Merge branch 'pe/cleanup'
[git/git.git] / pack-objects.c
index 32befa0..9346392 100644 (file)
@@ -1,9 +1,13 @@
 #include "cache.h"
 #include "object.h"
+#include "blob.h"
+#include "commit.h"
+#include "tag.h"
+#include "tree.h"
 #include "delta.h"
 #include "pack.h"
 #include "csum-file.h"
-#include "diff.h"
+#include "tree-walk.h"
 #include <sys/time.h>
 #include <signal.h>
 
@@ -32,9 +36,6 @@ 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.
-                                */
 };
 
 /*
@@ -61,7 +62,7 @@ static int nr_objects = 0, nr_alloc = 0, nr_result = 0;
 static const char *base_name;
 static unsigned char pack_file_sha1[20];
 static int progress = 1;
-static volatile int progress_update = 0;
+static volatile sig_atomic_t progress_update = 0;
 
 /*
  * The object names in objects array are hashed with this hashtable,
@@ -99,7 +100,7 @@ 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 +463,74 @@ 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<<DIRBITS) | (hash & ((1U<<DIRBITS )-1));
+
+       if (0) { /* debug */
+               n = name + strlen(name);
+               if (n != name && n[-1] == '\n')
+                       n--;
+               while (name <= --n)
+                       fputc(*n, stderr);
+               for (p = path; p; p = p->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 +550,6 @@ 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 +569,7 @@ 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 +584,15 @@ 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 +602,23 @@ static void add_pbase_tree(struct tree_desc *tree)
                        continue;
                if (sha1_object_info(sha1, type, &size))
                        continue;
-               add_object_entry(sha1, name, 1);
-               if (!strcmp(type, "tree")) {
+
+               hash = name_hash(up, name);
+               if (!add_object_entry(sha1, hash, 1))
+                       continue;
+
+               if (!strcmp(type, tree_type)) {
                        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 +629,13 @@ static void add_preferred_base(unsigned char *sha1)
 {
        struct tree_desc tree;
        void *elem;
-       elem = read_object_with_reference(sha1, "tree", &tree.size, NULL);
+
+       elem = read_object_with_reference(sha1, tree_type, &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);
 }
 
@@ -624,13 +688,13 @@ static void check_object(struct object_entry *entry)
                die("unable to get type of object %s",
                    sha1_to_hex(entry->sha1));
 
-       if (!strcmp(type, "commit")) {
+       if (!strcmp(type, commit_type)) {
                entry->type = OBJ_COMMIT;
-       } else if (!strcmp(type, "tree")) {
+       } else if (!strcmp(type, tree_type)) {
                entry->type = OBJ_TREE;
-       } else if (!strcmp(type, "blob")) {
+       } else if (!strcmp(type, blob_type)) {
                entry->type = OBJ_BLOB;
-       } else if (!strcmp(type, "tag")) {
+       } else if (!strcmp(type, tag_type)) {
                entry->type = OBJ_TAG;
        } else
                die("unable to pack object %s of type %s",
@@ -658,10 +722,23 @@ 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 +769,7 @@ 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 +825,6 @@ 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 +866,8 @@ 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,
@@ -821,14 +877,12 @@ 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;
 }
 
 static void progress_interval(int signum)
 {
-       signal(SIGALRM, progress_interval);
        progress_update = 1;
 }
 
@@ -890,6 +944,15 @@ 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;
@@ -965,6 +1028,23 @@ static int reuse_cached_pack(unsigned char *sha1, int pack_to_stdout)
        return 1;
 }
 
+static void setup_progress_signal(void)
+{
+       struct sigaction sa;
+       struct itimerval v;
+
+       memset(&sa, 0, sizeof(sa));
+       sa.sa_handler = progress_interval;
+       sigemptyset(&sa.sa_mask);
+       sa.sa_flags = SA_RESTART;
+       sigaction(SIGALRM, &sa, NULL);
+
+       v.it_interval.tv_sec = 1;
+       v.it_interval.tv_usec = 0;
+       v.it_value = v.it_interval;
+       setitimer(ITIMER_REAL, &v, NULL);
+}
+
 int main(int argc, char **argv)
 {
        SHA_CTX ctx;
@@ -1030,18 +1110,24 @@ int main(int argc, char **argv)
        prepare_packed_git();
 
        if (progress) {
-               struct itimerval v;
-               v.it_interval.tv_sec = 1;
-               v.it_interval.tv_usec = 0;
-               v.it_value = v.it_interval;
-               signal(SIGALRM, progress_interval);
-               setitimer(ITIMER_REAL, &v, NULL);
                fprintf(stderr, "Generating pack...\n");
+               setup_progress_signal();
        }
 
-       while (fgets(line, sizeof(line), stdin) != NULL) {
+       for (;;) {
                unsigned char sha1[20];
 
+               if (!fgets(line, sizeof(line), stdin)) {
+                       if (feof(stdin))
+                               break;
+                       if (!ferror(stdin))
+                               die("fgets returned NULL, not EOF, not error!");
+                       if (errno != EINTR)
+                               die("fgets: %s", strerror(errno));
+                       clearerr(stdin);
+                       continue;
+               }
+
                if (line[0] == '-') {
                        if (get_sha1_hex(line+1, sha1))
                                die("expected edge sha1, got garbage:\n %s",
@@ -1051,7 +1137,7 @@ 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);