user/sven/linux.git/kernel/audit.c, branch v4.14.5

audit: update the function comments

2017-09-05T13:46:59Z

Update the function comments to match the code. Signed-off-by: Geliang Tang Signed-off-by: Paul Moore

audit: Reduce overhead using a coarse clock

2017-09-05T13:46:54Z

Commit 2115bb250f26 ("audit: Use timespec64 to represent audit timestamps") noted that audit timestamps were not y2038 safe and used a 64-bit timestamp. In itself, this makes sense but the conversion was from CURRENT_TIME to ktime_get_real_ts64() which is a heavier call to record an accurate timestamp which is required in some, but not all, cases. The impact is that when auditd is running without any rules that all syscalls have higher overhead. This is visible in the sysbench-thread benchmark as a 11.5% performance hit. That benchmark is dumb as rocks but it's also visible in redis as an 8-10% hit on all operations which is of greater concern. It is somewhat stupid of audit to track syscalls without any rules related to syscalls but that is how it behaves. The overhead can be directly measured with perf comparing 4.9 with 4.12 4.9 7.76% sysbench [kernel.vmlinux] [k] __schedule 7.62% sysbench [kernel.vmlinux] [k] _raw_spin_lock 7.37% sysbench libpthread-2.22.so [.] __lll_lock_elision 7.29% sysbench [kernel.vmlinux] [.] syscall_return_via_sysret 6.59% sysbench [kernel.vmlinux] [k] native_sched_clock 5.21% sysbench libc-2.22.so [.] __sched_yield 4.38% sysbench [kernel.vmlinux] [k] entry_SYSCALL_64 4.28% sysbench [kernel.vmlinux] [k] do_syscall_64 3.49% sysbench libpthread-2.22.so [.] __lll_unlock_elision 3.13% sysbench [kernel.vmlinux] [k] __audit_syscall_exit 2.87% sysbench [kernel.vmlinux] [k] update_curr 2.73% sysbench [kernel.vmlinux] [k] pick_next_task_fair 2.31% sysbench [kernel.vmlinux] [k] syscall_trace_enter 2.20% sysbench [kernel.vmlinux] [k] __audit_syscall_entry ..... 0.00% swapper [kernel.vmlinux] [k] read_tsc 4.12 7.84% sysbench [kernel.vmlinux] [k] __schedule 7.05% sysbench [kernel.vmlinux] [k] _raw_spin_lock 6.57% sysbench libpthread-2.22.so [.] __lll_lock_elision 6.50% sysbench [kernel.vmlinux] [.] syscall_return_via_sysret 5.95% sysbench [kernel.vmlinux] [k] read_tsc 5.71% sysbench [kernel.vmlinux] [k] native_sched_clock 4.78% sysbench libc-2.22.so [.] __sched_yield 4.30% sysbench [kernel.vmlinux] [k] entry_SYSCALL_64 3.94% sysbench [kernel.vmlinux] [k] do_syscall_64 3.37% sysbench libpthread-2.22.so [.] __lll_unlock_elision 3.32% sysbench [kernel.vmlinux] [k] __audit_syscall_exit 2.91% sysbench [kernel.vmlinux] [k] __getnstimeofday64 Note the additional overhead from read_tsc which goes from 0% to 5.95%. This is on a single-socket E3-1230 but similar overheads have been measured on an older machine which the patch also eliminates. The patch in question has no explanation as to why a fully-accurate timestamp is required and is likely an oversight. Using a coarser, but monotically increasing, timestamp the overhead can be eliminated. While it can be worked around by configuring or disabling audit, it's tricky enough to detect that a kernel fix is justified. With this patch, we see the following; sysbenchthread 4.9.0 4.12.0 4.12.0 vanilla vanilla coarse-v1r1 Amean 1 1.49 ( 0.00%) 1.66 ( -11.42%) 1.51 ( -1.34%) Amean 3 1.48 ( 0.00%) 1.65 ( -11.45%) 1.50 ( -0.96%) Amean 5 1.49 ( 0.00%) 1.67 ( -12.31%) 1.51 ( -1.83%) Amean 7 1.49 ( 0.00%) 1.66 ( -11.72%) 1.50 ( -0.67%) Amean 12 1.48 ( 0.00%) 1.65 ( -11.57%) 1.52 ( -2.89%) Amean 16 1.49 ( 0.00%) 1.65 ( -11.13%) 1.51 ( -1.73%) The benchmark is reporting the time required for different thread counts to lock/unlock a private mutex which, while dense, demonstrates the syscall overhead. This is showing that 4.12 took a 11-12% hit but the overhead is almost eliminated by the patch. While the variance is not reported here, it's well within the noise with the patch applied. Signed-off-by: Mel Gorman Acked-by: Arnd Bergmann Acked-by: Deepa Dinamani Signed-off-by: Paul Moore

Merge branch 'stable-4.13' of git://git.infradead.org/users/pcmoore/audit

2017-07-20T17:22:26Z

Pull audit fix from Paul Moore: "A small audit fix, just a single line, to plug a memory leak in some audit error handling code" * 'stable-4.13' of git://git.infradead.org/users/pcmoore/audit: audit: fix memleak in auditd_send_unicast_skb.

audit: fix memleak in auditd_send_unicast_skb.

2017-07-19T14:28:54Z

Found this issue by kmemleak report, auditd_send_unicast_skb did not free skb if rcu_dereference(auditd_conn) returns null. unreferenced object 0xffff88082568ce00 (size 256): comm "auditd", pid 1119, jiffies 4294708499 backtrace: [] kmemleak_alloc+0x4a/0xa0 [] kmem_cache_alloc_node+0xcc/0x210 [] __alloc_skb+0x5d/0x290 [] audit_make_reply+0x54/0xd0 [] audit_receive_msg+0x967/0xd70 ---------------- (gdb) list *audit_receive_msg+0x967 0xffffffff8113dff7 is in audit_receive_msg (kernel/audit.c:1133). 1132 skb = audit_make_reply(0, AUDIT_REPLACE, 0, 0, &pvnr, sizeof(pvnr)); --------------- [] audit_receive+0x52/0xa0 [] netlink_unicast+0x181/0x240 [] netlink_sendmsg+0x2c2/0x3b0 [] sock_sendmsg+0x38/0x50 [] SYSC_sendto+0x102/0x190 [] SyS_sendto+0xe/0x10 [] entry_SYSCALL_64_fastpath+0x1a/0xa5 [] 0xffffffffffffffff Signed-off-by: Shu Wang Signed-off-by: Paul Moore

Merge branch 'stable-4.13' of git://git.infradead.org/users/pcmoore/audit

2017-07-05T18:24:05Z

Pull audit updates from Paul Moore: "Things are relatively quiet on the audit front for v4.13, just five patches for a total diffstat of 102 lines. There are two patches from Richard to consistently record the POSIX capabilities and add the ambient capability information as well. I also chipped in two patches to fix a race condition with the auditd tracking code and ensure we don't skip sending any records to the audit multicast group. Finally a single style fix that I accepted because I must have been in a good mood that day. Everything passes our test suite, and should be relatively harmless, please merge for v4.13" * 'stable-4.13' of git://git.infradead.org/users/pcmoore/audit: audit: make sure we never skip the multicast broadcast audit: fix a race condition with the auditd tracking code audit: style fix audit: add ambient capabilities to CAPSET and BPRM_FCAPS records audit: unswing cap_* fields in PATH records

audit: make sure we never skip the multicast broadcast

2017-06-16T15:51:00Z

When the auditd connection is reset, either intentionally or due to a failure, any records that were in the main backlog queue would not be sent in a multicast broadcast. This patch fixes this problem by not flushing the main backlog queue on a connection reset, the main kauditd_thread() will take care of that normally. Resolves: https://github.com/linux-audit/audit-kernel/issues/41 Reviewed-by: Richard Guy Briggs Signed-off-by: Paul Moore

audit: fix a race condition with the auditd tracking code

2017-06-13T19:19:43Z

Originally reported by Adam and Dusty, it appears we have a small race window in kauditd_thread(), as documented in the Fedora BZ: * https://bugzilla.redhat.com/show_bug.cgi?id=1459326#c35 "This issue is partly due to the read-copy nature of RCU, and partly due to how we sync the auditd_connection state across kauditd_thread and the audit control channel. The kauditd_thread thread is always running so it can service the record queues and emit the multicast messages, if it happens to be just past the "main_queue" label, but before the "if (sk == NULL || ...)" if-statement which calls auditd_reset() when the new auditd connection is registered it could end up resetting the auditd connection, regardless of if it is valid or not. This is a rather small window and the variable nature of multi-core scheduling explains why this is proving rather difficult to reproduce." The fix is to have functions only call auditd_reset() when they believe that the kernel/auditd connection is still valid, e.g. non-NULL, and to have these callers pass their local copy of the auditd_connection pointer to auditd_reset() where it can be compared with the current connection state before resetting. If the caller has a stale state tracking pointer then the reset is ignored. We also make a small change to kauditd_thread() so that if the kernel/auditd connection is dead we skip the retry queue and send the records straight to the hold queue. This is necessary as we used to rely on auditd_reset() to occasionally purge the retry queue but we are going to be calling the reset function much less now and we want to make sure the retry queue doesn't grow unbounded. Reported-by: Adam Williamson Reported-by: Dusty Mabe Reviewed-by: Richard Guy Briggs Signed-off-by: Paul Moore

audit: unswing cap_* fields in PATH records

2017-05-23T20:50:02Z

The cap_* fields swing in and out of PATH records. If no capabilities are set, the cap_* fields are completely missing and when one of the cap_fi or cap_fp values is empty, that field is omitted. Original: type=PATH msg=audit(04/20/2017 12:17:11.222:193) : item=1 name=/lib64/ld-linux-x86-64.so.2 inode=787694 dev=08:03 mode=file,755 ouid=root ogid=root rdev=00:00 obj=system_u:object_r:ld_so_t:s0 nametype=NORMAL type=PATH msg=audit(04/20/2017 12:17:11.222:193) : item=0 name=/home/sleep inode=1319469 dev=08:03 mode=file,suid,755 ouid=root ogid=root rdev=00:00 obj=system_u:object_r:bin_t:s0 nametype=NORMAL cap_fp=sys_admin cap_fe=1 cap_fver=2 Normalize the PATH record by always printing all 4 cap_* fields. Fixed: type=PATH msg=audit(04/20/2017 13:01:31.679:201) : item=1 name=/lib64/ld-linux-x86-64.so.2 inode=787694 dev=08:03 mode=file,755 ouid=root ogid=root rdev=00:00 obj=system_u:object_r:ld_so_t:s0 nametype=NORMAL cap_fp=none cap_fi=none cap_fe=0 cap_fver=0 type=PATH msg=audit(04/20/2017 13:01:31.679:201) : item=0 name=/home/sleep inode=1319469 dev=08:03 mode=file,suid,755 ouid=root ogid=root rdev=00:00 obj=system_u:object_r:bin_t:s0 nametype=NORMAL cap_fp=sys_admin cap_fi=none cap_fe=1 cap_fver=2 See: https://github.com/linux-audit/audit-kernel/issues/42 Signed-off-by: Richard Guy Briggs Signed-off-by: Paul Moore

Merge branch 'stable-4.12' of git://git.infradead.org/users/pcmoore/audit

2017-05-03T16:21:59Z

Pull audit updates from Paul Moore: "Fourteen audit patches for v4.12 that span the full range of fixes, new features, and internal cleanups. We have a patches to move to 64-bit timestamps, convert refcounts from atomic_t to refcount_t, track PIDs using the pid struct instead of pid_t, convert our own private audit buffer cache to a standard kmem_cache, log kernel module names when they are unloaded, and normalize the NETFILTER_PKT to make the userspace folks happier. From a fixes perspective, the most important is likely the auditd connection tracking RCU fix; it was a rather brain dead bug that I'll take the blame for, but thankfully it didn't seem to affect many people (only one report). I think the patch subject lines and commit descriptions do a pretty good job of explaining the details and why the changes are important so I'll point you there instead of duplicating it here; as usual, if you have any questions you know where to find us. We also manage to take out more code than we put in this time, that always makes me happy :)" * 'stable-4.12' of git://git.infradead.org/users/pcmoore/audit: audit: fix the RCU locking for the auditd_connection structure audit: use kmem_cache to manage the audit_buffer cache audit: Use timespec64 to represent audit timestamps audit: store the auditd PID as a pid struct instead of pid_t audit: kernel generated netlink traffic should have a portid of 0 audit: combine audit_receive() and audit_receive_skb() audit: convert audit_watch.count from atomic_t to refcount_t audit: convert audit_tree.count from atomic_t to refcount_t audit: normalize NETFILTER_PKT netfilter: use consistent ipv4 network offset in xt_AUDIT audit: log module name on delete_module audit: remove unnecessary semicolon in audit_watch_handle_event() audit: remove unnecessary semicolon in audit_mark_handle_event() audit: remove unnecessary semicolon in audit_field_valid()

audit: fix the RCU locking for the auditd_connection structure

2017-05-02T14:16:05Z

Cong Wang correctly pointed out that the RCU read locking of the auditd_connection struct was wrong, this patch correct this by adopting a more traditional, and correct RCU locking model. This patch is heavily based on an earlier prototype by Cong Wang. Cc: # 4.11.x- Reported-by: Cong Wang Signed-off-by: Cong Wang Signed-off-by: Paul Moore