Re: [RFC] Userspace top-level Makefile (was Re: [PATCH] Fix LIBDIR usage for load_policy)

From: Guido Trentalancia <guido@trentalancia.com>
To: Eric Paris <eparis@redhat.com>
Cc: Joshua Brindle <method@manicmethod.com>,
	Eric Paris <eparis@parisplace.org>,
	SELinux Mail List <selinux@tycho.nsa.gov>
Subject: Re: [RFC] Userspace top-level Makefile (was Re: [PATCH] Fix LIBDIR usage for load_policy)
Date: Sat, 10 Sep 2011 01:25:19 +0200	[thread overview]
Message-ID: <1315610720.2482.42.camel@vortex> (raw)
In-Reply-To: <4E6A9E26.9080806@redhat.com>

On Fri, 2011-09-09 at 19:15 -0400, Eric Paris wrote:
> On 09/09/2011 07:12 PM, Guido Trentalancia wrote:
> > On Fri, 2011-09-09 at 19:07 -0400, Eric Paris wrote:
> >> On 09/09/2011 06:35 PM, Guido Trentalancia wrote:
> >>> The following patch is probably even better as it actually sorts out the
> >>> problem rather than just printing out a warning (it can still be
> >>> combined with the creation of a top-level README file):
> >>>
> >>> diff -x '*.po' -x '*.pot' -pru selinux-09092011-orig/checkpolicy/Makefile selinux-09092011-local-headers/checkpolicy/Makefile
> >>> --- selinux-09092011-orig/checkpolicy/Makefile	2011-09-09 20:12:55.978662153 +0200
> >>> +++ selinux-09092011-local-headers/checkpolicy/Makefile	2011-09-10 00:21:16.242852130 +0200
> >>> @@ -12,7 +12,7 @@ YACC = bison -y
> >>>  
> >>>  CFLAGS ?= -g -Wall -Werror -Wshadow -O2 -pipe -fno-strict-aliasing
> >>>  
> >>> -override CFLAGS += -I. -I${INCLUDEDIR}
> >>> +override CFLAGS += -I. -I../libsepol/include -I${INCLUDEDIR}
> >>
> >> I haven't checked, but can the makefiles then work when there is no
> >> ../libsepol/include?
> > 
> > If ../libsepol/include does not exist and libsepol headers are not
> > installed under the standard location (${INCLUDEDIR}), then the above is
> > going to fail.
> > 
> > But why should libsepol/include be missing from git in the first
> > place ??
> 
> I agree that they must be in one or the other for it to build.  A number
> of distros build these packages separately, rather than all at once.
> Thus there would be no ../libsepol/include directory at all, but the
> headers should be picked up by -I$(INCLUDEDIR).  As long as gcc doesn't
> care that the directory doesn't exist, I actually like it.
> 
> -Eric

gcc should only care that at least one version of the file can be found
if #include'd.

It still needs to be tested though. Can you try building the whole
userspace git using that patch ?

Regards,

Guido

--
This message was distributed to subscribers of the selinux mailing list.
If you no longer wish to subscribe, send mail to majordomo@tycho.nsa.gov with
the words "unsubscribe selinux" without quotes as the message.