Auto merge of #115507 - cjgillot:relative-source-file, r=oli-obk

Use relative positions inside a SourceFile.

This allows to remove the normalization of start positions for hashing, and simplify allocation of global address space.

cc `@Zoxc`

Author: bors

compiler/rustc_codegen_cranelift/src/debuginfo/line_info.rs

@@ -82,12 +82,9 @@ impl DebugContext {
         match tcx.sess.source_map().lookup_line(span.lo()) {
             Ok(SourceFileAndLine { sf: file, line }) => {
                 let line_pos = file.lines(|lines| lines[line]);
+                let col = file.relative_position(span.lo()) - line_pos;
 
-                (
-                    file,
-                    u64::try_from(line).unwrap() + 1,
-                    u64::from((span.lo() - line_pos).to_u32()) + 1,
-                )
+                (file, u64::try_from(line).unwrap() + 1, u64::from(col.to_u32()) + 1)
             }
             Err(file) => (file, 0, 0),
         }

compiler/rustc_codegen_llvm/src/debuginfo/create_scope_map.rs

@@ -68,7 +68,7 @@ fn make_mir_scope<'ll, 'tcx>(
         let file = cx.sess().source_map().lookup_source_file(mir.span.lo());
         debug_context.scopes[scope] = DebugScope {
             file_start_pos: file.start_pos,
-            file_end_pos: file.end_pos,
+            file_end_pos: file.end_position(),
             ..debug_context.scopes[scope]
         };
         instantiated.insert(scope);
@@ -120,7 +120,7 @@ fn make_mir_scope<'ll, 'tcx>(
         dbg_scope,
         inlined_at: inlined_at.or(parent_scope.inlined_at),
         file_start_pos: loc.file.start_pos,
-        file_end_pos: loc.file.end_pos,
+        file_end_pos: loc.file.end_position(),
     };
     instantiated.insert(scope);
 }

compiler/rustc_codegen_llvm/src/debuginfo/mod.rs

@@ -267,7 +267,7 @@ impl CodegenCx<'_, '_> {
 
                 // Use 1-based indexing.
                 let line = (line + 1) as u32;
-                let col = (pos - line_pos).to_u32() + 1;
+                let col = (file.relative_position(pos) - line_pos).to_u32() + 1;
 
                 (file, line, col)
             }

compiler/rustc_metadata/src/rmeta/decoder.rs

@@ -1501,11 +1501,12 @@ impl<'a, 'tcx> CrateMetadataRef<'a> {
 
                 // We can't reuse an existing SourceFile, so allocate a new one
                 // containing the information we need.
+                let original_end_pos = source_file_to_import.end_position();
                 let rustc_span::SourceFile {
                     mut name,
                     src_hash,
-                    start_pos,
-                    end_pos,
+                    start_pos: original_start_pos,
+                    source_len,
                     lines,
                     multibyte_chars,
                     non_narrow_chars,
@@ -1547,35 +1548,32 @@ impl<'a, 'tcx> CrateMetadataRef<'a> {
                 // on `try_to_translate_virtual_to_real`).
                 try_to_translate_virtual_to_real(&mut name);
 
-                let source_length = (end_pos - start_pos).to_usize();
-
                 let local_version = sess.source_map().new_imported_source_file(
                     name,
                     src_hash,
                     name_hash,
-                    source_length,
+                    source_len.to_u32(),
                     self.cnum,
                     lines,
                     multibyte_chars,
                     non_narrow_chars,
                     normalized_pos,
-                    start_pos,
                     source_file_index,
                 );
                 debug!(
                     "CrateMetaData::imported_source_files alloc \
-                         source_file {:?} original (start_pos {:?} end_pos {:?}) \
-                         translated (start_pos {:?} end_pos {:?})",
+                         source_file {:?} original (start_pos {:?} source_len {:?}) \
+                         translated (start_pos {:?} source_len {:?})",
                     local_version.name,
-                    start_pos,
-                    end_pos,
+                    original_start_pos,
+                    source_len,
                     local_version.start_pos,
-                    local_version.end_pos
+                    local_version.source_len
                 );
 
                 ImportedSourceFile {
-                    original_start_pos: start_pos,
-                    original_end_pos: end_pos,
+                    original_start_pos,
+                    original_end_pos,
                     translated_source_file: local_version,
                 }
             })

compiler/rustc_middle/src/query/on_disk_cache.rs

@@ -22,7 +22,7 @@ use rustc_span::hygiene::{
     ExpnId, HygieneDecodeContext, HygieneEncodeContext, SyntaxContext, SyntaxContextData,
 };
 use rustc_span::source_map::{SourceMap, StableSourceFileId};
-use rustc_span::{BytePos, ExpnData, ExpnHash, Pos, SourceFile, Span};
+use rustc_span::{BytePos, ExpnData, ExpnHash, Pos, RelativeBytePos, SourceFile, Span};
 use rustc_span::{CachingSourceMapView, Symbol};
 use std::collections::hash_map::Entry;
 use std::io;
@@ -688,11 +688,12 @@ impl<'a, 'tcx> Decodable<CacheDecoder<'a, 'tcx>> for Span {
 
         let file_lo_index = SourceFileIndex::decode(decoder);
         let line_lo = usize::decode(decoder);
-        let col_lo = BytePos::decode(decoder);
+        let col_lo = RelativeBytePos::decode(decoder);
         let len = BytePos::decode(decoder);
 
         let file_lo = decoder.file_index_to_file(file_lo_index);
         let lo = file_lo.lines(|lines| lines[line_lo - 1] + col_lo);
+        let lo = file_lo.absolute_position(lo);
         let hi = lo + len;
 
         Span::new(lo, hi, ctxt, parent)

compiler/rustc_mir_transform/src/coverage/mod.rs

@@ -28,7 +28,7 @@ use rustc_middle::mir::{
 use rustc_middle::ty::TyCtxt;
 use rustc_span::def_id::DefId;
 use rustc_span::source_map::SourceMap;
-use rustc_span::{CharPos, ExpnKind, Pos, SourceFile, Span, Symbol};
+use rustc_span::{ExpnKind, SourceFile, Span, Symbol};
 
 /// A simple error message wrapper for `coverage::Error`s.
 #[derive(Debug)]
@@ -314,8 +314,7 @@ impl<'a, 'tcx> Instrumentor<'a, 'tcx> {
             };
             graphviz_data.add_bcb_coverage_span_with_counter(bcb, &covspan, &counter_kind);
 
-            let code_region =
-                make_code_region(source_map, file_name, &self.source_file, span, body_span);
+            let code_region = make_code_region(source_map, file_name, span, body_span);
 
             inject_statement(
                 self.mir_body,
@@ -510,40 +509,36 @@ fn inject_intermediate_expression(mir_body: &mut mir::Body<'_>, expression: Cove
 fn make_code_region(
     source_map: &SourceMap,
     file_name: Symbol,
-    source_file: &Lrc<SourceFile>,
     span: Span,
     body_span: Span,
 ) -> CodeRegion {
     debug!(
-        "Called make_code_region(file_name={}, source_file={:?}, span={}, body_span={})",
+        "Called make_code_region(file_name={}, span={}, body_span={})",
         file_name,
-        source_file,
         source_map.span_to_diagnostic_string(span),
         source_map.span_to_diagnostic_string(body_span)
     );
 
-    let (start_line, mut start_col) = source_file.lookup_file_pos(span.lo());
-    let (end_line, end_col) = if span.hi() == span.lo() {
-        let (end_line, mut end_col) = (start_line, start_col);
+    let (file, mut start_line, mut start_col, mut end_line, mut end_col) =
+        source_map.span_to_location_info(span);
+    if span.hi() == span.lo() {
         // Extend an empty span by one character so the region will be counted.
-        let CharPos(char_pos) = start_col;
         if span.hi() == body_span.hi() {
-            start_col = CharPos(char_pos.saturating_sub(1));
+            start_col = start_col.saturating_sub(1);
         } else {
-            end_col = CharPos(char_pos + 1);
+            end_col = start_col + 1;
         }
-        (end_line, end_col)
-    } else {
-        source_file.lookup_file_pos(span.hi())
     };
-    let start_line = source_map.doctest_offset_line(&source_file.name, start_line);
-    let end_line = source_map.doctest_offset_line(&source_file.name, end_line);
+    if let Some(file) = file {
+        start_line = source_map.doctest_offset_line(&file.name, start_line);
+        end_line = source_map.doctest_offset_line(&file.name, end_line);
+    }
     CodeRegion {
         file_name,
         start_line: start_line as u32,
-        start_col: start_col.to_u32() + 1,
+        start_col: start_col as u32,
         end_line: end_line as u32,
-        end_col: end_col.to_u32() + 1,
+        end_col: end_col as u32,
     }
 }
 

compiler/rustc_parse/src/lib.rs

@@ -132,7 +132,7 @@ fn maybe_source_file_to_parser(
     sess: &ParseSess,
     source_file: Lrc<SourceFile>,
 ) -> Result<Parser<'_>, Vec<Diagnostic>> {
-    let end_pos = source_file.end_pos;
+    let end_pos = source_file.end_position();
     let stream = maybe_file_to_stream(sess, source_file, None)?;
     let mut parser = stream_to_parser(sess, stream, None);
     if parser.token == token::Eof {

compiler/rustc_query_system/src/ich/impls_syntax.rs

@@ -5,7 +5,7 @@ use crate::ich::StableHashingContext;
 
 use rustc_ast as ast;
 use rustc_data_structures::stable_hasher::{HashStable, StableHasher};
-use rustc_span::{BytePos, NormalizedPos, SourceFile};
+use rustc_span::SourceFile;
 use std::assert_matches::assert_matches;
 
 use smallvec::SmallVec;
@@ -67,8 +67,8 @@ impl<'a> HashStable<StableHashingContext<'a>> for SourceFile {
             src: _,
             ref src_hash,
             external_src: _,
-            start_pos,
-            end_pos: _,
+            start_pos: _,
+            source_len: _,
             lines: _,
             ref multibyte_chars,
             ref non_narrow_chars,
@@ -85,56 +85,30 @@ impl<'a> HashStable<StableHashingContext<'a>> for SourceFile {
             // We only hash the relative position within this source_file
             lines.len().hash_stable(hcx, hasher);
             for &line in lines.iter() {
-                stable_byte_pos(line, start_pos).hash_stable(hcx, hasher);
+                line.hash_stable(hcx, hasher);
             }
         });
 
         // We only hash the relative position within this source_file
         multibyte_chars.len().hash_stable(hcx, hasher);
         for &char_pos in multibyte_chars.iter() {
-            stable_multibyte_char(char_pos, start_pos).hash_stable(hcx, hasher);
+            char_pos.hash_stable(hcx, hasher);
         }
 
         non_narrow_chars.len().hash_stable(hcx, hasher);
         for &char_pos in non_narrow_chars.iter() {
-            stable_non_narrow_char(char_pos, start_pos).hash_stable(hcx, hasher);
+            char_pos.hash_stable(hcx, hasher);
         }
 
         normalized_pos.len().hash_stable(hcx, hasher);
         for &char_pos in normalized_pos.iter() {
-            stable_normalized_pos(char_pos, start_pos).hash_stable(hcx, hasher);
+            char_pos.hash_stable(hcx, hasher);
         }
 
         cnum.hash_stable(hcx, hasher);
     }
 }
 
-fn stable_byte_pos(pos: BytePos, source_file_start: BytePos) -> u32 {
-    pos.0 - source_file_start.0
-}
-
-fn stable_multibyte_char(mbc: rustc_span::MultiByteChar, source_file_start: BytePos) -> (u32, u32) {
-    let rustc_span::MultiByteChar { pos, bytes } = mbc;
-
-    (pos.0 - source_file_start.0, bytes as u32)
-}
-
-fn stable_non_narrow_char(
-    swc: rustc_span::NonNarrowChar,
-    source_file_start: BytePos,
-) -> (u32, u32) {
-    let pos = swc.pos();
-    let width = swc.width();
-
-    (pos.0 - source_file_start.0, width as u32)
-}
-
-fn stable_normalized_pos(np: NormalizedPos, source_file_start: BytePos) -> (u32, u32) {
-    let NormalizedPos { pos, diff } = np;
-
-    (pos.0 - source_file_start.0, diff)
-}
-
 impl<'tcx> HashStable<StableHashingContext<'tcx>> for rustc_feature::Features {
     fn hash_stable(&self, hcx: &mut StableHashingContext<'tcx>, hasher: &mut StableHasher) {
         // Unfortunately we cannot exhaustively list fields here, since the