[lit] Improve tool substitution in lit.

This addresses two sources of inconsistency in test configuration
files.

1. Substitution boundaries.  Previously you would specify a
   substitution, such as 'lli', and then additionally a set
   of characters that should fail to match before and after
   the tool.  This was used, for example, so that matches that
   are parts of full paths would not be replaced.  But not all
   tools did this, and those that did would often re-invent
   the set of characters themselves, leading to inconsistency.
   Now, every tool substitution defaults to using a sane set
   of reasonable defaults and you have to explicitly opt out
   of it.  This actually fixed a few latent bugs that were
   never being surfaced, but only on accident.

2. There was no standard way for the system to decide how to
   locate a tool.  Sometimes you have an explicit path, sometimes
   we would search for it and build up a path ourselves, and
   sometimes we would build up a full command line.  Furthermore,
   there was no standardized way to handle missing tools.  Do we
   warn, fail, ignore, etc?  All of this is now encapsulated in
   the ToolSubst class.  You either specify an exact command to
   run, or an instance of FindTool('<tool-name>') and everything
   else just works.  Furthermore, you can specify an action to
   take if the tool cannot be resolved.

Differential Revision: https://reviews.llvm.org/D38565

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@315085 91177308-0d34-0410-b5e6-96231b3b80d8

Author: Zachary Turner

test/lit.cfg.py

@@ -11,7 +11,8 @@
 import lit.util
 import lit.formats
 from lit.llvm import llvm_config
-from lit.llvm import ToolFilter
+from lit.llvm.subst import FindTool
+from lit.llvm.subst import ToolSubst
 
 # name: The name of this test suite.
 config.name = 'LLVM'
@@ -82,31 +83,30 @@ def get_asan_rtlib():
     return found_dylibs[0]
 
 
-lli = 'lli'
+llvm_config.use_default_substitutions()
+
+# Add site-specific substitutions.
+config.substitutions.append(('%llvmshlibdir', config.llvm_shlib_dir))
+config.substitutions.append(('%shlibext', config.llvm_shlib_ext))
+config.substitutions.append(('%exeext', config.llvm_exe_ext))
+config.substitutions.append(('%host_cc', config.host_cc))
+
+
+lli_args = []
 # The target triple used by default by lli is the process target triple (some
 # triple appropriate for generating code for the current process) but because
 # we don't support COFF in MCJIT well enough for the tests, force ELF format on
 # Windows.  FIXME: the process target triple should be used here, but this is
 # difficult to obtain on Windows.
 if re.search(r'cygwin|mingw32|windows-gnu|windows-msvc|win32', config.host_triple):
-    lli += ' -mtriple=' + config.host_triple + '-elf'
-config.substitutions.append(('%lli', lli))
+    lli_args = ['-mtriple=' + config.host_triple + '-elf']
+
+llc_args = []
 
 # Similarly, have a macro to use llc with DWARF even when the host is win32.
-llc_dwarf = 'llc'
 if re.search(r'win32', config.target_triple):
-    llc_dwarf += ' -mtriple=' + \
-        config.target_triple.replace('-win32', '-mingw32')
-config.substitutions.append(('%llc_dwarf', llc_dwarf))
-
-# Add site-specific substitutions.
-config.substitutions.append(('%gold', config.gold_executable))
-config.substitutions.append(('%go', config.go_executable))
-config.substitutions.append(('%llvmshlibdir', config.llvm_shlib_dir))
-config.substitutions.append(('%shlibext', config.llvm_shlib_ext))
-config.substitutions.append(('%exeext', config.llvm_exe_ext))
-config.substitutions.append(('%python', config.python_executable))
-config.substitutions.append(('%host_cc', config.host_cc))
+    llc_args = [' -mtriple=' +
+                config.target_triple.replace('-win32', '-mingw32')]
 
 # Provide the path to asan runtime lib if available. On darwin, this lib needs
 # to be loaded via DYLD_INSERT_LIBRARIES before libLTO.dylib in case the files
@@ -115,65 +115,50 @@ def get_asan_rtlib():
 asan_rtlib = get_asan_rtlib()
 if asan_rtlib:
     ld64_cmd = 'DYLD_INSERT_LIBRARIES={} {}'.format(asan_rtlib, ld64_cmd)
-config.substitutions.append(('%ld64', ld64_cmd))
 
-# OCaml substitutions.
-# Support tests for both native and bytecode builds.
-config.substitutions.append(('%ocamlc',
-                             '%s ocamlc -cclib -L%s %s' %
-                             (config.ocamlfind_executable, config.llvm_lib_dir, config.ocaml_flags)))
+ocamlc_command = '%s ocamlc -cclib -L%s %s' % (
+    config.ocamlfind_executable, config.llvm_lib_dir, config.ocaml_flags)
+ocamlopt_command = 'true'
 if config.have_ocamlopt:
-    config.substitutions.append(('%ocamlopt',
-                                 '%s ocamlopt -cclib -L%s -cclib -Wl,-rpath,%s %s' %
-                                 (config.ocamlfind_executable, config.llvm_lib_dir, config.llvm_lib_dir, config.ocaml_flags)))
-else:
-    config.substitutions.append(('%ocamlopt', 'true'))
-
-# For each occurrence of an llvm tool name as its own word, replace it
-# with the full path to the build directory holding that tool.  This
-# ensures that we are testing the tools just built and not some random
-# tools that might happen to be in the user's PATH.  Thus this list
-# includes every tool placed in $(LLVM_OBJ_ROOT)/$(BuildMode)/bin
-# (llvm_tools_dir in lit parlance).
-
-# Avoid matching RUN line fragments that are actually part of
-# path names or options or whatever.
-# The regex is a pre-assertion to avoid matching a preceding
-# dot, hyphen, carat, or slash (.foo, -foo, etc.).  Some patterns
-# also have a post-assertion to not match a trailing hyphen (foo-).
-JUNKCHARS = r".-^/<"
-
-required_tools = [
-    'lli', 'llvm-ar', 'llvm-as', 'llvm-bcanalyzer', 'llvm-config', 'llvm-cov',
+    ocamlopt_command = '%s ocamlopt -cclib -L%s -cclib -Wl,-rpath,%s %s' % (
+        config.ocamlfind_executable, config.llvm_lib_dir, config.llvm_lib_dir, config.ocaml_flags)
+
+
+tools = [
+    ToolSubst('%lli', FindTool('lli'), post='.', extra_args=lli_args),
+    ToolSubst('%llc_dwarf', FindTool('llc'), extra_args=llc_args),
+    ToolSubst('%go', config.go_executable, unresolved='ignore'),
+    ToolSubst('%gold', config.gold_executable, unresolved='ignore'),
+    ToolSubst('%ld64', ld64_cmd, unresolved='ignore'),
+    ToolSubst('%ocamlc', ocamlc_command, unresolved='ignore'),
+    ToolSubst('%ocamlopt', ocamlopt_command, unresolved='ignore'),
+]
+
+# FIXME: Why do we have both `lli` and `%lli` that do slightly different things?
+tools.extend([
+    'lli', 'lli-child-target', 'llvm-ar', 'llvm-as', 'llvm-bcanalyzer', 'llvm-config', 'llvm-cov',
     'llvm-cxxdump', 'llvm-cvtres', 'llvm-diff', 'llvm-dis', 'llvm-dsymutil',
     'llvm-dwarfdump', 'llvm-extract', 'llvm-isel-fuzzer', 'llvm-lib',
     'llvm-link', 'llvm-lto', 'llvm-lto2', 'llvm-mc', 'llvm-mcmarkup',
     'llvm-modextract', 'llvm-nm', 'llvm-objcopy', 'llvm-objdump',
     'llvm-pdbutil', 'llvm-profdata', 'llvm-ranlib', 'llvm-readobj',
     'llvm-rtdyld', 'llvm-size', 'llvm-split', 'llvm-strings', 'llvm-tblgen',
     'llvm-c-test', 'llvm-cxxfilt', 'llvm-xray', 'yaml2obj', 'obj2yaml',
-    'FileCheck', 'yaml-bench', 'verify-uselistorder',
-    ToolFilter('bugpoint', post='-'),
-    ToolFilter('llc', pre=JUNKCHARS),
-    ToolFilter('llvm-symbolizer', pre=JUNKCHARS),
-    ToolFilter('opt', JUNKCHARS),
-    ToolFilter('sancov', pre=JUNKCHARS),
-    ToolFilter('sanstats', pre=JUNKCHARS),
-    # Handle these specially as they are strings searched for during testing.
-    ToolFilter(r'\| \bcount\b', verbatim=True),
-    ToolFilter(r'\| \bnot\b', verbatim=True)]
-
-llvm_config.add_tool_substitutions(required_tools, config.llvm_tools_dir)
-
-# For tools that are optional depending on the config, we won't warn
-# if they're missing.
-
-optional_tools = [
-    'llvm-go', 'llvm-mt', 'Kaleidoscope-Ch3', 'Kaleidoscope-Ch4',
-    'Kaleidoscope-Ch5', 'Kaleidoscope-Ch6', 'Kaleidoscope-Ch7',
-    'Kaleidoscope-Ch8']
-llvm_config.add_tool_substitutions(optional_tools, config.llvm_tools_dir,
-                                   warn_missing=False)
+    'yaml-bench', 'verify-uselistorder',
+    'bugpoint', 'llc', 'llvm-symbolizer', 'opt', 'sancov', 'sanstats'])
+
+# The following tools are optional
+tools.extend([
+    ToolSubst('llvm-go', unresolved='ignore'),
+    ToolSubst('llvm-mt', unresolved='ignore'),
+    ToolSubst('Kaleidoscope-Ch3', unresolved='ignore'),
+    ToolSubst('Kaleidoscope-Ch4', unresolved='ignore'),
+    ToolSubst('Kaleidoscope-Ch5', unresolved='ignore'),
+    ToolSubst('Kaleidoscope-Ch6', unresolved='ignore'),
+    ToolSubst('Kaleidoscope-Ch7', unresolved='ignore'),
+    ToolSubst('Kaleidoscope-Ch8', unresolved='ignore')])
+
+llvm_config.add_tool_substitutions(tools, config.llvm_tools_dir)
 
 # Targets
 

utils/lit/lit/llvm/__init__.py

@@ -1,51 +1,9 @@
 from lit.llvm import config
-import lit.util
-import re
 
 llvm_config = None
 
-class ToolFilter(object):
-    """
-        String-like class used to build regex substitution patterns for
-        llvm tools.  Handles things like adding word-boundary patterns,
-        and filtering characters from the beginning an end of a tool name
-    """
-
-    def __init__(self, name, pre=None, post=None, verbatim=False):
-        """
-            Construct a ToolFilter.
-
-            name: the literal name of the substitution to look for.
-
-            pre: If specified, the substitution will not find matches where
-            the character immediately preceding the word-boundary that begins
-            `name` is any of the characters in the string `pre`.
-
-            post: If specified, the substitution will not find matches where
-            the character immediately after the word-boundary that ends `name`
-            is any of the characters specified in the string `post`.
-
-            verbatim: If True, `name` is an exact regex that is passed to the
-            underlying substitution
-        """
-        if verbatim:
-            self.regex = name
-            return
-
-        def not_in(chars, where=''):
-            if not chars:
-                return ''
-            pattern_str = '|'.join(re.escape(x) for x in chars)
-            return r'(?{}!({}))'.format(where, pattern_str)
-
-        self.regex = not_in(pre, '<') + r'\b' + name + r'\b' + not_in(post)
-
-    def __str__(self):
-        return self.regex
-
 
 def initialize(lit_config, test_config):
     global llvm_config
 
     llvm_config = config.LLVMConfig(lit_config, test_config)
-

utils/lit/lit/llvm/config.py

@@ -5,6 +5,8 @@
 import sys
 
 import lit.util
+from lit.llvm.subst import FindTool
+from lit.llvm.subst import ToolSubst
 
 
 def binary_feature(on, feature, off_prefix):
@@ -225,41 +227,47 @@ def make_msabi_triple(self, triple):
         # -win32 is not supported for non-x86 targets; use a default.
         return 'i686-pc-win32'
 
-    def add_tool_substitutions(self, tools, search_dirs, warn_missing=True):
+    def add_tool_substitutions(self, tools, search_dirs=None):
+        if not search_dirs:
+            search_dirs = [self.config.llvm_tools_dir]
+
         if lit.util.is_string(search_dirs):
             search_dirs = [search_dirs]
 
+        tools = [x if isinstance(x, ToolSubst) else ToolSubst(x)
+                 for x in tools]
+
         search_dirs = os.pathsep.join(search_dirs)
+        substitutions = []
+
         for tool in tools:
-            # Extract the tool name from the pattern.  This relies on the tool
-            # name being surrounded by \b word match operators.  If the
-            # pattern starts with "| ", include it in the string to be
-            # substituted.
-            if lit.util.is_string(tool):
-                tool = lit.util.make_word_regex(tool)
-            else:
-                tool = str(tool)
+            match = tool.resolve(self, search_dirs)
 
-            tool_match = re.match(r"^(\\)?((\| )?)\W+b([0-9A-Za-z-_\.]+)\\b\W*$",
-                                  tool)
-            if not tool_match:
+            # Either no match occurred, or there was an unresolved match that
+            # is ignored.
+            if not match:
                 continue
 
-            tool_pipe = tool_match.group(2)
-            tool_name = tool_match.group(4)
-            tool_path = lit.util.which(tool_name, search_dirs)
-            if not tool_path:
-                if warn_missing:
-                    # Warn, but still provide a substitution.
-                    self.lit_config.note(
-                        'Did not find ' + tool_name + ' in %s' % search_dirs)
-                tool_path = self.config.llvm_tools_dir + '/' + tool_name
-
-            if tool_name == 'llc' and os.environ.get('LLVM_ENABLE_MACHINE_VERIFIER') == '1':
-                tool_path += ' -verify-machineinstrs'
-            if tool_name == 'llvm-go':
-                exe = getattr(self.config, 'go_executable', None)
-                if exe:
-                    tool_path += ' go=' + exe
-
-            self.config.substitutions.append((tool, tool_pipe + tool_path))
+            subst_key, tool_pipe, command = match
+
+            # An unresolved match occurred that can't be ignored.  Fail without
+            # adding any of the previously-discovered substitutions.
+            if not command:
+                return False
+
+            substitutions.append((subst_key, tool_pipe + command))
+
+        self.config.substitutions.extend(substitutions)
+        return True
+
+    def use_default_substitutions(self):
+        tool_patterns = [
+            ToolSubst('FileCheck', unresolved='fatal'),
+            # Handle these specially as they are strings searched for during testing.
+            ToolSubst(r'\| \bcount\b', command=FindTool(
+                'count'), verbatim=True, unresolved='fatal'),
+            ToolSubst(r'\| \bnot\b', command=FindTool('not'), verbatim=True, unresolved='fatal')]
+
+        self.config.substitutions.append(('%python', sys.executable))
+        self.add_tool_substitutions(
+            tool_patterns, [self.config.llvm_tools_dir])