bulk-checkin.c

   1 /*
   2  * Copyright (c) 2011, Google Inc.
   3  */
   4 #include "bulk-checkin.h"
   5 #include "csum-file.h"
   6 #include "pack.h"
   7 #include "strbuf.h"
   8
   9 static int pack_compression_level = Z_DEFAULT_COMPRESSION;
  10
  11 static struct bulk_checkin_state {
  12         unsigned plugged:1;
  13
  14         char *pack_tmp_name;
  15         struct sha1file *f;
  16         off_t offset;
  17         struct pack_idx_option pack_idx_opts;
  18
  19         struct pack_idx_entry **written;
  20         uint32_t alloc_written;
  21         uint32_t nr_written;
  22 } state;
  23
  24 static void finish_bulk_checkin(struct bulk_checkin_state *state)
  25 {
  26         unsigned char sha1[20];
  27         struct strbuf packname = STRBUF_INIT;
  28         int i;
  29
  30         if (!state->f)
  31                 return;
  32
  33         if (state->nr_written == 0) {
  34                 close(state->f->fd);
  35                 unlink(state->pack_tmp_name);
  36                 goto clear_exit;
  37         } else if (state->nr_written == 1) {
  38                 sha1close(state->f, sha1, CSUM_FSYNC);
  39         } else {
  40                 int fd = sha1close(state->f, sha1, 0);
  41                 fixup_pack_header_footer(fd, sha1, state->pack_tmp_name,
  42                                          state->nr_written, sha1,
  43                                          state->offset);
  44                 close(fd);
  45         }
  46
  47         strbuf_addf(&packname, "%s/pack/pack-", get_object_directory());
  48         finish_tmp_packfile(&packname, state->pack_tmp_name,
  49                             state->written, state->nr_written,
  50                             &state->pack_idx_opts, sha1);
  51         for (i = 0; i < state->nr_written; i++)
  52                 free(state->written[i]);
  53
  54 clear_exit:
  55         free(state->written);
  56         memset(state, 0, sizeof(*state));
  57
  58         strbuf_release(&packname);
  59         /* Make objects we just wrote available to ourselves */
  60         reprepare_packed_git();
  61 }
  62
  63 static int already_written(struct bulk_checkin_state *state, unsigned char sha1[])
  64 {
  65         int i;
  66
  67         /* The object may already exist in the repository */
  68         if (has_sha1_file(sha1))
  69                 return 1;
  70
  71         /* Might want to keep the list sorted */
  72         for (i = 0; i < state->nr_written; i++)
  73                 if (!hashcmp(state->written[i]->sha1, sha1))
  74                         return 1;
  75
  76         /* This is a new object we need to keep */
  77         return 0;
  78 }
  79
  80 /*
  81  * Read the contents from fd for size bytes, streaming it to the
  82  * packfile in state while updating the hash in ctx. Signal a failure
  83  * by returning a negative value when the resulting pack would exceed
  84  * the pack size limit and this is not the first object in the pack,
  85  * so that the caller can discard what we wrote from the current pack
  86  * by truncating it and opening a new one. The caller will then call
  87  * us again after rewinding the input fd.
  88  *
  89  * The already_hashed_to pointer is kept untouched by the caller to
  90  * make sure we do not hash the same byte when we are called
  91  * again. This way, the caller does not have to checkpoint its hash
  92  * status before calling us just in case we ask it to call us again
  93  * with a new pack.
  94  */
  95 static int stream_to_pack(struct bulk_checkin_state *state,
  96                           git_SHA_CTX *ctx, off_t *already_hashed_to,
  97                           int fd, size_t size, enum object_type type,
  98                           const char *path, unsigned flags)
  99 {
 100         git_zstream s;
 101         unsigned char obuf[16384];
 102         unsigned hdrlen;
 103         int status = Z_OK;
 104         int write_object = (flags & HASH_WRITE_OBJECT);
 105         off_t offset = 0;
 106
 107         git_deflate_init(&s, pack_compression_level);
 108
 109         hdrlen = encode_in_pack_object_header(type, size, obuf);
 110         s.next_out = obuf + hdrlen;
 111         s.avail_out = sizeof(obuf) - hdrlen;
 112
 113         while (status != Z_STREAM_END) {
 114                 unsigned char ibuf[16384];
 115
 116                 if (size && !s.avail_in) {
 117                         ssize_t rsize = size < sizeof(ibuf) ? size : sizeof(ibuf);
 118                         if (read_in_full(fd, ibuf, rsize) != rsize)
 119                                 die("failed to read %d bytes from '%s'",
 120                                     (int)rsize, path);
 121                         offset += rsize;
 122                         if (*already_hashed_to < offset) {
 123                                 size_t hsize = offset - *already_hashed_to;
 124                                 if (rsize < hsize)
 125                                         hsize = rsize;
 126                                 if (hsize)
 127                                         git_SHA1_Update(ctx, ibuf, hsize);
 128                                 *already_hashed_to = offset;
 129                         }
 130                         s.next_in = ibuf;
 131                         s.avail_in = rsize;
 132                         size -= rsize;
 133                 }
 134
 135                 status = git_deflate(&s, size ? 0 : Z_FINISH);
 136
 137                 if (!s.avail_out || status == Z_STREAM_END) {
 138                         if (write_object) {
 139                                 size_t written = s.next_out - obuf;
 140
 141                                 /* would we bust the size limit? */
 142                                 if (state->nr_written &&
 143                                     pack_size_limit_cfg &&
 144                                     pack_size_limit_cfg < state->offset + written) {
 145                                         git_deflate_abort(&s);
 146                                         return -1;
 147                                 }
 148
 149                                 sha1write(state->f, obuf, written);
 150                                 state->offset += written;
 151                         }
 152                         s.next_out = obuf;
 153                         s.avail_out = sizeof(obuf);
 154                 }
 155
 156                 switch (status) {
 157                 case Z_OK:
 158                 case Z_BUF_ERROR:
 159                 case Z_STREAM_END:
 160                         continue;
 161                 default:
 162                         die("unexpected deflate failure: %d", status);
 163                 }
 164         }
 165         git_deflate_end(&s);
 166         return 0;
 167 }
 168
 169 /* Lazily create backing packfile for the state */
 170 static void prepare_to_stream(struct bulk_checkin_state *state,
 171                               unsigned flags)
 172 {
 173         if (!(flags & HASH_WRITE_OBJECT) || state->f)
 174                 return;
 175
 176         state->f = create_tmp_packfile(&state->pack_tmp_name);
 177         reset_pack_idx_option(&state->pack_idx_opts);
 178
 179         /* Pretend we are going to write only one object */
 180         state->offset = write_pack_header(state->f, 1);
 181         if (!state->offset)
 182                 die_errno("unable to write pack header");
 183 }
 184
 185 static int deflate_to_pack(struct bulk_checkin_state *state,
 186                            unsigned char result_sha1[],
 187                            int fd, size_t size,
 188                            enum object_type type, const char *path,
 189                            unsigned flags)
 190 {
 191         off_t seekback, already_hashed_to;
 192         git_SHA_CTX ctx;
 193         unsigned char obuf[16384];
 194         unsigned header_len;
 195         struct sha1file_checkpoint checkpoint;
 196         struct pack_idx_entry *idx = NULL;
 197
 198         seekback = lseek(fd, 0, SEEK_CUR);
 199         if (seekback == (off_t) -1)
 200                 return error("cannot find the current offset");
 201
 202         header_len = sprintf((char *)obuf, "%s %" PRIuMAX,
 203                              typename(type), (uintmax_t)size) + 1;
 204         git_SHA1_Init(&ctx);
 205         git_SHA1_Update(&ctx, obuf, header_len);
 206
 207         /* Note: idx is non-NULL when we are writing */
 208         if ((flags & HASH_WRITE_OBJECT) != 0)
 209                 idx = xcalloc(1, sizeof(*idx));
 210
 211         already_hashed_to = 0;
 212
 213         while (1) {
 214                 prepare_to_stream(state, flags);
 215                 if (idx) {
 216                         sha1file_checkpoint(state->f, &checkpoint);
 217                         idx->offset = state->offset;
 218                         crc32_begin(state->f);
 219                 }
 220                 if (!stream_to_pack(state, &ctx, &already_hashed_to,
 221                                     fd, size, type, path, flags))
 222                         break;
 223                 /*
 224                  * Writing this object to the current pack will make
 225                  * it too big; we need to truncate it, start a new
 226                  * pack, and write into it.
 227                  */
 228                 if (!idx)
 229                         die("BUG: should not happen");
 230                 sha1file_truncate(state->f, &checkpoint);
 231                 state->offset = checkpoint.offset;
 232                 finish_bulk_checkin(state);
 233                 if (lseek(fd, seekback, SEEK_SET) == (off_t) -1)
 234                         return error("cannot seek back");
 235         }
 236         git_SHA1_Final(result_sha1, &ctx);
 237         if (!idx)
 238                 return 0;
 239
 240         idx->crc32 = crc32_end(state->f);
 241         if (already_written(state, result_sha1)) {
 242                 sha1file_truncate(state->f, &checkpoint);
 243                 state->offset = checkpoint.offset;
 244                 free(idx);
 245         } else {
 246                 hashcpy(idx->sha1, result_sha1);
 247                 ALLOC_GROW(state->written,
 248                            state->nr_written + 1,
 249                            state->alloc_written);
 250                 state->written[state->nr_written++] = idx;
 251         }
 252         return 0;
 253 }
 254
 255 int index_bulk_checkin(unsigned char *sha1,
 256                        int fd, size_t size, enum object_type type,
 257                        const char *path, unsigned flags)
 258 {
 259         int status = deflate_to_pack(&state, sha1, fd, size, type,
 260                                      path, flags);
 261         if (!state.plugged)
 262                 finish_bulk_checkin(&state);
 263         return status;
 264 }
 265
 266 void plug_bulk_checkin(void)
 267 {
 268         state.plugged = 1;
 269 }
 270
 271 void unplug_bulk_checkin(void)
 272 {
 273         state.plugged = 0;
 274         if (state.f)
 275                 finish_bulk_checkin(&state);
 276 }