fetch/push: readd rsync support
[gitweb.git] / builtin-pack-objects.c
index 5e9d1fd86ce5ee18dcb9ace9595b925f7630b6e8..a15906bdb2021e68a014344cad4e73e9de3367ca 100644 (file)
 #include "list-objects.h"
 #include "progress.h"
 
+#ifdef THREADED_DELTA_SEARCH
+#include <pthread.h>
+#endif
+
 static const char pack_usage[] = "\
 git-pack-objects [{ -q | --progress | --all-progress }] \n\
        [--max-pack-size=N] [--local] [--incremental] \n\
        [--window=N] [--window-memory=N] [--depth=N] \n\
        [--no-reuse-delta] [--no-reuse-object] [--delta-base-offset] \n\
-       [--non-empty] [--revs [--unpacked | --all]*] [--reflog] \n\
+       [--threads=N] [--non-empty] [--revs [--unpacked | --all]*] [--reflog] \n\
        [--stdout | base-name] [<ref-list | <object-list]";
 
 struct object_entry {
@@ -68,6 +72,7 @@ static int progress = 1;
 static int window = 10;
 static uint32_t pack_size_limit;
 static int depth = 50;
+static int delta_search_threads = 1;
 static int pack_to_stdout;
 static int num_preferred_base;
 static struct progress progress_state;
@@ -78,7 +83,6 @@ static unsigned long delta_cache_size = 0;
 static unsigned long max_delta_cache_size = 0;
 static unsigned long cache_max_small_delta_size = 1000;
 
-static unsigned long window_memory_usage = 0;
 static unsigned long window_memory_limit = 0;
 
 /*
@@ -586,7 +590,7 @@ static off_t write_one(struct sha1file *f,
 static int open_object_dir_tmp(const char *path)
 {
     snprintf(tmpname, sizeof(tmpname), "%s/%s", get_object_directory(), path);
-    return mkstemp(tmpname);
+    return xmkstemp(tmpname);
 }
 
 /* forward declaration for write_pack_file */
@@ -612,8 +616,6 @@ static void write_pack_file(void)
                        f = sha1fd(1, "<stdout>");
                } else {
                        int fd = open_object_dir_tmp("tmp_pack_XXXXXX");
-                       if (fd < 0)
-                               die("unable to create %s: %s\n", tmpname, strerror(errno));
                        pack_tmp_name = xstrdup(tmpname);
                        f = sha1fd(fd, pack_tmp_name);
                }
@@ -981,6 +983,8 @@ static void add_pbase_object(struct tree_desc *tree,
        int cmp;
 
        while (tree_entry(tree,&entry)) {
+               if (S_ISGITLINK(entry.mode))
+                       continue;
                cmp = tree_entry_len(entry.path, entry.sha1) != cmplen ? 1 :
                      memcmp(name, entry.path, cmplen);
                if (cmp > 0)
@@ -1275,9 +1279,8 @@ struct unpacked {
        unsigned depth;
 };
 
-static int delta_cacheable(struct unpacked *trg, struct unpacked *src,
-                           unsigned long src_size, unsigned long trg_size,
-                           unsigned long delta_size)
+static int delta_cacheable(unsigned long src_size, unsigned long trg_size,
+                          unsigned long delta_size)
 {
        if (max_delta_cache_size && delta_cache_size + delta_size > max_delta_cache_size)
                return 0;
@@ -1292,6 +1295,31 @@ static int delta_cacheable(struct unpacked *trg, struct unpacked *src,
        return 0;
 }
 
+#ifdef THREADED_DELTA_SEARCH
+
+static pthread_mutex_t read_mutex = PTHREAD_MUTEX_INITIALIZER;
+#define read_lock()            pthread_mutex_lock(&read_mutex)
+#define read_unlock()          pthread_mutex_unlock(&read_mutex)
+
+static pthread_mutex_t cache_mutex = PTHREAD_MUTEX_INITIALIZER;
+#define cache_lock()           pthread_mutex_lock(&cache_mutex)
+#define cache_unlock()         pthread_mutex_unlock(&cache_mutex)
+
+static pthread_mutex_t progress_mutex = PTHREAD_MUTEX_INITIALIZER;
+#define progress_lock()                pthread_mutex_lock(&progress_mutex)
+#define progress_unlock()      pthread_mutex_unlock(&progress_mutex)
+
+#else
+
+#define read_lock()            (void)0
+#define read_unlock()          (void)0
+#define cache_lock()           (void)0
+#define cache_unlock()         (void)0
+#define progress_lock()                (void)0
+#define progress_unlock()      (void)0
+
+#endif
+
 /*
  * We search for deltas _backwards_ in a list sorted by type and
  * by size, so that we see progressively smaller and smaller files.
@@ -1301,7 +1329,7 @@ static int delta_cacheable(struct unpacked *trg, struct unpacked *src,
  * one.
  */
 static int try_delta(struct unpacked *trg, struct unpacked *src,
-                    unsigned max_depth)
+                    unsigned max_depth, unsigned long *mem_usage)
 {
        struct object_entry *trg_entry = trg->entry;
        struct object_entry *src_entry = src->entry;
@@ -1314,12 +1342,6 @@ static int try_delta(struct unpacked *trg, struct unpacked *src,
        if (trg_entry->type != src_entry->type)
                return -1;
 
-       /* We do not compute delta to *create* objects we are not
-        * going to pack.
-        */
-       if (trg_entry->preferred_base)
-               return -1;
-
        /*
         * We do not bother to try a delta that we discarded
         * on an earlier try, but only when reusing delta data.
@@ -1356,18 +1378,28 @@ static int try_delta(struct unpacked *trg, struct unpacked *src,
 
        /* Load data if not already done */
        if (!trg->data) {
+               read_lock();
                trg->data = read_sha1_file(trg_entry->idx.sha1, &type, &sz);
+               read_unlock();
+               if (!trg->data)
+                       die("object %s cannot be read",
+                           sha1_to_hex(trg_entry->idx.sha1));
                if (sz != trg_size)
                        die("object %s inconsistent object length (%lu vs %lu)",
                            sha1_to_hex(trg_entry->idx.sha1), sz, trg_size);
-               window_memory_usage += sz;
+               *mem_usage += sz;
        }
        if (!src->data) {
+               read_lock();
                src->data = read_sha1_file(src_entry->idx.sha1, &type, &sz);
+               read_unlock();
+               if (!src->data)
+                       die("object %s cannot be read",
+                           sha1_to_hex(src_entry->idx.sha1));
                if (sz != src_size)
                        die("object %s inconsistent object length (%lu vs %lu)",
                            sha1_to_hex(src_entry->idx.sha1), sz, src_size);
-               window_memory_usage += sz;
+               *mem_usage += sz;
        }
        if (!src->index) {
                src->index = create_delta_index(src->data, src_size);
@@ -1377,33 +1409,47 @@ static int try_delta(struct unpacked *trg, struct unpacked *src,
                                warning("suboptimal pack - out of memory");
                        return 0;
                }
-               window_memory_usage += sizeof_delta_index(src->index);
+               *mem_usage += sizeof_delta_index(src->index);
        }
 
        delta_buf = create_delta(src->index, trg->data, trg_size, &delta_size, max_size);
        if (!delta_buf)
                return 0;
 
-       if (trg_entry->delta_data) {
+       if (trg_entry->delta) {
                /* Prefer only shallower same-sized deltas. */
                if (delta_size == trg_entry->delta_size &&
                    src->depth + 1 >= trg->depth) {
                        free(delta_buf);
                        return 0;
                }
-               delta_cache_size -= trg_entry->delta_size;
-               free(trg_entry->delta_data);
-               trg_entry->delta_data = NULL;
        }
+
        trg_entry->delta = src_entry;
        trg_entry->delta_size = delta_size;
        trg->depth = src->depth + 1;
 
-       if (delta_cacheable(src, trg, src_size, trg_size, delta_size)) {
-               trg_entry->delta_data = xrealloc(delta_buf, delta_size);
+       /*
+        * Handle memory allocation outside of the cache
+        * accounting lock.  Compiler will optimize the strangeness
+        * away when THREADED_DELTA_SEARCH is not defined.
+        */
+       if (trg_entry->delta_data)
+               free(trg_entry->delta_data);
+       cache_lock();
+       if (trg_entry->delta_data) {
+               delta_cache_size -= trg_entry->delta_size;
+               trg_entry->delta_data = NULL;
+       }
+       if (delta_cacheable(src_size, trg_size, delta_size)) {
                delta_cache_size += trg_entry->delta_size;
-       } else
+               cache_unlock();
+               trg_entry->delta_data = xrealloc(delta_buf, delta_size);
+       } else {
+               cache_unlock();
                free(delta_buf);
+       }
+
        return 1;
 }
 
@@ -1420,68 +1466,60 @@ static unsigned int check_delta_limit(struct object_entry *me, unsigned int n)
        return m;
 }
 
-static void free_unpacked(struct unpacked *n)
+static unsigned long free_unpacked(struct unpacked *n)
 {
-       window_memory_usage -= sizeof_delta_index(n->index);
+       unsigned long freed_mem = sizeof_delta_index(n->index);
        free_delta_index(n->index);
        n->index = NULL;
        if (n->data) {
+               freed_mem += n->entry->size;
                free(n->data);
                n->data = NULL;
-               window_memory_usage -= n->entry->size;
        }
        n->entry = NULL;
        n->depth = 0;
+       return freed_mem;
 }
 
-static void find_deltas(struct object_entry **list, int window, int depth)
+static void find_deltas(struct object_entry **list, unsigned list_size,
+                       int window, int depth, unsigned *processed)
 {
-       uint32_t i = nr_objects, idx = 0, count = 0, processed = 0;
+       uint32_t i = list_size, idx = 0, count = 0;
        unsigned int array_size = window * sizeof(struct unpacked);
        struct unpacked *array;
-       int max_depth;
+       unsigned long mem_usage = 0;
 
-       if (!nr_objects)
-               return;
        array = xmalloc(array_size);
        memset(array, 0, array_size);
-       if (progress)
-               start_progress(&progress_state, "Deltifying %u objects...", "", nr_result);
 
        do {
                struct object_entry *entry = list[--i];
                struct unpacked *n = array + idx;
-               int j;
-
-               if (!entry->preferred_base)
-                       processed++;
-
-               if (progress)
-                       display_progress(&progress_state, processed);
-
-               if (entry->delta)
-                       /* This happens if we decided to reuse existing
-                        * delta from a pack.  "!no_reuse_delta &&" is implied.
-                        */
-                       continue;
-
-               if (entry->size < 50)
-                       continue;
-
-               if (entry->no_try_delta)
-                       continue;
+               int j, max_depth, best_base = -1;
 
-               free_unpacked(n);
+               mem_usage -= free_unpacked(n);
                n->entry = entry;
 
                while (window_memory_limit &&
-                      window_memory_usage > window_memory_limit &&
+                      mem_usage > window_memory_limit &&
                       count > 1) {
                        uint32_t tail = (idx + window - count) % window;
-                       free_unpacked(array + tail);
+                       mem_usage -= free_unpacked(array + tail);
                        count--;
                }
 
+               /* We do not compute delta to *create* objects we are not
+                * going to pack.
+                */
+               if (entry->preferred_base)
+                       goto next;
+
+               progress_lock();
+               (*processed)++;
+               if (progress)
+                       display_progress(&progress_state, *processed);
+               progress_unlock();
+
                /*
                 * If the current object is at pack edge, take the depth the
                 * objects that depend on the current object into account
@@ -1496,6 +1534,7 @@ static void find_deltas(struct object_entry **list, int window, int depth)
 
                j = window;
                while (--j > 0) {
+                       int ret;
                        uint32_t other_idx = idx + j;
                        struct unpacked *m;
                        if (other_idx >= window)
@@ -1503,8 +1542,11 @@ static void find_deltas(struct object_entry **list, int window, int depth)
                        m = array + other_idx;
                        if (!m->entry)
                                break;
-                       if (try_delta(n, m, max_depth) < 0)
+                       ret = try_delta(n, m, max_depth, &mem_usage);
+                       if (ret < 0)
                                break;
+                       else if (ret > 0)
+                               best_base = other_idx;
                }
 
                /* if we made n a delta, and if n is already at max
@@ -1514,6 +1556,23 @@ static void find_deltas(struct object_entry **list, int window, int depth)
                if (entry->delta && depth <= n->depth)
                        continue;
 
+               /*
+                * Move the best delta base up in the window, after the
+                * currently deltified object, to keep it longer.  It will
+                * be the first base object to be attempted next.
+                */
+               if (entry->delta) {
+                       struct unpacked swap = array[best_base];
+                       int dist = (window + idx - best_base) % window;
+                       int dst = best_base;
+                       while (dist--) {
+                               int src = (dst + 1) % window;
+                               array[dst] = array[src];
+                               dst = src;
+                       }
+                       array[dst] = swap;
+               }
+
                next:
                idx++;
                if (count + 1 < window)
@@ -1522,9 +1581,6 @@ static void find_deltas(struct object_entry **list, int window, int depth)
                        idx = 0;
        } while (i > 0);
 
-       if (progress)
-               stop_progress(&progress_state);
-
        for (i = 0; i < window; ++i) {
                free_delta_index(array[i].index);
                free(array[i].data);
@@ -1532,21 +1588,145 @@ static void find_deltas(struct object_entry **list, int window, int depth)
        free(array);
 }
 
+#ifdef THREADED_DELTA_SEARCH
+
+struct thread_params {
+       pthread_t thread;
+       struct object_entry **list;
+       unsigned list_size;
+       int window;
+       int depth;
+       unsigned *processed;
+};
+
+static pthread_mutex_t data_request  = PTHREAD_MUTEX_INITIALIZER;
+static pthread_mutex_t data_ready    = PTHREAD_MUTEX_INITIALIZER;
+static pthread_mutex_t data_provider = PTHREAD_MUTEX_INITIALIZER;
+static struct thread_params *data_requester;
+
+static void *threaded_find_deltas(void *arg)
+{
+       struct thread_params *me = arg;
+
+       for (;;) {
+               pthread_mutex_lock(&data_request);
+               data_requester = me;
+               pthread_mutex_unlock(&data_provider);
+               pthread_mutex_lock(&data_ready);
+               pthread_mutex_unlock(&data_request);
+
+               if (!me->list_size)
+                       return NULL;
+
+               find_deltas(me->list, me->list_size,
+                           me->window, me->depth, me->processed);
+       }
+}
+
+static void ll_find_deltas(struct object_entry **list, unsigned list_size,
+                          int window, int depth, unsigned *processed)
+{
+       struct thread_params *target, p[delta_search_threads];
+       int i, ret;
+       unsigned chunk_size;
+
+       if (delta_search_threads <= 1) {
+               find_deltas(list, list_size, window, depth, processed);
+               return;
+       }
+
+       pthread_mutex_lock(&data_provider);
+       pthread_mutex_lock(&data_ready);
+
+       for (i = 0; i < delta_search_threads; i++) {
+               p[i].window = window;
+               p[i].depth = depth;
+               p[i].processed = processed;
+               ret = pthread_create(&p[i].thread, NULL,
+                                    threaded_find_deltas, &p[i]);
+               if (ret)
+                       die("unable to create thread: %s", strerror(ret));
+       }
+
+       /* this should be auto-tuned somehow */
+       chunk_size = window * 1000;
+
+       do {
+               unsigned sublist_size = chunk_size;
+               if (sublist_size > list_size)
+                       sublist_size = list_size;
+
+               /* try to split chunks on "path" boundaries */
+               while (sublist_size < list_size && list[sublist_size]->hash &&
+                      list[sublist_size]->hash == list[sublist_size-1]->hash)
+                       sublist_size++;
+
+               pthread_mutex_lock(&data_provider);
+               target = data_requester;
+               target->list = list;
+               target->list_size = sublist_size;
+               pthread_mutex_unlock(&data_ready);
+
+               list += sublist_size;
+               list_size -= sublist_size;
+               if (!sublist_size) {
+                       pthread_join(target->thread, NULL);
+                       i--;
+               }
+       } while (i);
+}
+
+#else
+#define ll_find_deltas find_deltas
+#endif
+
 static void prepare_pack(int window, int depth)
 {
        struct object_entry **delta_list;
-       uint32_t i;
+       uint32_t i, n, nr_deltas;
 
        get_object_details();
 
-       if (!window || !depth)
+       if (!nr_objects || !window || !depth)
                return;
 
        delta_list = xmalloc(nr_objects * sizeof(*delta_list));
-       for (i = 0; i < nr_objects; i++)
-               delta_list[i] = objects + i;
-       qsort(delta_list, nr_objects, sizeof(*delta_list), type_size_sort);
-       find_deltas(delta_list, window+1, depth);
+       nr_deltas = n = 0;
+
+       for (i = 0; i < nr_objects; i++) {
+               struct object_entry *entry = objects + i;
+
+               if (entry->delta)
+                       /* This happens if we decided to reuse existing
+                        * delta from a pack.  "!no_reuse_delta &&" is implied.
+                        */
+                       continue;
+
+               if (entry->size < 50)
+                       continue;
+
+               if (entry->no_try_delta)
+                       continue;
+
+               if (!entry->preferred_base)
+                       nr_deltas++;
+
+               delta_list[n++] = entry;
+       }
+
+       if (nr_deltas) {
+               unsigned nr_done = 0;
+               if (progress)
+                       start_progress(&progress_state,
+                                      "Deltifying %u objects...", "",
+                                      nr_deltas);
+               qsort(delta_list, n, sizeof(*delta_list), type_size_sort);
+               ll_find_deltas(delta_list, n, window+1, depth, &nr_done);
+               if (progress)
+                       stop_progress(&progress_state);
+               if (nr_done != nr_deltas)
+                       die("inconsistency with delta count");
+       }
        free(delta_list);
 }
 
@@ -1582,6 +1762,17 @@ static int git_pack_config(const char *k, const char *v)
                cache_max_small_delta_size = git_config_int(k, v);
                return 0;
        }
+       if (!strcmp(k, "pack.threads")) {
+               delta_search_threads = git_config_int(k, v);
+               if (delta_search_threads < 1)
+                       die("invalid number of threads specified (%d)",
+                           delta_search_threads);
+#ifndef THREADED_DELTA_SEARCH
+               if (delta_search_threads > 1)
+                       warning("no threads support, ignoring %s", k);
+#endif
+               return 0;
+       }
        return git_default_config(k, v);
 }
 
@@ -1741,6 +1932,18 @@ int cmd_pack_objects(int argc, const char **argv, const char *prefix)
                                usage(pack_usage);
                        continue;
                }
+               if (!prefixcmp(arg, "--threads=")) {
+                       char *end;
+                       delta_search_threads = strtoul(arg+10, &end, 0);
+                       if (!arg[10] || *end || delta_search_threads < 1)
+                               usage(pack_usage);
+#ifndef THREADED_DELTA_SEARCH
+                       if (delta_search_threads > 1)
+                               warning("no threads support, "
+                                       "ignoring %s", arg);
+#endif
+                       continue;
+               }
                if (!prefixcmp(arg, "--depth=")) {
                        char *end;
                        depth = strtoul(arg+8, &end, 0);