solaris patch

[Benjamin Kosnik]

> > Phil, I'd like you to try current CVS, and see how David's suggestions
> > work on solaris (WRT using os_defines vs libio.h ifdefs) . If it's
> > pretty straightforward, then I'd appreciate hearing this before you
> > try to auto-gen _G_config.h.
> 
> It's pretty straightforward.

great

> > Also, perhaps this would be a good time to make sense of the
> > libio/Makefile.am based on yesterday's email.........
> 
> I've tweaked that like I mentioned (merging the need_xtra stuff).  No build
> problems so far (except for that freaky modff thing), but I won't be able
> to test the changes on Linux until this afternoon.

hmm. If you post your patch I can test on linux.

> The testsuite results are hideous.  Just about every shared executable
> cores with a bus error:
> 
> 
> 52% gdb -nw testsuite/header_ciso646.sh-exe core.header_ciso646.sh-exe.20770 
> Core was generated by `/tec/build/3test/sparc-sun-solaris2.8/libstdc++-v3/testsuite/header_ciso646.sh-'.
> Program terminated with signal 10, Bus Error.
> Reading symbols from /tec/build/Ebuild/lib/libstdc++.so.3...done.
> Loaded symbols for /tec/build/Ebuild/lib/libstdc++.so.3
> Reading symbols from /opt/SUNWspro/lib/libm.so.1...done.
> Loaded symbols for /opt/SUNWspro/lib/libm.so.1
> Reading symbols from /usr/lib/libc.so.1...done.
> Loaded symbols for /usr/lib/libc.so.1
> Reading symbols from /usr/lib/libdl.so.1...done.
> Loaded symbols for /usr/lib/libdl.so.1
> Reading symbols from /usr/platform/SUNW,Ultra-30/lib/libc_psr.so.1...done.
> Loaded symbols for /usr/platform/SUNW,Ultra-30/lib/libc_psr.so.1
> #0  0xff2fa2a4 in _IO_file_overflow (f=0x22a17, ch=3)
>     at ../../../../unified/libstdc++-v3/libio/fileops.c:489
> 489       if (f->_flags & _IO_NO_WRITES) /* SET ERROR */
> Current language:  auto; currently c
> (gdb) bt
> #0  0xff2fa2a4 in _IO_file_overflow (f=0x22a17, ch=3)
>     at ../../../../unified/libstdc++-v3/libio/fileops.c:489
> #1  0xff33f42c in std::basic_filebuf<char, std::char_traits<char> >::close (
>     this=0x22a50)
>     at ../../../../unified/libstdc++-v3/include/bits/fstream.tcc:174
> #2  0xff31e8f4 in std::basic_filebuf<char, std::char_traits<char> >::~basic_filebuf (this=0x22a50, __in_chrg=3)
>     at ../../../../unified/libstdc++-v3/include/bits/std_fstream.h:94
> #3  0xff306470 in std::ios_base::Init::~Init (this=0xff38ba68, __in_chrg=2)
>     at ../../../../unified/libstdc++-v3/src/ios.cc:165
> #4  0xff307540 in __static_initialization_and_destruction_0 (__initialize_p=0, 
>     __priority=65535)
>     at ../../../../unified/libstdc++-v3/include/bits/std_fstream.h:55
> #5  0xff307580 in global destructors keyed to std::__ios_flags::_S_boolalpha ()
>     at ../../../../unified/libstdc++-v3/include/bits/std_fstream.h:94
> #6  0xff2f2058 in __do_global_dtors_aux ()
>    from /tec/build/Ebuild/lib/libstdc++.so.3
> #7  0xff35e780 in _fini () from /tec/build/Ebuild/lib/libstdc++.so.3
> #8  0xff3ca964 in ?? ()
> #9  0xff3caad4 in ?? ()
> #10 0xff19b4b8 in _exithandle () from /usr/lib/libc.so.1
> #11 0xff21a380 in exit () from /usr/lib/libc.so.1
> 
> It is however an improvement over last week, where shared executables
> couldn't be built at all.

Right.

> Static executables often fail to link:
> 
> /tec/build/Ebuild/bin/g++ -g3 -DDEBUG_ASSERT -ffunction-sections
>   -fdata-sections  -static -I../../../unified/libstdc++-v3/testsuite
>   -L/tec/build/Ebuild/lib          -Wl,--rpath -Wl,/tec/build/Ebuild/lib
>   ../../../unified/libstdc++-v3/testsuite/17_intro/header_iostream.cc -o
>   /tec/build/3test/sparc-sun-solaris2.8/libstdc++-v3/testsuite/header_iostream.st-exe
> collect2: ld terminated with signal 10 [Bus Error]
> /usr/lib/libc.a(iconv.o): In function `iconv_open_private':
> iconv.o(.text+0x

Hmm. Wow. The wide character stuff compiled on solaris? I'm not quite 
sure about this, and won't have a solaris box to test on till monday.

-benjamin