form value resolution; line table header parsing

src/dwarf/dwarf.c

@@ -194,7 +194,7 @@ dw_attrib_class_from_attrib_mips(DW_AttribKind k)
 }
 
 internal DW_AttribClass
-dw_attrib_class_from_attrib(DW_Version ver, DW_Ext ext, DW_AttribKind k)
+dw_attrib_class_from_kind(DW_Version ver, DW_Ext ext, DW_AttribKind k)
 {
   DW_AttribClass result = DW_AttribClass_Null;
   
@@ -344,11 +344,11 @@ dw_pick_attrib_value_class(DW_Version ver, DW_Ext ext, DW_AttribKind attrib_kind
   // This function's purpose is to find the overlapping class between an
   // DW_AttribKind and DW_FormKind.
   
-  DW_AttribClass attrib_class = dw_attrib_class_from_attrib(ver, ext, attrib_kind);
+  DW_AttribClass attrib_class = dw_attrib_class_from_kind(ver, ext, attrib_kind);
   DW_AttribClass form_class   = dw_attrib_class_from_form_kind(ver, form_kind);
   if(attrib_class == DW_AttribClass_Null || form_class == DW_AttribClass_Null)
   {
-    attrib_class = dw_attrib_class_from_attrib(DW_Version_Last, ext, attrib_kind);
+    attrib_class = dw_attrib_class_from_kind(DW_Version_Last, ext, attrib_kind);
     form_class   = dw_attrib_class_from_form_kind(DW_Version_Last, form_kind);
   }
   

src/dwarf/dwarf.h

@@ -478,7 +478,7 @@ X(GNU_RefAlt,    DW_AttribClass_Undefined) \
 X(GNU_StrpAlt,   DW_AttribClass_String)
 
 typedef U64 DW_FormKind;
-typedef enum DW_FormEnum
+typedef enum DW_FormKindEnum
 {
   DW_Form_Null,
 #define X(_N, _ID) DW_Form_##_N = _ID,
@@ -487,7 +487,7 @@ typedef enum DW_FormEnum
     DW_Form_V5_XList(X)
     DW_Form_GNU_XList(X)
 #undef X
-} DW_FormEnum;
+} DW_FormKindEnum;
 
 //- Attributes DWARF2
 
@@ -1762,7 +1762,7 @@ internal DW_AttribClass dw_attrib_class_from_attrib_llvm (DW_AttribKind k);
 internal DW_AttribClass dw_attrib_class_from_attrib_apple(DW_AttribKind k);
 internal DW_AttribClass dw_attrib_class_from_attrib_mips (DW_AttribKind k);
 
-internal DW_AttribClass dw_attrib_class_from_attrib(DW_Version ver, DW_Ext ext, DW_AttribKind v);
+internal DW_AttribClass dw_attrib_class_from_kind(DW_Version ver, DW_Ext ext, DW_AttribKind v);
 
 //- Form Class Encodings
 

src/dwarf/dwarf_parse_2.c

@@ -246,8 +246,9 @@ internal U64
 dw2_read_form_val(DW2_ParseCtx *ctx, String8 data, U64 off, DW_FormKind form_kind, U64 implicit_const, DW2_FormVal *out)
 {
   U64 start_off = off;
-  DW2_FormVal val = {0};
+  DW2_FormVal val = {form_kind};
   {
+    //- rjf: read value bytes
     U64 bytes_to_read = 0;
     switch(form_kind)
     {
@@ -351,8 +352,7 @@ dw2_read_form_val(DW2_ParseCtx *ctx, String8 data, U64 off, DW_FormKind form_kin
         U64 size = 0;
         U64 og_read_off = off;
         off += str8_deserial_read_uleb128(data, off, &size);
-        val.u128.u64[0] = size;
+        val.string = str8_substr(data, r1u64(off, off+size));
-        val.u128.u64[1] = og_read_off;
         off += size;
       }break;
       
@@ -362,8 +362,7 @@ dw2_read_form_val(DW2_ParseCtx *ctx, String8 data, U64 off, DW_FormKind form_kin
         String8 string = {0};
         U64 og_read_off = off;
         off += str8_deserial_read_cstr(data, off, &string);
-        val.u128.u64[0] = og_read_off;
+        val.string = string;
-        val.u128.u64[1] = og_read_off + string.size - 1;
       }break;
       
       //- rjf: implicit constants (no reading in .debug_info; value comes from .debug_abbrev)
@@ -389,21 +388,41 @@ dw2_read_form_val(DW2_ParseCtx *ctx, String8 data, U64 off, DW_FormKind form_kin
         val.u128.u64[0] = 1;
       }break;
     }
+    
+    //- rjf: in some cases, resolve numeric values -> strings
+    {
+      switch(form_kind)
+      {
+        default:{}break;
+        case DW_Form_Strx1:
+        case DW_Form_Strx2:
+        case DW_Form_Strx3:
+        case DW_Form_Strx4:
+        case DW_Form_Strx:
+        {
+          // TODO(rjf)
+        }break;
+        case DW_Form_LineStrp:
+        {
+          String8 string_section_data = ctx->raw->sec[DW_Section_LineStr].data;
+          val.string = str8_cstring_capped(string_section_data.str + val.u128.u64[0],
+                                           string_section_data.str + string_section_data.size);
+        }break;
+        case DW_Form_Strp:
+        case DW_Form_StrpSup:
+        {
+          String8 string_section_data = ctx->raw->sec[DW_Section_Str].data;
+          val.string = str8_cstring_capped(string_section_data.str + val.u128.u64[0],
+                                           string_section_data.str + string_section_data.size);
+        }break;
+      }
+    }
   }
   *out = val;
   U64 bytes_read = (off - start_off);
   return bytes_read;
 }
 
-internal String8
-dw2_string_from_form_val(DW2_ParseCtx *ctx, DW2_FormVal val)
-{
-  String8 section_data = ctx->raw->sec[val.section_kind].data;
-  Rng1U64 range = r1u64(val.u128.u64[0], val.u128.u64[1]);
-  String8 string = str8_substr(section_data, range);
-  return string;
-}
-
 ////////////////////////////////
 //~ rjf: Tag Parsing
 
@@ -490,7 +509,7 @@ dw2_read_tag(Arena *arena, DW2_ParseCtx *ctx, String8 data, U64 off, DW2_Tag *ta
         SLLQueuePush(attribs.first, attribs.last, n);
         attribs.count += 1;
         n->v.attrib_kind = attrib_kind;
-        n->v.form_kind   = attrib_form_kind;
+        n->v.val         = val;
       }
       
       // rjf: no movement, or no attrib kind -> done
@@ -511,6 +530,21 @@ dw2_read_tag(Arena *arena, DW2_ParseCtx *ctx, String8 data, U64 off, DW2_Tag *ta
   return bytes_read;
 }
 
+internal DW2_Attrib *
+dw2_attrib_from_kind(DW2_Tag *tag, DW_AttribKind kind)
+{
+  DW2_Attrib *result = &dw2_attrib_nil;
+  for EachNode(n, DW2_AttribNode, tag->attribs.first)
+  {
+    if(n->v.attrib_kind == kind)
+    {
+      result = &n->v;
+      break;
+    }
+  }
+  return result;
+}
+
 ////////////////////////////////
 //~ rjf: Line Table Parsing
 
@@ -695,7 +729,7 @@ dw2_read_line_table_header(Arena *arena, DW2_ParseCtx *ctx, String8 data, U64 of
     case DW_Version_5:
     {
       //- rjf: read directory & file tables
-      for(B32 files = 0, dirs = 1; files <= 1; files = 1, dirs = 0)
+      for(B32 do_files = 0, do_dirs = 1; do_files <= 1; do_files += 1, do_dirs = 0)
       {
         // rjf: read header
         U8 entry_formats_count = 0;
@@ -703,7 +737,8 @@ dw2_read_line_table_header(Arena *arena, DW2_ParseCtx *ctx, String8 data, U64 of
         U64 *entry_formats = push_array(scratch.arena, U64, entry_formats_count*2);
         for EachIndex(idx, entry_formats_count)
         {
-          off += str8_deserial_read_uleb128(data, off, &entry_formats[idx]);
+          off += str8_deserial_read_uleb128(data, off, &entry_formats[idx*2 + 0]);
+          off += str8_deserial_read_uleb128(data, off, &entry_formats[idx*2 + 1]);
         }
         U64 table_count = 0;
         off += str8_deserial_read_uleb128(data, off, &table_count);
@@ -731,7 +766,7 @@ dw2_read_line_table_header(Arena *arena, DW2_ParseCtx *ctx, String8 data, U64 of
                 }break;
                 case DW_LNCT_Path:
                 {
-                  entry_out->file_name = dw2_string_from_form_val(ctx, form_val);
+                  entry_out->file_name = form_val.string;
                 }break;
                 case DW_LNCT_DirectoryIndex:
                 {
@@ -751,12 +786,17 @@ dw2_read_line_table_header(Arena *arena, DW2_ParseCtx *ctx, String8 data, U64 of
                 }break;
                 case DW_LNCT_LLVM_Source:
                 {
-                  entry_out->source = dw2_string_from_form_val(ctx, form_val);
+                  entry_out->source = form_val.string;
                 }break;
               }
             }
           }
         }
+        
+        // rjf: store
+        if(0){}
+        else if(do_files) { files = table; }
+        else if(do_dirs)  { dirs = table; }
       }
       
     }break;

src/dwarf/dwarf_parse_2.h

@@ -59,7 +59,8 @@ struct DW2_ParseCtx
 typedef struct DW2_FormVal DW2_FormVal;
 struct DW2_FormVal
 {
-  DW_SectionKind section_kind;
+  DW_FormKind kind;
+  String8 string;
   U128 u128;
 };
 
@@ -67,7 +68,6 @@ typedef struct DW2_Attrib DW2_Attrib;
 struct DW2_Attrib
 {
   DW_AttribKind attrib_kind;
-  DW_FormKind form_kind;
   DW2_FormVal val;
 };
 
@@ -154,6 +154,11 @@ struct DW2_LineTableHeader
   DW2_LineTableFileArray files;
 };
 
+////////////////////////////////
+//~ rjf: Globals
+
+global read_only DW2_Attrib dw2_attrib_nil = {0};
+
 ////////////////////////////////
 //~ rjf: Basic Parsing Helpers
 
@@ -174,12 +179,12 @@ internal DW2_AbbrevMap dw2_abbrev_map_from_data(Arena *arena, String8 data, U64
 //~ rjf: Form Value Parsing
 
 internal U64 dw2_read_form_val(DW2_ParseCtx *ctx, String8 data, U64 off, DW_FormKind form_kind, U64 implicit_const, DW2_FormVal *out);
-internal String8 dw2_string_from_form_val(DW2_ParseCtx *ctx, DW2_FormVal val);
 
 ////////////////////////////////
 //~ rjf: Tag Parsing
 
 internal U64 dw2_read_tag(Arena *arena, DW2_ParseCtx *ctx, String8 data, U64 off, DW2_Tag *tag_out);
+internal DW2_Attrib *dw2_attrib_from_kind(DW2_Tag *tag, DW_AttribKind kind);
 
 ////////////////////////////////
 //~ rjf: Line Table Parsing

src/rdi_from_dwarf/rdi_from_dwarf_2.c

@@ -298,6 +298,31 @@ d2r2_convert(Arena *arena, D2R2_ConvertParams *params)
   }
   lane_sync();
   
+  ////////////////////////////
+  //- rjf: parse each unit's line table
+  //
+  {
+    U64 unit_take_idx = 0;
+    U64 *unit_take_idx_ptr = &unit_take_idx;
+    lane_sync_u64(&unit_take_idx_ptr, 0);
+    for(;;)
+    {
+      U64 unit_idx = ins_atomic_u64_inc_eval(unit_take_idx_ptr) - 1;
+      if(unit_idx >= unit_count)
+      {
+        break;
+      }
+      DW2_ParseCtx *ctx = &unit_parse_ctxs[unit_idx];
+      DW2_Tag *unit_root_tag = &unit_root_tags[unit_idx];
+      DW2_Attrib *stmt_list = dw2_attrib_from_kind(unit_root_tag, DW_AttribKind_StmtList);
+      U64 line_info_off = stmt_list->val.u128.u64[0];
+      DW2_LineTableHeader line_table_header = {0};
+      String8 line_info_data = raw->sec[DW_Section_Line].data;
+      dw2_read_line_table_header(scratch.arena, ctx, line_info_data, line_info_off, &line_table_header);
+    }
+  }
+  lane_sync();
+  
   ////////////////////////////
   //- rjf: fill result
   //