shallow.con commit smart-http: support shallow fetch/clone (1609488)
   1#include "cache.h"
   2#include "commit.h"
   3#include "tag.h"
   4#include "pkt-line.h"
   5#include "remote.h"
   6#include "refs.h"
   7#include "sha1-array.h"
   8#include "diff.h"
   9#include "revision.h"
  10#include "commit-slab.h"
  11
  12static int is_shallow = -1;
  13static struct stat shallow_stat;
  14static char *alternate_shallow_file;
  15
  16void set_alternate_shallow_file(const char *path, int override)
  17{
  18        if (is_shallow != -1)
  19                die("BUG: is_repository_shallow must not be called before set_alternate_shallow_file");
  20        if (alternate_shallow_file && !override)
  21                return;
  22        free(alternate_shallow_file);
  23        alternate_shallow_file = path ? xstrdup(path) : NULL;
  24}
  25
  26int register_shallow(const unsigned char *sha1)
  27{
  28        struct commit_graft *graft =
  29                xmalloc(sizeof(struct commit_graft));
  30        struct commit *commit = lookup_commit(sha1);
  31
  32        hashcpy(graft->sha1, sha1);
  33        graft->nr_parent = -1;
  34        if (commit && commit->object.parsed)
  35                commit->parents = NULL;
  36        return register_commit_graft(graft, 0);
  37}
  38
  39int is_repository_shallow(void)
  40{
  41        FILE *fp;
  42        char buf[1024];
  43        const char *path = alternate_shallow_file;
  44
  45        if (is_shallow >= 0)
  46                return is_shallow;
  47
  48        if (!path)
  49                path = git_path("shallow");
  50        /*
  51         * fetch-pack sets '--shallow-file ""' as an indicator that no
  52         * shallow file should be used. We could just open it and it
  53         * will likely fail. But let's do an explicit check instead.
  54         */
  55        if (!*path ||
  56            stat(path, &shallow_stat) ||
  57            (fp = fopen(path, "r")) == NULL) {
  58                is_shallow = 0;
  59                return is_shallow;
  60        }
  61        is_shallow = 1;
  62
  63        while (fgets(buf, sizeof(buf), fp)) {
  64                unsigned char sha1[20];
  65                if (get_sha1_hex(buf, sha1))
  66                        die("bad shallow line: %s", buf);
  67                register_shallow(sha1);
  68        }
  69        fclose(fp);
  70        return is_shallow;
  71}
  72
  73struct commit_list *get_shallow_commits(struct object_array *heads, int depth,
  74                int shallow_flag, int not_shallow_flag)
  75{
  76        int i = 0, cur_depth = 0;
  77        struct commit_list *result = NULL;
  78        struct object_array stack = OBJECT_ARRAY_INIT;
  79        struct commit *commit = NULL;
  80        struct commit_graft *graft;
  81
  82        while (commit || i < heads->nr || stack.nr) {
  83                struct commit_list *p;
  84                if (!commit) {
  85                        if (i < heads->nr) {
  86                                commit = (struct commit *)
  87                                        deref_tag(heads->objects[i++].item, NULL, 0);
  88                                if (!commit || commit->object.type != OBJ_COMMIT) {
  89                                        commit = NULL;
  90                                        continue;
  91                                }
  92                                if (!commit->util)
  93                                        commit->util = xmalloc(sizeof(int));
  94                                *(int *)commit->util = 0;
  95                                cur_depth = 0;
  96                        } else {
  97                                commit = (struct commit *)
  98                                        stack.objects[--stack.nr].item;
  99                                cur_depth = *(int *)commit->util;
 100                        }
 101                }
 102                if (parse_commit(commit))
 103                        die("invalid commit");
 104                cur_depth++;
 105                if ((depth != INFINITE_DEPTH && cur_depth >= depth) ||
 106                    (is_repository_shallow() && !commit->parents &&
 107                     (graft = lookup_commit_graft(commit->object.sha1)) != NULL &&
 108                     graft->nr_parent < 0)) {
 109                        commit_list_insert(commit, &result);
 110                        commit->object.flags |= shallow_flag;
 111                        commit = NULL;
 112                        continue;
 113                }
 114                commit->object.flags |= not_shallow_flag;
 115                for (p = commit->parents, commit = NULL; p; p = p->next) {
 116                        if (!p->item->util) {
 117                                int *pointer = xmalloc(sizeof(int));
 118                                p->item->util = pointer;
 119                                *pointer =  cur_depth;
 120                        } else {
 121                                int *pointer = p->item->util;
 122                                if (cur_depth >= *pointer)
 123                                        continue;
 124                                *pointer = cur_depth;
 125                        }
 126                        if (p->next)
 127                                add_object_array(&p->item->object,
 128                                                NULL, &stack);
 129                        else {
 130                                commit = p->item;
 131                                cur_depth = *(int *)commit->util;
 132                        }
 133                }
 134        }
 135
 136        return result;
 137}
 138
 139void check_shallow_file_for_update(void)
 140{
 141        struct stat st;
 142
 143        if (!is_shallow)
 144                return;
 145        else if (is_shallow == -1)
 146                die("BUG: shallow must be initialized by now");
 147
 148        if (stat(git_path("shallow"), &st))
 149                die("shallow file was removed during fetch");
 150        else if (st.st_mtime != shallow_stat.st_mtime
 151#ifdef USE_NSEC
 152                 || ST_MTIME_NSEC(st) != ST_MTIME_NSEC(shallow_stat)
 153#endif
 154                   )
 155                die("shallow file was changed during fetch");
 156}
 157
 158struct write_shallow_data {
 159        struct strbuf *out;
 160        int use_pack_protocol;
 161        int count;
 162};
 163
 164static int write_one_shallow(const struct commit_graft *graft, void *cb_data)
 165{
 166        struct write_shallow_data *data = cb_data;
 167        const char *hex = sha1_to_hex(graft->sha1);
 168        if (graft->nr_parent != -1)
 169                return 0;
 170        data->count++;
 171        if (data->use_pack_protocol)
 172                packet_buf_write(data->out, "shallow %s", hex);
 173        else {
 174                strbuf_addstr(data->out, hex);
 175                strbuf_addch(data->out, '\n');
 176        }
 177        return 0;
 178}
 179
 180int write_shallow_commits(struct strbuf *out, int use_pack_protocol,
 181                          const struct sha1_array *extra)
 182{
 183        struct write_shallow_data data;
 184        int i;
 185        data.out = out;
 186        data.use_pack_protocol = use_pack_protocol;
 187        data.count = 0;
 188        for_each_commit_graft(write_one_shallow, &data);
 189        if (!extra)
 190                return data.count;
 191        for (i = 0; i < extra->nr; i++) {
 192                strbuf_addstr(out, sha1_to_hex(extra->sha1[i]));
 193                strbuf_addch(out, '\n');
 194                data.count++;
 195        }
 196        return data.count;
 197}
 198
 199char *setup_temporary_shallow(const struct sha1_array *extra)
 200{
 201        struct strbuf sb = STRBUF_INIT;
 202        int fd;
 203
 204        if (write_shallow_commits(&sb, 0, extra)) {
 205                struct strbuf path = STRBUF_INIT;
 206                strbuf_addstr(&path, git_path("shallow_XXXXXX"));
 207                fd = xmkstemp(path.buf);
 208                if (write_in_full(fd, sb.buf, sb.len) != sb.len)
 209                        die_errno("failed to write to %s",
 210                                  path.buf);
 211                close(fd);
 212                strbuf_release(&sb);
 213                return strbuf_detach(&path, NULL);
 214        }
 215        /*
 216         * is_repository_shallow() sees empty string as "no shallow
 217         * file".
 218         */
 219        return xstrdup("");
 220}
 221
 222void setup_alternate_shallow(struct lock_file *shallow_lock,
 223                             const char **alternate_shallow_file,
 224                             const struct sha1_array *extra)
 225{
 226        struct strbuf sb = STRBUF_INIT;
 227        int fd;
 228
 229        check_shallow_file_for_update();
 230        fd = hold_lock_file_for_update(shallow_lock, git_path("shallow"),
 231                                       LOCK_DIE_ON_ERROR);
 232        if (write_shallow_commits(&sb, 0, extra)) {
 233                if (write_in_full(fd, sb.buf, sb.len) != sb.len)
 234                        die_errno("failed to write to %s",
 235                                  shallow_lock->filename);
 236                *alternate_shallow_file = shallow_lock->filename;
 237        } else
 238                /*
 239                 * is_repository_shallow() sees empty string as "no
 240                 * shallow file".
 241                 */
 242                *alternate_shallow_file = "";
 243        strbuf_release(&sb);
 244}
 245
 246static int advertise_shallow_grafts_cb(const struct commit_graft *graft, void *cb)
 247{
 248        int fd = *(int *)cb;
 249        if (graft->nr_parent == -1)
 250                packet_write(fd, "shallow %s\n", sha1_to_hex(graft->sha1));
 251        return 0;
 252}
 253
 254void advertise_shallow_grafts(int fd)
 255{
 256        if (!is_repository_shallow())
 257                return;
 258        for_each_commit_graft(advertise_shallow_grafts_cb, &fd);
 259}
 260
 261#define TRACE_KEY "GIT_TRACE_SHALLOW"
 262
 263/*
 264 * Step 1, split sender shallow commits into "ours" and "theirs"
 265 * Step 2, clean "ours" based on .git/shallow
 266 */
 267void prepare_shallow_info(struct shallow_info *info, struct sha1_array *sa)
 268{
 269        int i;
 270        trace_printf_key(TRACE_KEY, "shallow: prepare_shallow_info\n");
 271        memset(info, 0, sizeof(*info));
 272        info->shallow = sa;
 273        if (!sa)
 274                return;
 275        info->ours = xmalloc(sizeof(*info->ours) * sa->nr);
 276        info->theirs = xmalloc(sizeof(*info->theirs) * sa->nr);
 277        for (i = 0; i < sa->nr; i++) {
 278                if (has_sha1_file(sa->sha1[i])) {
 279                        struct commit_graft *graft;
 280                        graft = lookup_commit_graft(sa->sha1[i]);
 281                        if (graft && graft->nr_parent < 0)
 282                                continue;
 283                        info->ours[info->nr_ours++] = i;
 284                } else
 285                        info->theirs[info->nr_theirs++] = i;
 286        }
 287}
 288
 289void clear_shallow_info(struct shallow_info *info)
 290{
 291        free(info->ours);
 292        free(info->theirs);
 293}
 294
 295/* Step 4, remove non-existent ones in "theirs" after getting the pack */
 296
 297void remove_nonexistent_theirs_shallow(struct shallow_info *info)
 298{
 299        unsigned char (*sha1)[20] = info->shallow->sha1;
 300        int i, dst;
 301        trace_printf_key(TRACE_KEY, "shallow: remove_nonexistent_theirs_shallow\n");
 302        for (i = dst = 0; i < info->nr_theirs; i++) {
 303                if (i != dst)
 304                        info->theirs[dst] = info->theirs[i];
 305                if (has_sha1_file(sha1[info->theirs[i]]))
 306                        dst++;
 307        }
 308        info->nr_theirs = dst;
 309}
 310
 311/* Step 5, remove non-existent ones in "ours" in the pack */
 312void remove_nonexistent_ours_in_pack(struct shallow_info *info,
 313                                     struct packed_git *p)
 314{
 315        unsigned char (*sha1)[20] = info->shallow->sha1;
 316        int i, dst;
 317        trace_printf_key(TRACE_KEY, "shallow: remove_nonexistent_ours_in_pack\n");
 318        for (i = dst = 0; i < info->nr_ours; i++) {
 319                if (i != dst)
 320                        info->ours[dst] = info->ours[i];
 321                if (find_pack_entry_one(sha1[info->ours[i]], p))
 322                        dst++;
 323        }
 324        info->nr_ours = dst;
 325}
 326
 327define_commit_slab(ref_bitmap, uint32_t *);
 328
 329struct paint_info {
 330        struct ref_bitmap ref_bitmap;
 331        unsigned nr_bits;
 332        char **slab;
 333        char *free, *end;
 334        unsigned slab_count;
 335};
 336
 337static uint32_t *paint_alloc(struct paint_info *info)
 338{
 339        unsigned nr = (info->nr_bits + 31) / 32;
 340        unsigned size = nr * sizeof(uint32_t);
 341        void *p;
 342        if (!info->slab_count || info->free + size > info->end) {
 343                info->slab_count++;
 344                info->slab = xrealloc(info->slab,
 345                                      info->slab_count * sizeof(*info->slab));
 346                info->free = xmalloc(COMMIT_SLAB_SIZE);
 347                info->slab[info->slab_count - 1] = info->free;
 348                info->end = info->free + COMMIT_SLAB_SIZE;
 349        }
 350        p = info->free;
 351        info->free += size;
 352        return p;
 353}
 354
 355/*
 356 * Given a commit SHA-1, walk down to parents until either SEEN,
 357 * UNINTERESTING or BOTTOM is hit. Set the id-th bit in ref_bitmap for
 358 * all walked commits.
 359 */
 360static void paint_down(struct paint_info *info, const unsigned char *sha1,
 361                       int id)
 362{
 363        unsigned int i, nr;
 364        struct commit_list *head = NULL;
 365        int bitmap_nr = (info->nr_bits + 31) / 32;
 366        int bitmap_size = bitmap_nr * sizeof(uint32_t);
 367        uint32_t *tmp = xmalloc(bitmap_size); /* to be freed before return */
 368        uint32_t *bitmap = paint_alloc(info);
 369        struct commit *c = lookup_commit_reference_gently(sha1, 1);
 370        if (!c)
 371                return;
 372        memset(bitmap, 0, bitmap_size);
 373        bitmap[id / 32] |= (1 << (id % 32));
 374        commit_list_insert(c, &head);
 375        while (head) {
 376                struct commit_list *p;
 377                struct commit *c = head->item;
 378                uint32_t **refs = ref_bitmap_at(&info->ref_bitmap, c);
 379
 380                p = head;
 381                head = head->next;
 382                free(p);
 383
 384                /* XXX check "UNINTERESTING" from pack bitmaps if available */
 385                if (c->object.flags & (SEEN | UNINTERESTING))
 386                        continue;
 387                else
 388                        c->object.flags |= SEEN;
 389
 390                if (*refs == NULL)
 391                        *refs = bitmap;
 392                else {
 393                        memcpy(tmp, *refs, bitmap_size);
 394                        for (i = 0; i < bitmap_nr; i++)
 395                                tmp[i] |= bitmap[i];
 396                        if (memcmp(tmp, *refs, bitmap_size)) {
 397                                *refs = paint_alloc(info);
 398                                memcpy(*refs, tmp, bitmap_size);
 399                        }
 400                }
 401
 402                if (c->object.flags & BOTTOM)
 403                        continue;
 404
 405                if (parse_commit(c))
 406                        die("unable to parse commit %s",
 407                            sha1_to_hex(c->object.sha1));
 408
 409                for (p = c->parents; p; p = p->next) {
 410                        uint32_t **p_refs = ref_bitmap_at(&info->ref_bitmap,
 411                                                          p->item);
 412                        if (p->item->object.flags & SEEN)
 413                                continue;
 414                        if (*p_refs == NULL || *p_refs == *refs)
 415                                *p_refs = *refs;
 416                        commit_list_insert(p->item, &head);
 417                }
 418        }
 419
 420        nr = get_max_object_index();
 421        for (i = 0; i < nr; i++) {
 422                struct object *o = get_indexed_object(i);
 423                if (o && o->type == OBJ_COMMIT)
 424                        o->flags &= ~SEEN;
 425        }
 426
 427        free(tmp);
 428}
 429
 430static int mark_uninteresting(const char *refname,
 431                              const unsigned char *sha1,
 432                              int flags, void *cb_data)
 433{
 434        struct commit *commit = lookup_commit_reference_gently(sha1, 1);
 435        if (!commit)
 436                return 0;
 437        commit->object.flags |= UNINTERESTING;
 438        mark_parents_uninteresting(commit);
 439        return 0;
 440}
 441
 442static void post_assign_shallow(struct shallow_info *info,
 443                                struct ref_bitmap *ref_bitmap,
 444                                int *ref_status);
 445/*
 446 * Step 6(+7), associate shallow commits with new refs
 447 *
 448 * info->ref must be initialized before calling this function.
 449 *
 450 * If used is not NULL, it's an array of info->shallow->nr
 451 * bitmaps. The n-th bit set in the m-th bitmap if ref[n] needs the
 452 * m-th shallow commit from info->shallow.
 453 *
 454 * If used is NULL, "ours" and "theirs" are updated. And if ref_status
 455 * is not NULL it's an array of ref->nr ints. ref_status[i] is true if
 456 * the ref needs some shallow commits from either info->ours or
 457 * info->theirs.
 458 */
 459void assign_shallow_commits_to_refs(struct shallow_info *info,
 460                                    uint32_t **used, int *ref_status)
 461{
 462        unsigned char (*sha1)[20] = info->shallow->sha1;
 463        struct sha1_array *ref = info->ref;
 464        unsigned int i, nr;
 465        int *shallow, nr_shallow = 0;
 466        struct paint_info pi;
 467
 468        trace_printf_key(TRACE_KEY, "shallow: assign_shallow_commits_to_refs\n");
 469        shallow = xmalloc(sizeof(*shallow) * (info->nr_ours + info->nr_theirs));
 470        for (i = 0; i < info->nr_ours; i++)
 471                shallow[nr_shallow++] = info->ours[i];
 472        for (i = 0; i < info->nr_theirs; i++)
 473                shallow[nr_shallow++] = info->theirs[i];
 474
 475        /*
 476         * Prepare the commit graph to track what refs can reach what
 477         * (new) shallow commits.
 478         */
 479        nr = get_max_object_index();
 480        for (i = 0; i < nr; i++) {
 481                struct object *o = get_indexed_object(i);
 482                if (!o || o->type != OBJ_COMMIT)
 483                        continue;
 484
 485                o->flags &= ~(UNINTERESTING | BOTTOM | SEEN);
 486        }
 487
 488        memset(&pi, 0, sizeof(pi));
 489        init_ref_bitmap(&pi.ref_bitmap);
 490        pi.nr_bits = ref->nr;
 491
 492        /*
 493         * "--not --all" to cut short the traversal if new refs
 494         * connect to old refs. If not (e.g. force ref updates) it'll
 495         * have to go down to the current shallow commits.
 496         */
 497        head_ref(mark_uninteresting, NULL);
 498        for_each_ref(mark_uninteresting, NULL);
 499
 500        /* Mark potential bottoms so we won't go out of bound */
 501        for (i = 0; i < nr_shallow; i++) {
 502                struct commit *c = lookup_commit(sha1[shallow[i]]);
 503                c->object.flags |= BOTTOM;
 504        }
 505
 506        for (i = 0; i < ref->nr; i++)
 507                paint_down(&pi, ref->sha1[i], i);
 508
 509        if (used) {
 510                int bitmap_size = ((pi.nr_bits + 31) / 32) * sizeof(uint32_t);
 511                memset(used, 0, sizeof(*used) * info->shallow->nr);
 512                for (i = 0; i < nr_shallow; i++) {
 513                        const struct commit *c = lookup_commit(sha1[shallow[i]]);
 514                        uint32_t **map = ref_bitmap_at(&pi.ref_bitmap, c);
 515                        if (*map)
 516                                used[shallow[i]] = xmemdupz(*map, bitmap_size);
 517                }
 518                /*
 519                 * unreachable shallow commits are not removed from
 520                 * "ours" and "theirs". The user is supposed to run
 521                 * step 7 on every ref separately and not trust "ours"
 522                 * and "theirs" any more.
 523                 */
 524        } else
 525                post_assign_shallow(info, &pi.ref_bitmap, ref_status);
 526
 527        clear_ref_bitmap(&pi.ref_bitmap);
 528        for (i = 0; i < pi.slab_count; i++)
 529                free(pi.slab[i]);
 530        free(pi.slab);
 531        free(shallow);
 532}
 533
 534struct commit_array {
 535        struct commit **commits;
 536        int nr, alloc;
 537};
 538
 539static int add_ref(const char *refname,
 540                   const unsigned char *sha1, int flags, void *cb_data)
 541{
 542        struct commit_array *ca = cb_data;
 543        ALLOC_GROW(ca->commits, ca->nr + 1, ca->alloc);
 544        ca->commits[ca->nr] = lookup_commit_reference_gently(sha1, 1);
 545        if (ca->commits[ca->nr])
 546                ca->nr++;
 547        return 0;
 548}
 549
 550static void update_refstatus(int *ref_status, int nr, uint32_t *bitmap)
 551{
 552        int i;
 553        if (!ref_status)
 554                return;
 555        for (i = 0; i < nr; i++)
 556                if (bitmap[i / 32] & (1 << (i % 32)))
 557                        ref_status[i]++;
 558}
 559
 560/*
 561 * Step 7, reachability test on "ours" at commit level
 562 */
 563static void post_assign_shallow(struct shallow_info *info,
 564                                struct ref_bitmap *ref_bitmap,
 565                                int *ref_status)
 566{
 567        unsigned char (*sha1)[20] = info->shallow->sha1;
 568        struct commit *c;
 569        uint32_t **bitmap;
 570        int dst, i, j;
 571        int bitmap_nr = (info->ref->nr + 31) / 32;
 572        struct commit_array ca;
 573
 574        trace_printf_key(TRACE_KEY, "shallow: post_assign_shallow\n");
 575        if (ref_status)
 576                memset(ref_status, 0, sizeof(*ref_status) * info->ref->nr);
 577
 578        /* Remove unreachable shallow commits from "theirs" */
 579        for (i = dst = 0; i < info->nr_theirs; i++) {
 580                if (i != dst)
 581                        info->theirs[dst] = info->theirs[i];
 582                c = lookup_commit(sha1[info->theirs[i]]);
 583                bitmap = ref_bitmap_at(ref_bitmap, c);
 584                if (!*bitmap)
 585                        continue;
 586                for (j = 0; j < bitmap_nr; j++)
 587                        if (bitmap[0][j]) {
 588                                update_refstatus(ref_status, info->ref->nr, *bitmap);
 589                                dst++;
 590                                break;
 591                        }
 592        }
 593        info->nr_theirs = dst;
 594
 595        memset(&ca, 0, sizeof(ca));
 596        head_ref(add_ref, &ca);
 597        for_each_ref(add_ref, &ca);
 598
 599        /* Remove unreachable shallow commits from "ours" */
 600        for (i = dst = 0; i < info->nr_ours; i++) {
 601                if (i != dst)
 602                        info->ours[dst] = info->ours[i];
 603                c = lookup_commit(sha1[info->ours[i]]);
 604                bitmap = ref_bitmap_at(ref_bitmap, c);
 605                if (!*bitmap)
 606                        continue;
 607                for (j = 0; j < bitmap_nr; j++)
 608                        if (bitmap[0][j] &&
 609                            /* Step 7, reachability test at commit level */
 610                            !in_merge_bases_many(c, ca.nr, ca.commits)) {
 611                                update_refstatus(ref_status, info->ref->nr, *bitmap);
 612                                dst++;
 613                                break;
 614                        }
 615        }
 616        info->nr_ours = dst;
 617
 618        free(ca.commits);
 619}
 620
 621/* (Delayed) step 7, reachability test at commit level */
 622int delayed_reachability_test(struct shallow_info *si, int c)
 623{
 624        if (si->need_reachability_test[c]) {
 625                struct commit *commit = lookup_commit(si->shallow->sha1[c]);
 626
 627                if (!si->commits) {
 628                        struct commit_array ca;
 629                        memset(&ca, 0, sizeof(ca));
 630                        head_ref(add_ref, &ca);
 631                        for_each_ref(add_ref, &ca);
 632                        si->commits = ca.commits;
 633                        si->nr_commits = ca.nr;
 634                }
 635
 636                si->reachable[c] = in_merge_bases_many(commit,
 637                                                       si->nr_commits,
 638                                                       si->commits);
 639                si->need_reachability_test[c] = 0;
 640        }
 641        return si->reachable[c];
 642}