From 46a61f02f12d697c6f25eb03371bbb22a5b8504a Mon Sep 17 00:00:00 2001
From: Raiki Tamura <tamaron1203@gmail.com>
Date: Tue, 4 Jul 2023 18:21:48 +0900
Subject: [PATCH] gccrs: add utf-8 validation for input source

gcc/rust/ChangeLog:

	* lex/rust-lex.cc (Lexer::input_source_is_valid_utf8): New method of `Lexer`.
	* lex/rust-lex.h: Likewise.
	* rust-session-manager.cc (Session::compile_crate): Add error.

gcc/testsuite/ChangeLog:

	* rust/compile/broken_utf8.rs: New test.

Signed-off-by: Raiki Tamura <tamaron1203@gmail.com>
---
 gcc/rust/lex/rust-lex.cc                  |  6 ++++++
 gcc/rust/lex/rust-lex.h                   | 15 ++++++++-------
 gcc/rust/rust-session-manager.cc          |  8 ++++++++
 gcc/testsuite/rust/compile/broken_utf8.rs |  2 ++
 4 files changed, 24 insertions(+), 7 deletions(-)
 create mode 100644 gcc/testsuite/rust/compile/broken_utf8.rs

diff --git a/gcc/rust/lex/rust-lex.cc b/gcc/rust/lex/rust-lex.cc
index 3593ee05f8f5..19bf5c46d951 100644
--- a/gcc/rust/lex/rust-lex.cc
+++ b/gcc/rust/lex/rust-lex.cc
@@ -167,6 +167,12 @@ Lexer::~Lexer ()
   // line_map->stop();
 }
 
+bool
+Lexer::input_source_is_valid_utf8 ()
+{
+  return raw_input_source->is_valid ();
+}
+
 /* TODO: need to optimise somehow to avoid the virtual function call in the
  * tight loop. Best idea at the moment is CRTP, but that might make lexer
  * implementation annoying when storing the "base class" (i.e. would need
diff --git a/gcc/rust/lex/rust-lex.h b/gcc/rust/lex/rust-lex.h
index 4355394acde1..2034fd77bd8d 100644
--- a/gcc/rust/lex/rust-lex.h
+++ b/gcc/rust/lex/rust-lex.h
@@ -175,6 +175,8 @@ class Lexer
   Lexer (Lexer &&other) = default;
   Lexer &operator= (Lexer &&other) = default;
 
+  bool input_source_is_valid_utf8 ();
+
   // Returns token n tokens ahead of current position.
   const_TokenPtr peek_token (int n) { return token_queue.peek (n); }
   // Peeks the current token.
@@ -217,9 +219,9 @@ class Lexer
 
     Codepoint next_codepoint ()
     {
-      uint8_t input = next_byte ();
+      uint32_t input = next_byte ();
 
-      if ((int8_t) input == EOF)
+      if ((int32_t) input == EOF)
 	return Codepoint::eof ();
       else if (input < 128)
 	{
@@ -246,11 +248,13 @@ class Lexer
 	  // 3 bytes or UTF-8 BOM
 	  uint8_t input2 = next_byte ();
 	  // If the second byte is equal to 0xBB then the input is no longer a
-	  // valid UTF-8 char.
+	  // valid UTF-8 char. Then, we check if the third byte makes up a UTF
+	  // BOM.
 	  if (input == 0xEF && input2 == 0xBB)
 	    {
 	      uint8_t input3 = next_byte ();
 	      if (input3 == 0xBF)
+		// found BOM
 		return next_codepoint ();
 	      else
 		return {0xFFFE};
@@ -289,8 +293,6 @@ class Lexer
 	}
       else
 	{
-	  // rust_error_at (get_current_location (),
-	  //   "invalid UTF-8 [SECND] (too long)");
 	  return {0xFFFE};
 	}
     }
@@ -362,8 +364,7 @@ class Lexer
     {
       if (offs >= buffer.size ())
 	return EOF;
-
-      return buffer.at (offs++);
+      return (uint8_t) buffer.at (offs++);
     }
 
   public:
diff --git a/gcc/rust/rust-session-manager.cc b/gcc/rust/rust-session-manager.cc
index 08e8cdb8ae37..0fe6aa52e2d3 100644
--- a/gcc/rust/rust-session-manager.cc
+++ b/gcc/rust/rust-session-manager.cc
@@ -497,6 +497,14 @@ Session::compile_crate (const char *filename)
 
   Lexer lex (filename, std::move (file_wrap), linemap, dump_lex_opt);
 
+  if (!lex.input_source_is_valid_utf8 ())
+    {
+      rust_error_at (Linemap::unknown_location (),
+		     "cannot read %s; stream did not contain valid UTF-8",
+		     filename);
+      return;
+    }
+
   Parser<Lexer> parser (lex);
 
   // generate crate from parser
diff --git a/gcc/testsuite/rust/compile/broken_utf8.rs b/gcc/testsuite/rust/compile/broken_utf8.rs
new file mode 100644
index 000000000000..8053b83ebd76
--- /dev/null
+++ b/gcc/testsuite/rust/compile/broken_utf8.rs
@@ -0,0 +1,2 @@
+// { dg-excess-errors "stream did not contain valid UTF-8" }
+ÿ
\ No newline at end of file