1/*
2 * Copyright (c) 2000-2001,2005 Silicon Graphics, Inc.
3 * All Rights Reserved.
4 *
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.
8 *
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.
13 *
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
17 */
18#include "xfs.h"
19#include "xfs_fs.h"
20#include "xfs_types.h"
21#include "xfs_bit.h"
22#include "xfs_log.h"
23#include "xfs_inum.h"
24#include "xfs_trans.h"
25#include "xfs_sb.h"
26#include "xfs_ag.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_dinode.h"
32#include "xfs_inode.h"
33#include "xfs_btree.h"
34#include "xfs_btree_trace.h"
35#include "xfs_alloc.h"
36#include "xfs_error.h"
37#include "xfs_trace.h"
38
39
40STATIC struct xfs_btree_cur *
41xfs_allocbt_dup_cursor(
42	struct xfs_btree_cur	*cur)
43{
44	return xfs_allocbt_init_cursor(cur->bc_mp, cur->bc_tp,
45			cur->bc_private.a.agbp, cur->bc_private.a.agno,
46			cur->bc_btnum);
47}
48
49STATIC void
50xfs_allocbt_set_root(
51	struct xfs_btree_cur	*cur,
52	union xfs_btree_ptr	*ptr,
53	int			inc)
54{
55	struct xfs_buf		*agbp = cur->bc_private.a.agbp;
56	struct xfs_agf		*agf = XFS_BUF_TO_AGF(agbp);
57	xfs_agnumber_t		seqno = be32_to_cpu(agf->agf_seqno);
58	int			btnum = cur->bc_btnum;
59	struct xfs_perag	*pag = xfs_perag_get(cur->bc_mp, seqno);
60
61	ASSERT(ptr->s != 0);
62
63	agf->agf_roots[btnum] = ptr->s;
64	be32_add_cpu(&agf->agf_levels[btnum], inc);
65	pag->pagf_levels[btnum] += inc;
66	xfs_perag_put(pag);
67
68	xfs_alloc_log_agf(cur->bc_tp, agbp, XFS_AGF_ROOTS | XFS_AGF_LEVELS);
69}
70
71STATIC int
72xfs_allocbt_alloc_block(
73	struct xfs_btree_cur	*cur,
74	union xfs_btree_ptr	*start,
75	union xfs_btree_ptr	*new,
76	int			length,
77	int			*stat)
78{
79	int			error;
80	xfs_agblock_t		bno;
81
82	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
83
84	/* Allocate the new block from the freelist. If we can't, give up.  */
85	error = xfs_alloc_get_freelist(cur->bc_tp, cur->bc_private.a.agbp,
86				       &bno, 1);
87	if (error) {
88		XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
89		return error;
90	}
91
92	if (bno == NULLAGBLOCK) {
93		XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
94		*stat = 0;
95		return 0;
96	}
97
98	xfs_trans_agbtree_delta(cur->bc_tp, 1);
99	new->s = cpu_to_be32(bno);
100
101	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
102	*stat = 1;
103	return 0;
104}
105
106STATIC int
107xfs_allocbt_free_block(
108	struct xfs_btree_cur	*cur,
109	struct xfs_buf		*bp)
110{
111	struct xfs_buf		*agbp = cur->bc_private.a.agbp;
112	struct xfs_agf		*agf = XFS_BUF_TO_AGF(agbp);
113	xfs_agblock_t		bno;
114	int			error;
115
116	bno = xfs_daddr_to_agbno(cur->bc_mp, XFS_BUF_ADDR(bp));
117	error = xfs_alloc_put_freelist(cur->bc_tp, agbp, NULL, bno, 1);
118	if (error)
119		return error;
120
121	/*
122	 * Since blocks move to the free list without the coordination used in
123	 * xfs_bmap_finish, we can't allow block to be available for
124	 * reallocation and non-transaction writing (user data) until we know
125	 * that the transaction that moved it to the free list is permanently
126	 * on disk. We track the blocks by declaring these blocks as "busy";
127	 * the busy list is maintained on a per-ag basis and each transaction
128	 * records which entries should be removed when the iclog commits to
129	 * disk. If a busy block is allocated, the iclog is pushed up to the
130	 * LSN that freed the block.
131	 */
132	xfs_alloc_busy_insert(cur->bc_tp, be32_to_cpu(agf->agf_seqno), bno, 1);
133	xfs_trans_agbtree_delta(cur->bc_tp, -1);
134	return 0;
135}
136
137/*
138 * Update the longest extent in the AGF
139 */
140STATIC void
141xfs_allocbt_update_lastrec(
142	struct xfs_btree_cur	*cur,
143	struct xfs_btree_block	*block,
144	union xfs_btree_rec	*rec,
145	int			ptr,
146	int			reason)
147{
148	struct xfs_agf		*agf = XFS_BUF_TO_AGF(cur->bc_private.a.agbp);
149	xfs_agnumber_t		seqno = be32_to_cpu(agf->agf_seqno);
150	struct xfs_perag	*pag;
151	__be32			len;
152	int			numrecs;
153
154	ASSERT(cur->bc_btnum == XFS_BTNUM_CNT);
155
156	switch (reason) {
157	case LASTREC_UPDATE:
158		/*
159		 * If this is the last leaf block and it's the last record,
160		 * then update the size of the longest extent in the AG.
161		 */
162		if (ptr != xfs_btree_get_numrecs(block))
163			return;
164		len = rec->alloc.ar_blockcount;
165		break;
166	case LASTREC_INSREC:
167		if (be32_to_cpu(rec->alloc.ar_blockcount) <=
168		    be32_to_cpu(agf->agf_longest))
169			return;
170		len = rec->alloc.ar_blockcount;
171		break;
172	case LASTREC_DELREC:
173		numrecs = xfs_btree_get_numrecs(block);
174		if (ptr <= numrecs)
175			return;
176		ASSERT(ptr == numrecs + 1);
177
178		if (numrecs) {
179			xfs_alloc_rec_t *rrp;
180
181			rrp = XFS_ALLOC_REC_ADDR(cur->bc_mp, block, numrecs);
182			len = rrp->ar_blockcount;
183		} else {
184			len = 0;
185		}
186
187		break;
188	default:
189		ASSERT(0);
190		return;
191	}
192
193	agf->agf_longest = len;
194	pag = xfs_perag_get(cur->bc_mp, seqno);
195	pag->pagf_longest = be32_to_cpu(len);
196	xfs_perag_put(pag);
197	xfs_alloc_log_agf(cur->bc_tp, cur->bc_private.a.agbp, XFS_AGF_LONGEST);
198}
199
200STATIC int
201xfs_allocbt_get_minrecs(
202	struct xfs_btree_cur	*cur,
203	int			level)
204{
205	return cur->bc_mp->m_alloc_mnr[level != 0];
206}
207
208STATIC int
209xfs_allocbt_get_maxrecs(
210	struct xfs_btree_cur	*cur,
211	int			level)
212{
213	return cur->bc_mp->m_alloc_mxr[level != 0];
214}
215
216STATIC void
217xfs_allocbt_init_key_from_rec(
218	union xfs_btree_key	*key,
219	union xfs_btree_rec	*rec)
220{
221	ASSERT(rec->alloc.ar_startblock != 0);
222
223	key->alloc.ar_startblock = rec->alloc.ar_startblock;
224	key->alloc.ar_blockcount = rec->alloc.ar_blockcount;
225}
226
227STATIC void
228xfs_allocbt_init_rec_from_key(
229	union xfs_btree_key	*key,
230	union xfs_btree_rec	*rec)
231{
232	ASSERT(key->alloc.ar_startblock != 0);
233
234	rec->alloc.ar_startblock = key->alloc.ar_startblock;
235	rec->alloc.ar_blockcount = key->alloc.ar_blockcount;
236}
237
238STATIC void
239xfs_allocbt_init_rec_from_cur(
240	struct xfs_btree_cur	*cur,
241	union xfs_btree_rec	*rec)
242{
243	ASSERT(cur->bc_rec.a.ar_startblock != 0);
244
245	rec->alloc.ar_startblock = cpu_to_be32(cur->bc_rec.a.ar_startblock);
246	rec->alloc.ar_blockcount = cpu_to_be32(cur->bc_rec.a.ar_blockcount);
247}
248
249STATIC void
250xfs_allocbt_init_ptr_from_cur(
251	struct xfs_btree_cur	*cur,
252	union xfs_btree_ptr	*ptr)
253{
254	struct xfs_agf		*agf = XFS_BUF_TO_AGF(cur->bc_private.a.agbp);
255
256	ASSERT(cur->bc_private.a.agno == be32_to_cpu(agf->agf_seqno));
257	ASSERT(agf->agf_roots[cur->bc_btnum] != 0);
258
259	ptr->s = agf->agf_roots[cur->bc_btnum];
260}
261
262STATIC __int64_t
263xfs_allocbt_key_diff(
264	struct xfs_btree_cur	*cur,
265	union xfs_btree_key	*key)
266{
267	xfs_alloc_rec_incore_t	*rec = &cur->bc_rec.a;
268	xfs_alloc_key_t		*kp = &key->alloc;
269	__int64_t		diff;
270
271	if (cur->bc_btnum == XFS_BTNUM_BNO) {
272		return (__int64_t)be32_to_cpu(kp->ar_startblock) -
273				rec->ar_startblock;
274	}
275
276	diff = (__int64_t)be32_to_cpu(kp->ar_blockcount) - rec->ar_blockcount;
277	if (diff)
278		return diff;
279
280	return (__int64_t)be32_to_cpu(kp->ar_startblock) - rec->ar_startblock;
281}
282
283STATIC int
284xfs_allocbt_kill_root(
285	struct xfs_btree_cur	*cur,
286	struct xfs_buf		*bp,
287	int			level,
288	union xfs_btree_ptr	*newroot)
289{
290	int			error;
291
292	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
293	XFS_BTREE_STATS_INC(cur, killroot);
294
295	/*
296	 * Update the root pointer, decreasing the level by 1 and then
297	 * free the old root.
298	 */
299	xfs_allocbt_set_root(cur, newroot, -1);
300	error = xfs_allocbt_free_block(cur, bp);
301	if (error) {
302		XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
303		return error;
304	}
305
306	XFS_BTREE_STATS_INC(cur, free);
307
308	xfs_btree_setbuf(cur, level, NULL);
309	cur->bc_nlevels--;
310
311	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
312	return 0;
313}
314
315#ifdef DEBUG
316STATIC int
317xfs_allocbt_keys_inorder(
318	struct xfs_btree_cur	*cur,
319	union xfs_btree_key	*k1,
320	union xfs_btree_key	*k2)
321{
322	if (cur->bc_btnum == XFS_BTNUM_BNO) {
323		return be32_to_cpu(k1->alloc.ar_startblock) <
324		       be32_to_cpu(k2->alloc.ar_startblock);
325	} else {
326		return be32_to_cpu(k1->alloc.ar_blockcount) <
327			be32_to_cpu(k2->alloc.ar_blockcount) ||
328			(k1->alloc.ar_blockcount == k2->alloc.ar_blockcount &&
329			 be32_to_cpu(k1->alloc.ar_startblock) <
330			 be32_to_cpu(k2->alloc.ar_startblock));
331	}
332}
333
334STATIC int
335xfs_allocbt_recs_inorder(
336	struct xfs_btree_cur	*cur,
337	union xfs_btree_rec	*r1,
338	union xfs_btree_rec	*r2)
339{
340	if (cur->bc_btnum == XFS_BTNUM_BNO) {
341		return be32_to_cpu(r1->alloc.ar_startblock) +
342			be32_to_cpu(r1->alloc.ar_blockcount) <=
343			be32_to_cpu(r2->alloc.ar_startblock);
344	} else {
345		return be32_to_cpu(r1->alloc.ar_blockcount) <
346			be32_to_cpu(r2->alloc.ar_blockcount) ||
347			(r1->alloc.ar_blockcount == r2->alloc.ar_blockcount &&
348			 be32_to_cpu(r1->alloc.ar_startblock) <
349			 be32_to_cpu(r2->alloc.ar_startblock));
350	}
351}
352#endif	/* DEBUG */
353
354#ifdef XFS_BTREE_TRACE
355ktrace_t	*xfs_allocbt_trace_buf;
356
357STATIC void
358xfs_allocbt_trace_enter(
359	struct xfs_btree_cur	*cur,
360	const char		*func,
361	char			*s,
362	int			type,
363	int			line,
364	__psunsigned_t		a0,
365	__psunsigned_t		a1,
366	__psunsigned_t		a2,
367	__psunsigned_t		a3,
368	__psunsigned_t		a4,
369	__psunsigned_t		a5,
370	__psunsigned_t		a6,
371	__psunsigned_t		a7,
372	__psunsigned_t		a8,
373	__psunsigned_t		a9,
374	__psunsigned_t		a10)
375{
376	ktrace_enter(xfs_allocbt_trace_buf, (void *)(__psint_t)type,
377		(void *)func, (void *)s, NULL, (void *)cur,
378		(void *)a0, (void *)a1, (void *)a2, (void *)a3,
379		(void *)a4, (void *)a5, (void *)a6, (void *)a7,
380		(void *)a8, (void *)a9, (void *)a10);
381}
382
383STATIC void
384xfs_allocbt_trace_cursor(
385	struct xfs_btree_cur	*cur,
386	__uint32_t		*s0,
387	__uint64_t		*l0,
388	__uint64_t		*l1)
389{
390	*s0 = cur->bc_private.a.agno;
391	*l0 = cur->bc_rec.a.ar_startblock;
392	*l1 = cur->bc_rec.a.ar_blockcount;
393}
394
395STATIC void
396xfs_allocbt_trace_key(
397	struct xfs_btree_cur	*cur,
398	union xfs_btree_key	*key,
399	__uint64_t		*l0,
400	__uint64_t		*l1)
401{
402	*l0 = be32_to_cpu(key->alloc.ar_startblock);
403	*l1 = be32_to_cpu(key->alloc.ar_blockcount);
404}
405
406STATIC void
407xfs_allocbt_trace_record(
408	struct xfs_btree_cur	*cur,
409	union xfs_btree_rec	*rec,
410	__uint64_t		*l0,
411	__uint64_t		*l1,
412	__uint64_t		*l2)
413{
414	*l0 = be32_to_cpu(rec->alloc.ar_startblock);
415	*l1 = be32_to_cpu(rec->alloc.ar_blockcount);
416	*l2 = 0;
417}
418#endif /* XFS_BTREE_TRACE */
419
420static const struct xfs_btree_ops xfs_allocbt_ops = {
421	.rec_len		= sizeof(xfs_alloc_rec_t),
422	.key_len		= sizeof(xfs_alloc_key_t),
423
424	.dup_cursor		= xfs_allocbt_dup_cursor,
425	.set_root		= xfs_allocbt_set_root,
426	.kill_root		= xfs_allocbt_kill_root,
427	.alloc_block		= xfs_allocbt_alloc_block,
428	.free_block		= xfs_allocbt_free_block,
429	.update_lastrec		= xfs_allocbt_update_lastrec,
430	.get_minrecs		= xfs_allocbt_get_minrecs,
431	.get_maxrecs		= xfs_allocbt_get_maxrecs,
432	.init_key_from_rec	= xfs_allocbt_init_key_from_rec,
433	.init_rec_from_key	= xfs_allocbt_init_rec_from_key,
434	.init_rec_from_cur	= xfs_allocbt_init_rec_from_cur,
435	.init_ptr_from_cur	= xfs_allocbt_init_ptr_from_cur,
436	.key_diff		= xfs_allocbt_key_diff,
437
438#ifdef DEBUG
439	.keys_inorder		= xfs_allocbt_keys_inorder,
440	.recs_inorder		= xfs_allocbt_recs_inorder,
441#endif
442
443#ifdef XFS_BTREE_TRACE
444	.trace_enter		= xfs_allocbt_trace_enter,
445	.trace_cursor		= xfs_allocbt_trace_cursor,
446	.trace_key		= xfs_allocbt_trace_key,
447	.trace_record		= xfs_allocbt_trace_record,
448#endif
449};
450
451/*
452 * Allocate a new allocation btree cursor.
453 */
454struct xfs_btree_cur *			/* new alloc btree cursor */
455xfs_allocbt_init_cursor(
456	struct xfs_mount	*mp,		/* file system mount point */
457	struct xfs_trans	*tp,		/* transaction pointer */
458	struct xfs_buf		*agbp,		/* buffer for agf structure */
459	xfs_agnumber_t		agno,		/* allocation group number */
460	xfs_btnum_t		btnum)		/* btree identifier */
461{
462	struct xfs_agf		*agf = XFS_BUF_TO_AGF(agbp);
463	struct xfs_btree_cur	*cur;
464
465	ASSERT(btnum == XFS_BTNUM_BNO || btnum == XFS_BTNUM_CNT);
466
467	cur = kmem_zone_zalloc(xfs_btree_cur_zone, KM_SLEEP);
468
469	cur->bc_tp = tp;
470	cur->bc_mp = mp;
471	cur->bc_nlevels = be32_to_cpu(agf->agf_levels[btnum]);
472	cur->bc_btnum = btnum;
473	cur->bc_blocklog = mp->m_sb.sb_blocklog;
474
475	cur->bc_ops = &xfs_allocbt_ops;
476	if (btnum == XFS_BTNUM_CNT)
477		cur->bc_flags = XFS_BTREE_LASTREC_UPDATE;
478
479	cur->bc_private.a.agbp = agbp;
480	cur->bc_private.a.agno = agno;
481
482	return cur;
483}
484
485/*
486 * Calculate number of records in an alloc btree block.
487 */
488int
489xfs_allocbt_maxrecs(
490	struct xfs_mount	*mp,
491	int			blocklen,
492	int			leaf)
493{
494	blocklen -= XFS_ALLOC_BLOCK_LEN(mp);
495
496	if (leaf)
497		return blocklen / sizeof(xfs_alloc_rec_t);
498	return blocklen / (sizeof(xfs_alloc_key_t) + sizeof(xfs_alloc_ptr_t));
499}
500