SELinux musl support

SELinux musl support

[Jason Zaman]

This series fixes compiling and running on musl libc.

patches 1-2 are fairly trivial.

patches 3-4 are a feature change on that platform since it does not
support GLOB_TILDE and GLOB_BRACE. tilde is coming in musl 1.1.21
according to [1]. brace support is not documented anywhere or in the
example configs so that is probably not a big problem.

patch 5 fixes a bug and it just happens that glibc returns a value for
sysconf and the error handling was wrong but never noticed.

[1]: https://wiki.musl-libc.org/roadmap.html

[PATCH 1/5] sestatus: include limits.h for PATH_MAX
[PATCH 2/5] libselinux: enable linking to musl-fts
[PATCH 3/5] setfiles: Musl compatibility for GLOB_BRACE and
[PATCH 4/5] restorecond: Musl compatibility for GLOB_BRACE and
[PATCH 5/5] genhomedircon: sysconf can return -1 without failure

[PATCH 1/5] sestatus: include limits.h for PATH_MAX

[Jason Zaman]

compile fails on musl libc because it cant find PATH_MAX.

Signed-off-by: Jason Zaman <jason@perfinion.com>
---
 policycoreutils/sestatus/sestatus.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/policycoreutils/sestatus/sestatus.c b/policycoreutils/sestatus/sestatus.c
index 9a92e72f..b37f0353 100644
--- a/policycoreutils/sestatus/sestatus.c
+++ b/policycoreutils/sestatus/sestatus.c
@@ -17,6 +17,7 @@
 #include <unistd.h>
 #include <libgen.h>
 #include <ctype.h>
+#include <limits.h>
 
 #define PROC_BASE "/proc"
 #define MAX_CHECK 50
-- 
2.16.1

[PATCH 2/5] libselinux: enable linking to musl-fts

[Jason Zaman]

Musl libc does not include the fts(3) functions so need to link to the
musl-fts library
https://github.com/pullmoll/musl-fts

Signed-off-by: Jason Zaman <jason@perfinion.com>
---
 libselinux/src/Makefile | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/libselinux/src/Makefile b/libselinux/src/Makefile
index 8af04aab..977b5c8c 100644
--- a/libselinux/src/Makefile
+++ b/libselinux/src/Makefile
@@ -98,6 +98,8 @@ LD_SONAME_FLAGS=-install_name,$(LIBSO)
 endif
 
 PCRE_LDLIBS ?= -lpcre
+# override with -lfts when building on Musl libc to use fts-standalone
+FTS_LDLIBS ?=
 
 override CFLAGS += -I../include -D_GNU_SOURCE $(DISABLE_FLAGS) $(PCRE_CFLAGS)
 
@@ -149,7 +151,7 @@ $(LIBA): $(OBJS)
 	$(RANLIB) $@
 
 $(LIBSO): $(LOBJS)
-	$(CC) $(CFLAGS) $(LDFLAGS) -shared -o $@ $^ $(PCRE_LDLIBS) -ldl -Wl,$(LD_SONAME_FLAGS)
+	$(CC) $(CFLAGS) $(LDFLAGS) -shared -o $@ $^ $(PCRE_LDLIBS) $(FTS_LDLIBS) -ldl -Wl,$(LD_SONAME_FLAGS)
 	ln -sf $@ $(TARGET)
 
 $(LIBPC): $(LIBPC).in ../VERSION
-- 
2.16.1

[PATCH 3/5] setfiles: Musl compatibility for GLOB_BRACE and GLOB_TILDE

[Jason Zaman]

From: Luis Ressel <aranea@aixah.de>

musl doesn't implement GLOB_BRACE and GLOB_TILDE, so simply don't use
them there. This only affects "setfiles -f", which I don't expect many
people use, and it's undocumented anyway that it expands globs.

Signed-off-by: Luis Ressel <aranea@aixah.de>
Signed-off-by: Jason Zaman <jason@perfinion.com>
---
 policycoreutils/setfiles/restore.c | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/policycoreutils/setfiles/restore.c b/policycoreutils/setfiles/restore.c
index 50d192a2..9dea5656 100644
--- a/policycoreutils/setfiles/restore.c
+++ b/policycoreutils/setfiles/restore.c
@@ -6,6 +6,14 @@
 #include "restore.h"
 #include <glob.h>
 
+#ifndef GLOB_BRACE
+#define GLOB_BRACE 0
+#endif
+
+#ifndef GLOB_TILDE
+#define GLOB_TILDE 0
+#endif
+
 char **exclude_list;
 int exclude_count;
 
-- 
2.16.1

[PATCH 4/5] restorecond: Musl compatibility for GLOB_BRACE and GLOB_TILDE

[Jason Zaman]

musl doesn't implement GLOB_BRACE and GLOB_TILDE, so simply don't use
them there. This affects restorecond -u but braces are not used in the
example configs. GLOB_TILDE is on the roadmap[1] for musl 1.1.21 so
restorecond -u should be fine soon.

[1]: https://wiki.musl-libc.org/roadmap.html

Signed-off-by: Jason Zaman <jason@perfinion.com>
---
 restorecond/restore.c | 13 ++++++++-----
 restorecond/watch.c   |  8 ++++++++
 2 files changed, 16 insertions(+), 5 deletions(-)

diff --git a/restorecond/restore.c b/restorecond/restore.c
index cf04e962..f6e30001 100644
--- a/restorecond/restore.c
+++ b/restorecond/restore.c
@@ -1,11 +1,14 @@
-/*
- * Note that the restorecond(8) service build links with these functions.
- * Therefore any changes here should also be tested against that utility.
- */
-
 #include "restore.h"
 #include <glob.h>
 
+#ifndef GLOB_TILDE
+#define GLOB_TILDE 0
+#endif
+
+#ifndef GLOB_BRACE
+#define GLOB_BRACE 0
+#endif
+
 char **exclude_list;
 int exclude_count;
 
diff --git a/restorecond/watch.c b/restorecond/watch.c
index 691df824..98ff797b 100644
--- a/restorecond/watch.c
+++ b/restorecond/watch.c
@@ -20,6 +20,14 @@
 #include "stringslist.h"
 #include "utmpwatcher.h"
 
+#ifndef GLOB_TILDE
+#define GLOB_TILDE 0
+#endif
+
+#ifndef GLOB_BRACE
+#define GLOB_BRACE 0
+#endif
+
 /* size of the event structure, not counting name */
 #define EVENT_SIZE  (sizeof (struct inotify_event))
 /* reasonable guess as to size of 1024 events */
-- 
2.16.1

[PATCH 5/5] genhomedircon: sysconf can return -1 without failure

[Jason Zaman]

from getpwnam_r(3): "The call sysconf(_SC_GETPW_R_SIZE_MAX) returns
either -1, without changing errno, or an initial suggested size for buf.
(If this size is too small, the call fails with ERANGE, in which case
the caller can retry with a larger buffer.)"

The same can happen for _SC_GETGR_R_SIZE_MAX. 1024 appears to be a good
fallback but may need revisiting in the future.

This triggered an error on musl libc but could happen other places too.

Signed-off-by: Jason Zaman <jason@perfinion.com>
---
 libsemanage/src/genhomedircon.c | 12 ++++++++++--
 1 file changed, 10 insertions(+), 2 deletions(-)

diff --git a/libsemanage/src/genhomedircon.c b/libsemanage/src/genhomedircon.c
index d09d82ff..3e61b510 100644
--- a/libsemanage/src/genhomedircon.c
+++ b/libsemanage/src/genhomedircon.c
@@ -972,9 +972,13 @@ static int add_user(genhomedircon_settings_t * s,
 	char uid[11];
 	char gid[11];
 
+	errno = 0;
 	/* Allocate space for the getpwnam_r buffer */
 	rbuflen = sysconf(_SC_GETPW_R_SIZE_MAX);
-	if (rbuflen <= 0)
+	if (rbuflen == -1 && errno == 0)
+		/* sysconf returning -1 with no errno means indeterminate size */
+		rbuflen = 1024;
+	else if (rbuflen <= 0)
 		goto cleanup;
 	rbuf = malloc(rbuflen);
 	if (rbuf == NULL)
@@ -1057,8 +1061,12 @@ static int get_group_users(genhomedircon_settings_t * s,
 	struct group grstorage, *group = NULL;
 	struct passwd *pw = NULL;
 
+	errno = 0;
 	grbuflen = sysconf(_SC_GETGR_R_SIZE_MAX);
-	if (grbuflen <= 0)
+	if (grbuflen == -1 && errno == 0)
+		/* sysconf returning -1 with no errno means indeterminate size */
+		grbuflen = 1024;
+	else if (grbuflen <= 0)
 		goto cleanup;
 	grbuf = malloc(grbuflen);
 	if (grbuf == NULL)
-- 
2.16.1

Re: SELinux musl support

[Nicolas Iooss]

On Thu, May 17, 2018 at 7:11 AM, Jason Zaman <jason@perfinion.com> wrote:
> This series fixes compiling and running on musl libc.
>
> patches 1-2 are fairly trivial.
>
> patches 3-4 are a feature change on that platform since it does not
> support GLOB_TILDE and GLOB_BRACE. tilde is coming in musl 1.1.21
> according to [1]. brace support is not documented anywhere or in the
> example configs so that is probably not a big problem.
>
> patch 5 fixes a bug and it just happens that glibc returns a value for
> sysconf and the error handling was wrong but never noticed.
>
> [1]: https://wiki.musl-libc.org/roadmap.html
>
> [PATCH 1/5] sestatus: include limits.h for PATH_MAX
> [PATCH 2/5] libselinux: enable linking to musl-fts
> [PATCH 3/5] setfiles: Musl compatibility for GLOB_BRACE and
> [PATCH 4/5] restorecond: Musl compatibility for GLOB_BRACE and
> [PATCH 5/5] genhomedircon: sysconf can return -1 without failure

Thanks! These patches look good to me and I would also appreciate if
they are merged in 2.8.

Nicolas

Re: SELinux musl support

[Jason Zaman]

On Thu, May 17, 2018 at 09:22:01PM +0200, Nicolas Iooss wrote:
> On Thu, May 17, 2018 at 7:11 AM, Jason Zaman <jason@perfinion.com> wrote:
> > This series fixes compiling and running on musl libc.
> >
> > patches 1-2 are fairly trivial.
> >
> > patches 3-4 are a feature change on that platform since it does not
> > support GLOB_TILDE and GLOB_BRACE. tilde is coming in musl 1.1.21
> > according to [1]. brace support is not documented anywhere or in the
> > example configs so that is probably not a big problem.
> >
> > patch 5 fixes a bug and it just happens that glibc returns a value for
> > sysconf and the error handling was wrong but never noticed.
> >
> > [1]: https://wiki.musl-libc.org/roadmap.html
> >
> > [PATCH 1/5] sestatus: include limits.h for PATH_MAX
> > [PATCH 2/5] libselinux: enable linking to musl-fts
> > [PATCH 3/5] setfiles: Musl compatibility for GLOB_BRACE and
> > [PATCH 4/5] restorecond: Musl compatibility for GLOB_BRACE and
> > [PATCH 5/5] genhomedircon: sysconf can return -1 without failure
> 
> Thanks! These patches look good to me and I would also appreciate if
> they are merged in 2.8.

Awesome, I'll push them on monday then unless there are any issues
before then :)
> 
> Nicolas
> 

Re: SELinux musl support

[Stephen Smalley]

On 05/18/2018 01:03 AM, Jason Zaman wrote:
> On Thu, May 17, 2018 at 09:22:01PM +0200, Nicolas Iooss wrote:
>> On Thu, May 17, 2018 at 7:11 AM, Jason Zaman <jason@perfinion.com> wrote:
>>> This series fixes compiling and running on musl libc.
>>>
>>> patches 1-2 are fairly trivial.
>>>
>>> patches 3-4 are a feature change on that platform since it does not
>>> support GLOB_TILDE and GLOB_BRACE. tilde is coming in musl 1.1.21
>>> according to [1]. brace support is not documented anywhere or in the
>>> example configs so that is probably not a big problem.
>>>
>>> patch 5 fixes a bug and it just happens that glibc returns a value for
>>> sysconf and the error handling was wrong but never noticed.
>>>
>>> [1]: https://wiki.musl-libc.org/roadmap.html
>>>
>>> [PATCH 1/5] sestatus: include limits.h for PATH_MAX
>>> [PATCH 2/5] libselinux: enable linking to musl-fts
>>> [PATCH 3/5] setfiles: Musl compatibility for GLOB_BRACE and
>>> [PATCH 4/5] restorecond: Musl compatibility for GLOB_BRACE and
>>> [PATCH 5/5] genhomedircon: sysconf can return -1 without failure
>>
>> Thanks! These patches look good to me and I would also appreciate if
>> they are merged in 2.8.
> 
> Awesome, I'll push them on monday then unless there are any issues
> before then :)

FWIW, the patches look fine to me as well and I did a test PR to trigger travis CI testing
and they passed,
https://github.com/SELinuxProject/selinux/pull/96

One possible area for improvement (but not necessary to merge this) would be to fix the
genhomedircon code to also then handle the case where getpwnam_r/getgrnam_r returns ERANGE,
in which case we are supposed to realloc a larger buffer and try again per the man page.

Re: SELinux musl support

[Jason Zaman]

On Fri, May 18, 2018 at 08:58:58AM -0400, Stephen Smalley wrote:
> On 05/18/2018 01:03 AM, Jason Zaman wrote:
> > On Thu, May 17, 2018 at 09:22:01PM +0200, Nicolas Iooss wrote:
> >> On Thu, May 17, 2018 at 7:11 AM, Jason Zaman <jason@perfinion.com> wrote:
> >>> This series fixes compiling and running on musl libc.
> >>>
> >>> patches 1-2 are fairly trivial.
> >>>
> >>> patches 3-4 are a feature change on that platform since it does not
> >>> support GLOB_TILDE and GLOB_BRACE. tilde is coming in musl 1.1.21
> >>> according to [1]. brace support is not documented anywhere or in the
> >>> example configs so that is probably not a big problem.
> >>>
> >>> patch 5 fixes a bug and it just happens that glibc returns a value for
> >>> sysconf and the error handling was wrong but never noticed.
> >>>
> >>> [1]: https://wiki.musl-libc.org/roadmap.html
> >>>
> >>> [PATCH 1/5] sestatus: include limits.h for PATH_MAX
> >>> [PATCH 2/5] libselinux: enable linking to musl-fts
> >>> [PATCH 3/5] setfiles: Musl compatibility for GLOB_BRACE and
> >>> [PATCH 4/5] restorecond: Musl compatibility for GLOB_BRACE and
> >>> [PATCH 5/5] genhomedircon: sysconf can return -1 without failure
> >>
> >> Thanks! These patches look good to me and I would also appreciate if
> >> they are merged in 2.8.
> > 
> > Awesome, I'll push them on monday then unless there are any issues
> > before then :)
> 
> FWIW, the patches look fine to me as well and I did a test PR to trigger travis CI testing
> and they passed,
> https://github.com/SELinuxProject/selinux/pull/96
> 
> One possible area for improvement (but not necessary to merge this) would be to fix the
> genhomedircon code to also then handle the case where getpwnam_r/getgrnam_r returns ERANGE,
> in which case we are supposed to realloc a larger buffer and try again per the man page.

Yeah i thought about that, but you're supposed to do that even when
sysconf does return a number and we've apparently never hit this before.
and the example code in the man pages doesnt really do it either. so as
small changes as possible before release is better. after release we could
think about it but probably still not really high importance.

-- Jason

Re: SELinux musl support

[Jason Zaman]

On Fri, May 18, 2018 at 11:58:58PM +0800, Jason Zaman wrote:
> On Fri, May 18, 2018 at 08:58:58AM -0400, Stephen Smalley wrote:
> > On 05/18/2018 01:03 AM, Jason Zaman wrote:
> > > On Thu, May 17, 2018 at 09:22:01PM +0200, Nicolas Iooss wrote:
> > >> On Thu, May 17, 2018 at 7:11 AM, Jason Zaman <jason@perfinion.com> wrote:
> > >>> This series fixes compiling and running on musl libc.
> > >>>
> > >>> patches 1-2 are fairly trivial.
> > >>>
> > >>> patches 3-4 are a feature change on that platform since it does not
> > >>> support GLOB_TILDE and GLOB_BRACE. tilde is coming in musl 1.1.21
> > >>> according to [1]. brace support is not documented anywhere or in the
> > >>> example configs so that is probably not a big problem.
> > >>>
> > >>> patch 5 fixes a bug and it just happens that glibc returns a value for
> > >>> sysconf and the error handling was wrong but never noticed.
> > >>>
> > >>> [1]: https://wiki.musl-libc.org/roadmap.html
> > >>>
> > >>> [PATCH 1/5] sestatus: include limits.h for PATH_MAX
> > >>> [PATCH 2/5] libselinux: enable linking to musl-fts
> > >>> [PATCH 3/5] setfiles: Musl compatibility for GLOB_BRACE and
> > >>> [PATCH 4/5] restorecond: Musl compatibility for GLOB_BRACE and
> > >>> [PATCH 5/5] genhomedircon: sysconf can return -1 without failure
> > >>
> > >> Thanks! These patches look good to me and I would also appreciate if
> > >> they are merged in 2.8.
> > > 
> > > Awesome, I'll push them on monday then unless there are any issues
> > > before then :)
> > 
> > FWIW, the patches look fine to me as well and I did a test PR to trigger travis CI testing
> > and they passed,
> > https://github.com/SELinuxProject/selinux/pull/96
> > 
> > One possible area for improvement (but not necessary to merge this) would be to fix the
> > genhomedircon code to also then handle the case where getpwnam_r/getgrnam_r returns ERANGE,
> > in which case we are supposed to realloc a larger buffer and try again per the man page.
> 
> Yeah i thought about that, but you're supposed to do that even when
> sysconf does return a number and we've apparently never hit this before.
> and the example code in the man pages doesnt really do it either. so as
> small changes as possible before release is better. after release we could
> think about it but probably still not really high importance.

This is all pushed to master now, travis-ci passes on master too.

-- Jason