Merge pull request #338 from slothy-optimizer/multi-line-comments

Add support for single-line and multi-line /* */ comments

Author: hanno-becker

slothy/helper.py

@@ -37,6 +37,10 @@
 
 from unicorn import Uc, UcError
 
+# Markers for /* */ comment preservation
+_BLOCK_COMMENT_MARKER = "__SLOTHY_BLOCK__"
+_NEWLINE_MARKER = "__SLOTHY_NL__"
+
 
 class SourceLine:
     """Representation of a single line of source code"""
@@ -232,7 +236,21 @@ def to_string(self, indentation=True, comments=True, tags=True):
         additional = []
 
         if comments is True:
-            additional += list(map(lambda s: f"// {s}", double_comments))
+            # Convert comments with markers back to /* */ format
+            def format_comment(s):
+                if _NEWLINE_MARKER in s:
+                    # Restore as multi-line /* */ comment
+                    lines = s.split(_NEWLINE_MARKER)
+                    lines = [line.strip() for line in lines]
+                    return "/* " + "\n   ".join(lines) + " */"
+                elif s.startswith(_BLOCK_COMMENT_MARKER):
+                    # Restore as single-line /* */ comment
+                    return f"/* {s[len(_BLOCK_COMMENT_MARKER):]} */"
+                else:
+                    # Regular // comment - keep as is
+                    return f"// {s}"
+
+            additional += list(map(format_comment, double_comments))
             additional += list(map(lambda s: f"///{s}", triple_comments))
 
         if tags is True:
@@ -346,6 +364,21 @@ def read_multiline(s, reduce=True):
         if isinstance(s, str):
             # Retain newline termination
             terminated_by_newline = len(s) > 0 and s[-1] == "\n"
+
+            # Convert /* */ comments to // style with markers for restoration
+            def replace_comment(match):
+                content = match.group(1)
+                if "\n" in content:
+                    # Multi-line /* */ comment - mark newlines
+                    content = content.replace("\n", f" {_NEWLINE_MARKER} ")
+                    content = " ".join(content.split())
+                    return f"// {content}" if content else ""
+                else:
+                    # Single-line /* */ comment - mark with prefix
+                    content = content.strip()
+                    return f"// {_BLOCK_COMMENT_MARKER}{content}" if content else ""
+
+            s = re.sub(r"/\*(.*?)\*/", replace_comment, s, flags=re.DOTALL)
             s = s.splitlines()
             if terminated_by_newline:
                 s.append("")

tests/naive/aarch64/_test.py

@@ -287,6 +287,21 @@ def core(self, slothy):
         slothy.fusion_region(start="start", end="end", ssa=False)
 
 
+class AArch64CStyleComments(OptimizationRunner):
+    def __init__(self, var="", arch=AArch64_Neon, target=Target_CortexA55):
+        name = "aarch64_cstyle_comments"
+        infile = name
+
+        super().__init__(
+            infile, name, rename=True, arch=arch, target=target, base_dir="tests"
+        )
+
+    def core(self, slothy):
+        slothy.config.variable_size = True
+        slothy.config.constraints.stalls_first_attempt = 32
+        slothy.optimize(start="start", end="end")
+
+
 test_instances = [
     Instructions(),
     Instructions(target=Target_CortexA72),
@@ -309,4 +324,5 @@ def core(self, slothy):
     AArch64LoopLabels(),
     AArch64LoopBranch(),
     AArch64FusionVeor(),
+    AArch64CStyleComments(),
 ]

tests/naive/aarch64/aarch64_cstyle_comments.s

@@ -0,0 +1,34 @@
+/* Comment before optimization region */
+// Another comment before
+ldr q0, [x1, #0]  /* This should be preserved */
+
+start:
+/* Single line C-style comment */
+ldr q1, [x2, #0]  /* inline comment */
+
+/* Multi-line with
+   // what looks like a comment
+   but should be ignored */
+ldr q8, [x0]
+ldr q9, [x0, #1*16]
+ldr q10, [x0, #2*16]
+ldr q11, [x0, #3*16]
+
+mul v24.8h, v9.8h, v0.h[0]  /* first */ /* second */ // third
+sqrdmulh v9.8h, v9.8h, v0.h[1]
+mls v24.8h, v9.8h, v1.h[0]
+sub v9.8h, v8.8h, v24.8h
+add v8.8h, v8.8h, v24.8h
+
+/**
+ * Javadoc style before stores
+ */
+str q8, [x0], #4*16
+str q9, [x0, #-3*16]  /* store with offset */
+str q10, [x0, #-2*16]
+str q11, [x0, #-1*16]
+end:
+
+/* Comment after optimization region */
+// Another comment after
+str q0, [x1]  /* This should also be preserved */