@matheusbsbs / matheus-bagliano / commits / 7481d2bfca

sequencer: truncate labels to accommodate loose refs

Some commits may have unusually long subject lines. When those subject lines are used as labels in the `--rebase-merges` mode of `git rebase`, they can cause errors when writing the corresponding loose refs because most file systems have a maximal file name length of 255 (`NAME_MAX`). The symptom looks like this: $ git rebase --continue error: cannot lock ref 'refs/rewritten/SANITIZED-SUBJECT': Unable to create '.git/refs/rewritten/SANITIZED-SUBJECT.lock': File name too long - where SANITIZED-SUBJECT is very long Let's accommodate this situation by truncating the labels. Care must be taken in case the subject line contains multi-byte characters so as not to truncate in the middle of a character. Signed-off-by: Mark Ruvald Pedersen <mped@demant.com> Signed-off-by: Johannes Schindelin <Johannes.Schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>

Mark Ruvald Pedersen committed Aug 10, 2023 at 16:34 UTC 7481d2bfca7fd36f63fd437508be2bca338c9477
2 files changed +40 -5
git-compat-util.h
+4
@@ -422,6 +422,10 @@ char *gitdirname(char *);
422 #define PATH_MAX 4096
423 #endif
424
425 +#ifndef NAME_MAX
426 +#define NAME_MAX 255
427 +#endif
428 +
429 typedef uintmax_t timestamp_t;
430 #define PRItime PRIuMAX
431 #define parse_timestamp strtoumax
sequencer.c
+36 -5
@@ -50,6 +50,15 @@
50
51 #define GIT_REFLOG_ACTION "GIT_REFLOG_ACTION"
52
53 +/*
54 + * To accommodate common filesystem limitations, where the loose refs' file
55 + * names must not exceed `NAME_MAX`, the labels generated by `git rebase
56 + * --rebase-merges` need to be truncated if the corresponding commit subjects
57 + * are too long.
58 + * Add some margin to stay clear from reaching `NAME_MAX`.
59 + */
60 +#define GIT_MAX_LABEL_LENGTH ((NAME_MAX) - (LOCK_SUFFIX_LEN) - 16)
61 +
62 static const char sign_off_header[] = "Signed-off-by: ";
63 static const char cherry_picked_prefix[] = "(cherry picked from commit ";
64
@@ -5386,6 +5395,8 @@ static const char *label_oid(struct object_id *oid, const char *label,
5395 }
5396 } else {
5397 struct strbuf *buf = &state->buf;
5398 + int label_is_utf8 = 1; /* start with this assumption */
5399 + size_t max_len = buf->len + GIT_MAX_LABEL_LENGTH;
5400
5401 /*
5402 * Sanitize labels by replacing non-alpha-numeric characters
@@ -5394,14 +5405,34 @@ static const char *label_oid(struct object_id *oid, const char *label,
5405 *
5406 * Note that we retain non-ASCII UTF-8 characters (identified
5407 * via the most significant bit). They should be all acceptable
5397 - * in file names. We do not validate the UTF-8 here, that's not
5398 - * the job of this function.
5408 + * in file names.
5409 + *
5410 + * As we will use the labels as names of (loose) refs, it is
5411 + * vital that the name not be longer than the maximum component
5412 + * size of the file system (`NAME_MAX`). We are careful to
5413 + * truncate the label accordingly, allowing for the `.lock`
5414 + * suffix and for the label to be UTF-8 encoded (i.e. we avoid
5415 + * truncating in the middle of a character).
5416 */
5400 - for (; *label; label++)
5401 - if ((*label & 0x80) || isalnum(*label))
5417 + for (; *label && buf->len + 1 < max_len; label++)
5418 + if (isalnum(*label) ||
5419 + (!label_is_utf8 && (*label & 0x80)))
5420 strbuf_addch(buf, *label);
5421 + else if (*label & 0x80) {
5422 + const char *p = label;
5423 +
5424 + utf8_width(&p, NULL);
5425 + if (p) {
5426 + if (buf->len + (p - label) > max_len)
5427 + break;
5428 + strbuf_add(buf, label, p - label);
5429 + label = p - 1;
5430 + } else {
5431 + label_is_utf8 = 0;
5432 + strbuf_addch(buf, *label);
5433 + }
5434 /* avoid leading dash and double-dashes */
5404 - else if (buf->len && buf->buf[buf->len - 1] != '-')
5435 + } else if (buf->len && buf->buf[buf->len - 1] != '-')
5436 strbuf_addch(buf, '-');
5437 if (!buf->len) {
5438 strbuf_addstr(buf, "rev-");