inode->i_sem); +} + +/* + * Basic procedure for returning a delegation to the server + */ +int nfs_inode_return_delegation(struct inode *inode) +{ + struct nfs4_client *clp = NFS_SERVER(inode)->nfs4_state; + struct nfs_inode *nfsi = NFS_I(inode); + struct nfs_delegation *delegation; + int res = 0; + + nfs_msync_inode(inode); + down_read(&clp->cl_sem); + /* Guard against new delegated open calls */ + down_write(&nfsi->rwsem); + spin_lock(&clp->cl_lock); + delegation = nfsi->delegation; + if (delegation != NULL) { + list_del_init(&delegation->super_list); + nfsi->delegation = NULL; + nfsi->delegation_state = 0; + } + spin_unlock(&clp->cl_lock); + nfs_delegation_claim_opens(inode); + up_write(&nfsi->rwsem); + up_read(&clp->cl_sem); + nfs_msync_inode(inode); + + if (delegation != NULL) + res = nfs_do_return_delegation(inode, delegation); + return res; +} + +/* + * Return all delegations associated to a super block + */ +void nfs_return_all_delegations(struct super_block *sb) +{ + struct nfs4_client *clp = NFS_SB(sb)->nfs4_state; + struct nfs_delegation *delegation; + struct inode *inode; + + if (clp == NULL) + return; +restart: + spin_lock(&clp->cl_lock); + list_for_each_entry(delegation, &clp->cl_delegations, super_list) { + if (delegation->inode->i_sb != sb) + continue; + inode = igrab(delegation->inode); + if (inode == NULL) + continue; + spin_unlock(&clp->cl_lock); + nfs_inode_return_delegation(inode); + iput(inode); + goto restart; + } + spin_unlock(&clp->cl_lock); +} + +struct recall_threadargs { + struct inode *inode; + struct nfs4_client *clp; + const nfs4_stateid *stateid; + + struct completion started; + int result; +}; + +static int recall_thread(void *data) +{ + struct recall_threadargs *args = (struct recall_threadargs *)data; + struct inode *inode = igrab(args->inode); + struct nfs4_client *clp = NFS_SERVER(inode)->nfs4_state; + struct nfs_inode *nfsi = NFS_I(inode); + struct nfs_delegation *delegation; + + daemonize("nfsv4-delegreturn"); + + nfs_msync_inode(inode); + down_read(&clp->cl_sem); + down_write(&nfsi->rwsem); + spin_lock(&clp->cl_lock); + delegation = nfsi->delegation; + if (delegation != NULL && memcmp(delegation->stateid.data, + args->stateid->data, + sizeof(delegation->stateid.data)) == 0) { + list_del_init(&delegation->super_list); + nfsi->delegation = NULL; + nfsi->delegation_state = 0; + args->result = 0; + } else { + delegation = NULL; + args->result = -ENOENT; + } + spin_unlock(&clp->cl_lock); + complete(&args->started); + nfs_delegation_claim_opens(inode); + up_write(&nfsi->rwsem); + up_read(&clp->cl_sem); + nfs_msync_inode(inode); + + if (delegation != NULL) + nfs_do_return_delegation(inode, delegation); + iput(inode); + module_put_and_exit(0); +} + +/* + * Asynchronous delegation recall! + */ +int nfs_async_inode_return_delegation(struct inode *inode, const nfs4_stateid *stateid) +{ + struct recall_threadargs data = { + .inode = inode, + .stateid = stateid, + }; + int status; + + init_completion(&data.started); + __module_get(THIS_MODULE); + status = kernel_thread(recall_thread, &data, CLONE_KERNEL); + if (status < 0) + goto out_module_put; + wait_for_completion(&data.started); + return data.result; +out_module_put: + module_put(THIS_MODULE); + return status; +} + +/* + * Retrieve the inode associated with a delegation + */ +struct inode *nfs_delegation_find_inode(struct nfs4_client *clp, const struct nfs_fh *fhandle) +{ + struct nfs_delegation *delegation; + struct inode *res = NULL; + spin_lock(&clp->cl_lock); + list_for_each_entry(delegation, &clp->cl_delegations, super_list) { + if (nfs_compare_fh(fhandle, &NFS_I(delegation->inode)->fh) == 0) { + res = igrab(delegation->inode); + break; + } + } + spin_unlock(&clp->cl_lock); + return res; +} + +/* + * Mark all delegations as needing to be reclaimed + */ +void nfs_delegation_mark_reclaim(struct nfs4_client *clp) +{ + struct nfs_delegation *delegation; + spin_lock(&clp->cl_lock); + list_for_each_entry(delegation, &clp->cl_delegations, super_list) + delegation->flags |= NFS_DELEGATION_NEED_RECLAIM; + spin_unlock(&clp->cl_lock); +} + +/* + * Reap all unclaimed delegations after reboot recovery is done + */ +void nfs_delegation_reap_unclaimed(struct nfs4_client *clp) +{ + struct nfs_delegation *delegation, *n; + LIST_HEAD(head); + spin_lock(&clp->cl_lock); + list_for_each_entry_safe(delegation, n, &clp->cl_delegations, super_list) { + if ((delegation->flags & NFS_DELEGATION_NEED_RECLAIM) == 0) + continue; + list_move(&delegation->super_list, &head); + NFS_I(delegation->inode)->delegation = NULL; + NFS_I(delegation->inode)->delegation_state = 0; + } + spin_unlock(&clp->cl_lock); + while(!list_empty(&head)) { + delegation = list_entry(head.next, struct nfs_delegation, super_list); + list_del(&delegation->super_list); + nfs_free_delegation(delegation); + } +} diff -u --recursive --new-file --show-c-function linux-2.6.7/fs/nfs/delegation.h linux-2.6.7-23-rpc_queue_lock/fs/nfs/delegation.h --- linux-2.6.7/fs/nfs/delegation.h 1969-12-31 19:00:00.000000000 -0500 +++ linux-2.6.7-23-rpc_queue_lock/fs/nfs/delegation.h 2004-06-22 16:49:41.000000000 -0400 @@ -0,0 +1,56 @@ +/* + * linux/fs/nfs/delegation.h + * + * Copyright (c) Trond Myklebust + * + * Definitions pertaining to NFS delegated files + */ +#ifndef FS_NFS_DELEGATION_H +#define FS_NFS_DELEGATION_H + +#if defined(CONFIG_NFS_V4) +/* + * NFSv4 delegation + */ +struct nfs_delegation { + struct list_head super_list; + struct rpc_cred *cred; + struct inode *inode; + nfs4_stateid stateid; + int type; +#define NFS_DELEGATION_NEED_RECLAIM 1 + long flags; + loff_t maxsize; +}; + +int nfs_inode_set_delegation(struct inode *inode, struct rpc_cred *cred, struct nfs_openres *res); +void nfs_inode_reclaim_delegation(struct inode *inode, struct rpc_cred *cred, struct nfs_openres *res); +int nfs_inode_return_delegation(struct inode *inode); +int nfs_async_inode_return_delegation(struct inode *inode, const nfs4_stateid *stateid); + +struct inode *nfs_delegation_find_inode(struct nfs4_client *clp, const struct nfs_fh *fhandle); +void nfs_return_all_delegations(struct super_block *sb); + +void nfs_delegation_mark_reclaim(struct nfs4_client *clp); +void nfs_delegation_reap_unclaimed(struct nfs4_client *clp); + +/* NFSv4 delegation-related procedures */ +int nfs4_proc_delegreturn(struct inode *inode, struct rpc_cred *cred, const nfs4_stateid *stateid); +int nfs4_open_delegation_recall(struct dentry *dentry, struct nfs4_state *state); + +static inline int nfs_have_delegation(struct inode *inode, int flags) +{ + flags &= FMODE_READ|FMODE_WRITE; + rmb(); + if ((NFS_I(inode)->delegation_state & flags) == flags) + return 1; + return 0; +} +#else +static inline int nfs_have_delegation(struct inode *inode, int flags) +{ + return 0; +} +#endif + +#endif diff -u --recursive --new-file --show-c-function linux-2.6.7/fs/nfs/dir.c linux-2.6.7-23-rpc_queue_lock/fs/nfs/dir.c --- linux-2.6.7/fs/nfs/dir.c 2004-06-16 14:58:05.000000000 -0400 +++ linux-2.6.7-23-rpc_queue_lock/fs/nfs/dir.c 2004-06-22 16:49:41.000000000 -0400 @@ -32,6 +32,8 @@ #include #include +#include "delegation.h" + #define NFS_PARANOIA 1 /* #define NFS_DEBUG_VERBOSE 1 */ @@ -850,22 +852,22 @@ static int nfs_open_revalidate(struct de unsigned long verifier; int openflags, ret = 0; - /* NFS only supports OPEN for regular files */ - if (inode && !S_ISREG(inode->i_mode)) - goto no_open; parent = dget_parent(dentry); dir = parent->d_inode; if (!is_atomic_open(dir, nd)) goto no_open; + /* We can't create new files in nfs_open_revalidate(), so we + * optimize away revalidation of negative dentries. + */ + if (inode == NULL) + goto out; + /* NFS only supports OPEN on regular files */ + if (!S_ISREG(inode->i_mode)) + goto no_open; openflags = nd->intent.open.flags; - if (openflags & O_CREAT) { - /* If this is a negative dentry, just drop it */ - if (!inode) - goto out; - /* If this is exclusive open, just revalidate */ - if (openflags & O_EXCL) - goto no_open; - } + /* We cannot do exclusive creation on a positive dentry */ + if ((openflags & (O_CREAT|O_EXCL)) == (O_CREAT|O_EXCL)) + goto no_open; /* We can't create new files, or truncate existing ones here */ openflags &= ~(O_CREAT|O_TRUNC); @@ -887,6 +889,8 @@ out: return ret; no_open: dput(parent); + if (inode != NULL && nfs_have_delegation(inode, FMODE_READ)) + return 1; return nfs_lookup_revalidate(dentry, nd); } #endif /* CONFIG_NFSV4 */ @@ -1299,19 +1303,6 @@ nfs_symlink(struct inode *dir, struct de dfprintk(VFS, "NFS: symlink(%s/%ld, %s, %s)\n", dir->i_sb->s_id, dir->i_ino, dentry->d_name.name, symname); - error = -ENAMETOOLONG; - switch (NFS_PROTO(dir)->version) { - case 2: - if (strlen(symname) > NFS2_MAXPATHLEN) - goto out; - break; - case 3: - if (strlen(symname) > NFS3_MAXPATHLEN) - goto out; - default: - break; - } - #ifdef NFS_PARANOIA if (dentry->d_inode) printk("nfs_proc_symlink: %s/%s not negative!\n", @@ -1341,8 +1332,6 @@ dentry->d_parent->d_name.name, dentry->d d_drop(dentry); } unlock_kernel(); - -out: return error; } @@ -1498,10 +1487,56 @@ out: return error; } -int -nfs_permission(struct inode *inode, int mask, struct nameidata *nd) +int nfs_access_get_cached(struct inode *inode, struct rpc_cred *cred, struct nfs_access_entry *res) +{ + struct nfs_access_entry *cache = &NFS_I(inode)->cache_access; + + if (cache->cred != cred + || time_after(jiffies, cache->jiffies + NFS_ATTRTIMEO(inode)) + || (NFS_FLAGS(inode) & NFS_INO_INVALID_ATTR)) + return -ENOENT; + memcpy(res, cache, sizeof(*res)); + return 0; +} + +void nfs_access_add_cache(struct inode *inode, struct nfs_access_entry *set) +{ + struct nfs_access_entry *cache = &NFS_I(inode)->cache_access; + + if (cache->cred != set->cred) { + if (cache->cred) + put_rpccred(cache->cred); + cache->cred = get_rpccred(set->cred); + } + cache->jiffies = set->jiffies; + cache->mask = set->mask; +} + +static int nfs_do_access(struct inode *inode, struct rpc_cred *cred, int mask) +{ + struct nfs_access_entry cache; + int status; + + status = nfs_access_get_cached(inode, cred, &cache); + if (status == 0) + goto out; + + /* Be clever: ask server to check for all possible rights */ + cache.mask = MAY_EXEC | MAY_WRITE | MAY_READ; + cache.cred = cred; + cache.jiffies = jiffies; + status = NFS_PROTO(inode)->access(inode, &cache); + if (status != 0) + return status; + nfs_access_add_cache(inode, &cache); +out: + if ((cache.mask & mask) == mask) + return 0; + return -EACCES; +} + +int nfs_permission(struct inode *inode, int mask, struct nameidata *nd) { - struct nfs_access_cache *cache = &NFS_I(inode)->cache_access; struct rpc_cred *cred; int mode = inode->i_mode; int res; @@ -1542,24 +1577,7 @@ nfs_permission(struct inode *inode, int goto out_notsup; cred = rpcauth_lookupcred(NFS_CLIENT(inode)->cl_auth, 0); - if (cache->cred == cred - && time_before(jiffies, cache->jiffies + NFS_ATTRTIMEO(inode)) - && !(NFS_FLAGS(inode) & NFS_INO_INVALID_ATTR)) { - if (!(res = cache->err)) { - /* Is the mask a subset of an accepted mask? */ - if ((cache->mask & mask) == mask) - goto out; - } else { - /* ...or is it a superset of a rejected mask? */ - if ((cache->mask & mask) == cache->mask) - goto out; - } - } - - res = NFS_PROTO(inode)->access(inode, cred, mask); - if (!res || res == -EACCES) - goto add_cache; -out: + res = nfs_do_access(inode, cred, mask); put_rpccred(cred); unlock_kernel(); return res; @@ -1568,15 +1586,6 @@ out_notsup: res = vfs_permission(inode, mask); unlock_kernel(); return res; -add_cache: - cache->jiffies = jiffies; - if (cache->cred) - put_rpccred(cache->cred); - cache->cred = cred; - cache->mask = mask; - cache->err = res; - unlock_kernel(); - return res; } /* diff -u --recursive --new-file --show-c-function linux-2.6.7/fs/nfs/direct.c linux-2.6.7-23-rpc_queue_lock/fs/nfs/direct.c --- linux-2.6.7/fs/nfs/direct.c 2004-06-16 14:58:26.000000000 -0400 +++ linux-2.6.7-23-rpc_queue_lock/fs/nfs/direct.c 2004-06-22 16:48:55.000000000 -0400 @@ -110,7 +110,7 @@ nfs_free_user_pages(struct page **pages, * nfs_direct_read_seg - Read in one iov segment. Generate separate * read RPCs for each "rsize" bytes. * @inode: target inode - * @file: target file (may be NULL) + * @ctx: target file open context * user_addr: starting address of this segment of user's buffer * count: size of this segment * file_offset: offset in file to begin the operation @@ -118,7 +118,7 @@ nfs_free_user_pages(struct page **pages, * nr_pages: size of pages array */ static int -nfs_direct_read_seg(struct inode *inode, struct file *file, +nfs_direct_read_seg(struct inode *inode, struct nfs_open_context *ctx, unsigned long user_addr, size_t count, loff_t file_offset, struct page **pages, int nr_pages) { @@ -127,9 +127,10 @@ nfs_direct_read_seg(struct inode *inode, int curpage = 0; struct nfs_read_data rdata = { .inode = inode, + .cred = ctx->cred, .args = { .fh = NFS_FH(inode), - .lockowner = current->files, + .context = ctx, }, .res = { .fattr = &rdata.fattr, @@ -151,7 +152,7 @@ nfs_direct_read_seg(struct inode *inode, user_addr + tot_bytes, rdata.args.pgbase, curpage); lock_kernel(); - result = NFS_PROTO(inode)->read(&rdata, file); + result = NFS_PROTO(inode)->read(&rdata); unlock_kernel(); if (result <= 0) { @@ -183,7 +184,7 @@ nfs_direct_read_seg(struct inode *inode, * nfs_direct_read - For each iov segment, map the user's buffer * then generate read RPCs. * @inode: target inode - * @file: target file (may be NULL) + * @ctx: target file open context * @iov: array of vectors that define I/O buffer * file_offset: offset in file to begin the operation * nr_segs: size of iovec array @@ -193,7 +194,7 @@ nfs_direct_read_seg(struct inode *inode, * server. */ static ssize_t -nfs_direct_read(struct inode *inode, struct file *file, +nfs_direct_read(struct inode *inode, struct nfs_open_context *ctx, const struct iovec *iov, loff_t file_offset, unsigned long nr_segs) { @@ -216,7 +217,7 @@ nfs_direct_read(struct inode *inode, str return page_count; } - result = nfs_direct_read_seg(inode, file, user_addr, size, + result = nfs_direct_read_seg(inode, ctx, user_addr, size, file_offset, pages, page_count); nfs_free_user_pages(pages, page_count, 1); @@ -239,7 +240,7 @@ nfs_direct_read(struct inode *inode, str * nfs_direct_write_seg - Write out one iov segment. Generate separate * write RPCs for each "wsize" bytes, then commit. * @inode: target inode - * @file: target file (may be NULL) + * @ctx: target file open context * user_addr: starting address of this segment of user's buffer * count: size of this segment * file_offset: offset in file to begin the operation @@ -247,7 +248,7 @@ nfs_direct_read(struct inode *inode, str * nr_pages: size of pages array */ static int -nfs_direct_write_seg(struct inode *inode, struct file *file, +nfs_direct_write_seg(struct inode *inode, struct nfs_open_context *ctx, unsigned long user_addr, size_t count, loff_t file_offset, struct page **pages, int nr_pages) { @@ -257,9 +258,10 @@ nfs_direct_write_seg(struct inode *inode struct nfs_writeverf first_verf; struct nfs_write_data wdata = { .inode = inode, + .cred = ctx->cred, .args = { .fh = NFS_FH(inode), - .lockowner = current->files, + .context = ctx, }, .res = { .fattr = &wdata.fattr, @@ -290,7 +292,7 @@ retry: user_addr + tot_bytes, wdata.args.pgbase, curpage); lock_kernel(); - result = NFS_PROTO(inode)->write(&wdata, file); + result = NFS_PROTO(inode)->write(&wdata); unlock_kernel(); if (result <= 0) { @@ -325,7 +327,7 @@ retry: wdata.args.offset = file_offset; lock_kernel(); - result = NFS_PROTO(inode)->commit(&wdata, file); + result = NFS_PROTO(inode)->commit(&wdata); unlock_kernel(); if (result < 0 || memcmp(&first_verf.verifier, @@ -349,7 +351,7 @@ sync_retry: * nfs_direct_write - For each iov segment, map the user's buffer * then generate write and commit RPCs. * @inode: target inode - * @file: target file (may be NULL) + * @ctx: target file open context * @iov: array of vectors that define I/O buffer * file_offset: offset in file to begin the operation * nr_segs: size of iovec array @@ -358,8 +360,7 @@ sync_retry: * that non-direct readers might access, so they will pick up these * writes immediately. */ -static ssize_t -nfs_direct_write(struct inode *inode, struct file *file, +static int nfs_direct_write(struct inode *inode, struct nfs_open_context *ctx, const struct iovec *iov, loff_t file_offset, unsigned long nr_segs) { @@ -382,7 +383,7 @@ nfs_direct_write(struct inode *inode, st return page_count; } - result = nfs_direct_write_seg(inode, file, user_addr, size, + result = nfs_direct_write_seg(inode, ctx, user_addr, size, file_offset, pages, page_count); nfs_free_user_pages(pages, page_count, 0); @@ -414,6 +415,7 @@ nfs_direct_IO(int rw, struct kiocb *iocb { ssize_t result = -EINVAL; struct file *file = iocb->ki_filp; + struct nfs_open_context *ctx; struct dentry *dentry = file->f_dentry; struct inode *inode = dentry->d_inode; @@ -423,19 +425,20 @@ nfs_direct_IO(int rw, struct kiocb *iocb if (!is_sync_kiocb(iocb)) return result; + ctx = (struct nfs_open_context *)file->private_data; switch (rw) { case READ: dprintk("NFS: direct_IO(read) (%s) off/no(%Lu/%lu)\n", dentry->d_name.name, file_offset, nr_segs); - result = nfs_direct_read(inode, file, iov, + result = nfs_direct_read(inode, ctx, iov, file_offset, nr_segs); break; case WRITE: dprintk("NFS: direct_IO(write) (%s) off/no(%Lu/%lu)\n", dentry->d_name.name, file_offset, nr_segs); - result = nfs_direct_write(inode, file, iov, + result = nfs_direct_write(inode, ctx, iov, file_offset, nr_segs); break; default: @@ -471,6 +474,8 @@ nfs_file_direct_read(struct kiocb *iocb, ssize_t retval = -EINVAL; loff_t *ppos = &iocb->ki_pos; struct file *file = iocb->ki_filp; + struct nfs_open_context *ctx = + (struct nfs_open_context *) file->private_data; struct dentry *dentry = file->f_dentry; struct address_space *mapping = file->f_mapping; struct inode *inode = mapping->host; @@ -502,7 +507,7 @@ nfs_file_direct_read(struct kiocb *iocb, goto out; } - retval = nfs_direct_read(inode, file, &iov, pos, 1); + retval = nfs_direct_read(inode, ctx, &iov, pos, 1); if (retval > 0) *ppos = pos + retval; @@ -542,6 +547,8 @@ nfs_file_direct_write(struct kiocb *iocb loff_t *ppos = &iocb->ki_pos; unsigned long limit = current->rlim[RLIMIT_FSIZE].rlim_cur; struct file *file = iocb->ki_filp; + struct nfs_open_context *ctx = + (struct nfs_open_context *) file->private_data; struct dentry *dentry = file->f_dentry; struct address_space *mapping = file->f_mapping; struct inode *inode = mapping->host; @@ -589,7 +596,7 @@ nfs_file_direct_write(struct kiocb *iocb goto out; } - retval = nfs_direct_write(inode, file, &iov, pos, 1); + retval = nfs_direct_write(inode, ctx, &iov, pos, 1); if (mapping->nrpages) invalidate_inode_pages2(mapping); if (retval > 0) diff -u --recursive --new-file --show-c-function linux-2.6.7/fs/nfs/file.c linux-2.6.7-23-rpc_queue_lock/fs/nfs/file.c --- linux-2.6.7/fs/nfs/file.c 2004-06-16 14:58:08.000000000 -0400 +++ linux-2.6.7-23-rpc_queue_lock/fs/nfs/file.c 2004-06-22 16:49:41.000000000 -0400 @@ -31,6 +31,8 @@ #include #include +#include "delegation.h" + #define NFSDBG_FACILITY NFSDBG_FILE static long nfs_file_fcntl(int fd, unsigned int cmd, @@ -127,6 +129,7 @@ nfs_file_release(struct inode *inode, st static int nfs_file_flush(struct file *file) { + struct nfs_open_context *ctx = (struct nfs_open_context *)file->private_data; struct inode *inode = file->f_dentry->d_inode; int status; @@ -138,9 +141,9 @@ nfs_file_flush(struct file *file) /* Ensure that data+attribute caches are up to date after close() */ status = nfs_wb_all(inode); if (!status) { - status = file->f_error; - file->f_error = 0; - if (!status) + status = ctx->error; + ctx->error = 0; + if (!status && !nfs_have_delegation(inode, FMODE_READ)) __nfs_revalidate_inode(NFS_SERVER(inode), inode); } unlock_kernel(); @@ -211,6 +214,7 @@ nfs_file_mmap(struct file * file, struct static int nfs_fsync(struct file *file, struct dentry *dentry, int datasync) { + struct nfs_open_context *ctx = (struct nfs_open_context *)file->private_data; struct inode *inode = dentry->d_inode; int status; @@ -219,8 +223,8 @@ nfs_fsync(struct file *file, struct dent lock_kernel(); status = nfs_wb_all(inode); if (!status) { - status = file->f_error; - file->f_error = 0; + status = ctx->error; + ctx->error = 0; } unlock_kernel(); return status; @@ -302,6 +306,90 @@ out_swapfile: goto out; } +static int do_getlk(struct file *filp, int cmd, struct file_lock *fl) +{ + struct inode *inode = filp->f_mapping->host; + int status; + + lock_kernel(); + status = NFS_PROTO(inode)->lock(filp, cmd, fl); + unlock_kernel(); + return status; +} + +static int do_unlk(struct file *filp, int cmd, struct file_lock *fl) +{ + struct inode *inode = filp->f_mapping->host; + sigset_t oldset; + int status; + + rpc_clnt_sigmask(NFS_CLIENT(inode), &oldset); + /* + * Flush all pending writes before doing anything + * with locks.. + */ + filemap_fdatawrite(filp->f_mapping); + down(&inode->i_sem); + nfs_wb_all(inode); + up(&inode->i_sem); + filemap_fdatawait(filp->f_mapping); + + /* NOTE: special case + * If we're signalled while cleaning up locks on process exit, we + * still need to complete the unlock. + */ + lock_kernel(); + status = NFS_PROTO(inode)->lock(filp, cmd, fl); + rpc_clnt_sigunmask(NFS_CLIENT(inode), &oldset); + return status; +} + +static int do_setlk(struct file *filp, int cmd, struct file_lock *fl) +{ + struct inode *inode = filp->f_mapping->host; + int status; + + /* + * Flush all pending writes before doing anything + * with locks.. + */ + status = filemap_fdatawrite(filp->f_mapping); + if (status == 0) { + down(&inode->i_sem); + status = nfs_wb_all(inode); + up(&inode->i_sem); + if (status == 0) + status = filemap_fdatawait(filp->f_mapping); + } + if (status < 0) + return status; + + lock_kernel(); + status = NFS_PROTO(inode)->lock(filp, cmd, fl); + /* If we were signalled we still need to ensure that + * we clean up any state on the server. We therefore + * record the lock call as having succeeded in order to + * ensure that locks_remove_posix() cleans it out when + * the process exits. + */ + if (status == -EINTR || status == -ERESTARTSYS) + posix_lock_file(filp, fl); + unlock_kernel(); + if (status < 0) + return status; + /* + * Make sure we clear the cache whenever we try to get the lock. + * This makes locking act as a cache coherency point. + */ + filemap_fdatawrite(filp->f_mapping); + down(&inode->i_sem); + nfs_wb_all(inode); /* we may have slept */ + up(&inode->i_sem); + filemap_fdatawait(filp->f_mapping); + nfs_zap_caches(inode); + return 0; +} + /* * Lock a (portion of) a file */ @@ -309,8 +397,6 @@ int nfs_lock(struct file *filp, int cmd, struct file_lock *fl) { struct inode * inode = filp->f_mapping->host; - int status = 0; - int status2; dprintk("NFS: nfs_lock(f=%s/%ld, t=%x, fl=%x, r=%Ld:%Ld)\n", inode->i_sb->s_id, inode->i_ino, @@ -328,8 +414,8 @@ nfs_lock(struct file *filp, int cmd, str /* Fake OK code if mounted without NLM support */ if (NFS_SERVER(inode)->flags & NFS_MOUNT_NONLM) { if (IS_GETLK(cmd)) - status = LOCK_USE_CLNT; - goto out_ok; + return LOCK_USE_CLNT; + return 0; } } @@ -340,45 +426,12 @@ nfs_lock(struct file *filp, int cmd, str * Not sure whether that would be unique, though, or whether * that would break in other places. */ - if (!fl->fl_owner || !(fl->fl_flags & FL_POSIX)) + if (!(fl->fl_flags & FL_POSIX)) return -ENOLCK; - /* - * Flush all pending writes before doing anything - * with locks.. - */ - status = filemap_fdatawrite(filp->f_mapping); - down(&inode->i_sem); - status2 = nfs_wb_all(inode); - if (!status) - status = status2; - up(&inode->i_sem); - status2 = filemap_fdatawait(filp->f_mapping); - if (!status) - status = status2; - if (status < 0) - return status; - - lock_kernel(); - status = NFS_PROTO(inode)->lock(filp, cmd, fl); - unlock_kernel(); - if (status < 0) - return status; - - status = 0; - - /* - * Make sure we clear the cache whenever we try to get the lock. - * This makes locking act as a cache coherency point. - */ - out_ok: - if ((IS_SETLK(cmd) || IS_SETLKW(cmd)) && fl->fl_type != F_UNLCK) { - filemap_fdatawrite(filp->f_mapping); - down(&inode->i_sem); - nfs_wb_all(inode); /* we may have slept */ - up(&inode->i_sem); - filemap_fdatawait(filp->f_mapping); - nfs_zap_caches(inode); - } - return status; + if (IS_GETLK(cmd)) + return do_getlk(filp, cmd, fl); + if (fl->fl_type == F_UNLCK) + return do_unlk(filp, cmd, fl); + return do_setlk(filp, cmd, fl); } diff -u --recursive --new-file --show-c-function linux-2.6.7/fs/nfs/inode.c linux-2.6.7-23-rpc_queue_lock/fs/nfs/inode.c --- linux-2.6.7/fs/nfs/inode.c 2004-06-16 14:58:18.000000000 -0400 +++ linux-2.6.7-23-rpc_queue_lock/fs/nfs/inode.c 2004-06-22 16:49:41.000000000 -0400 @@ -39,6 +39,8 @@ #include #include +#include "delegation.h" + #define NFSDBG_FACILITY NFSDBG_VFS #define NFS_PARANOIA 1 @@ -123,8 +125,9 @@ nfs_delete_inode(struct inode * inode) { dprintk("NFS: delete_inode(%s/%ld)\n", inode->i_sb->s_id, inode->i_ino); + nfs_wb_all(inode); /* - * The following can never actually happen... + * The following should never happen... */ if (nfs_have_writebacks(inode)) { printk(KERN_ERR "nfs_delete_inode: inode %ld has pending RPC requests\n", inode->i_ino); @@ -141,10 +144,10 @@ static void nfs_clear_inode(struct inode *inode) { struct nfs_inode *nfsi = NFS_I(inode); - struct rpc_cred *cred = nfsi->mm_cred; + struct rpc_cred *cred; - if (cred) - put_rpccred(cred); + nfs_wb_all(inode); + BUG_ON (!list_empty(&nfsi->open_files)); cred = nfsi->cache_access.cred; if (cred) put_rpccred(cred); @@ -859,53 +862,114 @@ int nfs_getattr(struct vfsmount *mnt, st return err; } +struct nfs_open_context *alloc_nfs_open_context(struct dentry *dentry, struct rpc_cred *cred) +{ + struct nfs_open_context *ctx; + + ctx = (struct nfs_open_context *)kmalloc(sizeof(*ctx), GFP_KERNEL); + if (ctx != NULL) { + atomic_set(&ctx->count, 1); + ctx->dentry = dget(dentry); + ctx->cred = get_rpccred(cred); + ctx->state = NULL; + ctx->pid = current->tgid; + ctx->error = 0; + init_waitqueue_head(&ctx->waitq); + } + return ctx; +} + +struct nfs_open_context *get_nfs_open_context(struct nfs_open_context *ctx) +{ + if (ctx != NULL) + atomic_inc(&ctx->count); + return ctx; +} + +void put_nfs_open_context(struct nfs_open_context *ctx) +{ + if (atomic_dec_and_test(&ctx->count)) { + if (ctx->state != NULL) + nfs4_close_state(ctx->state, ctx->mode); + if (ctx->cred != NULL) + put_rpccred(ctx->cred); + dput(ctx->dentry); + kfree(ctx); + } +} + /* * Ensure that mmap has a recent RPC credential for use when writing out * shared pages */ -void -nfs_set_mmcred(struct inode *inode, struct rpc_cred *cred) +void nfs_file_set_open_context(struct file *filp, struct nfs_open_context *ctx) +{ + struct inode *inode = filp->f_dentry->d_inode; + struct nfs_inode *nfsi = NFS_I(inode); + + filp->private_data = get_nfs_open_context(ctx); + spin_lock(&inode->i_lock); + list_add(&ctx->list, &nfsi->open_files); + spin_unlock(&inode->i_lock); +} + +struct nfs_open_context *nfs_find_open_context(struct inode *inode, int mode) +{ + struct nfs_inode *nfsi = NFS_I(inode); + struct nfs_open_context *pos, *ctx = NULL; + + spin_lock(&inode->i_lock); + list_for_each_entry(pos, &nfsi->open_files, list) { + if ((pos->mode & mode) == mode) { + ctx = get_nfs_open_context(pos); + break; + } + } + spin_unlock(&inode->i_lock); + return ctx; +} + +void nfs_file_clear_open_context(struct file *filp) { - struct rpc_cred **p = &NFS_I(inode)->mm_cred, - *oldcred = *p; + struct inode *inode = filp->f_dentry->d_inode; + struct nfs_open_context *ctx = (struct nfs_open_context *)filp->private_data; - *p = get_rpccred(cred); - if (oldcred) - put_rpccred(oldcred); + if (ctx) { + filp->private_data = NULL; + spin_lock(&inode->i_lock); + list_del(&ctx->list); + spin_unlock(&inode->i_lock); + put_nfs_open_context(ctx); + } } /* - * These are probably going to contain hooks for - * allocating and releasing RPC credentials for - * the file. I'll have to think about Tronds patch - * a bit more.. + * These allocate and release file read/write context information. */ int nfs_open(struct inode *inode, struct file *filp) { - struct rpc_auth *auth; + struct nfs_open_context *ctx; struct rpc_cred *cred; - auth = NFS_CLIENT(inode)->cl_auth; - cred = rpcauth_lookupcred(auth, 0); - filp->private_data = cred; - if ((filp->f_mode & FMODE_WRITE) != 0) { - nfs_set_mmcred(inode, cred); + if ((cred = rpcauth_lookupcred(NFS_CLIENT(inode)->cl_auth, 0)) == NULL) + return -ENOMEM; + ctx = alloc_nfs_open_context(filp->f_dentry, cred); + put_rpccred(cred); + if (ctx == NULL) + return -ENOMEM; + ctx->mode = filp->f_mode; + nfs_file_set_open_context(filp, ctx); + put_nfs_open_context(ctx); + if ((filp->f_mode & FMODE_WRITE) != 0) nfs_begin_data_update(inode); - } return 0; } int nfs_release(struct inode *inode, struct file *filp) { - struct rpc_cred *cred; - - lock_kernel(); if ((filp->f_mode & FMODE_WRITE) != 0) nfs_end_data_update(inode); - cred = nfs_file_cred(filp); - if (cred) - put_rpccred(cred); - unlock_kernel(); + nfs_file_clear_open_context(filp); return 0; } @@ -1002,6 +1066,30 @@ out: return status; } +int nfs_attribute_timeout(struct inode *inode) +{ + struct nfs_inode *nfsi = NFS_I(inode); + + if (nfs_have_delegation(inode, FMODE_READ)) + return 0; + return time_after(jiffies, nfsi->read_cache_jiffies+nfsi->attrtimeo); +} + +/** + * nfs_revalidate_inode - Revalidate the inode attributes + * @server - pointer to nfs_server struct + * @inode - pointer to inode struct + * + * Updates inode attribute information by retrieving the data from the server. + */ +int nfs_revalidate_inode(struct nfs_server *server, struct inode *inode) +{ + if (!(NFS_FLAGS(inode) & (NFS_INO_INVALID_ATTR|NFS_INO_INVALID_DATA)) + && !nfs_attribute_timeout(inode)) + return NFS_STALE(inode) ? -ESTALE : 0; + return __nfs_revalidate_inode(server, inode); +} + /** * nfs_begin_data_update * @inode - pointer to inode @@ -1023,11 +1111,13 @@ void nfs_end_data_update(struct inode *i { struct nfs_inode *nfsi = NFS_I(inode); - /* Mark the attribute cache for revalidation */ - nfsi->flags |= NFS_INO_INVALID_ATTR; - /* Directories and symlinks: invalidate page cache too */ - if (S_ISDIR(inode->i_mode) || S_ISLNK(inode->i_mode)) - nfsi->flags |= NFS_INO_INVALID_DATA; + if (!nfs_have_delegation(inode, FMODE_READ)) { + /* Mark the attribute cache for revalidation */ + nfsi->flags |= NFS_INO_INVALID_ATTR; + /* Directories and symlinks: invalidate page cache too */ + if (S_ISDIR(inode->i_mode) || S_ISLNK(inode->i_mode)) + nfsi->flags |= NFS_INO_INVALID_DATA; + } nfsi->cache_change_attribute ++; atomic_dec(&nfsi->data_updates); } @@ -1068,6 +1158,10 @@ int nfs_refresh_inode(struct inode *inod loff_t cur_size, new_isize; int data_unstable; + /* Do we hold a delegation? */ + if (nfs_have_delegation(inode, FMODE_READ)) + return 0; + /* Are we in the process of updating data on the server? */ data_unstable = nfs_caches_unstable(inode); @@ -1265,7 +1359,8 @@ static int nfs_update_inode(struct inode if (!(S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) || S_ISLNK(inode->i_mode))) invalid &= ~NFS_INO_INVALID_DATA; - nfsi->flags |= invalid; + if (!nfs_have_delegation(inode, FMODE_READ)) + nfsi->flags |= invalid; return 0; out_changed: @@ -1402,6 +1497,7 @@ static struct file_system_type nfs_fs_ty static void nfs4_clear_inode(struct inode *); + static struct super_operations nfs4_sops = { .alloc_inode = nfs_alloc_inode, .destroy_inode = nfs_destroy_inode, @@ -1423,6 +1519,12 @@ static void nfs4_clear_inode(struct inod { struct nfs_inode *nfsi = NFS_I(inode); + /* If we are holding a delegation, return it! */ + if (nfsi->delegation != NULL) + nfs_inode_return_delegation(inode); + /* First call standard NFS clear_inode() code */ + nfs_clear_inode(inode); + /* Now clear out any remaining state */ while (!list_empty(&nfsi->open_states)) { struct nfs4_state *state; @@ -1437,8 +1539,6 @@ static void nfs4_clear_inode(struct inod BUG_ON(atomic_read(&state->count) != 1); nfs4_close_state(state, state->state); } - /* Now call standard NFS clear_inode() code */ - nfs_clear_inode(inode); } @@ -1536,8 +1636,13 @@ static int nfs4_fill_super(struct super_ memcpy(clp->cl_ipaddr, server->ip_addr, sizeof(clp->cl_ipaddr)); nfs_idmap_new(clp); } - if (list_empty(&clp->cl_superblocks)) - clear_bit(NFS4CLNT_OK, &clp->cl_state); + 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)) @@ -1709,22 +1814,31 @@ out_free: return s; } +static void nfs4_kill_super(struct super_block *sb) +{ + nfs_return_all_delegations(sb); + nfs_kill_super(sb); +} + static struct file_system_type nfs4_fs_type = { .owner = THIS_MODULE, .name = "nfs4", .get_sb = nfs4_get_sb, - .kill_sb = nfs_kill_super, + .kill_sb = nfs4_kill_super, .fs_flags = FS_ODD_RENAME|FS_REVAL_DOT|FS_BINARY_MOUNTDATA, }; -#define nfs4_zero_state(nfsi) \ +#define nfs4_init_once(nfsi) \ do { \ INIT_LIST_HEAD(&(nfsi)->open_states); \ + nfsi->delegation = NULL; \ + 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) #else -#define nfs4_zero_state(nfsi) \ +#define nfs4_init_once(nfsi) \ do { } while (0) #define register_nfs4fs() (0) #define unregister_nfs4fs() @@ -1746,8 +1860,6 @@ static struct inode *nfs_alloc_inode(str if (!nfsi) return NULL; nfsi->flags = 0; - nfsi->mm_cred = NULL; - nfs4_zero_state(nfsi); return &nfsi->vfs_inode; } @@ -1765,12 +1877,14 @@ static void init_once(void * foo, kmem_c inode_init_once(&nfsi->vfs_inode); INIT_LIST_HEAD(&nfsi->dirty); INIT_LIST_HEAD(&nfsi->commit); + INIT_LIST_HEAD(&nfsi->open_files); INIT_RADIX_TREE(&nfsi->nfs_page_tree, GFP_ATOMIC); atomic_set(&nfsi->data_updates, 0); nfsi->ndirty = 0; nfsi->ncommit = 0; nfsi->npages = 0; init_waitqueue_head(&nfsi->nfs_i_wait); + nfs4_init_once(nfsi); } } diff -u --recursive --new-file --show-c-function linux-2.6.7/fs/nfs/Makefile linux-2.6.7-23-rpc_queue_lock/fs/nfs/Makefile --- linux-2.6.7/fs/nfs/Makefile 2004-06-16 14:57:53.000000000 -0400 +++ linux-2.6.7-23-rpc_queue_lock/fs/nfs/Makefile 2004-06-22 16:49:06.000000000 -0400 @@ -9,6 +9,7 @@ nfs-y := dir.o file.o inode.o nfs2xdr nfs-$(CONFIG_ROOT_NFS) += nfsroot.o mount_clnt.o nfs-$(CONFIG_NFS_V3) += nfs3proc.o nfs3xdr.o nfs-$(CONFIG_NFS_V4) += nfs4proc.o nfs4xdr.o nfs4state.o nfs4renewd.o \ - idmap.o + delegation.o idmap.o \ + callback.o callback_xdr.o callback_proc.o nfs-$(CONFIG_NFS_DIRECTIO) += direct.o nfs-objs := $(nfs-y) diff -u --recursive --new-file --show-c-function linux-2.6.7/fs/nfs/nfs3proc.c linux-2.6.7-23-rpc_queue_lock/fs/nfs/nfs3proc.c --- linux-2.6.7/fs/nfs/nfs3proc.c 2004-06-16 14:58:05.000000000 -0400 +++ linux-2.6.7-23-rpc_queue_lock/fs/nfs/nfs3proc.c 2004-06-22 16:48:51.000000000 -0400 @@ -68,18 +68,6 @@ nfs3_async_handle_jukebox(struct rpc_tas return 1; } -static struct rpc_cred * -nfs_cred(struct inode *inode, struct file *filp) -{ - struct rpc_cred *cred = NULL; - - if (filp) - cred = (struct rpc_cred *)filp->private_data; - if (!cred) - cred = NFS_I(inode)->mm_cred; - return cred; -} - /* * Bare-bones access to getattr: this is for nfs_read_super. */ @@ -164,8 +152,7 @@ nfs3_proc_lookup(struct inode *dir, stru return status; } -static int -nfs3_proc_access(struct inode *inode, struct rpc_cred *cred, int mode) +static int nfs3_proc_access(struct inode *inode, struct nfs_access_entry *entry) { struct nfs_fattr fattr; struct nfs3_accessargs arg = { @@ -178,9 +165,10 @@ nfs3_proc_access(struct inode *inode, st .rpc_proc = &nfs3_procedures[NFS3PROC_ACCESS], .rpc_argp = &arg, .rpc_resp = &res, - .rpc_cred = cred + .rpc_cred = entry->cred }; - int status; + int mode = entry->mask; + int status; dprintk("NFS call access\n"); fattr.valid = 0; @@ -200,10 +188,16 @@ nfs3_proc_access(struct inode *inode, st } status = rpc_call_sync(NFS_CLIENT(inode), &msg, 0); nfs_refresh_inode(inode, &fattr); - dprintk("NFS reply access\n"); - - if (status == 0 && (arg.access & res.access) != arg.access) - status = -EACCES; + if (status == 0) { + entry->mask = 0; + if (res.access & NFS3_ACCESS_READ) + entry->mask |= MAY_READ; + if (res.access & (NFS3_ACCESS_MODIFY | NFS3_ACCESS_EXTEND | NFS3_ACCESS_DELETE)) + entry->mask |= MAY_WRITE; + if (res.access & (NFS3_ACCESS_LOOKUP|NFS3_ACCESS_EXECUTE)) + entry->mask |= MAY_EXEC; + } + dprintk("NFS reply access, status = %d\n", status); return status; } @@ -227,8 +221,7 @@ nfs3_proc_readlink(struct inode *inode, return status; } -static int -nfs3_proc_read(struct nfs_read_data *rdata, struct file *filp) +static int nfs3_proc_read(struct nfs_read_data *rdata) { int flags = rdata->flags; struct inode * inode = rdata->inode; @@ -237,13 +230,13 @@ nfs3_proc_read(struct nfs_read_data *rda .rpc_proc = &nfs3_procedures[NFS3PROC_READ], .rpc_argp = &rdata->args, .rpc_resp = &rdata->res, + .rpc_cred = rdata->cred, }; int status; dprintk("NFS call read %d @ %Ld\n", rdata->args.count, (long long) rdata->args.offset); fattr->valid = 0; - msg.rpc_cred = nfs_cred(inode, filp); status = rpc_call_sync(NFS_CLIENT(inode), &msg, flags); if (status >= 0) nfs_refresh_inode(inode, fattr); @@ -251,8 +244,7 @@ nfs3_proc_read(struct nfs_read_data *rda return status; } -static int -nfs3_proc_write(struct nfs_write_data *wdata, struct file *filp) +static int nfs3_proc_write(struct nfs_write_data *wdata) { int rpcflags = wdata->flags; struct inode * inode = wdata->inode; @@ -261,13 +253,13 @@ nfs3_proc_write(struct nfs_write_data *w .rpc_proc = &nfs3_procedures[NFS3PROC_WRITE], .rpc_argp = &wdata->args, .rpc_resp = &wdata->res, + .rpc_cred = wdata->cred, }; int status; dprintk("NFS call write %d @ %Ld\n", wdata->args.count, (long long) wdata->args.offset); fattr->valid = 0; - msg.rpc_cred = nfs_cred(inode, filp); status = rpc_call_sync(NFS_CLIENT(inode), &msg, rpcflags); if (status >= 0) nfs_refresh_inode(inode, fattr); @@ -275,8 +267,7 @@ nfs3_proc_write(struct nfs_write_data *w return status < 0? status : wdata->res.count; } -static int -nfs3_proc_commit(struct nfs_write_data *cdata, struct file *filp) +static int nfs3_proc_commit(struct nfs_write_data *cdata) { struct inode * inode = cdata->inode; struct nfs_fattr * fattr = cdata->res.fattr; @@ -284,13 +275,13 @@ nfs3_proc_commit(struct nfs_write_data * .rpc_proc = &nfs3_procedures[NFS3PROC_COMMIT], .rpc_argp = &cdata->args, .rpc_resp = &cdata->res, + .rpc_cred = cdata->cred, }; int status; dprintk("NFS call commit %d @ %Ld\n", cdata->args.count, (long long) cdata->args.offset); fattr->valid = 0; - msg.rpc_cred = nfs_cred(inode, filp); status = rpc_call_sync(NFS_CLIENT(inode), &msg, 0); if (status >= 0) nfs_refresh_inode(inode, fattr); @@ -534,6 +525,8 @@ nfs3_proc_symlink(struct inode *dir, str }; int status; + if (path->len > NFS3_MAXPATHLEN) + return -ENAMETOOLONG; dprintk("NFS call symlink %s -> %s\n", name->name, path->name); dir_attr.valid = 0; fattr->valid = 0; @@ -832,27 +825,6 @@ nfs3_proc_commit_setup(struct nfs_write_ rpc_call_setup(task, &msg, 0); } -/* - * Set up the nfspage struct with the right credentials - */ -void -nfs3_request_init(struct nfs_page *req, struct file *filp) -{ - req->wb_cred = get_rpccred(nfs_cred(req->wb_inode, filp)); -} - -static int -nfs3_request_compatible(struct nfs_page *req, struct file *filp, struct page *page) -{ - if (req->wb_file != filp) - return 0; - if (req->wb_page != page) - return 0; - if (req->wb_cred != nfs_file_cred(filp)) - return 0; - return 1; -} - static int nfs3_proc_lock(struct file *filp, int cmd, struct file_lock *fl) { @@ -892,7 +864,5 @@ struct nfs_rpc_ops nfs_v3_clientops = { .commit_setup = nfs3_proc_commit_setup, .file_open = nfs_open, .file_release = nfs_release, - .request_init = nfs3_request_init, - .request_compatible = nfs3_request_compatible, .lock = nfs3_proc_lock, }; diff -u --recursive --new-file --show-c-function linux-2.6.7/fs/nfs/nfs4proc.c linux-2.6.7-23-rpc_queue_lock/fs/nfs/nfs4proc.c --- linux-2.6.7/fs/nfs/nfs4proc.c 2004-06-16 14:58:17.000000000 -0400 +++ linux-2.6.7-23-rpc_queue_lock/fs/nfs/nfs4proc.c 2004-06-22 16:49:34.000000000 -0400 @@ -47,12 +47,16 @@ #include #include +#include "delegation.h" + #define NFSDBG_FACILITY NFSDBG_PROC -#define NFS4_POLL_RETRY_TIME (15*HZ) +#define NFS4_POLL_RETRY_MIN (1*HZ) +#define NFS4_POLL_RETRY_MAX (15*HZ) static int nfs4_do_fsinfo(struct nfs_server *, struct nfs_fh *, struct nfs_fsinfo *); static int nfs4_async_handle_error(struct rpc_task *, struct nfs_server *); +static int _nfs4_proc_access(struct inode *inode, struct nfs_access_entry *entry); extern u32 *nfs4_decode_dirent(u32 *p, struct nfs_entry *entry, int plus); extern struct rpc_procinfo nfs4_procedures[]; @@ -189,53 +193,296 @@ static void update_changeattr(struct ino * reclaim state on the server after a reboot. * Assumes caller is holding the sp->so_sem */ -int -nfs4_open_reclaim(struct nfs4_state_owner *sp, struct nfs4_state *state) +static int _nfs4_open_reclaim(struct nfs4_state_owner *sp, struct nfs4_state *state) { struct inode *inode = state->inode; struct nfs_server *server = NFS_SERVER(inode); - struct nfs_fattr fattr = { - .valid = 0, - }; - struct nfs_open_reclaimargs o_arg = { + struct nfs_delegation *delegation = NFS_I(inode)->delegation; + struct nfs_openargs o_arg = { .fh = NFS_FH(inode), .seqid = sp->so_seqid, .id = sp->so_id, - .share_access = state->state, + .open_flags = state->state, .clientid = server->nfs4_state->cl_clientid, .claim = NFS4_OPEN_CLAIM_PREVIOUS, .bitmask = server->attr_bitmask, }; struct nfs_openres o_res = { - .f_attr = &fattr, .server = server, /* Grrr */ }; struct rpc_message msg = { - .rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OPEN_RECLAIM], + .rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OPEN_NOATTR], .rpc_argp = &o_arg, .rpc_resp = &o_res, .rpc_cred = sp->so_cred, }; int status; + if (delegation != NULL) { + if (!(delegation->flags & NFS_DELEGATION_NEED_RECLAIM)) { + memcpy(&state->stateid, &delegation->stateid, + sizeof(state->stateid)); + set_bit(NFS_DELEGATED_STATE, &state->flags); + return 0; + } + o_arg.u.delegation_type = delegation->type; + } status = rpc_call_sync(server->client, &msg, 0); nfs4_increment_seqid(status, sp); - if (status == 0) + if (status == 0) { memcpy(&state->stateid, &o_res.stateid, sizeof(state->stateid)); - /* Update the inode attributes */ - nfs_refresh_inode(inode, &fattr); + if (o_res.delegation_type != 0) { + nfs_inode_reclaim_delegation(inode, sp->so_cred, &o_res); + /* Did the server issue an immediate delegation recall? */ + if (o_res.do_recall) + nfs_async_inode_return_delegation(inode, &o_res.stateid); + } + } + clear_bit(NFS_DELEGATED_STATE, &state->flags); + /* Ensure we update the inode attributes */ + NFS_CACHEINV(inode); + return status; +} + +int nfs4_open_reclaim(struct nfs4_state_owner *sp, struct nfs4_state *state) +{ + struct nfs_server *server = NFS_SERVER(state->inode); + struct nfs4_exception exception = { }; + int err; + do { + err = _nfs4_open_reclaim(sp, state); + switch (err) { + case 0: + case -NFS4ERR_STALE_CLIENTID: + case -NFS4ERR_STALE_STATEID: + case -NFS4ERR_EXPIRED: + return err; + } + err = nfs4_handle_exception(server, err, &exception); + } while (exception.retry); + return err; +} + +static int _nfs4_open_delegation_recall(struct dentry *dentry, struct nfs4_state *state) +{ + struct nfs4_state_owner *sp = state->owner; + struct inode *inode = dentry->d_inode; + struct nfs_server *server = NFS_SERVER(inode); + struct dentry *parent = dget_parent(dentry); + struct nfs_openargs arg = { + .fh = NFS_FH(parent->d_inode), + .clientid = server->nfs4_state->cl_clientid, + .name = &dentry->d_name, + .id = sp->so_id, + .server = server, + .bitmask = server->attr_bitmask, + .claim = NFS4_OPEN_CLAIM_DELEGATE_CUR, + }; + struct nfs_openres res = { + .server = server, + }; + struct rpc_message msg = { + .rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OPEN_NOATTR], + .rpc_argp = &arg, + .rpc_resp = &res, + .rpc_cred = sp->so_cred, + }; + int status = 0; + + down(&sp->so_sema); + if (!test_bit(NFS_DELEGATED_STATE, &state->flags)) + goto out; + if (state->state == 0) + goto out; + arg.seqid = sp->so_seqid; + arg.open_flags = state->state; + memcpy(arg.u.delegation.data, state->stateid.data, sizeof(arg.u.delegation.data)); + status = rpc_call_sync(server->client, &msg, 0); + nfs4_increment_seqid(status, sp); + if (status >= 0) { + memcpy(state->stateid.data, res.stateid.data, + sizeof(state->stateid.data)); + clear_bit(NFS_DELEGATED_STATE, &state->flags); + } +out: + up(&sp->so_sema); + dput(parent); + return status; +} + +int nfs4_open_delegation_recall(struct dentry *dentry, struct nfs4_state *state) +{ + struct nfs4_exception exception = { }; + struct nfs_server *server = NFS_SERVER(dentry->d_inode); + int err; + do { + err = _nfs4_open_delegation_recall(dentry, state); + switch (err) { + case 0: + return err; + case -NFS4ERR_STALE_CLIENTID: + case -NFS4ERR_STALE_STATEID: + case -NFS4ERR_EXPIRED: + /* Don't recall a delegation if it was lost */ + nfs4_schedule_state_recovery(server->nfs4_state); + return err; + } + err = nfs4_handle_exception(server, err, &exception); + } while (exception.retry); + return err; +} + +static int _nfs4_proc_open_confirm(struct rpc_clnt *clnt, const struct nfs_fh *fh, struct nfs4_state_owner *sp, nfs4_stateid *stateid) +{ + struct nfs_open_confirmargs arg = { + .fh = fh, + .seqid = sp->so_seqid, + .stateid = *stateid, + }; + struct nfs_open_confirmres res; + struct rpc_message msg = { + .rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OPEN_CONFIRM], + .rpc_argp = &arg, + .rpc_resp = &res, + .rpc_cred = sp->so_cred, + }; + int status; + + status = rpc_call_sync(clnt, &msg, 0); + nfs4_increment_seqid(status, sp); + if (status >= 0) + memcpy(stateid, &res.stateid, sizeof(*stateid)); return status; } +static int _nfs4_do_access(struct inode *inode, struct rpc_cred *cred, int mask) +{ + struct nfs_access_entry cache; + int status; + + status = nfs_access_get_cached(inode, cred, &cache); + if (status == 0) + goto out; + + /* Be clever: ask server to check for all possible rights */ + cache.mask = MAY_EXEC | MAY_WRITE | MAY_READ; + cache.cred = cred; + cache.jiffies = jiffies; + status = _nfs4_proc_access(inode, &cache); + if (status != 0) + return status; + nfs_access_add_cache(inode, &cache); +out: + if ((cache.mask & mask) == mask) + return 0; + return -EACCES; +} + +/* + * Returns an nfs4_state + an extra reference to the inode + */ +int _nfs4_open_delegated(struct inode *inode, int flags, struct rpc_cred *cred, struct nfs4_state **res) +{ + struct nfs_delegation *delegation; + struct nfs_server *server = NFS_SERVER(inode); + struct nfs4_client *clp = server->nfs4_state; + struct nfs_inode *nfsi = NFS_I(inode); + struct nfs4_state_owner *sp = NULL; + struct nfs4_state *state = NULL; + int open_flags = flags & (FMODE_READ|FMODE_WRITE); + int mask = 0; + int err; + + /* Protect against reboot recovery - NOTE ORDER! */ + down_read(&clp->cl_sem); + /* Protect against delegation recall */ + down_read(&nfsi->rwsem); + delegation = NFS_I(inode)->delegation; + err = -ENOENT; + if (delegation == NULL || (delegation->type & open_flags) != open_flags) + goto out_err; + err = -ENOMEM; + if (!(sp = nfs4_get_state_owner(server, cred))) { + dprintk("%s: nfs4_get_state_owner failed!\n", __FUNCTION__); + goto out_err; + } + down(&sp->so_sema); + state = nfs4_get_open_state(inode, sp); + if (state == NULL) + goto out_err; + + err = -ENOENT; + if ((state->state & open_flags) == open_flags) { + spin_lock(&inode->i_lock); + if (open_flags & FMODE_READ) + state->nreaders++; + if (open_flags & FMODE_WRITE) + state->nwriters++; + spin_unlock(&inode->i_lock); + goto out_ok; + } else if (state->state != 0) + goto out_err; + + lock_kernel(); + err = _nfs4_do_access(inode, cred, mask); + unlock_kernel(); + if (err != 0) + goto out_err; + spin_lock(&inode->i_lock); + memcpy(state->stateid.data, delegation->stateid.data, + sizeof(state->stateid.data)); + state->state |= open_flags; + if (open_flags & FMODE_READ) + state->nreaders++; + if (open_flags & FMODE_WRITE) + state->nwriters++; + set_bit(NFS_DELEGATED_STATE, &state->flags); + spin_unlock(&inode->i_lock); +out_ok: + up(&sp->so_sema); + nfs4_put_state_owner(sp); + up_read(&nfsi->rwsem); + up_read(&clp->cl_sem); + igrab(inode); + *res = state; + return 0; +out_err: + if (sp != NULL) { + if (state != NULL) + nfs4_put_open_state(state); + up(&sp->so_sema); + nfs4_put_state_owner(sp); + } + up_read(&nfsi->rwsem); + up_read(&clp->cl_sem); + return err; +} + +static struct nfs4_state *nfs4_open_delegated(struct inode *inode, int flags, struct rpc_cred *cred) +{ + struct nfs4_exception exception = { }; + struct nfs4_state *res; + int err; + + do { + err = _nfs4_open_delegated(inode, flags, cred, &res); + if (err == 0) + break; + res = ERR_PTR(nfs4_handle_exception(NFS_SERVER(inode), + err, &exception)); + } while (exception.retry); + return res; +} + /* * Returns an nfs4_state + an referenced inode */ -struct nfs4_state * -nfs4_do_open(struct inode *dir, struct qstr *name, int flags, struct iattr *sattr, struct rpc_cred *cred) +static int _nfs4_do_open(struct inode *dir, struct qstr *name, int flags, struct iattr *sattr, struct rpc_cred *cred, struct nfs4_state **res) { struct nfs4_state_owner *sp; struct nfs4_state *state = NULL; struct nfs_server *server = NFS_SERVER(dir); + struct nfs4_client *clp = server->nfs4_state; struct inode *inode = NULL; int status; struct nfs_fattr f_attr = { @@ -243,12 +490,11 @@ nfs4_do_open(struct inode *dir, struct q }; struct nfs_openargs o_arg = { .fh = NFS_FH(dir), - .share_access = flags & (FMODE_READ|FMODE_WRITE), - .opentype = (flags & O_CREAT) ? NFS4_OPEN_CREATE : NFS4_OPEN_NOCREATE, - .createmode = (flags & O_EXCL) ? NFS4_CREATE_EXCLUSIVE : NFS4_CREATE_UNCHECKED, + .open_flags = flags, .name = name, .server = server, .bitmask = server->attr_bitmask, + .claim = NFS4_OPEN_CLAIM_NULL, }; struct nfs_openres o_res = { .f_attr = &f_attr, @@ -261,60 +507,44 @@ nfs4_do_open(struct inode *dir, struct q .rpc_cred = cred, }; -retry: + /* Protect against reboot recovery conflicts */ + down_read(&clp->cl_sem); status = -ENOMEM; - if (!(sp = nfs4_get_state_owner(NFS_SERVER(dir), cred))) { + if (!(sp = nfs4_get_state_owner(server, cred))) { dprintk("nfs4_do_open: nfs4_get_state_owner failed!\n"); - goto out; + goto out_err; } - if (o_arg.createmode & NFS4_CREATE_EXCLUSIVE){ + if (flags & O_EXCL) { u32 *p = (u32 *) o_arg.u.verifier.data; p[0] = jiffies; p[1] = current->pid; - } else if (o_arg.createmode == NFS4_CREATE_UNCHECKED) { + } else o_arg.u.attrs = sattr; - } /* Serialization for the sequence id */ down(&sp->so_sema); o_arg.seqid = sp->so_seqid; o_arg.id = sp->so_id; - o_arg.clientid = NFS_SERVER(dir)->nfs4_state->cl_clientid, + o_arg.clientid = clp->cl_clientid; status = rpc_call_sync(server->client, &msg, 0); nfs4_increment_seqid(status, sp); if (status) - goto out_up; + goto out_err; update_changeattr(dir, &o_res.cinfo); + if(o_res.rflags & NFS4_OPEN_RESULT_CONFIRM) { + status = _nfs4_proc_open_confirm(server->client, &o_res.fh, sp, &o_res.stateid); + if (status) + goto out_err; + } status = -ENOMEM; inode = nfs_fhget(dir->i_sb, &o_res.fh, &f_attr); if (!inode) - goto out_up; + goto out_err; state = nfs4_get_open_state(inode, sp); if (!state) - goto out_up; - - if(o_res.rflags & NFS4_OPEN_RESULT_CONFIRM) { - struct nfs_open_confirmargs oc_arg = { - .fh = &o_res.fh, - .seqid = sp->so_seqid, - }; - struct nfs_open_confirmres oc_res; - struct rpc_message msg = { - .rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OPEN_CONFIRM], - .rpc_argp = &oc_arg, - .rpc_resp = &oc_res, - .rpc_cred = cred, - }; - - memcpy(&oc_arg.stateid, &o_res.stateid, sizeof(oc_arg.stateid)); - status = rpc_call_sync(server->client, &msg, 0); - nfs4_increment_seqid(status, sp); - if (status) - goto out_up; - memcpy(&state->stateid, &oc_res.stateid, sizeof(state->stateid)); - } else - memcpy(&state->stateid, &o_res.stateid, sizeof(state->stateid)); + goto out_err; + memcpy(&state->stateid, &o_res.stateid, sizeof(state->stateid)); spin_lock(&inode->i_lock); if (flags & FMODE_READ) state->nreaders++; @@ -322,47 +552,62 @@ retry: state->nwriters++; state->state |= flags & (FMODE_READ|FMODE_WRITE); spin_unlock(&inode->i_lock); - + if (o_res.delegation_type != 0) + nfs_inode_set_delegation(inode, cred, &o_res); up(&sp->so_sema); nfs4_put_state_owner(sp); - return state; - -out_up: - up(&sp->so_sema); - nfs4_put_state_owner(sp); - if (state) { - nfs4_put_open_state(state); - state = NULL; - } - if (inode) { + up_read(&clp->cl_sem); + *res = state; + return 0; +out_err: + if (sp != NULL) { + if (state != NULL) + nfs4_put_open_state(state); + up(&sp->so_sema); + nfs4_put_state_owner(sp); + } + /* Note: clp->cl_sem must be released before nfs4_put_open_state()! */ + up_read(&clp->cl_sem); + if (inode != NULL) iput(inode); - inode = NULL; - } - /* NOTE: BAD_SEQID means the server and client disagree about the - * book-keeping w.r.t. state-changing operations - * (OPEN/CLOSE/LOCK/LOCKU...) - * It is actually a sign of a bug on the client or on the server. - * - * If we receive a BAD_SEQID error in the particular case of - * doing an OPEN, we assume that nfs4_increment_seqid() will - * have unhashed the old state_owner for us, and that we can - * therefore safely retry using a new one. We should still warn - * the user though... - */ - if (status == -NFS4ERR_BAD_SEQID) { - printk(KERN_WARNING "NFS: v4 server returned a bad sequence-id error!\n"); - goto retry; - } - status = nfs4_handle_error(server, status); - if (!status) - goto retry; - BUG_ON(status < -1000 || status > 0); -out: - return ERR_PTR(status); + *res = NULL; + return status; } -int -nfs4_do_setattr(struct nfs_server *server, struct nfs_fattr *fattr, + +struct nfs4_state *nfs4_do_open(struct inode *dir, struct qstr *name, int flags, struct iattr *sattr, struct rpc_cred *cred) +{ + struct nfs4_exception exception = { }; + struct nfs4_state *res; + int status; + + do { + status = _nfs4_do_open(dir, name, flags, sattr, cred, &res); + if (status == 0) + break; + /* NOTE: BAD_SEQID means the server and client disagree about the + * book-keeping w.r.t. state-changing operations + * (OPEN/CLOSE/LOCK/LOCKU...) + * It is actually a sign of a bug on the client or on the server. + * + * If we receive a BAD_SEQID error in the particular case of + * doing an OPEN, we assume that nfs4_increment_seqid() will + * have unhashed the old state_owner for us, and that we can + * therefore safely retry using a new one. We should still warn + * the user though... + */ + if (status == -NFS4ERR_BAD_SEQID) { + printk(KERN_WARNING "NFS: v4 server returned a bad sequence-id error!\n"); + exception.retry = 1; + continue; + } + res = ERR_PTR(nfs4_handle_exception(NFS_SERVER(dir), + status, &exception)); + } while (exception.retry); + return res; +} + +static int _nfs4_do_setattr(struct nfs_server *server, struct nfs_fattr *fattr, struct nfs_fh *fhandle, struct iattr *sattr, struct nfs4_state *state) { @@ -381,9 +626,7 @@ nfs4_do_setattr(struct nfs_server *serve .rpc_argp = &arg, .rpc_resp = &res, }; - int status; -retry: fattr->valid = 0; if (sattr->ia_valid & ATTR_SIZE) @@ -391,13 +634,22 @@ retry: else memcpy(&arg.stateid, &zero_stateid, sizeof(arg.stateid)); - status = rpc_call_sync(server->client, &msg, 0); - if (status) { - status = nfs4_handle_error(server, status); - if (!status) - goto retry; - } - return status; + return rpc_call_sync(server->client, &msg, 0); +} + +int nfs4_do_setattr(struct nfs_server *server, struct nfs_fattr *fattr, + struct nfs_fh *fhandle, struct iattr *sattr, + struct nfs4_state *state) +{ + struct nfs4_exception exception = { }; + int err; + do { + err = nfs4_handle_exception(server, + _nfs4_do_setattr(server, fattr, fhandle, sattr, + state), + &exception); + } while (exception.retry); + return err; } /* @@ -411,8 +663,7 @@ retry: * * NOTE: Caller must be holding the sp->so_owner semaphore! */ -int -nfs4_do_close(struct inode *inode, struct nfs4_state *state) +static int _nfs4_do_close(struct inode *inode, struct nfs4_state *state) { struct nfs4_state_owner *sp = state->owner; int status = 0; @@ -426,6 +677,8 @@ nfs4_do_close(struct inode *inode, struc .rpc_resp = &res, }; + if (test_bit(NFS_DELEGATED_STATE, &state->flags)) + return 0; memcpy(&arg.stateid, &state->stateid, sizeof(arg.stateid)); /* Serialization for the sequence id */ arg.seqid = sp->so_seqid, @@ -441,15 +694,34 @@ nfs4_do_close(struct inode *inode, struc return status; } -int -nfs4_do_downgrade(struct inode *inode, struct nfs4_state *state, mode_t mode) +int nfs4_do_close(struct inode *inode, struct nfs4_state *state) +{ + struct nfs_server *server = NFS_SERVER(state->inode); + struct nfs4_exception exception = { }; + int err; + do { + err = _nfs4_do_close(inode, state); + switch (err) { + case -NFS4ERR_STALE_STATEID: + case -NFS4ERR_EXPIRED: + nfs4_schedule_state_recovery(server->nfs4_state); + case 0: + state->state = 0; + return 0; + } + err = nfs4_handle_exception(server, err, &exception); + } while (exception.retry); + return err; +} + +static int _nfs4_do_downgrade(struct inode *inode, struct nfs4_state *state, mode_t mode) { struct nfs4_state_owner *sp = state->owner; int status = 0; struct nfs_closeargs arg = { .fh = NFS_FH(inode), .seqid = sp->so_seqid, - .share_access = mode, + .open_flags = mode, }; struct nfs_closeres res; struct rpc_message msg = { @@ -458,6 +730,8 @@ nfs4_do_downgrade(struct inode *inode, s .rpc_resp = &res, }; + if (test_bit(NFS_DELEGATED_STATE, &state->flags)) + return 0; memcpy(&arg.stateid, &state->stateid, sizeof(arg.stateid)); status = rpc_call_sync(NFS_SERVER(inode)->client, &msg, 0); nfs4_increment_seqid(status, sp); @@ -467,6 +741,26 @@ nfs4_do_downgrade(struct inode *inode, s return status; } +int nfs4_do_downgrade(struct inode *inode, struct nfs4_state *state, mode_t mode) +{ + struct nfs_server *server = NFS_SERVER(state->inode); + struct nfs4_exception exception = { }; + int err; + do { + err = _nfs4_do_downgrade(inode, state, mode); + switch (err) { + case -NFS4ERR_STALE_STATEID: + case -NFS4ERR_EXPIRED: + nfs4_schedule_state_recovery(server->nfs4_state); + case 0: + state->state = mode; + return 0; + } + err = nfs4_handle_exception(server, err, &exception); + } while (exception.retry); + return err; +} + struct inode * nfs4_atomic_open(struct inode *dir, struct dentry *dentry, struct nameidata *nd) { @@ -500,7 +794,9 @@ nfs4_open_revalidate(struct inode *dir, struct inode *inode; cred = rpcauth_lookupcred(NFS_SERVER(dir)->client->cl_auth, 0); - state = nfs4_do_open(dir, &dentry->d_name, openflags, NULL, cred); + state = nfs4_open_delegated(dentry->d_inode, openflags, cred); + if (IS_ERR(state)) + state = nfs4_do_open(dir, &dentry->d_name, openflags, NULL, cred); put_rpccred(cred); if (state == ERR_PTR(-ENOENT) && dentry->d_inode == 0) return 1; @@ -518,7 +814,7 @@ nfs4_open_revalidate(struct inode *dir, } -static int nfs4_server_capabilities(struct nfs_server *server, struct nfs_fh *fhandle) +static int _nfs4_server_capabilities(struct nfs_server *server, struct nfs_fh *fhandle) { struct nfs4_server_caps_res res = {}; struct rpc_message msg = { @@ -542,7 +838,19 @@ static int nfs4_server_capabilities(stru return status; } -static int nfs4_lookup_root(struct nfs_server *server, struct nfs_fh *fhandle, +static int nfs4_server_capabilities(struct nfs_server *server, struct nfs_fh *fhandle) +{ + struct nfs4_exception exception = { }; + int err; + do { + err = nfs4_handle_exception(server, + _nfs4_server_capabilities(server, fhandle), + &exception); + } while (exception.retry); + return err; +} + +static int _nfs4_lookup_root(struct nfs_server *server, struct nfs_fh *fhandle, struct nfs_fsinfo *info) { struct nfs_fattr * fattr = info->fattr; @@ -563,6 +871,19 @@ static int nfs4_lookup_root(struct nfs_s return rpc_call_sync(server->client, &msg, 0); } +static int nfs4_lookup_root(struct nfs_server *server, struct nfs_fh *fhandle, + struct nfs_fsinfo *info) +{ + struct nfs4_exception exception = { }; + int err; + do { + err = nfs4_handle_exception(server, + _nfs4_lookup_root(server, fhandle, info), + &exception); + } while (exception.retry); + return err; +} + static int nfs4_proc_get_root(struct nfs_server *server, struct nfs_fh *fhandle, struct nfs_fsinfo *info) { @@ -597,6 +918,8 @@ static int nfs4_proc_get_root(struct nfs p = server->mnt_path; for (;;) { + struct nfs4_exception exception = { }; + while (*p == '/') p++; if (!*p) @@ -606,9 +929,13 @@ static int nfs4_proc_get_root(struct nfs p++; q.len = p - q.name; - fattr->valid = 0; - status = rpc_call_sync(server->client, &msg, 0); - if (!status) + do { + fattr->valid = 0; + status = nfs4_handle_exception(server, + rpc_call_sync(server->client, &msg, 0), + &exception); + } while (exception.retry); + if (status == 0) continue; if (status == -ENOENT) { printk(KERN_NOTICE "NFS: mount path %s does not exist!\n", server->mnt_path); @@ -621,10 +948,10 @@ static int nfs4_proc_get_root(struct nfs if (status == 0) status = nfs4_do_fsinfo(server, fhandle, info); out: - return nfs4_map_errors(status); + return status; } -static int nfs4_proc_getattr(struct inode *inode, struct nfs_fattr *fattr) +static int _nfs4_proc_getattr(struct inode *inode, struct nfs_fattr *fattr) { struct nfs_server *server = NFS_SERVER(inode); struct nfs4_getattr_arg args = { @@ -642,8 +969,19 @@ static int nfs4_proc_getattr(struct inod }; fattr->valid = 0; + return rpc_call_sync(NFS_CLIENT(inode), &msg, 0); +} - return nfs4_map_errors(rpc_call_sync(NFS_CLIENT(inode), &msg, 0)); +static int nfs4_proc_getattr(struct inode *inode, struct nfs_fattr *fattr) +{ + struct nfs4_exception exception = { }; + int err; + do { + err = nfs4_handle_exception(NFS_SERVER(inode), + _nfs4_proc_getattr(inode, fattr), + &exception); + } while (exception.retry); + return err; } /* @@ -678,9 +1016,13 @@ nfs4_proc_setattr(struct dentry *dentry, if (size_change) { struct rpc_cred *cred = rpcauth_lookupcred(NFS_SERVER(inode)->client->cl_auth, 0); state = nfs4_find_state(inode, cred, FMODE_WRITE); - if (!state) { - state = nfs4_do_open(dentry->d_parent->d_inode, - &dentry->d_name, FMODE_WRITE, NULL, cred); + if (state == NULL) { + state = nfs4_open_delegated(dentry->d_inode, + FMODE_WRITE, cred); + if (IS_ERR(state)) + state = nfs4_do_open(dentry->d_parent->d_inode, + &dentry->d_name, FMODE_WRITE, + NULL, cred); need_iput = 1; } put_rpccred(cred); @@ -705,7 +1047,7 @@ out: return status; } -static int nfs4_proc_lookup(struct inode *dir, struct qstr *name, +static int _nfs4_proc_lookup(struct inode *dir, struct qstr *name, struct nfs_fh *fhandle, struct nfs_fattr *fattr) { int status; @@ -731,12 +1073,23 @@ static int nfs4_proc_lookup(struct inode dprintk("NFS call lookup %s\n", name->name); status = rpc_call_sync(NFS_CLIENT(dir), &msg, 0); dprintk("NFS reply lookup: %d\n", status); - return nfs4_map_errors(status); + return status; } -static int nfs4_proc_access(struct inode *inode, struct rpc_cred *cred, int mode) +static int nfs4_proc_lookup(struct inode *dir, struct qstr *name, struct nfs_fh *fhandle, struct nfs_fattr *fattr) +{ + struct nfs4_exception exception = { }; + int err; + do { + err = nfs4_handle_exception(NFS_SERVER(dir), + _nfs4_proc_lookup(dir, name, fhandle, fattr), + &exception); + } while (exception.retry); + return err; +} + +static int _nfs4_proc_access(struct inode *inode, struct nfs_access_entry *entry) { - int status; struct nfs4_accessargs args = { .fh = NFS_FH(inode), }; @@ -745,8 +1098,10 @@ static int nfs4_proc_access(struct inode .rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_ACCESS], .rpc_argp = &args, .rpc_resp = &res, - .rpc_cred = cred, + .rpc_cred = entry->cred, }; + int mode = entry->mask; + int status; /* * Determine which access bits we want to ask for... @@ -758,8 +1113,7 @@ static int nfs4_proc_access(struct inode args.access |= NFS4_ACCESS_MODIFY | NFS4_ACCESS_EXTEND | NFS4_ACCESS_DELETE; if (mode & MAY_EXEC) args.access |= NFS4_ACCESS_LOOKUP; - } - else { + } else { if (mode & MAY_WRITE) args.access |= NFS4_ACCESS_MODIFY | NFS4_ACCESS_EXTEND; if (mode & MAY_EXEC) @@ -767,13 +1121,27 @@ static int nfs4_proc_access(struct inode } status = rpc_call_sync(NFS_CLIENT(inode), &msg, 0); if (!status) { - if (args.access != res.supported) { - printk(KERN_NOTICE "NFS: server didn't support all access bits!\n"); - status = -ENOTSUPP; - } else if ((args.access & res.access) != args.access) - status = -EACCES; + entry->mask = 0; + if (res.access & NFS4_ACCESS_READ) + entry->mask |= MAY_READ; + if (res.access & (NFS4_ACCESS_MODIFY | NFS4_ACCESS_EXTEND | NFS4_ACCESS_DELETE)) + entry->mask |= MAY_WRITE; + if (res.access & (NFS4_ACCESS_LOOKUP|NFS4_ACCESS_EXECUTE)) + entry->mask |= MAY_EXEC; } - return nfs4_map_errors(status); + return status; +} + +static int nfs4_proc_access(struct inode *inode, struct nfs_access_entry *entry) +{ + struct nfs4_exception exception = { }; + int err; + do { + err = nfs4_handle_exception(NFS_SERVER(inode), + _nfs4_proc_access(inode, entry), + &exception); + } while (exception.retry); + return err; } /* @@ -800,7 +1168,7 @@ static int nfs4_proc_access(struct inode * Both of these changes to the XDR layer would in fact be quite * minor, but I decided to leave them for a subsequent patch. */ -static int nfs4_proc_readlink(struct inode *inode, struct page *page) +static int _nfs4_proc_readlink(struct inode *inode, struct page *page) { struct nfs4_readlink args = { .fh = NFS_FH(inode), @@ -813,11 +1181,22 @@ static int nfs4_proc_readlink(struct ino .rpc_resp = NULL, }; - return nfs4_map_errors(rpc_call_sync(NFS_CLIENT(inode), &msg, 0)); + return rpc_call_sync(NFS_CLIENT(inode), &msg, 0); } -static int -nfs4_proc_read(struct nfs_read_data *rdata, struct file *filp) +static int nfs4_proc_readlink(struct inode *inode, struct page *page) +{ + struct nfs4_exception exception = { }; + int err; + do { + err = nfs4_handle_exception(NFS_SERVER(inode), + _nfs4_proc_readlink(inode, page), + &exception); + } while (exception.retry); + return err; +} + +static int _nfs4_proc_read(struct nfs_read_data *rdata) { int flags = rdata->flags; struct inode *inode = rdata->inode; @@ -827,6 +1206,7 @@ nfs4_proc_read(struct nfs_read_data *rda .rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_READ], .rpc_argp = &rdata->args, .rpc_resp = &rdata->res, + .rpc_cred = rdata->cred, }; unsigned long timestamp = jiffies; int status; @@ -834,29 +1214,27 @@ nfs4_proc_read(struct nfs_read_data *rda dprintk("NFS call read %d @ %Ld\n", rdata->args.count, (long long) rdata->args.offset); - /* - * Try first to use O_RDONLY, then O_RDWR stateid. - */ - if (filp) { - struct nfs4_state *state; - state = (struct nfs4_state *)filp->private_data; - rdata->args.state = state; - msg.rpc_cred = state->owner->so_cred; - } else { - rdata->args.state = NULL; - msg.rpc_cred = NFS_I(inode)->mm_cred; - } - fattr->valid = 0; status = rpc_call_sync(server->client, &msg, flags); if (!status) renew_lease(server, timestamp); dprintk("NFS reply read: %d\n", status); - return nfs4_map_errors(status); + return status; } -static int -nfs4_proc_write(struct nfs_write_data *wdata, struct file *filp) +static int nfs4_proc_read(struct nfs_read_data *rdata) +{ + struct nfs4_exception exception = { }; + int err; + do { + err = nfs4_handle_exception(NFS_SERVER(rdata->inode), + _nfs4_proc_read(rdata), + &exception); + } while (exception.retry); + return err; +} + +static int _nfs4_proc_write(struct nfs_write_data *wdata) { int rpcflags = wdata->flags; struct inode *inode = wdata->inode; @@ -866,33 +1244,32 @@ nfs4_proc_write(struct nfs_write_data *w .rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_WRITE], .rpc_argp = &wdata->args, .rpc_resp = &wdata->res, + .rpc_cred = wdata->cred, }; int status; dprintk("NFS call write %d @ %Ld\n", wdata->args.count, (long long) wdata->args.offset); - /* - * Try first to use O_WRONLY, then O_RDWR stateid. - */ - if (filp) { - struct nfs4_state *state; - state = (struct nfs4_state *)filp->private_data; - wdata->args.state = state; - msg.rpc_cred = state->owner->so_cred; - } else { - wdata->args.state = NULL; - msg.rpc_cred = NFS_I(inode)->mm_cred; - } - fattr->valid = 0; status = rpc_call_sync(server->client, &msg, rpcflags); dprintk("NFS reply write: %d\n", status); - return nfs4_map_errors(status); + return status; } -static int -nfs4_proc_commit(struct nfs_write_data *cdata, struct file *filp) +static int nfs4_proc_write(struct nfs_write_data *wdata) +{ + struct nfs4_exception exception = { }; + int err; + do { + err = nfs4_handle_exception(NFS_SERVER(wdata->inode), + _nfs4_proc_write(wdata), + &exception); + } while (exception.retry); + return err; +} + +static int _nfs4_proc_commit(struct nfs_write_data *cdata) { struct inode *inode = cdata->inode; struct nfs_fattr *fattr = cdata->res.fattr; @@ -901,24 +1278,29 @@ nfs4_proc_commit(struct nfs_write_data * .rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_COMMIT], .rpc_argp = &cdata->args, .rpc_resp = &cdata->res, + .rpc_cred = cdata->cred, }; int status; dprintk("NFS call commit %d @ %Ld\n", cdata->args.count, (long long) cdata->args.offset); - /* - * Try first to use O_WRONLY, then O_RDWR stateid. - */ - if (filp) - msg.rpc_cred = ((struct nfs4_state *)filp->private_data)->owner->so_cred; - else - msg.rpc_cred = NFS_I(inode)->mm_cred; - fattr->valid = 0; status = rpc_call_sync(server->client, &msg, 0); dprintk("NFS reply commit: %d\n", status); - return nfs4_map_errors(status); + return status; +} + +static int nfs4_proc_commit(struct nfs_write_data *cdata) +{ + struct nfs4_exception exception = { }; + int err; + do { + err = nfs4_handle_exception(NFS_SERVER(cdata->inode), + _nfs4_proc_commit(cdata), + &exception); + } while (exception.retry); + return err; } /* @@ -965,7 +1347,7 @@ nfs4_proc_create(struct inode *dir, stru return inode; } -static int nfs4_proc_remove(struct inode *dir, struct qstr *name) +static int _nfs4_proc_remove(struct inode *dir, struct qstr *name) { struct nfs4_remove_arg args = { .fh = NFS_FH(dir), @@ -982,7 +1364,19 @@ static int nfs4_proc_remove(struct inode status = rpc_call_sync(NFS_CLIENT(dir), &msg, 0); if (status == 0) update_changeattr(dir, &res); - return nfs4_map_errors(status); + return status; +} + +static int nfs4_proc_remove(struct inode *dir, struct qstr *name) +{ + struct nfs4_exception exception = { }; + int err; + do { + err = nfs4_handle_exception(NFS_SERVER(dir), + _nfs4_proc_remove(dir, name), + &exception); + } while (exception.retry); + return err; } struct unlink_desc { @@ -1023,7 +1417,7 @@ static int nfs4_proc_unlink_done(struct return 0; } -static int nfs4_proc_rename(struct inode *old_dir, struct qstr *old_name, +static int _nfs4_proc_rename(struct inode *old_dir, struct qstr *old_name, struct inode *new_dir, struct qstr *new_name) { struct nfs4_rename_arg arg = { @@ -1046,10 +1440,24 @@ static int nfs4_proc_rename(struct inode update_changeattr(old_dir, &res.old_cinfo); update_changeattr(new_dir, &res.new_cinfo); } - return nfs4_map_errors(status); + return status; } -static int nfs4_proc_link(struct inode *inode, struct inode *dir, struct qstr *name) +static int nfs4_proc_rename(struct inode *old_dir, struct qstr *old_name, + struct inode *new_dir, struct qstr *new_name) +{ + struct nfs4_exception exception = { }; + int err; + do { + err = nfs4_handle_exception(NFS_SERVER(old_dir), + _nfs4_proc_rename(old_dir, old_name, + new_dir, new_name), + &exception); + } while (exception.retry); + return err; +} + +static int _nfs4_proc_link(struct inode *inode, struct inode *dir, struct qstr *name) { struct nfs4_link_arg arg = { .fh = NFS_FH(inode), @@ -1068,10 +1476,22 @@ static int nfs4_proc_link(struct inode * if (!status) update_changeattr(dir, &cinfo); - return nfs4_map_errors(status); + return status; } -static int nfs4_proc_symlink(struct inode *dir, struct qstr *name, +static int nfs4_proc_link(struct inode *inode, struct inode *dir, struct qstr *name) +{ + struct nfs4_exception exception = { }; + int err; + do { + err = nfs4_handle_exception(NFS_SERVER(inode), + _nfs4_proc_link(inode, dir, name), + &exception); + } while (exception.retry); + return err; +} + +static int _nfs4_proc_symlink(struct inode *dir, struct qstr *name, struct qstr *path, struct iattr *sattr, struct nfs_fh *fhandle, struct nfs_fattr *fattr) { @@ -1090,22 +1510,39 @@ static int nfs4_proc_symlink(struct inod .fattr = fattr, }; struct rpc_message msg = { - .rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_CREATE], + .rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SYMLINK], .rpc_argp = &arg, .rpc_resp = &res, }; int status; + if (path->len > NFS4_MAXPATHLEN) + return -ENAMETOOLONG; arg.u.symlink = path; fattr->valid = 0; status = rpc_call_sync(NFS_CLIENT(dir), &msg, 0); if (!status) update_changeattr(dir, &res.dir_cinfo); - return nfs4_map_errors(status); + return status; } -static int nfs4_proc_mkdir(struct inode *dir, struct qstr *name, +static int nfs4_proc_symlink(struct inode *dir, struct qstr *name, + struct qstr *path, struct iattr *sattr, struct nfs_fh *fhandle, + struct nfs_fattr *fattr) +{ + struct nfs4_exception exception = { }; + int err; + do { + err = nfs4_handle_exception(NFS_SERVER(dir), + _nfs4_proc_symlink(dir, name, path, sattr, + fhandle, fattr), + &exception); + } while (exception.retry); + return err; +} + +static int _nfs4_proc_mkdir(struct inode *dir, struct qstr *name, struct iattr *sattr, struct nfs_fh *fhandle, struct nfs_fattr *fattr) { @@ -1135,10 +1572,25 @@ static int nfs4_proc_mkdir(struct inode status = rpc_call_sync(NFS_CLIENT(dir), &msg, 0); if (!status) update_changeattr(dir, &res.dir_cinfo); - return nfs4_map_errors(status); + return status; } -static int nfs4_proc_readdir(struct dentry *dentry, struct rpc_cred *cred, +static int nfs4_proc_mkdir(struct inode *dir, struct qstr *name, + struct iattr *sattr, struct nfs_fh *fhandle, + struct nfs_fattr *fattr) +{ + struct nfs4_exception exception = { }; + int err; + do { + err = nfs4_handle_exception(NFS_SERVER(dir), + _nfs4_proc_mkdir(dir, name, sattr, + fhandle, fattr), + &exception); + } while (exception.retry); + return err; +} + +static int _nfs4_proc_readdir(struct dentry *dentry, struct rpc_cred *cred, u64 cookie, struct page *page, unsigned int count, int plus) { struct inode *dir = dentry->d_inode; @@ -1164,10 +1616,24 @@ static int nfs4_proc_readdir(struct dent if (status == 0) memcpy(NFS_COOKIEVERF(dir), res.verifier.data, NFS4_VERIFIER_SIZE); unlock_kernel(); - return nfs4_map_errors(status); + return status; } -static int nfs4_proc_mknod(struct inode *dir, struct qstr *name, +static int nfs4_proc_readdir(struct dentry *dentry, struct rpc_cred *cred, + u64 cookie, struct page *page, unsigned int count, int plus) +{ + struct nfs4_exception exception = { }; + int err; + do { + err = nfs4_handle_exception(NFS_SERVER(dentry->d_inode), + _nfs4_proc_readdir(dentry, cred, cookie, + page, count, plus), + &exception); + } while (exception.retry); + return err; +} + +static int _nfs4_proc_mknod(struct inode *dir, struct qstr *name, struct iattr *sattr, dev_t rdev, struct nfs_fh *fh, struct nfs_fattr *fattr) { @@ -1214,10 +1680,25 @@ static int nfs4_proc_mknod(struct inode status = rpc_call_sync(NFS_CLIENT(dir), &msg, 0); if (!status) update_changeattr(dir, &res.dir_cinfo); - return nfs4_map_errors(status); + return status; +} + +static int nfs4_proc_mknod(struct inode *dir, struct qstr *name, + struct iattr *sattr, dev_t rdev, struct nfs_fh *fh, + struct nfs_fattr *fattr) +{ + struct nfs4_exception exception = { }; + int err; + do { + err = nfs4_handle_exception(NFS_SERVER(dir), + _nfs4_proc_mknod(dir, name, sattr, rdev, + fh, fattr), + &exception); + } while (exception.retry); + return err; } -static int nfs4_proc_statfs(struct nfs_server *server, struct nfs_fh *fhandle, +static int _nfs4_proc_statfs(struct nfs_server *server, struct nfs_fh *fhandle, struct nfs_fsstat *fsstat) { struct nfs4_statfs_arg args = { @@ -1231,10 +1712,22 @@ static int nfs4_proc_statfs(struct nfs_s }; fsstat->fattr->valid = 0; - return nfs4_map_errors(rpc_call_sync(server->client, &msg, 0)); + return rpc_call_sync(server->client, &msg, 0); } -static int nfs4_do_fsinfo(struct nfs_server *server, struct nfs_fh *fhandle, +static int nfs4_proc_statfs(struct nfs_server *server, struct nfs_fh *fhandle, struct nfs_fsstat *fsstat) +{ + struct nfs4_exception exception = { }; + int err; + do { + err = nfs4_handle_exception(server, + _nfs4_proc_statfs(server, fhandle, fsstat), + &exception); + } while (exception.retry); + return err; +} + +static int _nfs4_do_fsinfo(struct nfs_server *server, struct nfs_fh *fhandle, struct nfs_fsinfo *fsinfo) { struct nfs4_fsinfo_arg args = { @@ -1247,16 +1740,29 @@ static int nfs4_do_fsinfo(struct nfs_ser .rpc_resp = fsinfo, }; - return nfs4_map_errors(rpc_call_sync(server->client, &msg, 0)); + return rpc_call_sync(server->client, &msg, 0); +} + +static int nfs4_do_fsinfo(struct nfs_server *server, struct nfs_fh *fhandle, struct nfs_fsinfo *fsinfo) +{ + struct nfs4_exception exception = { }; + int err; + + do { + err = nfs4_handle_exception(server, + _nfs4_do_fsinfo(server, fhandle, fsinfo), + &exception); + } while (exception.retry); + return err; } static int nfs4_proc_fsinfo(struct nfs_server *server, struct nfs_fh *fhandle, struct nfs_fsinfo *fsinfo) { fsinfo->fattr->valid = 0; - return nfs4_map_errors(nfs4_do_fsinfo(server, fhandle, fsinfo)); + return nfs4_do_fsinfo(server, fhandle, fsinfo); } -static int nfs4_proc_pathconf(struct nfs_server *server, struct nfs_fh *fhandle, +static int _nfs4_proc_pathconf(struct nfs_server *server, struct nfs_fh *fhandle, struct nfs_pathconf *pathconf) { struct nfs4_pathconf_arg args = { @@ -1276,7 +1782,21 @@ static int nfs4_proc_pathconf(struct nfs } pathconf->fattr->valid = 0; - return nfs4_map_errors(rpc_call_sync(server->client, &msg, 0)); + return rpc_call_sync(server->client, &msg, 0); +} + +static int nfs4_proc_pathconf(struct nfs_server *server, struct nfs_fh *fhandle, + struct nfs_pathconf *pathconf) +{ + struct nfs4_exception exception = { }; + int err; + + do { + err = nfs4_handle_exception(server, + _nfs4_proc_pathconf(server, fhandle, pathconf), + &exception); + } while (exception.retry); + return err; } static void @@ -1467,8 +1987,10 @@ static int nfs4_proc_file_open(struct inode *inode, struct file *filp) { struct dentry *dentry = filp->f_dentry; - struct nfs4_state *state; + struct nfs_open_context *ctx; + struct nfs4_state *state = NULL; struct rpc_cred *cred; + int status = -ENOMEM; dprintk("nfs4_proc_file_open: starting on (%.*s/%.*s)\n", (int)dentry->d_parent->d_name.len, @@ -1478,21 +2000,28 @@ nfs4_proc_file_open(struct inode *inode, /* Find our open stateid */ cred = rpcauth_lookupcred(NFS_SERVER(inode)->client->cl_auth, 0); - state = nfs4_find_state(inode, cred, filp->f_mode); + if (unlikely(cred == NULL)) + return -ENOMEM; + ctx = alloc_nfs_open_context(dentry, cred); put_rpccred(cred); - if (state == NULL) { - printk(KERN_WARNING "NFS: v4 raced in function %s\n", __FUNCTION__); - return -EIO; /* ERACE actually */ - } + if (unlikely(ctx == NULL)) + return -ENOMEM; + status = -EIO; /* ERACE actually */ + state = nfs4_find_state(inode, cred, filp->f_mode); + if (unlikely(state == NULL)) + goto no_state; + ctx->state = state; nfs4_close_state(state, filp->f_mode); - if (filp->f_mode & FMODE_WRITE) { - lock_kernel(); - nfs_set_mmcred(inode, state->owner->so_cred); + ctx->mode = filp->f_mode; + nfs_file_set_open_context(filp, ctx); + put_nfs_open_context(ctx); + if (filp->f_mode & FMODE_WRITE) nfs_begin_data_update(inode); - unlock_kernel(); - } - filp->private_data = state; return 0; +no_state: + printk(KERN_WARNING "NFS: v4 raced in function %s\n", __FUNCTION__); + put_nfs_open_context(ctx); + return status; } /* @@ -1501,37 +2030,12 @@ nfs4_proc_file_open(struct inode *inode, static int nfs4_proc_file_release(struct inode *inode, struct file *filp) { - struct nfs4_state *state = (struct nfs4_state *)filp->private_data; - - if (state) - nfs4_close_state(state, filp->f_mode); - if (filp->f_mode & FMODE_WRITE) { - lock_kernel(); + if (filp->f_mode & FMODE_WRITE) nfs_end_data_update(inode); - unlock_kernel(); - } + nfs_file_clear_open_context(filp); return 0; } -/* - * Set up the nfspage struct with the right state info and credentials - */ -static void -nfs4_request_init(struct nfs_page *req, struct file *filp) -{ - struct nfs4_state *state; - - if (!filp) { - req->wb_cred = get_rpccred(NFS_I(req->wb_inode)->mm_cred); - req->wb_state = NULL; - return; - } - state = (struct nfs4_state *)filp->private_data; - req->wb_state = state; - req->wb_cred = get_rpccred(state->owner->so_cred); - req->wb_lockowner = current->files; -} - static int nfs4_async_handle_error(struct rpc_task *task, struct nfs_server *server) { @@ -1545,11 +2049,13 @@ nfs4_async_handle_error(struct rpc_task case -NFS4ERR_EXPIRED: rpc_sleep_on(&clp->cl_rpcwaitq, task, NULL, NULL); nfs4_schedule_state_recovery(clp); + if (test_bit(NFS4CLNT_OK, &clp->cl_state)) + rpc_wake_up_task(task); task->tk_status = 0; return -EAGAIN; case -NFS4ERR_GRACE: case -NFS4ERR_DELAY: - rpc_delay(task, NFS4_POLL_RETRY_TIME); + rpc_delay(task, NFS4_POLL_RETRY_MAX); task->tk_status = 0; return -EAGAIN; case -NFS4ERR_OLD_STATEID: @@ -1560,12 +2066,11 @@ nfs4_async_handle_error(struct rpc_task return 0; } -int -nfs4_wait_clnt_recover(struct rpc_clnt *clnt, struct nfs4_client *clp) +int nfs4_wait_clnt_recover(struct rpc_clnt *clnt, struct nfs4_client *clp) { DEFINE_WAIT(wait); sigset_t oldset; - int interruptible, res; + int interruptible, res = 0; might_sleep(); @@ -1573,101 +2078,85 @@ nfs4_wait_clnt_recover(struct rpc_clnt * interruptible = TASK_UNINTERRUPTIBLE; if (clnt->cl_intr) interruptible = TASK_INTERRUPTIBLE; - do { - res = 0; - prepare_to_wait(&clp->cl_waitq, &wait, interruptible); - nfs4_schedule_state_recovery(clp); - if (test_bit(NFS4CLNT_OK, &clp->cl_state) && - !test_bit(NFS4CLNT_SETUP_STATE, &clp->cl_state)) - break; - if (clnt->cl_intr && signalled()) { - res = -ERESTARTSYS; - break; - } + prepare_to_wait(&clp->cl_waitq, &wait, interruptible); + nfs4_schedule_state_recovery(clp); + if (clnt->cl_intr && signalled()) + res = -ERESTARTSYS; + else if (!test_bit(NFS4CLNT_OK, &clp->cl_state)) schedule(); - } while(!test_bit(NFS4CLNT_OK, &clp->cl_state)); finish_wait(&clp->cl_waitq, &wait); rpc_clnt_sigunmask(clnt, &oldset); return res; } -static int -nfs4_delay(struct rpc_clnt *clnt) +static int nfs4_delay(struct rpc_clnt *clnt, long *timeout) { sigset_t oldset; int res = 0; might_sleep(); + if (*timeout <= 0) + *timeout = NFS4_POLL_RETRY_MIN; + if (*timeout > NFS4_POLL_RETRY_MAX) + *timeout = NFS4_POLL_RETRY_MAX; rpc_clnt_sigmask(clnt, &oldset); if (clnt->cl_intr) { set_current_state(TASK_INTERRUPTIBLE); - schedule_timeout(NFS4_POLL_RETRY_TIME); + schedule_timeout(*timeout); if (signalled()) res = -ERESTARTSYS; } else { set_current_state(TASK_UNINTERRUPTIBLE); - schedule_timeout(NFS4_POLL_RETRY_TIME); + schedule_timeout(*timeout); } rpc_clnt_sigunmask(clnt, &oldset); + *timeout <<= 1; return res; } /* This is the error handling routine for processes that are allowed * to sleep. */ -int -nfs4_handle_error(struct nfs_server *server, int errorcode) +int nfs4_handle_exception(struct nfs_server *server, int errorcode, struct nfs4_exception *exception) { struct nfs4_client *clp = server->nfs4_state; int ret = errorcode; + exception->retry = 0; switch(errorcode) { + case 0: + return 0; case -NFS4ERR_STALE_CLIENTID: case -NFS4ERR_STALE_STATEID: case -NFS4ERR_EXPIRED: ret = nfs4_wait_clnt_recover(server->client, clp); + if (ret == 0) + exception->retry = 1; break; case -NFS4ERR_GRACE: case -NFS4ERR_DELAY: - ret = nfs4_delay(server->client); + ret = nfs4_delay(server->client, &exception->timeout); + if (ret == 0) + exception->retry = 1; break; case -NFS4ERR_OLD_STATEID: - ret = 0; + if (ret == 0) + exception->retry = 1; } /* We failed to handle the error */ return nfs4_map_errors(ret); } - -static int -nfs4_request_compatible(struct nfs_page *req, struct file *filp, struct page *page) -{ - struct nfs4_state *state = NULL; - struct rpc_cred *cred = NULL; - - if (req->wb_file != filp) - return 0; - if (req->wb_page != page) - return 0; - state = (struct nfs4_state *)filp->private_data; - if (req->wb_state != state) - return 0; - if (req->wb_lockowner != current->files) - return 0; - cred = state->owner->so_cred; - if (req->wb_cred != cred) - return 0; - return 1; -} - -int -nfs4_proc_setclientid(struct nfs4_client *clp, - u32 program, unsigned short port) +int nfs4_proc_setclientid(struct nfs4_client *clp, u32 program, unsigned short port) { - u32 *p; - struct nfs4_setclientid setclientid; - struct timespec tv; + static nfs4_verifier sc_verifier; + static int initialized; + + struct nfs4_setclientid setclientid = { + .sc_verifier = &sc_verifier, + .sc_prog = program, + }; struct rpc_message msg = { .rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SETCLIENTID], .rpc_argp = &setclientid, @@ -1675,15 +2164,24 @@ nfs4_proc_setclientid(struct nfs4_client .rpc_cred = clp->cl_cred, }; - tv = CURRENT_TIME; - p = (u32*)setclientid.sc_verifier.data; - *p++ = (u32)tv.tv_sec; - *p = (u32)tv.tv_nsec; - setclientid.sc_name = clp->cl_ipaddr; - sprintf(setclientid.sc_netid, "tcp"); - sprintf(setclientid.sc_uaddr, "%s.%d.%d", clp->cl_ipaddr, port >> 8, port & 255); - setclientid.sc_prog = htonl(program); - setclientid.sc_cb_ident = 0; + if (!initialized) { + struct timespec boot_time; + u32 *p; + + initialized = 1; + boot_time = CURRENT_TIME; + p = (u32*)sc_verifier.data; + *p++ = htonl((u32)boot_time.tv_sec); + *p = htonl((u32)boot_time.tv_nsec); + } + setclientid.sc_name_len = scnprintf(setclientid.sc_name, + sizeof(setclientid.sc_name), "%s/%u.%u.%u.%u", + clp->cl_ipaddr, NIPQUAD(clp->cl_addr.s_addr)); + setclientid.sc_netid_len = scnprintf(setclientid.sc_netid, + sizeof(setclientid.sc_netid), "tcp"); + setclientid.sc_uaddr_len = scnprintf(setclientid.sc_uaddr, + sizeof(setclientid.sc_uaddr), "%s.%d.%d", + clp->cl_ipaddr, port >> 8, port & 255); return rpc_call_sync(clp->cl_rpcclient, &msg, 0); } @@ -1712,6 +2210,40 @@ nfs4_proc_setclientid_confirm(struct nfs return status; } +static int _nfs4_proc_delegreturn(struct inode *inode, struct rpc_cred *cred, const nfs4_stateid *stateid) +{ + struct nfs4_delegreturnargs args = { + .fhandle = NFS_FH(inode), + .stateid = stateid, + }; + struct rpc_message msg = { + .rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_DELEGRETURN], + .rpc_argp = &args, + .rpc_cred = cred, + }; + + return rpc_call_sync(NFS_CLIENT(inode), &msg, 0); +} + +int nfs4_proc_delegreturn(struct inode *inode, struct rpc_cred *cred, const nfs4_stateid *stateid) +{ + struct nfs_server *server = NFS_SERVER(inode); + struct nfs4_exception exception = { }; + int err; + do { + err = _nfs4_proc_delegreturn(inode, cred, stateid); + switch (err) { + case -NFS4ERR_STALE_STATEID: + case -NFS4ERR_EXPIRED: + nfs4_schedule_state_recovery(server->nfs4_state); + case 0: + return 0; + } + err = nfs4_handle_exception(server, err, &exception); + } while (exception.retry); + return err; +} + #define NFS4_LOCK_MINTIMEOUT (1 * HZ) #define NFS4_LOCK_MAXTIMEOUT (30 * HZ) @@ -1753,8 +2285,7 @@ nfs4_lck_length(struct file_lock *reques return request->fl_end - request->fl_start + 1; } -int -nfs4_proc_getlk(struct nfs4_state *state, int cmd, struct file_lock *request) +static int _nfs4_proc_getlk(struct nfs4_state *state, int cmd, struct file_lock *request) { struct inode *inode = state->inode; struct nfs_server *server = NFS_SERVER(inode); @@ -1778,9 +2309,10 @@ nfs4_proc_getlk(struct nfs4_state *state struct nfs4_lock_state *lsp; int status; + down_read(&clp->cl_sem); nlo.clientid = clp->cl_clientid; down(&state->lock_sema); - lsp = nfs4_find_lock_state(state, request->fl_owner); + lsp = nfs4_find_lock_state(state, request->fl_pid); if (lsp) nlo.id = lsp->ls_id; else { @@ -1811,14 +2343,28 @@ nfs4_proc_getlk(struct nfs4_state *state if (lsp) nfs4_put_lock_state(lsp); up(&state->lock_sema); - return nfs4_map_errors(status); + up_read(&clp->cl_sem); + return status; } -int -nfs4_proc_unlck(struct nfs4_state *state, int cmd, struct file_lock *request) +static int nfs4_proc_getlk(struct nfs4_state *state, int cmd, struct file_lock *request) +{ + struct nfs4_exception exception = { }; + int err; + + do { + err = nfs4_handle_exception(NFS_SERVER(state->inode), + _nfs4_proc_getlk(state, cmd, request), + &exception); + } while (exception.retry); + return err; +} + +static int _nfs4_proc_unlck(struct nfs4_state *state, int cmd, struct file_lock *request) { struct inode *inode = state->inode; struct nfs_server *server = NFS_SERVER(inode); + struct nfs4_client *clp = server->nfs4_state; struct nfs_lockargs arg = { .fh = NFS_FH(inode), .type = nfs4_lck_type(cmd, request), @@ -1838,29 +2384,46 @@ nfs4_proc_unlck(struct nfs4_state *state struct nfs_locku_opargs luargs; int status = 0; + down_read(&clp->cl_sem); down(&state->lock_sema); - lsp = nfs4_find_lock_state(state, request->fl_owner); + lsp = nfs4_find_lock_state(state, request->fl_pid); if (!lsp) goto out; - luargs.seqid = lsp->ls_seqid; - memcpy(&luargs.stateid, &lsp->ls_stateid, sizeof(luargs.stateid)); - arg.u.locku = &luargs; - status = rpc_call_sync(server->client, &msg, 0); - nfs4_increment_lock_seqid(status, lsp); + /* We might have lost the locks! */ + if ((lsp->flags & NFS_LOCK_INITIALIZED) != 0) { + luargs.seqid = lsp->ls_seqid; + memcpy(&luargs.stateid, &lsp->ls_stateid, sizeof(luargs.stateid)); + arg.u.locku = &luargs; + status = rpc_call_sync(server->client, &msg, 0); + nfs4_increment_lock_seqid(status, lsp); + } if (status == 0) { memcpy(&lsp->ls_stateid, &res.u.stateid, sizeof(lsp->ls_stateid)); - nfs4_notify_unlck(inode, request, lsp); + nfs4_notify_unlck(state, request, lsp); } nfs4_put_lock_state(lsp); out: up(&state->lock_sema); - return nfs4_map_errors(status); + up_read(&clp->cl_sem); + return status; } -static int -nfs4_proc_setlk(struct nfs4_state *state, int cmd, struct file_lock *request) +static int nfs4_proc_unlck(struct nfs4_state *state, int cmd, struct file_lock *request) +{ + struct nfs4_exception exception = { }; + int err; + + do { + err = nfs4_handle_exception(NFS_SERVER(state->inode), + _nfs4_proc_unlck(state, cmd, request), + &exception); + } while (exception.retry); + return err; +} + +static int _nfs4_do_setlk(struct nfs4_state *state, int cmd, struct file_lock *request, int reclaim) { struct inode *inode = state->inode; struct nfs_server *server = NFS_SERVER(inode); @@ -1881,23 +2444,22 @@ nfs4_proc_setlk(struct nfs4_state *state .rpc_cred = state->owner->so_cred, }; struct nfs_lock_opargs largs = { + .reclaim = reclaim, .new_lock_owner = 0, }; int status; - down(&state->lock_sema); - lsp = nfs4_find_lock_state(state, request->fl_owner); - if (lsp == NULL) { + lsp = nfs4_get_lock_state(state, request->fl_pid); + if (lsp == NULL) + return -ENOMEM; + if (!(lsp->flags & NFS_LOCK_INITIALIZED)) { struct nfs4_state_owner *owner = state->owner; struct nfs_open_to_lock otl = { .lock_owner = { .clientid = server->nfs4_state->cl_clientid, }, }; - status = -ENOMEM; - lsp = nfs4_alloc_lock_state(state, request->fl_owner); - if (!lsp) - goto out; + otl.lock_seqid = lsp->ls_seqid; otl.lock_owner.id = lsp->ls_id; memcpy(&otl.open_stateid, &state->stateid, sizeof(otl.open_stateid)); @@ -1926,25 +2488,56 @@ nfs4_proc_setlk(struct nfs4_state *state /* save the returned stateid. */ if (status == 0) { memcpy(&lsp->ls_stateid, &res.u.stateid, sizeof(nfs4_stateid)); - nfs4_notify_setlk(inode, reque