Re: [sigc] Incorrect memory read in libsigc++-2.0.3 when compiled without optimization

[Martin Schulze <martin-ml hippogriff de>]

Am 02.06.2004 09:38:34 schrieb(en) Andris Pavenis:
> Noticed incorrect memory read in included example when it was  
> compiled
> without optimizations (-O0). It's most likely a bug in GCC, as
> it doesn't appear with -O1 or -O2. Tested under Linux.
>
> I have libsigc++-2.0.3, gcc-3.3.3, valgrind-2.1.1

I will investigate. Thanks for the pointer!
(Ah, debian sarge rocks on my notebook since yesterday :-))

Regards,

 Martin

> Andris
>
> #include <string>
> #include <iostream>
> #include <sigc++/sigc++.h>
>
> struct Foo
> {
>    bool f1 (std::string s, bool x)
>    {
>       std::cout << x << " : " << s << std::endl;
>    }
> };
>
> int main (int argc, const char * argv[])
> {
>    Foo foo;
>    sigc::slot<void,bool,std::string> slot_1;
>    slot_1 = sigc::group(
>            sigc::hide_return(sigc::mem_fun(foo, &Foo::f1)),
>            sigc::_2,
>            sigc::_1);
>
>    slot_1 (false, argv[0]);
>    return 0;
> }
>
> CXX = g++
> CXXFLAGS = -O0 -g $(shell pkg-config --cflags sigc++-2.0)
> LDFLAGS = $(shell pkg-config --libs sigc++-2.0)
>
> all:	prog
> 	valgrind --tool=memcheck ./foo *
>
> prog:	foo
>
> foo:	foo.o
> 	g++ -o foo foo.o $(LDFLAGS)
>
> g++ -O0 -g -I/usr/include/sigc++-2.0 -I/usr/lib/sigc++-2.0/include
> -c -o foo.o foo.cc
> g++ -o foo foo.o -lsigc-2.0
> valgrind --tool=memcheck ./foo *
> ==19457== Memcheck, a memory error detector for x86-linux.
> ==19457== Copyright (C) 2002-2004, and GNU GPL'd, by Julian Seward.
> ==19457== Using valgrind-2.1.1, a program supervision framework for
> x86-linux.
> ==19457== Copyright (C) 2000-2004, and GNU GPL'd, by Julian Seward.
> ==19457== For more details, rerun with: -v
> ==19457==
> ==19457== Invalid read of size 1
> ==19457==    at 0x80495C6:  
> _ZNK4sigc13lambda_group2INS_21retype_return_functorIvNS_18bound_mem_functor2Ib3FooSsbEEEENS_6lambdaINS_8internal14lambda_select2EEENS6_INS7_14lambda_select1EEEEclIRKbRKSsEENSC_18deduce_result_typeIT_T0_vvvvvE4typeESJ_SK_
> (group.h:227)
> ==19457==    by 0x8049586:  
> _ZNK4sigc8internal11lambda_coreINS_13lambda_group2INS_21retype_return_functorIvNS_18bound_mem_functor2Ib3FooSsbEEEENS_6lambdaINS0_14lambda_select2EEENS8_INS0_14lambda_select1EEEEELb1EEclIRKbRKSsEENSE_18deduce_result_typeIT_T0_vvvvvE4typeESL_SM_
> (base.h:71)
> ==19457==    by 0x8049566: sigc::internal::slot_call2<sigc:: 
> lambda<sigc::lambda_group2<sigc::retype_return_functor<void,
> sigc::bound_mem_functor2<bool, Foo, std::string, bool> >,
> sigc::lambda<sigc::internal::lambda_select2>,
> sigc::lambda<sigc::internal::lambda_select1> > >, void, bool,
> std::string>::call_it(sigc::internal::slot_rep*, bool const&,
> std::string const&) (slot.h:158)
> ==19457==    by 0x8049045: sigc::slot2<void, bool,
> std::string>::operator()(bool const&, std::string const&) const
> (slot.h:573)
> ==19457==  Address 0x4FFFE247 is just below %esp.  Possibly a bug in
> GCC/G++
> ==19457==   v 2.96 or 3.0.X.  To suppress, use:
> --workaround-gcc296-bugs=yes
> 0 : ./foo
> ==19457==
> ==19457== ERROR SUMMARY: 1 errors from 1 contexts (suppressed: 19  
> from
> 1)
> ==19457== malloc/free: in use at exit: 960 bytes in 1 blocks.
> ==19457== malloc/free: 3 allocs, 2 frees, 1080 bytes allocated.
> ==19457== For a detailed leak analysis,  rerun with: --leak-check=yes
> ==19457== For counts of detected errors, rerun with: -v
>
> _______________________________________________
> libsigc-list mailing list
> libsigc-list gnome org
> http://mail.gnome.org/mailman/listinfo/libsigc-list