[PATCH] fix scalability problems with git-deltafy-script
[gitweb.git] / fsck-cache.c
index a00702b79f7169ef3da7d60e9bbbe9ccf237e673..d69c426b360db21820a4e180202ae4d71e0ba965 100644 (file)
@@ -6,15 +6,46 @@
 #include "tree.h"
 #include "blob.h"
 #include "tag.h"
+#include "delta.h"
 
 #define REACHABLE 0x0001
 
 static int show_root = 0;
 static int show_tags = 0;
 static int show_unreachable = 0;
+static int show_max_delta_depth = 0;
 static int keep_cache_objects = 0; 
 static unsigned char head_sha1[20];
 
+static void expand_deltas(void)
+{
+       int i, max_depth = 0;
+
+       /*
+        * To be as efficient as possible we look for delta heads and
+        * recursively process them going backward, and parsing
+        * resulting objects along the way.  This allows for processing
+        * each delta objects only once regardless of the delta depth.
+        */
+       for (i = 0; i < nr_objs; i++) {
+               struct object *obj = objs[i];
+               if (obj->parsed && !obj->delta && obj->attached_deltas) {
+                       int depth = 0;
+                       char type[10];
+                       unsigned long size;
+                       void *buf = read_sha1_file(obj->sha1, type, &size);
+                       if (!buf)
+                               continue;
+                       depth = process_deltas(buf, size, obj->type,
+                                              obj->attached_deltas);
+                       if (max_depth < depth)
+                               max_depth = depth;
+               }
+       }
+       if (show_max_delta_depth)
+               printf("maximum delta depth = %d\n", max_depth);
+}
+                                                                                                                       
 static void check_connectivity(void)
 {
        int i;
@@ -25,7 +56,12 @@ static void check_connectivity(void)
                struct object_list *refs;
 
                if (!obj->parsed) {
-                       printf("missing %s %s\n", obj->type, sha1_to_hex(obj->sha1));
+                       if (obj->delta)
+                               printf("unresolved delta %s\n",
+                                      sha1_to_hex(obj->sha1));
+                       else
+                               printf("missing %s %s\n",
+                                      obj->type, sha1_to_hex(obj->sha1));
                        continue;
                }
 
@@ -43,7 +79,12 @@ static void check_connectivity(void)
                        continue;
 
                if (show_unreachable && !(obj->flags & REACHABLE)) {
-                       printf("unreachable %s %s\n", obj->type, sha1_to_hex(obj->sha1));
+                       if (obj->attached_deltas)
+                               printf("foreign delta reference %s\n", 
+                                      sha1_to_hex(obj->sha1));
+                       else
+                               printf("unreachable %s %s\n",
+                                      obj->type, sha1_to_hex(obj->sha1));
                        continue;
                }
 
@@ -152,7 +193,7 @@ static int fsck_tree(struct tree *item)
        }
 
        if (has_full_path) {
-               fprintf(stderr, "warning: fsck-cache: tree %s "
+               fprintf(stderr, "warning: git-fsck-cache: tree %s "
                        "has full pathnames in it\n", 
                        sha1_to_hex(item->object.sha1));
        }
@@ -162,6 +203,8 @@ static int fsck_tree(struct tree *item)
 
 static int fsck_commit(struct commit *commit)
 {
+       free(commit->buffer);
+       commit->buffer = NULL;
        if (!commit->tree)
                return -1;
        if (!commit->parents && show_root)
@@ -201,6 +244,8 @@ static int fsck_sha1(unsigned char *sha1)
                return fsck_commit((struct commit *) obj);
        if (obj->type == tag_type)
                return fsck_tag((struct tag *) obj);
+       if (!obj->type && obj->delta)
+               return 0;
        return -1;
 }
 
@@ -296,6 +341,76 @@ static int fsck_dir(int i, char *path)
        return 0;
 }
 
+static int read_sha1_reference(const char *path)
+{
+       char hexname[60];
+       unsigned char sha1[20];
+       int fd = open(path, O_RDONLY), len;
+       struct object *obj;
+
+       if (fd < 0)
+               return -1;
+
+       len = read(fd, hexname, sizeof(hexname));
+       close(fd);
+       if (len < 40)
+               return -1;
+
+       if (get_sha1_hex(hexname, sha1) < 0)
+               return -1;
+
+       obj = lookup_object(sha1);
+       if (!obj)
+               return error("%s: invalid sha1 pointer %.40s", path, hexname);
+
+       obj->used = 1;
+       mark_reachable(obj, REACHABLE);
+       return 0;
+}
+
+static int find_file_objects(const char *base, const char *name)
+{
+       int baselen = strlen(base);
+       int namelen = strlen(name);
+       char *path = xmalloc(baselen + namelen + 2);
+       struct stat st;
+
+       memcpy(path, base, baselen);
+       path[baselen] = '/';
+       memcpy(path + baselen + 1, name, namelen+1);
+       if (stat(path, &st) < 0)
+               return 0;
+
+       /*
+        * Recurse into directories
+        */
+       if (S_ISDIR(st.st_mode)) {
+               int count = 0;
+               DIR *dir = opendir(path);
+               if (dir) {
+                       struct dirent *de;
+                       while ((de = readdir(dir)) != NULL) {
+                               if (de->d_name[0] == '.')
+                                       continue;
+                               count += find_file_objects(path, de->d_name);
+                       }
+                       closedir(dir);
+               }
+               return count;
+       }
+       if (S_ISREG(st.st_mode))
+               return read_sha1_reference(path) == 0;
+       return 0;
+}
+
+static void get_default_heads(void)
+{
+       char *git_dir = gitenv(GIT_DIR_ENVIRONMENT) ? : DEFAULT_GIT_DIR_ENVIRONMENT;
+       int count = find_file_objects(git_dir, "refs");
+       if (!count)
+               die("No default references");
+}
+
 int main(int argc, char **argv)
 {
        int i, heads;
@@ -316,12 +431,16 @@ int main(int argc, char **argv)
                        show_root = 1;
                        continue;
                }
+               if (!strcmp(arg, "--delta-depth")) {
+                       show_max_delta_depth = 1;
+                       continue;
+               }
                if (!strcmp(arg, "--cache")) {
                        keep_cache_objects = 1;
                        continue;
                }
                if (*arg == '-')
-                       usage("fsck-cache [--tags] [[--unreachable] [--cache] <head-sha1>*]");
+                       usage("git-fsck-cache [--tags] [[--unreachable] [--cache] <head-sha1>*]");
        }
 
        sha1_dir = get_object_directory();
@@ -332,6 +451,8 @@ int main(int argc, char **argv)
        }
        fsck_sha1_list();
 
+       expand_deltas();
+
        heads = 0;
        for (i = 1; i < argc; i++) {
                const char *arg = argv[i]; 
@@ -354,6 +475,16 @@ int main(int argc, char **argv)
                error("expected sha1, got %s", arg);
        }
 
+       /*
+        * If we've not been given any explicit head information, do the
+        * default ones from .git/refs. We also consider the index file
+        * in this case (ie this implies --cache).
+        */
+       if (!heads) {
+               get_default_heads();
+               keep_cache_objects = 1;
+       }
+
        if (keep_cache_objects) {
                int i;
                read_cache();
@@ -368,15 +499,6 @@ int main(int argc, char **argv)
                }
        }
 
-       if (!heads && !keep_cache_objects) {
-               if (show_unreachable) {
-                       fprintf(stderr, "unable to do reachability without a head nor --cache\n");
-                       show_unreachable = 0; 
-               }
-               if (!heads)
-                       fprintf(stderr, "expect dangling commits - potential heads - due to lack of head information\n");
-       }
-
        check_connectivity();
        return 0;
 }