Cache symbol lists used by LLD_REPORT_UNDEFINED. NFC

This means that the JS libraries only only need to be processed when
there is cache miss.  The cost of processing the JS libraries is about
300ms on my machine which is about 30% of the link time for hello
world.  When there is cache hit this cost is reduced to 3ms.

This change is in prepartion for switching this mode on my default in.

See: #16003

Author: sbc100

emcc.py

@@ -23,6 +23,8 @@
 from tools.toolchain_profiler import ToolchainProfiler
 
 import base64
+import glob
+import hashlib
 import json
 import logging
 import os
@@ -500,8 +502,7 @@ def ensure_archive_index(archive_file):
     run_process([shared.LLVM_RANLIB, archive_file])
 
 
-@ToolchainProfiler.profile_block('JS symbol generation')
-def get_all_js_syms():
+def generate_js_symbols():
   # Runs the js compiler to generate a list of all symbols available in the JS
   # libraries.  This must be done separately for each linker invokation since the
   # list of symbols depends on what settings are used.
@@ -516,6 +517,59 @@ def get_all_js_syms():
     if shared.is_c_symbol(name):
       name = shared.demangle_c_symbol_name(name)
       library_syms.add(name)
+  return library_syms
+
+
+@ToolchainProfiler.profile_block('JS symbol generation')
+def get_all_js_syms():
+  # Avoiding using the cache when generating struct info since
+  # this step is performed while the cache is locked.
+  if settings.BOOTSTRAPPING_STRUCT_INFO or config.FROZEN_CACHE:
+    return generate_js_symbols()
+
+  # To avoid the cost of calling generate_js_symbols each time an executable is
+  # linked we cache symbol lists for the N most recently used configs.
+  # We define a cache hit as when the settings and `--js-library` contents are
+  # identical.
+  input_files = {}
+  # The JS library files are all absolute paths so can't conflict with this one.
+  input_files['settings.json'] = json.dumps(settings.dict(), sort_keys=True, indent=2)
+  for jslib in sorted(glob.glob(utils.path_from_root('src') + '/library*.js')):
+    input_files[jslib] = read_file(jslib)
+  for jslib in settings.JS_LIBRARIES:
+    if not os.path.isabs(jslib):
+      jslib = utils.path_from_root('src', jslib)
+    input_files[jslib] = read_file(jslib)
+  input_data = []
+  for name, content in input_files.items():
+    content_hash = hashlib.sha1(content.encode('utf-8')).hexdigest()
+    input_data.append(f'{name}: {content_hash}')
+
+  input_data = '\n'.join(input_data) + '\n'
+  cache_filename = None
+  num_cache_entries = 20
+
+  with cache.lock('js_symbol_lists'):
+    oldest_timestamp = 0
+    for i in range(num_cache_entries):
+      input_file = cache.get_path(f'js_symbol_list_{i}.inputs')
+      list_file = cache.get_path(f'js_symbol_list_{i}.txt')
+      if not os.path.exists(input_file) or not os.path.exists(list_file):
+        cache_filename = list_file
+        break
+      timestamp = os.path.getmtime(input_file)
+      if timestamp < oldest_timestamp or not oldest_timestamp:
+        oldest_timestamp = timestamp
+        cache_filename = list_file
+      if read_file(input_file) == input_data:
+        # Cache hit, read the symbol list from the list_file
+        return read_file(list_file).splitlines()
+
+    # Cache miss. Generate a new symbol list and write to the the cache
+    library_syms = generate_js_symbols()
+
+    write_file(cache_filename, '\n'.join(library_syms) + '\n')
+    write_file(shared.replace_suffix(cache_filename, '.inputs'), input_data)
 
   return library_syms