Fix ICE when include_str! reads binary files

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),

src/tools/tidy/src/ui_tests.rs

@@ -157,6 +157,7 @@ fn check_unexpected_extension(check: &mut RunningCheck, file_path: &Path, ext: &
         "tests/ui/crate-loading/auxiliary/libfoo.rlib", // testing loading a manually created rlib
         "tests/ui/include-macros/data.bin", // testing including data with the include macros
         "tests/ui/include-macros/file.txt", // testing including data with the include macros
+        "tests/ui/include-macros/invalid-utf8-binary-file.bin", // testing including data with the include macros
         "tests/ui/macros/macro-expanded-include/file.txt", // testing including data with the include macros
         "tests/ui/macros/not-utf8.bin", // testing including data with the include macros
         "tests/ui/macros/syntax-extension-source-utils-files/includeme.fragment", // more include

tests/ui/include-macros/invalid-utf8-binary-file.rs

@@ -0,0 +1,10 @@
+//@ normalize-stderr: "at byte `\d+`" -> "at byte `$$BYTE`"
+//@ normalize-stderr: "`[^`\n]*invalid-utf8-binary-file\.bin`" -> "`$DIR/invalid-utf8-binary-file.bin`"
+//@ rustc-env:INVALID_UTF8_BIN={{src-base}}/include-macros/invalid-utf8-binary-file.bin
+
+//! Ensure that ICE does not occur when reading an invalid UTF8 file with an absolute path.
+//! regression test for issue <https://github.com/rust-lang/rust/issues/149304>
+
+#![doc = include_str!(concat!(env!("INVALID_UTF8_BIN")))] //~ ERROR: wasn't a utf-8 file
+
+fn main() {}

tests/ui/include-macros/invalid-utf8-binary-file.stderr

@@ -0,0 +1,10 @@
+error: `/invalid-utf8-binary-file.bin` wasn't a utf-8 file
+  --> $DIR/invalid-utf8-binary-file.rs:8:10
+   |
+LL | #![doc = include_str!(concat!(env!("INVALID_UTF8_BIN")))]
+   |          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+   |
+   = note: invalid utf-8 at byte `$BYTE`
+
+error: aborting due to 1 previous error
+