1 /*
2  *   Copyright (c) International Business Machines Corp., 2000-2002
3  *
4  *   This program is free software;  you can redistribute it and/or modify
5  *   it under the terms of the GNU General Public License as published by
6  *   the Free Software Foundation; either version 2 of the License, or
7  *   (at your option) any later version.
8  *
9  *   This program is distributed in the hope that it will be useful,
10  *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
11  *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
12  *   the 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 to the Free Software
16  *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17  */
18 #ifndef	_H_JFS_DMAP
19 #define _H_JFS_DMAP
20 
21 #include "jfs_txnmgr.h"
22 
23 #define BMAPVERSION	1	/* version number */
24 #define	TREESIZE	(256+64+16+4+1)	/* size of a dmap tree */
25 #define	LEAFIND		(64+16+4+1)	/* index of 1st leaf of a dmap tree */
26 #define LPERDMAP	256	/* num leaves per dmap tree */
27 #define L2LPERDMAP	8	/* l2 number of leaves per dmap tree */
28 #define	DBWORD		32	/* # of blks covered by a map word */
29 #define	L2DBWORD	5	/* l2 # of blks covered by a mword */
30 #define BUDMIN  	L2DBWORD	/* max free string in a map word */
31 #define BPERDMAP	(LPERDMAP * DBWORD)	/* num of blks per dmap */
32 #define L2BPERDMAP	13	/* l2 num of blks per dmap */
33 #define CTLTREESIZE	(1024+256+64+16+4+1)	/* size of a dmapctl tree */
34 #define CTLLEAFIND	(256+64+16+4+1)	/* idx of 1st leaf of a dmapctl tree */
35 #define LPERCTL		1024	/* num of leaves per dmapctl tree */
36 #define L2LPERCTL	10	/* l2 num of leaves per dmapctl tree */
37 #define	ROOT		0	/* index of the root of a tree */
38 #define	NOFREE		((s8) -1)	/* no blocks free */
39 #define	MAXAG		128	/* max number of allocation groups */
40 #define L2MAXAG		7	/* l2 max num of AG */
41 #define L2MINAGSZ	25	/* l2 of minimum AG size in bytes */
42 #define	BMAPBLKNO	0	/* lblkno of bmap within the map */
43 
44 /*
45  * maximum l2 number of disk blocks at the various dmapctl levels.
46  */
47 #define	L2MAXL0SIZE	(L2BPERDMAP + 1 * L2LPERCTL)
48 #define	L2MAXL1SIZE	(L2BPERDMAP + 2 * L2LPERCTL)
49 #define	L2MAXL2SIZE	(L2BPERDMAP + 3 * L2LPERCTL)
50 
51 /*
52  * maximum number of disk blocks at the various dmapctl levels.
53  */
54 #define	MAXL0SIZE	((s64)1 << L2MAXL0SIZE)
55 #define	MAXL1SIZE	((s64)1 << L2MAXL1SIZE)
56 #define	MAXL2SIZE	((s64)1 << L2MAXL2SIZE)
57 
58 #define	MAXMAPSIZE	MAXL2SIZE	/* maximum aggregate map size */
59 
60 /*
61  * determine the maximum free string for four (lower level) nodes
62  * of the tree.
63  */
TREEMAX(signed char * cp)64 static __inline signed char TREEMAX(signed char *cp)
65 {
66 	signed char tmp1, tmp2;
67 
68 	tmp1 = max(*(cp+2), *(cp+3));
69 	tmp2 = max(*(cp), *(cp+1));
70 
71 	return max(tmp1, tmp2);
72 }
73 
74 /*
75  * convert disk block number to the logical block number of the dmap
76  * describing the disk block.  s is the log2(number of logical blocks per page)
77  *
78  * The calculation figures out how many logical pages are in front of the dmap.
79  *	- the number of dmaps preceding it
80  *	- the number of L0 pages preceding its L0 page
81  *	- the number of L1 pages preceding its L1 page
82  *	- 3 is added to account for the L2, L1, and L0 page for this dmap
83  *	- 1 is added to account for the control page of the map.
84  */
85 #define BLKTODMAP(b,s)    \
86         ((((b) >> 13) + ((b) >> 23) + ((b) >> 33) + 3 + 1) << (s))
87 
88 /*
89  * convert disk block number to the logical block number of the LEVEL 0
90  * dmapctl describing the disk block.  s is the log2(number of logical blocks
91  * per page)
92  *
93  * The calculation figures out how many logical pages are in front of the L0.
94  *	- the number of dmap pages preceding it
95  *	- the number of L0 pages preceding it
96  *	- the number of L1 pages preceding its L1 page
97  *	- 2 is added to account for the L2, and L1 page for this L0
98  *	- 1 is added to account for the control page of the map.
99  */
100 #define BLKTOL0(b,s)      \
101         (((((b) >> 23) << 10) + ((b) >> 23) + ((b) >> 33) + 2 + 1) << (s))
102 
103 /*
104  * convert disk block number to the logical block number of the LEVEL 1
105  * dmapctl describing the disk block.  s is the log2(number of logical blocks
106  * per page)
107  *
108  * The calculation figures out how many logical pages are in front of the L1.
109  *	- the number of dmap pages preceding it
110  *	- the number of L0 pages preceding it
111  *	- the number of L1 pages preceding it
112  *	- 1 is added to account for the L2 page
113  *	- 1 is added to account for the control page of the map.
114  */
115 #define BLKTOL1(b,s)      \
116      (((((b) >> 33) << 20) + (((b) >> 33) << 10) + ((b) >> 33) + 1 + 1) << (s))
117 
118 /*
119  * convert disk block number to the logical block number of the dmapctl
120  * at the specified level which describes the disk block.
121  */
122 #define BLKTOCTL(b,s,l)   \
123         (((l) == 2) ? 1 : ((l) == 1) ? BLKTOL1((b),(s)) : BLKTOL0((b),(s)))
124 
125 /*
126  * convert aggregate map size to the zero origin dmapctl level of the
127  * top dmapctl.
128  */
129 #define	BMAPSZTOLEV(size)	\
130 	(((size) <= MAXL0SIZE) ? 0 : ((size) <= MAXL1SIZE) ? 1 : 2)
131 
132 /* convert disk block number to allocation group number.
133  */
134 #define BLKTOAG(b,sbi)	((b) >> ((sbi)->bmap->db_agl2size))
135 
136 /* convert allocation group number to starting disk block
137  * number.
138  */
139 #define AGTOBLK(a,ip)	\
140 	((s64)(a) << (JFS_SBI((ip)->i_sb)->bmap->db_agl2size))
141 
142 /*
143  *	dmap summary tree
144  *
145  * dmaptree must be consistent with dmapctl.
146  */
147 struct dmaptree {
148 	s32 nleafs;		/* 4: number of tree leafs      */
149 	s32 l2nleafs;		/* 4: l2 number of tree leafs   */
150 	s32 leafidx;		/* 4: index of first tree leaf  */
151 	s32 height;		/* 4: height of the tree        */
152 	s8 budmin;		/* 1: min l2 tree leaf value to combine */
153 	s8 stree[TREESIZE];	/* TREESIZE: tree               */
154 	u8 pad[2];		/* 2: pad to word boundary      */
155 };				/* - 360 -                      */
156 
157 /*
158  *	dmap page per 8K blocks bitmap
159  */
160 struct dmap {
161 	s32 nblocks;		/* 4: num blks covered by this dmap     */
162 	s32 nfree;		/* 4: num of free blks in this dmap     */
163 	s64 start;		/* 8: starting blkno for this dmap      */
164 	struct dmaptree tree;	/* 360: dmap tree                       */
165 	u8 pad[1672];		/* 1672: pad to 2048 bytes              */
166 	u32 wmap[LPERDMAP];	/* 1024: bits of the working map        */
167 	u32 pmap[LPERDMAP];	/* 1024: bits of the persistent map     */
168 };				/* - 4096 -                             */
169 
170 /*
171  *	disk map control page per level.
172  *
173  * dmapctl must be consistent with dmaptree.
174  */
175 struct dmapctl {
176 	s32 nleafs;		/* 4: number of tree leafs      */
177 	s32 l2nleafs;		/* 4: l2 number of tree leafs   */
178 	s32 leafidx;		/* 4: index of the first tree leaf      */
179 	s32 height;		/* 4: height of tree            */
180 	s8 budmin;		/* 1: minimum l2 tree leaf value        */
181 	s8 stree[CTLTREESIZE];	/* CTLTREESIZE: dmapctl tree    */
182 	u8 pad[2714];		/* 2714: pad to 4096            */
183 };				/* - 4096 -                     */
184 
185 /*
186  *	common definition for dmaptree within dmap and dmapctl
187  */
188 typedef union dmtree {
189 	struct dmaptree t1;
190 	struct dmapctl t2;
191 } dmtree_t;
192 
193 /* macros for accessing fields within dmtree */
194 #define	dmt_nleafs	t1.nleafs
195 #define	dmt_l2nleafs 	t1.l2nleafs
196 #define	dmt_leafidx 	t1.leafidx
197 #define	dmt_height 	t1.height
198 #define	dmt_budmin 	t1.budmin
199 #define	dmt_stree 	t1.stree
200 
201 /*
202  *	on-disk aggregate disk allocation map descriptor.
203  */
204 struct dbmap {
205 	s64 dn_mapsize;		/* 8: number of blocks in aggregate     */
206 	s64 dn_nfree;		/* 8: num free blks in aggregate map    */
207 	s32 dn_l2nbperpage;	/* 4: number of blks per page           */
208 	s32 dn_numag;		/* 4: total number of ags               */
209 	s32 dn_maxlevel;	/* 4: number of active ags              */
210 	s32 dn_maxag;		/* 4: max active alloc group number     */
211 	s32 dn_agpref;		/* 4: preferred alloc group (hint)      */
212 	s32 dn_aglevel;		/* 4: dmapctl level holding the AG      */
213 	s32 dn_agheigth;	/* 4: height in dmapctl of the AG       */
214 	s32 dn_agwidth;		/* 4: width in dmapctl of the AG        */
215 	s32 dn_agstart;		/* 4: start tree index at AG height     */
216 	s32 dn_agl2size;	/* 4: l2 num of blks per alloc group    */
217 	s64 dn_agfree[MAXAG];	/* 8*MAXAG: per AG free count           */
218 	s64 dn_agsize;		/* 8: num of blks per alloc group       */
219 	s8 dn_maxfreebud;	/* 1: max free buddy system             */
220 	u8 pad[3007];		/* 3007: pad to 4096                    */
221 };				/* - 4096 -                             */
222 
223 /*
224  *	in-memory aggregate disk allocation map descriptor.
225  */
226 struct bmap {
227 	struct dbmap db_bmap;	/* on-disk aggregate map descriptor */
228 	struct inode *db_ipbmap;	/* ptr to aggregate map incore inode */
229 	struct semaphore db_bmaplock;	/* aggregate map lock */
230 	atomic_t db_active[MAXAG];	/* count of active, open files in AG */
231 	u32 *db_DBmap;
232 };
233 
234 /* macros for accessing fields within in-memory aggregate map descriptor */
235 #define	db_mapsize	db_bmap.dn_mapsize
236 #define	db_nfree	db_bmap.dn_nfree
237 #define	db_agfree	db_bmap.dn_agfree
238 #define	db_agsize	db_bmap.dn_agsize
239 #define	db_agl2size	db_bmap.dn_agl2size
240 #define	db_agwidth	db_bmap.dn_agwidth
241 #define	db_agheigth	db_bmap.dn_agheigth
242 #define	db_agstart	db_bmap.dn_agstart
243 #define	db_numag	db_bmap.dn_numag
244 #define	db_maxlevel	db_bmap.dn_maxlevel
245 #define	db_aglevel	db_bmap.dn_aglevel
246 #define	db_agpref	db_bmap.dn_agpref
247 #define	db_maxag	db_bmap.dn_maxag
248 #define	db_maxfreebud	db_bmap.dn_maxfreebud
249 #define	db_l2nbperpage	db_bmap.dn_l2nbperpage
250 
251 /*
252  * macros for various conversions needed by the allocators.
253  * blkstol2(), cntlz(), and cnttz() are operating system dependent functions.
254  */
255 /* convert number of blocks to log2 number of blocks, rounding up to
256  * the next log2 value if blocks is not a l2 multiple.
257  */
258 #define	BLKSTOL2(d)		(blkstol2(d))
259 
260 /* convert number of leafs to log2 leaf value */
261 #define	NLSTOL2BSZ(n)		(31 - cntlz((n)) + BUDMIN)
262 
263 /* convert leaf index to log2 leaf value */
264 #define	LITOL2BSZ(n,m,b)	((((n) == 0) ? (m) : cnttz((n))) + (b))
265 
266 /* convert a block number to a dmap control leaf index */
267 #define BLKTOCTLLEAF(b,m)	\
268 	(((b) & (((s64)1 << ((m) + L2LPERCTL)) - 1)) >> (m))
269 
270 /* convert log2 leaf value to buddy size */
271 #define	BUDSIZE(s,m)		(1 << ((s) - (m)))
272 
273 /*
274  *	external references.
275  */
276 extern int dbMount(struct inode *ipbmap);
277 
278 extern int dbUnmount(struct inode *ipbmap, int mounterror);
279 
280 extern int dbFree(struct inode *ipbmap, s64 blkno, s64 nblocks);
281 
282 extern int dbUpdatePMap(struct inode *ipbmap,
283 			int free, s64 blkno, s64 nblocks, struct tblock * tblk);
284 
285 extern int dbNextAG(struct inode *ipbmap);
286 
287 extern int dbAlloc(struct inode *ipbmap, s64 hint, s64 nblocks, s64 * results);
288 
289 extern int dbReAlloc(struct inode *ipbmap,
290 		     s64 blkno, s64 nblocks, s64 addnblocks, s64 * results);
291 
292 extern int dbSync(struct inode *ipbmap);
293 extern int dbAllocBottomUp(struct inode *ip, s64 blkno, s64 nblocks);
294 extern int dbExtendFS(struct inode *ipbmap, s64 blkno, s64 nblocks);
295 extern void dbFinalizeBmap(struct inode *ipbmap);
296 extern s64 dbMapFileSizeToMapSize(struct inode *ipbmap);
297 #endif				/* _H_JFS_DMAP */
298