From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <gcc-bugs-return-431239-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Received: (qmail 24939 invoked by alias); 7 Oct 2013 09:19:57 -0000
Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Id: <gcc-bugs.gcc.gnu.org>
List-Archive: <http://gcc.gnu.org/ml/gcc-bugs/>
List-Post: <mailto:gcc-bugs@gcc.gnu.org>
List-Help: <mailto:gcc-bugs-help@gcc.gnu.org>
Sender: gcc-bugs-owner@gcc.gnu.org
Received: (qmail 24773 invoked by uid 48); 7 Oct 2013 09:19:54 -0000
From: "gjl at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug middle-end/57503] [4.7/4.8 Regression] Expand uses wrong multiply routine
Date: Mon, 07 Oct 2013 09:19:00 -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: 4.7.2
X-Bugzilla-Keywords: wrong-code
X-Bugzilla-Severity: normal
X-Bugzilla-Who: gjl at gcc dot gnu.org
X-Bugzilla-Status: NEW
X-Bugzilla-Priority: P3
X-Bugzilla-Assigned-To: unassigned at gcc dot gnu.org
X-Bugzilla-Target-Milestone: 4.7.4
X-Bugzilla-Flags:
X-Bugzilla-Changed-Fields: bug_status
Message-ID: <bug-57503-4-V9JCywflur@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-57503-4@http.gcc.gnu.org/bugzilla/>
References: <bug-57503-4@http.gcc.gnu.org/bugzilla/>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: 7bit
X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/
Auto-Submitted: auto-generated
MIME-Version: 1.0
X-SW-Source: 2013-10/txt/msg00383.txt.bz2

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57503

Georg-Johann Lay <gjl at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|WAITING                     |NEW

--- Comment #8 from Georg-Johann Lay <gjl at gcc dot gnu.org> ---
(In reply to Richard Biener from comment #7)
> Can you please investigate the behavior on current trunk and the top of the
> 4.8 branch?

Using the code from comment #7 the problem persists

$ avr-gcc -S -Os pr57503.c -mmcu=atmega8 -dP


long
func1 (unsigned char a, unsigned char b, int c)
{
    unsigned ab = a * b;
    return (long) ab * c;
}


The code is a bit different but still wrong:


func1:
 ; (insn 10 5 25 (set (reg:HI 18 r18 [orig:44 D.1450 ] [44])
 ;         (mult:HI (zero_extend:HI (reg:QI 24 r24 [ a ]))
 ;             (zero_extend:HI (reg/v:QI 22 r22 [orig:50 b ] [50]))))
pr57503.c:4 168 {umulqihi3}
 ;      (expr_list:REG_DEAD (reg:QI 24 r24 [ a ])
 ;         (expr_list:REG_DEAD (reg/v:QI 22 r22 [orig:50 b ] [50])
 ;             (nil))))
    mul r24,r22     ;  10    umulqihi3    [length = 3]
    movw r18,r0
    clr __zero_reg__
 ; (insn 25 10 26 (set (reg:HI 26 r26)
 ;         (reg/v:HI 20 r20 [orig:51 c ] [51])) pr57503.c:5 82 {*movhi}
 ;      (expr_list:REG_DEAD (reg/v:HI 20 r20 [orig:51 c ] [51])
 ;         (nil)))
    movw r26,r20     ;  25    *movhi/1    [length = 1]
 ; (insn 26 25 21 (set (reg:SI 22 r22)
 ;         (mult:SI (sign_extend:SI (reg:HI 18 r18))
 ;             (sign_extend:SI (reg:HI 26 r26)))) pr57503.c:5 231
{*mulhisi3_call}
 ;      (expr_list:REG_DEAD (reg:HI 26 r26)
 ;         (expr_list:REG_DEAD (reg:HI 18 r18)
 ;             (nil))))
    rcall __mulhisi3     ;  26    *mulhisi3_call    [length = 1]
 ; (jump_insn 31 21 30 (return) pr57503.c:6 451 {return}
 ;      (nil)
 ;  -> return)
    ret     ;  31    return    [length = 1]


Insn 26 sign-extends both inputs but R18 (unsigned ab) should be zero-extended.

Tested with SVN 203240

gcc version 4.9.0 20131007 (experimental) (GCC)