There are still a few places where profile replacement fails to update
and a stale profile is used for mediation. Fix this by moving to
accessing the current label through a critical section that will
always ensure mediation is using the current label regardless of
whether the tasks cred has been updated or not.
Signed-off-by: NJohn Johansen <john.johansen@canonical.com>

There is no reason to have the small stubs that don't use domain
private functions in domain.c, instead move them to lsm.c and make
them static.
Signed-off-by: NJohn Johansen <john.johansen@canonical.com>

The ns name being displayed should go through an ns view lookup.
Signed-off-by: NJohn Johansen <john.johansen@canonical.com>

The data being queried isn't always the current profile and a lookup
relative to the current profile should be done.
Signed-off-by: NJohn Johansen <john.johansen@canonical.com>

The namespace being passed into the replace/remove profiles fns() is
not the view, but the namespace specified by the inode from the
file hook (if present) or the loading tasks ns, if accessing the
top level virtualized load/replace file interface.
Signed-off-by: NJohn Johansen <john.johansen@canonical.com>

Currently lookups are restricted to a single ns component in the
path. However when namespaces are allowed to have separate views, and
scopes this will not be sufficient, as it will be possible to have
a multiple component ns path in scope.

Add some ns lookup fns() to allow this and use them.
Signed-off-by: NJohn Johansen <john.johansen@canonical.com>

Signed-off-by: NJohn Johansen <john.johansen@canonical.com>

Signed-off-by: NJohn Johansen <john.johansen@canonical.com>

Allow userspace to query a profile about permissions, through the
transaction interface that is already used to allow userspace to
query about key,value data.
Signed-off-by: NJohn Johansen <john.johansen@canonical.com>

Signed-off-by: NJohn Johansen <john.johansen@canonical.com>

Signed-off-by: NJohn Johansen <john.johansen@canonical.com>

Signed-off-by: NJohn Johansen <john.johansen@canonical.com>

The simple_transaction interface is slow. It requires 4 syscalls
(open, write, read, close) per query and shares a single lock for each
queries.

So replace its use with a compatible in multi_transaction interface.
It allows for a faster 2 syscall pattern per query. After an initial
open, an arbitrary number of writes and reads can be issued. Each
write will reset the query with new data that can be read. Reads do
not clear the data, and can be issued multiple times, and used with
seek, until a new write is performed which will reset the data
available and the seek position.

Note: this keeps the single lock design, if needed moving to a per
file lock will have to come later.
Signed-off-by: NJohn Johansen <john.johansen@canonical.com>

gsettings mediation needs to be able to determine if apparmor supports
label data queries. A label data query can be done to test for support
but its failure is indistinguishable from other failures, making it an
unreliable indicator.

Fix by making support of label data queries available as a flag in the
apparmorfs features dir tree.
Signed-off-by: NJohn Johansen <john.johansen@canonical.com>

When setting up namespaces for containers its easier for them to use
an fs interface to create the namespace for the containers
policy. Allow mkdir/rmdir under the policy/namespaces/ dir to be used
to create and remove namespaces.

BugLink: http://bugs.launchpad.net/bugs/1611078Signed-off-by: NJohn Johansen <john.johansen@canonical.com>

Add a policy revision file to find the current revision of a ns's policy.
There is a revision file per ns, as well as a virtualized global revision
file in the base apparmor fs directory. The global revision file when
opened will provide the revision of the opening task namespace.

The revision file can be waited on via select/poll to detect apparmor
policy changes from the last read revision of the opened file. This
means that the revision file must be read after the select/poll other
wise update data will remain ready for reading.
Signed-off-by: NJohn Johansen <john.johansen@canonical.com>

Signed-off-by: NJohn Johansen <john.johansen@canonical.com>

Signed-off-by: NJohn Johansen <john.johansen@canonical.com>

Signed-off-by: NJohn Johansen <john.johansen@canonical.com>

Virtualize the apparmor policy/ directory so that the current
namespace affects what part of policy is seen. To do this convert to
using apparmorfs for policy namespace files and setup a magic symlink
in the securityfs apparmor dir to access those files.
Signed-off-by: NJohn Johansen <john.johansen@canonical.com>
Reviewed-by: NSeth Arnold <seth.arnold@canonical.com>
Reviewed-by: NKees Cook <keescook@chromium.org>

Signed-off-by: NJohn Johansen <john.johansen@canonical.com>
Reviewed-by: NSeth Arnold <seth.arnold@canonical.com>
Reviewed-by: NKees Cook <keescook@chromium.org>

prefixes are used for fns/data that are not static to apparmorfs.c
with the prefixes being
  aafs   - special magic apparmorfs for policy namespace data
  aa_sfs - for fns/data that go into securityfs
  aa_fs  - for fns/data that may be used in the either of aafs or
           securityfs
Signed-off-by: NJohn Johansen <john.johansen@canonical.com>
Reviewed-by: NSeth Arnold <seth.arnold@canonical.com>
Reviewed-by: NKees Cook <keescook@chromium.org>

AppArmor policy needs to be able to be resolved based on the policy
namespace a task is confined by. Add a base apparmorfs filesystem that
(like nsfs) will exist as a kern mount and be accessed via jump_link
through a securityfs file.

Setup the base apparmorfs fns and data, but don't use it yet.
Signed-off-by: NJohn Johansen <john.johansen@canonical.com>
Reviewed-by: NSeth Arnold <seth.arnold@canonical.com>
Reviewed-by: NKees Cook <keescook@chromium.org>

Signed-off-by: NJohn Johansen <john.johansen@canonical.com>
Reviewed-by: NSeth Arnold <seth.arnold@canonical.com>
Reviewed-by: NKees Cook <keescook@chromium.org>

Signed-off-by: NJohn Johansen <john.johansen@canonical.com>
Reviewed-by: NSeth Arnold <seth.arnold@canonical.com>
Reviewed-by: NKees Cook <keescook@chromium.org>

The loaddata sets cover more than just a single profile and should
be tracked at the ns level. Move the load data files under the namespace
and reference the files from the profiles via a symlink.
Signed-off-by: NJohn Johansen <john.johansen@canonical.com>
Reviewed-by: NSeth Arnold <seth.arnold@canonical.com>
Reviewed-by: NKees Cook <keescook@chromium.org>

Dynamically allocating buffers is problematic and is an extra layer
that is a potntial point of failure and can slow down mediation.
Change path lookup to use the preallocated per cpu buffers.
Signed-off-by: NJohn Johansen <john.johansen@canonical.com>

Signed-off-by: NJohn Johansen <john.johansen@canonical.com>

Signed-off-by: NJohn Johansen <john.johansen@canonical.com>

Signed-off-by: NJohn Johansen <john.johansen@canonical.com>

When using a strictly POSIX-compliant shell, "-n #define ..." gets
written into the file.  Use "printf '%s'" to avoid this.
Signed-off-by: NThomas Schneider <qsx@qsx.re>
Signed-off-by: NJohn Johansen <john.johansen@canonical.com>

We can either return PTR_ERR(NULL) or a PTR_ERR(a valid pointer) here.
Returning NULL is probably not good, but since this happens at boot
then we are probably already toasted if we were to hit this bug in real
life.  In other words, it seems like a very low severity bug to me.
Signed-off-by: NDan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: NJohn Johansen <john.johansen@canonical.com>

Two single characters (line breaks) should be put into a sequence.
Thus use the corresponding function "seq_putc".

This issue was detected by using the Coccinelle software.
Signed-off-by: NMarkus Elfring <elfring@users.sourceforge.net>
Signed-off-by: NJohn Johansen <john.johansen@canonical.com>

A bit of data was put into a sequence by two separate function calls.
Print the same data by a single function call instead.
Signed-off-by: NMarkus Elfring <elfring@users.sourceforge.net>
Signed-off-by: NJohn Johansen <john.johansen@canonical.com>

CURRENT_TIME macro is not y2038 safe on 32 bit systems.

The patch replaces all the uses of CURRENT_TIME by current_time().

This is also in preparation for the patch that transitions vfs
timestamps to use 64 bit time and hence make them y2038 safe.
current_time() is also planned to be transitioned to y2038 safe behavior
along with this change.

CURRENT_TIME macro will be deleted before merging the aforementioned
change.

Link: http://lkml.kernel.org/r/1491613030-11599-11-git-send-email-deepa.kernel@gmail.comSigned-off-by: NDeepa Dinamani <deepa.kernel@gmail.com>
Acked-by: NJohn Johansen <john.johansen@canonical.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Patch series "kvmalloc", v5.

There are many open coded kmalloc with vmalloc fallback instances in the
tree.  Most of them are not careful enough or simply do not care about
the underlying semantic of the kmalloc/page allocator which means that
a) some vmalloc fallbacks are basically unreachable because the kmalloc
part will keep retrying until it succeeds b) the page allocator can
invoke a really disruptive steps like the OOM killer to move forward
which doesn't sound appropriate when we consider that the vmalloc
fallback is available.

As it can be seen implementing kvmalloc requires quite an intimate
knowledge if the page allocator and the memory reclaim internals which
strongly suggests that a helper should be implemented in the memory
subsystem proper.

Most callers, I could find, have been converted to use the helper
instead.  This is patch 6.  There are some more relying on __GFP_REPEAT
in the networking stack which I have converted as well and Eric Dumazet
was not opposed [2] to convert them as well.

[1] http://lkml.kernel.org/r/20170130094940.13546-1-mhocko@kernel.org
[2] http://lkml.kernel.org/r/1485273626.16328.301.camel@edumazet-glaptop3.roam.corp.google.com

This patch (of 9):

Using kmalloc with the vmalloc fallback for larger allocations is a
common pattern in the kernel code.  Yet we do not have any common helper
for that and so users have invented their own helpers.  Some of them are
really creative when doing so.  Let's just add kv[mz]alloc and make sure
it is implemented properly.  This implementation makes sure to not make
a large memory pressure for > PAGE_SZE requests (__GFP_NORETRY) and also
to not warn about allocation failures.  This also rules out the OOM
killer as the vmalloc is a more approapriate fallback than a disruptive
user visible action.

This patch also changes some existing users and removes helpers which
are specific for them.  In some cases this is not possible (e.g.
ext4_kvmalloc, libcfs_kvzalloc) because those seems to be broken and
require GFP_NO{FS,IO} context which is not vmalloc compatible in general
(note that the page table allocation is GFP_KERNEL).  Those need to be
fixed separately.

While we are at it, document that __vmalloc{_node} about unsupported gfp
mask because there seems to be a lot of confusion out there.
kvmalloc_node will warn about GFP_KERNEL incompatible (which are not
superset) flags to catch new abusers.  Existing ones would have to die
slowly.

[sfr@canb.auug.org.au: f2fs fixup]
  Link: http://lkml.kernel.org/r/20170320163735.332e64b7@canb.auug.org.au
Link: http://lkml.kernel.org/r/20170306103032.2540-2-mhocko@kernel.orgSigned-off-by: NMichal Hocko <mhocko@suse.com>
Signed-off-by: NStephen Rothwell <sfr@canb.auug.org.au>
Reviewed-by: Andreas Dilger <adilger@dilger.ca>	[ext4 part]
Acked-by: NVlastimil Babka <vbabka@suse.cz>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: David Miller <davem@davemloft.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

The path_max parameter determines the max size of buffers allocated
but it should  not be setable at run time. If can be used to cause an
oops

root@ubuntu:~# echo 16777216 > /sys/module/apparmor/parameters/path_max
root@ubuntu:~# cat /sys/module/apparmor/parameters/path_max
Killed

[  122.141911] BUG: unable to handle kernel paging request at ffff880080945fff
[  122.143497] IP: [<ffffffff81228844>] d_absolute_path+0x44/0xa0
[  122.144742] PGD 220c067 PUD 0
[  122.145453] Oops: 0002 [#1] SMP
[  122.146204] Modules linked in: vmw_vsock_vmci_transport vsock ppdev vmw_balloon snd_ens1371 btusb snd_ac97_codec gameport snd_rawmidi btrtl snd_seq_device ac97_bus btbcm btintel snd_pcm input_leds bluetooth snd_timer snd joydev soundcore serio_raw coretemp shpchp nfit parport_pc i2c_piix4 8250_fintek vmw_vmci parport mac_hid ib_iser rdma_cm iw_cm ib_cm ib_sa ib_mad ib_core ib_addr iscsi_tcp libiscsi_tcp libiscsi scsi_transport_iscsi autofs4 btrfs raid10 raid456 async_raid6_recov async_memcpy async_pq async_xor async_tx xor raid6_pq libcrc32c raid1 raid0 multipath linear hid_generic usbhid hid crct10dif_pclmul crc32_pclmul ghash_clmulni_intel aesni_intel aes_x86_64 lrw gf128mul glue_helper ablk_helper cryptd vmwgfx psmouse mptspi ttm mptscsih drm_kms_helper mptbase syscopyarea scsi_transport_spi sysfillrect
[  122.163365]  ahci sysimgblt e1000 fb_sys_fops libahci drm pata_acpi fjes
[  122.164747] CPU: 3 PID: 1501 Comm: bash Not tainted 4.4.0-59-generic #80-Ubuntu
[  122.166250] Hardware name: VMware, Inc. VMware Virtual Platform/440BX Desktop Reference Platform, BIOS 6.00 07/02/2015
[  122.168611] task: ffff88003496aa00 ti: ffff880076474000 task.ti: ffff880076474000
[  122.170018] RIP: 0010:[<ffffffff81228844>]  [<ffffffff81228844>] d_absolute_path+0x44/0xa0
[  122.171525] RSP: 0018:ffff880076477b90  EFLAGS: 00010206
[  122.172462] RAX: ffff880080945fff RBX: 0000000000000000 RCX: 0000000001000000
[  122.173709] RDX: 0000000000ffffff RSI: ffff880080946000 RDI: ffff8800348a1010
[  122.174978] RBP: ffff880076477bb8 R08: ffff880076477c80 R09: 0000000000000000
[  122.176227] R10: 00007ffffffff000 R11: ffff88007f946000 R12: ffff88007f946000
[  122.177496] R13: ffff880076477c80 R14: ffff8800348a1010 R15: ffff8800348a2400
[  122.178745] FS:  00007fd459eb4700(0000) GS:ffff88007b6c0000(0000) knlGS:0000000000000000
[  122.180176] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  122.181186] CR2: ffff880080945fff CR3: 0000000073422000 CR4: 00000000001406e0
[  122.182469] Stack:
[  122.182843]  00ffffff00000001 ffff880080946000 0000000000000000 0000000000000000
[  122.184409]  00000000570f789c ffff880076477c30 ffffffff81385671 ffff88007a2e7a58
[  122.185810]  0000000000000000 ffff880076477c88 01000000008a1000 0000000000000000
[  122.187231] Call Trace:
[  122.187680]  [<ffffffff81385671>] aa_path_name+0x81/0x370
[  122.188637]  [<ffffffff813875dd>] profile_transition+0xbd/0xb80
[  122.190181]  [<ffffffff811af9bc>] ? zone_statistics+0x7c/0xa0
[  122.191674]  [<ffffffff81389b20>] apparmor_bprm_set_creds+0x9b0/0xac0
[  122.193288]  [<ffffffff812e1971>] ? ext4_xattr_get+0x81/0x220
[  122.194793]  [<ffffffff812e800c>] ? ext4_xattr_security_get+0x1c/0x30
[  122.196392]  [<ffffffff813449b9>] ? get_vfs_caps_from_disk+0x69/0x110
[  122.198004]  [<ffffffff81232d4f>] ? mnt_may_suid+0x3f/0x50
[  122.199737]  [<ffffffff81344b03>] ? cap_bprm_set_creds+0xa3/0x600
[  122.201377]  [<ffffffff81346e53>] security_bprm_set_creds+0x33/0x50
[  122.203024]  [<ffffffff81214ce5>] prepare_binprm+0x85/0x190
[  122.204515]  [<ffffffff81216545>] do_execveat_common.isra.33+0x485/0x710
[  122.206200]  [<ffffffff81216a6a>] SyS_execve+0x3a/0x50
[  122.207615]  [<ffffffff81838795>] stub_execve+0x5/0x5
[  122.208978]  [<ffffffff818384f2>] ? entry_SYSCALL_64_fastpath+0x16/0x71
[  122.210615] Code: f8 31 c0 48 63 c2 83 ea 01 48 c7 45 e8 00 00 00 00 48 01 c6 85 d2 48 c7 45 f0 00 00 00 00 48 89 75 e0 89 55 dc 78 0c 48 8d 46 ff <c6> 46 ff 00 48 89 45 e0 48 8d 55 e0 48 8d 4d dc 48 8d 75 e8 e8
[  122.217320] RIP  [<ffffffff81228844>] d_absolute_path+0x44/0xa0
[  122.218860]  RSP <ffff880076477b90>
[  122.219919] CR2: ffff880080945fff
[  122.220936] ---[ end trace 506cdbd85eb6c55e ]---
Reported-by: NTetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Signed-off-by: NJohn Johansen <john.johansen@canonical.com>
Signed-off-by: NJames Morris <james.l.morris@oracle.com>

Boot parameters are written before apparmor is ready to answer whether
the user is policy_view_capable(). Setting the parameters at boot results
in an oops and failure to boot. Setting the parameters at boot is
obviously allowed so skip the permission check when apparmor is not
initialized.

While we are at it move the more complicated check to last.
Signed-off-by: NJohn Johansen <john.johansen@canonical.com>
Signed-off-by: NJames Morris <james.l.morris@oracle.com>

Once the loop on lines 836-853 is complete and exits normally, ent is a
pointer to the dummy list head value.  The derefernces accessible from eg
the goto fail on line 860 or the various goto fail_lock's afterwards thus
seem incorrect.
Reported-by: NJulia Lawall <julia.lawall@lip6.fr>
Signed-off-by: NJohn Johansen <john.johansen@canonical.com>
Signed-off-by: NJames Morris <james.l.morris@oracle.com>

When building the kernel with clang, the compiler fails to build
security/apparmor/crypto.c with the following error:

    security/apparmor/crypto.c:36:8: error: fields must have a constant
    size: 'variable length array in structure' extension will never be
    supported
                    char ctx[crypto_shash_descsize(apparmor_tfm)];
                         ^

Since commit a0a77af1 ("crypto: LLVMLinux: Add macro to remove use
of VLAIS in crypto code"), include/crypto/hash.h defines
SHASH_DESC_ON_STACK to work around this issue. Use it in aa_calc_hash()
and aa_calc_profile_hash().
Signed-off-by: NNicolas Iooss <nicolas.iooss_linux@m4x.org>
Signed-off-by: NJohn Johansen <john.johansen@canonical.com>
Signed-off-by: NJames Morris <james.l.morris@oracle.com>