2 * Copyright (c) 2000-2005 Silicon Graphics, Inc.
5 * This program is free software; you can redistribute it and/or
6 * modify it under the terms of the GNU General Public License as
7 * published by the Free Software Foundation.
9 * This program is distributed in the hope that it would be useful,
10 * but WITHOUT ANY WARRANTY; without even the implied warranty of
11 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 * GNU General Public License for more details.
14 * You should have received a copy of the GNU General Public License
15 * along with this program; if not, write the Free Software Foundation,
16 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
25 #include "xfs_trans.h"
26 #include "xfs_dmapi.h"
27 #include "xfs_mount.h"
28 #include "xfs_bmap_btree.h"
29 #include "xfs_alloc_btree.h"
30 #include "xfs_ialloc_btree.h"
31 #include "xfs_alloc.h"
32 #include "xfs_btree.h"
33 #include "xfs_attr_sf.h"
34 #include "xfs_dir2_sf.h"
35 #include "xfs_dinode.h"
36 #include "xfs_inode.h"
37 #include "xfs_error.h"
39 #include "xfs_ioctl32.h"
40 #include "xfs_vnodeops.h"
41 #include "xfs_da_btree.h"
43 #include <linux/dcache.h>
44 #include <linux/smp_lock.h>
46 static struct vm_operations_struct xfs_file_vm_ops;
51 const struct iovec *iov,
52 unsigned long nr_segs,
56 struct file *file = iocb->ki_filp;
58 BUG_ON(iocb->ki_pos != pos);
59 if (unlikely(file->f_flags & O_DIRECT))
60 ioflags |= IO_ISDIRECT;
61 return xfs_read(XFS_I(file->f_path.dentry->d_inode), iocb, iov,
62 nr_segs, &iocb->ki_pos, ioflags);
68 const struct iovec *iov,
69 unsigned long nr_segs,
72 return __xfs_file_read(iocb, iov, nr_segs, IO_ISAIO, pos);
76 xfs_file_aio_read_invis(
78 const struct iovec *iov,
79 unsigned long nr_segs,
82 return __xfs_file_read(iocb, iov, nr_segs, IO_ISAIO|IO_INVIS, pos);
88 const struct iovec *iov,
89 unsigned long nr_segs,
93 struct file *file = iocb->ki_filp;
95 BUG_ON(iocb->ki_pos != pos);
96 if (unlikely(file->f_flags & O_DIRECT))
97 ioflags |= IO_ISDIRECT;
98 return xfs_write(XFS_I(file->f_mapping->host), iocb, iov, nr_segs,
99 &iocb->ki_pos, ioflags);
105 const struct iovec *iov,
106 unsigned long nr_segs,
109 return __xfs_file_write(iocb, iov, nr_segs, IO_ISAIO, pos);
113 xfs_file_aio_write_invis(
115 const struct iovec *iov,
116 unsigned long nr_segs,
119 return __xfs_file_write(iocb, iov, nr_segs, IO_ISAIO|IO_INVIS, pos);
123 xfs_file_splice_read(
126 struct pipe_inode_info *pipe,
130 return xfs_splice_read(XFS_I(infilp->f_path.dentry->d_inode),
131 infilp, ppos, pipe, len, flags, 0);
135 xfs_file_splice_read_invis(
138 struct pipe_inode_info *pipe,
142 return xfs_splice_read(XFS_I(infilp->f_path.dentry->d_inode),
143 infilp, ppos, pipe, len, flags, IO_INVIS);
147 xfs_file_splice_write(
148 struct pipe_inode_info *pipe,
149 struct file *outfilp,
154 return xfs_splice_write(XFS_I(outfilp->f_path.dentry->d_inode),
155 pipe, outfilp, ppos, len, flags, 0);
159 xfs_file_splice_write_invis(
160 struct pipe_inode_info *pipe,
161 struct file *outfilp,
166 return xfs_splice_write(XFS_I(outfilp->f_path.dentry->d_inode),
167 pipe, outfilp, ppos, len, flags, IO_INVIS);
175 if (!(file->f_flags & O_LARGEFILE) && i_size_read(inode) > MAX_NON_LFS)
177 if (XFS_FORCED_SHUTDOWN(XFS_M(inode->i_sb)))
187 struct xfs_inode *ip = XFS_I(inode);
191 error = xfs_file_open(inode, file);
196 * If there are any blocks, read-ahead block 0 as we're almost
197 * certain to have the next operation be a read there.
199 mode = xfs_ilock_map_shared(ip);
200 if (ip->i_d.di_nextents > 0)
201 xfs_da_reada_buf(NULL, ip, 0, XFS_DATA_FORK);
202 xfs_iunlock(ip, mode);
211 return -xfs_release(XFS_I(inode));
215 * We ignore the datasync flag here because a datasync is effectively
216 * identical to an fsync. That is, datasync implies that we need to write
217 * only the metadata needed to be able to access the data that is written
218 * if we crash after the call completes. Hence if we are writing beyond
219 * EOF we have to log the inode size change as well, which makes it a
220 * full fsync. If we don't write beyond EOF, the inode core will be
221 * clean in memory and so we don't need to log the inode, just like
227 struct dentry *dentry,
230 xfs_iflags_clear(XFS_I(dentry->d_inode), XFS_ITRUNCATED);
231 return -xfs_fsync(XFS_I(dentry->d_inode));
240 struct inode *inode = filp->f_path.dentry->d_inode;
241 xfs_inode_t *ip = XFS_I(inode);
246 * The Linux API doesn't pass down the total size of the buffer
247 * we read into down to the filesystem. With the filldir concept
248 * it's not needed for correct information, but the XFS dir2 leaf
249 * code wants an estimate of the buffer size to calculate it's
250 * readahead window and size the buffers used for mapping to
253 * Try to give it an estimate that's good enough, maybe at some
254 * point we can change the ->readdir prototype to include the
257 bufsize = (size_t)min_t(loff_t, PAGE_SIZE, inode->i_size);
259 error = xfs_readdir(ip, dirent, bufsize,
260 (xfs_off_t *)&filp->f_pos, filldir);
269 struct vm_area_struct *vma)
271 vma->vm_ops = &xfs_file_vm_ops;
272 vma->vm_flags |= VM_CAN_NONLINEAR;
285 struct inode *inode = filp->f_path.dentry->d_inode;
287 error = xfs_ioctl(XFS_I(inode), filp, 0, cmd, (void __user *)p);
288 xfs_iflags_set(XFS_I(inode), XFS_IMODIFIED);
290 /* NOTE: some of the ioctl's return positive #'s as a
291 * byte count indicating success, such as
292 * readlink_by_handle. So we don't "sign flip"
293 * like most other routines. This means true
294 * errors need to be returned as a negative value.
300 xfs_file_ioctl_invis(
306 struct inode *inode = filp->f_path.dentry->d_inode;
308 error = xfs_ioctl(XFS_I(inode), filp, IO_INVIS, cmd, (void __user *)p);
309 xfs_iflags_set(XFS_I(inode), XFS_IMODIFIED);
311 /* NOTE: some of the ioctl's return positive #'s as a
312 * byte count indicating success, such as
313 * readlink_by_handle. So we don't "sign flip"
314 * like most other routines. This means true
315 * errors need to be returned as a negative value.
321 * mmap()d file has taken write protection fault and is being made
322 * writable. We can set the page state up correctly for a writable
323 * page, which means we can do correct delalloc accounting (ENOSPC
324 * checking!) and unwritten extent mapping.
328 struct vm_area_struct *vma,
331 return block_page_mkwrite(vma, page, xfs_get_blocks);
334 const struct file_operations xfs_file_operations = {
335 .llseek = generic_file_llseek,
336 .read = do_sync_read,
337 .write = do_sync_write,
338 .aio_read = xfs_file_aio_read,
339 .aio_write = xfs_file_aio_write,
340 .splice_read = xfs_file_splice_read,
341 .splice_write = xfs_file_splice_write,
342 .unlocked_ioctl = xfs_file_ioctl,
344 .compat_ioctl = xfs_file_compat_ioctl,
346 .mmap = xfs_file_mmap,
347 .open = xfs_file_open,
348 .release = xfs_file_release,
349 .fsync = xfs_file_fsync,
350 #ifdef HAVE_FOP_OPEN_EXEC
351 .open_exec = xfs_file_open_exec,
355 const struct file_operations xfs_invis_file_operations = {
356 .llseek = generic_file_llseek,
357 .read = do_sync_read,
358 .write = do_sync_write,
359 .aio_read = xfs_file_aio_read_invis,
360 .aio_write = xfs_file_aio_write_invis,
361 .splice_read = xfs_file_splice_read_invis,
362 .splice_write = xfs_file_splice_write_invis,
363 .unlocked_ioctl = xfs_file_ioctl_invis,
365 .compat_ioctl = xfs_file_compat_invis_ioctl,
367 .mmap = xfs_file_mmap,
368 .open = xfs_file_open,
369 .release = xfs_file_release,
370 .fsync = xfs_file_fsync,
374 const struct file_operations xfs_dir_file_operations = {
375 .open = xfs_dir_open,
376 .read = generic_read_dir,
377 .readdir = xfs_file_readdir,
378 .llseek = generic_file_llseek,
379 .unlocked_ioctl = xfs_file_ioctl,
381 .compat_ioctl = xfs_file_compat_ioctl,
383 .fsync = xfs_file_fsync,
386 static struct vm_operations_struct xfs_file_vm_ops = {
387 .fault = filemap_fault,
388 .page_mkwrite = xfs_vm_page_mkwrite,