Re: c/7871: ICE on legal code, global register variables problems

[Jim Wilson <wilson@tuliptree.org>]

The following reply was made to PR c/7871; it has been noted by GNATS.

From: Jim Wilson <wilson@tuliptree.org>
To: Richard Zidlicky <rz@linux-m68k.org>
Cc: Richard Henderson <rth@redhat.com>, gcc-gnats@gcc.gnu.org,
   gcc-bugs@gcc.gnu.org, wilson@gcc.gnu.org
Subject: Re: c/7871: ICE on legal code, global register variables problems
Date: 06 Apr 2003 15:37:51 -0400

 On Wed, 2003-04-02 at 15:49, Richard Zidlicky wrote:
 > how does this differ from =C2=B4normal=C2=B4 global variables?
 
 Normal global variables are stored in memory instead of in registers.
 We don't add REG_NOTES and/or LOG_LINKS for values in memory like we do
 for registers.  Also, most optimization passes known that memory is
 volatile across function calls, but they assume (call-saved) registers
 will be preserved.  Thus we need special treatment for global register
 variables.
 
 > my application is still miscompiled with this patch.=20
 
 OK, so we need more work here.
 
 > Tested only gcc-3.2, should this make any difference?
 
 No.
 
 I haven't had a chance to do any more work on this problem since my
 original message, and won't for a while.  I'm in the middle of a 4000 km
 international move, and will be offline for the next week or two.
 
 Jim