This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: Fix minor bugs in expr.c
- To: jh at suse dot cz
- Subject: Re: Fix minor bugs in expr.c
- From: kenner at vlsi1 dot ultra dot nyu dot edu (Richard Kenner)
- Date: Tue, 13 Nov 01 06:35:44 EST
- Cc: gcc-patches at gcc dot gnu dot org
May I ask for rationale to conditionalize this by INSN_SCHEDULING? I
know that combine.c does too,
It needs to match recog.c. force_operand must create a valid operand and
it isn't due to not having this code.
but I still don't see why this should be scheduling specific. I see
that it avoids instruction to read/write more than they actually need.
This can cuase memory traps for unaligned data and possibly stalls if
machine does on-chip scheduling of loads/stores, but still...
I don't remember why this was done, but know it was long ago.