1 /*
2  *  Copyright 2000-2002 by Hans Reiser, licensing governed by reiserfs/README
3  */
4 
5 /*
6  *  Written by Anatoly P. Pinchuk pap@namesys.botik.ru
7  *  Programm System Institute
8  *  Pereslavl-Zalessky Russia
9  */
10 
11 /*
12  *  This file contains functions dealing with S+tree
13  *
14  * B_IS_IN_TREE
15  * copy_short_key
16  * copy_item_head
17  * comp_short_keys
18  * comp_keys
19  * comp_cpu_keys
20  * comp_short_le_keys
21  * comp_short_cpu_keys
22  * cpu_key2cpu_key
23  * le_key2cpu_key
24  * comp_le_keys
25  * bin_search
26  * get_lkey
27  * get_rkey
28  * key_in_buffer
29  * decrement_bcount
30  * decrement_counters_in_path
31  * reiserfs_check_path
32  * pathrelse_and_restore
33  * pathrelse
34  * search_by_key_reada
35  * search_by_key
36  * search_for_position_by_key
37  * comp_items
38  * prepare_for_direct_item
39  * prepare_for_direntry_item
40  * prepare_for_delete_or_cut
41  * calc_deleted_bytes_number
42  * init_tb_struct
43  * padd_item
44  * reiserfs_delete_item
45  * reiserfs_delete_solid_item
46  * reiserfs_delete_object
47  * maybe_indirect_to_direct
48  * indirect_to_direct_roll_back
49  * reiserfs_cut_from_item
50  * truncate_directory
51  * reiserfs_do_truncate
52  * reiserfs_paste_into_item
53  * reiserfs_insert_item
54  */
55 
56 #include <linux/config.h>
57 #include <linux/sched.h>
58 #include <linux/string.h>
59 #include <linux/locks.h>
60 #include <linux/pagemap.h>
61 #include <linux/reiserfs_fs.h>
62 #include <linux/smp_lock.h>
63 
64 /* Does the buffer contain a disk block which is in the tree. */
B_IS_IN_TREE(const struct buffer_head * p_s_bh)65 inline int B_IS_IN_TREE (const struct buffer_head * p_s_bh)
66 {
67 
68   RFALSE( B_LEVEL (p_s_bh) > MAX_HEIGHT,
69 	  "PAP-1010: block (%b) has too big level (%z)", p_s_bh, p_s_bh);
70 
71   return ( B_LEVEL (p_s_bh) != FREE_LEVEL );
72 }
73 
74 
75 
76 
copy_short_key(void * to,const void * from)77 inline void copy_short_key (void * to, const void * from)
78 {
79     memcpy (to, from, SHORT_KEY_SIZE);
80 }
81 
82 //
83 // to gets item head in le form
84 //
copy_item_head(struct item_head * p_v_to,const struct item_head * p_v_from)85 inline void copy_item_head(struct item_head * p_v_to,
86 			   const struct item_head * p_v_from)
87 {
88   memcpy (p_v_to, p_v_from, IH_SIZE);
89 }
90 
91 
92 /* k1 is pointer to on-disk structure which is stored in little-endian
93    form. k2 is pointer to cpu variable. For key of items of the same
94    object this returns 0.
95    Returns: -1 if key1 < key2
96    0 if key1 == key2
97    1 if key1 > key2 */
comp_short_keys(const struct key * le_key,const struct cpu_key * cpu_key)98 inline int  comp_short_keys (const struct key * le_key,
99 			     const struct cpu_key * cpu_key)
100 {
101   __u32 * p_s_le_u32, * p_s_cpu_u32;
102   int n_key_length = REISERFS_SHORT_KEY_LEN;
103 
104   p_s_le_u32 = (__u32 *)le_key;
105   p_s_cpu_u32 = (__u32 *)cpu_key;
106   for( ; n_key_length--; ++p_s_le_u32, ++p_s_cpu_u32 ) {
107     if ( le32_to_cpu (*p_s_le_u32) < *p_s_cpu_u32 )
108       return -1;
109     if ( le32_to_cpu (*p_s_le_u32) > *p_s_cpu_u32 )
110       return 1;
111   }
112 
113   return 0;
114 }
115 
116 
117 /* k1 is pointer to on-disk structure which is stored in little-endian
118    form. k2 is pointer to cpu variable.
119    Compare keys using all 4 key fields.
120    Returns: -1 if key1 < key2 0
121    if key1 = key2 1 if key1 > key2 */
comp_keys(const struct key * le_key,const struct cpu_key * cpu_key)122 inline int  comp_keys (const struct key * le_key, const struct cpu_key * cpu_key)
123 {
124   int retval;
125 
126   retval = comp_short_keys (le_key, cpu_key);
127   if (retval)
128       return retval;
129   if (le_key_k_offset (le_key_version(le_key), le_key) < cpu_key_k_offset (cpu_key))
130       return -1;
131   if (le_key_k_offset (le_key_version(le_key), le_key) > cpu_key_k_offset (cpu_key))
132       return 1;
133 
134   if (cpu_key->key_length == 3)
135       return 0;
136 
137   /* this part is needed only when tail conversion is in progress */
138   if (le_key_k_type (le_key_version(le_key), le_key) < cpu_key_k_type (cpu_key))
139     return -1;
140 
141   if (le_key_k_type (le_key_version(le_key), le_key) > cpu_key_k_type (cpu_key))
142     return 1;
143 
144   return 0;
145 }
146 
147 
148 //
149 // FIXME: not used yet
150 //
comp_cpu_keys(const struct cpu_key * key1,const struct cpu_key * key2)151 inline int comp_cpu_keys (const struct cpu_key * key1,
152 			  const struct cpu_key * key2)
153 {
154     if (key1->on_disk_key.k_dir_id < key2->on_disk_key.k_dir_id)
155 	return -1;
156     if (key1->on_disk_key.k_dir_id > key2->on_disk_key.k_dir_id)
157 	return 1;
158 
159     if (key1->on_disk_key.k_objectid < key2->on_disk_key.k_objectid)
160 	return -1;
161     if (key1->on_disk_key.k_objectid > key2->on_disk_key.k_objectid)
162 	return 1;
163 
164     if (cpu_key_k_offset (key1) < cpu_key_k_offset (key2))
165 	return -1;
166     if (cpu_key_k_offset (key1) > cpu_key_k_offset (key2))
167 	return 1;
168 
169     reiserfs_warning (NULL, "comp_cpu_keys: type are compared for %K and %K\n",
170 		      key1, key2);
171 
172     if (cpu_key_k_type (key1) < cpu_key_k_type (key2))
173 	return -1;
174     if (cpu_key_k_type (key1) > cpu_key_k_type (key2))
175 	return 1;
176     return 0;
177 }
178 
comp_short_le_keys(const struct key * key1,const struct key * key2)179 inline int comp_short_le_keys (const struct key * key1, const struct key * key2)
180 {
181   __u32 * p_s_1_u32, * p_s_2_u32;
182   int n_key_length = REISERFS_SHORT_KEY_LEN;
183 
184   p_s_1_u32 = (__u32 *)key1;
185   p_s_2_u32 = (__u32 *)key2;
186   for( ; n_key_length--; ++p_s_1_u32, ++p_s_2_u32 ) {
187     if ( le32_to_cpu (*p_s_1_u32) < le32_to_cpu (*p_s_2_u32) )
188       return -1;
189     if ( le32_to_cpu (*p_s_1_u32) > le32_to_cpu (*p_s_2_u32) )
190       return 1;
191   }
192   return 0;
193 }
194 
comp_short_cpu_keys(const struct cpu_key * key1,const struct cpu_key * key2)195 inline int comp_short_cpu_keys (const struct cpu_key * key1,
196 				const struct cpu_key * key2)
197 {
198   __u32 * p_s_1_u32, * p_s_2_u32;
199   int n_key_length = REISERFS_SHORT_KEY_LEN;
200 
201   p_s_1_u32 = (__u32 *)key1;
202   p_s_2_u32 = (__u32 *)key2;
203 
204   for( ; n_key_length--; ++p_s_1_u32, ++p_s_2_u32 ) {
205     if ( *p_s_1_u32 < *p_s_2_u32 )
206       return -1;
207     if ( *p_s_1_u32 > *p_s_2_u32 )
208       return 1;
209   }
210   return 0;
211 }
212 
213 
214 
cpu_key2cpu_key(struct cpu_key * to,const struct cpu_key * from)215 inline void cpu_key2cpu_key (struct cpu_key * to, const struct cpu_key * from)
216 {
217     memcpy (to, from, sizeof (struct cpu_key));
218 }
219 
220 
le_key2cpu_key(struct cpu_key * to,const struct key * from)221 inline void le_key2cpu_key (struct cpu_key * to, const struct key * from)
222 {
223     to->on_disk_key.k_dir_id = le32_to_cpu (from->k_dir_id);
224     to->on_disk_key.k_objectid = le32_to_cpu (from->k_objectid);
225 
226     // find out version of the key
227     to->version = le_key_version (from);
228     if (to->version == KEY_FORMAT_3_5) {
229 	to->on_disk_key.u.k_offset_v1.k_offset = le32_to_cpu (from->u.k_offset_v1.k_offset);
230 	to->on_disk_key.u.k_offset_v1.k_uniqueness = le32_to_cpu (from->u.k_offset_v1.k_uniqueness);
231     } else {
232 	to->on_disk_key.u.k_offset_v2.k_offset = offset_v2_k_offset(&from->u.k_offset_v2);
233 	to->on_disk_key.u.k_offset_v2.k_type = offset_v2_k_type(&from->u.k_offset_v2);
234     }
235 }
236 
237 
238 
239 // this does not say which one is bigger, it only returns 1 if keys
240 // are not equal, 0 otherwise
comp_le_keys(const struct key * k1,const struct key * k2)241 inline int comp_le_keys (const struct key * k1, const struct key * k2)
242 {
243     return memcmp (k1, k2, sizeof (struct key));
244 }
245 
246 /**************************************************************************
247  *  Binary search toolkit function                                        *
248  *  Search for an item in the array by the item key                       *
249  *  Returns:    1 if found,  0 if not found;                              *
250  *        *p_n_pos = number of the searched element if found, else the    *
251  *        number of the first element that is larger than p_v_key.        *
252  **************************************************************************/
253 /* For those not familiar with binary search: n_lbound is the leftmost item that it
254  could be, n_rbound the rightmost item that it could be.  We examine the item
255  halfway between n_lbound and n_rbound, and that tells us either that we can increase
256  n_lbound, or decrease n_rbound, or that we have found it, or if n_lbound <= n_rbound that
257  there are no possible items, and we have not found it. With each examination we
258  cut the number of possible items it could be by one more than half rounded down,
259  or we find it. */
bin_search(const void * p_v_key,const void * p_v_base,int p_n_num,int p_n_width,int * p_n_pos)260 inline	int bin_search (
261               const void * p_v_key, /* Key to search for.                   */
262 	      const void * p_v_base,/* First item in the array.             */
263 	      int       p_n_num,    /* Number of items in the array.        */
264 	      int       p_n_width,  /* Item size in the array.
265 				       searched. Lest the reader be
266 				       confused, note that this is crafted
267 				       as a general function, and when it
268 				       is applied specifically to the array
269 				       of item headers in a node, p_n_width
270 				       is actually the item header size not
271 				       the item size.                      */
272 	      int     * p_n_pos     /* Number of the searched for element. */
273             ) {
274     int   n_rbound, n_lbound, n_j;
275 
276    for ( n_j = ((n_rbound = p_n_num - 1) + (n_lbound = 0))/2; n_lbound <= n_rbound; n_j = (n_rbound + n_lbound)/2 )
277      switch( COMP_KEYS((struct key *)((char * )p_v_base + n_j * p_n_width), (struct cpu_key *)p_v_key) )  {
278      case -1: n_lbound = n_j + 1; continue;
279      case  1: n_rbound = n_j - 1; continue;
280      case  0: *p_n_pos = n_j;     return ITEM_FOUND; /* Key found in the array.  */
281         }
282 
283     /* bin_search did not find given key, it returns position of key,
284         that is minimal and greater than the given one. */
285     *p_n_pos = n_lbound;
286     return ITEM_NOT_FOUND;
287 }
288 
289 #ifdef CONFIG_REISERFS_CHECK
290 extern struct tree_balance * cur_tb;
291 #endif
292 
293 
294 
295 /* Minimal possible key. It is never in the tree. */
296 const struct key  MIN_KEY = {0, 0, {{0, 0},}};
297 
298 /* Maximal possible key. It is never in the tree. */
299 const struct key  MAX_KEY = {0xffffffff, 0xffffffff, {{0xffffffff, 0xffffffff},}};
300 
301 
302 /* Get delimiting key of the buffer by looking for it in the buffers in the path, starting from the bottom
303    of the path, and going upwards.  We must check the path's validity at each step.  If the key is not in
304    the path, there is no delimiting key in the tree (buffer is first or last buffer in tree), and in this
305    case we return a special key, either MIN_KEY or MAX_KEY. */
get_lkey(const struct path * p_s_chk_path,const struct super_block * p_s_sb)306 inline	const struct  key * get_lkey  (
307 	                const struct path         * p_s_chk_path,
308                         const struct super_block  * p_s_sb
309                       ) {
310   int                   n_position, n_path_offset = p_s_chk_path->path_length;
311   struct buffer_head  * p_s_parent;
312 
313   RFALSE( n_path_offset < FIRST_PATH_ELEMENT_OFFSET,
314 	  "PAP-5010: illegal offset in the path");
315 
316   /* While not higher in path than first element. */
317   while ( n_path_offset-- > FIRST_PATH_ELEMENT_OFFSET ) {
318 
319     RFALSE( ! buffer_uptodate(PATH_OFFSET_PBUFFER(p_s_chk_path, n_path_offset)),
320 	    "PAP-5020: parent is not uptodate");
321 
322     /* Parent at the path is not in the tree now. */
323     if ( ! B_IS_IN_TREE(p_s_parent = PATH_OFFSET_PBUFFER(p_s_chk_path, n_path_offset)) )
324       return &MAX_KEY;
325     /* Check whether position in the parent is correct. */
326     if ( (n_position = PATH_OFFSET_POSITION(p_s_chk_path, n_path_offset)) > B_NR_ITEMS(p_s_parent) )
327        return &MAX_KEY;
328     /* Check whether parent at the path really points to the child. */
329     if ( B_N_CHILD_NUM(p_s_parent, n_position) !=
330 	 PATH_OFFSET_PBUFFER(p_s_chk_path, n_path_offset + 1)->b_blocknr )
331       return &MAX_KEY;
332     /* Return delimiting key if position in the parent is not equal to zero. */
333     if ( n_position )
334       return B_N_PDELIM_KEY(p_s_parent, n_position - 1);
335   }
336   /* Return MIN_KEY if we are in the root of the buffer tree. */
337   if ( PATH_OFFSET_PBUFFER(p_s_chk_path, FIRST_PATH_ELEMENT_OFFSET)->b_blocknr ==
338        SB_ROOT_BLOCK (p_s_sb) )
339     return &MIN_KEY;
340   return  &MAX_KEY;
341 }
342 
343 
344 /* Get delimiting key of the buffer at the path and its right neighbor. */
get_rkey(const struct path * p_s_chk_path,const struct super_block * p_s_sb)345 inline	const struct  key * get_rkey  (
346 	                const struct path         * p_s_chk_path,
347                         const struct super_block  * p_s_sb
348                       ) {
349   int                   n_position,
350     			n_path_offset = p_s_chk_path->path_length;
351   struct buffer_head  * p_s_parent;
352 
353   RFALSE( n_path_offset < FIRST_PATH_ELEMENT_OFFSET,
354 	  "PAP-5030: illegal offset in the path");
355 
356   while ( n_path_offset-- > FIRST_PATH_ELEMENT_OFFSET ) {
357 
358     RFALSE( ! buffer_uptodate(PATH_OFFSET_PBUFFER(p_s_chk_path, n_path_offset)),
359 	    "PAP-5040: parent is not uptodate");
360 
361     /* Parent at the path is not in the tree now. */
362     if ( ! B_IS_IN_TREE(p_s_parent = PATH_OFFSET_PBUFFER(p_s_chk_path, n_path_offset)) )
363       return &MIN_KEY;
364     /* Check whether position in the parent is correct. */
365     if ( (n_position = PATH_OFFSET_POSITION(p_s_chk_path, n_path_offset)) > B_NR_ITEMS(p_s_parent) )
366       return &MIN_KEY;
367     /* Check whether parent at the path really points to the child. */
368     if ( B_N_CHILD_NUM(p_s_parent, n_position) !=
369                                         PATH_OFFSET_PBUFFER(p_s_chk_path, n_path_offset + 1)->b_blocknr )
370       return &MIN_KEY;
371     /* Return delimiting key if position in the parent is not the last one. */
372     if ( n_position != B_NR_ITEMS(p_s_parent) )
373       return B_N_PDELIM_KEY(p_s_parent, n_position);
374   }
375   /* Return MAX_KEY if we are in the root of the buffer tree. */
376   if ( PATH_OFFSET_PBUFFER(p_s_chk_path, FIRST_PATH_ELEMENT_OFFSET)->b_blocknr ==
377        SB_ROOT_BLOCK (p_s_sb) )
378     return &MAX_KEY;
379   return  &MIN_KEY;
380 }
381 
382 
383 /* Check whether a key is contained in the tree rooted from a buffer at a path. */
384 /* This works by looking at the left and right delimiting keys for the buffer in the last path_element in
385    the path.  These delimiting keys are stored at least one level above that buffer in the tree. If the
386    buffer is the first or last node in the tree order then one of the delimiting keys may be absent, and in
387    this case get_lkey and get_rkey return a special key which is MIN_KEY or MAX_KEY. */
key_in_buffer(struct path * p_s_chk_path,const struct cpu_key * p_s_key,struct super_block * p_s_sb)388 static  inline  int key_in_buffer (
389                       struct path         * p_s_chk_path, /* Path which should be checked.  */
390                       const struct cpu_key      * p_s_key,      /* Key which should be checked.   */
391                       struct super_block  * p_s_sb        /* Super block pointer.           */
392 		      ) {
393 
394   RFALSE( ! p_s_key || p_s_chk_path->path_length < FIRST_PATH_ELEMENT_OFFSET ||
395 	  p_s_chk_path->path_length > MAX_HEIGHT,
396 	  "PAP-5050: pointer to the key(%p) is NULL or illegal path length(%d)",
397 	  p_s_key, p_s_chk_path->path_length);
398   RFALSE( PATH_PLAST_BUFFER(p_s_chk_path)->b_dev == NODEV,
399 	  "PAP-5060: device must not be NODEV");
400 
401   if ( COMP_KEYS(get_lkey(p_s_chk_path, p_s_sb), p_s_key) == 1 )
402     /* left delimiting key is bigger, that the key we look for */
403     return 0;
404   //  if ( COMP_KEYS(p_s_key, get_rkey(p_s_chk_path, p_s_sb)) != -1 )
405   if ( COMP_KEYS(get_rkey(p_s_chk_path, p_s_sb), p_s_key) != 1 )
406     /* p_s_key must be less than right delimitiing key */
407     return 0;
408   return 1;
409 }
410 
411 
decrement_bcount(struct buffer_head * p_s_bh)412 inline void decrement_bcount(
413               struct buffer_head  * p_s_bh
414             ) {
415   if ( p_s_bh ) {
416     if ( atomic_read (&(p_s_bh->b_count)) ) {
417       put_bh(p_s_bh) ;
418       return;
419     }
420     reiserfs_panic(NULL, "PAP-5070: decrement_bcount: trying to free free buffer %b", p_s_bh);
421   }
422 }
423 
424 
425 /* Decrement b_count field of the all buffers in the path. */
decrement_counters_in_path(struct path * p_s_search_path)426 void decrement_counters_in_path (
427               struct path * p_s_search_path
428             ) {
429   int n_path_offset = p_s_search_path->path_length;
430 
431   RFALSE( n_path_offset < ILLEGAL_PATH_ELEMENT_OFFSET ||
432 	  n_path_offset > EXTENDED_MAX_HEIGHT - 1,
433 	  "PAP-5080: illegal path offset of %d", n_path_offset);
434 
435   while ( n_path_offset > ILLEGAL_PATH_ELEMENT_OFFSET ) {
436     struct buffer_head * bh;
437 
438     bh = PATH_OFFSET_PBUFFER(p_s_search_path, n_path_offset--);
439     decrement_bcount (bh);
440   }
441   p_s_search_path->path_length = ILLEGAL_PATH_ELEMENT_OFFSET;
442 }
443 
444 
reiserfs_check_path(struct path * p)445 int reiserfs_check_path(struct path *p) {
446   RFALSE( p->path_length != ILLEGAL_PATH_ELEMENT_OFFSET,
447 	  "path not properly relsed") ;
448   return 0 ;
449 }
450 
451 
452 /* Release all buffers in the path. Restore dirty bits clean
453 ** when preparing the buffer for the log
454 **
455 ** only called from fix_nodes()
456 */
pathrelse_and_restore(struct super_block * s,struct path * p_s_search_path)457 void  pathrelse_and_restore (
458 	struct super_block *s,
459         struct path * p_s_search_path
460       ) {
461   int n_path_offset = p_s_search_path->path_length;
462 
463   RFALSE( n_path_offset < ILLEGAL_PATH_ELEMENT_OFFSET,
464 	  "clm-4000: illegal path offset");
465 
466   while ( n_path_offset > ILLEGAL_PATH_ELEMENT_OFFSET )  {
467     reiserfs_restore_prepared_buffer(s, PATH_OFFSET_PBUFFER(p_s_search_path,
468                                      n_path_offset));
469     brelse(PATH_OFFSET_PBUFFER(p_s_search_path, n_path_offset--));
470   }
471   p_s_search_path->path_length = ILLEGAL_PATH_ELEMENT_OFFSET;
472 }
473 
474 /* Release all buffers in the path. */
pathrelse(struct path * p_s_search_path)475 void  pathrelse (
476         struct path * p_s_search_path
477       ) {
478   int n_path_offset = p_s_search_path->path_length;
479 
480   RFALSE( n_path_offset < ILLEGAL_PATH_ELEMENT_OFFSET,
481 	  "PAP-5090: illegal path offset");
482 
483   while ( n_path_offset > ILLEGAL_PATH_ELEMENT_OFFSET )
484     brelse(PATH_OFFSET_PBUFFER(p_s_search_path, n_path_offset--));
485 
486   p_s_search_path->path_length = ILLEGAL_PATH_ELEMENT_OFFSET;
487 }
488 
489 
490 
is_leaf(char * buf,int blocksize,struct buffer_head * bh)491 static int is_leaf (char * buf, int blocksize, struct buffer_head * bh)
492 {
493     struct block_head * blkh;
494     struct item_head * ih;
495     int used_space;
496     int prev_location;
497     int i;
498     int nr;
499 
500     blkh = (struct block_head *)buf;
501     if ( blkh_level(blkh) != DISK_LEAF_NODE_LEVEL) {
502 	printk ("is_leaf: this should be caught earlier\n");
503 	return 0;
504     }
505 
506     nr = blkh_nr_item(blkh);
507     if (nr < 1 || nr > ((blocksize - BLKH_SIZE) / (IH_SIZE + MIN_ITEM_LEN))) {
508 	/* item number is too big or too small */
509 	reiserfs_warning (NULL, "is_leaf: nr_item seems wrong: %z\n", bh);
510 	return 0;
511     }
512     ih = (struct item_head *)(buf + BLKH_SIZE) + nr - 1;
513     used_space = BLKH_SIZE + IH_SIZE * nr + (blocksize - ih_location (ih));
514     if (used_space != blocksize - blkh_free_space(blkh)) {
515 	/* free space does not match to calculated amount of use space */
516 	reiserfs_warning (NULL, "is_leaf: free space seems wrong: %z\n", bh);
517 	return 0;
518     }
519 
520     // FIXME: it is_leaf will hit performance too much - we may have
521     // return 1 here
522 
523     /* check tables of item heads */
524     ih = (struct item_head *)(buf + BLKH_SIZE);
525     prev_location = blocksize;
526     for (i = 0; i < nr; i ++, ih ++) {
527 	if ( le_ih_k_type(ih) == TYPE_ANY) {
528 	    reiserfs_warning (NULL, "is_leaf: wrong item type for item %h\n",ih);
529 	    return 0;
530 	}
531 	if (ih_location (ih) >= blocksize || ih_location (ih) < IH_SIZE * nr) {
532 	    reiserfs_warning (NULL, "is_leaf: item location seems wrong: %h\n", ih);
533 	    return 0;
534 	}
535 	if (ih_item_len (ih) < 1 || ih_item_len (ih) > MAX_ITEM_LEN (blocksize)) {
536 	    reiserfs_warning (NULL, "is_leaf: item length seems wrong: %h\n", ih);
537 	    return 0;
538 	}
539 	if (prev_location - ih_location (ih) != ih_item_len (ih)) {
540 	    reiserfs_warning (NULL, "is_leaf: item location seems wrong (second one): %h\n", ih);
541 	    return 0;
542 	}
543 	prev_location = ih_location (ih);
544     }
545 
546     // one may imagine much more checks
547     return 1;
548 }
549 
550 
551 /* returns 1 if buf looks like an internal node, 0 otherwise */
is_internal(char * buf,int blocksize,struct buffer_head * bh)552 static int is_internal (char * buf, int blocksize, struct buffer_head * bh)
553 {
554     struct block_head * blkh;
555     int nr;
556     int used_space;
557 
558     blkh = (struct block_head *)buf;
559     nr = blkh_level(blkh);
560     if (nr <= DISK_LEAF_NODE_LEVEL || nr > MAX_HEIGHT) {
561 	/* this level is not possible for internal nodes */
562 	printk ("is_internal: this should be caught earlier\n");
563 	return 0;
564     }
565 
566     nr = blkh_nr_item(blkh);
567     if (nr > (blocksize - BLKH_SIZE - DC_SIZE) / (KEY_SIZE + DC_SIZE)) {
568 	/* for internal which is not root we might check min number of keys */
569 	reiserfs_warning (NULL, "is_internal: number of key seems wrong: %z\n", bh);
570 	return 0;
571     }
572 
573     used_space = BLKH_SIZE + KEY_SIZE * nr + DC_SIZE * (nr + 1);
574     if (used_space != blocksize - blkh_free_space(blkh)) {
575 	reiserfs_warning (NULL, "is_internal: free space seems wrong: %z\n", bh);
576 	return 0;
577     }
578 
579     // one may imagine much more checks
580     return 1;
581 }
582 
583 
584 // make sure that bh contains formatted node of reiserfs tree of
585 // 'level'-th level
is_tree_node(struct buffer_head * bh,int level)586 static int is_tree_node (struct buffer_head * bh, int level)
587 {
588     if (B_LEVEL (bh) != level) {
589 	printk ("is_tree_node: node level %d does not match to the expected one %d\n",
590 		B_LEVEL (bh), level);
591 	return 0;
592     }
593     if (level == DISK_LEAF_NODE_LEVEL)
594 	return is_leaf (bh->b_data, bh->b_size, bh);
595 
596     return is_internal (bh->b_data, bh->b_size, bh);
597 }
598 
599 
600 
601 #ifdef SEARCH_BY_KEY_READA
602 
603 /* The function is NOT SCHEDULE-SAFE! */
search_by_key_reada(struct super_block * s,int blocknr)604 static void search_by_key_reada (struct super_block * s, int blocknr)
605 {
606     struct buffer_head * bh;
607 
608     if (blocknr == 0)
609 	return;
610 
611     bh = getblk (s->s_dev, blocknr, s->s_blocksize);
612 
613     if (!buffer_uptodate (bh)) {
614 	ll_rw_block (READA, 1, &bh);
615     }
616     bh->b_count --;
617 }
618 
619 #endif
620 
621 /**************************************************************************
622  * Algorithm   SearchByKey                                                *
623  *             look for item in the Disk S+Tree by its key                *
624  * Input:  p_s_sb   -  super block                                        *
625  *         p_s_key  - pointer to the key to search                        *
626  * Output: ITEM_FOUND, ITEM_NOT_FOUND or IO_ERROR                         *
627  *         p_s_search_path - path from the root to the needed leaf        *
628  **************************************************************************/
629 
630 /* This function fills up the path from the root to the leaf as it
631    descends the tree looking for the key.  It uses reiserfs_bread to
632    try to find buffers in the cache given their block number.  If it
633    does not find them in the cache it reads them from disk.  For each
634    node search_by_key finds using reiserfs_bread it then uses
635    bin_search to look through that node.  bin_search will find the
636    position of the block_number of the next node if it is looking
637    through an internal node.  If it is looking through a leaf node
638    bin_search will find the position of the item which has key either
639    equal to given key, or which is the maximal key less than the given
640    key.  search_by_key returns a path that must be checked for the
641    correctness of the top of the path but need not be checked for the
642    correctness of the bottom of the path */
643 /* The function is NOT SCHEDULE-SAFE! */
search_by_key(struct super_block * p_s_sb,const struct cpu_key * p_s_key,struct path * p_s_search_path,int n_stop_level)644 int search_by_key (struct super_block * p_s_sb,
645 		   const struct cpu_key * p_s_key, /* Key to search. */
646 		   struct path * p_s_search_path, /* This structure was
647 						     allocated and initialized
648 						     by the calling
649 						     function. It is filled up
650 						     by this function.  */
651 		   int n_stop_level /* How far down the tree to search. To
652                                        stop at leaf level - set to
653                                        DISK_LEAF_NODE_LEVEL */
654     ) {
655     int  n_block_number = SB_ROOT_BLOCK (p_s_sb),
656       expected_level = SB_TREE_HEIGHT (p_s_sb),
657       n_block_size    = p_s_sb->s_blocksize;
658     struct buffer_head  *       p_s_bh;
659     struct path_element *       p_s_last_element;
660     int				n_node_level, n_retval;
661     int 			right_neighbor_of_leaf_node;
662     int				fs_gen;
663 
664 #ifdef CONFIG_REISERFS_CHECK
665     int n_repeat_counter = 0;
666 #endif
667 
668     PROC_INFO_INC( p_s_sb, search_by_key );
669 
670     /* As we add each node to a path we increase its count.  This means that
671        we must be careful to release all nodes in a path before we either
672        discard the path struct or re-use the path struct, as we do here. */
673 
674     decrement_counters_in_path(p_s_search_path);
675 
676     right_neighbor_of_leaf_node = 0;
677 
678     /* With each iteration of this loop we search through the items in the
679        current node, and calculate the next current node(next path element)
680        for the next iteration of this loop.. */
681     while ( 1 ) {
682 
683 #ifdef CONFIG_REISERFS_CHECK
684 	if ( !(++n_repeat_counter % 50000) )
685 	    reiserfs_warning (p_s_sb, "PAP-5100: search_by_key: %s:"
686 			      "there were %d iterations of while loop "
687 			      "looking for key %K\n",
688 			      current->comm, n_repeat_counter, p_s_key);
689 #endif
690 
691 	/* prep path to have another element added to it. */
692 	p_s_last_element = PATH_OFFSET_PELEMENT(p_s_search_path, ++p_s_search_path->path_length);
693 	fs_gen = get_generation (p_s_sb);
694 	expected_level --;
695 
696 #ifdef SEARCH_BY_KEY_READA
697 	/* schedule read of right neighbor */
698 	search_by_key_reada (p_s_sb, right_neighbor_of_leaf_node);
699 #endif
700 
701 	/* Read the next tree node, and set the last element in the path to
702            have a pointer to it. */
703 	if ( ! (p_s_bh = p_s_last_element->pe_buffer =
704 		reiserfs_bread(p_s_sb, n_block_number, n_block_size)) ) {
705 	    p_s_search_path->path_length --;
706 	    pathrelse(p_s_search_path);
707 	    return IO_ERROR;
708 	}
709 
710  	if( fs_changed (fs_gen, p_s_sb) ) {
711  		PROC_INFO_INC( p_s_sb, search_by_key_fs_changed );
712  		PROC_INFO_INC( p_s_sb, sbk_fs_changed[ expected_level - 1 ] );
713  	}
714 
715 	/* It is possible that schedule occurred. We must check whether the key
716 	   to search is still in the tree rooted from the current buffer. If
717 	   not then repeat search from the root. */
718 	if ( fs_changed (fs_gen, p_s_sb) &&
719 	     (!B_IS_IN_TREE (p_s_bh) || !key_in_buffer(p_s_search_path, p_s_key, p_s_sb)) ) {
720  	    PROC_INFO_INC( p_s_sb, search_by_key_restarted );
721 	    PROC_INFO_INC( p_s_sb, sbk_restarted[ expected_level - 1 ] );
722 	    decrement_counters_in_path(p_s_search_path);
723 
724 	    /* Get the root block number so that we can repeat the search
725                starting from the root. */
726 	    n_block_number = SB_ROOT_BLOCK (p_s_sb);
727 	    expected_level = SB_TREE_HEIGHT (p_s_sb);
728 	    right_neighbor_of_leaf_node = 0;
729 
730 	    /* repeat search from the root */
731 	    continue;
732 	}
733 
734         /* only check that the key is in the buffer if p_s_key is not
735            equal to the MAX_KEY. Latter case is only possible in
736            "finish_unfinished()" processing during mount. */
737         RFALSE( COMP_KEYS( &MAX_KEY, p_s_key ) &&
738                 ! key_in_buffer(p_s_search_path, p_s_key, p_s_sb),
739 		"PAP-5130: key is not in the buffer");
740 #ifdef CONFIG_REISERFS_CHECK
741 	if ( cur_tb ) {
742 	    print_cur_tb ("5140");
743 	    reiserfs_panic(p_s_sb, "PAP-5140: search_by_key: schedule occurred in do_balance!");
744 	}
745 #endif
746 
747 	// make sure, that the node contents look like a node of
748 	// certain level
749 	if (!is_tree_node (p_s_bh, expected_level)) {
750 	    reiserfs_warning (p_s_sb, "vs-5150: search_by_key: "
751 			      "invalid format found in block %ld. Fsck?\n",
752 			      p_s_bh->b_blocknr);
753 	    pathrelse (p_s_search_path);
754 	    return IO_ERROR;
755 	}
756 
757 	/* ok, we have acquired next formatted node in the tree */
758 	n_node_level = B_LEVEL (p_s_bh);
759 
760 	PROC_INFO_BH_STAT( p_s_sb, p_s_bh, n_node_level - 1 );
761 
762 	RFALSE( n_node_level < n_stop_level,
763 		"vs-5152: tree level (%d) is less than stop level (%d)",
764 		n_node_level, n_stop_level);
765 
766 	n_retval = bin_search( p_s_key, B_N_PITEM_HEAD(p_s_bh, 0),
767                 B_NR_ITEMS(p_s_bh),
768                 ( n_node_level == DISK_LEAF_NODE_LEVEL ) ? IH_SIZE : KEY_SIZE,
769                 &(p_s_last_element->pe_position));
770 	if (n_node_level == n_stop_level) {
771 	    return n_retval;
772 	}
773 
774 	/* we are not in the stop level */
775 	if (n_retval == ITEM_FOUND)
776 	    /* item has been found, so we choose the pointer which is to the right of the found one */
777 	    p_s_last_element->pe_position++;
778 
779 	/* if item was not found we choose the position which is to
780 	   the left of the found item. This requires no code,
781 	   bin_search did it already.*/
782 
783 	/* So we have chosen a position in the current node which is
784 	   an internal node.  Now we calculate child block number by
785 	   position in the node. */
786 	n_block_number = B_N_CHILD_NUM(p_s_bh, p_s_last_element->pe_position);
787 
788 #ifdef SEARCH_BY_KEY_READA
789 	/* if we are going to read leaf node, then calculate its right neighbor if possible */
790 	if (n_node_level == DISK_LEAF_NODE_LEVEL + 1 && p_s_last_element->pe_position < B_NR_ITEMS (p_s_bh))
791 	    right_neighbor_of_leaf_node = B_N_CHILD_NUM(p_s_bh, p_s_last_element->pe_position + 1);
792 #endif
793     }
794 }
795 
796 
797 /* Form the path to an item and position in this item which contains
798    file byte defined by p_s_key. If there is no such item
799    corresponding to the key, we point the path to the item with
800    maximal key less than p_s_key, and *p_n_pos_in_item is set to one
801    past the last entry/byte in the item.  If searching for entry in a
802    directory item, and it is not found, *p_n_pos_in_item is set to one
803    entry more than the entry with maximal key which is less than the
804    sought key.
805 
806    Note that if there is no entry in this same node which is one more,
807    then we point to an imaginary entry.  for direct items, the
808    position is in units of bytes, for indirect items the position is
809    in units of blocknr entries, for directory items the position is in
810    units of directory entries.  */
811 
812 /* The function is NOT SCHEDULE-SAFE! */
search_for_position_by_key(struct super_block * p_s_sb,const struct cpu_key * p_cpu_key,struct path * p_s_search_path)813 int search_for_position_by_key (struct super_block  * p_s_sb,         /* Pointer to the super block.          */
814 				const struct cpu_key  * p_cpu_key,      /* Key to search (cpu variable)         */
815 				struct path         * p_s_search_path /* Filled up by this function.          */
816     ) {
817     struct item_head    * p_le_ih; /* pointer to on-disk structure */
818     int                   n_blk_size;
819     loff_t item_offset, offset;
820     struct reiserfs_dir_entry de;
821     int retval;
822 
823     /* If searching for directory entry. */
824     if ( is_direntry_cpu_key (p_cpu_key) )
825 	return  search_by_entry_key (p_s_sb, p_cpu_key, p_s_search_path, &de);
826 
827     /* If not searching for directory entry. */
828 
829     /* If item is found. */
830     retval = search_item (p_s_sb, p_cpu_key, p_s_search_path);
831     if (retval == IO_ERROR)
832 	return retval;
833     if ( retval == ITEM_FOUND )  {
834 
835 	RFALSE( ! ih_item_len(
836                 B_N_PITEM_HEAD(PATH_PLAST_BUFFER(p_s_search_path),
837 			       PATH_LAST_POSITION(p_s_search_path))),
838 	        "PAP-5165: item length equals zero");
839 
840 	pos_in_item(p_s_search_path) = 0;
841 	return POSITION_FOUND;
842     }
843 
844     RFALSE( ! PATH_LAST_POSITION(p_s_search_path),
845 	    "PAP-5170: position equals zero");
846 
847     /* Item is not found. Set path to the previous item. */
848     p_le_ih = B_N_PITEM_HEAD(PATH_PLAST_BUFFER(p_s_search_path), --PATH_LAST_POSITION(p_s_search_path));
849     n_blk_size = p_s_sb->s_blocksize;
850 
851     if (comp_short_keys (&(p_le_ih->ih_key), p_cpu_key)) {
852 	return FILE_NOT_FOUND;
853     }
854 
855     // FIXME: quite ugly this far
856 
857     item_offset = le_ih_k_offset (p_le_ih);
858     offset = cpu_key_k_offset (p_cpu_key);
859 
860     /* Needed byte is contained in the item pointed to by the path.*/
861     if (item_offset <= offset &&
862 	item_offset + op_bytes_number (p_le_ih, n_blk_size) > offset) {
863 	pos_in_item (p_s_search_path) = offset - item_offset;
864 	if ( is_indirect_le_ih(p_le_ih) ) {
865 	    pos_in_item (p_s_search_path) /= n_blk_size;
866 	}
867 	return POSITION_FOUND;
868     }
869 
870     /* Needed byte is not contained in the item pointed to by the
871      path. Set pos_in_item out of the item. */
872     if ( is_indirect_le_ih (p_le_ih) )
873 	pos_in_item (p_s_search_path) = ih_item_len(p_le_ih) / UNFM_P_SIZE;
874     else
875         pos_in_item (p_s_search_path) = ih_item_len( p_le_ih );
876 
877     return POSITION_NOT_FOUND;
878 }
879 
880 
881 /* Compare given item and item pointed to by the path. */
comp_items(const struct item_head * stored_ih,const struct path * p_s_path)882 int comp_items (const struct item_head * stored_ih, const struct path * p_s_path)
883 {
884     struct buffer_head  * p_s_bh;
885     struct item_head    * ih;
886 
887     /* Last buffer at the path is not in the tree. */
888     if ( ! B_IS_IN_TREE(p_s_bh = PATH_PLAST_BUFFER(p_s_path)) )
889 	return 1;
890 
891     /* Last path position is invalid. */
892     if ( PATH_LAST_POSITION(p_s_path) >= B_NR_ITEMS(p_s_bh) )
893 	return 1;
894 
895     /* we need only to know, whether it is the same item */
896     ih = get_ih (p_s_path);
897     return memcmp (stored_ih, ih, IH_SIZE);
898 }
899 
900 
901 /* unformatted nodes are not logged anymore, ever.  This is safe
902 ** now
903 */
904 #define held_by_others(bh) (atomic_read(&(bh)->b_count) > 1)
905 
906 // block can not be forgotten as it is in I/O or held by someone
907 #define block_in_use(bh) (buffer_locked(bh) || (held_by_others(bh)))
908 
909 
910 
911 // prepare for delete or cut of direct item
prepare_for_direct_item(struct path * path,struct item_head * le_ih,struct inode * inode,loff_t new_file_length,int * cut_size)912 static inline int prepare_for_direct_item (struct path * path,
913 					   struct item_head * le_ih,
914 					   struct inode * inode,
915 					   loff_t new_file_length,
916 					   int * cut_size)
917 {
918     loff_t round_len;
919 
920 
921     if ( new_file_length == max_reiserfs_offset (inode) ) {
922 	/* item has to be deleted */
923 	*cut_size = -(IH_SIZE + ih_item_len(le_ih));
924 	return M_DELETE;
925     }
926 
927     // new file gets truncated
928     if (get_inode_item_key_version (inode) == KEY_FORMAT_3_6) {
929 	//
930 	round_len = ROUND_UP (new_file_length);
931 	/* this was n_new_file_length < le_ih ... */
932 	if ( round_len < le_ih_k_offset (le_ih) )  {
933 	    *cut_size = -(IH_SIZE + ih_item_len(le_ih));
934 	    return M_DELETE; /* Delete this item. */
935 	}
936 	/* Calculate first position and size for cutting from item. */
937 	pos_in_item (path) = round_len - (le_ih_k_offset (le_ih) - 1);
938 	*cut_size = -(ih_item_len(le_ih) - pos_in_item(path));
939 
940 	return M_CUT; /* Cut from this item. */
941     }
942 
943 
944     // old file: items may have any length
945 
946     if ( new_file_length < le_ih_k_offset (le_ih) )  {
947 	*cut_size = -(IH_SIZE + ih_item_len(le_ih));
948 	return M_DELETE; /* Delete this item. */
949     }
950     /* Calculate first position and size for cutting from item. */
951     *cut_size = -(ih_item_len(le_ih) -
952 		      (pos_in_item (path) = new_file_length + 1 - le_ih_k_offset (le_ih)));
953     return M_CUT; /* Cut from this item. */
954 }
955 
956 
prepare_for_direntry_item(struct path * path,struct item_head * le_ih,struct inode * inode,loff_t new_file_length,int * cut_size)957 static inline int prepare_for_direntry_item (struct path * path,
958 					     struct item_head * le_ih,
959 					     struct inode * inode,
960 					     loff_t new_file_length,
961 					     int * cut_size)
962 {
963     if (le_ih_k_offset (le_ih) == DOT_OFFSET &&
964 	new_file_length == max_reiserfs_offset (inode)) {
965 	RFALSE( ih_entry_count (le_ih) != 2,
966 	        "PAP-5220: incorrect empty directory item (%h)", le_ih);
967 	*cut_size = -(IH_SIZE + ih_item_len(le_ih));
968 	return M_DELETE; /* Delete the directory item containing "." and ".." entry. */
969     }
970 
971     if ( ih_entry_count (le_ih) == 1 )  {
972 	/* Delete the directory item such as there is one record only
973 	   in this item*/
974 	*cut_size = -(IH_SIZE + ih_item_len(le_ih));
975 	return M_DELETE;
976     }
977 
978     /* Cut one record from the directory item. */
979     *cut_size = -(DEH_SIZE + entry_length (get_last_bh (path), le_ih, pos_in_item (path)));
980     return M_CUT;
981 }
982 
983 
984 /*  If the path points to a directory or direct item, calculate mode and the size cut, for balance.
985     If the path points to an indirect item, remove some number of its unformatted nodes.
986     In case of file truncate calculate whether this item must be deleted/truncated or last
987     unformatted node of this item will be converted to a direct item.
988     This function returns a determination of what balance mode the calling function should employ. */
prepare_for_delete_or_cut(struct reiserfs_transaction_handle * th,struct inode * inode,struct path * p_s_path,const struct cpu_key * p_s_item_key,int * p_n_removed,int * p_n_cut_size,unsigned long long n_new_file_length)989 static char  prepare_for_delete_or_cut(
990 				       struct reiserfs_transaction_handle *th,
991 				       struct inode * inode,
992 				       struct path         * p_s_path,
993 				       const struct cpu_key      * p_s_item_key,
994 				       int                 * p_n_removed,      /* Number of unformatted nodes which were removed
995 										  from end of the file. */
996 				       int                 * p_n_cut_size,
997 				       unsigned long long    n_new_file_length /* MAX_KEY_OFFSET in case of delete. */
998     ) {
999     struct super_block  * p_s_sb = inode->i_sb;
1000     struct item_head    * p_le_ih = PATH_PITEM_HEAD(p_s_path);
1001     struct buffer_head  * p_s_bh = PATH_PLAST_BUFFER(p_s_path);
1002 
1003     /* Stat_data item. */
1004     if ( is_statdata_le_ih (p_le_ih) ) {
1005 
1006 	RFALSE( n_new_file_length != max_reiserfs_offset (inode),
1007 		"PAP-5210: mode must be M_DELETE");
1008 
1009 	*p_n_cut_size = -(IH_SIZE + ih_item_len(p_le_ih));
1010 	return M_DELETE;
1011     }
1012 
1013 
1014     /* Directory item. */
1015     if ( is_direntry_le_ih (p_le_ih) )
1016 	return prepare_for_direntry_item (p_s_path, p_le_ih, inode, n_new_file_length, p_n_cut_size);
1017 
1018     /* Direct item. */
1019     if ( is_direct_le_ih (p_le_ih) )
1020 	return prepare_for_direct_item (p_s_path, p_le_ih, inode, n_new_file_length, p_n_cut_size);
1021 
1022 
1023     /* Case of an indirect item. */
1024     {
1025 	int                   n_unfm_number,    /* Number of the item unformatted nodes. */
1026 	    n_counter,
1027 	    n_blk_size;
1028 	__u32               * p_n_unfm_pointer; /* Pointer to the unformatted node number. */
1029 	__u32 tmp;
1030 	struct item_head      s_ih;           /* Item header. */
1031 	char                  c_mode;           /* Returned mode of the balance. */
1032 	int need_research;
1033 
1034 
1035 	n_blk_size = p_s_sb->s_blocksize;
1036 
1037 	/* Search for the needed object indirect item until there are no unformatted nodes to be removed. */
1038 	do  {
1039 	    need_research = 0;
1040             p_s_bh = PATH_PLAST_BUFFER(p_s_path);
1041 	    /* Copy indirect item header to a temp variable. */
1042 	    copy_item_head(&s_ih, PATH_PITEM_HEAD(p_s_path));
1043 	    /* Calculate number of unformatted nodes in this item. */
1044 	    n_unfm_number = I_UNFM_NUM(&s_ih);
1045 
1046 	    RFALSE( ! is_indirect_le_ih(&s_ih) || ! n_unfm_number ||
1047 		    pos_in_item (p_s_path) + 1 !=  n_unfm_number,
1048 		    "PAP-5240: illegal item %h "
1049 		    "n_unfm_number = %d *p_n_pos_in_item = %d",
1050 		    &s_ih, n_unfm_number, pos_in_item (p_s_path));
1051 
1052 	    /* Calculate balance mode and position in the item to remove unformatted nodes. */
1053 	    if ( n_new_file_length == max_reiserfs_offset (inode) ) {/* Case of delete. */
1054 		pos_in_item (p_s_path) = 0;
1055 		*p_n_cut_size = -(IH_SIZE + ih_item_len(&s_ih));
1056 		c_mode = M_DELETE;
1057 	    }
1058 	    else  { /* Case of truncate. */
1059 		if ( n_new_file_length < le_ih_k_offset (&s_ih) )  {
1060 		    pos_in_item (p_s_path) = 0;
1061 		    *p_n_cut_size = -(IH_SIZE + ih_item_len(&s_ih));
1062 		    c_mode = M_DELETE; /* Delete this item. */
1063 		}
1064 		else  {
1065 		    /* indirect item must be truncated starting from *p_n_pos_in_item-th position */
1066 		    pos_in_item (p_s_path) = (n_new_file_length + n_blk_size - le_ih_k_offset (&s_ih) ) >> p_s_sb->s_blocksize_bits;
1067 
1068 		    RFALSE( pos_in_item (p_s_path) > n_unfm_number,
1069 			    "PAP-5250: illegal position in the item");
1070 
1071 		    /* Either convert last unformatted node of indirect item to direct item or increase
1072 		       its free space.  */
1073 		    if ( pos_in_item (p_s_path) == n_unfm_number )  {
1074 			*p_n_cut_size = 0; /* Nothing to cut. */
1075 			return M_CONVERT; /* Maybe convert last unformatted node to the direct item. */
1076 		    }
1077 		    /* Calculate size to cut. */
1078 		    *p_n_cut_size = -(ih_item_len(&s_ih) - pos_in_item(p_s_path) * UNFM_P_SIZE);
1079 
1080 		    c_mode = M_CUT;     /* Cut from this indirect item. */
1081 		}
1082 	    }
1083 
1084 	    RFALSE( n_unfm_number <= pos_in_item (p_s_path),
1085 		    "PAP-5260: illegal position in the indirect item");
1086 
1087 	    /* pointers to be cut */
1088 	    n_unfm_number -= pos_in_item (p_s_path);
1089 	    /* Set pointer to the last unformatted node pointer that is to be cut. */
1090 	    p_n_unfm_pointer = (__u32 *)B_I_PITEM(p_s_bh, &s_ih) + I_UNFM_NUM(&s_ih) - 1 - *p_n_removed;
1091 
1092 
1093 	    /* We go through the unformatted nodes pointers of the indirect
1094 	       item and look for the unformatted nodes in the cache. If we
1095 	       found some of them we free it, zero corresponding indirect item
1096 	       entry and log buffer containing that indirect item. For this we
1097 	       need to prepare last path element for logging. If some
1098 	       unformatted node has b_count > 1 we must not free this
1099 	       unformatted node since it is in use. */
1100 	    reiserfs_prepare_for_journal(p_s_sb, p_s_bh, 1);
1101 	    // note: path could be changed, first line in for loop takes care
1102 	    // of it
1103 
1104 	    for (n_counter = *p_n_removed;
1105 		 n_counter < n_unfm_number; n_counter++, p_n_unfm_pointer-- ) {
1106 
1107 		if (item_moved (&s_ih, p_s_path)) {
1108 		    need_research = 1 ;
1109 		    break;
1110 		}
1111 		RFALSE( p_n_unfm_pointer < (__u32 *)B_I_PITEM(p_s_bh, &s_ih) ||
1112 			p_n_unfm_pointer > (__u32 *)B_I_PITEM(p_s_bh, &s_ih) + I_UNFM_NUM(&s_ih) - 1,
1113 			"vs-5265: pointer out of range");
1114 
1115 		/* Hole, nothing to remove. */
1116 		if ( ! get_block_num(p_n_unfm_pointer,0) )  {
1117 			(*p_n_removed)++;
1118 			continue;
1119 		}
1120 
1121 		(*p_n_removed)++;
1122 
1123 		tmp = get_block_num(p_n_unfm_pointer,0);
1124 		put_block_num(p_n_unfm_pointer, 0, 0);
1125 		journal_mark_dirty (th, p_s_sb, p_s_bh);
1126 		inode->i_blocks -= p_s_sb->s_blocksize / 512;
1127 		reiserfs_free_block(th, tmp);
1128 		/* In case of big fragmentation it is possible that each block
1129 		   freed will cause dirtying of one more bitmap and then we will
1130 		   quickly overflow our transaction space. This is a
1131 		   counter-measure against that scenario */
1132 		if (journal_transaction_should_end(th, th->t_blocks_allocated)) {
1133 		    int orig_len_alloc = th->t_blocks_allocated ;
1134 		    pathrelse(p_s_path) ;
1135 
1136 		    journal_end(th, p_s_sb, orig_len_alloc) ;
1137 		    journal_begin(th, p_s_sb, orig_len_alloc) ;
1138 		    reiserfs_update_inode_transaction(inode) ;
1139 		    need_research = 1;
1140 		    break;
1141 		}
1142 
1143 		if ( item_moved (&s_ih, p_s_path) )  {
1144 			need_research = 1;
1145 			break ;
1146 		}
1147 	    }
1148 
1149 	    /* a trick.  If the buffer has been logged, this
1150 	    ** will do nothing.  If we've broken the loop without
1151 	    ** logging it, it will restore the buffer
1152 	    **
1153 	    */
1154 	    reiserfs_restore_prepared_buffer(p_s_sb, p_s_bh);
1155 
1156 	    /* This loop can be optimized. */
1157 	} while ( (*p_n_removed < n_unfm_number || need_research) &&
1158 		  search_for_position_by_key(p_s_sb, p_s_item_key, p_s_path) == POSITION_FOUND );
1159 
1160 	RFALSE( *p_n_removed < n_unfm_number,
1161 		"PAP-5310: indirect item is not found");
1162 	RFALSE( item_moved (&s_ih, p_s_path),
1163 		"after while, comp failed, retry") ;
1164 
1165 	if (c_mode == M_CUT)
1166 	    pos_in_item (p_s_path) *= UNFM_P_SIZE;
1167 	return c_mode;
1168     }
1169 }
1170 
1171 
1172 /* Calculate bytes number which will be deleted or cutted in the balance. */
calc_deleted_bytes_number(struct tree_balance * p_s_tb,char c_mode)1173 int calc_deleted_bytes_number(
1174     struct  tree_balance  * p_s_tb,
1175     char                    c_mode
1176     ) {
1177     int                     n_del_size;
1178     struct  item_head     * p_le_ih = PATH_PITEM_HEAD(p_s_tb->tb_path);
1179 
1180     if ( is_statdata_le_ih (p_le_ih) )
1181 	return 0;
1182 
1183     if ( is_direntry_le_ih (p_le_ih) ) {
1184 	// return EMPTY_DIR_SIZE; /* We delete emty directoris only. */
1185 	// we can't use EMPTY_DIR_SIZE, as old format dirs have a different
1186 	// empty size.  ick. FIXME, is this right?
1187 	//
1188 	return ih_item_len(p_le_ih);
1189     }
1190     n_del_size = ( c_mode == M_DELETE ) ? ih_item_len(p_le_ih) : -p_s_tb->insert_size[0];
1191 
1192     if ( is_indirect_le_ih (p_le_ih) )
1193 	n_del_size = (n_del_size/UNFM_P_SIZE)*
1194 	  (PATH_PLAST_BUFFER(p_s_tb->tb_path)->b_size);// - get_ih_free_space (p_le_ih);
1195     return n_del_size;
1196 }
1197 
init_tb_struct(struct reiserfs_transaction_handle * th,struct tree_balance * p_s_tb,struct super_block * p_s_sb,struct path * p_s_path,int n_size)1198 static void init_tb_struct(
1199     struct reiserfs_transaction_handle *th,
1200     struct tree_balance * p_s_tb,
1201     struct super_block  * p_s_sb,
1202     struct path         * p_s_path,
1203     int                   n_size
1204     ) {
1205     memset (p_s_tb,'\0',sizeof(struct tree_balance));
1206     p_s_tb->transaction_handle = th ;
1207     p_s_tb->tb_sb = p_s_sb;
1208     p_s_tb->tb_path = p_s_path;
1209     PATH_OFFSET_PBUFFER(p_s_path, ILLEGAL_PATH_ELEMENT_OFFSET) = NULL;
1210     PATH_OFFSET_POSITION(p_s_path, ILLEGAL_PATH_ELEMENT_OFFSET) = 0;
1211     p_s_tb->insert_size[0] = n_size;
1212 }
1213 
1214 
1215 
padd_item(char * item,int total_length,int length)1216 void padd_item (char * item, int total_length, int length)
1217 {
1218     int i;
1219 
1220     for (i = total_length; i > length; )
1221 	item [--i] = 0;
1222 }
1223 
1224 
1225 /* Delete object item. */
reiserfs_delete_item(struct reiserfs_transaction_handle * th,struct path * p_s_path,const struct cpu_key * p_s_item_key,struct inode * p_s_inode,struct buffer_head * p_s_un_bh)1226 int reiserfs_delete_item (struct reiserfs_transaction_handle *th,
1227 			  struct path * p_s_path, /* Path to the deleted item. */
1228 			  const struct cpu_key * p_s_item_key, /* Key to search for the deleted item.  */
1229 			  struct inode * p_s_inode,/* inode is here just to update i_blocks */
1230 			  struct buffer_head  * p_s_un_bh)    /* NULL or unformatted node pointer.    */
1231 {
1232     struct super_block * p_s_sb = p_s_inode->i_sb;
1233     struct tree_balance   s_del_balance;
1234     struct item_head      s_ih;
1235     int                   n_ret_value,
1236 	n_del_size,
1237 	n_removed;
1238 
1239 #ifdef CONFIG_REISERFS_CHECK
1240     char                  c_mode;
1241     int			n_iter = 0;
1242 #endif
1243 
1244     init_tb_struct(th, &s_del_balance, p_s_sb, p_s_path, 0/*size is unknown*/);
1245 
1246     while ( 1 ) {
1247 	n_removed = 0;
1248 
1249 #ifdef CONFIG_REISERFS_CHECK
1250 	n_iter++;
1251 	c_mode =
1252 #endif
1253 	    prepare_for_delete_or_cut(th, p_s_inode, p_s_path, p_s_item_key, &n_removed, &n_del_size, max_reiserfs_offset (p_s_inode));
1254 
1255 	RFALSE( c_mode != M_DELETE, "PAP-5320: mode must be M_DELETE");
1256 
1257 	copy_item_head(&s_ih, PATH_PITEM_HEAD(p_s_path));
1258 	s_del_balance.insert_size[0] = n_del_size;
1259 
1260 	n_ret_value = fix_nodes(M_DELETE, &s_del_balance, NULL, 0);
1261 	if ( n_ret_value != REPEAT_SEARCH )
1262 	    break;
1263 
1264 	PROC_INFO_INC( p_s_sb, delete_item_restarted );
1265 
1266 	// file system changed, repeat search
1267 	n_ret_value = search_for_position_by_key(p_s_sb, p_s_item_key, p_s_path);
1268 	if (n_ret_value == IO_ERROR)
1269 	    break;
1270 	if (n_ret_value == FILE_NOT_FOUND) {
1271 	    reiserfs_warning (p_s_sb, "vs-5340: reiserfs_delete_item: "
1272 			      "no items of the file %K found\n", p_s_item_key);
1273 	    break;
1274 	}
1275     } /* while (1) */
1276 
1277     if ( n_ret_value != CARRY_ON ) {
1278 	unfix_nodes(&s_del_balance);
1279 	return 0;
1280     }
1281 
1282     // reiserfs_delete_item returns item length when success
1283     n_ret_value = calc_deleted_bytes_number(&s_del_balance, M_DELETE);
1284 
1285     if ( p_s_un_bh )  {
1286 	int off;
1287         char *data ;
1288 
1289 	/* We are in direct2indirect conversion, so move tail contents
1290            to the unformatted node */
1291 	/* note, we do the copy before preparing the buffer because we
1292 	** don't care about the contents of the unformatted node yet.
1293 	** the only thing we really care about is the direct item's data
1294 	** is in the unformatted node.
1295 	**
1296 	** Otherwise, we would have to call reiserfs_prepare_for_journal on
1297 	** the unformatted node, which might schedule, meaning we'd have to
1298 	** loop all the way back up to the start of the while loop.
1299 	**
1300 	** The unformatted node must be dirtied later on.  We can't be
1301 	** sure here if the entire tail has been deleted yet.
1302         **
1303         ** p_s_un_bh is from the page cache (all unformatted nodes are
1304         ** from the page cache) and might be a highmem page.  So, we
1305         ** can't use p_s_un_bh->b_data.  But, the page has already been
1306         ** kmapped, so we can use page_address()
1307 	** -clm
1308 	*/
1309 
1310         data = page_address(p_s_un_bh->b_page) ;
1311 	off = ((le_ih_k_offset (&s_ih) - 1) & (PAGE_CACHE_SIZE - 1));
1312 	memcpy(data + off,
1313 	       B_I_PITEM(PATH_PLAST_BUFFER(p_s_path), &s_ih), n_ret_value);
1314     }
1315 
1316     /* Perform balancing after all resources have been collected at once. */
1317     do_balance(&s_del_balance, NULL, NULL, M_DELETE);
1318 
1319     /* Return deleted body length */
1320     return n_ret_value;
1321 }
1322 
1323 
1324 /* Summary Of Mechanisms For Handling Collisions Between Processes:
1325 
1326  deletion of the body of the object is performed by iput(), with the
1327  result that if multiple processes are operating on a file, the
1328  deletion of the body of the file is deferred until the last process
1329  that has an open inode performs its iput().
1330 
1331  writes and truncates are protected from collisions by use of
1332  semaphores.
1333 
1334  creates, linking, and mknod are protected from collisions with other
1335  processes by making the reiserfs_add_entry() the last step in the
1336  creation, and then rolling back all changes if there was a collision.
1337  - Hans
1338 */
1339 
1340 
1341 /* this deletes item which never gets split */
reiserfs_delete_solid_item(struct reiserfs_transaction_handle * th,struct key * key)1342 void reiserfs_delete_solid_item (struct reiserfs_transaction_handle *th,
1343 				 struct key * key)
1344 {
1345     struct tree_balance tb;
1346     INITIALIZE_PATH (path);
1347     int item_len;
1348     int tb_init = 0 ;
1349     struct cpu_key cpu_key;
1350     int retval;
1351 
1352     le_key2cpu_key (&cpu_key, key);
1353 
1354     while (1) {
1355 	retval = search_item (th->t_super, &cpu_key, &path);
1356 	if (retval == IO_ERROR) {
1357 	    reiserfs_warning (th->t_super, "vs-5350: reiserfs_delete_solid_item: "
1358 			      "i/o failure occurred trying to delete %K\n", &cpu_key);
1359 	    break;
1360 	}
1361 	if (retval != ITEM_FOUND) {
1362 	    pathrelse (&path);
1363 	    // No need for a warning, if there is just no free space to insert '..' item into the newly-created subdir
1364 	    if ( !( (unsigned long long) GET_HASH_VALUE (le_key_k_offset (le_key_version (key), key)) == 0 && \
1365 		 GET_GENERATION_NUMBER (le_key_k_offset (le_key_version (key), key)) == 1 ) )
1366 		reiserfs_warning (th->t_super, "vs-5355: reiserfs_delete_solid_item: %k not found\n", key);
1367 	    break;
1368 	}
1369 	if (!tb_init) {
1370 	    tb_init = 1 ;
1371 	    item_len = ih_item_len( PATH_PITEM_HEAD(&path) );
1372 	    init_tb_struct (th, &tb, th->t_super, &path, - (IH_SIZE + item_len));
1373 	}
1374 
1375 	retval = fix_nodes (M_DELETE, &tb, NULL, 0);
1376 	if (retval == REPEAT_SEARCH) {
1377 	    PROC_INFO_INC( th -> t_super, delete_solid_item_restarted );
1378 	    continue;
1379 	}
1380 
1381 	if (retval == CARRY_ON) {
1382 	    do_balance (&tb, 0, 0, M_DELETE);
1383 	    break;
1384 	}
1385 
1386 	// IO_ERROR, NO_DISK_SPACE, etc
1387 	reiserfs_warning (th->t_super, "vs-5360: reiserfs_delete_solid_item: "
1388 			  "could not delete %K due to fix_nodes failure\n", &cpu_key);
1389 	unfix_nodes (&tb);
1390 	break;
1391     }
1392 
1393     reiserfs_check_path(&path) ;
1394 }
1395 
1396 
reiserfs_delete_object(struct reiserfs_transaction_handle * th,struct inode * inode)1397 void reiserfs_delete_object (struct reiserfs_transaction_handle *th, struct inode * inode)
1398 {
1399     inode->i_size = 0;
1400 
1401     /* for directory this deletes item containing "." and ".." */
1402     reiserfs_do_truncate (th, inode, NULL, 0/*no timestamp updates*/);
1403 
1404 #if defined( USE_INODE_GENERATION_COUNTER )
1405     if( !old_format_only ( th -> t_super ) )
1406       {
1407        __u32 *inode_generation;
1408 
1409        inode_generation =
1410          &th -> t_super -> u.reiserfs_sb.s_rs -> s_inode_generation;
1411        *inode_generation = cpu_to_le32( le32_to_cpu( *inode_generation ) + 1 );
1412       }
1413 /* USE_INODE_GENERATION_COUNTER */
1414 #endif
1415     reiserfs_delete_solid_item (th, INODE_PKEY (inode));
1416 }
1417 
1418 
maybe_indirect_to_direct(struct reiserfs_transaction_handle * th,struct inode * p_s_inode,struct page * page,struct path * p_s_path,const struct cpu_key * p_s_item_key,loff_t n_new_file_size,char * p_c_mode)1419 static int maybe_indirect_to_direct (struct reiserfs_transaction_handle *th,
1420 			      struct inode * p_s_inode,
1421 			      struct page *page,
1422 			      struct path         * p_s_path,
1423 			      const struct cpu_key      * p_s_item_key,
1424 			      loff_t         n_new_file_size,
1425 			      char                * p_c_mode
1426 			      ) {
1427     struct super_block * p_s_sb = p_s_inode->i_sb;
1428     int n_block_size = p_s_sb->s_blocksize;
1429     int cut_bytes;
1430 
1431     if (n_new_file_size != p_s_inode->i_size)
1432 	BUG ();
1433 
1434     /* the page being sent in could be NULL if there was an i/o error
1435     ** reading in the last block.  The user will hit problems trying to
1436     ** read the file, but for now we just skip the indirect2direct
1437     */
1438     if (atomic_read(&p_s_inode->i_count) > 1 ||
1439         !tail_has_to_be_packed (p_s_inode) ||
1440         !page || (p_s_inode->u.reiserfs_i.i_flags & i_nopack_mask)) {
1441 	// leave tail in an unformatted node
1442 	*p_c_mode = M_SKIP_BALANCING;
1443 	cut_bytes = n_block_size - (n_new_file_size & (n_block_size - 1));
1444 	pathrelse(p_s_path);
1445 	return cut_bytes;
1446     }
1447     /* Permorm the conversion to a direct_item. */
1448     /*return indirect_to_direct (p_s_inode, p_s_path, p_s_item_key, n_new_file_size, p_c_mode);*/
1449     return indirect2direct (th, p_s_inode, page, p_s_path, p_s_item_key, n_new_file_size, p_c_mode);
1450 }
1451 
1452 
1453 /* we did indirect_to_direct conversion. And we have inserted direct
1454    item successesfully, but there were no disk space to cut unfm
1455    pointer being converted. Therefore we have to delete inserted
1456    direct item(s) */
indirect_to_direct_roll_back(struct reiserfs_transaction_handle * th,struct inode * inode,struct path * path)1457 static void indirect_to_direct_roll_back (struct reiserfs_transaction_handle *th, struct inode * inode, struct path * path)
1458 {
1459     struct cpu_key tail_key;
1460     int tail_len;
1461     int removed;
1462 
1463     make_cpu_key (&tail_key, inode, inode->i_size + 1, TYPE_DIRECT, 4);// !!!!
1464     tail_key.key_length = 4;
1465 
1466     tail_len = (cpu_key_k_offset (&tail_key) & (inode->i_sb->s_blocksize - 1)) - 1;
1467     while (tail_len) {
1468 	/* look for the last byte of the tail */
1469 	if (search_for_position_by_key (inode->i_sb, &tail_key, path) == POSITION_NOT_FOUND)
1470 	    reiserfs_panic (inode->i_sb, "vs-5615: indirect_to_direct_roll_back: found invalid item");
1471 	RFALSE( path->pos_in_item != ih_item_len(PATH_PITEM_HEAD (path)) - 1,
1472 	        "vs-5616: appended bytes found");
1473 	PATH_LAST_POSITION (path) --;
1474 
1475 	removed = reiserfs_delete_item (th, path, &tail_key, inode, 0/*unbh not needed*/);
1476 	RFALSE( removed <= 0 || removed > tail_len,
1477 	        "vs-5617: there was tail %d bytes, removed item length %d bytes",
1478                 tail_len, removed);
1479 	tail_len -= removed;
1480 	set_cpu_key_k_offset (&tail_key, cpu_key_k_offset (&tail_key) - removed);
1481     }
1482     reiserfs_warning (inode->i_sb, "indirect_to_direct_roll_back: indirect_to_direct conversion has been rolled back due to lack of disk space\n");
1483     //mark_file_without_tail (inode);
1484     mark_inode_dirty (inode);
1485 }
1486 
1487 
1488 /* (Truncate or cut entry) or delete object item. Returns < 0 on failure */
reiserfs_cut_from_item(struct reiserfs_transaction_handle * th,struct path * p_s_path,struct cpu_key * p_s_item_key,struct inode * p_s_inode,struct page * page,loff_t n_new_file_size)1489 int reiserfs_cut_from_item (struct reiserfs_transaction_handle *th,
1490 			    struct path * p_s_path,
1491 			    struct cpu_key * p_s_item_key,
1492 			    struct inode * p_s_inode,
1493 			    struct page *page,
1494 			    loff_t n_new_file_size)
1495 {
1496     struct super_block * p_s_sb = p_s_inode->i_sb;
1497     /* Every function which is going to call do_balance must first
1498        create a tree_balance structure.  Then it must fill up this
1499        structure by using the init_tb_struct and fix_nodes functions.
1500        After that we can make tree balancing. */
1501     struct tree_balance s_cut_balance;
1502     int n_cut_size = 0,        /* Amount to be cut. */
1503 	n_ret_value = CARRY_ON,
1504 	n_removed = 0,     /* Number of the removed unformatted nodes. */
1505 	n_is_inode_locked = 0;
1506     char                c_mode;            /* Mode of the balance. */
1507     int retval2 = -1;
1508 
1509 
1510     init_tb_struct(th, &s_cut_balance, p_s_inode->i_sb, p_s_path, n_cut_size);
1511 
1512 
1513     /* Repeat this loop until we either cut the item without needing
1514        to balance, or we fix_nodes without schedule occuring */
1515     while ( 1 ) {
1516 	/* Determine the balance mode, position of the first byte to
1517 	   be cut, and size to be cut.  In case of the indirect item
1518 	   free unformatted nodes which are pointed to by the cut
1519 	   pointers. */
1520 
1521 	c_mode = prepare_for_delete_or_cut(th, p_s_inode, p_s_path, p_s_item_key, &n_removed,
1522 					   &n_cut_size, n_new_file_size);
1523 	if ( c_mode == M_CONVERT )  {
1524 	    /* convert last unformatted node to direct item or leave
1525                tail in the unformatted node */
1526 	    RFALSE( n_ret_value != CARRY_ON, "PAP-5570: can not convert twice");
1527 
1528 	    n_ret_value = maybe_indirect_to_direct (th, p_s_inode, page, p_s_path, p_s_item_key,
1529 						    n_new_file_size, &c_mode);
1530 	    if ( c_mode == M_SKIP_BALANCING )
1531 		/* tail has been left in the unformatted node */
1532 		return n_ret_value;
1533 
1534 	    n_is_inode_locked = 1;
1535 
1536 	    /* removing of last unformatted node will change value we
1537                have to return to truncate. Save it */
1538 	    retval2 = n_ret_value;
1539 	    /*retval2 = p_s_sb->s_blocksize - (n_new_file_size & (p_s_sb->s_blocksize - 1));*/
1540 
1541 	    /* So, we have performed the first part of the conversion:
1542 	       inserting the new direct item.  Now we are removing the
1543 	       last unformatted node pointer. Set key to search for
1544 	       it. */
1545       	    set_cpu_key_k_type (p_s_item_key, TYPE_INDIRECT);
1546 	    p_s_item_key->key_length = 4;
1547 	    n_new_file_size -= (n_new_file_size & (p_s_sb->s_blocksize - 1));
1548 	    set_cpu_key_k_offset (p_s_item_key, n_new_file_size + 1);
1549 	    if ( search_for_position_by_key(p_s_sb, p_s_item_key, p_s_path) == POSITION_NOT_FOUND ){
1550 		print_block (PATH_PLAST_BUFFER (p_s_path), 3, PATH_LAST_POSITION (p_s_path) - 1, PATH_LAST_POSITION (p_s_path) + 1);
1551 		reiserfs_panic(p_s_sb, "PAP-5580: reiserfs_cut_from_item: item to convert does not exist (%K)", p_s_item_key);
1552 	    }
1553 	    continue;
1554 	}
1555 	if (n_cut_size == 0) {
1556 	    pathrelse (p_s_path);
1557 	    return 0;
1558 	}
1559 
1560 	s_cut_balance.insert_size[0] = n_cut_size;
1561 
1562 	n_ret_value = fix_nodes(c_mode, &s_cut_balance, NULL, 0);
1563       	if ( n_ret_value != REPEAT_SEARCH )
1564 	    break;
1565 
1566 	PROC_INFO_INC( p_s_sb, cut_from_item_restarted );
1567 
1568 	n_ret_value = search_for_position_by_key(p_s_sb, p_s_item_key, p_s_path);
1569 	if (n_ret_value == POSITION_FOUND)
1570 	    continue;
1571 
1572 	reiserfs_warning (p_s_sb, "PAP-5610: reiserfs_cut_from_item: item %K not found\n", p_s_item_key);
1573 	unfix_nodes (&s_cut_balance);
1574 	return (n_ret_value == IO_ERROR) ? -EIO : -ENOENT;
1575     } /* while */
1576 
1577     // check fix_nodes results (IO_ERROR or NO_DISK_SPACE)
1578     if ( n_ret_value != CARRY_ON ) {
1579 	if ( n_is_inode_locked ) {
1580 	    // FIXME: this seems to be not needed: we are always able
1581 	    // to cut item
1582 	    indirect_to_direct_roll_back (th, p_s_inode, p_s_path);
1583 	}
1584 	if (n_ret_value == NO_DISK_SPACE)
1585 	    reiserfs_warning (p_s_sb, "NO_DISK_SPACE\n");
1586 	unfix_nodes (&s_cut_balance);
1587 	return -EIO;
1588     }
1589 
1590     /* go ahead and perform balancing */
1591 
1592     RFALSE( c_mode == M_PASTE || c_mode == M_INSERT, "illegal mode");
1593 
1594     /* Calculate number of bytes that need to be cut from the item. */
1595     if (retval2 == -1)
1596 	n_ret_value = calc_deleted_bytes_number(&s_cut_balance, c_mode);
1597     else
1598 	n_ret_value = retval2;
1599 
1600     if ( c_mode == M_DELETE ) {
1601 	struct item_head * p_le_ih = PATH_PITEM_HEAD (s_cut_balance.tb_path);
1602 
1603 	if ( is_direct_le_ih (p_le_ih) && (le_ih_k_offset (p_le_ih) & (p_s_sb->s_blocksize - 1)) == 1 ) {
1604 	    /* we delete first part of tail which was stored in direct
1605                item(s) */
1606 	    // FIXME: this is to keep 3.5 happy
1607 	    p_s_inode->u.reiserfs_i.i_first_direct_byte = U32_MAX;
1608 	    p_s_inode->i_blocks -= p_s_sb->s_blocksize / 512;
1609 	}
1610     }
1611 
1612 #ifdef CONFIG_REISERFS_CHECK
1613     if (n_is_inode_locked) {
1614 	struct item_head * le_ih = PATH_PITEM_HEAD (s_cut_balance.tb_path);
1615 	/* we are going to complete indirect2direct conversion. Make
1616            sure, that we exactly remove last unformatted node pointer
1617            of the item */
1618 	if (!is_indirect_le_ih (le_ih))
1619 	    reiserfs_panic (p_s_sb, "vs-5652: reiserfs_cut_from_item: "
1620 			    "item must be indirect %h", le_ih);
1621 
1622 	if (c_mode == M_DELETE && ih_item_len(le_ih) != UNFM_P_SIZE)
1623 	    reiserfs_panic (p_s_sb, "vs-5653: reiserfs_cut_from_item: "
1624 			    "completing indirect2direct conversion indirect item %h "
1625 			    "being deleted must be of 4 byte long", le_ih);
1626 
1627 	if (c_mode == M_CUT && s_cut_balance.insert_size[0] != -UNFM_P_SIZE) {
1628 	    reiserfs_panic (p_s_sb, "vs-5654: reiserfs_cut_from_item: "
1629 			    "can not complete indirect2direct conversion of %h (CUT, insert_size==%d)",
1630 			    le_ih, s_cut_balance.insert_size[0]);
1631 	}
1632 	/* it would be useful to make sure, that right neighboring
1633            item is direct item of this file */
1634     }
1635 #endif
1636 
1637     do_balance(&s_cut_balance, NULL, NULL, c_mode);
1638     if ( n_is_inode_locked ) {
1639 	/* we've done an indirect->direct conversion.  when the data block
1640 	** was freed, it was removed from the list of blocks that must
1641 	** be flushed before the transaction commits, so we don't need to
1642 	** deal with it here.
1643 	*/
1644 	p_s_inode->u.reiserfs_i.i_flags &= ~i_pack_on_close_mask;
1645     }
1646     return n_ret_value;
1647 }
1648 
1649 
truncate_directory(struct reiserfs_transaction_handle * th,struct inode * inode)1650 static void truncate_directory (struct reiserfs_transaction_handle *th, struct inode * inode)
1651 {
1652     if (inode->i_nlink)
1653 	reiserfs_warning (th->t_super, "vs-5655: truncate_directory: link count != 0\n");
1654 
1655     set_le_key_k_offset (KEY_FORMAT_3_5, INODE_PKEY (inode), DOT_OFFSET);
1656     set_le_key_k_type (KEY_FORMAT_3_5, INODE_PKEY (inode), TYPE_DIRENTRY);
1657     reiserfs_delete_solid_item (th, INODE_PKEY (inode));
1658 
1659     set_le_key_k_offset (KEY_FORMAT_3_5, INODE_PKEY (inode), SD_OFFSET);
1660     set_le_key_k_type (KEY_FORMAT_3_5, INODE_PKEY (inode), TYPE_STAT_DATA);
1661 }
1662 
1663 
1664 
1665 
1666 /* Truncate file to the new size. Note, this must be called with a transaction
1667    already started */
reiserfs_do_truncate(struct reiserfs_transaction_handle * th,struct inode * p_s_inode,struct page * page,int update_timestamps)1668 void reiserfs_do_truncate (struct reiserfs_transaction_handle *th,
1669 			   struct  inode * p_s_inode, /* ->i_size contains new
1670                                                          size */
1671 			   struct page *page, /* up to date for last block */
1672 			   int update_timestamps  /* when it is called by
1673 						     file_release to convert
1674 						     the tail - no timestamps
1675 						     should be updated */
1676     ) {
1677     INITIALIZE_PATH (s_search_path);       /* Path to the current object item. */
1678     struct item_head    * p_le_ih;         /* Pointer to an item header. */
1679     struct cpu_key      s_item_key;     /* Key to search for a previous file item. */
1680     loff_t         n_file_size,    /* Old file size. */
1681 	n_new_file_size;/* New file size. */
1682     int                   n_deleted;      /* Number of deleted or truncated bytes. */
1683     int retval;
1684 
1685     if ( ! (S_ISREG(p_s_inode->i_mode) || S_ISDIR(p_s_inode->i_mode) || S_ISLNK(p_s_inode->i_mode)) )
1686 	return;
1687 
1688     if (S_ISDIR(p_s_inode->i_mode)) {
1689 	// deletion of directory - no need to update timestamps
1690 	truncate_directory (th, p_s_inode);
1691 	return;
1692     }
1693 
1694     /* Get new file size. */
1695     n_new_file_size = p_s_inode->i_size;
1696 
1697     // FIXME: note, that key type is unimportant here
1698     make_cpu_key (&s_item_key, p_s_inode, max_reiserfs_offset (p_s_inode), TYPE_DIRECT, 3);
1699 
1700     retval = search_for_position_by_key(p_s_inode->i_sb, &s_item_key, &s_search_path);
1701     if (retval == IO_ERROR) {
1702 	reiserfs_warning (p_s_inode->i_sb, "vs-5657: reiserfs_do_truncate: "
1703 			  "i/o failure occurred trying to truncate %K\n", &s_item_key);
1704 	return;
1705     }
1706     if (retval == POSITION_FOUND || retval == FILE_NOT_FOUND) {
1707 	pathrelse (&s_search_path);
1708 	reiserfs_warning (p_s_inode->i_sb, "PAP-5660: reiserfs_do_truncate: "
1709 			  "wrong result %d of search for %K\n", retval, &s_item_key);
1710 	return;
1711     }
1712 
1713     s_search_path.pos_in_item --;
1714 
1715     /* Get real file size (total length of all file items) */
1716     p_le_ih = PATH_PITEM_HEAD(&s_search_path);
1717     if ( is_statdata_le_ih (p_le_ih) )
1718 	n_file_size = 0;
1719     else {
1720 	loff_t offset = le_ih_k_offset (p_le_ih);
1721 	int bytes = op_bytes_number (p_le_ih,p_s_inode->i_sb->s_blocksize);
1722 
1723 	/* this may mismatch with real file size: if last direct item
1724            had no padding zeros and last unformatted node had no free
1725            space, this file would have this file size */
1726 	n_file_size = offset + bytes - 1;
1727     }
1728 
1729     if ( n_file_size == 0 || n_file_size < n_new_file_size ) {
1730 	goto update_and_out ;
1731     }
1732 
1733     /* Update key to search for the last file item. */
1734     set_cpu_key_k_offset (&s_item_key, n_file_size);
1735 
1736     do  {
1737 	/* Cut or delete file item. */
1738 	n_deleted = reiserfs_cut_from_item(th, &s_search_path, &s_item_key, p_s_inode,  page, n_new_file_size);
1739 	if (n_deleted < 0) {
1740 	    reiserfs_warning (th->t_super, "vs-5665: reiserfs_truncate_file: cut_from_item failed\n");
1741 	    reiserfs_check_path(&s_search_path) ;
1742 	    return;
1743 	}
1744 
1745 	RFALSE( n_deleted > n_file_size,
1746 		"PAP-5670: reiserfs_truncate_file returns too big number: deleted %d, file_size %lu, item_key %K",
1747 		n_deleted, n_file_size, &s_item_key);
1748 
1749 	/* Change key to search the last file item. */
1750 	n_file_size -= n_deleted;
1751 
1752 	set_cpu_key_k_offset (&s_item_key, n_file_size);
1753 
1754 	/* While there are bytes to truncate and previous file item is presented in the tree. */
1755 
1756 	/*
1757 	** This loop could take a really long time, and could log
1758 	** many more blocks than a transaction can hold.  So, we do a polite
1759 	** journal end here, and if the transaction needs ending, we make
1760 	** sure the file is consistent before ending the current trans
1761 	** and starting a new one
1762 	*/
1763         if (journal_transaction_should_end(th, th->t_blocks_allocated)) {
1764 	  int orig_len_alloc = th->t_blocks_allocated ;
1765 	  decrement_counters_in_path(&s_search_path) ;
1766 
1767 	  if (update_timestamps) {
1768 	      p_s_inode->i_mtime = p_s_inode->i_ctime = CURRENT_TIME;
1769 	  }
1770 	  reiserfs_update_sd(th, p_s_inode) ;
1771 
1772 	  journal_end(th, p_s_inode->i_sb, orig_len_alloc) ;
1773 	  journal_begin(th, p_s_inode->i_sb, orig_len_alloc) ;
1774 	  reiserfs_update_inode_transaction(p_s_inode) ;
1775 	}
1776     } while ( n_file_size > ROUND_UP (n_new_file_size) &&
1777 	      search_for_position_by_key(p_s_inode->i_sb, &s_item_key, &s_search_path) == POSITION_FOUND )  ;
1778 
1779     RFALSE( n_file_size > ROUND_UP (n_new_file_size),
1780 	    "PAP-5680: truncate did not finish: new_file_size %Ld, current %Ld, oid %d\n",
1781 	    n_new_file_size, n_file_size, s_item_key.on_disk_key.k_objectid);
1782 
1783 update_and_out:
1784     if (update_timestamps) {
1785 	// this is truncate, not file closing
1786 	p_s_inode->i_mtime = p_s_inode->i_ctime = CURRENT_TIME;
1787     }
1788     reiserfs_update_sd (th, p_s_inode);
1789 
1790     pathrelse(&s_search_path) ;
1791 }
1792 
1793 
1794 #ifdef CONFIG_REISERFS_CHECK
1795 // this makes sure, that we __append__, not overwrite or add holes
check_research_for_paste(struct path * path,const struct cpu_key * p_s_key)1796 static void check_research_for_paste (struct path * path,
1797 				      const struct cpu_key * p_s_key)
1798 {
1799     struct item_head * found_ih = get_ih (path);
1800 
1801     if (is_direct_le_ih (found_ih)) {
1802 	if (le_ih_k_offset (found_ih) + op_bytes_number (found_ih, get_last_bh (path)->b_size) !=
1803 	    cpu_key_k_offset (p_s_key) ||
1804 	    op_bytes_number (found_ih, get_last_bh (path)->b_size) != pos_in_item (path))
1805 	    reiserfs_panic (0, "PAP-5720: check_research_for_paste: "
1806 			    "found direct item %h or position (%d) does not match to key %K",
1807 			    found_ih, pos_in_item (path), p_s_key);
1808     }
1809     if (is_indirect_le_ih (found_ih)) {
1810 	if (le_ih_k_offset (found_ih) + op_bytes_number (found_ih, get_last_bh (path)->b_size) != cpu_key_k_offset (p_s_key) ||
1811 	    I_UNFM_NUM (found_ih) != pos_in_item (path) ||
1812 	    get_ih_free_space (found_ih) != 0)
1813 	    reiserfs_panic (0, "PAP-5730: check_research_for_paste: "
1814 			    "found indirect item (%h) or position (%d) does not match to key (%K)",
1815 			    found_ih, pos_in_item (path), p_s_key);
1816     }
1817 }
1818 #endif /* config reiserfs check */
1819 
1820 
1821 /* Paste bytes to the existing item. Returns bytes number pasted into the item. */
reiserfs_paste_into_item(struct reiserfs_transaction_handle * th,struct path * p_s_search_path,const struct cpu_key * p_s_key,const char * p_c_body,int n_pasted_size)1822 int reiserfs_paste_into_item (struct reiserfs_transaction_handle *th,
1823 			      struct path         * p_s_search_path,	/* Path to the pasted item.          */
1824 			      const struct cpu_key      * p_s_key,        	/* Key to search for the needed item.*/
1825 			      const char          * p_c_body,       	/* Pointer to the bytes to paste.    */
1826 			      int                   n_pasted_size)  	/* Size of pasted bytes.             */
1827 {
1828     struct tree_balance s_paste_balance;
1829     int                 retval;
1830 
1831     init_tb_struct(th, &s_paste_balance, th->t_super, p_s_search_path, n_pasted_size);
1832 #ifdef DISPLACE_NEW_PACKING_LOCALITIES
1833     s_paste_balance.key = p_s_key->on_disk_key;
1834 #endif
1835 
1836     while ( (retval = fix_nodes(M_PASTE, &s_paste_balance, NULL, p_c_body)) == REPEAT_SEARCH ) {
1837 	/* file system changed while we were in the fix_nodes */
1838 	PROC_INFO_INC( th -> t_super, paste_into_item_restarted );
1839 	retval = search_for_position_by_key (th->t_super, p_s_key, p_s_search_path);
1840 	if (retval == IO_ERROR) {
1841 	    retval = -EIO ;
1842 	    goto error_out ;
1843 	}
1844 	if (retval == POSITION_FOUND) {
1845 	    reiserfs_warning (th->t_super, "PAP-5710: reiserfs_paste_into_item: entry or pasted byte (%K) exists\n", p_s_key);
1846 	    retval = -EEXIST ;
1847 	    goto error_out ;
1848 	}
1849 
1850 #ifdef CONFIG_REISERFS_CHECK
1851 	check_research_for_paste (p_s_search_path, p_s_key);
1852 #endif
1853     }
1854 
1855     /* Perform balancing after all resources are collected by fix_nodes, and
1856        accessing them will not risk triggering schedule. */
1857     if ( retval == CARRY_ON ) {
1858 	do_balance(&s_paste_balance, NULL/*ih*/, p_c_body, M_PASTE);
1859 	return 0;
1860     }
1861     retval = (retval == NO_DISK_SPACE) ? -ENOSPC : -EIO;
1862 error_out:
1863     /* this also releases the path */
1864     unfix_nodes(&s_paste_balance);
1865     return retval ;
1866 }
1867 
1868 
1869 /* Insert new item into the buffer at the path. */
reiserfs_insert_item(struct reiserfs_transaction_handle * th,struct path * p_s_path,const struct cpu_key * key,struct item_head * p_s_ih,const char * p_c_body)1870 int reiserfs_insert_item(struct reiserfs_transaction_handle *th,
1871 			 struct path         * 	p_s_path,         /* Path to the inserteded item.         */
1872 			 const struct cpu_key      * key,
1873 			 struct item_head    * 	p_s_ih,           /* Pointer to the item header to insert.*/
1874 			 const char          * 	p_c_body)         /* Pointer to the bytes to insert.      */
1875 {
1876     struct tree_balance s_ins_balance;
1877     int                 retval;
1878 
1879     init_tb_struct(th, &s_ins_balance, th->t_super, p_s_path, IH_SIZE + ih_item_len(p_s_ih));
1880 #ifdef DISPLACE_NEW_PACKING_LOCALITIES
1881     s_ins_balance.key = key->on_disk_key;
1882 #endif
1883 
1884     /*
1885     if (p_c_body == 0)
1886       n_zeros_num = ih_item_len(p_s_ih);
1887     */
1888     //    le_key2cpu_key (&key, &(p_s_ih->ih_key));
1889 
1890     while ( (retval = fix_nodes(M_INSERT, &s_ins_balance, p_s_ih, p_c_body)) == REPEAT_SEARCH) {
1891 	/* file system changed while we were in the fix_nodes */
1892 	PROC_INFO_INC( th -> t_super, insert_item_restarted );
1893 	retval = search_item (th->t_super, key, p_s_path);
1894 	if (retval == IO_ERROR) {
1895 	    retval = -EIO;
1896 	    goto error_out ;
1897 	}
1898 	if (retval == ITEM_FOUND) {
1899 	    reiserfs_warning (th->t_super, "PAP-5760: reiserfs_insert_item: "
1900 			      "key %K already exists in the tree\n", key);
1901 	    retval = -EEXIST ;
1902 	    goto error_out;
1903 	}
1904     }
1905 
1906     /* make balancing after all resources will be collected at a time */
1907     if ( retval == CARRY_ON ) {
1908 	do_balance (&s_ins_balance, p_s_ih, p_c_body, M_INSERT);
1909 	return 0;
1910     }
1911 
1912     retval = (retval == NO_DISK_SPACE) ? -ENOSPC : -EIO;
1913 error_out:
1914     /* also releases the path */
1915     unfix_nodes(&s_ins_balance);
1916     return retval;
1917 }
1918 
1919 
1920 
1921 
1922