Re: optimization/3770: ICE on -mcpu=ultrasparc -O2

Re: optimization/3770: ICE on -mcpu=ultrasparc -O2

[rodrigc]

The following reply was made to PR optimization/3770; it has been noted by GNATS.

From: rodrigc@gcc.gnu.org
To: gcc-bugs@gcc.gnu.org, gcc-gnats@gcc.gnu.org, gcc-prs@gcc.gnu.org,
  nobody@gcc.gnu.org, rth@redhat.com, wolfgang.bangerth@iwr.uni-heidelberg.de
Cc:  
Subject: Re: optimization/3770: ICE on -mcpu=ultrasparc -O2
Date: 1 Dec 2001 06:14:06 -0000

 Synopsis: ICE on -mcpu=ultrasparc -O2
 
 State-Changed-From-To: analyzed->closed
 State-Changed-By: rodrigc
 State-Changed-When: Fri Nov 30 22:14:06 2001
 State-Changed-Why:
     Duplicate of PR 3623.  Fixed.
     http://gcc.gnu.org/ml/gcc-patches/2001-11/msg02101.html
 
 http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&pr=3770&database=gcc

Re: optimization/3770: ICE on -mcpu=ultrasparc -O2

[rodrigc]

Synopsis: ICE on -mcpu=ultrasparc -O2

State-Changed-From-To: analyzed->closed
State-Changed-By: rodrigc
State-Changed-When: Fri Nov 30 22:14:06 2001
State-Changed-Why:
    Duplicate of PR 3623.  Fixed.
    http://gcc.gnu.org/ml/gcc-patches/2001-11/msg02101.html

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&pr=3770&database=gcc

Re: optimization/3770: ICE on -mcpu=ultrasparc -O2

[Wolfgang Bangerth]

The following reply was made to PR optimization/3770; it has been noted by GNATS.

From: Wolfgang Bangerth <wolfgang.bangerth@iwr.uni-heidelberg.de>
To: ciceron@gcc.gnu.org
Cc: gcc-bugs@gcc.gnu.org, gcc-gnats@gcc.gnu.org, rth@redhat.com
Subject: Re: optimization/3770: ICE on -mcpu=ultrasparc -O2
Date: Fri, 30 Nov 2001 12:11:20 +0100 (MET)

 > Synopsis: ICE on -mcpu=ultrasparc -O2
 > State-Changed-Why:
 >     Same problem as PR/3623, PR/4420, PR/3306.  Register %f34
 >     cannot handle a SFmode (see fix in PR/3623).  TBH, this PR
 >     cannot be reproduced because attachment cannot be downloaded.
 
 It should have been attached to the PR when setting it up, but I attach it
 to this mail again.
 Regards
   Wolfgang
 
 -------------------------------------------------------------------------
 Wolfgang Bangerth          email: wolfgang.bangerth@iwr.uni-heidelberg.de
                              www: http://gaia.iwr.uni-heidelberg.de/~wolf
 
 
 // compile with -mcpu=ultrasparc -O2
 struct FullMatrix
 {
     float & el (const unsigned int i, const unsigned int j);
     float* val;
     unsigned int num_cols;
     void invert (FullMatrix &M);
 };
 
 
 inline float &
 FullMatrix::el (const unsigned int i, const unsigned int j)
 {
   return val[i*num_cols+j];
 };
 
 
 void
 FullMatrix::invert (FullMatrix &M)
 {
   switch (num_cols)
     {
       case 4:
       {
         const float t14 = M.el(0,0)*M.el(1,1);
         const float t15 = M.el(2,2)*M.el(3,3);
         const float t17 = M.el(2,3)*M.el(3,2);
         const float t19 = M.el(0,0)*M.el(2,1);
         const float t20 = M.el(1,2)*M.el(3,3);
         const float t22 = M.el(1,3)*M.el(3,2);
         const float t24 = M.el(0,0)*M.el(3,1);
         const float t25 = M.el(1,2)*M.el(2,3);
         const float t27 = M.el(1,3)*M.el(2,2);
         const float t29 = M.el(1,0)*M.el(0,1);
         const float t32 = M.el(1,0)*M.el(2,1);
         const float t33 = M.el(0,2)*M.el(3,3);
         const float t35 = M.el(0,3)*M.el(3,2);
         const float t37 = M.el(1,0)*M.el(3,1);
         const float t38 = M.el(0,2)*M.el(2,3);
         const float t40 = M.el(0,3)*M.el(2,2);
         const float t42 = t14*t15-t14*t17-t19*t20+t19*t22+
                            t24*t25-t24*t27-t29*t15+t29*t17+
                            t32*t33-t32*t35-t37*t38+t37*t40;
         const float t43 = M.el(2,0)*M.el(0,1);
         const float t46 = M.el(2,0)*M.el(1,1);
         const float t49 = M.el(2,0)*M.el(3,1);
         const float t50 = M.el(0,2)*M.el(1,3);
         const float t52 = M.el(0,3)*M.el(1,2);
         const float t54 = M.el(3,0)*M.el(0,1);
         const float t57 = M.el(3,0)*M.el(1,1);
         const float t60 = M.el(3,0)*M.el(2,1);
         const float t63 = t43*t20-t43*t22-t46*t33+t46*t35+
                            t49*t50-t49*t52-t54*t25+t54*t27+
                            t57*t38-t57*t40-t60*t50+t60*t52;
         const float t65 = 1/(t42+t63);
         const float t71 = M.el(0,2)*M.el(2,1);
         const float t73 = M.el(0,3)*M.el(2,1);
         const float t75 = M.el(0,2)*M.el(3,1);
         const float t77 = M.el(0,3)*M.el(3,1);
         const float t81 = M.el(0,1)*M.el(1,2);
         const float t83 = M.el(0,1)*M.el(1,3);
         const float t85 = M.el(0,2)*M.el(1,1);
         const float t87 = M.el(0,3)*M.el(1,1);
         const float t101 = M.el(1,0)*M.el(2,2);
         const float t103 = M.el(1,0)*M.el(2,3);
         const float t105 = M.el(2,0)*M.el(1,2);
         const float t107 = M.el(2,0)*M.el(1,3);
         const float t109 = M.el(3,0)*M.el(1,2);
         const float t111 = M.el(3,0)*M.el(1,3);
         const float t115 = M.el(0,0)*M.el(2,2);
         const float t117 = M.el(0,0)*M.el(2,3);
         const float t119 = M.el(2,0)*M.el(0,2);
         const float t121 = M.el(2,0)*M.el(0,3);
         const float t123 = M.el(3,0)*M.el(0,2);
         const float t125 = M.el(3,0)*M.el(0,3);
         const float t129 = M.el(0,0)*M.el(1,2);
         const float t131 = M.el(0,0)*M.el(1,3);
         const float t133 = M.el(1,0)*M.el(0,2);
         const float t135 = M.el(1,0)*M.el(0,3);
         el(0,0) = (M.el(1,1)*M.el(2,2)*M.el(3,3)-M.el(1,1)*M.el(2,3)*M.el(3,2)-
                    M.el(2,1)*M.el(1,2)*M.el(3,3)+M.el(2,1)*M.el(1,3)*M.el(3,2)+
                    M.el(3,1)*M.el(1,2)*M.el(2,3)-M.el(3,1)*M.el(1,3)*M.el(2,2))*t65;
         el(0,1) = -(M.el(0,1)*M.el(2,2)*M.el(3,3)-M.el(0,1)*M.el(2,3)*M.el(3,2)-
                     t71*M.el(3,3)+t73*M.el(3,2)+t75*M.el(2,3)-t77*M.el(2,2))*t65;
         el(0,2) = (t81*M.el(3,3)-t83*M.el(3,2)-t85*M.el(3,3)+t87*M.el(3,2)+
                    t75*M.el(1,3)-t77*M.el(1,2))*t65;
         el(0,3) = -(t81*M.el(2,3)-t83*M.el(2,2)-t85*M.el(2,3)+t87*M.el(2,2)+
                     t71*M.el(1,3)-t73*M.el(1,2))*t65;
         el(1,0) = -(t101*M.el(3,3)-t103*M.el(3,2)-t105*M.el(3,3)+t107*M.el(3,2)+
                     t109*M.el(2,3)-t111*M.el(2,2))*t65;
 
         break;
       }
 
       default:
             *this = M;
     };
 };
 
 
 

Re: optimization/3770: ICE on -mcpu=ultrasparc -O2

[ciceron]

The following reply was made to PR optimization/3770; it has been noted by GNATS.

From: ciceron@gcc.gnu.org
To: gcc-bugs@gcc.gnu.org, gcc-gnats@gcc.gnu.org, gcc-prs@gcc.gnu.org,
  nobody@gcc.gnu.org, rth@redhat.com, wolfgang.bangerth@iwr.uni-heidelberg.de
Cc:  
Subject: Re: optimization/3770: ICE on -mcpu=ultrasparc -O2
Date: 30 Nov 2001 10:53:05 -0000

 Synopsis: ICE on -mcpu=ultrasparc -O2
 
 State-Changed-From-To: open->analyzed
 State-Changed-By: ciceron
 State-Changed-When: Fri Nov 30 02:53:04 2001
 State-Changed-Why:
     Same problem as PR/3623, PR/4420, PR/3306.  Register %f34
     cannot handle a SFmode (see fix in PR/3623).  TBH, this PR
     cannot be reproduced because attachment cannot be downloaded.
 
 http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&pr=3770&database=gcc

Re: optimization/3770: ICE on -mcpu=ultrasparc -O2

[ciceron]

Synopsis: ICE on -mcpu=ultrasparc -O2

State-Changed-From-To: open->analyzed
State-Changed-By: ciceron
State-Changed-When: Fri Nov 30 02:53:04 2001
State-Changed-Why:
    Same problem as PR/3623, PR/4420, PR/3306.  Register %f34
    cannot handle a SFmode (see fix in PR/3623).  TBH, this PR
    cannot be reproduced because attachment cannot be downloaded.

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&pr=3770&database=gcc

optimization/3770: ICE on -mcpu=ultrasparc -O2

[wolfgang.bangerth]

>Number:         3770
>Category:       optimization
>Synopsis:       ICE on -mcpu=ultrasparc -O2
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    unassigned
>State:          open
>Class:          ice-on-legal-code
>Submitter-Id:   net
>Arrival-Date:   Mon Jul 23 01:36:02 PDT 2001
>Closed-Date:
>Last-Modified:
>Originator:     Wolfgang Bangerth
>Release:        unknown-1.0
>Organization:
>Environment:
sparc-sun-solaris2.7, gcc CVS of 20010702, 3.0 branch
>Description:
Richard, you asked for bugs with -mcpu=ultrasparc, here is one:
gcc hits an ICE when using -mcpu=ultrasparc on the appended
code.

deal.II/lac> c++ -mcpu=ultrasparc -O2 -c a.cc
a.cc: In member function `void FullMatrix::invert(FullMatrix&)':
a.cc:97: Insn does not satisfy its constraints:
(insn 4430 199 243 (set (reg/v:SF 66 %f34 [150])
        (reg:SF 34 %f2)) 84 {*movsf_insn_novis} (nil)
    (nil))
a.cc:97: Internal compiler error in reload_cse_simplify_operands, at reload1.c:
   8355
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL: http://www.gnu.org/software/gcc/bugs.html > for instructions.
deal.II/lac> 
deal.II/lac> 
deal.II/lac> 
deal.II/lac> c++ -v
Reading specs from /home/people/wolf/Config/gcc-ss-sun/lib/gcc-lib/sparc-sun-solaris2.7/3.0.1/specs
Configured with: ../gcc/configure --prefix=/home/people/wolf/Config/gcc-ss-sun --with-gnu-as --with-as=/usr/local/bin/as --with-gnu-ld --with-ld=/usr/local/bin/ld : (reconfigured) 
Thread model: posix
gcc version 3.0.1 20010702 (prerelease)

Regards
  Wolfgang
>How-To-Repeat:
c++ -mcpu=ultrasparc -O2 -c a.cc
>Fix:

>Release-Note:
>Audit-Trail:
>Unformatted:
----gnatsweb-attachment----
Content-Type: application/octet-stream; name="a.cc"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="a.cc"