rpm_host_install: fix TypeError with SELinux bindings

[giuseppe]

It solves a TypeError exception:

Traceback (most recent call last): File "/bin/atomic", line 185, in sys.exit(_func()) File "/usr/lib/python2.7/site-packages/Atomic/install.py", line 134, in install return be.install(self.image, self.name) File "/usr/lib/python2.7/site-packages/Atomic/backends/_ostree.py", line 128, in install return self.syscontainers.install(image, name) File "/usr/lib/python2.7/site-packages/Atomic/syscontainers.py", line 563, in install return_value = self._install(image, name) File "/usr/lib/python2.7/site-packages/Atomic/syscontainers.py", line 686, in _install self._checkout_wrapper(repo, name, image, 0, SystemContainers.CHECKOUT_MODE_INSTALL, values=values, remote=self.args.remote, system_package=self.args.system_package) File "/usr/lib/python2.7/site-packages/Atomic/syscontainers.py", line 797, in _checkout_wrapper return self._checkout(repo, options) File "/usr/lib/python2.7/site-packages/Atomic/syscontainers.py", line 813, in _checkout return self._do_checkout(repo, options) File "/usr/lib/python2.7/site-packages/Atomic/syscontainers.py", line 1126, in _do_checkout rpm_install_content = self._handle_system_package_files(options, manifest, exports) File "/usr/lib/python2.7/site-packages/Atomic/syscontainers.py", line 469, in _handle_system_package_files new_installed_files_checksum = RPMHostInstall.rm_add_files_to_host(options["installed_files_checksum"], exports, options["prefix"] or "/", files_template=installed_files_template, values=options["values"], rename_files=rename_files, use_links=use_links) File "/usr/lib/python2.7/site-packages/Atomic/rpm_host_install.py", line 155, in rm_add_files_to_host created = RPMHostInstall._copyfile(selinux_hnd, src_file, dest_path, try_hardlink=try_hardlink) File "/usr/lib/python2.7/site-packages/Atomic/rpm_host_install.py", line 24, in _copyfile ctx = selinux.selabel_lookup_raw(selinux_hnd, dest, mode) TypeError: in method 'selabel_lookup_raw', argument 3 of type 'char const *'

Closes: https://bugzilla.redhat.com/show_bug.cgi?id=1542144

Signed-off-by: Giuseppe Scrivano gscrivan@redhat.com

[peterbaouoft]

EDIT: the output is from tests.log when running test_system_containers_rpm.sh with the command: make && make test TEST_INTEGRATION=system_containers_rpm

Before: ( with print out mode for debugging ) https://paste.fedoraproject.org/paste/srt5x9kLulqGj9Smicl1nA

Now: ( with patch applied) https://paste.fedoraproject.org/paste/GW5X8O8DQUBCc9kd-sqSWQ

[root@localhost atomic]# git diff HEAD~
diff --git a/Atomic/rpm_host_install.py b/Atomic/rpm_host_install.py
index ed3a123..680d575 100644
--- a/Atomic/rpm_host_install.py
+++ b/Atomic/rpm_host_install.py
@@ -21,7 +21,7 @@ class RPMHostInstall(object):
             except OSError:
                 pass

-            ctx = selinux.selabel_lookup_raw(selinux_hnd, dest, mode)
+            ctx = selinux.selabel_lookup_raw(selinux_hnd, str(dest), mode)
             selinux.setfscreatecon_raw(ctx[1])

         if os.path.isdir(src):
@@ -144,7 +144,7 @@ class RPMHostInstall(object):
                             data = src_file_obj.read()

                         if selinux_hnd is not None:
-                            ctx = selinux.selabel_lookup_raw(selinux_hnd, dest_path, os.stat(src_file).st_mode)
+                            ctx = selinux.selabel_lookup_raw(selinux_hnd, str(dest_path), os.stat(src_file).st_mode)
                             selinux.setfscreatecon_raw(ctx[1])

                         util.write_template(src_file, data, values or {}, dest_path)

P.S: if needed more info, I can try produce them :-), let me know

[giuseppe]

@rh-atomic-bot delegate=peterbaouoft

[rh-atomic-bot]

:v: @peterbaouoft can now approve this pull request

[peterbaouoft]

One quick question, is https://github.com/projectatomic/atomic/issues/1176 related to this patch? ( From the issue, it looks like to be a skopeo error and not related?)

If they are not related, then LGTM :).

P.S: side question, why casting destination to string fix the problem? Is it because the unicode string(i.e: https://docs.python.org/2/howto/unicode.html#the-unicode-type) can not be properly recognized?

[giuseppe]

no, #1176 is a different issue

[peterbaouoft]

no, #1176 is a different issue

sounds good, then LGTM, @rh-atomic-bot r+ c36f6a5

[rh-atomic-bot]

:zap: Test exempted: pull fully rebased and already tested.