summaryrefslogtreecommitdiff
path: root/ext2fs/truncate.c
blob: d55740b13097a64513cf587d4cd15f8c60540f27 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
/* File truncation

   Copyright (C) 1995, 1996 Free Software Foundation, Inc.

   Written by Miles Bader <miles@gnu.ai.mit.edu>

   This program is free software; you can redistribute it and/or
   modify it under the terms of the GNU General Public License as
   published by the Free Software Foundation; either version 2, or (at
   your option) any later version.

   This program is distributed in the hope that it will be useful, but
   WITHOUT ANY WARRANTY; without even the implied warranty of
   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
   General Public License for more details.

   You should have received a copy of the GNU General Public License
   along with this program; if not, write to the Free Software
   Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. */

#include "ext2fs.h"

#ifdef DONT_CACHE_MEMORY_OBJECTS
#define MAY_CACHE 0
#else
#define MAY_CACHE 1
#endif

/* ---------------------------------------------------------------- */

/* A sequence of blocks to be freed in NODE.  */
struct free_block_run
{
  block_t first_block;
  unsigned long num_blocks;
  struct node *node;
};

/* Initialize FBR, pointing to NODE.  */
static inline void
free_block_run_init (struct free_block_run *fbr, struct node *node)
{
  fbr->num_blocks = 0;
  fbr->node = node;
}

static inline void
_free_block_run_flush (struct free_block_run *fbr, unsigned long count)
{
  fbr->node->dn_stat.st_blocks -= count << log2_stat_blocks_per_fs_block;
  fbr->node->dn_stat_dirty = 1;
  ext2_free_blocks (fbr->first_block, count);
}

/* Add BLOCK to the list of blocks to be freed in FBR.  */
static inline void
free_block_run_add (struct free_block_run *fbr, block_t block)
{
  unsigned long count = fbr->num_blocks;
  if (count == 0)
    {
      fbr->first_block = block;
      fbr->num_blocks++;
    }
  else if (count > 0 && fbr->first_block == block - count)
    fbr->num_blocks++;
  else
    {
      _free_block_run_flush (fbr, count);
      fbr->first_block = block;
      fbr->num_blocks = 1;
    }
}

/* If *P is non-zero, set it to zero, and add the block it pointed to the
   list of blocks to be freed in FBR.  */
static inline void
free_block_run_free_ptr (struct free_block_run *fbr, block_t *p)
{
  block_t block = *p;
  if (block)
    {
      *p = 0;
      free_block_run_add (fbr, block);
    }
}

/* Free any blocks left in FBR, and cleanup any resources it's using.  */
static inline void
free_block_run_finish (struct free_block_run *fbr)
{
  unsigned long count = fbr->num_blocks;
  if (count > 0)
    _free_block_run_flush (fbr, count);
}

/* ---------------------------------------------------------------- */

/* Free any direct blocks starting with block END.  */
static void
trunc_direct (struct node *node, block_t end, struct free_block_run *fbr)
{
  block_t *blocks = node->dn->info.i_data;

  ext2_debug ("truncating direct blocks from %d", direct_block);

  while (end < EXT2_NDIR_BLOCKS)
    free_block_run_free_ptr (fbr, blocks + end++);
}

/* Free any blocks in NODE greater than or equal to END that are rooted in
   the indirect block *P; OFFSET should be the block position that *P
   corresponds to.  For each block pointer in *P that should be freed,
   FREE_BLOCK is called with a pointer to the entry for that block, and the
   index of the entry within *P.  If every block in *P is freed, then *P is
   set to 0, otherwise it is left alone.  */
static void
trunc_indirect (struct node *node, block_t end,
		block_t *p, block_t offset,
		void (*free_block)(block_t *p, unsigned index),
		struct free_block_run *fbr)
{
  if (*p)
    {
      unsigned index;
      int modified = 0, all_freed = 1;
      block_t *ind_bh = (block_t *)bptr (*p);
      unsigned first = end < offset ? 0 : end - offset;

      for (index = first; index < addr_per_block; index++)
	if (ind_bh[index])
	  {
	    (*free_block)(ind_bh + index, index);
	    if (ind_bh[index])
	      all_freed = 0;	/* Some descendent hasn't been freed.  */
	    else
	      modified = 1;
	  }

      if (first == 0 && all_freed)
	free_block_run_free_ptr (fbr, p);
      else if (modified)
	record_indir_poke (node, ind_bh);
    }
}

static void
trunc_single_indirect (struct node *node, block_t end,
		       block_t *p, block_t offset,
		       struct free_block_run *fbr)
{
  void free_block (block_t *p, unsigned index)
    {
      free_block_run_free_ptr (fbr, p);
    }
  trunc_indirect (node, end, p, offset, free_block, fbr);
}

static void
trunc_double_indirect (struct node *node, block_t end,
		       block_t *p, block_t offset,
		       struct free_block_run *fbr)
{
  void free_block (block_t *p, unsigned index)
    {
      block_t entry_offs = offset + (index * addr_per_block);
      trunc_single_indirect (node, end, p, entry_offs, fbr);
    }
  trunc_indirect (node, end, p, offset, free_block, fbr);
}

static void
trunc_triple_indirect (struct node *node, block_t end,
		       block_t *p, block_t offset,
		       struct free_block_run *fbr)
{
  void free_block (block_t *p, unsigned index)
    {
      block_t entry_offs = offset + (index * addr_per_block * addr_per_block);
      trunc_double_indirect (node, end, p, entry_offs, fbr);
    }
  trunc_indirect (node, end, p, offset, free_block, fbr);
}

/* ---------------------------------------------------------------- */

/* Write something to each page from START to END inclusive of memory
   object OBJ, but make sure the data doesns't actually change. */
static void
poke_pages (memory_object_t obj, vm_offset_t start, vm_offset_t end)
{
  while (start < end)
    {
      error_t err;
      vm_size_t len = 8 * vm_page_size;
      vm_address_t addr = 0;

      if (len > end - start)
	len = end - start;

      err = vm_map (mach_task_self (), &addr, len, 0, 1, obj, start, 0,
		    VM_PROT_WRITE|VM_PROT_READ, VM_PROT_READ|VM_PROT_WRITE, 0);
      if (!err)
	{
	  vm_address_t poke;
	  for (poke = addr; poke < addr + len; poke += vm_page_size)
	    *(volatile int *)poke = *(volatile int *)poke;
	  vm_deallocate (mach_task_self (), addr, len);
	}

      start += len;
    }
}

/* Flush all the data past the new size from the kernel.  Also force any
   delayed copies of this data to take place immediately.  (We are implicitly
   changing the data to zeros and doing it without the kernel's immediate
   knowledge; accordingl we must help out the kernel thusly.) */
static void
force_delayed_copies (struct node *node, off_t length)
{
  struct pager *pager;

  spin_lock (&node_to_page_lock);
  pager = node->dn->pager;
  if (pager)
    ports_port_ref (pager);
  spin_unlock (&node_to_page_lock);
 
  if (pager)
    {
      mach_port_t obj;
      
      pager_change_attributes (pager, MAY_CACHE, MEMORY_OBJECT_COPY_NONE, 1);
      obj = diskfs_get_filemap (node, VM_PROT_READ);
      poke_pages (obj, round_page (length), round_page (node->allocsize));
      mach_port_deallocate (mach_task_self (), obj);
      pager_flush_some (pager, round_page(length), node->allocsize - length, 1);
      ports_port_deref (pager);
    }
}

static void
enable_delayed_copies (struct node *node)
{
  struct pager *pager;

  spin_lock (&node_to_page_lock);
  pager = node->dn->pager;
  if (pager)
    ports_port_ref (pager);
  spin_unlock (&node_to_page_lock);

  if (pager)
    {
      pager_change_attributes (pager, MAY_CACHE, MEMORY_OBJECT_COPY_DELAY, 0);
      ports_port_deref (pager);
    }
}

/* ---------------------------------------------------------------- */

/* The user must define this function.  Truncate locked node NODE to be SIZE
   bytes long.  (If NODE is already less than or equal to SIZE bytes
   long, do nothing.)  If this is a symlink (and diskfs_shortcut_symlink
   is set) then this should clear the symlink, even if 
   diskfs_create_symlink_hook stores the link target elsewhere.  */
error_t
diskfs_truncate (struct node *node, off_t length)
{
  error_t err;
  off_t offset;

  diskfs_check_readonly ();
  assert (!diskfs_readonly);

  if (length >= node->dn_stat.st_size)
    return 0;

  /*
   * If the file is not being truncated to a block boundary, the
   * contents of the partial block following the end of the file must be
   * zeroed in case it ever becomes accessible again because of
   * subsequent file growth.
   */
  offset = length % block_size;
  if (offset > 0)
    {
      diskfs_node_rdwr (node, (void *)zeroblock, length, block_size - offset,
			1, 0, 0);
      diskfs_file_update (node, 1);
    }

  ext2_discard_prealloc(node);

  force_delayed_copies (node, length);

  rwlock_writer_lock (&node->dn->alloc_lock);

  RECORD_LAST (node->dn, LAST_TRUNCATE, length);

  /* Update the size on disk; fsck will finish freeing blocks if necessary
     should we crash. */
  node->dn_stat.st_size = length;
  node->dn_set_mtime = 1;
  node->dn_set_ctime = 1;
  diskfs_node_update (node, 1);

  err = diskfs_catch_exception();
  if (!err)
    {
      block_t end = boffs_block (round_block (length)), offs;
      block_t *bptrs = node->dn->info.i_data;
      struct free_block_run fbr;

      free_block_run_init (&fbr, node);

      trunc_direct (node, end, &fbr);

      offs = EXT2_NDIR_BLOCKS;
      trunc_single_indirect (node, end, bptrs + EXT2_IND_BLOCK, offs, &fbr);
      offs += addr_per_block;
      trunc_double_indirect (node, end, bptrs + EXT2_DIND_BLOCK, offs, &fbr);
      offs += addr_per_block * addr_per_block;
      trunc_triple_indirect (node, end, bptrs + EXT2_TIND_BLOCK, offs, &fbr);

      free_block_run_finish (&fbr);

      node->allocsize = round_block (length);

      /* Set our last_page_partially_writable to a pessimistic state -- it
	 won't hurt if is wrong.  */
      node->dn->last_page_partially_writable =
	trunc_page (node->allocsize) != node->allocsize;
    }

  node->dn_set_mtime = 1;
  node->dn_set_ctime = 1;
  node->dn_stat_dirty = 1;

  /* Now we can permit delayed copies again. */
  enable_delayed_copies (node);

  rwlock_writer_unlock (&node->dn->alloc_lock);

  return err;
}