apply: detect overflow when parsing hunk header

Commit Message

Phillip Wood Jan. 30, 2025, 11:08 a.m. UTC

From: Phillip Wood <[e-mail]>

"git apply" uses strtoul() to parse the numbers in the hunk header but
silently ignores overflows. As LONG_MAX is a legitimate return value for
strtoul() we need to set errno to zero before the call to strtoul() and
check that it is still zero afterwards. The error message we display is
not particularly helpful as it does not say what was wrong.  However, it
seems pretty unlikely that users are going to trigger this error in
practice and we can always improve it later if needed.

Signed-off-by: Phillip Wood <[e-mail]>
---
    apply: detect overflow when parsing hunk header
    
    We should do something similar in "git add -p" but I'll wait to see what
    happens with
    https://lore.kernel.org/git/[e-mail]/
    first

Published-As: https://github.com/gitgitgadget/git/releases/tag/pr-1858%2Fphillipwood%2Fapply-detect-hunk-header-overflow-v1
Fetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-1858/phillipwood/apply-detect-hunk-header-overflow-v1
Pull-Request: https://github.com/gitgitgadget/git/pull/1858

 apply.c               |  3 +++
 t/t4100-apply-stat.sh | 13 +++++++++++++
 2 files changed, 16 insertions(+)


base-commit: fbe8d3079d4a96aeb4e4529cc93cc0043b759a05

Comments

Junio C Hamano Jan. 30, 2025, 10:17 p.m. UTC | #1

"Phillip Wood via GitGitGadget" <[e-mail]> writes:

> From: Phillip Wood <[e-mail]>
>
> "git apply" uses strtoul() to parse the numbers in the hunk header but
> silently ignores overflows. As LONG_MAX is a legitimate return value for
> strtoul() we need to set errno to zero before the call to strtoul() and
> check that it is still zero afterwards. The error message we display is
> not particularly helpful as it does not say what was wrong.  However, it
> seems pretty unlikely that users are going to trigger this error in
> practice and we can always improve it later if needed.

Thanks.  We made an effort to use a type that is a bit wider than
"int", but we apparently ignored that the Git userbase will become a
lot wider some day and unfriendly and/or hostile folks would start
feeding malicious input to us X-<.  The check presented here look
good, and the fact that there was only one change needed shows how
well designed the base code was ;-)

Will queue.  Thanks.

> @@ -1423,7 +1423,10 @@ static int parse_num(const char *line, unsigned long *p)
>  
>  	if (!isdigit(*line))
>  		return 0;
> +	errno = 0;
>  	*p = strtoul(line, &ptr, 10);
> +	if (errno)
> +		return 0;
>  	return ptr - line;
>  }
>  
> diff --git a/t/t4100-apply-stat.sh b/t/t4100-apply-stat.sh
> index 146e73d8f55..a5664f3eb3c 100755
> --- a/t/t4100-apply-stat.sh
> +++ b/t/t4100-apply-stat.sh
> @@ -38,4 +38,17 @@ incomplete (1)
>  incomplete (2)
>  EOF
>  
> +test_expect_success 'applying a hunk header which overflows fails' '
> +	cat >patch <<-\EOF &&
> +	diff -u a/file b/file
> +	--- a/file
> +	+++ b/file
> +	@@ -98765432109876543210 +98765432109876543210 @@
> +	-a
> +	+b
> +	EOF
> +	test_must_fail git apply patch 2>err &&
> +	echo "error: corrupt patch at line 4" >expect &&
> +	test_cmp expect err
> +'
>  test_done
>
> base-commit: fbe8d3079d4a96aeb4e4529cc93cc0043b759a05

Patch

diff --git a/apply.c b/apply.c
index 4a7b6120ac8..b124678b93b 100644
--- a/apply.c
+++ b/apply.c
@@ -1423,7 +1423,10 @@  static int parse_num(const char *line, unsigned long *p)
 
 	if (!isdigit(*line))
 		return 0;
+	errno = 0;
 	*p = strtoul(line, &ptr, 10);
+	if (errno)
+		return 0;
 	return ptr - line;
 }
 
diff --git a/t/t4100-apply-stat.sh b/t/t4100-apply-stat.sh
index 146e73d8f55..a5664f3eb3c 100755
--- a/t/t4100-apply-stat.sh
+++ b/t/t4100-apply-stat.sh
@@ -38,4 +38,17 @@  incomplete (1)
 incomplete (2)
 EOF
 
+test_expect_success 'applying a hunk header which overflows fails' '
+	cat >patch <<-\EOF &&
+	diff -u a/file b/file
+	--- a/file
+	+++ b/file
+	@@ -98765432109876543210 +98765432109876543210 @@
+	-a
+	+b
+	EOF
+	test_must_fail git apply patch 2>err &&
+	echo "error: corrupt patch at line 4" >expect &&
+	test_cmp expect err
+'
 test_done