This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
[PATCH] Fix PR c/9262
- From: Eric Botcazou <ebotcazou at libertysurf dot fr>
- To: gcc-patches at gcc dot gnu dot org
- Date: Fri, 7 Feb 2003 01:28:58 +0100
- Subject: [PATCH] Fix PR c/9262
Hi,
I really wonder how one can write such a broken code but a segfault is a
little bit brutal and it's a regression from gcc 2.95.3 so...
The functions pushcase and pushcase_range from stmt.c (not used by the C and
C++ front-ends) are guarded by a sanity check, while add_case_node (directly
called by c-semantics.c) is not. I simply copy-n-pasted it from the formers
to the latter.
Bootstrapped/regtested (c,c++,objc,f77 3.3 branch) on i586-redhat-linux-gnu.
Ok for mainline and 3.3 ? for 3.2 too ?
--
Eric Botcazou
2003-02-07 Eric Botcazou <ebotcazou@libertysurf.fr>
PR c/9262
* stmt.c (add_case_node): Check that we are inside a real case
statement.
2003-02-07 Eric Botcazou <ebotcazou@libertysurf.fr>
* gcc.dg/switch-2.c: New test.
Index: stmt.c
===================================================================
RCS file: /cvs/gcc/gcc/gcc/stmt.c,v
retrieving revision 1.276.2.2
diff -u -r1.276.2.2 stmt.c
--- stmt.c 21 Jan 2003 07:40:02 -0000 1.276.2.2
+++ stmt.c 6 Feb 2003 23:51:32 -0000
@@ -4666,6 +4666,10 @@
{
struct case_node *p, **q, *r;
+ /* Fail if not inside a real case statement. */
+ if (! (case_stack && case_stack->data.case_stmt.start))
+ return 1;
+
/* If there's no HIGH value, then this is not a case range; it's
just a simple case label. But that's just a degenerate case
range. */
/* PR c/9262 */
/* Originator: Rasmus Hahn <rassahah@neofonie.de> */
/* { dg-do compile } */
int foo(int i)
{
switch (i)
case 3:
return 1,
case 4: /* { dg-error "parse error" } */
return 2;
}