environment: move access to "core.bigFileThreshold" into repo settings
The "core.bigFileThreshold" setting is stored in a global variable and populated via `git_default_core_config()`. This may cause issues in the case where one is handling multiple different repositories in a single process with different values for that config key, as we may or may not see the correct value in that case. Furthermore, global state blocks our path towards libification. Refactor the code so that we instead store the value in `struct repo_settings`, where the value is computed as-needed and cached. Note that this change requires us to adapt one test in t1050 that verifies that we die when parsing an invalid "core.bigFileThreshold" value. The exercised Git command doesn't use the value at all, and thus it won't hit the new code path that parses the value. This is addressed by using git-hash-object(1) instead, which does read the value. Signed-off-by: Patrick Steinhardt <ps@pks.im> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Patrick Steinhardt committed
Mar 10, 2025 at 08:13 UTC
7835ee75cdffbce925246cbacc83e8b4a932a681
15 files changed
+52
-22
archive.c
+1
-1
@@ -216,7 +216,7 @@ static int write_archive_entry(const struct object_id *oid, const char *base,
216
/* Stream it? */
217
if (S_ISREG(mode) && !args->convert &&
218
oid_object_info(args->repo, oid, &size) == OBJ_BLOB &&
219
- size > big_file_threshold)
219
+ size > repo_settings_get_big_file_threshold(the_repository))
220
return write_entry(args, oid, path.buf, path.len, mode, NULL, size);
221
222
buffer = object_file_to_archive(args, path.buf, oid, mode, &type, &size);
builtin/fast-import.c
+2
-2
@@ -2021,7 +2021,7 @@ static void parse_and_store_blob(
2021
static struct strbuf buf = STRBUF_INIT;
2022
uintmax_t len;
2023
2024
- if (parse_data(&buf, big_file_threshold, &len))
2024
+ if (parse_data(&buf, repo_settings_get_big_file_threshold(the_repository), &len))
2025
store_object(OBJ_BLOB, &buf, last, oidout, mark);
2026
else {
2027
if (last) {
@@ -3402,7 +3402,7 @@ static int parse_one_option(const char *option)
3402
unsigned long v;
3403
if (!git_parse_ulong(option, &v))
3404
return 0;
3405
- big_file_threshold = v;
3405
+ repo_settings_set_big_file_threshold(the_repository, v);
3406
} else if (skip_prefix(option, "depth=", &option)) {
3407
option_depth(option);
3408
} else if (skip_prefix(option, "active-branches=", &option)) {
builtin/index-pack.c
+4
-2
@@ -485,7 +485,8 @@ static void *unpack_entry_data(off_t offset, unsigned long size,
485
git_hash_update(&c, hdr, hdrlen);
486
} else
487
oid = NULL;
488
- if (type == OBJ_BLOB && size > big_file_threshold)
488
+ if (type == OBJ_BLOB &&
489
+ size > repo_settings_get_big_file_threshold(the_repository))
490
buf = fixed_buf;
491
else
492
buf = xmallocz(size);
@@ -799,7 +800,8 @@ static int check_collison(struct object_entry *entry)
800
enum object_type type;
801
unsigned long size;
802
802
- if (entry->size <= big_file_threshold || entry->type != OBJ_BLOB)
803
+ if (entry->size <= repo_settings_get_big_file_threshold(the_repository) ||
804
+ entry->type != OBJ_BLOB)
805
return -1;
806
807
memset(&data, 0, sizeof(data));
builtin/pack-objects.c
+4
-2
@@ -499,7 +499,8 @@ static unsigned long write_no_reuse_object(struct hashfile *f, struct object_ent
499
500
if (!usable_delta) {
501
if (oe_type(entry) == OBJ_BLOB &&
502
- oe_size_greater_than(&to_pack, entry, big_file_threshold) &&
502
+ oe_size_greater_than(&to_pack, entry,
503
+ repo_settings_get_big_file_threshold(the_repository)) &&
504
(st = open_istream(the_repository, &entry->idx.oid, &type,
505
&size, NULL)) != NULL)
506
buf = NULL;
@@ -2454,7 +2455,8 @@ static void get_object_details(void)
2455
struct object_entry *entry = sorted_by_offset[i];
2456
check_object(entry, i);
2457
if (entry->type_valid &&
2457
- oe_size_greater_than(&to_pack, entry, big_file_threshold))
2458
+ oe_size_greater_than(&to_pack, entry,
2459
+ repo_settings_get_big_file_threshold(the_repository)))
2460
entry->no_try_delta = 1;
2461
display_progress(progress_state, i + 1);
2462
}
builtin/unpack-objects.c
+2
-1
@@ -553,7 +553,8 @@ static void unpack_one(unsigned nr)
553
554
switch (type) {
555
case OBJ_BLOB:
556
- if (!dry_run && size > big_file_threshold) {
556
+ if (!dry_run &&
557
+ size > repo_settings_get_big_file_threshold(the_repository)) {
558
stream_blob(size, nr);
559
return;
560
}
config.c
-5
@@ -1490,11 +1490,6 @@ static int git_default_core_config(const char *var, const char *value,
1490
return 0;
1491
}
1492
1493
- if (!strcmp(var, "core.bigfilethreshold")) {
1494
- big_file_threshold = git_config_ulong(var, value, ctx->kvi);
1495
- return 0;
1496
- }
1497
-
1493
if (!strcmp(var, "core.autocrlf")) {
1494
if (value && !strcasecmp(value, "input")) {
1495
auto_crlf = AUTO_CRLF_INPUT;
diff.c
+4
-2
@@ -4193,7 +4193,8 @@ int diff_populate_filespec(struct repository *r,
4193
* is probably fine.
4194
*/
4195
if (check_binary &&
4196
- s->size > big_file_threshold && s->is_binary == -1) {
4196
+ s->size > repo_settings_get_big_file_threshold(the_repository) &&
4197
+ s->is_binary == -1) {
4198
s->is_binary = 1;
4199
return 0;
4200
}
@@ -4243,7 +4244,8 @@ object_read:
4244
if (size_only || check_binary) {
4245
if (size_only)
4246
return 0;
4246
- if (s->size > big_file_threshold && s->is_binary == -1) {
4247
+ if (s->size > repo_settings_get_big_file_threshold(the_repository) &&
4248
+ s->is_binary == -1) {
4249
s->is_binary = 1;
4250
return 0;
4251
}
environment.c
-1
@@ -49,7 +49,6 @@ int fsync_object_files = -1;
49
int use_fsync = -1;
50
enum fsync_method fsync_method = FSYNC_METHOD_DEFAULT;
51
enum fsync_component fsync_components = FSYNC_COMPONENTS_DEFAULT;
52
-unsigned long big_file_threshold = 512 * 1024 * 1024;
52
char *editor_program;
53
char *askpass_program;
54
char *excludes_file;
environment.h
-1
@@ -154,7 +154,6 @@ extern int zlib_compression_level;
154
extern int pack_compression_level;
155
extern size_t packed_git_window_size;
156
extern size_t packed_git_limit;
157
-extern unsigned long big_file_threshold;
157
extern unsigned long pack_size_limit_cfg;
158
extern int max_allowed_tree_depth;
159
object-file.c
+4
-2
@@ -2803,7 +2803,8 @@ int index_fd(struct index_state *istate, struct object_id *oid,
2803
ret = index_stream_convert_blob(istate, oid, fd, path, flags);
2804
else if (!S_ISREG(st->st_mode))
2805
ret = index_pipe(istate, oid, fd, type, path, flags);
2806
- else if (st->st_size <= big_file_threshold || type != OBJ_BLOB ||
2806
+ else if (st->st_size <= repo_settings_get_big_file_threshold(the_repository) ||
2807
+ type != OBJ_BLOB ||
2808
(path && would_convert_to_git(istate, path)))
2809
ret = index_core(istate, oid, fd, xsize_t(st->st_size),
2810
type, path, flags);
@@ -3137,7 +3138,8 @@ int read_loose_object(const char *path,
3138
goto out;
3139
}
3140
3140
- if (*oi->typep == OBJ_BLOB && *size > big_file_threshold) {
3141
+ if (*oi->typep == OBJ_BLOB &&
3142
+ *size > repo_settings_get_big_file_threshold(the_repository)) {
3143
if (check_stream_oid(&stream, hdr, *size, path, expected_oid) < 0)
3144
goto out;
3145
} else {
pack-check.c
+2
-1
@@ -131,7 +131,8 @@ static int verify_packfile(struct repository *r,
131
type = unpack_object_header(p, w_curs, &curpos, &size);
132
unuse_pack(w_curs);
133
134
- if (type == OBJ_BLOB && big_file_threshold <= size) {
134
+ if (type == OBJ_BLOB &&
135
+ repo_settings_get_big_file_threshold(the_repository) <= size) {
136
/*
137
* Let stream_object_signature() check it with
138
* the streaming interface; no point slurping
repo-settings.c
+20
@@ -20,6 +20,13 @@ static void repo_cfg_int(struct repository *r, const char *key, int *dest,
20
*dest = def;
21
}
22
23
+static void repo_cfg_ulong(struct repository *r, const char *key, unsigned long *dest,
24
+ unsigned long def)
25
+{
26
+ if (repo_config_get_ulong(r, key, dest))
27
+ *dest = def;
28
+}
29
+
30
void prepare_repo_settings(struct repository *r)
31
{
32
int experimental;
@@ -151,6 +158,19 @@ void repo_settings_clear(struct repository *r)
158
r->settings = empty;
159
}
160
161
+unsigned long repo_settings_get_big_file_threshold(struct repository *repo)
162
+{
163
+ if (!repo->settings.big_file_threshold)
164
+ repo_cfg_ulong(repo, "core.bigfilethreshold",
165
+ &repo->settings.big_file_threshold, 512 * 1024 * 1024);
166
+ return repo->settings.big_file_threshold;
167
+}
168
+
169
+void repo_settings_set_big_file_threshold(struct repository *repo, unsigned long value)
170
+{
171
+ repo->settings.big_file_threshold = value;
172
+}
173
+
174
enum log_refs_config repo_settings_get_log_all_ref_updates(struct repository *repo)
175
{
176
const char *value;
repo-settings.h
+5
@@ -64,6 +64,7 @@ struct repo_settings {
64
size_t delta_base_cache_limit;
65
size_t packed_git_window_size;
66
size_t packed_git_limit;
67
+ unsigned long big_file_threshold;
68
69
char *hooks_path;
70
};
@@ -88,6 +89,10 @@ int repo_settings_get_warn_ambiguous_refs(struct repository *repo);
89
/* Read the value for "core.hooksPath". */
90
const char *repo_settings_get_hooks_path(struct repository *repo);
91
92
+/* Read and set the value for "core.bigFileThreshold". */
93
+unsigned long repo_settings_get_big_file_threshold(struct repository *repo);
94
+void repo_settings_set_big_file_threshold(struct repository *repo, unsigned long value);
95
+
96
/* Read, set or reset the value for "core.sharedRepository". */
97
int repo_settings_get_shared_repository(struct repository *repo);
98
void repo_settings_set_shared_repository(struct repository *repo, int value);
streaming.c
+2
-1
@@ -431,7 +431,8 @@ static int istream_source(struct git_istream *st,
431
st->open = open_istream_loose;
432
return 0;
433
case OI_PACKED:
434
- if (!oi.u.packed.is_delta && big_file_threshold < size) {
434
+ if (!oi.u.packed.is_delta &&
435
+ repo_settings_get_big_file_threshold(the_repository) < size) {
436
st->u.in_pack.pack = oi.u.packed.pack;
437
st->u.in_pack.pos = oi.u.packed.offset;
438
st->open = open_istream_pack_non_delta;
t/t1050-large.sh
+2
-1
@@ -6,7 +6,8 @@ test_description='adding and checking out large blobs'
6
. ./test-lib.sh
7
8
test_expect_success 'core.bigFileThreshold must be non-negative' '
9
- test_must_fail git -c core.bigFileThreshold=-1 rev-parse >out 2>err &&
9
+ : >input &&
10
+ test_must_fail git -c core.bigFileThreshold=-1 hash-object input >out 2>err &&
11
grep "bad numeric config value" err &&
12
test_must_be_empty out
13
'