directories). It also disables the longpath test (after locking up
another machine from running it), but adds a 'make alltests' target
which will run all the tests plus that one (and any others defined in
the RISKY_TESTS variable.)
pam_apparmor and here's a patch to address most of them--
* header comment was incorrect
* use pam_get_user() instead of pam_get_item()
* return an error if we're unable to change to the DEFAULT hat
In addition, this has a fix to make sure that the magic token we read
from /dev/urandom is not null (which would cause the hat probing to fail
if we need to fall back to the DEFAULT hat).
if necessary. Currently used to suppress warnings about unsafe/unfiltered
environment 'u' exec transitions.
[Corresponds to rev 6415 in the old svn repo]
This commit adds a test that checks to see if exec unconfined -->
confined requires any access to the binary being run confined. Current
behavior is that it does not.
the old svn tree.]
This patch adds tests to verify the environment filtering done in the Px
and Ux cases. It tests the environment from both an elf executable and
from a shell script. Also verifies that with the apparmor module loaded,
environment filtering on setuid apps still happens.
Relicense file to GPL (don't know why it wasn't caught originally). Made
it look in the right location, but doesn't appear to function -- I don't
know if that's due to language changes or if it ever worked. Seth Arnold
will need to comment.
This (updated) patch to trunk adds support for Px and Ux (toggle
bprm_secure on exec) in the parser, As requested, lowercase p and u
corresponds to an unfiltered environmnet on exec, uppercase will filter
the environment. It applies after the 'm' patch.
As a side effect, I tried to reduce the use of hardcoded characters in
the debugging statements -- there are still a few warnings that have
hard coded letters in them; not sure I can fix them all.
This version issues a warning for every unsafe ux and issues a single
warning for the first 'R', 'W', 'X', 'L', and 'I' it encounters,
except when the "-q" or "--quiet" flag , "--remove" profile flag, or
"-N" report names flags are passed. Unfortunately, it made the logic
somewhat more convoluted. Wordsmithing improvements welcome.
This (updated) patch to trunk adds the m flag to the parser language. The
m flag explicitly does -not- conflict with px, ux, or ix.
It does not add exec mmap as implicit to inherited execs, as it was
asserted that the module should do this.
I have not fixed up the testcases to match.
