This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: AIX bootstrap failure due to libiberty configure ICE
- From: law at redhat dot com
- To: David Edelsohn <dje at watson dot ibm dot com>
- Cc: gcc at gcc dot gnu dot org
- Date: Tue, 11 Jun 2002 18:27:14 -0600
- Subject: Re: AIX bootstrap failure due to libiberty configure ICE
- Reply-to: law at redhat dot com
In message <200206120015.UAA36100@rios0.watson.ibm.com>, David Edelsohn writes:
> And the winner is: Jeff Law's RTL generation patch.
>
> 2002-06-10 Jeffrey Law <law@redhat.com>
>
> * alias.c (static_reg_base_value): New to hold RTL for
> items allocated once per function for the aliasing code.
> (init_alias_once_per_function): Initialize static_reg_base_value.
> (init_alias_analysis): Avoid throw-away allocations of RTL by
> using pre-computed values in static_reg_base_value.
> * function.c (prepare_function_start): Call
> init_alias_once_per_function appropriately.
> * rtl.h (init_alias_once_per_function): Declare.
> * caller-save (init_caller_save): Restructure slightly to
> avoid lots of silly RTL generation.
> * expr.c (init_expr_once): Likewise.
> * reload1.c (reload_cse_regs_1): Allocate throw-away register
> RTL object here. Pass it into children.
> (reload_cse_simplify_operands): Use passed-in register RTL
> object.
> (reload_cse_simplify): Pass through throw-away register
> RTL object.
>
> With this patch, stage2 GCC ICEs when compiling the "const" configure test
> from libiberty. stage1 GCC does not ICE. This is a regression and needs
> to be fixed or the patch reverted.
>
Can you check after Richard's patch to alias.c? That bug was causing a
number of "interesting" failures.
jeff