X-Git-Url: https://git.octo.it/?a=blobdiff_plain;f=pack-objects.c;h=49357c67357fdc96e0d98b193b88e2137453c7c8;hb=4bc87a28be020a6bf7387161c65ea3d8e4a0228b;hp=32befa002817273a55721d9b7fa43a33c3cf9e1d;hpb=597888adc6006cbca02175b6f070f8ef46b014e6;p=git.git diff --git a/pack-objects.c b/pack-objects.c index 32befa00..49357c67 100644 --- a/pack-objects.c +++ b/pack-objects.c @@ -32,9 +32,6 @@ struct object_entry { * be used as the base objectto delta huge * objects against. */ - int based_on_preferred; /* current delta candidate is a preferred - * one, or delta against a preferred one. - */ }; /* @@ -99,7 +96,7 @@ static int reused_delta = 0; static int pack_revindex_ix(struct packed_git *p) { - unsigned int ui = (unsigned int) p; + unsigned long ui = (unsigned long)p; int i; ui = ui ^ (ui >> 16); /* defeat structure alignment */ @@ -462,15 +459,74 @@ static void rehash_objects(void) } } -static int add_object_entry(const unsigned char *sha1, const char *name, int exclude) +struct name_path { + struct name_path *up; + const char *elem; + int len; +}; + +#define DIRBITS 12 + +static unsigned name_hash(struct name_path *path, const char *name) +{ + struct name_path *p = path; + const char *n = name + strlen(name); + unsigned hash = 0, name_hash = 0, name_done = 0; + + if (n != name && n[-1] == '\n') + n--; + while (name <= --n) { + unsigned char c = *n; + if (c == '/' && !name_done) { + name_hash = hash; + name_done = 1; + hash = 0; + } + hash = hash * 11 + c; + } + if (!name_done) { + name_hash = hash; + hash = 0; + } + for (p = path; p; p = p->up) { + hash = hash * 11 + '/'; + n = p->elem + p->len; + while (p->elem <= --n) { + unsigned char c = *n; + hash = hash * 11 + c; + } + } + /* + * Make sure "Makefile" and "t/Makefile" are hashed separately + * but close enough. + */ + hash = (name_hash<up) { + fputc('/', stderr); + n = p->elem + p->len; + while (p->elem <= --n) + fputc(*n, stderr); + } + fprintf(stderr, "\t%08x\n", hash); + } + return hash; +} + +static int add_object_entry(const unsigned char *sha1, unsigned hash, int exclude) { - unsigned int hash = 0; unsigned int idx = nr_objects; struct object_entry *entry; struct packed_git *p; unsigned int found_offset = 0; struct packed_git *found_pack = NULL; - int ix; + int ix, status = 0; if (!exclude) { for (p = packed_git; p; p = p->next) { @@ -490,13 +546,6 @@ static int add_object_entry(const unsigned char *sha1, const char *name, int exc if ((entry = locate_object_entry(sha1)) != NULL) goto already_added; - while (*name) { - unsigned char c = *name++; - if (isspace(c)) - continue; - hash = hash * 11 + c; - } - if (idx >= nr_alloc) { unsigned int needed = (idx + 1024) * 3 / 2; objects = xrealloc(objects, needed * sizeof(*entry)); @@ -516,6 +565,7 @@ static int add_object_entry(const unsigned char *sha1, const char *name, int exc die("internal error in object hashing."); object_ix[-1 - ix] = idx + 1; } + status = 1; already_added: if (progress_update) { @@ -530,15 +580,15 @@ static int add_object_entry(const unsigned char *sha1, const char *name, int exc entry->in_pack_offset = found_offset; } } - return 1; + return status; } -static void add_pbase_tree(struct tree_desc *tree) +static void add_pbase_tree(struct tree_desc *tree, struct name_path *up) { while (tree->size) { const unsigned char *sha1; const char *name; - unsigned mode; + unsigned mode, hash; unsigned long size; char type[20]; @@ -548,14 +598,23 @@ static void add_pbase_tree(struct tree_desc *tree) continue; if (sha1_object_info(sha1, type, &size)) continue; - add_object_entry(sha1, name, 1); + + hash = name_hash(up, name); + if (!add_object_entry(sha1, hash, 1)) + continue; + if (!strcmp(type, "tree")) { struct tree_desc sub; void *elem; + struct name_path me; + elem = read_sha1_file(sha1, type, &sub.size); sub.buf = elem; if (sub.buf) { - add_pbase_tree(&sub); + me.up = up; + me.elem = name; + me.len = strlen(name); + add_pbase_tree(&sub, &me); free(elem); } } @@ -566,12 +625,13 @@ static void add_preferred_base(unsigned char *sha1) { struct tree_desc tree; void *elem; + elem = read_object_with_reference(sha1, "tree", &tree.size, NULL); tree.buf = elem; if (!tree.buf) return; - add_object_entry(sha1, "", 1); - add_pbase_tree(&tree); + if (add_object_entry(sha1, name_hash(NULL, ""), 1)) + add_pbase_tree(&tree, NULL); free(elem); } @@ -658,10 +718,23 @@ static void get_object_details(void) prepare_pack_ix(); for (i = 0, entry = objects; i < nr_objects; i++, entry++) check_object(entry); - for (i = 0, entry = objects; i < nr_objects; i++, entry++) - if (!entry->delta && entry->delta_child) - entry->delta_limit = - check_delta_limit(entry, 1); + + if (nr_objects == nr_result) { + /* + * Depth of objects that depend on the entry -- this + * is subtracted from depth-max to break too deep + * delta chain because of delta data reusing. + * However, we loosen this restriction when we know we + * are creating a thin pack -- it will have to be + * expanded on the other end anyway, so do not + * artificially cut the delta chain and let it go as + * deep as it wants. + */ + for (i = 0, entry = objects; i < nr_objects; i++, entry++) + if (!entry->delta && entry->delta_child) + entry->delta_limit = + check_delta_limit(entry, 1); + } } typedef int (*entry_sort_t)(const struct object_entry *, const struct object_entry *); @@ -692,7 +765,7 @@ static int sha1_sort(const struct object_entry *a, const struct object_entry *b) return memcmp(a->sha1, b->sha1, 20); } -static struct object_entry **create_final_object_list() +static struct object_entry **create_final_object_list(void) { struct object_entry **list; int i, j; @@ -748,8 +821,6 @@ static int try_delta(struct unpacked *cur, struct unpacked *old, unsigned max_de { struct object_entry *cur_entry = cur->entry; struct object_entry *old_entry = old->entry; - int old_preferred = (old_entry->preferred_base || - old_entry->based_on_preferred); unsigned long size, oldsize, delta_size, sizediff; long max_size; void *delta_buf; @@ -791,27 +862,8 @@ static int try_delta(struct unpacked *cur, struct unpacked *old, unsigned max_de * delete). */ max_size = size / 2 - 20; - if (cur_entry->delta) { - if (cur_entry->based_on_preferred) { - if (old_preferred) - max_size = cur_entry->delta_size-1; - else - /* trying with non-preferred one when we - * already have a delta based on preferred - * one is pointless. - */ - return 0; - } - else if (!old_preferred) - max_size = cur_entry->delta_size-1; - else - /* otherwise... even if delta with a - * preferred one produces a bigger result than - * what we currently have, which is based on a - * non-preferred one, it is OK. - */ - ; - } + if (cur_entry->delta) + max_size = cur_entry->delta_size-1; if (sizediff >= max_size) return -1; delta_buf = diff_delta(old->data, oldsize, @@ -821,7 +873,6 @@ static int try_delta(struct unpacked *cur, struct unpacked *old, unsigned max_de cur_entry->delta = old_entry; cur_entry->delta_size = delta_size; cur_entry->depth = old_entry->depth + 1; - cur_entry->based_on_preferred = old_preferred; free(delta_buf); return 0; } @@ -890,6 +941,15 @@ static void find_deltas(struct object_entry **list, int window, int depth) if (try_delta(n, m, depth) < 0) break; } +#if 0 + /* if we made n a delta, and if n is already at max + * depth, leaving it in the window is pointless. we + * should evict it first. + * ... in theory only; somehow this makes things worse. + */ + if (entry->delta && depth <= entry->depth) + continue; +#endif idx++; if (idx >= window) idx = 0; @@ -1051,7 +1111,7 @@ int main(int argc, char **argv) } if (get_sha1_hex(line, sha1)) die("expected sha1, got garbage:\n %s", line); - add_object_entry(sha1, line+40, 0); + add_object_entry(sha1, name_hash(NULL, line+41), 0); } if (progress) fprintf(stderr, "Done counting %d objects.\n", nr_objects);