[tuples] Ada front-end changes for tuples

[Laurent GUERBY]

On Thu, 2008-06-19 at 23:52 +0200, Arnaud Charlet wrote:

> Note that if I disable the code that needs to be translated to your modified
> gimple API (the original code is actually not used that often), I then
> end up with a ICE in handling of exceptions compiling expr_attr.adb, which
> looks more like a possible bug/inconsistency in the middle-end. See below for
> the patch I applied to the branch (after having reverted your changes) to get
> to this ICE.
> 
> /users/charlet/gcc/obj/prev-gcc/gnat1 -nostdinc -I- -I. -Iada -I/users/charlet/gcc/gimple-tuples-branch/gcc/ada -O0 -gnatpg -quiet /users/charlet/gcc/gimple-tuples-branch/gcc/ada/exp_attr.adb -o exp_attr.o
> +===========================GNAT BUG DETECTED==============================+
> | 4.4.0 20080613 (experimental) (i686-pc-linux-gnu) GCC error:             |
> | in find_goto_replacement, at tree-eh.c:463                               |
> | Error detected around /chinon.a/users/charlet/gcc/gimple-tuples-branch/gcc/ada
> /rtsfind.ads:2848|             

I recently open a ticket about EH and middle-end on trunk Ada:

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36554
[4.4 regression] verify_flow_info ICE can not throw but has EH edges

It has a reduced test case of a dozen lines of Ada.

Note exactly the same error, but something is not quite right
in trunk EH handling for Ada.

Laurent