[PATCH 8/9] fsck: avoid parse_timestamp() on buffer that isn't NUL-terminated

From: Jeff King <peff@peff.net>
Date: Nov 12, 2025, 08:10 UTC
Message-ID: <20251112081040.GH979063@coredump.intra.peff.net>
In-Reply-To: <20251112075522.GA978866@coredump.intra.peff.net>

In fsck_ident(), we parse the timestamp with parse_timestamp(), which is really an alias for strtoumax(). But since our buffer may not be NUL-terminated, this can trigger a complaint from ASan's strict_string_checks mode. This is a false positive, since we know that the buffer contains a trailing newline (which we checked earlier in the function), and that strtoumax() would stop there.

But it is worth working around ASan's complaint. One is because that will let us turn on strict_string_checks by default, which has helped catch other real problems. And two is that the safety of the current code is very hard to reason about (it subtly depends on distant code which could change).

One option here is to just parse the number left-to-right ourselves. But we care about the size of a timestamp_t and detecting overflow, since that's part of the point of these checks. And doing that correctly is tricky. So we'll instead just pull the digits into a separate, NUL-terminated buffer, and use that to call parse_timestamp().

Signed-off-by: Jeff King <peff@peff.net>
---
There's one step we could take after this commit, which is to annotate
all of the spots that look at "p" to do:

  -if (*p != ' ')
  +if (p >= ident_end || *p != ' ')
	return report(..., "expected space");

or similar. And then I think it would be safe to call fsck_ident() with a buffer that is not NUL-terminated (and does not have our "safety" newline). I stopped short for this series because we are just trying to appease ASan (and not fixing real bugs), and because the result gets rather unwieldy. But it might be worth it in the long run. We could always do it later on top.

Again, I find the strto*() wrapping to be gross. Here we use the extra-buffer trick. But we still don't get away with avoiding custom logic (for example, if we ever want to support negative timestamps, the fsck code will have to recognize "-" signs). But it feels like the best we can do for now.

 fsck.c | 23 +++++++++++++++++++----
 1 file changed, 19 insertions(+), 4 deletions(-)

diff --git a/fsck.c b/fsck.c
index 266c965cec..8e8083e7c6 100644
--- a/fsck.c
+++ b/fsck.c
@@ -860,13 +860,28 @@ static int verify_headers(const void *data, unsigned long size,
 		FSCK_MSG_UNTERMINATED_HEADER, "unterminated header");
 }
 
+static timestamp_t parse_timestamp_from_buf(const char **start, const char *end)
+{
+	const char *p = *start;
+	char buf[24]; /* big enough for 2^64 */
+	size_t i = 0;
+
+	while (p < end && isdigit(*p)) {
+		if (i >= ARRAY_SIZE(buf) - 1)
+			return TIME_MAX;
+		buf[i++] = *p++;
+	}
+	buf[i] = '\0';
+	*start = p;
+	return parse_timestamp(buf, NULL, 10);
+}
+
 static int fsck_ident(const char **ident, const char *ident_end,
 		      const struct object_id *oid, enum object_type type,
 		      struct fsck_options *options)
 {
 	const char *p = *ident;
 	const char *nl;
-	char *end;
 
 	nl = memchr(p, '\n', ident_end - p);
 	if (!nl)
@@ -918,11 +933,11 @@ static int fsck_ident(const char **ident, const char *ident_end,
 			      "invalid author/committer line - bad date");
 	if (*p == '0' && p[1] != ' ')
 		return report(options, oid, type, FSCK_MSG_ZERO_PADDED_DATE, "invalid author/committer line - zero-padded date");
-	if (date_overflows(parse_timestamp(p, &end, 10)))
+	if (date_overflows(parse_timestamp_from_buf(&p, ident_end)))
 		return report(options, oid, type, FSCK_MSG_BAD_DATE_OVERFLOW, "invalid author/committer line - date causes integer overflow");
-	if (*end != ' ')
+	if (*p != ' ')
 		return report(options, oid, type, FSCK_MSG_BAD_DATE, "invalid author/committer line - bad date");
-	p = end + 1;
+	p++;
 	if ((*p != '+' && *p != '-') ||
 	    !isdigit(p[1]) ||
 	    !isdigit(p[2]) ||

-- 
2.52.0.rc1.260.g3e4993586f