On Wed, Apr 18, 2012 at 2:44 PM, Richard Henderson <[email protected]> wrote:
> On 04/18/2012 05:39 AM, Jan Hubicka wrote:
>> Well, if SJLJ lowering happens as gimple pass somewhere near the end of
>> gimple
>> queue, this should not be problem at all. (and implementation would be
>> cleaner)
>
> If you can find a clean way of separating sjlj expansion from dw2 expansion,
> please do. But there's a lot of code shared between the two. I see nothing
> wrong with always expanding via tablejump.
In that case, would something like the following patch be acceptable?
I have verified that all targets that always set
TARGET_EXCEPT_UNWIND_INFO define casesi and/or tablejump.
* toplev.c (process_options): Fail for sjlj exceptions if the
target machine
has no casesi insn and no tablejump insn.
Index: toplev.c
===================================================================
--- toplev.c (revision 187394)
+++ toplev.c (working copy)
@@ -1265,6 +1265,16 @@ process_options (void)
"and -ftree-loop-linear)");
#endif
+#if !defined(HAVE_casesi) && !defined(HAVE_tablejump)
+ if (flag_exceptions
+ && targetm_common.except_unwind_info (&global_options) == UI_SJLJ)
+ {
+ sorry ("setjmp/longjmp-based exception handling not supported "
+ "on this target machine "
+ "(required %<casesi%> or %<tablejump%> insns are not present)");
+ }
+#endif
+
if (flag_mudflap && flag_lto)
sorry ("mudflap cannot be used together with link-time optimization");
