This is the mail archive of the
libstdc++@gcc.gnu.org
mailing list for the libstdc++ project.
Re: gcc-3.1 cross compile ok? [was Re: gcc-3.1 + libstdc++-v3.0.97 cross compile ok?]
- From: Brad Spencer <spencer at infointeractive dot com>
- To: libstdc++ at gcc dot gnu dot org
- Date: Fri, 5 Jul 2002 11:49:55 -0300
- Subject: Re: gcc-3.1 cross compile ok? [was Re: gcc-3.1 + libstdc++-v3.0.97 cross compile ok?]
- References: <20020528184534.J9548@infointeractive.com> <20020529162546.A19754@disaster.basement.lan> <20020704162058.F29346@infointeractive.com> <20020704170358.I29346@infointeractive.com>
On Thu, Jul 04, 2002 at 05:03:59PM -0300, Brad Spencer wrote:
> Perhaps I spoke too soon. The original problem still shows up when
> cross compiling and using the supplied libstdc++-v3. I can repost if
> necessary.
Looking even more into this, it appears that when I try to build a
cross compiler on host i386-sun-solaris2.6 for sparc64-sun-solaris2.8
with gcc-3.1 (and stock libstdc++-v3, now), it's thinking I have
newlib instead of the Solaris libc. I put on my Automake hat, which
doesn't fit very well, and I think I found the problem.
It looks like this is an artifact of the hack that lets cross-compiles
work. There's some mumbling about it being done "by hand" and that it
will all go away soon :) Anyway, here's a patch that's been tested
for building a target=sparc64-sun-solaris2.8 cross compiler. Note
that the HAVE_* settings I chose was mainly guess work, but I did have
a config.h from a sparc64-sun-solaris2.8 of the same compiler and
libstdc++-v3.
Comments? Someone with a better knowledge of configure.in could probably
tweak it. Should I open this kind of thing as a PR?
Also, perhaps configure should log the OS_INCLUDE_DIR somewhere more
obvious to make this easier to detect.
*** configure.in.orig Thu Jul 4 20:17:46 2002
--- configure.in Thu Jul 4 20:48:48 2002
***************
*** 156,161 ****
--- 156,178 ----
GLIBCPP_CHECK_WCHAR_T_SUPPORT
os_include_dir="config/os/mingw32"
;;
+ *-solaris2.5)
+ os_include_dir="config/os/solaris/solaris2.5"
+ AC_DEFINE(HAVE_HYPOT)
+ ;;
+ *-solaris2.6)
+ os_include_dir="config/os/solaris/solaris2.6"
+ AC_DEFINE(HAVE_HYPOT)
+ ;;
+ *-solaris2.7)
+ os_include_dir="config/os/solaris/solaris2.7"
+ AC_DEFINE(HAVE_HYPOT)
+ ;;
+ *-solaris2.8)
+ # 2.7 and 2.8 are the same for now
+ os_include_dir="config/os/solaris/solaris2.7"
+ AC_DEFINE(HAVE_HYPOT)
+ ;;
*)
os_include_dir="config/os/newlib"
AC_DEFINE(HAVE_HYPOT)
***************
*** 165,170 ****
--- 182,199 ----
case "$target_alias" in
*-mingw32*)
;;
+ *-solaris*)
+ # This is a subset of the '*' case below, taken from a native
+ # build of sparc64-sun-solaris2.8
+ AC_DEFINE(HAVE_STRTOF)
+ AC_DEFINE(HAVE_STRTOLD)
+ AC_DEFINE(HAVE_MMAP)
+
+ AC_DEFINE(HAVE_COPYSIGN)
+ AC_DEFINE(HAVE_ISNAN)
+ AC_DEFINE(HAVE_ISNANF)
+ AC_DEFINE(HAVE_MODFF)
+ ;;
*)
# GLIBCPP_CHECK_STDLIB_SUPPORT
AC_DEFINE(HAVE_STRTOF)
--
-----------------------------------------------------------------
Brad Spencer - spencer@infointeractive.com - "It's quite nice..."
Systems Architect | InfoInterActive Corp. | An AOL Company