[patch] include/posix.h

[patch] include/posix.h

[Andreas Tobler]

Hello all,

to complete the HP-UX PA port we need to modify the file in the subject.

We conflict with the gcj/javaprims.h and the Boolean.h when we compile 
java/net/natVMNetworkInterfacePosix.cc

The reason comes from the system header sys/rw_lock.h:

-----


/*
  * For source compatibility, need to continue defining TRUE and FALSE for
  * user-level applications. Although this file defines a kernel only 
service,
  * some user-level applications are indirectly including this file and
  * depending on the defines for TRUE and FALSE. These defines need to
  * remain until all the user-level issues have been handled.
  */
#ifndef _KERNEL
#ifndef TRUE
#define TRUE    1
#define FALSE   0
#else
#if ((TRUE != 1) || (FALSE != 0))
This is probably not a good thing....
#endif
#endif /* !TRUE */

-----

Do you have a better idea or is the below ok?

Thanks,
Andreas


2006-05-20  Andreas Tobler  <a.tobler@schweiz.ch>

	* include/posix.h: Undefine TRUE and FALSE.


Index: include/posix.h
===================================================================
--- include/posix.h     (revision 113914)
+++ include/posix.h     (working copy)
@@ -47,6 +47,8 @@
  #else
  #define _Jv_platform_solib_suffix ".so"
  #endif
+#undef TRUE
+#undef FALSE

  // Some POSIX systems don't have O_SYNC and O_DYSNC so we define them 
here.
  // Needed in java/io/natFileDescriptorPosix.cc.

Re: [patch] include/posix.h

[Tom Tromey]

>>>>> "Andreas" == Andreas Tobler <toa@pop.agri.ch> writes:

Andreas> We conflict with the gcj/javaprims.h and the Boolean.h when we compile
Andreas> java/net/natVMNetworkInterfacePosix.cc

Thanks.

What if we handled this more directly and generated a Boolean.h that
had the #undefs at the top?  I think that would work a bit better in
more cases.  Otherwise we're dependent on the order of inclusion of
our posix.h, system headers, and Boolean.h.

This can be implemented with a small Makefile change.  Could you try
it?  Or is there a problem with this?

Tom

Re: [patch] include/posix.h

[Andreas Tobler]

Tom Tromey wrote:
>>>>>> "Andreas" == Andreas Tobler <toa@pop.agri.ch> writes:
> 
> Andreas> We conflict with the gcj/javaprims.h and the Boolean.h when we compile
> Andreas> java/net/natVMNetworkInterfacePosix.cc
> 
> Thanks.
> 
> What if we handled this more directly and generated a Boolean.h that
> had the #undefs at the top?  I think that would work a bit better in
> more cases.  Otherwise we're dependent on the order of inclusion of
> our posix.h, system headers, and Boolean.h.

Well, this does not help if Boolean.h is included before such magic 
system headers which define TRUE and FALSE again.

In our case sys/rw_lock.h, the file where this (re)-definition happens, 
is included after Boolean.h, implicitely. So I tend to agree with Dave 
that the workaround he proposes is probably the best one in regard of 
effort and gain.

Thanks,
Andreas

Re: [patch] include/posix.h

[Tom Tromey]

>>>>> "Andreas" == Andreas Tobler <toa@pop.agri.ch> writes:

Andreas> Well, this does not help if Boolean.h is included before such magic
Andreas> system headers which define TRUE and FALSE again.

Yeah.  However, any ordering will have an associated potential
failure mode, afaics.

Andreas> In our case sys/rw_lock.h, the file where this (re)-definition
Andreas> happens, is included after Boolean.h, implicitely. So I tend to agree
Andreas> with Dave that the workaround he proposes is probably the best one in
Andreas> regard of effort and gain.

That patch also looked fine to me.  Feel free to check it in.

Tom

Re: [patch] include/posix.h

[John David Anglin]

> Andreas> In our case sys/rw_lock.h, the file where this (re)-definition
> Andreas> happens, is included after Boolean.h, implicitely. So I tend to agree
> Andreas> with Dave that the workaround he proposes is probably the best one in
> Andreas> regard of effort and gain.
> 
> That patch also looked fine to me.  Feel free to check it in.

Done.

Dave
-- 
J. David Anglin                                  dave.anglin@nrc-cnrc.gc.ca
National Research Council of Canada              (613) 990-0752 (FAX: 952-6602)

2006-05-25  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>

        * configure.ac: Add sys/rw_lock.h to AC_CHECK_HEADERS.
        * configure: Rebuilt.
        * include/config.h.in: Likewise.
	* include/posix.h: If HAVE_SYS_RW_LOCK_H is defined, include
	<sys/rw_lock.h>.

Index: configure.ac
===================================================================
--- configure.ac	(revision 113941)
+++ configure.ac	(working copy)
@@ -1276,7 +1276,7 @@
 		  sys/ioctl.h sys/filio.h sys/stat.h sys/select.h \
 		  sys/socket.h netinet/in.h arpa/inet.h netdb.h net/if.h \
 		  pwd.h sys/config.h stdint.h langinfo.h locale.h \
-		  dirent.h])
+		  dirent.h sys/rw_lock.h])
 AC_CHECK_HEADERS(inttypes.h, [
     AC_DEFINE(HAVE_INTTYPES_H, 1, [Define if <inttypes.h> is available])
     AC_DEFINE(JV_HAVE_INTTYPES_H, 1, [Define if <inttypes.h> is available])
Index: include/posix.h
===================================================================
--- include/posix.h	(revision 113941)
+++ include/posix.h	(working copy)
@@ -37,6 +37,12 @@
 
 #include <fcntl.h>
 
+/* The header file <sys/rw_lock.h> needs to be included before javaprims.h
+   on HP-UX 11 to avoid a compilation error.  */
+#ifdef HAVE_SYS_RW_LOCK_H
+#include <sys/rw_lock.h>
+#endif
+
 #include <gcj/cni.h>
 #include <java/util/Properties.h>