Support file hashes in .debug_line

Author: l4l

src/debuginfo/line_info.rs

@@ -1,14 +1,15 @@
 use std::ffi::OsStr;
+use std::time::SystemTime;
 use std::path::{Component, Path};
 
 use crate::prelude::*;
 
-use rustc_span::{FileName, SourceFileAndLine, Pos};
+use rustc_span::{FileName, SourceFile, SourceFileAndLine, Pos, SourceFileHash, SourceFileHashAlgorithm};
 
 use cranelift_codegen::binemit::CodeOffset;
 
 use gimli::write::{
-    Address, AttributeValue, FileId, LineProgram, LineString, LineStringTable, UnitEntryId,
+    Address, AttributeValue, FileId, LineProgram, LineString, FileInfo, LineStringTable, UnitEntryId,
 };
 
 // OPTIMIZATION: It is cheaper to do this in one pass than using `.parent()` and `.file_name()`.
@@ -38,9 +39,9 @@ fn osstr_as_utf8_bytes(path: &OsStr) -> &[u8] {
 fn line_program_add_file(
     line_program: &mut LineProgram,
     line_strings: &mut LineStringTable,
-    file: &FileName,
+    file: &SourceFile,
 ) -> FileId {
-    match file {
+    match &file.name {
         FileName::Real(path) => {
             let (dir_path, file_name) = split_path_dir_and_file(path);
             let dir_name = osstr_as_utf8_bytes(dir_path.as_os_str());
@@ -57,13 +58,32 @@ fn line_program_add_file(
                 line_program.encoding(),
                 line_strings,
             );
-            line_program.add_file(file_name, dir_id, None)
+
+            let md5 = Some(file.src_hash)
+                .filter(|h| matches!(h, SourceFileHash { kind: SourceFileHashAlgorithm::Md5, .. }))
+                .map(|h| {
+                    let mut buf = [0u8; super::MD5_LEN];
+                    buf.copy_from_slice(h.hash_bytes());
+                    buf
+                });
+
+            line_program.file_has_timestamp = true;
+            line_program.file_has_md5 = md5.is_some();
+
+            line_program.add_file(file_name, dir_id, Some(FileInfo {
+                timestamp: SystemTime::now()
+                    .duration_since(SystemTime::UNIX_EPOCH)
+                    .map(|t| t.as_secs())
+                    .unwrap_or(0),
+                size: 0,
+                md5: md5.unwrap_or_default(),
+            }))
         }
         // FIXME give more appropriate file names
-        _ => {
+        filename => {
             let dir_id = line_program.default_directory();
             let dummy_file_name = LineString::new(
-                file.to_string().into_bytes(),
+                filename.to_string().into_bytes(),
                 line_program.encoding(),
                 line_strings,
             );
@@ -79,7 +99,7 @@ impl<'tcx> DebugContext<'tcx> {
         let file_id = line_program_add_file(
             &mut self.dwarf.unit.line_program,
             &mut self.dwarf.line_strings,
-            &loc.file.name,
+            &loc.file,
         );
 
         let entry = self.dwarf.unit.get_mut(entry_id);
@@ -167,7 +187,7 @@ impl<'a, 'tcx> FunctionDebugContext<'a, 'tcx> {
                 true
             };
             if current_file_changed {
-                let file_id = line_program_add_file(line_program, line_strings, &file.name);
+                let file_id = line_program_add_file(line_program, line_strings, &file);
                 line_program.row().file = file_id;
                 last_file = Some(file.clone());
             }

src/debuginfo/mod.rs

@@ -1,15 +1,19 @@
 mod emit;
 mod line_info;
 
+use std::time::SystemTime;
+
 use crate::prelude::*;
 
+use rustc_span::{FileName, SourceFileHash, SourceFileHashAlgorithm};
+
 use cranelift_codegen::ir::{StackSlots, ValueLabel, ValueLoc};
 use cranelift_codegen::isa::TargetIsa;
 use cranelift_codegen::ValueLocRange;
 
 use gimli::write::{
     self, Address, AttributeValue, DwarfUnit, Expression, LineProgram, LineString, Location,
-    LocationList, Range, RangeList, UnitEntryId, Writer,
+    LocationList, Range, RangeList, UnitEntryId, Writer, FileInfo,
 };
 use gimli::{Encoding, Format, LineEncoding, RunTimeEndian, X86_64};
 
@@ -24,6 +28,8 @@ fn target_endian(tcx: TyCtxt<'_>) -> RunTimeEndian {
     }
 }
 
+const MD5_LEN: usize = 16;
+
 pub(crate) struct DebugContext<'tcx> {
     tcx: TyCtxt<'tcx>,
 
@@ -42,7 +48,8 @@ impl<'tcx> DebugContext<'tcx> {
             format: Format::Dwarf32,
             // TODO: this should be configurable
             // macOS doesn't seem to support DWARF > 3
-            version: 3,
+            // 5 version is required for md5 file hash
+            version: 5,
             address_size,
         };
 
@@ -52,18 +59,41 @@ impl<'tcx> DebugContext<'tcx> {
         // Normally this would use option_env!("CFG_VERSION").
         let producer = format!("cg_clif (rustc {})", "unknown version");
         let comp_dir = tcx.sess.working_dir.0.to_string_lossy().into_owned();
-        let name = match tcx.sess.local_crate_source_file {
-            Some(ref path) => path.to_string_lossy().into_owned(),
-            None => tcx.crate_name(LOCAL_CRATE).to_string(),
+        let (name, md5) = match tcx.sess.local_crate_source_file.clone() {
+            Some(path) => {
+                let name = path.to_string_lossy().into_owned();
+                let hash = tcx.sess
+                    .source_map()
+                    .get_source_file(&FileName::Real(path))
+                    .map(|f| f.src_hash)
+                    .filter(|h| matches!(h, SourceFileHash { kind: SourceFileHashAlgorithm::Md5, .. }))
+                    .map(|h| {
+                        let mut buf = [0u8; MD5_LEN];
+                        buf.copy_from_slice(h.hash_bytes());
+                        buf
+                    });
+                (name, hash)
+            },
+            None => (tcx.crate_name(LOCAL_CRATE).to_string(), None),
         };
 
-        let line_program = LineProgram::new(
+        let mut line_program = LineProgram::new(
             encoding,
             LineEncoding::default(),
             LineString::new(comp_dir.as_bytes(), encoding, &mut dwarf.line_strings),
             LineString::new(name.as_bytes(), encoding, &mut dwarf.line_strings),
-            None,
+            Some(FileInfo {
+                timestamp: SystemTime::now()
+                    .duration_since(SystemTime::UNIX_EPOCH)
+                    .map(|t| t.as_secs())
+                    .unwrap_or(0),
+                size: 0,
+                md5: md5.unwrap_or_default(),
+            }),
         );
+        line_program.file_has_timestamp = true;
+        line_program.file_has_md5 = md5.is_some();
+
         dwarf.unit.line_program = line_program;
 
         {