This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
Re: mips-sgi-irix6.5SE doesn't bootstrap
- To: Joern Rennecke <amylaar at cygnus dot co dot uk>
- Subject: Re: mips-sgi-irix6.5SE doesn't bootstrap
- From: Jim Wilson <wilson at cygnus dot com>
- Date: Thu, 21 May 1998 11:44:24 -0700
- cc: law at cygnus dot com, mmitchell at usa dot net, egcs-bugs at cygnus dot com, amylaar at cygnus dot com, hjl at lucon dot org
! We compute TIME1 as the time when the register for the prospective
! new reload will become live, and TIME2 for each existing
! reload as the time when that the reload register of that reload
! ceases to be live.
I found this comment to be very confusing. TIME1 is when the current reload
(TYPE) dies, and hence its register becomes available for other reloads.
Similarly TIME2 is when other reloads will load a value into their register,
and hence it is when they become live.
Death must occur before a new lifetime can start, hence
if (time1 >= time2)
return 0;
Calling a reload register `live' when it has no value, and `dead' when it is
assigned a value is opposite the usual meaning for the terms, and hence
confusing.
Otherwise, the patch is OK.
Jim