Blame


1 7a4ee46d 2003-11-23 devnull #include "stdinc.h"
2 7a4ee46d 2003-11-23 devnull #include "dat.h"
3 7a4ee46d 2003-11-23 devnull #include "fns.h"
4 7a4ee46d 2003-11-23 devnull
5 7a4ee46d 2003-11-23 devnull static int
6 7a4ee46d 2003-11-23 devnull writebucket(Index *ix, u32int buck, IBucket *ib, ZBlock *b)
7 7a4ee46d 2003-11-23 devnull {
8 7a4ee46d 2003-11-23 devnull ISect *is;
9 7a4ee46d 2003-11-23 devnull
10 9ffbb5ad 2004-03-12 devnull is = findibucket(ix, buck, &buck);
11 9ffbb5ad 2004-03-12 devnull if(is == nil)
12 7a4ee46d 2003-11-23 devnull return -1;
13 7a4ee46d 2003-11-23 devnull qlock(&stats.lock);
14 7a4ee46d 2003-11-23 devnull stats.indexwrites++;
15 7a4ee46d 2003-11-23 devnull qunlock(&stats.lock);
16 7a4ee46d 2003-11-23 devnull packibucket(ib, b->data);
17 7a4ee46d 2003-11-23 devnull return writepart(is->part, is->blockbase + ((u64int)buck << is->blocklog), b->data, is->blocksize);
18 7a4ee46d 2003-11-23 devnull }
19 7a4ee46d 2003-11-23 devnull
20 7a4ee46d 2003-11-23 devnull static int
21 7a4ee46d 2003-11-23 devnull buildindex(Index *ix, Part *part, u64int off, u64int clumps, int zero)
22 7a4ee46d 2003-11-23 devnull {
23 7a4ee46d 2003-11-23 devnull IEStream *ies;
24 7a4ee46d 2003-11-23 devnull IBucket ib, zib;
25 7a4ee46d 2003-11-23 devnull ZBlock *z, *b;
26 7a4ee46d 2003-11-23 devnull u32int next, buck;
27 7a4ee46d 2003-11-23 devnull int ok;
28 7a4ee46d 2003-11-23 devnull u64int found = 0;
29 7a4ee46d 2003-11-23 devnull
30 7a4ee46d 2003-11-23 devnull //ZZZ make buffer size configurable
31 7a4ee46d 2003-11-23 devnull b = alloczblock(ix->blocksize, 0);
32 7a4ee46d 2003-11-23 devnull z = alloczblock(ix->blocksize, 1);
33 7a4ee46d 2003-11-23 devnull ies = initiestream(part, off, clumps, 64*1024);
34 7a4ee46d 2003-11-23 devnull if(b == nil || z == nil || ies == nil){
35 7a4ee46d 2003-11-23 devnull ok = 0;
36 7a4ee46d 2003-11-23 devnull goto breakout;
37 7a4ee46d 2003-11-23 devnull return -1;
38 7a4ee46d 2003-11-23 devnull }
39 7a4ee46d 2003-11-23 devnull ok = 0;
40 7a4ee46d 2003-11-23 devnull next = 0;
41 7a4ee46d 2003-11-23 devnull ib.data = b->data + IBucketSize;
42 7a4ee46d 2003-11-23 devnull zib.data = z->data + IBucketSize;
43 7a4ee46d 2003-11-23 devnull zib.n = 0;
44 9ffbb5ad 2004-03-12 devnull zib.depth = 0;
45 7a4ee46d 2003-11-23 devnull for(;;){
46 7a4ee46d 2003-11-23 devnull buck = buildbucket(ix, ies, &ib);
47 7a4ee46d 2003-11-23 devnull found += ib.n;
48 7a4ee46d 2003-11-23 devnull if(zero){
49 7a4ee46d 2003-11-23 devnull for(; next != buck; next++){
50 7a4ee46d 2003-11-23 devnull if(next == ix->buckets){
51 7a4ee46d 2003-11-23 devnull if(buck != TWID32){
52 7a4ee46d 2003-11-23 devnull fprint(2, "bucket out of range\n");
53 7a4ee46d 2003-11-23 devnull ok = -1;
54 7a4ee46d 2003-11-23 devnull }
55 7a4ee46d 2003-11-23 devnull goto breakout;
56 7a4ee46d 2003-11-23 devnull }
57 7a4ee46d 2003-11-23 devnull if(writebucket(ix, next, &zib, z) < 0){
58 7a4ee46d 2003-11-23 devnull fprint(2, "can't write zero bucket to buck=%d: %r", next);
59 7a4ee46d 2003-11-23 devnull ok = -1;
60 7a4ee46d 2003-11-23 devnull }
61 7a4ee46d 2003-11-23 devnull }
62 7a4ee46d 2003-11-23 devnull }
63 7a4ee46d 2003-11-23 devnull if(buck >= ix->buckets){
64 7a4ee46d 2003-11-23 devnull if(buck == TWID32)
65 7a4ee46d 2003-11-23 devnull break;
66 7a4ee46d 2003-11-23 devnull fprint(2, "bucket out of range\n");
67 7a4ee46d 2003-11-23 devnull ok = -1;
68 7a4ee46d 2003-11-23 devnull goto breakout;
69 7a4ee46d 2003-11-23 devnull }
70 7a4ee46d 2003-11-23 devnull if(writebucket(ix, buck, &ib, b) < 0){
71 7a4ee46d 2003-11-23 devnull fprint(2, "bad bucket found=%lld: %r\n", found);
72 7a4ee46d 2003-11-23 devnull ok = -1;
73 7a4ee46d 2003-11-23 devnull }
74 7a4ee46d 2003-11-23 devnull next = buck + 1;
75 7a4ee46d 2003-11-23 devnull }
76 7a4ee46d 2003-11-23 devnull breakout:;
77 7a4ee46d 2003-11-23 devnull fprint(2, "constructed index with %lld entries\n", found);
78 7a4ee46d 2003-11-23 devnull freeiestream(ies);
79 7a4ee46d 2003-11-23 devnull freezblock(z);
80 7a4ee46d 2003-11-23 devnull freezblock(b);
81 7a4ee46d 2003-11-23 devnull return ok;
82 7a4ee46d 2003-11-23 devnull }
83 7a4ee46d 2003-11-23 devnull
84 7a4ee46d 2003-11-23 devnull void
85 7a4ee46d 2003-11-23 devnull usage(void)
86 7a4ee46d 2003-11-23 devnull {
87 7a4ee46d 2003-11-23 devnull fprint(2, "usage: buildindex [-Z] [-B blockcachesize] config tmppart\n");
88 7a4ee46d 2003-11-23 devnull threadexitsall(0);
89 7a4ee46d 2003-11-23 devnull }
90 7a4ee46d 2003-11-23 devnull
91 7a4ee46d 2003-11-23 devnull void
92 7a4ee46d 2003-11-23 devnull threadmain(int argc, char *argv[])
93 7a4ee46d 2003-11-23 devnull {
94 7a4ee46d 2003-11-23 devnull Part *part;
95 7a4ee46d 2003-11-23 devnull u64int clumps, base;
96 7a4ee46d 2003-11-23 devnull u32int bcmem;
97 7a4ee46d 2003-11-23 devnull int zero;
98 7a4ee46d 2003-11-23 devnull
99 7a4ee46d 2003-11-23 devnull zero = 1;
100 7a4ee46d 2003-11-23 devnull bcmem = 0;
101 7a4ee46d 2003-11-23 devnull ARGBEGIN{
102 7a4ee46d 2003-11-23 devnull case 'B':
103 7a4ee46d 2003-11-23 devnull bcmem = unittoull(ARGF());
104 7a4ee46d 2003-11-23 devnull break;
105 7a4ee46d 2003-11-23 devnull case 'Z':
106 7a4ee46d 2003-11-23 devnull zero = 0;
107 7a4ee46d 2003-11-23 devnull break;
108 7a4ee46d 2003-11-23 devnull default:
109 7a4ee46d 2003-11-23 devnull usage();
110 7a4ee46d 2003-11-23 devnull break;
111 7a4ee46d 2003-11-23 devnull }ARGEND
112 7a4ee46d 2003-11-23 devnull
113 7a4ee46d 2003-11-23 devnull if(argc != 2)
114 7a4ee46d 2003-11-23 devnull usage();
115 7a4ee46d 2003-11-23 devnull
116 7a4ee46d 2003-11-23 devnull if(initventi(argv[0]) < 0)
117 7a4ee46d 2003-11-23 devnull sysfatal("can't init venti: %r");
118 7a4ee46d 2003-11-23 devnull
119 7a4ee46d 2003-11-23 devnull if(bcmem < maxblocksize * (mainindex->narenas + mainindex->nsects * 4 + 16))
120 7a4ee46d 2003-11-23 devnull bcmem = maxblocksize * (mainindex->narenas + mainindex->nsects * 4 + 16);
121 7a4ee46d 2003-11-23 devnull fprint(2, "initialize %d bytes of disk block cache\n", bcmem);
122 7a4ee46d 2003-11-23 devnull initdcache(bcmem);
123 7a4ee46d 2003-11-23 devnull
124 7a4ee46d 2003-11-23 devnull fprint(2, "building a new index %s using %s for temporary storage\n", mainindex->name, argv[1]);
125 7a4ee46d 2003-11-23 devnull
126 7a4ee46d 2003-11-23 devnull part = initpart(argv[1], 1);
127 7a4ee46d 2003-11-23 devnull if(part == nil)
128 7a4ee46d 2003-11-23 devnull sysfatal("can't initialize temporary partition: %r");
129 7a4ee46d 2003-11-23 devnull
130 7a4ee46d 2003-11-23 devnull clumps = sortrawientries(mainindex, part, &base);
131 7a4ee46d 2003-11-23 devnull if(clumps == TWID64)
132 7a4ee46d 2003-11-23 devnull sysfatal("can't build sorted index: %r");
133 7a4ee46d 2003-11-23 devnull fprint(2, "found and sorted index entries for clumps=%lld at %lld\n", clumps, base);
134 7a4ee46d 2003-11-23 devnull
135 7a4ee46d 2003-11-23 devnull if(buildindex(mainindex, part, base, clumps, zero) < 0)
136 7a4ee46d 2003-11-23 devnull sysfatal("can't build new index: %r");
137 7a4ee46d 2003-11-23 devnull
138 7a4ee46d 2003-11-23 devnull threadexitsall(0);
139 7a4ee46d 2003-11-23 devnull }