Make sure we get a matching number of #if and #endifs before copying function to output

glibc has some weird cornercases in `sysdeps/generic/ldsodefs.h` where
a struct definition depends of a macro definition (SHARED), but its
fields attributes does not.  Here we devise an heuristic to try to
detect such cases by looking if the number of #ifs and #endifs matches
in the function.  If not, then fallback to AST dumping.

Closes #147

Signed-off-by: Giuliano Belinassi <gbelinassi@suse.de>

Author: giulianobelinassi

libcextract/LLVMMisc.cpp

@@ -315,3 +315,55 @@ bool Is_Decl_Equivalent_To(Decl *a, Decl *b)
 
   return a_str == b_str;
 }
+
+#define TOKEN_VECTOR " ().,;+-*/^|&{}[]<>^&|!\r\n\t"
+
+/** Check if string has unmatched #if, #ifdef, #ifndef.  */
+bool Has_Balanced_Ifdef(const StringRef &string)
+{
+  /* Create a temporary buffer for strtok and copy the string.  */
+  size_t len = string.size();
+  char buf[len + 1];
+  memcpy(buf, string.data(), len);
+  buf[len] = '\0';
+
+  /* Count the number of parenthesis problem. ifdef, ifndef, if increases,
+     endif decreases.  */
+  int balance = 0;
+
+  /* Tokenize.  */
+  bool in_comments = false;
+  char *tok = strtok(buf, TOKEN_VECTOR);
+  while (tok != nullptr) {
+    /* There is the silly case in which #ifdef is written as `#  ifdef`.  */
+    if (*tok == '#') {
+      /* Now check if it got into the same token or if we need to pull another
+         token.  */
+      if (tok[1] == '\0') {
+        tok = strtok(nullptr, TOKEN_VECTOR);
+      } else {
+        tok++;
+      }
+
+      assert(tok != nullptr && "tok is null! why?");
+
+      /* Case 1: #ifdef, #ifndef, and #if has "if" as prefix.  */
+      if (prefix("if", tok)) {
+        balance++;
+      }
+
+      /* Case 2: #endif.  Decrease the balance counter.  */
+      else if (strcmp("endif", tok) == 0) {
+        balance--;
+        if (balance < 0) {
+          /* Impossible.  This means there is an #endif for no matching #if.  */
+          return false;
+        }
+      }
+    }
+
+    tok = strtok(nullptr, TOKEN_VECTOR);
+  }
+
+  return balance == 0;
+}

libcextract/LLVMMisc.hh

@@ -109,3 +109,6 @@ DeclContextLookupResult Get_Decl_From_Symtab(ASTUnit *ast, const StringRef &name
 
 /** Check if two Decls are equivalent.  */
 bool Is_Decl_Equivalent_To(Decl *a, Decl *b);
+
+/** Check if string has unmatched #if, #ifdef, #ifndef.  */
+bool Has_Balanced_Ifdef(const StringRef &string);

libcextract/PrettyPrint.cpp

@@ -191,7 +191,29 @@ void PrettyPrint::Print_Decl_Raw(Decl *decl)
         }
       }
     } else {
-      Out << decl_source;
+      /* Check if the output string has a balanced number of ifdefs.  This is
+         required because of the following construct from glibc:
+
+          #ifdef !SHARED
+          #else
+          struct rtld_global_ro {
+          #endif
+
+         In case we want to print rtld_global_ro, PrettyPrint will get the text
+         from struct rtld_global_ro and below, which would only include the
+         #endif, and not the #ifdef.  In case its not balanced we fallback to
+         AST dumping.  */
+      if (NamedDecl *d = dyn_cast<NamedDecl>(decl)) {
+        if (d->getName() == "rtld_global_ro") {
+          llvm::outs() << "debug me \n";
+        }
+      }
+      if (Has_Balanced_Ifdef(decl_source)) {
+        Out << decl_source;
+      } else {
+        /* In case its not balanced, fallback to AST dump.  */
+        decl->print(Out, LangOpts);
+      }
     }
   } else {
     /* Else, we fallback to AST Dumping.  */

testsuite/small/ifdef-1.c

@@ -1,7 +1,4 @@
 /* { dg-options "-DCE_EXTRACT_FUNCTIONS=function -DCE_NO_EXTERNALIZATION" }*/
-/* { dg-xfail } Tests fails because we currently can't track that the endif
-   comes from #ifdef AAA.  */
-
 
 //#define AA
 

testsuite/small/ifdef-2.c

@@ -0,0 +1,30 @@
+/* { dg-options "-DCE_EXTRACT_FUNCTIONS=f -DCE_NO_EXTERNALIZATION" }*/
+
+#define SHARED
+
+#define IS_IN(x) 0
+
+#ifndef SHARED
+# define GLRO(name) _##name
+#else
+# if IS_IN (rtld)
+#  define GLRO(name) _rtld_local_ro._##name
+# else
+#  define GLRO(name) _rtld_global_ro._##name
+# endif
+struct rtld_global_ro
+{
+#endif
+  int field1;
+#ifdef SHARED
+  int field2;
+};
+#endif
+
+
+int f(struct rtld_global_ro *r)
+{
+  return r->field1;
+}
+
+/* { dg-final { scan-tree-dump "struct rtld_global_ro ?\n?{" } } */