1 files changed, 90 insertions, 17 deletions

diff --git a/scripts/genksyms/lex.l b/scripts/genksyms/lex.l
index fe50ff9dacd..e583565f201 100644
--- a/scripts/genksyms/lex.l
+++ b/scripts/genksyms/lex.l
@@ -29,7 +29,7 @@
 #include <ctype.h>
 
 #include "genksyms.h"
-#include "parse.h"
+#include "parse.tab.h"
 
 /* We've got a two-level lexer here.  We let flex do basic tokenization
    and then we categorize those basic tokens in the second stage.  */
@@ -55,10 +55,6 @@ CHAR			L?\'([^\\\']*\\.)*[^\\\']*\'
 
 MC_TOKEN		([~%^&*+=|<>/-]=)|(&&)|("||")|(->)|(<<)|(>>)
 
-/* Version 2 checksumming does proper tokenization; version 1 wasn't
-   quite so pedantic.  */
-%s V2_TOKENS
-
 /* We don't do multiple input files.  */
 %option noyywrap
 
@@ -84,9 +80,9 @@ MC_TOKEN		([~%^&*+=|<>/-]=)|(&&)|("||")|(->)|(<<)|(>>)
     recognized as tokens.  We don't actually use them since we don't
     parse expressions, but we do want whitespace to be arranged
     around them properly.  */
-<V2_TOKENS>{MC_TOKEN}			return OTHER;
-<V2_TOKENS>{INT}			return INT;
-<V2_TOKENS>{REAL}			return REAL;
+{MC_TOKEN}				return OTHER;
+{INT}					return INT;
+{REAL}					return REAL;
 
 "..."					return DOTS;
 
@@ -98,17 +94,29 @@ MC_TOKEN		([~%^&*+=|<>/-]=)|(&&)|("||")|(->)|(<<)|(>>)
 
 /* Bring in the keyword recognizer.  */
 
-#include "keywords.c"
+#include "keywords.hash.c"
 
 
 /* Macros to append to our phrase collection list.  */
 
+/*
+ * We mark any token, that that equals to a known enumerator, as
+ * SYM_ENUM_CONST. The parser will change this for struct and union tags later,
+ * the only problem is struct and union members:
+ *    enum e { a, b }; struct s { int a, b; }
+ * but in this case, the only effect will be, that the ABI checksums become
+ * more volatile, which is acceptable. Also, such collisions are quite rare,
+ * so far it was only observed in include/linux/telephony.h.
+ */
 #define _APP(T,L)	do {						   \
 			  cur_node = next_node;				   \
 			  next_node = xmalloc(sizeof(*next_node));	   \
 			  next_node->next = cur_node;			   \
 			  cur_node->string = memcpy(xmalloc(L+1), T, L+1); \
-			  cur_node->tag = SYM_NORMAL;			   \
+			  cur_node->tag =				   \
+			    find_symbol(cur_node->string, SYM_ENUM_CONST, 1)?\
+			    SYM_ENUM_CONST : SYM_NORMAL ;		   \
+			  cur_node->in_source_file = in_source_file;       \
 			} while (0)
 
 #define APP		_APP(yytext, yyleng)
@@ -121,8 +129,9 @@ int
 yylex(void)
 {
   static enum {
-    ST_NOTSTARTED, ST_NORMAL, ST_ATTRIBUTE, ST_ASM, ST_BRACKET, ST_BRACE,
-    ST_EXPRESSION, ST_TABLE_1, ST_TABLE_2, ST_TABLE_3, ST_TABLE_4,
+    ST_NOTSTARTED, ST_NORMAL, ST_ATTRIBUTE, ST_ASM, ST_TYPEOF, ST_TYPEOF_1,
+    ST_BRACKET, ST_BRACE, ST_EXPRESSION,
+    ST_TABLE_1, ST_TABLE_2, ST_TABLE_3, ST_TABLE_4,
     ST_TABLE_5, ST_TABLE_6
   } lexstate = ST_NOTSTARTED;
 
@@ -134,7 +143,6 @@ yylex(void)
 
   if (lexstate == ST_NOTSTARTED)
     {
-      BEGIN(V2_TOKENS);
       next_node = xmalloc(sizeof(*next_node));
       next_node->next = NULL;
       lexstate = ST_NORMAL;
@@ -160,6 +168,13 @@ repeat:
       cur_filename = memcpy(xmalloc(e-file+1), file, e-file+1);
       cur_line = atoi(yytext+2);
 
+      if (!source_file) {
+        source_file = xstrdup(cur_filename);
+        in_source_file = 1;
+      } else {
+        in_source_file = (strcmp(cur_filename, source_file) == 0);
+      }
+
       goto repeat;
     }
 
@@ -184,11 +199,15 @@ repeat:
 		    lexstate = ST_ASM;
 		    count = 0;
 		    goto repeat;
+		  case TYPEOF_KEYW:
+		    lexstate = ST_TYPEOF;
+		    count = 0;
+		    goto repeat;
 
 		  case STRUCT_KEYW:
 		  case UNION_KEYW:
-		    dont_want_brace_phrase = 3;
 		  case ENUM_KEYW:
+		    dont_want_brace_phrase = 3;
 		    suppress_type_lookup = 2;
 		    goto fini;
 
@@ -198,8 +217,7 @@ repeat:
 	      }
 	    if (!suppress_type_lookup)
 	      {
-		struct symbol *sym = find_symbol(yytext, SYM_TYPEDEF);
-		if (sym && sym->type == SYM_TYPEDEF)
+		if (find_symbol(yytext, SYM_TYPEDEF, 1))
 		  token = TYPE;
 	      }
 	  }
@@ -271,6 +289,48 @@ repeat:
 	}
       break;
 
+    case ST_TYPEOF:
+      switch (token)
+	{
+	case '(':
+	  if ( ++count == 1 )
+	    lexstate = ST_TYPEOF_1;
+	  else
+	    APP;
+	  goto repeat;
+	case ')':
+	  APP;
+	  if (--count == 0)
+	    {
+	      lexstate = ST_NORMAL;
+	      token = TYPEOF_PHRASE;
+	      break;
+	    }
+	  goto repeat;
+	default:
+	  APP;
+	  goto repeat;
+	}
+      break;
+
+    case ST_TYPEOF_1:
+      if (token == IDENT)
+	{
+	  if (is_reserved_word(yytext, yyleng)
+	      || find_symbol(yytext, SYM_TYPEDEF, 1))
+	    {
+	      yyless(0);
+	      unput('(');
+	      lexstate = ST_NORMAL;
+	      token = TYPEOF_KEYW;
+	      break;
+	    }
+	  _APP("(", 1);
+	}
+	APP;
+	lexstate = ST_TYPEOF;
+	goto repeat;
+
     case ST_BRACKET:
       APP;
       switch (token)
@@ -318,7 +378,20 @@ repeat:
 	  ++count;
 	  APP;
 	  goto repeat;
-	case ')': case ']': case '}':
+	case '}':
+	  /* is this the last line of an enum declaration? */
+	  if (count == 0)
+	    {
+	      /* Put back the token we just read so's we can find it again
+		 after registering the expression.  */
+	      unput(token);
+
+	      lexstate = ST_NORMAL;
+	      token = EXPRESSION_PHRASE;
+	      break;
+	    }
+	  /* FALLTHRU */
+	case ')': case ']':
 	  --count;
 	  APP;
 	  goto repeat;