[Openais] [Corosync] Patches to fix BSD support

[Steven Dake]

On Wed, 2009-06-10 at 15:36 +0200, Jérôme Flesch wrote:
> > On Wed, 2009-06-10 at 09:40 +0200, Jérôme Flesch wrote:
> >>> Jerome,
> >>>
> >>> Thanks for the work. Few comments:
> >>>
> >>> First, NAME_MAX was changed to FILENAME_MAX to match posix.  Also these
> >>> are defined in /usr/include/limits.h.  Doesn't bsd have this file? (I
> >>> recently fixed this problem for the Solaris port I did)  If it does,
> >>> then you shouldn't have to do any defining since this is now fixed in
> >>> current trunk.
> >>>
> >> I confirm, FILENAME_MAX is defined on FreeBSD (in stdio.h), so this part 
> >> of my patch is actually not required. Do you want me to prepare a new one ?
> >>
> > Yes please.  Also the PATH_MAX variable should be defined somewhere in
> > your header files and doesn't need a definition in configure.ac.
> >
> I've attached a new version of the patch to this mail. You were right, 
> PATH_MAX and NAME_MAX defines are not needed anymore. I've also tried to 
> make the -ldl part a little bit less ugly.
> 
> 
> >>> I think the LDL part needs to be in some user defined flag in the
> >>> configure.ac.  The #if BSD with defining LDL is a bit hacky.  I'm not
> >>> sure where it should go however, so perhaps Jim or Fabio could comment.
> >>> If Jim or Fabio give that part the thumbs up, then this patch should be
> >>> good to go in.
> >>>
> >>> the if defined semun wont work because unions are not defines.  That
> >>> entire code segment is wrong.  Can you try the attached patch on bsd for
> >>> me?
> >>>
> >> I think you misunderstood my problem: semun *is* already defined on 
> >> FreeBSD, but the #if !defined(semun) doesn't work on my system (probably 
> >> due to a GCC version difference ?) and the compiler ends up with a 
> >> redefinition of semun.
> >>
> >> Is there actually any system where semun is not already defined ?
> >>
> >>
> > 
> > Linux doesn't define it and Solaris doesn't even have a description of
> > the definition.  Do you have _SEM_SEMUN_UNDEFINED in your header files?
> > Or SEM_SEMUN_DEFINED?  That can be used to do an if defined test.
> > 
> I haven't found any of them in the FreeBSD headers :/
> 

I believe the proper thing to do is put the union in ifdefs around
_SEM_SEMUN_UNDEFINED and define them in configure.ac for Solaris.

Can you try this patch on your os?

Regards
-steve

> 
> > Regards
> > -steve
> > 
> > 
> > 
> > 
> > 
> plain text document attachment (patch-fix-bsd-compilation)
> diff --git a/configure.ac b/configure.ac
> index de1ac78..b0b7c87 100644
> --- configure.ac
> +++ configure.ac
> @@ -194,6 +194,7 @@ AC_ARG_WITH([socket-dir],
>  # OS detection
>  # THIS SECTION MUST DIE!
>  CP=cp
> +OS_LDL="-ldl"
>  case "$host_os" in
>  	*linux*)
>  		AC_DEFINE_UNQUOTED([COROSYNC_LINUX], [1],
> @@ -225,11 +226,14 @@ case "$host_os" in
>  	*bsd*)
>  		AC_DEFINE_UNQUOTED([COROSYNC_BSD], [1],
>  				   [Compiling for BSD platform])
> +		AC_DEFINE_UNQUOTED([MAP_ANONYMOUS], [MAP_ANON],
> +				   [Shared memory define for Darwin platform])
>  		OS_CFLAGS=""
>  		OS_CPPFLAGS="-I/usr/local/include"
>  		OS_LDFLAGS="-L/usr/local/lib"
>  		OS_DYFLAGS="-export-dynamic"
>  		DARWIN_OPTS=""
> +		OS_LDL=""
>  	;;
>  	*solaris*)
>  		AC_DEFINE_UNQUOTED([COROSYNC_SOLARIS], [1],
> diff --git a/lib/Makefile.am b/lib/Makefile.am
> index 37afffd..31e8664 100644
> --- lib/Makefile.am
> +++ lib/Makefile.am
> @@ -114,7 +114,7 @@ libconfdb.so.$(SONAME): confdb.o sa-confdb.o ../lcr/lcr_ifact.o libcoroipcc.so.$
>  	$(CC) -shared -o $@ \
>  		-Wl,-soname=libconfdb.so.$(SOMAJOR) \
>  		-Wl,-version-script=$(srcdir)/libconfdb.versions \
> -		$^ $(LDFLAGS) $(OS_DYFLAGS) -ldl $(AM_LDFLAGS)
> +		$^ $(LDFLAGS) $(OS_DYFLAGS) $(OS_LDL) $(AM_LDFLAGS)
>  	ln -sf libconfdb.so.$(SONAME) libconfdb.so
>  	ln -sf libconfdb.so.$(SONAME) libconfdb.so.$(SOMAJOR)
>  
-------------- next part --------------
A non-text attachment was scrubbed...
Name: corosync-trunk-semun-cleanup.patch
Type: text/x-patch
Size: 1178 bytes
Desc: not available
Url : http://lists.linux-foundation.org/pipermail/openais/attachments/20090609/b2d52b35/attachment-0001.bin