On Sun, Nov 15, 2020 at 11:37:18AM +0100, Christian Brauner wrote:
This adds a whole test suite for idmapped mounts but in order to
ensure that
there are no regression for the vfs itself it also includes tests for correct
functionality on non-idmapped mounts. The following tests are currently
available with more to come in the future:
Awesome! :)
Some glitches in the build, though... something about the ordering or
the Make rules produces odd results on a failure:
$ make
gcc -g -I../../../../usr/include/ -Wall -O2 -pthread xattr.c internal.h utils.c utils.h
-lcap -o /home/kees/src/linux-build/seccomp/tools/testing/selftests/idmap_mounts/xattr
gcc -g -I../../../../usr/include/ -Wall -O2 -pthread core.c internal.h utils.c utils.h
-lcap -o /home/kees/src/linux-build/seccomp/tools/testing/selftests/idmap_mounts/core
core.c:19:10: fatal error: sys/acl.h: No such file or directory
19 | #include <sys/acl.h>
| ^~~~~~~~~~~
compilation terminated.
make: *** [../lib.mk:139:
/home/kees/src/linux-build/seccomp/tools/testing/selftests/idmap_mounts/core]
Error 1
$ make
make: Nothing to be done for 'all'.
$ file xattr core
xattr: ELF 64-bit LSB shared object, x86-64, version 1 (SYSV), dynamically linked,
interpreter /lib64/ld-linux-x86-64.so.2,
BuildID[sha1]=7a3c1951e54f20e657b4181c1be77c7183a54f81, for GNU/Linux 3.2.0, with
debug_info, not stripped
core: GCC precompiled header (version 014) for C
Even after I install libacl1-dev, I still get a "core" file output which
breaks attempts to build again. :)
Is there any way to have the test suite not depend on
__NR_mount_setattr? Running this test on older kernels fails everything.
--
Kees Cook