Re: [PATCH][LTO] Committed patch to fix a crash due to syntax error.

["Richard Guenther" <richard dot guenther at gmail dot com>]

On Thu, Oct 9, 2008 at 9:00 AM, Doug Kwan (關振紱) <dougkwan@google.com> wrote:
> I have just commited this trivial patch to make
> gimple_try_catch_may_fallthru more robust in the presence of syntax
> errors in input.

Does this mean -flto writes out translation units which contain
errors?

Thanks,
Richard.

> -Doug
>
> Index: gcc/gcc/ChangeLog.lto
> ===================================================================
> --- gcc/gcc/ChangeLog.lto       (revision 140999)
> +++ gcc/gcc/ChangeLog.lto       (working copy)
> @@ -1,3 +1,8 @@
> +2008-10-09  Doug Kwan  <dougkwan@google.com>
> +
> +       * gimple-low.c (gimple_try_catch_may_fallthru): Check for empty
> +       clean-up GIMPLE sequence.
> +
>  2008-10-08 Rafael Espindola  <espindola@google.com>
>
>        * lto-tree-in.h (lto_symtab_get_resolution): New.
> Index: gcc/gcc/gimple-low.c
> ===================================================================
> --- gcc/gcc/gimple-low.c        (revision 140999)
> +++ gcc/gcc/gimple-low.c        (working copy)
> @@ -453,6 +453,15 @@ gimple_try_catch_may_fallthru (gimple st
>     return true;
>
>   i = gsi_start (gimple_try_cleanup (stmt));
> +
> +  /* If we have a syntax error, the cleanup sequence can be empty.
> +     Return FALSE instead of crashing the compiler.  */
> +  if (gsi_end_p (i))
> +    {
> +      gcc_assert (errorcount);
> +      return false;
> +    }
> +
>   switch (gimple_code (gsi_stmt (i)))
>     {
>     case GIMPLE_CATCH:
>