clang-tools-extra/clangd/TidyFastChecks.py

   1 #!/usr/bin/env python3
   2 #
   3 # Determines which clang-tidy checks are "fast enough" to run in clangd.
   4 # This runs clangd --check --check-tidy-time and parses the output.
   5 # This program outputs a header fragment specifying which checks are fast:
   6 #   FAST(bugprone-argument-comment, 5)
   7 #   SLOW(misc-const-correctness, 200)
   8 # If given the old header fragment as input, we lean to preserve its choices.
   9 #
  10 # This is not deterministic or hermetic, but should be run occasionally to
  11 # update the list of allowed checks. From llvm-project:
  12 #   clang-tools-extra/clangd/TidyFastChecks.py --clangd=build-opt/bin/clangd
  13 # Be sure to use an optimized, no-asserts, tidy-enabled build of clangd!
  14
  15 import argparse
  16 import os
  17 import re
  18 import subprocess
  19 import sys
  20
  21 # Checks faster than FAST_THRESHOLD are fast, slower than SLOW_THRESHOLD slow.
  22 # If a check is in between, we stick with our previous decision. This avoids
  23 # enabling/disabling checks between releases due to random measurement jitter.
  24 FAST_THRESHOLD = 8  # percent
  25 SLOW_THRESHOLD = 15
  26
  27 parser = argparse.ArgumentParser()
  28 parser.add_argument(
  29     "--target",
  30     help="X-macro output file. "
  31     "If it exists, existing contents will be used for hysteresis",
  32     default="clang-tools-extra/clangd/TidyFastChecks.inc",
  33 )
  34 parser.add_argument(
  35     "--source",
  36     help="Source file to benchmark tidy checks",
  37     default="clang/lib/Sema/Sema.cpp",
  38 )
  39 parser.add_argument(
  40     "--clangd", help="clangd binary to invoke", default="build/bin/clangd"
  41 )
  42 parser.add_argument("--checks", help="check glob to run", default="*")
  43 parser.add_argument("--verbose", help="log clangd output", action="store_true")
  44 args = parser.parse_args()
  45
  46 # Use the preprocessor to extract the list of previously-fast checks.
  47 def read_old_fast(path):
  48     text = subprocess.check_output(
  49         [
  50             "cpp",
  51             "-P",  # Omit GNU line markers
  52             "-nostdinc",  # Don't include stdc-predef.h
  53             "-DFAST(C,T)=C",  # Print fast checks only
  54             path,
  55         ]
  56     )
  57     for line in text.splitlines():
  58         if line.strip():
  59             yield line.strip().decode("utf-8")
  60
  61
  62 old_fast = list(read_old_fast(args.target)) if os.path.exists(args.target) else []
  63 print(f"Old fast checks: {old_fast}", file=sys.stderr)
  64
  65 # Runs clangd --check --check-tidy-time.
  66 # Yields (check, percent-overhead) pairs.
  67 def measure():
  68     process = subprocess.Popen(
  69         [
  70             args.clangd,
  71             "--check=" + args.source,
  72             "--check-locations=0",  # Skip useless slow steps.
  73             "--check-tidy-time=" + args.checks,
  74         ],
  75         stderr=subprocess.PIPE,
  76     )
  77     recording = False
  78     for line in iter(process.stderr.readline, b""):
  79         if args.verbose:
  80             print("clangd> ", line, file=sys.stderr)
  81         if not recording:
  82             if b"Timing AST build with individual clang-tidy checks" in line:
  83                 recording = True
  84             continue
  85         if b"Finished individual clang-tidy checks" in line:
  86             return
  87         match = re.search(rb"(\S+) = (\S+)%", line)
  88         if match:
  89             yield (match.group(1).decode("utf-8"), float(match.group(2)))
  90
  91
  92 with open(args.target, "w", buffering=1) as target:
  93     # Produce an includable X-macros fragment with our decisions.
  94     print(
  95         f"""// This file is generated, do not edit it directly!
  96 // Deltas are percentage regression in parsing {args.source}
  97 #ifndef FAST
  98 #define FAST(CHECK, DELTA)
  99 #endif
 100 #ifndef SLOW
 101 #define SLOW(CHECK, DELTA)
 102 #endif
 103 """,
 104         file=target,
 105     )
 106
 107     for check, time in measure():
 108         threshold = SLOW_THRESHOLD if check in old_fast else FAST_THRESHOLD
 109         decision = "FAST" if time <= threshold else "SLOW"
 110         print(f"{decision} {check} {time}% <= {threshold}%", file=sys.stderr)
 111         print(f"{decision}({check}, {time})", file=target)
 112
 113     print(
 114         """
 115 #undef FAST
 116 #undef SLOW
 117 """,
 118         file=target,
 119     )