The default audit_backlog_limit is 64.  This was a reasonable limit at one time.

systemd causes so much audit queue activity on startup that auditd doesn't
start before the backlog queue has already overflowed by more than a factor of
2.  On a system with audit= not set on the kernel command line, this isn't an
issue since that history isn't kept for auditd when it is available.  On a
system with audit=1 set on the kernel command line, kaudit tries to keep that
history until auditd is able to drain the queue.

This default can be changed by the "-b" option in audit.rules once the system
has booted, but won't help with lost messages on boot.

One way to solve this would be to increase the default backlog queue size to
avoid losing any messages before auditd is able to consume them.  This would
be overkill to the embedded community and insufficient for some servers.

Another way to solve it might be to add a kconfig option to set the default
based on the system type.  An embedded system would get the current (or
smaller) default, while Workstations might get more than now and servers might
get more.

None of these solutions helps if a system's compiled default is too small to
see the lost messages without compiling a new kernel.

This patch adds a kernel set-up parameter (audit already has one to
enable/disable it) "audit_backlog_limit=<n>" that overrides the default to
allow the system administrator to set the backlog limit.
Signed-off-by: NRichard Guy Briggs <rgb@redhat.com>
Signed-off-by: NEric Paris <eparis@redhat.com>

Add the "audit=" kernel start-up parameter to Documentation/kernel-parameters.txt.
Signed-off-by: NRichard Guy Briggs <rgb@redhat.com>
Signed-off-by: NEric Paris <eparis@redhat.com>

These and similar errors were seen on a patched 3.8 kernel when the
audit subsystem was overrun during boot:

  udevd[876]: worker [887] unexpectedly returned with status 0x0100
  udevd[876]: worker [887] failed while handling
'/devices/pci0000:00/0000:00:03.0/0000:40:00.0'
  udevd[876]: worker [880] unexpectedly returned with status 0x0100
  udevd[876]: worker [880] failed while handling
'/devices/LNXSYSTM:00/LNXPWRBN:00/input/input1/event1'

  udevadm settle - timeout of 180 seconds reached, the event queue
contains:
    /sys/devices/LNXSYSTM:00/LNXPWRBN:00/input/input1/event1 (3995)
    /sys/devices/LNXSYSTM:00/LNXSYBUS:00/PNP0A08:00/INT3F0D:00 (4034)

  audit: audit_backlog=258 > audit_backlog_limit=256
  audit: audit_lost=1 audit_rate_limit=0 audit_backlog_limit=256

The change below increases the efficiency of the audit code and prevents it
from being overrun:

Use add_wait_queue_exclusive() in wait_for_auditd() to put the
thread on the wait queue.  When kauditd dequeues an skb, all
of the waiting threads are waiting for the same resource, but
only one is going to get it, so there's no need to wake up
more than one waiter.

See: https://lkml.org/lkml/2013/9/2/479Signed-off-by: NDan Duval <dan.duval@oracle.com>
Signed-off-by: NChuck Anderson <chuck.anderson@oracle.com>
Signed-off-by: NRichard Guy Briggs <rgb@redhat.com>
Signed-off-by: NEric Paris <eparis@redhat.com>

These and similar errors were seen on a patched 3.8 kernel when the
audit subsystem was overrun during boot:

  udevd[876]: worker [887] unexpectedly returned with status 0x0100
  udevd[876]: worker [887] failed while handling
'/devices/pci0000:00/0000:00:03.0/0000:40:00.0'
  udevd[876]: worker [880] unexpectedly returned with status 0x0100
  udevd[876]: worker [880] failed while handling
'/devices/LNXSYSTM:00/LNXPWRBN:00/input/input1/event1'

  udevadm settle - timeout of 180 seconds reached, the event queue
contains:
    /sys/devices/LNXSYSTM:00/LNXPWRBN:00/input/input1/event1 (3995)
    /sys/devices/LNXSYSTM:00/LNXSYBUS:00/PNP0A08:00/INT3F0D:00 (4034)

  audit: audit_backlog=258 > audit_backlog_limit=256
  audit: audit_lost=1 audit_rate_limit=0 audit_backlog_limit=256

The change below increases the efficiency of the audit code and prevents it
from being overrun:

Only issue a wake_up in kauditd if the length of the skb queue is less than the
backlog limit.  Otherwise, threads waiting in wait_for_auditd() will simply
wake up, discover that the queue is still too long for them to proceed, and go
back to sleep.  This results in wasted context switches and machine cycles.
kauditd_thread() is the only function that removes buffers from audit_skb_queue
so we can't race.  If we did, the timeout in wait_for_auditd() would expire and
the waiting thread would continue.

See: https://lkml.org/lkml/2013/9/2/479Signed-off-by: NDan Duval <dan.duval@oracle.com>
Signed-off-by: NChuck Anderson <chuck.anderson@oracle.com>
Signed-off-by: NRichard Guy Briggs <rgb@redhat.com>
Signed-off-by: NEric Paris <eparis@redhat.com>

If wait_for_auditd() times out, go immediately to the error function rather
than retesting the loop conditions.
Signed-off-by: NRichard Guy Briggs <rgb@redhat.com>
Signed-off-by: NEric Paris <eparis@redhat.com>

When the audit queue overflows and times out (audit_backlog_wait_time), the
audit queue overflow timeout is set to zero.  Once the audit queue overflow
timeout condition recovers, the timeout should be reset to the original value.

See also:
	https://lkml.org/lkml/2013/9/2/473

Cc: stable@vger.kernel.org # v3.8-rc4+
Signed-off-by: NLuiz Capitulino <lcapitulino@redhat.com>
Signed-off-by: NDan Duval <dan.duval@oracle.com>
Signed-off-by: NChuck Anderson <chuck.anderson@oracle.com>
Signed-off-by: NRichard Guy Briggs <rgb@redhat.com>
Signed-off-by: NEric Paris <eparis@redhat.com>

Convert audit from only listening in init_net to use register_pernet_subsys()
to dynamically manage the netlink socket list.
Signed-off-by: NRichard Guy Briggs <rgb@redhat.com>
Signed-off-by: NEric Paris <eparis@redhat.com>

When being refactored from audit_log_start() to audit_log_task_info(), in
commit e23eb920 the tty and ses fields in the log output got transposed.
Restore to original order to avoid breaking search tools.
Signed-off-by: NRichard Guy Briggs <rgb@redhat.com>
Signed-off-by: NEric Paris <eparis@redhat.com>

Normally, netlink ports use the PID of the userspace process as the port ID.
If the PID is already in use by a port, the kernel will allocate another port
ID to avoid conflict.  Re-name all references to netlink ports from pid to
portid to reflect this reality and avoid confusion with actual PIDs.  Ports
use the __u32 type, so re-type all portids accordingly.

(This patch is very similar to ebiederman's 5deadd69)
Signed-off-by: NRichard Guy Briggs <rgb@redhat.com>
Signed-off-by: NGao feng <gaofeng@cn.fujitsu.com>
Signed-off-by: NEric Paris <eparis@redhat.com>

- Always report the current process as capset now always only works on
  the current process.  This prevents reporting 0 or a random pid in
  a random pid namespace.

- Don't bother to pass the pid as is available.
Signed-off-by: N"Eric W. Biederman" <ebiederm@xmission.com>
(cherry picked from commit bcc85f0af31af123e32858069eb2ad8f39f90e67)
(cherry picked from commit f911cac4556a7a23e0b3ea850233d13b32328692)
Signed-off-by: NRichard Guy Briggs <rgb@redhat.com>
[eparis: fix build error when audit disabled]
Signed-off-by: NEric Paris <eparis@redhat.com>

Linux 3.12

Conflicts:
	fs/exec.c

sfr pointed out that with CONFIG_UIDGID_STRICT_TYPE_CHECKS set the audit
tree would not build.  This is because the oldsessionid in
audit_set_loginuid() was accidentally being declared as a kuid_t.  This
patch fixes that declaration mistake.

Example of problem:
kernel/auditsc.c: In function 'audit_set_loginuid':
kernel/auditsc.c:2003:15: error: incompatible types when assigning to
type 'kuid_t' from type 'int'
  oldsessionid = audit_get_sessionid(current);
Reported-by: NStephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: NEric Paris <eparis@redhat.com>

Move the audit_bprm() call from search_binary_handler() to exec_binprm().  This
allows us to get rid of the mm member of struct audit_aux_data_execve since
bprm->mm will equal current->mm.

This also mitigates the issue that ->argc could be modified by the
load_binary() call in search_binary_handler().

audit_bprm() was being called to add an AUDIT_EXECVE record to the audit
context every time search_binary_handler() was recursively called.  Only one
reference is necessary.
Reported-by: NOleg Nesterov <onestero@redhat.com>
Cc: Eric Paris <eparis@redhat.com>
Signed-off-by: NRichard Guy Briggs <rgb@redhat.com>
Signed-off-by: NEric Paris <eparis@redhat.com>
---
This patch is against 3.11, but was developed on Oleg's post-3.11 patches that
introduce exec_binprm().

audit_bprm() was being called to add an AUDIT_EXECVE record to the audit
context every time search_binary_handler() was recursively called.  Only one
reference is necessary, so just update it.  Move the the contents of
audit_aux_data_execve into the union in audit_context, removing dependence on a
kmalloc along the way.
Reported-by: NOleg Nesterov <onestero@redhat.com>
Cc: Eric Paris <eparis@redhat.com>
Signed-off-by: NRichard Guy Briggs <rgb@redhat.com>
Signed-off-by: NEric Paris <eparis@redhat.com>

Get rid of write-only audit_aux_data_exeve structure member envc.
Signed-off-by: NRichard Guy Briggs <rgb@redhat.com>
Signed-off-by: NEric Paris <eparis@redhat.com>

Signed-off-by: N"Eric W. Biederman" <ebiederm@xmission.com>
(cherry picked from ebiederman commit 6904431d6b41190e42d6b94430b67cb4e7e6a4b7)
Signed-off-by: NEric Paris <eparis@redhat.com>

commit ab61d38e tried to merge the
invalid filter checking into a single function.  However AUDIT_INODE
filters were not verified in the new generic checker.  Thus such rules
were being denied even though they were perfectly valid.

Ex:
$ auditctl -a exit,always -F arch=b64 -S open -F key=/foo -F inode=6955 -F devmajor=9 -F devminor=1
Error sending add rule data request (Invalid argument)
Signed-off-by: NEric Paris <eparis@redhat.com>
Signed-off-by: NRichard Guy Briggs <rgb@redhat.com>
Signed-off-by: NEric Paris <eparis@redhat.com>

Supress the stock memory allocation failure warnings for audit buffers
since audit alreay takes care of memory allocation failure warnings, including
rate-limiting, in audit_log_start().
Signed-off-by: NRichard Guy Briggs <rgb@redhat.com>
Signed-off-by: NEric Paris <eparis@redhat.com>

...to make it clear what the intent behind each record's operation was.

In many cases you can infer this, based on the context of the syscall
and the result. In other cases it's not so obvious. For instance, in
the case where you have a file being renamed over another, you'll have
two different records with the same filename but different inode info.
By logging this information we can clearly tell which one was created
and which was deleted.

This fixes what was broken in commit bfcec708.
Commit 79f6530c should also be backported to stable v3.7+.
Signed-off-by: NJeff Layton <jlayton@redhat.com>
Signed-off-by: NEric Paris <eparis@redhat.com>
Signed-off-by: NRichard Guy Briggs <rgb@redhat.com>
Signed-off-by: NEric Paris <eparis@redhat.com>

Historically, when a syscall that creates a dentry fails, you get an audit
record that looks something like this (when trying to create a file named
"new" in "/tmp/tmp.SxiLnCcv63"):

    type=PATH msg=audit(1366128956.279:965): item=0 name="/tmp/tmp.SxiLnCcv63/new" inode=2138308 dev=fd:02 mode=040700 ouid=0 ogid=0 rdev=00:00 obj=staff_u:object_r:user_tmp_t:s15:c0.c1023

This record makes no sense since it's associating the inode information for
"/tmp/tmp.SxiLnCcv63" with the path "/tmp/tmp.SxiLnCcv63/new". The recent
patch I posted to fix the audit_inode call in do_last fixes this, by making it
look more like this:

    type=PATH msg=audit(1366128765.989:13875): item=0 name="/tmp/tmp.DJ1O8V3e4f/" inode=141 dev=fd:02 mode=040700 ouid=0 ogid=0 rdev=00:00 obj=staff_u:object_r:user_tmp_t:s15:c0.c1023

While this is more correct, if the creation of the file fails, then we
have no record of the filename that the user tried to create.

This patch adds a call to audit_inode_child to may_create. This creates
an AUDIT_TYPE_CHILD_CREATE record that will sit in place until the
create succeeds. When and if the create does succeed, then this record
will be updated with the correct inode info from the create.

This fixes what was broken in commit bfcec708.
Commit 79f6530c should also be backported to stable v3.7+.
Signed-off-by: NJeff Layton <jlayton@redhat.com>
Signed-off-by: NEric Paris <eparis@redhat.com>
Signed-off-by: NRichard Guy Briggs <rgb@redhat.com>
Signed-off-by: NEric Paris <eparis@redhat.com>

In send/GET, we don't want the kernel to lie about what value is set.
Signed-off-by: NRichard Guy Briggs <rgb@redhat.com>
Signed-off-by: NEric Paris <eparis@redhat.com>

Using the nlmsg_len member of the netlink header to test if the message
is valid is wrong as it includes the size of the netlink header itself.
Thereby allowing to send short netlink messages that pass those checks.

Use nlmsg_len() instead to test for the right message length. The result
of nlmsg_len() is guaranteed to be non-negative as the netlink message
already passed the checks of nlmsg_ok().

Also switch to min_t() to please checkpatch.pl.

Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Eric Paris <eparis@redhat.com>
Cc: stable@vger.kernel.org  # v2.6.6+ for the 1st hunk, v2.6.23+ for the 2nd
Signed-off-by: NMathias Krause <minipli@googlemail.com>
Signed-off-by: NRichard Guy Briggs <rgb@redhat.com>
Signed-off-by: NEric Paris <eparis@redhat.com>

We currently are setting fields to 0 to initialize the structure
declared on the stack.  This is a bad idea as if the structure has holes
or unpacked space these will not be initialized.  Just use memset.  This
is not a performance critical section of code.
Signed-off-by: NEric Paris <eparis@redhat.com>

We leak 4 bytes of kernel stack in response to an AUDIT_GET request as
we miss to initialize the mask member of status_set. Fix that.

Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Eric Paris <eparis@redhat.com>
Cc: stable@vger.kernel.org  # v2.6.6+
Signed-off-by: NMathias Krause <minipli@googlemail.com>
Signed-off-by: NRichard Guy Briggs <rgb@redhat.com>
Signed-off-by: NEric Paris <eparis@redhat.com>

It appears this one comparison function got missed in f368c07d (and 9c937dcc).
Signed-off-by: NRichard Guy Briggs <rgb@redhat.com>
Signed-off-by: NEric Paris <eparis@redhat.com>

This adds a new 'audit_feature' bit which allows userspace to set it
such that the loginuid is absolutely immutable, even if you have
CAP_AUDIT_CONTROL.
Signed-off-by: NEric Paris <eparis@redhat.com>
Signed-off-by: NRichard Guy Briggs <rgb@redhat.com>
Signed-off-by: NEric Paris <eparis@redhat.com>

This is a new audit feature which only grants processes with
CAP_AUDIT_CONTROL the ability to unset their loginuid.  They cannot
directly set it from a valid uid to another valid uid.  The ability to
unset the loginuid is nice because a priviledged task, like that of
container creation, can unset the loginuid and then priv is not needed
inside the container when a login daemon needs to set the loginuid.
Signed-off-by: NEric Paris <eparis@redhat.com>
Signed-off-by: NRichard Guy Briggs <rgb@redhat.com>
Signed-off-by: NEric Paris <eparis@redhat.com>

If a task has CAP_AUDIT_CONTROL allow that task to unset their loginuid.
This would allow a child of that task to set their loginuid without
CAP_AUDIT_CONTROL.  Thus when launching a new login daemon, a
priviledged helper would be able to unset the loginuid and then the
daemon, which may be malicious user facing, do not need priv to function
correctly.
Signed-off-by: NEric Paris <eparis@redhat.com>
Signed-off-by: NRichard Guy Briggs <rgb@redhat.com>
Signed-off-by: NEric Paris <eparis@redhat.com>

After trying to use this feature in Fedora we found the hard coding
policy like this into the kernel was a bad idea.  Surprise surprise.
We ran into these problems because it was impossible to launch a
container as a logged in user and run a login daemon inside that container.
This reverts back to the old behavior before this option was added.  The
option will be re-added in a userspace selectable manor such that
userspace can choose when it is and when it is not appropriate.
Signed-off-by: NEric Paris <eparis@redhat.com>
Signed-off-by: NRichard Guy Briggs <rgb@redhat.com>
Signed-off-by: NEric Paris <eparis@redhat.com>

This is just a code rework.  It makes things more readable.  It does not
make any functional changes.

It does change the log messages to include both the old session id as
well the new and it includes a new res field, which means we get
messages even when the user did not have permission to change the
loginuid.
Signed-off-by: NEric Paris <eparis@redhat.com>
Signed-off-by: NRichard Guy Briggs <rgb@redhat.com>
Signed-off-by: NEric Paris <eparis@redhat.com>

We use the read check to get the feature set (like AUDIT_GET) and the
write check to set the features (like AUDIT_SET).
Signed-off-by: NEric Paris <eparis@redhat.com>
Signed-off-by: NRichard Guy Briggs <rgb@redhat.com>
Signed-off-by: NEric Paris <eparis@redhat.com>

The audit_status structure was not designed with extensibility in mind.
Define a new AUDIT_SET_FEATURE message type which takes a new structure
of bits where things can be enabled/disabled/locked one at a time.  This
structure should be able to grow in the future while maintaining forward
and backward compatibility (based loosly on the ideas from capabilities
and prctl)

This does not actually add any features, but is just infrastructure to
allow new on/off types of audit system features.
Signed-off-by: NEric Paris <eparis@redhat.com>
Signed-off-by: NRichard Guy Briggs <rgb@redhat.com>
Signed-off-by: NEric Paris <eparis@redhat.com>

SFR reported this 2013-05-15:

> After merging the final tree, today's linux-next build (i386 defconfig)
> produced this warning:
>
> kernel/auditfilter.c: In function 'audit_data_to_entry':
> kernel/auditfilter.c:426:3: warning: this decimal constant is unsigned only
> in ISO C90 [enabled by default]
>
> Introduced by commit 780a7654 ("audit: Make testing for a valid
> loginuid explicit") from Linus' tree.

Replace this decimal constant in the code with a macro to make it more readable
(add to the unsigned cast to quiet the warning).

Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: NRichard Guy Briggs <rgb@redhat.com>
Signed-off-by: NEric Paris <eparis@redhat.com>

When the audit=1 kernel parameter is absent and auditd is not running,
AUDIT_USER_AVC messages are being silently discarded.

AUDIT_USER_AVC messages should be sent to userspace using printk(), as
mentioned in the commit message of 4a4cd633 ("AUDIT: Optimise the
audit-disabled case for discarding user messages").

When audit_enabled is 0, audit_receive_msg() discards all user messages
except for AUDIT_USER_AVC messages. However, audit_log_common_recv_msg()
refuses to allocate an audit_buffer if audit_enabled is 0. The fix is to
special case AUDIT_USER_AVC messages in both functions.

It looks like commit 50397bd1 ("[AUDIT] clean up audit_receive_msg()")
introduced this bug.

Cc: <stable@kernel.org> # v2.6.25+
Signed-off-by: NTyler Hicks <tyhicks@canonical.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Eric Paris <eparis@redhat.com>
Cc: linux-audit@redhat.com
Acked-by: NKees Cook <keescook@chromium.org>
Signed-off-by: NRichard Guy Briggs <rgb@redhat.com>
Signed-off-by: NEric Paris <eparis@redhat.com>

If audit_filter_task() nacks the new thread it makes sense
to clear TIF_SYSCALL_AUDIT which can be copied from parent
by dup_task_struct().

A wrong TIF_SYSCALL_AUDIT is not really bad but it triggers
the "slow" audit paths in entry.S to ensure the task can not
miss audit_syscall_*() calls, this is pointless if the task
has no ->audit_context.
Signed-off-by: NOleg Nesterov <oleg@redhat.com>
Acked-by: NSteve Grubb <sgrubb@redhat.com>
Acked-by: NEric Paris <eparis@redhat.com>
Signed-off-by: NRichard Guy Briggs <rgb@redhat.com>
Signed-off-by: NEric Paris <eparis@redhat.com>

Remove it.
Signed-off-by: NGao feng <gaofeng@cn.fujitsu.com>
Signed-off-by: NRichard Guy Briggs <rgb@redhat.com>
Signed-off-by: NEric Paris <eparis@redhat.com>

A newline was accidentally added during session ID helper refactorization in
commit 4d3fb709.  This needlessly uses up buffer space, messes up syslog
formatting and makes userspace processing less efficient.  Remove it.
Signed-off-by: NRichard Guy Briggs <rgb@redhat.com>
Acked-by: NEric Paris <eparis@redhat.com>
Signed-off-by: NEric Paris <eparis@redhat.com>

Signed-off-by: NIlya V. Matveychikov <matvejchikov@gmail.com>
Signed-off-by: NRichard Guy Briggs <rgb@redhat.com>
Signed-off-by: NEric Paris <eparis@redhat.com>

Messages of type AUDIT_USER_TTY were being formatted to 1024 octets,
truncating messages approaching MAX_AUDIT_MESSAGE_LENGTH (8970 octets).

Set the formatting to 8560 characters, given maximum estimates for prefix and
suffix budgets.

See the problem discussion:
https://www.redhat.com/archives/linux-audit/2009-January/msg00030.html

And the new size rationale:
https://www.redhat.com/archives/linux-audit/2013-September/msg00016.html

Test ~8k messages with:
auditctl -m "$(for i in $(seq -w 001 820);do echo -n "${i}0______";done)"
Reported-by: NLC Bruzenak <lenny@magitekltd.com>
Reported-by: NJustin Stephenson <jstephen@redhat.com>
Signed-off-by: NRichard Guy Briggs <rgb@redhat.com>
Signed-off-by: NEric Paris <eparis@redhat.com>