From 8475c8064d2cccba1790ebeb134b283d0ac912e8 Mon Sep 17 00:00:00 2001
From: Christopher Haster <chaster@utexas.edu>
Date: Mon, 14 Mar 2022 00:35:43 -0500
Subject: [PATCH] Limit ./scripts/structs.py to report structs in local .h
 files

This requires parsing an additional section of the dwarfinfo (--dwarf=rawlines)
to get the declaration file info.

---

Interpreting the results of ./scripts/structs.py reporting is a bit more
complicated than other scripts, structs aren't used in a consistent
manner so the cost of a large struct depends on the context in which it
is used.

But that being said, there really isn't much reason to report
internal-only structs. These structs really only exist for type-checking
in internal algorithms, and their cost will end up reflected in other RAM
measurements, either stack, heap, or other.
---
 scripts/structs.py | 63 +++++++++++++++++++++++++++++++++++++++-------
 1 file changed, 54 insertions(+), 9 deletions(-)

diff --git a/scripts/structs.py b/scripts/structs.py
index e56ce9d..dfa65dd 100755
--- a/scripts/structs.py
+++ b/scripts/structs.py
@@ -16,16 +16,48 @@ import collections as co
 OBJ_PATHS = ['*.o']
 
 def collect(paths, **args):
-    results = co.defaultdict(lambda: 0)
-    pattern = re.compile(
+    decl_pattern = re.compile(
+        '^\s+(?P<no>[0-9]+)'
+            '\s+(?P<dir>[0-9]+)'
+            '\s+.*'
+            '\s+(?P<file>[^\s]+)$')
+    struct_pattern = re.compile(
         '^(?:.*DW_TAG_(?P<tag>[a-z_]+).*'
             '|^.*DW_AT_name.*:\s*(?P<name>[^:\s]+)\s*'
+            '|^.*DW_AT_decl_file.*:\s*(?P<decl>[0-9]+)\s*'
             '|^.*DW_AT_byte_size.*:\s*(?P<size>[0-9]+)\s*)$')
 
+    results = co.defaultdict(lambda: 0)
     for path in paths:
+        # find decl, we want to filter by structs in .h files
+        decls = {}
+        # note objdump-tool may contain extra args
+        cmd = args['objdump_tool'] + ['--dwarf=rawline', path]
+        if args.get('verbose'):
+            print(' '.join(shlex.quote(c) for c in cmd))
+        proc = sp.Popen(cmd,
+            stdout=sp.PIPE,
+            stderr=sp.PIPE if not args.get('verbose') else None,
+            universal_newlines=True,
+            errors='replace')
+        for line in proc.stdout:
+            # find file numbers
+            m = decl_pattern.match(line)
+            if m:
+                decls[int(m.group('no'))] = (
+                    m.group('file'),
+                    int(m.group('dir')))
+        proc.wait()
+        if proc.returncode != 0:
+            if not args.get('verbose'):
+                for line in proc.stderr:
+                    sys.stdout.write(line)
+            sys.exit(-1)
+
         # collect structs as we parse dwarf info
         found = False
         name = None
+        decl = None
         size = None
 
         # note objdump-tool may contain extra args
@@ -39,16 +71,22 @@ def collect(paths, **args):
             errors='replace')
         for line in proc.stdout:
             # state machine here to find structs
-            m = pattern.match(line)
+            m = struct_pattern.match(line)
             if m:
                 if m.group('tag'):
-                    if name is not None and size is not None:
-                        results[(path, name)] = size
+                    if (name is not None
+                            and decl is not None
+                            and size is not None):
+                        decl_file, decl_dir = decls.get(decl, ('', 0))
+                        results[(path, name)] = (size, decl_file, decl_dir)
                     found = (m.group('tag') == 'structure_type')
                     name = None
+                    decl = None
                     size = None
                 elif found and m.group('name'):
                     name = m.group('name')
+                elif found and name and m.group('decl'):
+                    decl = int(m.group('decl'))
                 elif found and name and m.group('size'):
                     size = int(m.group('size'))
         proc.wait()
@@ -59,18 +97,25 @@ def collect(paths, **args):
             sys.exit(-1)
 
     flat_results = []
-    for (file, struct), size in results.items():
+    for (path, struct), (size, decl_file, decl_dir) in results.items():
         # map to source files
         if args.get('build_dir'):
-            file = re.sub('%s/*' % re.escape(args['build_dir']), '', file)
+            path = re.sub('%s/*' % re.escape(args['build_dir']), '', path)
+        # only include structs declared in header files in the current
+        # directory, ignore internal-only # structs (these are represented
+        # in other measurements)
+        if not args.get('everything'):
+            if not (decl_file.endswith('.h') and decl_dir == 0):
+                continue
         # replace .o with .c, different scripts report .o/.c, we need to
         # choose one if we want to deduplicate csv files
-        file = re.sub('\.o$', '.c', file)
+        path = re.sub('\.o$', '.c', path)
 
-        flat_results.append((file, struct, size))
+        flat_results.append((path, struct, size))
 
     return flat_results
 
+
 def main(**args):
     def openio(path, mode='r'):
         if path == '-':