Potential bug in stmt.c (expand_value_return)
Richard Henderson
rth@cygnus.com
Wed Feb 2 13:28:00 GMT 2000
On Wed, Feb 02, 2000 at 06:43:23PM +0000, Joern Rennecke wrote:
> stupid was removed, so it stands reason that all the USEs that were
> emitted only for the benefit of stuid are no longer needed.
> However, the USEs of the return value are not only needed by stupid, but
> also by some optimizations, so I suppose this removal was accidential.
No, it wasn't accidental. The information is available instead
from EXIT_BLOCK_PTR->global_live_at_start.
What optimizations need the actual use insn?
r~
More information about the Gcc
mailing list