Commit fe1ed56f authored by Derrick Stolee's avatar Derrick Stolee Committed by Junio C Hamano

midx: sort and deduplicate objects from packfiles

Before writing a list of objects and their offsets to a multi-pack-index,
we need to collect the list of objects contained in the packfiles. There
may be multiple copies of some objects, so this list must be deduplicated.

It is possible to artificially get into a state where there are many
duplicate copies of objects. That can create high memory pressure if we
are to create a list of all objects before de-duplication. To reduce
this memory pressure without a significant performance drop,
automatically group objects by the first byte of their object id. Use
the IDX fanout tables to group the data, copy to a local array, then
sort.

Copy only the de-duplicated entries. Select the duplicate based on the
most-recent modified time of a packfile containing the object.
Signed-off-by: default avatarDerrick Stolee <dstolee@microsoft.com>
Signed-off-by: default avatarJunio C Hamano <gitster@pobox.com>
parent 3227565c
......@@ -4,6 +4,7 @@
#include "lockfile.h"
#include "packfile.h"
#include "object-store.h"
#include "packfile.h"
#include "midx.h"
#define MIDX_SIGNATURE 0x4d494458 /* "MIDX" */
......@@ -182,12 +183,21 @@ static void add_pack_to_midx(const char *full_path, size_t full_path_len,
packs->list[packs->nr] = add_packed_git(full_path,
full_path_len,
0);
if (!packs->list[packs->nr]) {
warning(_("failed to add packfile '%s'"),
full_path);
return;
}
if (open_pack_index(packs->list[packs->nr])) {
warning(_("failed to open pack-index '%s'"),
full_path);
close_pack(packs->list[packs->nr]);
FREE_AND_NULL(packs->list[packs->nr]);
return;
}
packs->names[packs->nr] = xstrdup(file_name);
packs->pack_name_concat_len += strlen(file_name) + 1;
packs->nr++;
......@@ -228,6 +238,119 @@ static void sort_packs_by_name(char **pack_names, uint32_t nr_packs, uint32_t *p
free(pairs);
}
struct pack_midx_entry {
struct object_id oid;
uint32_t pack_int_id;
time_t pack_mtime;
uint64_t offset;
};
static int midx_oid_compare(const void *_a, const void *_b)
{
const struct pack_midx_entry *a = (const struct pack_midx_entry *)_a;
const struct pack_midx_entry *b = (const struct pack_midx_entry *)_b;
int cmp = oidcmp(&a->oid, &b->oid);
if (cmp)
return cmp;
if (a->pack_mtime > b->pack_mtime)
return -1;
else if (a->pack_mtime < b->pack_mtime)
return 1;
return a->pack_int_id - b->pack_int_id;
}
static void fill_pack_entry(uint32_t pack_int_id,
struct packed_git *p,
uint32_t cur_object,
struct pack_midx_entry *entry)
{
if (!nth_packed_object_oid(&entry->oid, p, cur_object))
die(_("failed to locate object %d in packfile"), cur_object);
entry->pack_int_id = pack_int_id;
entry->pack_mtime = p->mtime;
entry->offset = nth_packed_object_offset(p, cur_object);
}
/*
* It is possible to artificially get into a state where there are many
* duplicate copies of objects. That can create high memory pressure if
* we are to create a list of all objects before de-duplication. To reduce
* this memory pressure without a significant performance drop, automatically
* group objects by the first byte of their object id. Use the IDX fanout
* tables to group the data, copy to a local array, then sort.
*
* Copy only the de-duplicated entries (selected by most-recent modified time
* of a packfile containing the object).
*/
static struct pack_midx_entry *get_sorted_entries(struct packed_git **p,
uint32_t *perm,
uint32_t nr_packs,
uint32_t *nr_objects)
{
uint32_t cur_fanout, cur_pack, cur_object;
uint32_t alloc_fanout, alloc_objects, total_objects = 0;
struct pack_midx_entry *entries_by_fanout = NULL;
struct pack_midx_entry *deduplicated_entries = NULL;
for (cur_pack = 0; cur_pack < nr_packs; cur_pack++)
total_objects += p[cur_pack]->num_objects;
/*
* As we de-duplicate by fanout value, we expect the fanout
* slices to be evenly distributed, with some noise. Hence,
* allocate slightly more than one 256th.
*/
alloc_objects = alloc_fanout = total_objects > 3200 ? total_objects / 200 : 16;
ALLOC_ARRAY(entries_by_fanout, alloc_fanout);
ALLOC_ARRAY(deduplicated_entries, alloc_objects);
*nr_objects = 0;
for (cur_fanout = 0; cur_fanout < 256; cur_fanout++) {
uint32_t nr_fanout = 0;
for (cur_pack = 0; cur_pack < nr_packs; cur_pack++) {
uint32_t start = 0, end;
if (cur_fanout)
start = get_pack_fanout(p[cur_pack], cur_fanout - 1);
end = get_pack_fanout(p[cur_pack], cur_fanout);
for (cur_object = start; cur_object < end; cur_object++) {
ALLOC_GROW(entries_by_fanout, nr_fanout + 1, alloc_fanout);
fill_pack_entry(perm[cur_pack], p[cur_pack], cur_object, &entries_by_fanout[nr_fanout]);
nr_fanout++;
}
}
QSORT(entries_by_fanout, nr_fanout, midx_oid_compare);
/*
* The batch is now sorted by OID and then mtime (descending).
* Take only the first duplicate.
*/
for (cur_object = 0; cur_object < nr_fanout; cur_object++) {
if (cur_object && !oidcmp(&entries_by_fanout[cur_object - 1].oid,
&entries_by_fanout[cur_object].oid))
continue;
ALLOC_GROW(deduplicated_entries, *nr_objects + 1, alloc_objects);
memcpy(&deduplicated_entries[*nr_objects],
&entries_by_fanout[cur_object],
sizeof(struct pack_midx_entry));
(*nr_objects)++;
}
}
free(entries_by_fanout);
return deduplicated_entries;
}
static size_t write_midx_pack_names(struct hashfile *f,
char **pack_names,
uint32_t num_packs)
......@@ -271,6 +394,8 @@ int write_midx_file(const char *object_dir)
uint64_t written = 0;
uint32_t chunk_ids[MIDX_MAX_CHUNKS + 1];
uint64_t chunk_offsets[MIDX_MAX_CHUNKS + 1];
uint32_t nr_entries;
struct pack_midx_entry *entries = NULL;
midx_name = get_midx_filename(object_dir);
if (safe_create_leading_directories(midx_name)) {
......@@ -296,6 +421,8 @@ int write_midx_file(const char *object_dir)
ALLOC_ARRAY(pack_perm, packs.nr);
sort_packs_by_name(packs.names, packs.nr, pack_perm);
entries = get_sorted_entries(packs.list, pack_perm, packs.nr, &nr_entries);
hold_lock_file_for_update(&lk, midx_name, LOCK_DIE_ON_ERROR);
f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
FREE_AND_NULL(midx_name);
......@@ -365,5 +492,6 @@ int write_midx_file(const char *object_dir)
free(packs.list);
free(packs.names);
free(entries);
return 0;
}
......@@ -196,6 +196,23 @@ int open_pack_index(struct packed_git *p)
return ret;
}
uint32_t get_pack_fanout(struct packed_git *p, uint32_t value)
{
const uint32_t *level1_ofs = p->index_data;
if (!level1_ofs) {
if (open_pack_index(p))
return 0;
level1_ofs = p->index_data;
}
if (p->index_version > 1) {
level1_ofs += 2;
}
return ntohl(level1_ofs[value]);
}
static struct packed_git *alloc_packed_git(int extra)
{
struct packed_git *p = xmalloc(st_add(sizeof(*p), extra));
......
......@@ -69,6 +69,8 @@ extern int open_pack_index(struct packed_git *);
*/
extern void close_pack_index(struct packed_git *);
extern uint32_t get_pack_fanout(struct packed_git *p, uint32_t value);
extern unsigned char *use_pack(struct packed_git *, struct pack_window **, off_t, unsigned long *);
extern void close_pack_windows(struct packed_git *);
extern void close_pack(struct packed_git *);
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment