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"
42 #include <linux/dcache.h>
43 #include <linux/smp_lock.h>
45 static struct vm_operations_struct xfs_file_vm_ops
;
50 const struct iovec
*iov
,
51 unsigned long nr_segs
,
55 struct file
*file
= iocb
->ki_filp
;
57 BUG_ON(iocb
->ki_pos
!= pos
);
58 if (unlikely(file
->f_flags
& O_DIRECT
))
59 ioflags
|= IO_ISDIRECT
;
60 return xfs_read(XFS_I(file
->f_path
.dentry
->d_inode
), iocb
, iov
,
61 nr_segs
, &iocb
->ki_pos
, ioflags
);
67 const struct iovec
*iov
,
68 unsigned long nr_segs
,
71 return __xfs_file_read(iocb
, iov
, nr_segs
, IO_ISAIO
, pos
);
75 xfs_file_aio_read_invis(
77 const struct iovec
*iov
,
78 unsigned long nr_segs
,
81 return __xfs_file_read(iocb
, iov
, nr_segs
, IO_ISAIO
|IO_INVIS
, pos
);
87 const struct iovec
*iov
,
88 unsigned long nr_segs
,
92 struct file
*file
= iocb
->ki_filp
;
94 BUG_ON(iocb
->ki_pos
!= pos
);
95 if (unlikely(file
->f_flags
& O_DIRECT
))
96 ioflags
|= IO_ISDIRECT
;
97 return xfs_write(XFS_I(file
->f_mapping
->host
), iocb
, iov
, nr_segs
,
98 &iocb
->ki_pos
, ioflags
);
104 const struct iovec
*iov
,
105 unsigned long nr_segs
,
108 return __xfs_file_write(iocb
, iov
, nr_segs
, IO_ISAIO
, pos
);
112 xfs_file_aio_write_invis(
114 const struct iovec
*iov
,
115 unsigned long nr_segs
,
118 return __xfs_file_write(iocb
, iov
, nr_segs
, IO_ISAIO
|IO_INVIS
, pos
);
122 xfs_file_splice_read(
125 struct pipe_inode_info
*pipe
,
129 return xfs_splice_read(XFS_I(infilp
->f_path
.dentry
->d_inode
),
130 infilp
, ppos
, pipe
, len
, flags
, 0);
134 xfs_file_splice_read_invis(
137 struct pipe_inode_info
*pipe
,
141 return xfs_splice_read(XFS_I(infilp
->f_path
.dentry
->d_inode
),
142 infilp
, ppos
, pipe
, len
, flags
, IO_INVIS
);
146 xfs_file_splice_write(
147 struct pipe_inode_info
*pipe
,
148 struct file
*outfilp
,
153 return xfs_splice_write(XFS_I(outfilp
->f_path
.dentry
->d_inode
),
154 pipe
, outfilp
, ppos
, len
, flags
, 0);
158 xfs_file_splice_write_invis(
159 struct pipe_inode_info
*pipe
,
160 struct file
*outfilp
,
165 return xfs_splice_write(XFS_I(outfilp
->f_path
.dentry
->d_inode
),
166 pipe
, outfilp
, ppos
, len
, flags
, IO_INVIS
);
174 if (!(filp
->f_flags
& O_LARGEFILE
) && i_size_read(inode
) > MAX_NON_LFS
)
176 return -xfs_open(XFS_I(inode
));
184 return -xfs_release(XFS_I(inode
));
188 * We ignore the datasync flag here because a datasync is effectively
189 * identical to an fsync. That is, datasync implies that we need to write
190 * only the metadata needed to be able to access the data that is written
191 * if we crash after the call completes. Hence if we are writing beyond
192 * EOF we have to log the inode size change as well, which makes it a
193 * full fsync. If we don't write beyond EOF, the inode core will be
194 * clean in memory and so we don't need to log the inode, just like
200 struct dentry
*dentry
,
203 xfs_iflags_clear(XFS_I(dentry
->d_inode
), XFS_ITRUNCATED
);
204 return -xfs_fsync(XFS_I(dentry
->d_inode
));
208 * Unfortunately we can't just use the clean and simple readdir implementation
209 * below, because nfs might call back into ->lookup from the filldir callback
210 * and that will deadlock the low-level btree code.
212 * Hopefully we'll find a better workaround that allows to use the optimal
213 * version at least for local readdirs for 2.6.25.
222 struct inode
*inode
= filp
->f_path
.dentry
->d_inode
;
223 xfs_inode_t
*ip
= XFS_I(inode
);
228 * The Linux API doesn't pass down the total size of the buffer
229 * we read into down to the filesystem. With the filldir concept
230 * it's not needed for correct information, but the XFS dir2 leaf
231 * code wants an estimate of the buffer size to calculate it's
232 * readahead window and size the buffers used for mapping to
235 * Try to give it an estimate that's good enough, maybe at some
236 * point we can change the ->readdir prototype to include the
239 bufsize
= (size_t)min_t(loff_t
, PAGE_SIZE
, inode
->i_size
);
241 error
= xfs_readdir(ip
, dirent
, bufsize
,
242 (xfs_off_t
*)&filp
->f_pos
, filldir
);
257 struct hack_callback
{
272 struct hack_callback
*buf
= __buf
;
273 struct hack_dirent
*de
= (struct hack_dirent
*)(buf
->dirent
+ buf
->used
);
276 reclen
= ALIGN(sizeof(struct hack_dirent
) + namlen
, sizeof(u64
));
277 if (buf
->used
+ reclen
> buf
->len
)
284 memcpy(de
->name
, name
, namlen
);
295 struct inode
*inode
= filp
->f_path
.dentry
->d_inode
;
296 xfs_inode_t
*ip
= XFS_I(inode
);
297 struct hack_callback buf
;
298 struct hack_dirent
*de
;
302 xfs_off_t start_offset
, curr_offset
, offset
;
305 * Try fairly hard to get memory
307 buf
.len
= PAGE_CACHE_SIZE
;
309 buf
.dirent
= kmalloc(buf
.len
, GFP_KERNEL
);
313 } while (buf
.len
>= 1024);
318 curr_offset
= filp
->f_pos
;
319 if (curr_offset
== 0x7fffffff)
322 offset
= filp
->f_pos
;
327 start_offset
= offset
;
330 error
= -xfs_readdir(ip
, &buf
, buf
.len
, &offset
,
332 if (error
|| offset
== start_offset
) {
338 de
= (struct hack_dirent
*)buf
.dirent
;
340 curr_offset
= de
->offset
/* & 0x7fffffff */;
341 if (filldir(dirent
, de
->name
, de
->namlen
,
342 curr_offset
& 0x7fffffff,
343 de
->ino
, de
->d_type
)) {
347 reclen
= ALIGN(sizeof(struct hack_dirent
) + de
->namlen
,
350 de
= (struct hack_dirent
*)((char *)de
+ reclen
);
357 filp
->f_pos
= offset
& 0x7fffffff;
359 filp
->f_pos
= curr_offset
;
370 struct vm_area_struct
*vma
)
372 vma
->vm_ops
= &xfs_file_vm_ops
;
373 vma
->vm_flags
|= VM_CAN_NONLINEAR
;
386 struct inode
*inode
= filp
->f_path
.dentry
->d_inode
;
388 error
= xfs_ioctl(XFS_I(inode
), filp
, 0, cmd
, (void __user
*)p
);
389 xfs_iflags_set(XFS_I(inode
), XFS_IMODIFIED
);
391 /* NOTE: some of the ioctl's return positive #'s as a
392 * byte count indicating success, such as
393 * readlink_by_handle. So we don't "sign flip"
394 * like most other routines. This means true
395 * errors need to be returned as a negative value.
401 xfs_file_ioctl_invis(
407 struct inode
*inode
= filp
->f_path
.dentry
->d_inode
;
409 error
= xfs_ioctl(XFS_I(inode
), filp
, IO_INVIS
, cmd
, (void __user
*)p
);
410 xfs_iflags_set(XFS_I(inode
), XFS_IMODIFIED
);
412 /* NOTE: some of the ioctl's return positive #'s as a
413 * byte count indicating success, such as
414 * readlink_by_handle. So we don't "sign flip"
415 * like most other routines. This means true
416 * errors need to be returned as a negative value.
422 * mmap()d file has taken write protection fault and is being made
423 * writable. We can set the page state up correctly for a writable
424 * page, which means we can do correct delalloc accounting (ENOSPC
425 * checking!) and unwritten extent mapping.
429 struct vm_area_struct
*vma
,
432 return block_page_mkwrite(vma
, page
, xfs_get_blocks
);
435 const struct file_operations xfs_file_operations
= {
436 .llseek
= generic_file_llseek
,
437 .read
= do_sync_read
,
438 .write
= do_sync_write
,
439 .aio_read
= xfs_file_aio_read
,
440 .aio_write
= xfs_file_aio_write
,
441 .splice_read
= xfs_file_splice_read
,
442 .splice_write
= xfs_file_splice_write
,
443 .unlocked_ioctl
= xfs_file_ioctl
,
445 .compat_ioctl
= xfs_file_compat_ioctl
,
447 .mmap
= xfs_file_mmap
,
448 .open
= xfs_file_open
,
449 .release
= xfs_file_release
,
450 .fsync
= xfs_file_fsync
,
451 #ifdef HAVE_FOP_OPEN_EXEC
452 .open_exec
= xfs_file_open_exec
,
456 const struct file_operations xfs_invis_file_operations
= {
457 .llseek
= generic_file_llseek
,
458 .read
= do_sync_read
,
459 .write
= do_sync_write
,
460 .aio_read
= xfs_file_aio_read_invis
,
461 .aio_write
= xfs_file_aio_write_invis
,
462 .splice_read
= xfs_file_splice_read_invis
,
463 .splice_write
= xfs_file_splice_write_invis
,
464 .unlocked_ioctl
= xfs_file_ioctl_invis
,
466 .compat_ioctl
= xfs_file_compat_invis_ioctl
,
468 .mmap
= xfs_file_mmap
,
469 .open
= xfs_file_open
,
470 .release
= xfs_file_release
,
471 .fsync
= xfs_file_fsync
,
475 const struct file_operations xfs_dir_file_operations
= {
476 .read
= generic_read_dir
,
477 .readdir
= xfs_file_readdir
,
478 .unlocked_ioctl
= xfs_file_ioctl
,
480 .compat_ioctl
= xfs_file_compat_ioctl
,
482 .fsync
= xfs_file_fsync
,
485 static struct vm_operations_struct xfs_file_vm_ops
= {
486 .fault
= filemap_fault
,
487 .page_mkwrite
= xfs_vm_page_mkwrite
,