X-Git-Url: https://mattmccutchen.net/rsync/rsync.git/blobdiff_plain/d9bca0c32fe5c76668ec89c64dc5ec26a838f40e..fdd10da6f7995296f6bb5ee3dd7231d6dbc20b5e:/match.c diff --git a/match.c b/match.c index 4fe23fd5..72947d0a 100644 --- a/match.c +++ b/match.c @@ -1,21 +1,24 @@ /* - Copyright (C) Andrew Tridgell 1996 - Copyright (C) Paul Mackerras 1996 - - This program is free software; you can redistribute it and/or modify - it under the terms of the GNU General Public License as published by - the Free Software Foundation; either version 2 of the License, or - (at your option) any later version. - - This program is distributed in the hope that it will be useful, - but WITHOUT ANY WARRANTY; without even the implied warranty of - MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - GNU General Public License for more details. - - You should have received a copy of the GNU General Public License - along with this program; if not, write to the Free Software - Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. -*/ + * Block matching used by the file-transfer code. + * + * Copyright (C) 1996 Andrew Tridgell + * Copyright (C) 1996 Paul Mackerras + * Copyright (C) 2003, 2004, 2005, 2006 Wayne Davison + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License along + * with this program; if not, write to the Free Software Foundation, Inc., + * 51 Franklin Street - Fifth Floor, Boston, MA 02110-1301, USA. + */ #include "rsync.h" @@ -27,46 +30,39 @@ extern int append_mode; int updating_basis_file; static int false_alarms; -static int tag_hits; +static int hash_hits; static int matches; static int64 data_transfer; static int total_false_alarms; -static int total_tag_hits; +static int total_hash_hits; static int total_matches; extern struct stats stats; -static uint32 tablesize; -static int32 *sum_table; +#define TABLESIZE (1<<16) + +static int32 *hash_table; -#define gettag2(s1,s2) gettag((s1) + ((s2)<<16)) -#define gettag(sum) ((sum)%tablesize) +#define SUM2HASH2(s1,s2) (((s1) + (s2)) & 0xFFFF) +#define SUM2HASH(sum) SUM2HASH2((sum)&0xFFFF,(sum)>>16) static void build_hash_table(struct sum_struct *s) { int32 i; - uint32 prior_size = tablesize; - - /* Dynamically calculate the hash table size so that the hash load - * for big files is about 80%. This number must be odd or s2 will - * not be able to span the entire set. */ - tablesize = (uint32)(s->count/8) * 10 + 11; - if (tablesize < 65537) - tablesize = 65537; /* a prime number */ - if (tablesize != prior_size) { - free(sum_table); - sum_table = new_array(int32, tablesize); - if (!sum_table) + + if (!hash_table) { + hash_table = new_array(int32, TABLESIZE); + if (!hash_table) out_of_memory("build_hash_table"); } - memset(sum_table, 0xFF, tablesize * sizeof sum_table[0]); + memset(hash_table, 0xFF, TABLESIZE * sizeof hash_table[0]); for (i = 0; i < s->count; i++) { - uint32 t = gettag(s->sums[i].sum1); - s->sums[i].chain = sum_table[t]; - sum_table[t] = i; + uint32 t = SUM2HASH(s->sums[i].sum1); + s->sums[i].chain = hash_table[t]; + hash_table[t] = i; } } @@ -161,16 +157,21 @@ static void hash_search(int f,struct sum_struct *s, } do { - uint32 t = gettag2(s1,s2); int done_csum2 = 0; int32 i; - if (verbose > 4) - rprintf(FINFO,"offset=%.0f sum=%08x\n",(double)offset,sum); + if (verbose > 4) { + rprintf(FINFO, "offset=%.0f sum=%04x%04x\n", + (double)offset, s2 & 0xFFFF, s1 & 0xFFFF); + } + + i = hash_table[SUM2HASH2(s1,s2)]; + if (i < 0) + goto null_hash; sum = (s1 & 0xffff) | (s2 << 16); - tag_hits++; - for (i = sum_table[t]; i >= 0; i = s->sums[i].chain) { + hash_hits++; + do { int32 l; if (sum != s->sums[i].sum1) @@ -250,8 +251,9 @@ static void hash_search(int f,struct sum_struct *s, s2 = sum >> 16; matches++; break; - } + } while ((i = s->sums[i].chain) >= 0); + null_hash: backup = offset - last_match; /* We sometimes read 1 byte prior to last_match... */ if (backup < 0) @@ -306,7 +308,7 @@ void match_sums(int f, struct sum_struct *s, struct map_struct *buf, OFF_T len) last_match = 0; false_alarms = 0; - tag_hits = 0; + hash_hits = 0; matches = 0; data_transfer = 0; @@ -359,10 +361,10 @@ void match_sums(int f, struct sum_struct *s, struct map_struct *buf, OFF_T len) write_buf(f,file_sum,MD4_SUM_LENGTH); if (verbose > 2) - rprintf(FINFO, "false_alarms=%d tag_hits=%d matches=%d\n", - false_alarms, tag_hits, matches); + rprintf(FINFO, "false_alarms=%d hash_hits=%d matches=%d\n", + false_alarms, hash_hits, matches); - total_tag_hits += tag_hits; + total_hash_hits += hash_hits; total_false_alarms += false_alarms; total_matches += matches; stats.literal_data += data_transfer; @@ -374,8 +376,7 @@ void match_report(void) return; rprintf(FINFO, - "total: matches=%d tag_hits=%d false_alarms=%d data=%.0f\n", - total_matches,total_tag_hits, - total_false_alarms, + "total: matches=%d hash_hits=%d false_alarms=%d data=%.0f\n", + total_matches, total_hash_hits, total_false_alarms, (double)stats.literal_data); }