This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: SSA_NAME remaining at RTL generation
- From: Daniel Berlin <dberlin at dberlin dot org>
- To: kenner at vlsi1 dot ultra dot nyu dot edu (Richard Kenner)
- Cc: gcc at gcc dot gnu dot org
- Date: Sun, 4 Jul 2004 10:31:56 -0400
- Subject: Re: SSA_NAME remaining at RTL generation
- References: <10407041407.AA13182@vlsi1.ultra.nyu.edu>
On Jul 4, 2004, at 10:07 AM, Richard Kenner wrote:
This is a new one for me. It started showing up in the last 24 hours
or so when compiling gnatls.adb.
From t55.optimized:
<L11>:;
object_name$P_ARRAY = (character[(long int)<PLACEHOLDER_EXPR struct
gnat__str\
ings__string_access>.P_BOUNDS->LB0 .. (long int)<PLACEHOLDER_EXPR
struct gnat__\
strings__string_access>.P_BOUNDS->UB0] *)&CHAIN.1129->no_obj.ARRAY;
This is invalid GIMPLE right now, actually.
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15848
The addr_expr should be separate from the cast.
We seem to screw this up right now when the ADDR_EXPR is of a constant.
Is CHAIN.1129->no_obj.ARRAY constant?
--Dan