[Utils] Reuse argument variable names in the body

[jdoerfert]

Summary

If we have int foo(int a) { return a; } and we run with --function-signature
enabled, we want a single variable declaration for a which is reused
later.

[lebedev.ri]

Can you show the effect by regenerating some test?

[jdoerfert]

In D69722#1733607, @lebedev.ri wrote:

Can you show the effect by regenerating some test?

Take line 8 here https://reviews.llvm.org/D68766#change-4geuqNmmUCZL

`; ARGPROMOTION-NEXT:    ret i32 [[X_VAL]]`

It uses [[X_VAL]] which is an argument of the function.
Without this check line would look like this:

`; ARGPROMOTION-NEXT:    ret i32 [[X:.%*]]`

which has no relationship to the argument.

[lebedev.ri]

Now that i believe there's some testing infra for these utils, add/update tests for this?

[jdoerfert]

Add and adjust regression tests

[merge_guards_bot]

Unit tests: unknown.

clang-tidy: unknown.

clang-format: unknown.

Build artifacts: diff.json, console-log.txt

[lebedev.ri]

This is not opt-in and will cause a new wave of test regeneration,
BUT @signature suggests to me that this could be viewed as a fix and not a new feature.

Any other opinions, @spatel @RKSimon ?
(please don't commit until more generally-favorable review outcome)

[jdoerfert]

Add an explicit test without --function-signature

[jdoerfert]

In D69722#1799669, @lebedev.ri wrote:

This is not opt-in and will cause a new wave of test regeneration,

This is "opt-in" with the --function-signature option. I added an explicit test and the basic test should cover that as well, e.g., only the .funcsig. version is affected.

[lebedev.ri]

Right, we couldn't have already matched function arg names because we don't normally do that.

[merge_guards_bot]

Unit tests: fail. 61155 tests passed, 1 failed and 728 were skipped.

failed: LLVM.tools/UpdateTestChecks/update_test_checks/argument_name_reuse.test

clang-tidy: pass.

clang-format: pass.

Build artifacts: diff.json, clang-tidy.txt, clang-format.patch, CMakeCache.txt, console-log.txt, test-results.xml

llvm/test/tools/UpdateTestChecks/update_test_checks/Inputs/argument_name_reuse.ll

+; RUN: opt < %s -S | FileCheck %s
+
+define i32 @reuse_arg_names(i32 %X, i32 %Y) {
+  %Z = sub i32 %X, %Y
+  %Q = add i32 %Z, %Y
+  ret i32 %Q
+}

llvm/test/tools/UpdateTestChecks/update_test_checks/Inputs/argument_name_reuse.ll.expected

+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --function-signature
+; RUN: opt < %s -S | FileCheck %s
+
+define i32 @reuse_arg_names(i32 %X, i32 %Y) {
+; CHECK-LABEL: define {{[^@]+}}@reuse_arg_names
+; CHECK-SAME: (i32 [[X:%.*]], i32 [[Y:%.*]])
+; CHECK-NEXT:    [[Z:%.*]] = sub i32 [[X]], [[Y]]
+; CHECK-NEXT:    [[Q:%.*]] = add i32 [[Z]], [[Y]]
+; CHECK-NEXT:    ret i32 [[Q]]
+;
+  %Z = sub i32 %X, %Y
+  %Q = add i32 %Z, %Y
+  ret i32 %Q
+}

llvm/test/tools/UpdateTestChecks/update_test_checks/Inputs/argument_name_reuse.ll.plain.expected

+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt < %s -S | FileCheck %s
+
+define i32 @reuse_arg_names(i32 %X, i32 %Y) {
+; CHECK-LABEL: @reuse_arg_names(
+; CHECK-NEXT:    [[Z:%.*]] = sub i32 [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT:    [[Q:%.*]] = add i32 [[Z]], [[Y]]
+; CHECK-NEXT:    ret i32 [[Q]]
+;
+  %Z = sub i32 %X, %Y
+  %Q = add i32 %Z, %Y
+  ret i32 %Q
+}

llvm/test/tools/UpdateTestChecks/update_test_checks/Inputs/basic.ll.funcsig.expected

 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
 ; Example input for update_llc_test_checks (taken from test/Transforms/InstSimplify/add.ll)
 ; RUN: opt < %s -instsimplify -S | FileCheck %s
 
 define i32 @common_sub_operand(i32 %X, i32 %Y) {
 ; CHECK-LABEL: define {{[^@]+}}@common_sub_operand
 ; CHECK-SAME: (i32 [[X:%.*]], i32 [[Y:%.*]])
-; CHECK-NEXT:    ret i32 [[X:%.*]]
+; CHECK-NEXT:    ret i32 [[X]]
 ;
   %Z = sub i32 %X, %Y
   %Q = add i32 %Z, %Y
   ret i32 %Q
 }
 
 define i32 @negated_operand(i32 %x) {
 ; CHECK-LABEL: define {{[^@]+}}@negated_operand

llvm/test/tools/UpdateTestChecks/update_test_checks/argument_name_reuse.test

+# Verify we reuse the argument regexps with --function-signature
+# RUN: cp -f %S/Inputs/argument_name_reuse.ll %t.ll && %update_test_checks %t.ll --function-signature
+# RUN: diff -u %t.ll %S/Inputs/argument_name_reuse.ll.expected
+# Verify we do not reuse the argument regexps, as there are none, without --function-signature
+# RUN: cp -f %S/Inputs/argument_name_reuse.ll %t.ll && %update_test_checks %t.ll
+# RUN: diff -u %t.ll %S/Inputs/argument_name_reuse.ll.plain.expected

llvm/utils/UpdateTestChecks/common.py

def get_value_definition(var):
   return '[[' + get_value_name(var) + ':%.*]]'
 
 
 # Use a FileCheck variable.
 def get_value_use(var):
   return '[[' + get_value_name(var) + ']]'
 
 # Replace IR value defs and uses with FileCheck variables.
-def genericize_check_lines(lines, is_analyze):
+def genericize_check_lines(lines, is_analyze, vars_seen):
   # This gets called for each match that occurs in
   # a line. We transform variables we haven't seen
   # into defs, and variables we have seen into uses.
   def transform_line_vars(match):
     var = match.group(2)
     if var in vars_seen:
       rv = get_value_use(var)
     else:
       vars_seen.add(var)
       rv = get_value_definition(var)
     # re.sub replaces the entire regex match
     # with whatever you return, so we have
     # to make sure to hand it back everything
     # including the commas and spaces.
     return match.group(1) + rv + match.group(3)
 
-  vars_seen = set()
   lines_with_def = []
 
   for i, line in enumerate(lines):
     # An IR variable named '%.' matches the FileCheck regex string.
     line = line.replace('%.', '%dot')
     # Ignore any comments, since the check lines will too.
     scrubbed_line = SCRUB_IR_COMMENT_RE.sub(r'', line)
     if is_analyze:
for checkprefix in checkprefixes:
         continue
 
       # Add some space between different check prefixes, but not after the last
       # check line (before the test code).
       if is_asm:
         if len(printed_prefixes) != 0:
           output_lines.append(comment_marker)
 
+      vars_seen = set()
       printed_prefixes.append(checkprefix)
       args_and_sig = str(func_dict[checkprefix][func_name].args_and_sig)
-      args_and_sig = genericize_check_lines([args_and_sig], is_analyze)[0]
+      args_and_sig = genericize_check_lines([args_and_sig], is_analyze, vars_seen)[0]
       if '[[' in args_and_sig:
         output_lines.append(check_label_format % (checkprefix, func_name, ''))
         output_lines.append('%s %s-SAME: %s' % (comment_marker, checkprefix, args_and_sig))
       else:
         output_lines.append(check_label_format % (checkprefix, func_name, args_and_sig))
       func_body = str(func_dict[checkprefix][func_name]).splitlines()
 
       # For ASM output, just emit the check lines.
       if is_asm:
         output_lines.append('%s %s:       %s' % (comment_marker, checkprefix, func_body[0]))
         for func_line in func_body[1:]:
           if func_line.strip() == '':
             output_lines.append('%s %s-EMPTY:' % (comment_marker, checkprefix))
           else:
             output_lines.append('%s %s-NEXT:  %s' % (comment_marker, checkprefix, func_line))
         break
 
       # For IR output, change all defs to FileCheck variables, so we're immune
       # to variable naming fashions.
-      func_body = genericize_check_lines(func_body, is_analyze)
+      func_body = genericize_check_lines(func_body, is_analyze, vars_seen)
 
       # This could be selectively enabled with an optional invocation argument.
       # Disabled for now: better to check everything. Be safe rather than sorry.
 
       # Handle the first line of the function body as a special case because
       # it's often just noise (a useless asm comment or entry label).
       #if func_body[0].startswith("#") or func_body[0].startswith("entry:"):
       #  is_blank_line = True