From 56bcf04c496f5df1599f41e6bfd35a9a6b9a1ffc Mon Sep 17 00:00:00 2001
From: "Unknown W. Brackets" <checkins@unknownbrackets.org>
Date: Fri, 23 Apr 2021 23:24:47 -0700
Subject: [PATCH] Debugger: Try harder to validate UTF-8.

When sending this in the websocket debugger, it needs to be valid utf-8 or
it will cause clients to abort the connection.  We want to reject invalid
utf-8 anyway.
---
 Core/Debugger/DisassemblyManager.cpp | 10 ++++++++++
 1 file changed, 10 insertions(+)

diff --git a/Core/Debugger/DisassemblyManager.cpp b/Core/Debugger/DisassemblyManager.cpp
index d950fb5357..40daa83fef 100644
--- a/Core/Debugger/DisassemblyManager.cpp
+++ b/Core/Debugger/DisassemblyManager.cpp
@@ -64,13 +64,23 @@ bool IsLikelyStringAt(uint32_t addr) {
 	if (utf.end())
 		return false;
 
+	char verify[4];
 	while (!utf.end()) {
 		if (utf.invalid())
 			return false;
 
+		int pos = utf.byteIndex();
 		uint32_t c = utf.next();
+		int len = UTF8::encode(verify, c);
+		// Our decoder is a bit lax, so let's verify this is a normal encoding.
+		// This prevents us from trying to output invalid encodings in the debugger.
+		if (memcmp(p + pos, verify, len) != 0 || pos + len != utf.byteIndex())
+			return false;
+
 		if (c < ARRAY_SIZE(validControl) && !validControl[c])
 			return false;
+		if (c > 0x0010FFFF)
+			return false;
 	}
 
 	return true;