Re: optimization/7884: code compiled with optimization flag causes segmentation violation on powerpc

Re: optimization/7884: code compiled with optimization flag causes segmentation violation on powerpc

[nathan]

Synopsis: code compiled with optimization flag causes segmentation violation on powerpc

State-Changed-From-To: open->analyzed
State-Changed-By: nathan
State-Changed-When: Wed Sep 11 02:18:59 2002
State-Changed-Why:
    This is not a bug, but I expected a warning (which we used
    to give). '(char *)foo' is an rvalue. the call through
    to 'char *&ref' binds an rvalue to a non-const reference.
    That is ill-formed. It binds to a temporary copy, thus
    the pointer seen by the function is not the object
    you cast.
    
    The correct solution is the cast to an lvalue,
    '(char *&)foo'

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=7884

optimization/7884: code compiled with optimization flag causes segmentation violation on powerpc

[aloeff]

>Number:         7884
>Category:       optimization
>Synopsis:       code compiled with optimization flag causes segmentation violation on powerpc
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    unassigned
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Wed Sep 11 01:36:01 PDT 2002
>Closed-Date:
>Last-Modified:
>Originator:     Andreas Loeffler
>Release:        3.1 and 3.2
>Organization:
>Environment:
redhat 7.1
[lefty@trinity - 09:34am: ~]$ powerpc-linux-gcc -v
Reading specs from /opt/fsp/i586-pc-linux-gnu/local/lib/gcc-lib/powerpc-linux/3.2/specs
Configured with: ../configure --host=i586-pc-linux-gnu --target=powerpc-linux --prefix=/opt/fsp/i586-pc-linux-gnu/local --enable-shared --enable-languages=c,c++,f77 --with-headers=/opt/fsp/powerpc-linux/include --with-libs=/opt/fsp/powerpc-linux/lib --with-cpu=403 --without-fp
Thread model: posix
gcc version 3.2
>Description:
We have encountered problems compiling our code with the -Os
flag (and -O, -O2, -O3) and attempting to run on the hardware.
We've tracked it down to one particular 
scenario that causes the problem

If you have a function defined like this

        int f(char*& data);

and some code that calls it that looks like this

        rc = f((char*)myData);
        x = myData->value;

The value myData does not get set and we get a segmentation
violation. To fix it, it needs to be changed to

        rc = f((char*&)myData);

>How-To-Repeat:
powerpc-linux-g++ -o test-opt test-opt.cc
without optimization myData gets assigned a value

powerpc-linux-g++ -O -o test-opt test-opt.cc
powerpc-linux-g++ -Os -o test-opt test-opt.cc
powerpc-linux-g++ -O3 -o test-opt test-opt.cc
with optimization myData becomes a null pointer which
can cause seg faults
>Fix:

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

I2luY2x1ZGUgPHN0ZGlvLmg+Cgp2b2lkIGYoY2hhciomIGRhdGEpCnsKICAgIGNoYXIqIHRtcCA9
IG5ldyBjaGFyWzI1Nl07CiAgICBkYXRhID0gdG1wOwp9CgoKaW50IG1haW4gKGludCBhcmdjLCBj
aGFyICphcmd2W10pCnsKICAgIHZvaWQqIG15RGF0YSA9IDA7CgogICAgLy9mKChjaGFyKiYpbXlE
YXRhKTsKICAgIGYoKGNoYXIqKW15RGF0YSk7CgogICAgcHJpbnRmKCJteURhdGEoMSkgPSAleFxu
IixteURhdGEpOwoKICAgIHJldHVybiAwOwp9Cgo=