From: Willy Tarreau <w@1wt.eu>
Date: Wed, 4 Nov 2026 09:58:53 +0000 (+0100)
Subject: MINOR: http: speed up http_header_has_forbidden_char() to make it work alone
X-Git-Tag: v3.5-dev11~11
X-Git-Url: https://git.kaiwu.me/?a=commitdiff_plain;h=de80a2699cddb10a475d9c90c0ad1225f83bd556;p=haproxy.git

MINOR: http: speed up http_header_has_forbidden_char() to make it work alone

The function reads one char at a time and is systematically hidden behind
a first call to ist_find_http_ctl() to speed up the lookup of control chars.
Unfortunately, this construct doesn't permit to find other control chars
such as 0x7F. Better make http_header_has_forbidden_char() self-sustaining,
but for this we need to implement the speedup there. For this, we start
with a loop using is_char8_outside(0x20,0x7e) that allows to fly over most
non-offending bytes (SP, VCHAR). Only HT (9) and obs-text (0x80..0xFF) will
exit early, which is not a problem given that they are extremely rare in
header values. After that we continue the loop to search for offending
bytes starting from the block of 4 or 8 that triggered.
---

diff --git a/include/haproxy/http.h b/include/haproxy/http.h
index 2f70a7719..4c269c477 100644
--- a/include/haproxy/http.h
+++ b/include/haproxy/http.h
@@ -220,18 +220,34 @@ static inline struct http_uri_parser http_uri_parser_init(const struct ist uri)
 
 /* Looks into <ist> for forbidden characters for header values (0x00, 0x0A,
  * 0x0D), starting at pointer <start> which must be within <ist>. Returns
- * non-zero if such a character is found, 0 otherwise. When run on unlikely
- * header match, it's recommended to first check for the presence of control
- * chars using ist_find_http_ctl().
+ * non-zero if such a character is found, 0 otherwise.
  */
 static inline int http_header_has_forbidden_char(const struct ist ist, const char *start)
 {
-	do {
+	/* speed-up: skip 4 or 8 non-offending chars at once when possible.
+	 * We only skip 0x20-0x7e here. The rest is handled one byte at a time.
+	 */
+#ifdef HA_UNALIGNED_LE64
+	while (start <= istend(ist) - sizeof(long)) {
+		if (is_char8_outside(*(ulong *)start, 0x20, 0x7e))
+			break;
+		start += sizeof(long);
+	}
+#elif defined(HA_UNALIGNED_LE)
+	while (start <= istend(ist) - sizeof(int)) {
+		if (is_char4_outside(*(uint *)start, 0x20, 0x7e))
+			break;
+		start += sizeof(int);
+	}
+#endif
+	/* here we might have skipped most SP+VCHAR (0x20-0x7e). What commonly
+	 * remains is HT (rare), CTL, CR/LF (forbidden), or obs-text.
+	 */
+	for (; start < istend(ist); start++) {
 		if ((uint8_t)*start <= 0x0d &&
 		    (1U << (uint8_t)*start) & ((1<<13) | (1<<10) | (1<<0)))
 			return 1;
-		start++;
-	} while (start < istend(ist));
+	}
 	return 0;
 }