From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <gcc-bugzilla@gcc.gnu.org>
Received: by sourceware.org (Postfix, from userid 48)
	id 865D23858D20; Sun, 21 May 2023 09:09:51 +0000 (GMT)
DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org 865D23858D20
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gcc.gnu.org;
	s=default; t=1684660191;
	bh=KmF7ISRcJKmFRWXCJAbv2XX5nOKhaA4QEmrP/xKhy7A=;
	h=From:To:Subject:Date:In-Reply-To:References:From;
	b=bwS+txTr4dXkexZ3wRRMyJnPyL4fQPkRHDaSdSWs0ae5SMQNlYjhB7V+oHI4mVuFY
	 8E04yE2WBsulbfs6KXmFVb9+qZRqWsmaRKAqIfio0LHmNJpqWnn77N/h6OwqMhGyFV
	 fcPGhB21lXIzVjorKPhhtZg88DkzeZWrq7lWWsm0=
From: "gjl at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug middle-end/109907] Missed optimization for bit extraction (uses
 shift instead of single bit-test)
Date: Sun, 21 May 2023 09:09:49 +0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: middle-end
X-Bugzilla-Version: 14.0
X-Bugzilla-Keywords: missed-optimization
X-Bugzilla-Severity: normal
X-Bugzilla-Who: gjl at gcc dot gnu.org
X-Bugzilla-Status: NEW
X-Bugzilla-Resolution: 
X-Bugzilla-Priority: P3
X-Bugzilla-Assigned-To: unassigned at gcc dot gnu.org
X-Bugzilla-Target-Milestone: ---
X-Bugzilla-Flags: 
X-Bugzilla-Changed-Fields: attachments.created
Message-ID: <bug-109907-4-xtei0mbiUn@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-109907-4@http.gcc.gnu.org/bugzilla/>
References: <bug-109907-4@http.gcc.gnu.org/bugzilla/>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/
Auto-Submitted: auto-generated
MIME-Version: 1.0
List-Id: <gcc-bugs.sourceware.org>

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=3D109907

--- Comment #17 from Georg-Johann Lay <gjl at gcc dot gnu.org> ---
Created attachment 55129
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=3D55129&action=3Dedit
Patch for AVR backend: combine patterns, "extzv<mode>", test case

For now I have the attaches patch that resolves all cases of the included
test-case.  Iknow that the maintainers are not very fond of such combine
patterns though...

<log>
Provide more patterns to improve extracting (negated) single bits.
Deprecated insn "extzv" is replaced by "extzv<mode>".  The patterns now
also allow lower I/O memory because both the insns and the instructions
for I/O like SBIC/SBIS and the ones for REGs like SBRC/SBRS.
Even in the presence of "extzv<mode>" there is no canonical form for
single bit extractions, even less for extracting the insersed bit.
So more than one mathod is provided.

gcc/
        PR target/109907
        * config/avr/avr-protos.h (avr_out_extr, avr_out_extr_not): New pro=
tos.
        * config/avr/avr.cc (avr_out_extr, avr_out_extr_not): New functions.
        (avr_adjust_insn_length) [ADJUST_LEN_EXTR_NOT, ADJUST_LEN_EXTR]:
        Handle cases.
        * config/avr/avr.md (adjust_len) [extr, extr_not]: Add to define_at=
tr.
        (extzv): Turn into extzv<mode>.
        (MSB): New mode attribute.
        (*extzv_split, *extzv): Allow lower I/O addresses in operand 1.
        Unify constraints to a single case, as avr_out_extr() will now
        handle the alternatives.
        (*extzv.not_split, *extzv_not): New insn and its post-reload split.
        (*extzv.subreg.<mode>, *extzv.neg.subreg-msb.<mode>, *extzv<mode>.g=
e):
        New insns and post-reload splits.
        (*extzv.xor, *extzv.io.lsr): New insns and pre-reload split.
        * config/avr/constraints.md (Yil): New constraint for reg or low I/=
O.
        * config/avr/predicates.md (reg_or_low_io_operand, const7_operand)
        (const15_operand, const23_operand, const31_operand)
        (const_0_to_7_operand, const_0_to_15_operand, const_0_to_23_operand)
        (const_0_to_31_operand): New predicates.

gcc/testsuite/
        PR target/109907
        * gcc.target/avr/pr109907.c: New test.
</log>=