/* * linux/fs/nfs/file.c * * Copyright (C) 1992 Rick Sladkey */ #include #include #include #include #include /* BTRFS_IOC_CLONE/BTRFS_IOC_CLONE_RANGE */ #include "delegation.h" #include "internal.h" #include "iostat.h" #include "fscache.h" #include "pnfs.h" #include "nfstrace.h" #ifdef CONFIG_NFS_V4_2 #include "nfs42.h" #endif #define NFSDBG_FACILITY NFSDBG_FILE static int nfs4_file_open(struct inode *inode, struct file *filp) { struct nfs_open_context *ctx; struct dentry *dentry = file_dentry(filp); struct dentry *parent = NULL; struct inode *dir; unsigned openflags = filp->f_flags; struct iattr attr; int err; /* * If no cached dentry exists or if it's negative, NFSv4 handled the * opens in ->lookup() or ->create(). * * We only get this far for a cached positive dentry. We skipped * revalidation, so handle it here by dropping the dentry and returning * -EOPENSTALE. The VFS will retry the lookup/create/open. */ dprintk("NFS: open file(%pd2)\n", dentry); err = nfs_check_flags(openflags); if (err) return err; if ((openflags & O_ACCMODE) == 3) openflags--; /* We can't create new files here */ openflags &= ~(O_CREAT|O_EXCL); parent = dget_parent(dentry); dir = d_inode(parent); ctx = alloc_nfs_open_context(file_dentry(filp), filp->f_mode, filp); err = PTR_ERR(ctx); if (IS_ERR(ctx)) goto out; attr.ia_valid = ATTR_OPEN; if (openflags & O_TRUNC) { attr.ia_valid |= ATTR_SIZE; attr.ia_size = 0; filemap_write_and_wait(inode->i_mapping); } inode = NFS_PROTO(dir)->open_context(dir, ctx, openflags, &attr, NULL); if (IS_ERR(inode)) { err = PTR_ERR(inode); switch (err) { case -EPERM: case -EACCES: case -EDQUOT: case -ENOSPC: case -EROFS: goto out_put_ctx; default: goto out_drop; } } if (inode != d_inode(dentry)) goto out_drop; nfs_set_verifier(dentry, nfs_save_change_attribute(dir)); nfs_file_set_open_context(filp, ctx); nfs_fscache_open_file(inode, filp); err = 0; out_put_ctx: put_nfs_open_context(ctx); out: dput(parent); return err; out_drop: d_drop(dentry); err = -EOPENSTALE; goto out_put_ctx; } /* * Flush all dirty pages, and check for write errors. */ static int nfs4_file_flush(struct file *file, fl_owner_t id) { struct inode *inode = file_inode(file); dprintk("NFS: flush(%pD2)\n", file); nfs_inc_stats(inode, NFSIOS_VFSFLUSH); if ((file->f_mode & FMODE_WRITE) == 0) return 0; /* * If we're holding a write delegation, then check if we're required * to flush the i/o on close. If not, then just start the i/o now. */ if (!nfs4_delegation_flush_on_close(inode)) return filemap_fdatawrite(file->f_mapping); /* Flush writes to the server and return any errors */ return vfs_fsync(file, 0); } #ifdef CONFIG_NFS_V4_2 static ssize_t nfs4_copy_file_range(struct file *file_in, loff_t pos_in, struct file *file_out, loff_t pos_out, size_t count, unsigned int flags) { if (file_inode(file_in) == file_inode(file_out)) return -EINVAL; return nfs42_proc_copy(file_in, pos_in, file_out, pos_out, count); } static loff_t nfs4_file_llseek(struct file *filep, loff_t offset, int whence) { loff_t ret; switch (whence) { case SEEK_HOLE: case SEEK_DATA: ret = nfs42_proc_llseek(filep, offset, whence); if (ret != -ENOTSUPP) return ret; default: return nfs_file_llseek(filep, offset, whence); } } static long nfs42_fallocate(struct file *filep, int mode, loff_t offset, loff_t len) { struct inode *inode = file_inode(filep); long ret; if (!S_ISREG(inode->i_mode)) return -EOPNOTSUPP; if ((mode != 0) && (mode != (FALLOC_FL_PUNCH_HOLE | FALLOC_FL_KEEP_SIZE))) return -EOPNOTSUPP; ret = inode_newsize_ok(inode, offset + len); if (ret < 0) return ret; if (mode & FALLOC_FL_PUNCH_HOLE) return nfs42_proc_deallocate(filep, offset, len); return nfs42_proc_allocate(filep, offset, len); } static int nfs42_clone_file_range(struct file *src_file, loff_t src_off, struct file *dst_file, loff_t dst_off, u64 count) { struct inode *dst_inode = file_inode(dst_file); struct nfs_server *server = NFS_SERVER(dst_inode); struct inode *src_inode = file_inode(src_file); unsigned int bs = server->clone_blksize; bool same_inode = false; int ret; /* check alignment w.r.t. clone_blksize */ ret = -EINVAL; if (bs) { if (!IS_ALIGNED(src_off, bs) || !IS_ALIGNED(dst_off, bs)) goto out; if (!IS_ALIGNED(count, bs) && i_size_read(src_inode) != (src_off + count)) goto out; } if (src_inode == dst_inode) same_inode = true; /* XXX: do we lock at all? what if server needs CB_RECALL_LAYOUT? */ if (same_inode) { inode_lock(src_inode); } else if (dst_inode < src_inode) { inode_lock_nested(dst_inode, I_MUTEX_PARENT); inode_lock_nested(src_inode, I_MUTEX_CHILD); } else { inode_lock_nested(src_inode, I_MUTEX_PARENT); inode_lock_nested(dst_inode, I_MUTEX_CHILD); } /* flush all pending writes on both src and dst so that server * has the latest data */ ret = nfs_sync_inode(src_inode); if (ret) goto out_unlock; ret = nfs_sync_inode(dst_inode); if (ret) goto out_unlock; ret = nfs42_proc_clone(src_file, dst_file, src_off, dst_off, count); /* truncate inode page cache of the dst range so that future reads can fetch * new data from server */ if (!ret) truncate_inode_pages_range(&dst_inode->i_data, dst_off, dst_off + count - 1); out_unlock: if (same_inode) { inode_unlock(src_inode); } else if (dst_inode < src_inode) { inode_unlock(src_inode); inode_unlock(dst_inode); } else { inode_unlock(dst_inode); inode_unlock(src_inode); } out: return ret; } #endif /* CONFIG_NFS_V4_2 */ const struct file_operations nfs4_file_operations = { .read_iter = nfs_file_read, .write_iter = nfs_file_write, .mmap = nfs_file_mmap, .open = nfs4_file_open, .flush = nfs4_file_flush, .release = nfs_file_release, .fsync = nfs_file_fsync, .lock = nfs_lock, .flock = nfs_flock, .splice_read = generic_file_splice_read, .splice_write = iter_file_splice_write, .check_flags = nfs_check_flags, .setlease = simple_nosetlease, #ifdef CONFIG_NFS_V4_2 .copy_file_range = nfs4_copy_file_range, .llseek = nfs4_file_llseek, .fallocate = nfs42_fallocate, .clone_file_range = nfs42_clone_file_range, #else .llseek = nfs_file_llseek, #endif }; nt counting if NLM_F_EXCL is is not set. Moreover, don't decrenent set->nelems from abort patch if -ENFILE which leaks a spare slot in the set. This includes a patch to deconstify the set walk callback to update set->ndeact. 3) Two fixes for the fwmark_reflect sysctl feature: Propagate mark to reply packets both from nf_reject and local stack, from Pau Espin Pedrol. 4) Fix incorrect handling of loopback traffic in rpfilter and nf_tables fib expression, from Liping Zhang. 5) Fix oops on stateful objects netlink dump, when no filter is specified. Also from Liping Zhang. 6) Fix a build error if proc is not available in ipt_CLUSTERIP, related to fix that was applied in the previous batch for net. From Arnd Bergmann. 7) Fix lack of string validation in table, chain, set and stateful object names in nf_tables, from Liping Zhang. Moreover, restrict maximum log prefix length to 127 bytes, otherwise explicitly bail out. 8) Two patches to fix spelling and typos in nf_tables uapi header file and Kconfig, patches from Alexander Alemayhu and William Breathitt Gray. ==================== Signed-off-by: David S. Miller <davem@davemloft.net>
Diffstat