This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: The "980505-1.c" bug (The Story Continued)
- To: Carlo Wood <carlo at runaway dot xs4all dot nl>
- Subject: Re: The "980505-1.c" bug (The Story Continued)
- From: Jeffrey A Law <law at cygnus dot com>
- Date: Fri, 10 Jul 1998 23:54:18 -0600
- cc: egcs at cygnus dot com (egcs at cygnus dot com)
- Reply-To: law at cygnus dot com
In message <199807100056.CAA11845@jolan.ppro>you write:
> It might therefore be sufficient to just detect the addition of an expression
> with an _uninitialized_ register, and than not add it to the table at all.
> (This can be detected by checking if reg_in_table[reg_no] becomes 0 after
> incrementing it).
Note determining whether or not a reg or memory location is initialized
is not generally a solveable problem.
Even within the context of GCC's local CSE pass it is not a solveable
problem since the reg or memory could be set in a flow predecessor
block -- remember local CSE throws away all information at block boundaries
So it will not know that the register was set in a previous block.
jeff