git/delta-islands.c

555 lines
13 KiB
C
Raw Normal View History

#include "git-compat-util.h"
2018-08-16 06:13:07 +00:00
#include "object.h"
#include "commit.h"
#include "gettext.h"
#include "hex.h"
2018-08-16 06:13:07 +00:00
#include "tag.h"
#include "tree.h"
#include "pack.h"
#include "tree-walk.h"
#include "diff.h"
#include "progress.h"
#include "refs.h"
#include "khash.h"
#include "pack-bitmap.h"
#include "pack-objects.h"
#include "delta-islands.h"
#include "oid-array.h"
2018-08-16 06:13:07 +00:00
#include "config.h"
KHASH_INIT(str, const char *, void *, 1, kh_str_hash_func, kh_str_hash_equal)
static kh_oid_map_t *island_marks;
2018-08-16 06:13:07 +00:00
static unsigned island_counter;
static unsigned island_counter_core;
struct remote_island {
uint64_t hash;
struct oid_array oids;
};
struct island_bitmap {
uint32_t refcount;
uint32_t bits[FLEX_ARRAY];
};
static uint32_t island_bitmap_size;
/*
* Allocate a new bitmap; if "old" is not NULL, the new bitmap will be a copy
* of "old". Otherwise, the new bitmap is empty.
*/
static struct island_bitmap *island_bitmap_new(const struct island_bitmap *old)
{
size_t size = sizeof(struct island_bitmap) + (island_bitmap_size * 4);
struct island_bitmap *b = xcalloc(1, size);
if (old)
memcpy(b, old, size);
b->refcount = 1;
return b;
}
static void island_bitmap_or(struct island_bitmap *a, const struct island_bitmap *b)
{
uint32_t i;
for (i = 0; i < island_bitmap_size; ++i)
a->bits[i] |= b->bits[i];
}
static int island_bitmap_is_subset(struct island_bitmap *self,
struct island_bitmap *super)
{
uint32_t i;
if (self == super)
return 1;
for (i = 0; i < island_bitmap_size; ++i) {
if ((self->bits[i] & super->bits[i]) != self->bits[i])
return 0;
}
return 1;
}
#define ISLAND_BITMAP_BLOCK(x) (x / 32)
#define ISLAND_BITMAP_MASK(x) (1 << (x % 32))
static void island_bitmap_set(struct island_bitmap *self, uint32_t i)
{
self->bits[ISLAND_BITMAP_BLOCK(i)] |= ISLAND_BITMAP_MASK(i);
}
static int island_bitmap_get(struct island_bitmap *self, uint32_t i)
{
return (self->bits[ISLAND_BITMAP_BLOCK(i)] & ISLAND_BITMAP_MASK(i)) != 0;
}
int in_same_island(const struct object_id *trg_oid, const struct object_id *src_oid)
{
khiter_t trg_pos, src_pos;
/* If we aren't using islands, assume everything goes together. */
if (!island_marks)
return 1;
/*
* If we don't have a bitmap for the target, we can delta it
* against anything -- it's not an important object
*/
trg_pos = kh_get_oid_map(island_marks, *trg_oid);
2018-08-16 06:13:07 +00:00
if (trg_pos >= kh_end(island_marks))
return 1;
/*
* if the source (our delta base) doesn't have a bitmap,
* we don't want to base any deltas on it!
*/
src_pos = kh_get_oid_map(island_marks, *src_oid);
2018-08-16 06:13:07 +00:00
if (src_pos >= kh_end(island_marks))
return 0;
return island_bitmap_is_subset(kh_value(island_marks, trg_pos),
kh_value(island_marks, src_pos));
}
int island_delta_cmp(const struct object_id *a, const struct object_id *b)
{
khiter_t a_pos, b_pos;
struct island_bitmap *a_bitmap = NULL, *b_bitmap = NULL;
if (!island_marks)
return 0;
a_pos = kh_get_oid_map(island_marks, *a);
2018-08-16 06:13:07 +00:00
if (a_pos < kh_end(island_marks))
a_bitmap = kh_value(island_marks, a_pos);
b_pos = kh_get_oid_map(island_marks, *b);
2018-08-16 06:13:07 +00:00
if (b_pos < kh_end(island_marks))
b_bitmap = kh_value(island_marks, b_pos);
if (a_bitmap) {
if (!b_bitmap || !island_bitmap_is_subset(a_bitmap, b_bitmap))
return -1;
}
if (b_bitmap) {
if (!a_bitmap || !island_bitmap_is_subset(b_bitmap, a_bitmap))
return 1;
}
return 0;
}
static struct island_bitmap *create_or_get_island_marks(struct object *obj)
{
khiter_t pos;
int hash_ret;
pos = kh_put_oid_map(island_marks, obj->oid, &hash_ret);
2018-08-16 06:13:07 +00:00
if (hash_ret)
kh_value(island_marks, pos) = island_bitmap_new(NULL);
return kh_value(island_marks, pos);
}
static void set_island_marks(struct object *obj, struct island_bitmap *marks)
{
struct island_bitmap *b;
khiter_t pos;
int hash_ret;
pos = kh_put_oid_map(island_marks, obj->oid, &hash_ret);
2018-08-16 06:13:07 +00:00
if (hash_ret) {
/*
* We don't have one yet; make a copy-on-write of the
* parent.
*/
marks->refcount++;
kh_value(island_marks, pos) = marks;
return;
}
/*
* We do have it. Make sure we split any copy-on-write before
* updating.
*/
b = kh_value(island_marks, pos);
if (b->refcount > 1) {
b->refcount--;
b = kh_value(island_marks, pos) = island_bitmap_new(b);
}
island_bitmap_or(b, marks);
}
static void mark_remote_island_1(struct repository *r,
struct remote_island *rl,
int is_core_island)
2018-08-16 06:13:07 +00:00
{
uint32_t i;
for (i = 0; i < rl->oids.nr; ++i) {
struct island_bitmap *marks;
struct object *obj = parse_object(r, &rl->oids.oid[i]);
2018-08-16 06:13:07 +00:00
if (!obj)
continue;
marks = create_or_get_island_marks(obj);
island_bitmap_set(marks, island_counter);
if (is_core_island && obj->type == OBJ_COMMIT)
obj->flags |= NEEDS_BITMAP;
/* If it was a tag, also make sure we hit the underlying object. */
while (obj && obj->type == OBJ_TAG) {
obj = ((struct tag *)obj)->tagged;
if (obj) {
parse_object(r, &obj->oid);
2018-08-16 06:13:07 +00:00
marks = create_or_get_island_marks(obj);
island_bitmap_set(marks, island_counter);
}
}
}
if (is_core_island)
island_counter_core = island_counter;
island_counter++;
}
struct tree_islands_todo {
struct object_entry *entry;
unsigned int depth;
};
static int tree_depth_compare(const void *a, const void *b)
2018-08-16 06:13:07 +00:00
{
const struct tree_islands_todo *todo_a = a;
const struct tree_islands_todo *todo_b = b;
return todo_a->depth - todo_b->depth;
2018-08-16 06:13:07 +00:00
}
void resolve_tree_islands(struct repository *r,
int progress,
struct packing_data *to_pack)
2018-08-16 06:13:07 +00:00
{
struct progress *progress_state = NULL;
struct tree_islands_todo *todo;
2018-08-16 06:13:07 +00:00
int nr = 0;
int i;
if (!island_marks)
return;
/*
* We process only trees, as commits and tags have already been handled
* (and passed their marks on to root trees, as well. We must make sure
* to process them in descending tree-depth order so that marks
* propagate down the tree properly, even if a sub-tree is found in
* multiple parent trees.
*/
ALLOC_ARRAY(todo, to_pack->nr_objects);
for (i = 0; i < to_pack->nr_objects; i++) {
if (oe_type(&to_pack->objects[i]) == OBJ_TREE) {
todo[nr].entry = &to_pack->objects[i];
todo[nr].depth = oe_tree_depth(to_pack, &to_pack->objects[i]);
nr++;
}
2018-08-16 06:13:07 +00:00
}
QSORT(todo, nr, tree_depth_compare);
2018-08-16 06:13:07 +00:00
if (progress)
progress_state = start_progress(_("Propagating island marks"), nr);
for (i = 0; i < nr; i++) {
struct object_entry *ent = todo[i].entry;
2018-08-16 06:13:07 +00:00
struct island_bitmap *root_marks;
struct tree *tree;
struct tree_desc desc;
struct name_entry entry;
khiter_t pos;
pos = kh_get_oid_map(island_marks, ent->idx.oid);
2018-08-16 06:13:07 +00:00
if (pos >= kh_end(island_marks))
continue;
root_marks = kh_value(island_marks, pos);
tree = lookup_tree(r, &ent->idx.oid);
2018-08-16 06:13:07 +00:00
if (!tree || parse_tree(tree) < 0)
die(_("bad tree object %s"), oid_to_hex(&ent->idx.oid));
init_tree_desc(&desc, &tree->object.oid, tree->buffer, tree->size);
2018-08-16 06:13:07 +00:00
while (tree_entry(&desc, &entry)) {
struct object *obj;
if (S_ISGITLINK(entry.mode))
continue;
obj = lookup_object(r, &entry.oid);
2018-08-16 06:13:07 +00:00
if (!obj)
continue;
set_island_marks(obj, root_marks);
}
free_tree_buffer(tree);
display_progress(progress_state, i+1);
}
stop_progress(&progress_state);
free(todo);
}
struct island_load_data {
kh_str_t *remote_islands;
regex_t *rx;
size_t nr;
size_t alloc;
};
2018-08-16 06:13:07 +00:00
static const char *core_island_name;
static void free_config_regexes(struct island_load_data *ild)
2018-08-16 06:13:07 +00:00
{
for (size_t i = 0; i < ild->nr; i++)
regfree(&ild->rx[i]);
free(ild->rx);
}
static void free_remote_islands(kh_str_t *remote_islands)
{
const char *island_name;
struct remote_island *rl;
kh_foreach(remote_islands, island_name, rl, {
free((void *)island_name);
oid_array_clear(&rl->oids);
free(rl);
});
kh_destroy_str(remote_islands);
}
config: add ctx arg to config_fn_t Add a new "const struct config_context *ctx" arg to config_fn_t to hold additional information about the config iteration operation. config_context has a "struct key_value_info kvi" member that holds metadata about the config source being read (e.g. what kind of config source it is, the filename, etc). In this series, we're only interested in .kvi, so we could have just used "struct key_value_info" as an arg, but config_context makes it possible to add/adjust members in the future without changing the config_fn_t signature. We could also consider other ways of organizing the args (e.g. moving the config name and value into config_context or key_value_info), but in my experiments, the incremental benefit doesn't justify the added complexity (e.g. a config_fn_t will sometimes invoke another config_fn_t but with a different config value). In subsequent commits, the .kvi member will replace the global "struct config_reader" in config.c, making config iteration a global-free operation. It requires much more work for the machinery to provide meaningful values of .kvi, so for now, merely change the signature and call sites, pass NULL as a placeholder value, and don't rely on the arg in any meaningful way. Most of the changes are performed by contrib/coccinelle/config_fn_ctx.pending.cocci, which, for every config_fn_t: - Modifies the signature to accept "const struct config_context *ctx" - Passes "ctx" to any inner config_fn_t, if needed - Adds UNUSED attributes to "ctx", if needed Most config_fn_t instances are easily identified by seeing if they are called by the various config functions. Most of the remaining ones are manually named in the .cocci patch. Manual cleanups are still needed, but the majority of it is trivial; it's either adjusting config_fn_t that the .cocci patch didn't catch, or adding forward declarations of "struct config_context ctx" to make the signatures make sense. The non-trivial changes are in cases where we are invoking a config_fn_t outside of config machinery, and we now need to decide what value of "ctx" to pass. These cases are: - trace2/tr2_cfg.c:tr2_cfg_set_fl() This is indirectly called by git_config_set() so that the trace2 machinery can notice the new config values and update its settings using the tr2 config parsing function, i.e. tr2_cfg_cb(). - builtin/checkout.c:checkout_main() This calls git_xmerge_config() as a shorthand for parsing a CLI arg. This might be worth refactoring away in the future, since git_xmerge_config() can call git_default_config(), which can do much more than just parsing. Handle them by creating a KVI_INIT macro that initializes "struct key_value_info" to a reasonable default, and use that to construct the "ctx" arg. Signed-off-by: Glen Choo <chooglen@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2023-06-28 19:26:22 +00:00
static int island_config_callback(const char *k, const char *v,
const struct config_context *ctx UNUSED,
void *cb)
{
struct island_load_data *ild = cb;
2018-08-16 06:13:07 +00:00
if (!strcmp(k, "pack.island")) {
struct strbuf re = STRBUF_INIT;
if (!v)
return config_error_nonbool(k);
ALLOC_GROW(ild->rx, ild->nr + 1, ild->alloc);
2018-08-16 06:13:07 +00:00
if (*v != '^')
strbuf_addch(&re, '^');
strbuf_addstr(&re, v);
if (regcomp(&ild->rx[ild->nr], re.buf, REG_EXTENDED))
2018-08-16 06:13:07 +00:00
die(_("failed to load island regex for '%s': %s"), k, re.buf);
strbuf_release(&re);
ild->nr++;
2018-08-16 06:13:07 +00:00
return 0;
}
if (!strcmp(k, "pack.islandcore"))
return git_config_string(&core_island_name, k, v);
return 0;
}
static void add_ref_to_island(kh_str_t *remote_islands, const char *island_name,
const struct object_id *oid)
2018-08-16 06:13:07 +00:00
{
uint64_t sha_core;
struct remote_island *rl = NULL;
int hash_ret;
khiter_t pos = kh_put_str(remote_islands, island_name, &hash_ret);
if (hash_ret) {
kh_key(remote_islands, pos) = xstrdup(island_name);
kh_value(remote_islands, pos) = xcalloc(1, sizeof(struct remote_island));
}
rl = kh_value(remote_islands, pos);
oid_array_append(&rl->oids, oid);
memcpy(&sha_core, oid->hash, sizeof(uint64_t));
rl->hash += sha_core;
}
static int find_island_for_ref(const char *refname, const struct object_id *oid,
int flags UNUSED, void *cb)
2018-08-16 06:13:07 +00:00
{
struct island_load_data *ild = cb;
2018-08-16 06:13:07 +00:00
/*
* We should advertise 'ARRAY_SIZE(matches) - 2' as the max,
* so we can diagnose below a config with more capture groups
* than we support.
*/
regmatch_t matches[16];
int i, m;
struct strbuf island_name = STRBUF_INIT;
/* walk backwards to get last-one-wins ordering */
for (i = ild->nr - 1; i >= 0; i--) {
if (!regexec(&ild->rx[i], refname,
2018-08-16 06:13:07 +00:00
ARRAY_SIZE(matches), matches, 0))
break;
}
if (i < 0)
return 0;
if (matches[ARRAY_SIZE(matches) - 1].rm_so != -1)
warning(_("island regex from config has "
"too many capture groups (max=%d)"),
(int)ARRAY_SIZE(matches) - 2);
for (m = 1; m < ARRAY_SIZE(matches); m++) {
regmatch_t *match = &matches[m];
if (match->rm_so == -1)
continue;
if (island_name.len)
strbuf_addch(&island_name, '-');
strbuf_add(&island_name, refname + match->rm_so, match->rm_eo - match->rm_so);
}
add_ref_to_island(ild->remote_islands, island_name.buf, oid);
2018-08-16 06:13:07 +00:00
strbuf_release(&island_name);
return 0;
}
static struct remote_island *get_core_island(kh_str_t *remote_islands)
2018-08-16 06:13:07 +00:00
{
if (core_island_name) {
khiter_t pos = kh_get_str(remote_islands, core_island_name);
if (pos < kh_end(remote_islands))
return kh_value(remote_islands, pos);
}
return NULL;
}
static void deduplicate_islands(kh_str_t *remote_islands, struct repository *r)
2018-08-16 06:13:07 +00:00
{
struct remote_island *island, *core = NULL, **list;
unsigned int island_count, dst, src, ref, i = 0;
island_count = kh_size(remote_islands);
ALLOC_ARRAY(list, island_count);
kh_foreach_value(remote_islands, island, {
list[i++] = island;
});
for (ref = 0; ref + 1 < island_count; ref++) {
for (src = ref + 1, dst = src; src < island_count; src++) {
if (list[ref]->hash == list[src]->hash)
continue;
if (src != dst)
list[dst] = list[src];
dst++;
}
island_count = dst;
}
island_bitmap_size = (island_count / 32) + 1;
core = get_core_island(remote_islands);
2018-08-16 06:13:07 +00:00
for (i = 0; i < island_count; ++i) {
mark_remote_island_1(r, list[i], core && list[i]->hash == core->hash);
2018-08-16 06:13:07 +00:00
}
free(list);
}
void load_delta_islands(struct repository *r, int progress)
2018-08-16 06:13:07 +00:00
{
struct island_load_data ild = { 0 };
island_marks = kh_init_oid_map();
2018-08-16 06:13:07 +00:00
git_config(island_config_callback, &ild);
ild.remote_islands = kh_init_str();
for_each_ref(find_island_for_ref, &ild);
free_config_regexes(&ild);
deduplicate_islands(ild.remote_islands, r);
free_remote_islands(ild.remote_islands);
2018-08-16 06:13:07 +00:00
if (progress)
fprintf(stderr, _("Marked %d islands, done.\n"), island_counter);
2018-08-16 06:13:07 +00:00
}
void propagate_island_marks(struct commit *commit)
{
khiter_t pos = kh_get_oid_map(island_marks, commit->object.oid);
2018-08-16 06:13:07 +00:00
if (pos < kh_end(island_marks)) {
struct commit_list *p;
struct island_bitmap *root_marks = kh_value(island_marks, pos);
repo_parse_commit(the_repository, commit);
set_island_marks(&repo_get_commit_tree(the_repository, commit)->object,
root_marks);
2018-08-16 06:13:07 +00:00
for (p = commit->parents; p; p = p->next)
set_island_marks(&p->item->object, root_marks);
}
}
void free_island_marks(void)
{
struct island_bitmap *bitmap;
if (island_marks) {
kh_foreach_value(island_marks, bitmap, {
if (!--bitmap->refcount)
free(bitmap);
});
kh_destroy_oid_map(island_marks);
}
/* detect use-after-free with a an address which is never valid: */
island_marks = (void *)-1;
}
2018-08-16 06:13:07 +00:00
int compute_pack_layers(struct packing_data *to_pack)
{
uint32_t i;
if (!core_island_name || !island_marks)
return 1;
for (i = 0; i < to_pack->nr_objects; ++i) {
struct object_entry *entry = &to_pack->objects[i];
khiter_t pos = kh_get_oid_map(island_marks, entry->idx.oid);
2018-08-16 06:13:07 +00:00
oe_set_layer(to_pack, entry, 1);
2018-08-16 06:13:07 +00:00
if (pos < kh_end(island_marks)) {
struct island_bitmap *bitmap = kh_value(island_marks, pos);
if (island_bitmap_get(bitmap, island_counter_core))
oe_set_layer(to_pack, entry, 0);
2018-08-16 06:13:07 +00:00
}
}
return 2;
}