It seems to be unused since its initial addition in 76ba6eaa
("Squashed 'libsepol/cil/' changes from 08520e9..28ad56e").
Signed-off-by: Christian Göttsche <cgzones@googlemail.com>
Acked-by: James Carter <jwcart2@gmail.com>
Make it more obvious which parameters are read-only and not being
modified and allow callers to pass const pointers.
Signed-off-by: Christian Göttsche <cgzones@googlemail.com>
Make it more obvious which parameters are read-only and not being
modified and allow callers to pass const pointers.
Signed-off-by: Christian Göttsche <cgzones@googlemail.com>
Acked-by: James Carter <jwcart2@gmail.com>
Check return values of memory allocation functions and propagate their
failure.
Signed-off-by: Christian Göttsche <cgzones@googlemail.com>
Acked-by: James Carter <jwcart2@gmail.com>
GCC 12 produces an array-bounds warning:
In file included from ../include/sepol/policydb/context.h:23,
from ../include/sepol/policydb/policydb.h:62,
from ../cil/src/cil_binary.c:41:
In function ‘mls_level_init’,
inlined from ‘mls_level_destroy’ at ../include/sepol/policydb/mls_types.h:99:2,
inlined from ‘mls_level_destroy’ at ../include/sepol/policydb/mls_types.h:92:20,
inlined from ‘mls_range_destroy’ at ../include/sepol/policydb/mls_types.h:149:2,
inlined from ‘cil_rangetransition_to_policydb’ at ../cil/src/cil_binary.c:3231:6:
../include/sepol/policydb/mls_types.h:89:9: error: ‘memset’ offset [0, 23] is out of the bounds [0, 0] [-Werror=array-bounds]
89 | memset(level, 0, sizeof(mls_level_t));
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
../include/sepol/policydb/mls_types.h:89:9: error: ‘memset’ offset [0, 23] is out of the bounds [0, 0] [-Werror=array-bounds]
cc1: all warnings being treated as errors
This is a false positive, by inspecting the code and compiling with -O3
and -flto.
Closes: https://github.com/SELinuxProject/selinux/issues/339
Signed-off-by: Christian Göttsche <cgzones@googlemail.com>
Acked-by: James Carter <jwcart2@gmail.com>
In some circumstances, like semanage-store being on overlayfs, rename()
could fail with EXDEV - Invalid cross-device link. This is due to the
fact that overlays doesn't support rename() if source and target are not
on the same layer, e.g. in containers built from several layers. Even
though it's not atomic operation, it's better to try to copy files from
src to dst on our own in this case. Next rebuild will probably not fail
as the new directories will be on the same layer.
Fixes: https://github.com/SELinuxProject/selinux/issues/343
Reproducer:
$ cd selinux1
$ cat Dockerfile
FROM fedora:35
RUN dnf install -y selinux-policy selinux-policy-targeted
$ podman build -t localhost/selinux . --no-cache
$ cd ../selinux2
$ cat Dockerfile
FROM localhost/selinux
RUN semodule -B
$ podman build -t localhost/selinux2 . --no-cache
STEP 2/2: RUN semodule -B
libsemanage.semanage_commit_sandbox: Error while renaming /var/lib/selinux/targeted/active to /var/lib/selinux/targeted/previous. (Invalid cross-device link).
semodule: Failed!
Error: error building at STEP "RUN semodule -B": error while running runtime: exit status 1
With the fix:
$ podman build -t localhost/selinux2 . --no-cache
STEP 2/2: RUN semodule -B
libsemanage.semanage_rename: Warning: rename(/var/lib/selinux/targeted/active, /var/lib/selinux/targeted/previous) failed: Invalid cross-device link, fall back to non-atomic semanage_copy_dir_flags()
COMMIT localhost/selinux2
--> d2cfcebc1a1
Successfully tagged localhost/selinux2:latest
d2cfcebc1a1b34f1c2cd661ac18292b0612c3e5fa71d6fa1441be244da91b1af
Reported-by: Joseph Marrero Corchado <jmarrero@redhat.com>
Signed-off-by: Petr Lautrbach <plautrba@redhat.com>
Acked-by: Ondrej Mosnacek <omosnace@redhat.com>
Add the file SECURITY.md which describes the SELinux userspace
security vulnerability handling process including who to contact.
Signed-off-by: James Carter <jwcart2@gmail.com>
When validating a policydb, validate the conditional expressions
including the values of the booleans within them.
Found by oss-fuzz (#45523)
Signed-off-by: James Carter <jwcart2@gmail.com>
Use calloc() instead of mallocarray() so that everything is
initialized to zero to prevent the use of unitialized memory when
validating malformed binary policies.
Found by oss-fuzz (#45493)
Signed-off-by: James Carter <jwcart2@gmail.com>
The MAX_LOG_SIZE is 512. It is possible that a log message could
exceed the max size (such as for neverallowx rules). If so, then
write out "<LOG MESSAGE TRUNCATED>", so that it is obvious that
the log message has been truncated.
Reported-by: Jonathan Hettwer <j2468h@googlemail.com>
Signed-off-by: James Carter <jwcart2@gmail.com>
When writing a conf file or CIL policy out from a kernel binary,
do not write out a constraint rule if it has no permissions.
Signed-off-by: James Carter <jwcart2@gmail.com>
Since CIL allows permission expressions, it is possible for the
expression to evaluate to no permissions. If this is the case,
then don't add the constraint.
Signed-off-by: James Carter <jwcart2@gmail.com>
Shorten "CAPABILITY" to "CAP" following the kernel naming convention.
The SELinux policy capability enum names should now follow the
"POLICYDB_CAP_XXX" format.
Signed-off-by: Richard Haines <richard_c_haines@btinternet.com>
Acked-by: James Carter <jwcart2@gmail.com>
Extended permission and neverallow rules are not permitted in
conditional statements.
This causes issues on policy optimization where avtab_search() might
return a non extended permission rule when searching for one.
Found by oss-fuzz (#45327)
Signed-off-by: Christian Göttsche <cgzones@googlemail.com>
When validating a policydb, do a more thorough validation of the
constraints.
- No permissions if it is a (mls)validatetrans.
- Only mlsvalidatetrans can use u3, r3, and t3.
- Expressions not involving types should have an empty type set.
- Only "==" and "!=" are allowed when there are names.
- If names are not used in an expression then both the names bitmap
and the type set should be empty.
- Only roles and mls expressions can used "dom", "domby", and "incomp".
- An mls expression cannot use names.
- If the expression is "not", "and", or "or", then the names bitmap
and the type set should be empty.
Signed-off-by: James Carter <jwcart2@gmail.com>
In f0a5f6e, calls to reallocarray were introduced. Ensure that the
correct header (private.h) is included when necessary.
Fixes: f0a5f6e ("libsepol: use reallocarray wrapper to avoid overflows")
Signed-off-by: Thiébaud Weksteen <tweek@google.com>
Acked-by: James Carter <jwcart2@gmail.com>
Test: Built using Android CI (glibc 2.17)
Commit 93902fc834 ("setfiles/restorecon: support parallel relabeling")
implemented support for parallel relabeling in setfiles. This is
available for fixfiles now.
Signed-off-by: Petr Lautrbach <plautrba@redhat.com>
Acked-by: Ondrej Mosnacek <omosnace@redhat.com>
Compiler can optimize calls to memset(3), due to the as-if rule, away if
the object is not accessed later on. Use a wrapper using volatile
pointers to ensure the memory is guaranteed to be erased. Also erase
the encrypted password.
Signed-off-by: Christian Göttsche <cgzones@googlemail.com>
Depending on the implementation crypt(3) can fail either by returning
NULL, or returning a pointer to an invalid hash and setting errno.
Signed-off-by: Christian Göttsche <cgzones@googlemail.com>
newrole.c:636:12: warning: function declaration isn’t a prototype [-Wstrict-prototypes]
636 | static int transition_to_caller_uid()
| ^~~~~~~~~~~~~~~~~~~~~~~~
newrole.c:103:9: warning: macro is not used [-Wunused-macros]
#define DEFAULT_CONTEXT_SIZE 255 /* first guess at context size */
^
newrole.c:862:4: warning: 'break' will never be executed [-Wunreachable-code-break]
break;
^~~~~
newrole.c:168:13: warning: no previous extern declaration for non-static variable 'service_name' [-Wmissing-variable-declarations]
const char *service_name = "newrole";
^
hashtab.c:53:11: warning: implicit conversion changes signedness: 'unsigned int' to 'int' [-Wsign-conversion]
hvalue = h->hash_value(h, key);
~ ^~~~~~~~~~~~~~~~~~~~~
hashtab.c:92:11: warning: implicit conversion changes signedness: 'unsigned int' to 'int' [-Wsign-conversion]
hvalue = h->hash_value(h, key);
~ ^~~~~~~~~~~~~~~~~~~~~
hashtab.c:124:11: warning: implicit conversion changes signedness: 'unsigned int' to 'int' [-Wsign-conversion]
hvalue = h->hash_value(h, key);
~ ^~~~~~~~~~~~~~~~~~~~~
hashtab.c:172:10: warning: implicit conversion changes signedness: 'int' to 'unsigned int' [-Wsign-conversion]
ret = apply(cur->key, cur->datum, args);
~ ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
hashtab.c:174:12: warning: implicit conversion changes signedness: 'unsigned int' to 'int' [-Wsign-conversion]
return ret;
~~~~~~ ^~~
Signed-off-by: Christian Göttsche <cgzones@googlemail.com>
On the first loop iteration the variables `r_buf` and `reason_buf_used`
are NULL respective 0. Please UBSAN by not adding them but instead
directly assign NULL.
services.c:800:16: runtime error: applying zero offset to null pointer
#0 0x4d4fce in constraint_expr_eval_reason ./libsepol/src/services.c:800:16
#1 0x4cf31a in sepol_validate_transition_reason_buffer ./libsepol/src/services.c:1079:8
Signed-off-by: Christian Göttsche <cgzones@googlemail.com>
"semanage login -a" accepts whitespaces in user/group name
(e.g. users/groups from Active Directory), which may lead to issues down
the line since libsemanage doesn't expect whitespaces in
/var/lib/selinux/targeted/active/seusers and other config files.
Fixes:
Artificial but simple reproducer
# groupadd server_admins
# sed -i "s/^server_admins/server admins/" /etc/group
# semanage login -a -s staff_u %server\ admins
# semanage login -l (or "semodule -B")
libsemanage.parse_assert_ch: expected character ':', but found 'a' (/var/lib/selinux/targeted/active/seusers: 6):
%server admins:staff_u:s0-s0:c0.c1023 (No such file or directory).
libsemanage.seuser_parse: could not parse seuser record (No such file or directory).
libsemanage.dbase_file_cache: could not cache file database (No such file or directory).
libsemanage.enter_ro: could not enter read-only section (No such file or directory).
FileNotFoundError: [Errno 2] No such file or directory
Signed-off-by: Vit Mojzis <vmojzis@redhat.com>
Spaces before values in /etc/selinux/config should be ignored just as
spaces after them are.
E.g. "SELINUXTYPE= targeted" should be a valid value.
Fixes:
# sed -i 's/^SELINUXTYPE=/SELINUXTYPE= /g' /etc/selinux/config
# dnf install <any_package>
...
RPM: error: selabel_open: (/etc/selinux/ targeted/contexts/files/file_contexts) No such file or directory
RPM: error: Plugin selinux: hook tsm_pre failed
...
Error: Could not run transaction.
Signed-off-by: Vit Mojzis <vmojzis@redhat.com>
When parsing Reference Policy style files accept square brackets in file
names. The FILENAME token is used in the TYPE_TRANSITION grammar rule
for the optional name based argument. This name can contain square
brackets, e.g. for anonymous inodes like "[io_uring]".
Signed-off-by: Christian Göttsche <cgzones@googlemail.com>
When an assertion fails, the error message refers to a generic
"policy.conf" file. When parsing a policy in checkpolicy, populate its
name using the original filename (source_filename is still build using
the #line directives within the policy).
Signed-off-by: Thiébaud Weksteen <tweek@google.com>
[Merge conflicts fixed by: James Carter <jwcart2@gmail.com>]
Signed-off-by: James Carter <jwcart2@gmail.com>
When there are conflicting context rules, the location of the
conflicting rules are written out. If there are many duplicates of
the same context rule, there will be many pairs of conflicts written
out. This hides the fact that all of the rules are the same and can
make it hard to see the different conflicts.
First, since these are warnings and not reported at the default log
verbosity level (which only reports errors), only search for the
locations of the conflicting rules when the verbosity level means
that the warnings will actually be reported.
Second, Report all the duplicate conflicting rules together.
Third, Report the first four conflicts of the same rule if when
the verbosity level is at CIL_WARN ("-v") and report all of them
when the verbosity level is at CIL_INFO or higher ("-v -v").
Fixes problem found by oss-fuzz (#39735)
Signed-off-by: James Carter <jwcart2@gmail.com>
When there is a neverallow violation, a search is made for all of
the rules that violate the neverallow. The violating rules as well
as their parents are written out to make it easier to find these
rules.
If there is a lot of rules that violate a neverallow, then this
amount of reporting is too much. Instead, only print out the first
four rules (with their parents) that match the violated neverallow
rule along with the total number of rules that violate the
neverallow at the default log level. Report all the violations when
at a higher verbosity level.
Signed-off-by: James Carter <jwcart2@gmail.com>
Commit 4b2e2a248e (libsepol/cil: Limit
the amount of reporting for bounds failures) limited the number of
bounds failures that were reported to the first two matching rules
for the first two bad rules.
Instead, report the first two matching rules for the first four bad
rules at the default log level and report all matching rules for all
bad rules for higher verbosity levels.
Signed-off-by: James Carter <jwcart2@gmail.com>
Not all violations of neverallowxperm rules were being reported.
In check_assertion_extended_permissions_avtab(), a break was
performed after finding a match rather than just returning right
away. This means that if other src and tgt pairs were checked
afterward that did not match, then no match would be reported.
Example:
allow attr attr:CLASS ioctl;
allowxperm attr attr:CLASS ioctl 0x9401;
allowxperm t1 self:CLASS ioctl 0x9421;
neverallowxperm attr self:CLASS ioctl 0x9421;
Would result in no assertion violations being found.
Another problem was that the reporting function did not properly
recognize when there was a valid allowxperm rule and falsely
reported additional violations that did not exist. (There had
to be at least one legitimate violation.)
Using the same example as above (and assuming t1 and t2 both have
attribute attr), the following would be reported as:
neverallowxperm on line 4 of policy.conf (or line 4 of policy.conf)
violated by
allowxperm t1 t1:CLASS ioctl { 0x9421 };
neverallowxperm on line 4 of policy.conf (or line 4 of policy.conf)
violated by
allow t2 t2:CLASS4 { ioctl };
There is no violation for t2 because there is a valid allowxperm
rule for it.
With this patch, only the first error message (which is the correct
one) is printed.
Signed-off-by: James Carter <jwcart2@gmail.com>
The changes are the same as in a patch sent by Christian Göttsche
<cgzones@googlemail.com> to support adding not-self to neverallowxperm
checking, but it is needed for normal neverallowxperm checking as well
and the following explanation reflects that.
When reporting neverallowxperm violations, the avtab is searched to
find the rule that violates the assertion. If the avtab pointer of
the args is not set, then it will report the error as if no extended
permissions existed for the source and target (so allowing the ioctl
permission at all violates the neverallowxperm).
Example (where t1 has attribute attr):
allow attr attr:CLASS ioctl;
allowxperm attr attr:CLASS ioctl 0x9411;
neverallowxperm t1 self:CLASS ioctl 0x9411;
Would be reported as:
neverallowxperm on line 3 of policy.conf (or line 3 of policy.conf)
violated by
allow t1 t1:CLASS { ioctl };
Instead of:
neverallowxperm on line 3 of policy.conf (or line 3 of policy.conf)
violated by
allowxperm attr attr:CLASS ioctl { 0x9411 };
Reported-by: Christian Göttsche <cgzones@googlemail.com>
Signed-off-by: James Carter <jwcart2@gmail.com>
When checking for violations of neverallow rules, if the neverallow
uses self, then the src and tgt must be the same when checking
extended permissions and when reporting violations.
Example:
allow attr attr : CLASS PERM;
neverallow attr self : CLASS PERM;
If the types t1 and t2 have attribute attr, then the violations
that would be reported would be:
allow t1 t1 : CLASS PERM;
allow t1 t2 : CLASS PERM;
allow t2 t1 : CLASS PERM;
allow t2 t2 : CLASS PERM;
instead of:
allow t1 t1 : CLASS PERM;
allow t2 t2 : CLASS PERM;
Signed-off-by: James Carter <jwcart2@gmail.com>
The value returned from report_assertion_extended_permissions() is
the nubmer of errors, so call it that instead of ret.
Signed-off-by: James Carter <jwcart2@gmail.com>
In both check_assertion_extended_permissions() and
report_assertion_avtab_matches(), when checking for a match involving
a rule using self, the matches between the source and target of the
rule being checked are found using ebitmap_and() and then the matches
between that result and the source of the neverallow are found using
another ebitmap_and() call.
Since the matches between the sources of the rule being checked and
the neverallow have already been found, just find the matches between
that result and the target of the rule being checked. This only
requires one call to ebitmap_and() instead of two.
Signed-off-by: James Carter <jwcart2@gmail.com>
When check_assertion_extended_permissions() is called, it has already
been determined that there is a match, and, since neither the class
nor the permissions are used, there is no need for the check.
Signed-off-by: James Carter <jwcart2@gmail.com>
Inorder to differentiate errors from matches, use "(rc < 0)" when
calling ebitmap_* functions while checking neverallow rules.
Also, just use rc instead of having a separate variable (ret) in
check_assertion_extended_permissions().
Signed-off-by: James Carter <jwcart2@gmail.com>
If a neverallow has target types as well as using self and a match
is found with the target types, then self does not even need to
be checked, since the rule is already in violation of the assertion.
So move the check for a match of the target types before dealing with
self.
Signed-off-by: James Carter <jwcart2@gmail.com>
In check_assertion_avtab_match(), for the functions that do not return
an error, but only returns 0 or 1 depending on if a match is found,
call the function in an if statement.
Signed-off-by: James Carter <jwcart2@gmail.com>
Return an error if check_assertion_extended_permissions() returns
an error instead of treating it as an assertion violation.
Signed-off-by: James Carter <jwcart2@gmail.com>
Christian Göttsche
Petr Lautrbach
James Carter
Topi Miettinen
Richard Haines
Thiébaud Weksteen
Vit Mojzis