Review pass.

scripts/elf_dwarf.lua

@@ -616,7 +616,7 @@ end
 --- - ELF32 symtab entry = 16 bytes (`st_name:4 + st_value:4 + st_size:4 + st_info:1 + st_other:1 + st_shndx:2`); offsets within each entry are zero-based wire offsets.
 --- - Direct Lua `string.byte`/`string.sub`/`string.find` boundaries receive `+ 1`.
 --- - We filter on STB_GLOBAL (high nibble of st_info = 1) to match `nm`'s default (external symbols only). STB_WEAK excluded.
---- - We strip the `code_` prefix to match the previous `read_nm` output. 
+--- - The `code_` prefix is stripped (MipsAtom_ macros emit bare atom names, no `code_` prefix).
 --- - `st_size > 0` filter excludes undefined/imported symbols.
 --- @param elf_path Path
 --- @return table<string, {integer, integer}>
@@ -654,7 +654,7 @@ function M.read_nm(elf_path)
 				-- Extract the name from .strtab (null-terminated C string).
 				local name_end = strtab:find("\0", st_name_off + 1, true) or (st_name_off + 1)
 				local name     = strtab:sub(st_name_off + 1, name_end - 1)
-				-- Filter: keep all symbol-table symbols (atoms emit their name as the bare `<name>` since the `code_` prefix was removed from the MipsAtom_ macro).
+				-- Filter: keep all symbol-table symbols (atoms emit their name as the bare `<name>` — MipsAtom_ macros strip the `code_` prefix).
 				-- The atoms_source_map pass already filters out non-atom symbols via the source-map.txt cross-ref.
 				if name and #name > 0 then
 					local st_value = M.read_u32_le(symtab, entry_off + SYM_ST_VALUE)
@@ -791,132 +791,4 @@ end
 -- I/O helpers: atoms source-map + native directory glob
 -- ════════════════════════════════════════════════════════════════════════════
 
---- Parse a FORMAT_VERSION <expected_version> atoms-meta file (sourcemap or provenance).
---- Shared by M.parse_source_map_file + M.parse_provenance_file.
---- The two callers differ only in how they parse WORD lines; that's `extract_word(line)`.
---- Returns the standard `{name -> {total, words}}` shape.
---- Returns `{}` on format-version mismatch (and logs to stderr).
---- @param path             string
---- @param expected_version integer
---- @param extract_word     fun(line: string): table|nil  -- caller-supplied per-line parser
---- @return table<string, table>
-function M.parse_atom_records(path, expected_version, extract_word)
-	local out = {}
-	local cur_name, cur_words = nil, {}
-	for raw in io.lines(path) do
-		local line = raw
-		if line:match("^#") then
-			local ver = line:match("^# FORMAT_VERSION%s+(%d+)")
-			if ver and tonumber(ver) ~= expected_version then
-				io.stderr:write(string.format(
-					"[elf_dwarf.parse_atom_records] version mismatch (got %s, expected %d) in %s\n",
-					ver, expected_version, path))
-				return {}
-			end
-			-- skip other comments
-		elseif line:sub(1, 4) == "ATOM" then
-			-- ATOM <name> "<abs-source-path>" <total>
-			local _, _, name = line:find("ATOM%s+(%S+)%s+\"[^\"]*\"%s+(%d+)")
-			if name then
-				cur_name  = name
-				cur_words = {}
-				out[name] = { total = 0, words = cur_words }
-			end
-		elseif line == "ENDATOM" then
-			-- Update the recorded total from the entries count
-			-- (matches the `lines[1] = lines[1]:gsub(" 0$", " " .. total)` patch in atoms_source_map.lua:170).
-			if cur_name and out[cur_name] then
-				out[cur_name].total = #cur_words
-			end
-			cur_name, cur_words = nil, {}
-		elseif line:sub(1, 4) == "WORD" and cur_name then
-			local field = extract_word(line)
-			if field then
-				cur_words[#cur_words + 1] = field
-			end
-		end
-	end
-	return out
-end
-
---- Parse a FORMAT_VERSION <expected_version> `*.atoms.sourcemap.txt` file.
---- Returns `{name -> {total = N, words = {{pos, line}, ...}}}`.
---- Returns `{}` on format-version mismatch (and logs to stderr).
----
---- **Wire format** (emitted by `passes/atoms_source_map.lua`):
---- ```
---- # FORMAT_VERSION <n>
---- ATOM <name> "<abs-source-path>" <total>
---- WORD <n> LINE <line> TEXT <text...>
---- ...
---- ENDATOM
---- ```
----
---- **Conventions:** the in-memory shape uses `{pos, line, text}`
---- (`atoms_source_map.lua:142`); the `.txt` file uses `WORD <n>` so the parser maps `n` → `pos` field name.
---- @param sm_path Path
---- @param expected_version integer  -- expected FORMAT_VERSION line
---- @return table<string, table>
-function M.parse_source_map_file(sm_path, expected_version)
-	return M.parse_atom_records(sm_path, expected_version, function(line)
-		local _, n, _, src_line = line:find("WORD%s+(%d+)%s+LINE%s+(%d+)")
-		if n and src_line then
-			return { pos = tonumber(n), line = tonumber(src_line) }
-		end
-	end)
-end
-
---- Parse a FORMAT_VERSION <expected_version> `*.atoms.provenance.txt` file.
---- Returns `{name -> {total = N, words = {{pos, call_file, call_line, comp_name, comp_file, comp_line}, ...}}}`.
---- Returns `{}` on format-version mismatch (and logs to stderr).
----
---- **Wire format** (emitted by `passes/atoms_source_map.lua`):
---- ```
---- # FORMAT_VERSION <n>
---- ATOM <name> "<abs-source-path>" <total>
---- WORD <n>  CALL <src-file>:<src-line>  RAW
---- WORD <n>  CALL <src-file>:<src-line>  MACRO <comp_name> "<comp-file>:<comp-line>"
---- ...
---- ENDATOM
---- ```
----
---- **Used by** `passes/dwarf_injection.lua` to:
----   - group consecutive MACRO rows into component invocations (one `DW_TAG_inlined_subroutine` each)
----   - emit abstract `DW_TAG_subprogram` per unique component name
----   - extend `.debug_line` so stepping into a `mac_X(...)` lands on the component's source line.
---- @param prov_path string  -- path to *.atoms.provenance.txt
---- @param expected_version integer  -- expected FORMAT_VERSION line
---- @return table<string, table>
-function M.parse_provenance_file(prov_path, expected_version)
-	return M.parse_atom_records(prov_path, expected_version, function(line)
-		-- Two accepted shapes:
-		--   WORD <n>  CALL <call-file>:<call-line>  RAW
-		--   WORD <n>  CALL <call-file>:<call-line>  MACRO <comp_name> "<comp-file>:<comp-line>"
-		local pos, call_file, call_line, comp_name, comp_file, comp_line =
-			line:match('WORD%s+(%d+)%s+CALL%s+(.-):(%d+)%s+MACRO%s+(%S+)%s+"([^"]*):(%d+)"')
-		if pos then
-			return {
-				pos       = tonumber(pos),
-				call_file = call_file,
-				call_line = tonumber(call_line),
-				comp_name = comp_name,
-				comp_file = comp_file,
-				comp_line = tonumber(comp_line),
-			}
-		end
-		-- RAW row.
-		local raw_pos, raw_file, raw_line = line:match('WORD%s+(%d+)%s+CALL%s+(.-):(%d+)%s+RAW')
-		if raw_pos then
-			return {
-				pos       = tonumber(raw_pos),
-				call_file = raw_file,
-				call_line = tonumber(raw_line),
-				comp_name = nil,
-				comp_file = nil,
-				comp_line = nil,
-			}
-		end
-	end)
-end
-
 return M