@matheusbsbs / matheus-bagliano / commits / 4b3511b0f8
Fork 0

ce_smudge_racily_clean_entry: explain why it works.

This is a tricky code and warrants extra commenting. I wasted 30 minutes trying to break it until I realized why it works. Signed-off-by: Junio C Hamano <junkio@cox.net>

Junio C Hamano committed Dec 20, 2005 at 14:18 UTC 4b3511b0f8422fd2c5b1b37c9655ae2ce904bca5
1 file changed +25 -1
read-cache.c
+25 -1
@@ -616,7 +616,31 @@ static void ce_smudge_racily_clean_entry(struct cache_entry *ce)
616 if (ce_match_stat_basic(ce, &st))
617 return;
618 if (ce_modified_check_fs(ce, &st)) {
619 - /* This is "racily clean"; smudge it */
619 + /* This is "racily clean"; smudge it. Note that this
620 + * is a tricky code. At first glance, it may appear
621 + * that it can break with this sequence:
622 + *
623 + * $ echo xyzzy >frotz
624 + * $ git-update-index --add frotz
625 + * $ : >frotz
626 + * $ sleep 3
627 + * $ echo filfre >nitfol
628 + * $ git-update-index --add nitfol
629 + *
630 + * but it does not. Whe the second update-index runs,
631 + * it notices that the entry "frotz" has the same timestamp
632 + * as index, and if we were to smudge it by resetting its
633 + * size to zero here, then the object name recorded
634 + * in index is the 6-byte file but the cached stat information
635 + * becomes zero --- which would then match what we would
636 + * obtain from the filesystem next time we stat("frotz").
637 + *
638 + * However, the second update-index, before calling
639 + * this function, notices that the cached size is 6
640 + * bytes and what is on the filesystem is an empty
641 + * file, and never calls us, so the cached size information
642 + * for "frotz" stays 6 which does not match the filesystem.
643 + */
644 ce->ce_size = htonl(0);
645 }
646 }