From b11855821e5848fdbb6f66b609b2e1c25c7ddb8a Mon Sep 17 00:00:00 2001
From: "Roger A. Light" <roger@atchoo.org>
Date: Tue, 16 May 2017 14:51:29 +0100
Subject: [PATCH] Update UTF-8 validation checks.

---
 lib/utf8_mosq.c                        | 12 ++++++------
 test/lib/c/09-util-utf8-validate.c     |  3 ++-
 test/lib/cpp/09-util-utf8-validate.cpp |  3 ++-
 3 files changed, 10 insertions(+), 8 deletions(-)

diff --git a/lib/utf8_mosq.c b/lib/utf8_mosq.c
index f20e9931..dd5eedd9 100644
--- a/lib/utf8_mosq.c
+++ b/lib/utf8_mosq.c
@@ -60,11 +60,11 @@ int mosquitto_validate_utf8(const char *str, int len)
 		}
 
 		/* Reconstruct full code point */
+		if(i == len-codelen+1){
+			/* Not enough data */
+			return MOSQ_ERR_MALFORMED_UTF8;
+		}
 		for(j=0; j<codelen-1; j++){
-			if(i == len-1){
-				/* Not enough data */
-				return MOSQ_ERR_MALFORMED_UTF8;
-			}
 			if((ustr[++i] & 0xC0) != 0x80){
 				/* Not a continuation byte */
 				return MOSQ_ERR_MALFORMED_UTF8;
@@ -77,12 +77,12 @@ int mosquitto_validate_utf8(const char *str, int len)
 			return MOSQ_ERR_MALFORMED_UTF8;
 		}
 
-		/* Check for overlong encodings */
+		/* Check for overlong or out of range encodings */
 		if(codelen == 2 && codepoint < 0x0080){
 			return MOSQ_ERR_MALFORMED_UTF8;
 		}else if(codelen == 3 && codepoint < 0x0800){
 			return MOSQ_ERR_MALFORMED_UTF8;
-		}else if(codelen == 4 && codepoint < 0x10000){
+		}else if(codelen == 4 && (codepoint < 0x10000 || codepoint > 0x10FFFF)){
 			return MOSQ_ERR_MALFORMED_UTF8;
 		}
 	}
diff --git a/test/lib/c/09-util-utf8-validate.c b/test/lib/c/09-util-utf8-validate.c
index 1a472328..8a8e85b4 100644
--- a/test/lib/c/09-util-utf8-validate.c
+++ b/test/lib/c/09-util-utf8-validate.c
@@ -53,7 +53,8 @@ int main(int argc, char *argv[])
 	assert_valid("2.3.2  U-0000E000 = ee 80 80 = \"\"");
 	assert_valid("2.3.3  U-0000FFFD = ef bf bd = \"�\"");
 	assert_valid("2.3.4  U-0010FFFF = f4 8f bf bf = \"􏿿\"");
-	assert_valid("2.3.5  U-00110000 = f4 90 80 80 = \"����\"");
+	/* This used to be valid in pre-2003 utf-8 */
+	assert_invalid("2.3.5  U-00110000 = f4 90 80 80 = \"����\"");
 
 	/* 3  Malformed sequences */
 	/* 3.1  Unexpected continuation bytes */
diff --git a/test/lib/cpp/09-util-utf8-validate.cpp b/test/lib/cpp/09-util-utf8-validate.cpp
index ea8c0218..2decafcb 100644
--- a/test/lib/cpp/09-util-utf8-validate.cpp
+++ b/test/lib/cpp/09-util-utf8-validate.cpp
@@ -53,7 +53,8 @@ int main(int argc, char *argv[])
 	assert_valid("2.3.2  U-0000E000 = ee 80 80 = \"\"");
 	assert_valid("2.3.3  U-0000FFFD = ef bf bd = \"�\"");
 	assert_valid("2.3.4  U-0010FFFF = f4 8f bf bf = \"􏿿\"");
-	assert_valid("2.3.5  U-00110000 = f4 90 80 80 = \"����\"");
+	/* This used to be valid in pre-2003 utf-8 */
+	assert_invalid("2.3.5  U-00110000 = f4 90 80 80 = \"����\"");
 
 	/* 3  Malformed sequences */
 	/* 3.1  Unexpected continuation bytes */