[UTC] Support to check generated comments

Author: Camsyn

llvm/utils/UpdateTestChecks/common.py

@@ -607,7 +607,7 @@ def invoke_tool(exe, cmd_args, ir, preprocess_cmd=None, verbose=False):
 DEBUG_ONLY_ARG_RE = re.compile(r"-debug-only[= ]([^ ]+)")
 
 IS_DEBUG_RECORD_RE = re.compile(r"^(\s+)#dbg_")
-IS_SWITCH_CASE_RE = re.compile(r"^\s+i\d+ \d+, label %\w+")
+IS_SWITCH_CASE_RE = re.compile(r"^\s+i\d+ \d+, label %\S+")
 
 SCRUB_LEADING_WHITESPACE_RE = re.compile(r"^(\s+)")
 SCRUB_WHITESPACE_RE = re.compile(r"(?!^(|  \w))[ \t]+", flags=re.M)
@@ -1123,6 +1123,8 @@ def processed_prefixes(self, prefixes):
 ##### Generator of LLVM IR CHECK lines
 
 SCRUB_IR_COMMENT_RE = re.compile(r"\s*;.*")
+# Comments to indicate the predecessors of a block in the IR.
+SCRUB_PRED_COMMENT_RE = re.compile(r"\s*; preds = .*")
 SCRUB_IR_FUNC_META_RE = re.compile(r"((?:\!(?!dbg\b)[a-zA-Z_]\w*(?:\s+![0-9]+)?)\s*)+")
 
 # TODO: We should also derive check lines for global, debug, loop declarations, etc..
@@ -1877,6 +1879,7 @@ def generalize_check_lines(
     *,
     unstable_globals_only=False,
     no_meta_details=False,
+    ignore_all_comments=True,  # If False, only ignore comments of predecessors
 ):
     if unstable_globals_only:
         regexp = ginfo.get_unstable_globals_regexp()
@@ -1904,8 +1907,13 @@ def escape_braces(match_obj):
                         line,
                     )
                     break
-            # Ignore any comments, since the check lines will too.
-            scrubbed_line = SCRUB_IR_COMMENT_RE.sub(r"", line)
+            scrubbed_line = (
+                # Ignore any comments, since the check lines will too.
+                SCRUB_IR_COMMENT_RE.sub(r"", line)
+                if ignore_all_comments
+                # Ignore comments of predecessors only.
+                else SCRUB_PRED_COMMENT_RE.sub(r"", line)
+            )
             # Ignore the metadata details if check global is none
             if no_meta_details:
                 scrubbed_line = SCRUB_IR_FUNC_META_RE.sub(r"{{.*}}", scrubbed_line)
@@ -2083,6 +2091,7 @@ def add_checks(
     global_tbaa_records_for_prefixes={},
     preserve_names=False,
     original_check_lines: Mapping[str, List[str]] = {},
+    check_comments=True,
 ):
     # prefix_exclusions are prefixes we cannot use to print the function because it doesn't exist in run lines that use these prefixes as well.
     prefix_exclusions = set()
@@ -2280,6 +2289,8 @@ def add_checks(
                     global_tbaa_records,
                     preserve_names,
                     original_check_lines=original_check_lines.get(checkprefix),
+                    # IR output might require comments checks, e.g., predicate info, memssa
+                    ignore_all_comments=not check_comments,
                 )
 
                 # This could be selectively enabled with an optional invocation argument.
@@ -2299,8 +2310,9 @@ def add_checks(
                     if func_line.strip() == "":
                         is_blank_line = True
                         continue
-                    # Do not waste time checking IR comments.
-                    func_line = SCRUB_IR_COMMENT_RE.sub(r"", func_line)
+                    if not check_comments:
+                        # Do not waste time checking IR comments unless necessary.
+                        func_line = SCRUB_IR_COMMENT_RE.sub(r"", func_line)
 
                     # Skip blank lines instead of checking them.
                     if is_blank_line:
@@ -2342,6 +2354,7 @@ def add_ir_checks(
     global_vars_seen_dict,
     global_tbaa_records_for_prefixes,
     is_filtered,
+    check_comments=False,
     original_check_lines={},
 ):
     assert ginfo.is_ir()
@@ -2368,6 +2381,7 @@ def add_ir_checks(
         global_tbaa_records_for_prefixes,
         preserve_names,
         original_check_lines=original_check_lines,
+        check_comments=check_comments,
     )
 
 

llvm/utils/update_test_checks.py

@@ -197,6 +197,7 @@ def update_test(ti: common.TestInfo):
                     global_tbaa_records_for_prefixes,
                     is_filtered=builder.is_filtered(),
                     original_check_lines=original_check_lines.get(func, {}),
+                    check_comments=args.check_comments,
                 ),
             )
         )
@@ -230,6 +231,7 @@ def update_test(ti: common.TestInfo):
                         global_tbaa_records_for_prefixes,
                         is_filtered=builder.is_filtered(),
                         original_check_lines=original_check_lines.get(func_name, {}),
+                        check_comments=args.check_comments,
                     )
                 )
                 is_in_function_start = False
@@ -362,6 +364,12 @@ def main():
         choices=["none", "smart", "all"],
         help="Check global entries (global variables, metadata, attribute sets, ...) for functions",
     )
+    parser.add_argument(
+        "--check-comments",
+        action="store_true",
+        default=False,
+        help="Check the generated comments (e.g., PredicateInfo/MemSSA) for functions",
+    )
     parser.add_argument(
         "--reset-variable-names",
         action="store_true",