Brad Bishop | 004d499 | 2018-10-02 23:54:45 +0200 | [diff] [blame] | 1 | From 014281e62b7920a6d710a85089e00ca012b0744c Mon Sep 17 00:00:00 2001 |
| 2 | From: Jeff King <peff@peff.net> |
| 3 | Date: Sun, 13 May 2018 12:09:42 -0400 |
| 4 | Subject: [PATCH] is_ntfs_dotgit: use a size_t for traversing string |
| 5 | |
| 6 | We walk through the "name" string using an int, which can |
| 7 | wrap to a negative value and cause us to read random memory |
| 8 | before our array (e.g., by creating a tree with a name >2GB, |
| 9 | since "int" is still 32 bits even on most 64-bit platforms). |
| 10 | Worse, this is easy to trigger during the fsck_tree() check, |
| 11 | which is supposed to be protecting us from malicious |
| 12 | garbage. |
| 13 | |
| 14 | Note one bit of trickiness in the existing code: we |
| 15 | sometimes assign -1 to "len" at the end of the loop, and |
| 16 | then rely on the "len++" in the for-loop's increment to take |
| 17 | it back to 0. This is still legal with a size_t, since |
| 18 | assigning -1 will turn into SIZE_MAX, which then wraps |
| 19 | around to 0 on increment. |
| 20 | |
| 21 | Signed-off-by: Jeff King <peff@peff.net> |
| 22 | CVE: CVE-2018-11233 |
| 23 | Upstream-Status: Backport[https://github.com/git/git/commit/11a9f4d807a0d71dc6eff51bb87baf4ca2cccf1d] |
| 24 | Signed-off-by: Sinan Kaya <okaya@kernel.org> |
| 25 | --- |
| 26 | path.c | 2 +- |
| 27 | 1 file changed, 1 insertion(+), 1 deletion(-) |
| 28 | |
| 29 | diff --git a/path.c b/path.c |
| 30 | index da8b65573..d31c795ff 100644 |
| 31 | --- a/path.c |
| 32 | +++ b/path.c |
| 33 | @@ -1305,7 +1305,7 @@ static int only_spaces_and_periods(const char *path, size_t len, size_t skip) |
| 34 | |
| 35 | int is_ntfs_dotgit(const char *name) |
| 36 | { |
| 37 | - int len; |
| 38 | + size_t len; |
| 39 | |
| 40 | for (len = 0; ; len++) |
| 41 | if (!name[len] || name[len] == '\\' || is_dir_sep(name[len])) { |
| 42 | -- |
| 43 | 2.19.0 |
| 44 | |