commit c27c6bd2c4d6b6bb779f9b722d5607993e1d5e5c upstream.
Currently it is possible to specify a state machine table with 0 length,
this is not valid as optional tables are specified by not defining
the table as present. Further this allows by-passing the base tables
range check against the next/check tables.
Fixes: d901d6a298dc ("apparmor: dfa split verification of table headers")
Reported-by: Mike Salvatore <mike.salvatore@canonical.com>
Signed-off-by: John Johansen <john.johansen@canonical.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[ Upstream commit ca3fde5214e1d24f78269b337d3f22afd6bf445e ]
begin_current_label_crit_section() must run in sleepable context because
when label_is_stale() is true, aa_replace_current_label() runs, which uses
prepare_creds(), which can sleep.
Until now, the ptraceme access check (which runs with tasklist_lock held)
violated this rule.
Fixes: b2d09ae449ced ("apparmor: move ptrace checks to using labels")
Reported-by: Cyrill Gorcunov <gorcunov@gmail.com>
Reported-by: kernel test robot <rong.a.chen@intel.com>
Signed-off-by: Jann Horn <jannh@google.com>
Signed-off-by: John Johansen <john.johansen@canonical.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 3ed4aaa94fc07db3cd0c91be95e3e1b9782a2710 ] The subset test is not taking into account the unconfined exception which will cause profile transitions in the stacked confinement case to fail when no_new_privs is applied. This fixes a regression introduced in the fix for https://bugs.launchpad.net/bugs/1839037 BugLink: https://bugs.launchpad.net/bugs/1844186 Signed-off-by: John Johansen <john.johansen@canonical.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 3b646abc5bc6c0df649daea4c2c976bd4d47e4c8 ]
Currently apparmor_sk_clone_security() does not check for existing
label/peer in the 'new' struct sock; it just overwrites it, if any
(with another reference to the label of the source sock.)
static void apparmor_sk_clone_security(const struct sock *sk,
struct sock *newsk)
{
struct aa_sk_ctx *ctx = SK_CTX(sk);
struct aa_sk_ctx *new = SK_CTX(newsk);
new->label = aa_get_label(ctx->label);
new->peer = aa_get_label(ctx->peer);
}
This might leak label references, which might overflow under load.
Thus, check for and put labels, to prevent such errors.
Note this is similarly done on:
static int apparmor_socket_post_create(struct socket *sock, ...)
...
if (sock->sk) {
struct aa_sk_ctx *ctx = SK_CTX(sock->sk);
aa_put_label(ctx->label);
ctx->label = aa_get_label(label);
}
...
Context:
-------
The label reference count leak is observed if apparmor_sock_graft()
is called previously: this sets the 'ctx->label' field by getting
a reference to the current label (later overwritten, without put.)
static void apparmor_sock_graft(struct sock *sk, ...)
{
struct aa_sk_ctx *ctx = SK_CTX(sk);
if (!ctx->label)
ctx->label = aa_get_current_label();
}
And that is the case on crypto/af_alg.c:af_alg_accept():
int af_alg_accept(struct sock *sk, struct socket *newsock, ...)
...
struct sock *sk2;
...
sk2 = sk_alloc(...);
...
security_sock_graft(sk2, newsock);
security_sk_clone(sk, sk2);
...
Apparently both calls are done on their own right, especially for
other LSMs, being introduced in 2010/2014, before apparmor socket
mediation in 2017 (see commits [1,2,3,4]).
So, it looks OK there! Let's fix the reference leak in apparmor.
Test-case:
---------
Exercise that code path enough to overflow label reference count.
$ cat aa-refcnt-af_alg.c
#include <stdio.h>
#include <string.h>
#include <unistd.h>
#include <sys/socket.h>
#include <linux/if_alg.h>
int main() {
int sockfd;
struct sockaddr_alg sa;
/* Setup the crypto API socket */
sockfd = socket(AF_ALG, SOCK_SEQPACKET, 0);
if (sockfd < 0) {
perror("socket");
return 1;
}
memset(&sa, 0, sizeof(sa));
sa.salg_family = AF_ALG;
strcpy((char *) sa.salg_type, "rng");
strcpy((char *) sa.salg_name, "stdrng");
if (bind(sockfd, (struct sockaddr *) &sa, sizeof(sa)) < 0) {
perror("bind");
return 1;
}
/* Accept a "connection" and close it; repeat. */
while (!close(accept(sockfd, NULL, 0)));
return 0;
}
$ gcc -o aa-refcnt-af_alg aa-refcnt-af_alg.c
$ ./aa-refcnt-af_alg
<a few hours later>
[ 9928.475953] refcount_t overflow at apparmor_sk_clone_security+0x37/0x70 in aa-refcnt-af_alg[1322], uid/euid: 1000/1000
...
[ 9928.507443] RIP: 0010:apparmor_sk_clone_security+0x37/0x70
...
[ 9928.514286] security_sk_clone+0x33/0x50
[ 9928.514807] af_alg_accept+0x81/0x1c0 [af_alg]
[ 9928.516091] alg_accept+0x15/0x20 [af_alg]
[ 9928.516682] SYSC_accept4+0xff/0x210
[ 9928.519609] SyS_accept+0x10/0x20
[ 9928.520190] do_syscall_64+0x73/0x130
[ 9928.520808] entry_SYSCALL_64_after_hwframe+0x3d/0xa2
Note that other messages may be seen, not just overflow, depending on
the value being incremented by kref_get(); on another run:
[ 7273.182666] refcount_t: saturated; leaking memory.
...
[ 7273.185789] refcount_t: underflow; use-after-free.
Kprobes:
-------
Using kprobe events to monitor sk -> sk_security -> label -> count (kref):
Original v5.7 (one reference leak every iteration)
... (af_alg_accept+0x0/0x1c0) label=0xffff8a0f36c25eb0 label_refcnt=0x11fd2
... (af_alg_release_parent+0x0/0xd0) label=0xffff8a0f36c25eb0 label_refcnt=0x11fd4
... (af_alg_accept+0x0/0x1c0) label=0xffff8a0f36c25eb0 label_refcnt=0x11fd3
... (af_alg_release_parent+0x0/0xd0) label=0xffff8a0f36c25eb0 label_refcnt=0x11fd5
... (af_alg_accept+0x0/0x1c0) label=0xffff8a0f36c25eb0 label_refcnt=0x11fd4
... (af_alg_release_parent+0x0/0xd0) label=0xffff8a0f36c25eb0 label_refcnt=0x11fd6
Patched v5.7 (zero reference leak per iteration)
... (af_alg_accept+0x0/0x1c0) label=0xffff9ff376c25eb0 label_refcnt=0x593
... (af_alg_release_parent+0x0/0xd0) label=0xffff9ff376c25eb0 label_refcnt=0x594
... (af_alg_accept+0x0/0x1c0) label=0xffff9ff376c25eb0 label_refcnt=0x593
... (af_alg_release_parent+0x0/0xd0) label=0xffff9ff376c25eb0 label_refcnt=0x594
... (af_alg_accept+0x0/0x1c0) label=0xffff9ff376c25eb0 label_refcnt=0x593
... (af_alg_release_parent+0x0/0xd0) label=0xffff9ff376c25eb0 label_refcnt=0x594
Commits:
-------
[1] commit 507cad355fc9 ("crypto: af_alg - Make sure sk_security is initialized on accept()ed sockets")
[2] commit 4c63f83c2c2e ("crypto: af_alg - properly label AF_ALG socket")
[3] commit 2acce6aa9f65 ("Networking") a.k.a ("crypto: af_alg - Avoid sock_graft call warning)
[4] commit 56974a6fcfef ("apparmor: add base infastructure for socket mediation")
Fixes: 56974a6fcfef ("apparmor: add base infastructure for socket mediation")
Reported-by: Brian Moyles <bmoyles@netflix.com>
Signed-off-by: Mauricio Faria de Oliveira <mfo@canonical.com>
Signed-off-by: John Johansen <john.johansen@canonical.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit dd2569fbb053719f7df7ef8fdbb45cf47156a701 ]
Fix two issues with introspecting the task mode.
1. If a task is attached to a unconfined profile that is not the
ns->unconfined profile then. Mode the mode is always reported
as -
$ ps -Z
LABEL PID TTY TIME CMD
unconfined 1287 pts/0 00:00:01 bash
test (-) 1892 pts/0 00:00:00 ps
instead of the correct value of (unconfined) as shown below
$ ps -Z
LABEL PID TTY TIME CMD
unconfined 2483 pts/0 00:00:01 bash
test (unconfined) 3591 pts/0 00:00:00 ps
2. if a task is confined by a stack of profiles that are unconfined
the output of label mode is again the incorrect value of (-) like
above, instead of (unconfined). This is because the visibile
profile count increment is skipped by the special casing of
unconfined.
Fixes: f1bd904175e8 ("apparmor: add the base fns() for domain labels")
Signed-off-by: John Johansen <john.johansen@canonical.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
commit c6b39f070722ea9963ffe756bfe94e89218c5e63 upstream.
policy_update() invokes begin_current_label_crit_section(), which
returns a reference of the updated aa_label object to "label" with
increased refcount.
When policy_update() returns, "label" becomes invalid, so the refcount
should be decreased to keep refcount balanced.
The reference counting issue happens in one exception handling path of
policy_update(). When aa_may_manage_policy() returns not NULL, the
refcnt increased by begin_current_label_crit_section() is not decreased,
causing a refcnt leak.
Fix this issue by jumping to "end_section" label when
aa_may_manage_policy() returns not NULL.
Fixes: 5ac8c355ae00 ("apparmor: allow introspecting the loaded policy pre internal transform")
Signed-off-by: Xiyu Yang <xiyuyang19@fudan.edu.cn>
Signed-off-by: Xin Tan <tanxin.ctf@gmail.com>
Signed-off-by: John Johansen <john.johansen@canonical.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit a0b845ffa0d91855532b50fc040aeb2d8338dca4 upstream.
aa_change_profile() invokes aa_get_current_label(), which returns
a reference of the current task's label.
According to the comment of aa_get_current_label(), the returned
reference must be put with aa_put_label().
However, when the original object pointed by "label" becomes
unreachable because aa_change_profile() returns or a new object
is assigned to "label", reference count increased by
aa_get_current_label() is not decreased, causing a refcnt leak.
Fix this by calling aa_put_label() before aa_change_profile() return
and dropping unnecessary aa_get_current_label().
Fixes: 9fcf78cca198 ("apparmor: update domain transitions that are subsets of confinement at nnp")
Signed-off-by: Xiyu Yang <xiyuyang19@fudan.edu.cn>
Signed-off-by: Xin Tan <tanxin.ctf@gmail.com>
Signed-off-by: John Johansen <john.johansen@canonical.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit c54d481d71c6849e044690d3960aaebc730224cc upstream.
In the implementation of aa_audit_rule_init(), when aa_label_parse()
fails the allocated memory for rule is released using
aa_audit_rule_free(). But after this release, the return statement
tries to access the label field of the rule which results in
use-after-free. Before releasing the rule, copy errNo and return it
after release.
Fixes: 52e8c38001d8 ("apparmor: Fix memory leak of rule on error exit path")
Signed-off-by: Navid Emamdoost <navid.emamdoost@gmail.com>
Signed-off-by: John Johansen <john.johansen@canonical.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[ Upstream commit 1f8266ff58840d698a1e96d2274189de1bdf7969 ]
As a comment above begin_current_label_crit_section() explains,
begin_current_label_crit_section() must run in sleepable context because
when label_is_stale() is true, aa_replace_current_label() runs, which uses
prepare_creds(), which can sleep.
Until now, the ptrace access check (which runs with a task lock held)
violated this rule.
Also add a might_sleep() assertion to begin_current_label_crit_section(),
because asserts are less likely to be ignored than comments.
Fixes: b2d09ae449ced ("apparmor: move ptrace checks to using labels")
Signed-off-by: Jann Horn <jannh@google.com>
Signed-off-by: John Johansen <john.johansen@canonical.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 5f997580e8b12b9f585e34cc16304925d26ce49e ]
The netperf benchmark shows a 5.73% reduction in throughput for
small (64 byte) transfers by unconfined tasks.
DEFINE_AUDIT_SK() in aa_label_sk_perm() should not be performed
unconditionally, rather only when the label is confined.
netperf-tcp
56974a6fc^ 56974a6fc
Min 64 563.48 ( 0.00%) 531.17 ( -5.73%)
Min 128 1056.92 ( 0.00%) 999.44 ( -5.44%)
Min 256 1945.95 ( 0.00%) 1867.97 ( -4.01%)
Min 1024 6761.40 ( 0.00%) 6364.23 ( -5.87%)
Min 2048 11110.53 ( 0.00%) 10606.20 ( -4.54%)
Min 3312 13692.67 ( 0.00%) 13158.41 ( -3.90%)
Min 4096 14926.29 ( 0.00%) 14457.46 ( -3.14%)
Min 8192 18399.34 ( 0.00%) 18091.65 ( -1.67%)
Min 16384 21384.13 ( 0.00%) 21158.05 ( -1.06%)
Hmean 64 564.96 ( 0.00%) 534.38 ( -5.41%)
Hmean 128 1064.42 ( 0.00%) 1010.12 ( -5.10%)
Hmean 256 1965.85 ( 0.00%) 1879.16 ( -4.41%)
Hmean 1024 6839.77 ( 0.00%) 6478.70 ( -5.28%)
Hmean 2048 11154.80 ( 0.00%) 10671.13 ( -4.34%)
Hmean 3312 13838.12 ( 0.00%) 13249.01 ( -4.26%)
Hmean 4096 15009.99 ( 0.00%) 14561.36 ( -2.99%)
Hmean 8192 18975.57 ( 0.00%) 18326.54 ( -3.42%)
Hmean 16384 21440.44 ( 0.00%) 21324.59 ( -0.54%)
Stddev 64 1.24 ( 0.00%) 2.85 (-130.64%)
Stddev 128 4.51 ( 0.00%) 6.53 ( -44.84%)
Stddev 256 11.67 ( 0.00%) 8.50 ( 27.16%)
Stddev 1024 48.33 ( 0.00%) 75.07 ( -55.34%)
Stddev 2048 54.82 ( 0.00%) 65.16 ( -18.86%)
Stddev 3312 153.57 ( 0.00%) 56.29 ( 63.35%)
Stddev 4096 100.25 ( 0.00%) 88.50 ( 11.72%)
Stddev 8192 358.13 ( 0.00%) 169.99 ( 52.54%)
Stddev 16384 43.99 ( 0.00%) 141.82 (-222.39%)
Signed-off-by: Tony Jones <tonyj@suse.de>
Fixes: 56974a6fcfef ("apparmor: add base infastructure for socket
mediation")
Signed-off-by: John Johansen <john.johansen@canonical.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>