src/pack.h

   1 /*
   2  * Copyright (C) the libgit2 contributors. All rights reserved.
   3  *
   4  * This file is part of libgit2, distributed under the GNU GPL v2 with
   5  * a Linking Exception. For full terms see the included COPYING file.
   6  */
   7
   8 #ifndef INCLUDE_pack_h__
   9 #define INCLUDE_pack_h__
  10
  11 #include "common.h"
  12
  13 #include "git2/oid.h"
  14
  15 #include "array.h"
  16 #include "map.h"
  17 #include "mwindow.h"
  18 #include "odb.h"
  19 #include "offmap.h"
  20 #include "oidmap.h"
  21 #include "zstream.h"
  22
  23 /**
  24  * Function type for callbacks from git_pack_foreach_entry_offset.
  25  */
  26 typedef int git_pack_foreach_entry_offset_cb(
  27                 const git_oid *id,
  28                 off64_t offset,
  29                 void *payload);
  30
  31 #define GIT_PACK_FILE_MODE 0444
  32
  33 #define PACK_SIGNATURE 0x5041434b       /* "PACK" */
  34 #define PACK_VERSION 2
  35 #define pack_version_ok(v) ((v) == htonl(2) || (v) == htonl(3))
  36 struct git_pack_header {
  37         uint32_t hdr_signature;
  38         uint32_t hdr_version;
  39         uint32_t hdr_entries;
  40 };
  41
  42 /*
  43  * The first four bytes of index formats later than version 1 should
  44  * start with this signature, as all older git binaries would find this
  45  * value illegal and abort reading the file.
  46  *
  47  * This is the case because the number of objects in a packfile
  48  * cannot exceed 1,431,660,000 as every object would need at least
  49  * 3 bytes of data and the overall packfile cannot exceed 4 GiB with
  50  * version 1 of the index file due to the offsets limited to 32 bits.
  51  * Clearly the signature exceeds this maximum.
  52  *
  53  * Very old git binaries will also compare the first 4 bytes to the
  54  * next 4 bytes in the index and abort with a "non-monotonic index"
  55  * error if the second 4 byte word is smaller than the first 4
  56  * byte word. This would be true in the proposed future index
  57  * format as idx_signature would be greater than idx_version.
  58  */
  59
  60 #define PACK_IDX_SIGNATURE 0xff744f63   /* "\377tOc" */
  61
  62 struct git_pack_idx_header {
  63         uint32_t idx_signature;
  64         uint32_t idx_version;
  65 };
  66
  67 typedef struct git_pack_cache_entry {
  68         size_t last_usage; /* enough? */
  69         git_atomic32 refcount;
  70         git_rawobj raw;
  71 } git_pack_cache_entry;
  72
  73 struct pack_chain_elem {
  74         off64_t base_key;
  75         off64_t offset;
  76         size_t size;
  77         git_object_t type;
  78 };
  79
  80 typedef git_array_t(struct pack_chain_elem) git_dependency_chain;
  81
  82 #define GIT_PACK_CACHE_MEMORY_LIMIT 16 * 1024 * 1024
  83 #define GIT_PACK_CACHE_SIZE_LIMIT 1024 * 1024 /* don't bother caching anything over 1MB */
  84
  85 typedef struct {
  86         size_t memory_used;
  87         size_t memory_limit;
  88         size_t use_ctr;
  89         git_mutex lock;
  90         git_offmap *entries;
  91 } git_pack_cache;
  92
  93 struct git_pack_file {
  94         git_mwindow_file mwf;
  95         git_map index_map;
  96         git_mutex lock; /* protect updates to index_map */
  97         git_atomic32 refcount;
  98
  99         uint32_t num_objects;
 100         uint32_t num_bad_objects;
 101         git_oid *bad_object_sha1; /* array of git_oid */
 102
 103         int index_version;
 104         git_time_t mtime;
 105         unsigned pack_local:1, pack_keep:1, has_cache:1;
 106         git_oidmap *idx_cache;
 107         git_oid **oids;
 108
 109         git_pack_cache bases; /* delta base cache */
 110
 111         time_t last_freshen; /* last time the packfile was freshened */
 112
 113         /* something like ".git/objects/pack/xxxxx.pack" */
 114         char pack_name[GIT_FLEX_ARRAY]; /* more */
 115 };
 116
 117 /**
 118  * Return the position where an OID (or a prefix) would be inserted within the
 119  * OID Lookup Table of an .idx file. This performs binary search between the lo
 120  * and hi indices.
 121  *
 122  * The stride parameter is provided because .idx files version 1 store the OIDs
 123  * interleaved with the 4-byte file offsets of the objects within the .pack
 124  * file (stride = 24), whereas files with version 2 store them in a contiguous
 125  * flat array (stride = 20).
 126  */
 127 int git_pack__lookup_sha1(const void *oid_lookup_table, size_t stride, unsigned lo,
 128                 unsigned hi, const unsigned char *oid_prefix);
 129
 130 struct git_pack_entry {
 131         off64_t offset;
 132         git_oid sha1;
 133         struct git_pack_file *p;
 134 };
 135
 136 typedef struct git_packfile_stream {
 137         off64_t curpos;
 138         int done;
 139         git_zstream zstream;
 140         struct git_pack_file *p;
 141         git_mwindow *mw;
 142 } git_packfile_stream;
 143
 144 int git_packfile__object_header(size_t *out, unsigned char *hdr, size_t size, git_object_t type);
 145
 146 int git_packfile__name(char **out, const char *path);
 147
 148 int git_packfile_unpack_header(
 149                 size_t *size_p,
 150                 git_object_t *type_p,
 151                 struct git_pack_file *p,
 152                 git_mwindow **w_curs,
 153                 off64_t *curpos);
 154
 155 int git_packfile_resolve_header(
 156                 size_t *size_p,
 157                 git_object_t *type_p,
 158                 struct git_pack_file *p,
 159                 off64_t offset);
 160
 161 int git_packfile_unpack(git_rawobj *obj, struct git_pack_file *p, off64_t *obj_offset);
 162
 163 int git_packfile_stream_open(git_packfile_stream *obj, struct git_pack_file *p, off64_t curpos);
 164 ssize_t git_packfile_stream_read(git_packfile_stream *obj, void *buffer, size_t len);
 165 void git_packfile_stream_dispose(git_packfile_stream *obj);
 166
 167 int get_delta_base(
 168                 off64_t *delta_base_out,
 169                 struct git_pack_file *p,
 170                 git_mwindow **w_curs,
 171                 off64_t *curpos,
 172                 git_object_t type,
 173                 off64_t delta_obj_offset);
 174
 175 void git_packfile_free(struct git_pack_file *p, bool unlink_packfile);
 176 int git_packfile_alloc(struct git_pack_file **pack_out, const char *path);
 177
 178 int git_pack_entry_find(
 179                 struct git_pack_entry *e,
 180                 struct git_pack_file *p,
 181                 const git_oid *short_oid,
 182                 size_t len);
 183 int git_pack_foreach_entry(
 184                 struct git_pack_file *p,
 185                 git_odb_foreach_cb cb,
 186                 void *data);
 187 /**
 188  * Similar to git_pack_foreach_entry, but:
 189  * - It also provides the offset of the object within the
 190  *   packfile.
 191  * - It does not sort the objects in any order.
 192  * - It retains the lock while invoking the callback.
 193  */
 194 int git_pack_foreach_entry_offset(
 195                 struct git_pack_file *p,
 196                 git_pack_foreach_entry_offset_cb cb,
 197                 void *data);
 198
 199 #endif