test DWARF32 and DWARF64 unit encodings

src/base/base_core.c

@@ -300,6 +300,12 @@ memory_write32(void *ptr, U32 v)
   MemoryCopy(ptr, &v, sizeof(v));
 }
 
+internal void UBSAN_NO_ALIGN
+memory_write64(void *ptr, U64 v)
+{
+  MemoryCopy(ptr, &v, sizeof(v));
+}
+
 internal U8 UBSAN_NO_ALIGN
 memory_read8(void *ptr)
 {

src/base/base_core.h

@@ -1011,6 +1011,7 @@ internal B32 memory_is_zero(void *ptr, U64 size);
 
 internal void memory_write16(void *ptr, U16 v);
 internal void memory_write32(void *ptr, U32 v);
+internal void memory_write64(void *ptr, U64 v);
 
 internal U8  memory_read8(void *ptr);
 internal U16 memory_read16(void *ptr);

src/dwarf/dwarf.h

@@ -29,7 +29,6 @@ typedef enum DW_ExtEnum
   DW_Ext_All   = DW_Ext_GNU|DW_Ext_LLVM|DW_Ext_APPLE|DW_Ext_MIPS,
 } DW_ExtEnum;
 
-#define DW_FormatFromSize(size) ((size) >= max_U32 ? DW_Format_64Bit : DW_Format_32Bit)
 typedef enum DW_Format
 {
   DW_Format_Null,

src/dwarf/dwarf_coff.c

@@ -35,7 +35,7 @@ dw_input_from_coff_section_table(Arena              *arena,
   
   for (U64 i = 0; i < section_count; ++i) {
     COFF_SectionHeader *header         = &section_table[i];
-    Rng1U64             raw_data_range = rng_1u64(header->foff, header->foff + header->fsize);
+    Rng1U64             raw_data_range = rng_1u64(header->foff, header->vsize ? header->foff + header->vsize : header->foff + header->fsize);
     String8             name           = coff_name_from_section_header(string_table, header);
     
     DW_SectionKind s      = dw_section_kind_from_string(name);

src/dwarf/dwarf_parse.c

@@ -12,13 +12,14 @@ internal U64
 str8_deserial_read_dwarf_packed_size(String8 string, U64 off, U64 *size_out)
 {
   U64 bytes_read = 0;
-  if (str8_deserial_read(string, off, size_out, sizeof(U32), sizeof(U32))) {
+  U32 first_four_bytes;
-    if (*size_out == max_U32) {
+  if (str8_deserial_read_struct(string, off, &first_four_bytes) == sizeof(first_four_bytes)) {
-      if (str8_deserial_read_struct(string, off+sizeof(U32), size_out)) {
+    if (first_four_bytes == max_U32) {
+      if (str8_deserial_read_struct(string, off+sizeof(U32), size_out) == sizeof(U64)) {
         bytes_read = sizeof(U32) + sizeof(U64);
       }
     } else {
-      *size_out &= (U64)max_U32;
+      *size_out = first_four_bytes;
       bytes_read = sizeof(U32);
     }
   }
@@ -274,6 +275,10 @@ dw_read_list_unit_header_str_offsets(String8 unit_data, DW_ListUnit *lu_out)
 {
   U64 header_size = 0;
 
+  U32 first_four_bytes = 0;
+  if (str8_deserial_read_struct(unit_data, 0, &first_four_bytes) != sizeof(first_four_bytes)) { goto exit; }
+  DW_Format format = first_four_bytes == max_U32 ? DW_Format_64Bit : DW_Format_32Bit;
+  
   U64 unit_length      = 0;
   U64 unit_length_size = str8_deserial_read_dwarf_packed_size(unit_data, 0, &unit_length);
   
@@ -291,12 +296,13 @@ dw_read_list_unit_header_str_offsets(String8 unit_data, DW_ListUnit *lu_out)
         lu_out->version               = version;
         lu_out->address_size          = 0;
         lu_out->segment_selector_size = 0;
-        lu_out->entry_size            = dw_size_from_format(DW_FormatFromSize(unit_length));
+        lu_out->entry_size            = dw_size_from_format(format);
         lu_out->entries               = str8_skip(unit_data, header_size);
       }
     }
   }
   
+  exit:;
   return header_size;
 }
 
@@ -305,6 +311,10 @@ dw_read_list_unit_header_list(String8 unit_data, DW_ListUnit *lu_out)
 {
   U64 header_size = 0;
 
+  U32 first_four_bytes = 0;
+  if (str8_deserial_read_struct(unit_data, 0, &first_four_bytes) != sizeof(first_four_bytes)) { goto exit; }
+  DW_Format format = first_four_bytes == max_U32 ? DW_Format_64Bit : DW_Format_32Bit;
+  
   U64 unit_length      = 0;
   U64 unit_length_size = str8_deserial_read_dwarf_packed_size(unit_data, 0, &unit_length);
   
@@ -330,7 +340,7 @@ dw_read_list_unit_header_list(String8 unit_data, DW_ListUnit *lu_out)
             lu_out->version               = version;
             lu_out->address_size          = address_size;
             lu_out->segment_selector_size = segment_selector_size;
-            lu_out->entry_size            = dw_size_from_format(DW_FormatFromSize(unit_length));
+            lu_out->entry_size            = dw_size_from_format(format);
             lu_out->entries               = str8_skip(unit_data, header_size);
           }
         }
@@ -338,6 +348,7 @@ dw_read_list_unit_header_list(String8 unit_data, DW_ListUnit *lu_out)
     }
   }
   
+  exit:;
   return header_size;
 }
 
@@ -1973,6 +1984,10 @@ dw_cu_from_info_off(Arena *arena, DW_Input *input, DW_ListUnitInput lu_input, U6
 {
   DW_CompUnit cu = {0};
 
+  U32 first_four_bytes = 0;
+  if (str8_deserial_read_struct(input->sec[DW_Section_Info].data, 0, &first_four_bytes) != sizeof(first_four_bytes)) { goto exit; }
+  DW_Format format = first_four_bytes == max_U32 ? DW_Format_64Bit : DW_Format_32Bit;
+  
   // read unit size in bytes
   U64 length      = 0;
   U64 length_size = str8_deserial_read_dwarf_packed_size(input->sec[DW_Section_Info].data, cu_header_offset, &length);
@@ -1987,7 +2002,6 @@ dw_cu_from_info_off(Arena *arena, DW_Input *input, DW_ListUnitInput lu_input, U6
   DW_Version version;
   TryRead(str8_deserial_read_struct(info, info_cursor, &version), info_cursor, exit);
   
-  DW_Format       format       = DW_FormatFromSize(length);
   U64             abbrev_base  = max_U64;
   U8              address_size = 0;
   DW_CompUnitKind unit_kind    = DW_CompUnitKind_Reserved;
@@ -2202,11 +2216,14 @@ dw_read_line_vm_header(Arena           *arena,
 {
   Temp scratch = scratch_begin(&arena, 1);
 
+  U32 first_four_bytes = 0;
+  if (str8_deserial_read_struct(cu_stmt_list, 0, &first_four_bytes) != sizeof(U32)) { goto exit; }
+
   // read unit length
   U64 length      = 0;
   U64 length_size = str8_deserial_read_dwarf_packed_size(cu_stmt_list, 0, &length);
   
-  DW_Format format = DW_FormatFromSize(length);
+  DW_Format format = first_four_bytes == max_U32 ? DW_Format_64Bit : DW_Format_32Bit;
   U64       cursor = length_size;
   String8   data   = str8_substr(cu_stmt_list, r1u64(0, length + length_size));
   
@@ -2723,6 +2740,9 @@ dw_v4_pub_strings_table_from_section_kind(Arena *arena, DW_Input *input, DW_Sect
   
   String8 section_data = input->sec[section_kind].data;
   for(U64 cursor = 0; cursor < section_data.size; ) {
+    U32 first_four_bytes = 0;
+    if (str8_deserial_read_struct(input->sec[DW_Section_Info].data, 0, &first_four_bytes) != sizeof(first_four_bytes)) { break; }
+    DW_Format format = first_four_bytes == max_U32 ? DW_Format_64Bit : DW_Format_32Bit;
 
     U64 unit_length      = 0;
     U64 unit_length_size = str8_deserial_read_dwarf_packed_size(section_data, cursor, &unit_length);
@@ -2742,8 +2762,6 @@ dw_v4_pub_strings_table_from_section_kind(Arena *arena, DW_Input *input, DW_Sect
       break;
     }
     
-    DW_Format format = DW_FormatFromSize(unit_length);
-    
     U64 debug_info_off = 0;
     cursor += str8_deserial_read_dwarf_uint(section_data, cursor, format, &debug_info_off);
     if (cursor >= cursor_opl) {

src/dwarf/dwarf_writer.c

@@ -1212,15 +1212,15 @@ dw_writer_emit(DW_Writer *writer)
     for EachElement(i, writer->sections) {
       if (writer->sections[i].length) {
         if (writer->format == DW_Format_64Bit) {
-          U64 *length      = writer->sections[i].length;
+          U64 *length      = (U64 *)((U8 *)writer->sections[i].length + sizeof(U32));
           U64  length_size = sizeof(U64) + sizeof(U32);
           Assert(writer->sections[i].srl.total_size >= length_size);
-          *length = writer->sections[i].srl.total_size - length_size;
+          memory_write64(length, writer->sections[i].srl.total_size - length_size);
         } else {
           U32 *length      = writer->sections[i].length;
           U32  length_size = sizeof(U32);
           Assert(writer->sections[i].srl.total_size >= length_size);
-          *length = safe_cast_u32(writer->sections[i].srl.total_size - length_size);
+          memory_write32(length, safe_cast_u32(writer->sections[i].srl.total_size - length_size));
         }
       }
     }

src/torture/torture_dwarf.c

@@ -135,6 +135,64 @@ dw_input_from_writer(Arena *arena, DW_Writer *writer)
   return input;
 }
 
+T_BeginTest(dwarf_32bit)
+{
+  DW_Writer *writer = dw_writer_begin(DW_Format_32Bit, DW_Version_5, DW_CompUnitKind_Compile, Arch_x64);
+  dw_writer_tag_begin(writer, DW_TagKind_CompileUnit);
+    dw_writer_push_attrib_string(writer, DW_AttribKind_Producer, str8_lit("RAD DWARF WRITER"));
+  dw_writer_tag_end(writer);
+
+  DW_Input input = dw_input_from_writer(scratch.arena, writer);
+
+  for EachElement(sec_idx, input.sec) {
+    if (sec_idx == DW_Section_Abbrev) continue;
+    Rng1U64Array unit_ranges = dw_unit_ranges_from_data_arr(scratch.arena, input.sec[sec_idx].data);
+    for EachIndex(range_idx, unit_ranges.count) {
+      Rng1U64 range = unit_ranges.v[range_idx];
+
+      U32 first_four_bytes = 0;
+      T_Ok(str8_deserial_read_struct(input.sec[sec_idx].data, range.min, &first_four_bytes) == sizeof(first_four_bytes));
+      T_Ok(first_four_bytes + 4 == dim_1u64(range));
+
+      U32 unit_length = 0;
+      T_Ok(str8_deserial_read_struct(input.sec[sec_idx].data, range.min, &unit_length) == sizeof(U32));
+      T_Ok(unit_length + 4 == dim_1u64(range));
+    }
+  }
+
+  dw_writer_end(&writer);
+}
+T_EndTest;
+
+T_BeginTest(dwarf_64bit)
+{
+  DW_Writer *writer = dw_writer_begin(DW_Format_64Bit, DW_Version_5, DW_CompUnitKind_Compile, Arch_x64);
+  dw_writer_tag_begin(writer, DW_TagKind_CompileUnit);
+  dw_writer_push_attrib_string(writer, DW_AttribKind_Producer, str8_lit("RAD DWARF WRITER"));
+  dw_writer_tag_end(writer);
+
+  DW_Input input = dw_input_from_writer(scratch.arena, writer);
+
+  for EachElement(sec_idx, input.sec) {
+    if (sec_idx == DW_Section_Abbrev) continue;
+    Rng1U64Array unit_ranges = dw_unit_ranges_from_data_arr(scratch.arena, input.sec[sec_idx].data);
+    for EachIndex(range_idx, unit_ranges.count) {
+      Rng1U64 range = unit_ranges.v[range_idx];
+
+      U32 first_four_bytes = 0;
+      T_Ok(str8_deserial_read_struct(input.sec[sec_idx].data, range.min, &first_four_bytes) == sizeof(first_four_bytes));
+      T_Ok(first_four_bytes == max_U32);
+
+      U64 unit_length = 0;
+      T_Ok(str8_deserial_read_struct(input.sec[sec_idx].data, range.min + sizeof(U32), &unit_length) == sizeof(U64));
+      T_Ok(unit_length + 12 == dim_1u64(range));
+    }
+  }
+
+  dw_writer_end(&writer);
+}
+T_EndTest;
+
 T_BeginTest(dwarf_line_opcodes)
 {
   DW_Writer *writer = dw_writer_begin(DW_Format_32Bit, DW_Version_5, DW_CompUnitKind_Compile, Arch_x64);
@@ -574,7 +632,6 @@ T_BeginTest(dwarf_writer)
   DW_TagTree          tag_tree      = dw_tag_tree_from_cu(scratch.arena, &input, &cu);
   AssertAlways(dwt_tags_must_match(writer->root, tag_tree.root));
 
-
   // validate the writer
 
   T_Ok(writer->current == 0);