This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [rtl PATCH] PR/37889 and PR/38921, wrong code motion bugs
- From: Hans-Peter Nilsson <hans-peter dot nilsson at axis dot com>
- To: bonzini at gnu dot org
- Cc: gcc-patches at gcc dot gnu dot org, hans-peter dot nilsson at axis dot com
- Date: Wed, 4 Feb 2009 17:14:47 +0100
- Subject: Re: [rtl PATCH] PR/37889 and PR/38921, wrong code motion bugs
> Date: Wed, 04 Feb 2009 16:42:54 +0100
> From: Paolo Bonzini <bonzini@gnu.org>
Thanks for giving the patch and PR's love and care.
> This patch fixes two code motion bugs. The fix for PR/38921 comes
> first: basically the may_trap_p flags were overengineered, and the only
> combinations that made sense are 0 and
> MTP_UNALIGNED_MEMS|MTP_AFTER_MOVE. So H-P's part of the attached patch
> changes may_trap_p_1 to accept 0/1 as flags, and unifies
> may_trap_or_fault_p with may_trap_after_code_motion_p.
FWIW, this was how may_trap_or_fault_p used to be, at some
point. Its semantics was changed without updating the callers,
the cause for PR38921. (The "after_code_motion" was redundant;
we're always considering code where something has moved!) To
the reviewer: Zdenek suggested this re-unification, see the
audit trail.
brgds, H-P