Merge branch 'np/verify-pack' into maint
[gitweb.git] / pack-redundant.c
index 0a43278924edae9d6e3ecdf4bb100bea10faef97..25b81a445c8fafe0c00ce30082b7d9a7c22ccf1e 100644 (file)
@@ -8,14 +8,16 @@
 
 #include "cache.h"
 
+#define BLKSIZE 512
+
 static const char pack_redundant_usage[] =
-"git-pack-redundant [ --verbose ] [ --alt-odb ] < --all | <.pack filename> ...>";
+"git pack-redundant [ --verbose ] [ --alt-odb ] < --all | <.pack filename> ...>";
 
-static int load_all_packs = 0, verbose = 0, alt_odb = 0;
+static int load_all_packs, verbose, alt_odb;
 
 struct llist_item {
        struct llist_item *next;
-       unsigned char *sha1;
+       const unsigned char *sha1;
 };
 static struct llist {
        struct llist_item *front;
@@ -33,29 +35,32 @@ static struct pack_list {
 struct pll {
        struct pll *next;
        struct pack_list *pl;
-       size_t pl_size;
 };
 
-static struct llist_item *free_nodes = NULL;
+static struct llist_item *free_nodes;
+
+static inline void llist_item_put(struct llist_item *item)
+{
+       item->next = free_nodes;
+       free_nodes = item;
+}
 
-static inline struct llist_item *llist_item_get()
+static inline struct llist_item *llist_item_get(void)
 {
        struct llist_item *new;
        if ( free_nodes ) {
                new = free_nodes;
                free_nodes = free_nodes->next;
-       } else
-               new = xmalloc(sizeof(struct llist_item));
-
+       } else {
+               int i = 1;
+               new = xmalloc(sizeof(struct llist_item) * BLKSIZE);
+               for(;i < BLKSIZE; i++) {
+                       llist_item_put(&new[i]);
+               }
+       }
        return new;
 }
 
-static inline void llist_item_put(struct llist_item *item)
-{
-       item->next = free_nodes;
-       free_nodes = item;
-}
-
 static void llist_free(struct llist *list)
 {
        while((list->back = list->front)) {
@@ -76,7 +81,7 @@ static struct llist * llist_copy(struct llist *list)
 {
        struct llist *ret;
        struct llist_item *new, *old, *prev;
-       
+
        llist_init(&ret);
 
        if ((ret->size = list->size) == 0)
@@ -95,13 +100,13 @@ static struct llist * llist_copy(struct llist *list)
        }
        new->next = NULL;
        ret->back = new;
-       
+
        return ret;
 }
 
-static inline struct llist_item * llist_insert(struct llist *list,
-                                              struct llist_item *after,
-                                              unsigned char *sha1)
+static inline struct llist_item *llist_insert(struct llist *list,
+                                             struct llist_item *after,
+                                              const unsigned char *sha1)
 {
        struct llist_item *new = llist_item_get();
        new->sha1 = sha1;
@@ -123,18 +128,20 @@ static inline struct llist_item * llist_insert(struct llist *list,
        return new;
 }
 
-static inline struct llist_item *llist_insert_back(struct llist *list, unsigned char *sha1)
+static inline struct llist_item *llist_insert_back(struct llist *list,
+                                                  const unsigned char *sha1)
 {
        return llist_insert(list, list->back, sha1);
 }
 
-static inline struct llist_item *llist_insert_sorted_unique(struct llist *list, unsigned char *sha1, struct llist_item *hint)
+static inline struct llist_item *llist_insert_sorted_unique(struct llist *list,
+                       const unsigned char *sha1, struct llist_item *hint)
 {
        struct llist_item *prev = NULL, *l;
 
        l = (hint == NULL) ? list->front : hint;
        while (l) {
-               int cmp = memcmp(l->sha1, sha1, 20);
+               int cmp = hashcmp(l->sha1, sha1);
                if (cmp > 0) { /* we insert before this entry */
                        return llist_insert(list, prev, sha1);
                }
@@ -157,7 +164,7 @@ static inline struct llist_item * llist_sorted_remove(struct llist *list, const
        l = (hint == NULL) ? list->front : hint;
        prev = NULL;
        while (l) {
-               int cmp = memcmp(l->sha1, sha1, 20);
+               int cmp = hashcmp(l->sha1, sha1);
                if (cmp > 0) /* not in list, since sorted */
                        return prev;
                if(!cmp) { /* found */
@@ -240,107 +247,91 @@ static struct pack_list * pack_list_difference(const struct pack_list *A,
 
 static void cmp_two_packs(struct pack_list *p1, struct pack_list *p2)
 {
-       int p1_off, p2_off;
-       void *p1_base, *p2_base;
+       unsigned long p1_off = 0, p2_off = 0, p1_step, p2_step;
+       const unsigned char *p1_base, *p2_base;
        struct llist_item *p1_hint = NULL, *p2_hint = NULL;
-       
-       p1_off = p2_off = 256 * 4 + 4;
-       p1_base = (void *)p1->pack->index_base;
-       p2_base = (void *)p2->pack->index_base;
 
-       while (p1_off <= p1->pack->index_size - 3 * 20 &&
-              p2_off <= p2->pack->index_size - 3 * 20)
+       p1_base = p1->pack->index_data;
+       p2_base = p2->pack->index_data;
+       p1_base += 256 * 4 + ((p1->pack->index_version < 2) ? 4 : 8);
+       p2_base += 256 * 4 + ((p2->pack->index_version < 2) ? 4 : 8);
+       p1_step = (p1->pack->index_version < 2) ? 24 : 20;
+       p2_step = (p2->pack->index_version < 2) ? 24 : 20;
+
+       while (p1_off < p1->pack->num_objects * p1_step &&
+              p2_off < p2->pack->num_objects * p2_step)
        {
-               int cmp = memcmp(p1_base + p1_off, p2_base + p2_off, 20);
+               int cmp = hashcmp(p1_base + p1_off, p2_base + p2_off);
                /* cmp ~ p1 - p2 */
                if (cmp == 0) {
                        p1_hint = llist_sorted_remove(p1->unique_objects,
                                        p1_base + p1_off, p1_hint);
                        p2_hint = llist_sorted_remove(p2->unique_objects,
                                        p1_base + p1_off, p2_hint);
-                       p1_off+=24;
-                       p2_off+=24;
+                       p1_off += p1_step;
+                       p2_off += p2_step;
                        continue;
                }
                if (cmp < 0) { /* p1 has the object, p2 doesn't */
-                       p1_off+=24;
+                       p1_off += p1_step;
                } else { /* p2 has the object, p1 doesn't */
-                       p2_off+=24;
+                       p2_off += p2_step;
                }
        }
 }
 
-static void pll_insert(struct pll **pll, struct pll **hint_table)
+static void pll_free(struct pll *l)
 {
-       struct pll *prev;
-       int i = (*pll)->pl_size - 1;
-
-       if (hint_table[i] == NULL) {
-               hint_table[i--] = *pll;
-               for (; i >= 0; --i) {
-                       if (hint_table[i] != NULL)
-                               break;
+       struct pll *old;
+       struct pack_list *opl;
+
+       while (l) {
+               old = l;
+               while (l->pl) {
+                       opl = l->pl;
+                       l->pl = opl->next;
+                       free(opl);
                }
-               if (hint_table[i] == NULL) /* no elements in list */
-                       die("Why did this happen?");
+               l = l->next;
+               free(old);
        }
-
-       prev = hint_table[i];
-       while (prev->next && prev->next->pl_size < (*pll)->pl_size)
-               prev = prev->next;
-
-       (*pll)->next = prev->next;
-       prev->next = *pll;
 }
 
 /* all the permutations have to be free()d at the same time,
  * since they refer to each other
  */
-static struct pll * get_all_permutations(struct pack_list *list)
+static struct pll * get_permutations(struct pack_list *list, int n)
 {
-       struct pll *subset, *pll, *new_pll = NULL; /*silence warning*/
-       static struct pll **hint = NULL;
-       if (hint == NULL)
-               hint = xcalloc(pack_list_size(list), sizeof(struct pll *));
-               
-       if (list == NULL)
+       struct pll *subset, *ret = NULL, *new_pll = NULL, *pll;
+
+       if (list == NULL || pack_list_size(list) < n || n == 0)
                return NULL;
 
-       if (list->next == NULL) {
-               new_pll = xmalloc(sizeof(struct pll));
-               hint[0] = new_pll;
-               new_pll->next = NULL;
-               new_pll->pl = list;
-               new_pll->pl_size = 1;
-               return new_pll;
+       if (n == 1) {
+               while (list) {
+                       new_pll = xmalloc(sizeof(pll));
+                       new_pll->pl = NULL;
+                       pack_list_insert(&new_pll->pl, list);
+                       new_pll->next = ret;
+                       ret = new_pll;
+                       list = list->next;
+               }
+               return ret;
        }
 
-       pll = subset = get_all_permutations(list->next);
-       while (pll) {
-               if (pll->pl->pack == list->pack) {
-                       pll = pll->next;
-                       continue;
+       while (list->next) {
+               subset = get_permutations(list->next, n - 1);
+               while (subset) {
+                       new_pll = xmalloc(sizeof(pll));
+                       new_pll->pl = subset->pl;
+                       pack_list_insert(&new_pll->pl, list);
+                       new_pll->next = ret;
+                       ret = new_pll;
+                       subset = subset->next;
                }
-               new_pll = xmalloc(sizeof(struct pll));
-
-               new_pll->pl = xmalloc(sizeof(struct pack_list));
-               memcpy(new_pll->pl, list, sizeof(struct pack_list));
-               new_pll->pl->next = pll->pl;
-               new_pll->pl_size = pll->pl_size + 1;
-               
-               pll_insert(&new_pll, hint);
-
-               pll = pll->next;
-       }
-       /* add ourself */
-       new_pll = xmalloc(sizeof(struct pll));
-       new_pll->pl = xmalloc(sizeof(struct pack_list));
-       memcpy(new_pll->pl, list, sizeof(struct pack_list));
-       new_pll->pl->next = NULL;
-       new_pll->pl_size = 1;
-       pll_insert(&new_pll, hint);
-
-       return hint[0];
+               list = list->next;
+       }
+       return ret;
 }
 
 static int is_superset(struct pack_list *pl, struct llist *list)
@@ -364,28 +355,31 @@ static int is_superset(struct pack_list *pl, struct llist *list)
 static size_t sizeof_union(struct packed_git *p1, struct packed_git *p2)
 {
        size_t ret = 0;
-       int p1_off, p2_off;
-       void *p1_base, *p2_base;
-
-       p1_off = p2_off = 256 * 4 + 4;
-       p1_base = (void *)p1->index_base;
-       p2_base = (void *)p2->index_base;
-
-       while (p1_off <= p1->index_size - 3 * 20 &&
-              p2_off <= p2->index_size - 3 * 20)
+       unsigned long p1_off = 0, p2_off = 0, p1_step, p2_step;
+       const unsigned char *p1_base, *p2_base;
+
+       p1_base = p1->index_data;
+       p2_base = p2->index_data;
+       p1_base += 256 * 4 + ((p1->index_version < 2) ? 4 : 8);
+       p2_base += 256 * 4 + ((p2->index_version < 2) ? 4 : 8);
+       p1_step = (p1->index_version < 2) ? 24 : 20;
+       p2_step = (p2->index_version < 2) ? 24 : 20;
+
+       while (p1_off < p1->num_objects * p1_step &&
+              p2_off < p2->num_objects * p2_step)
        {
-               int cmp = memcmp(p1_base + p1_off, p2_base + p2_off, 20);
+               int cmp = hashcmp(p1_base + p1_off, p2_base + p2_off);
                /* cmp ~ p1 - p2 */
                if (cmp == 0) {
                        ret++;
-                       p1_off+=24;
-                       p2_off+=24;
+                       p1_off += p1_step;
+                       p2_off += p2_step;
                        continue;
                }
                if (cmp < 0) { /* p1 has the object, p2 doesn't */
-                       p1_off+=24;
+                       p1_off += p1_step;
                } else { /* p2 has the object, p1 doesn't */
-                       p2_off+=24;
+                       p2_off += p2_step;
                }
        }
        return ret;
@@ -410,9 +404,9 @@ static size_t get_pack_redundancy(struct pack_list *pl)
        return ret;
 }
 
-static inline size_t pack_set_bytecount(struct pack_list *pl)
+static inline off_t pack_set_bytecount(struct pack_list *pl)
 {
-       size_t ret = 0;
+       off_t ret = 0;
        while (pl) {
                ret += pl->pack->pack_size;
                ret += pl->pack->index_size;
@@ -427,7 +421,8 @@ static void minimize(struct pack_list **min)
                *non_unique = NULL, *min_perm = NULL;
        struct pll *perm, *perm_all, *perm_ok = NULL, *new_perm;
        struct llist *missing;
-       size_t min_perm_size = (size_t)-1, perm_size;
+       off_t min_perm_size = 0, perm_size;
+       int n;
 
        pl = local_packs;
        while (pl) {
@@ -441,8 +436,7 @@ static void minimize(struct pack_list **min)
        missing = llist_copy(all_objects);
        pl = unique;
        while (pl) {
-               llist_sorted_difference_inplace(missing,
-                                               pl->all_objects);
+               llist_sorted_difference_inplace(missing, pl->all_objects);
                pl = pl->next;
        }
 
@@ -453,19 +447,21 @@ static void minimize(struct pack_list **min)
        }
 
        /* find the permutations which contain all missing objects */
-       perm_all = perm = get_all_permutations(non_unique);
-       while (perm) {
-               if (perm_ok && perm->pl_size > perm_ok->pl_size)
-                       break; /* ignore all larger permutations */
-               if (is_superset(perm->pl, missing)) {
-                       new_perm = xmalloc(sizeof(struct pll));
-                       memcpy(new_perm, perm, sizeof(struct pll));
-                       new_perm->next = perm_ok;
-                       perm_ok = new_perm;
+       for (n = 1; n <= pack_list_size(non_unique) && !perm_ok; n++) {
+               perm_all = perm = get_permutations(non_unique, n);
+               while (perm) {
+                       if (is_superset(perm->pl, missing)) {
+                               new_perm = xmalloc(sizeof(struct pll));
+                               memcpy(new_perm, perm, sizeof(struct pll));
+                               new_perm->next = perm_ok;
+                               perm_ok = new_perm;
+                       }
+                       perm = perm->next;
                }
-               perm = perm->next;
+               if (perm_ok)
+                       break;
+               pll_free(perm_all);
        }
-       
        if (perm_ok == NULL)
                die("Internal error: No complete sets found!\n");
 
@@ -473,7 +469,7 @@ static void minimize(struct pack_list **min)
        perm = perm_ok;
        while (perm) {
                perm_size = pack_set_bytecount(perm->pl);
-               if (min_perm_size > perm_size) {
+               if (!min_perm_size || min_perm_size > perm_size) {
                        min_perm_size = perm_size;
                        min_perm = perm->pl;
                }
@@ -537,6 +533,7 @@ static void scan_alt_odb_packs(void)
                                                        alt->all_objects);
                        local = local->next;
                }
+               llist_sorted_difference_inplace(all_objects, alt->all_objects);
                alt = alt->next;
        }
 }
@@ -544,8 +541,8 @@ static void scan_alt_odb_packs(void)
 static struct pack_list * add_pack(struct packed_git *p)
 {
        struct pack_list l;
-       size_t off;
-       void *base;
+       unsigned long off = 0, step;
+       const unsigned char *base;
 
        if (!p->pack_local && !(alt_odb || verbose))
                return NULL;
@@ -553,11 +550,15 @@ static struct pack_list * add_pack(struct packed_git *p)
        l.pack = p;
        llist_init(&l.all_objects);
 
-       off = 256 * 4 + 4;
-       base = (void *)p->index_base;
-       while (off <= p->index_size - 3 * 20) {
+       if (open_pack_index(p))
+               return NULL;
+
+       base = p->index_data;
+       base += 256 * 4 + ((p->index_version < 2) ? 4 : 8);
+       step = (p->index_version < 2) ? 24 : 20;
+       while (off < p->num_objects * step) {
                llist_insert_back(l.all_objects, base + off);
-               off += 24;
+               off += step;
        }
        /* this list will be pruned in cmp_two_packs later */
        l.unique_objects = llist_copy(l.all_objects);