This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Combine deletes non noop insns
- From: "John David Anglin" <dave at hiauly1 dot hia dot nrc dot ca>
- To: rth at redhat dot com (Richard Henderson)
- Cc: gcc at gcc dot gnu dot org
- Date: Tue, 10 Sep 2002 15:11:16 -0400 (EDT)
- Subject: Re: Combine deletes non noop insns
> On Tue, Sep 10, 2002 at 01:26:41PM -0400, John David Anglin wrote:
> > (insn 42 41 43 0 (nil) (set (zero_extract:SI (subreg:SI (reg:QI 125) 0)
> > (const_int 8 [0x8])
> > (const_int 24 [0x18]))
> > (reg:SI 97)) 220 {insv_32} (insn_list 4 (insn_list 41 (nil)))
>
> This seems odd to me. Inserting data into the ignored
> portion of a paradoxical subreg?
The data is inserted to the correct place but I see that the
subreg number is wrong. It's probably another problem
related to the right justification of small BLKmode types. How
the code that generated this managed to get the insert location
and the number bits to insert correct, but not the subreg is
strange.
Dave
--
J. David Anglin dave.anglin@nrc.ca
National Research Council of Canada (613) 990-0752 (FAX: 952-6605)