From mboxrd@z Thu Jan 1 00:00:00 1970 From: Richard Henderson To: law@redhat.com Cc: Jan Hubicka , gcc@gcc.gnu.org, m.hayes@elec.canterbury.ac.nz, matzmich@cs.tu-berlin.de, dan@cgsoftware.com Subject: Re: df.c and partial writes/REG_EQUAL notes Date: Wed, 24 Oct 2001 13:00:00 -0000 Message-id: <20011024130019.A20204@redhat.com> References: <20010925161602.C13734@atrey.karlin.mff.cuni.cz> <3257.1003952277@localhost.localdomain> X-SW-Source: 2001-10/msg01248.html On Wed, Oct 24, 2001 at 01:37:57PM -0600, law@redhat.com wrote: > Right. Review the movxx section in the Machine Descriptions part of the > manual. Quoting: [...] > If operand 0 is a @code{subreg} with mode @var{m} of a register whose > own mode is wider than @var{m}, the effect of this instruction is > to store the specified value in the part of the register that corresponds > to mode @var{m}. The effect on the rest of the register is undefined. > > Based on that I believe there should be no dependency since insn #2 > leaves the rest of the DI register undefined. The documentation is incomplete. The rest of the register up to BITS_PER_WORD are undefined. Other words of a multi-word pseudo are no undefined. r~