Fix ICE when include_str! reads binary files

Author: reddevilmidzy

compiler/rustc_parse/src/lib.rs

@@ -130,6 +130,13 @@ pub fn utf8_error<E: EmissionGuarantee>(
     };
     let contents = String::from_utf8_lossy(contents).to_string();
     let source = sm.new_source_file(PathBuf::from(path).into(), contents);
+
+    // Avoid out-of-bounds span from lossy UTF-8 conversion.
+    if start as u32 > source.normalized_source_len.0 {
+        err.note(note);
+        return;
+    }
+
     let span = Span::with_root_ctxt(
         source.normalized_byte_pos(start as u32),
         source.normalized_byte_pos(start as u32),

tests/ui/include-macros/binary-file-linux.rs

@@ -0,0 +1,8 @@
+//@ only-linux
+//@ normalize-stderr: "at byte `\d+`" -> "at byte `$$BYTE`"
+// Test ensure reading a binary file with include_str! doesn't cause an ICE
+// regression test for issue <https://github.com/rust-lang/rust/issues/149304>
+
+#![doc = include_str!("/bin/id")] //~ ERROR: `/bin/id` wasn't a utf-8 file
+
+fn main() {}

tests/ui/include-macros/binary-file-linux.stderr

@@ -0,0 +1,10 @@
+error: `/bin/id` wasn't a utf-8 file
+  --> $DIR/binary-file-linux.rs:6:10
+   |
+LL | #![doc = include_str!("/bin/id")]
+   |          ^^^^^^^^^^^^^^^^^^^^^^^
+   |
+   = note: invalid utf-8 at byte `$BYTE`
+
+error: aborting due to 1 previous error
+

tests/ui/include-macros/binary-file-windows.rs

@@ -0,0 +1,8 @@
+//@ only-windows
+//@ normalize-stderr: "at byte `\d+`" -> "at byte `$$BYTE`"
+// Test ensure reading a binary file with include_str! doesn't cause an ICE
+// regression test for issue <https://github.com/rust-lang/rust/issues/149304>
+
+#![doc = include_str!("C:\\Windows\\System32\\cmd.exe")] //~ ERROR: `C:/Windows/System32/cmd.exe` wasn't a utf-8 file
+
+fn main() {}

tests/ui/include-macros/binary-file-windows.stderr

@@ -0,0 +1,10 @@
+error: `C:/Windows/System32/cmd.exe` wasn't a utf-8 file
+  --> $DIR/binary-file-windows.rs:6:10
+   |
+LL | #![doc = include_str!("C:/Windows/System32/cmd.exe")]
+   |          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+   |
+   = note: invalid utf-8 at byte `$BYTE`
+
+error: aborting due to 1 previous error
+