Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

Fixes: 1dc2f87b33ce ("util: shutdown log in cr_system_userns if error fd is negative")
Cc: Stanislav Kinsburskiy <skinsbursky@virtuozzo.com>
Signed-off-by: Andrew Vagin <avagin@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

While most are handled from the scratch there is a significant
problem with python setup. So I added some preliminaty solution,
probably someohe with good knowledge of how setup.py works
improve it later.
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Reviewed-by: Dmitry Safonov <dsafonov@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

Will need it for uninstall action.
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Reviewed-by: Dmitry Safonov <dsafonov@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

Otherwise error in case of exec error won't be printed
The problem is that when err fd is negative, it's replaced by log fd.
Then err is moved to STDERR (that means, that log_fd is _closed_).
But log facility still consider log fd as valid and tries to use it to print
error message in case of exec failure.
Which is equal to writing to /dev/null, basically.
This patch shutdown log, if err fd was negative, thus forcing criu to output
exec error to STDERR (which was replaced by log fs, btw).
Signed-off-by: Stanislav Kinsburskiy <skinsbursky@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

If several copies of binfmt_misc test are running, we may fail because of duplicate names.
Fix that by giving a custom name for registered /proc/sys/fs/binfmt_misc/xxx.

################## Test zdtm/static/binfmt_misc FAIL at start ##################
Test output: ================================
18:03:33.594: 4: string: :magic_file:M:9:\x73\x51\xff\x4a\xec\x29\xcd\xba\xab\xf2::/bin/interpreter:OCP
18:03:33.594: 4: FAIL: binfmt_misc.c:133: write magic_file (errno = 22 (Invalid argument))
18:03:33.594: 3: ERR: test.c:204: Test exited unexpectedly with code 1
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

When user namespace are stepping in we should wait for their preparation
to complete before start using userns daemon (internally the kernel
checks for uids and if uids are not set -EINVAL will be returned
when usersn calls for sendmsg()).

v2: change stages correctly
v3: Wait prepare_userns() before making any calls to usernsd()

Cc: Cyrill Gorcunov <gorcunov@gmail.com>
Acked-by: Cyrill Gorcunov <gorcunov@gmail.com>
Signed-off-by: Andrew Vagin <avagin@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

Fixes small bug:
__open_proc() macro is called from parse_pid_loginuid as:
>	fd = __open_proc(pid, (ignore_noent) ? ENOENT : 0,
>			O_RDONLY, "loginuid");

So, ier parameter is badly expanded with current version:
>	if (__fd < 0 && (errno != (ignore_noent) ? ENOENT : 0)

Which in result does not hide "No such file" error on feature
test (at least with arm-gcc). Not a big deal, tho.
Reported-by: alex vk <avankemp@gmail.com>
Signed-off-by: Dmitry Safonov <dsafonov@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

Impact: little optimization
Signed-off-by: Dmitry Safonov <dsafonov@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

Signed-off-by: Andrew Vagin <avagin@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

Signed-off-by: Pavel Tikhomirov <ptikhomirov@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

Signed-off-by: Pavel Tikhomirov <ptikhomirov@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

drop_unicast_in_l2_multicast and drop_gratuitous_arp appeared in MS:

commit 12b74dfadb5a ("ipv4: add option to drop unicast encapsulated in
L2 multicast")
commit 97daf3314550 ("ipv4: add option to drop gratuitous ARP packets")
Signed-off-by: Pavel Tikhomirov <ptikhomirov@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

We found that linkat for "unsafe" files doesn't work in userns
if a file uid isn't equal to the currect fsuid. This issue was
fixed by changing fsuid before calling linkat. But in this
case we are not able to createa link if a target directory doesn't
have write premissions.

Starting with the 4.3 kernel, it's possible to create links of
"unsafe files":

f2ca379642d7 ("namei: permit linking with CAP_FOWNER in userns")

So we can try to call linkat() without changing fsuid and make one
more attempt with changing fsuid if the first one failed with EPERM.
Signed-off-by: Andrew Vagin <avagin@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

When not set up it leaves empty making setup of
pycriu python library wrong. Define it as root.
Reported-by: Pavel Emelyanov <xemul@virtuozzo.com>
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

Cc: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrew Vagin <avagin@virtuozzo.com>
Acked-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

Do not fail, if task doesn't have open files. This fixes
zdtm/static/fd test, which is broken now:

(00.004411) Error (cr-dump.c:1312): Dump files (pid: 25) failed with -1
(00.004548) Error (cr-dump.c:1614): Dumping FAILED.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Acked-by: Cyrill Gorcunov <gorcunov@openvz.org>
Acked-by: Andrew Vagin <avagin@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

It might be nil on predump, so dereference only when present.
Signed-off-by: Cyrill Gorcunov <gorcunov@virtuozzo.com>
Acked-by: Andrew Vagin <avagin@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

Signed-off-by: Andrew Vagin <avagin@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

The only one (so far) "post-resume" script needs some pid to join its
Signed-off-by: Stanislav Kinsburskiy <skinsbursky@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

TCP code doesn't need one any longer.
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

Don't mess with sysctl, just try sending queues in greedy mode
shrinking the max_chunk eventually.

This is needed for 2 reasons -- first, to get rig of reading the
max_rshare sysctl and to make libsoccr possible and simple.
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

CID 159475 (#1 of 1): Unchecked return value from library (CHECKED_RETURN)
1. check_return: Calling lseek(img_raw_fd(img), pkt->entry->length, 1)
   without checking return value. This library function may fail and return an error code.
Signed-off-by: Andrew Vagin <avagin@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

The only one (so far) "post-resume" script needs some pid to join its
namespaces. Let it be containers root.

v4:
1) Enviroonment setup has been was moved to run_scripts
2) Environment variable NS_PID was renamed to CRTOOLS_INIT_PID
Signed-off-by: Stanislav Kinsburskiy <skinsbursky@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

This hook is called at the very end, when everything is restored and processes
were resumed.
Can be used for some actions, which require operation container, like
restarting of systemd autofs services.

v3:
Call "post-resume" scripts before detach option check (to make sure, that
restored process(es) are still alive

v2:
Remove code chunk, escaped from debugging
Signed-off-by: Stanislav Kinsburskiy <skinsbursky@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

Here's the first once-a-month release with whatever is there in
the master branch. We have quite a few new features, but a lot of
bugifxes :)
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

Signed-off-by: Andrew Vagin <avagin@virtuozzo.com>
Acked-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

Currently criu can't handle a processes with more than 1024
file descriptors.

In this patch, criu dumps file descriptors for a few iterations.

Fixes https://github.com/xemul/criu/issues/145

v2: don't move the collect stage
Signed-off-by: Andrew Vagin <avagin@virtuozzo.com>
Reviewed-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

We are going to remove the PARASITE_MAX_FDS limit and
this patch is a preparation for this.
Signed-off-by: Andrew Vagin <avagin@virtuozzo.com>
Acked-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

Signed-off-by: Deng Guangxing <dengguangxing@huawei.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

These devices are injected into every network namespace when the ip_gre
module is loaded. They cannot be configured as usable tunnels, so let's
just ignore them.
Signed-off-by: Tycho Andersen <tycho.andersen@canonical.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Reviewed-by: Dmitry Safonov <dsafonov@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

In case of error inside collect_cgroups we need

- @ctls list has been spliced into @cs->ctls so we cant just free it,
   but rather use put_ctls on @cs->ctls

- delete @cs->l from global @cg_sets list before free it
Signed-off-by: Cyrill Gorcunov <gorcunov@virtuozzo.com>
Reviewed-by: Andrew Vagin <avagin@openvz.org>
Acked-by: Tycho Andersen <tycho.andersen@canonical.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

If add_cgroup_properties failed we have @ncd sitting
in @match->children or @current_controller->heads
list, and then we free @ncd entry leaving those
lists carrying dead pointers.

Add proper rollback action.
Signed-off-by: Cyrill Gorcunov <gorcunov@virtuozzo.com>
Acked-by: Tycho Andersen <tycho.andersen@canonical.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

Since we're freeing list entries don't forget to
initialize list then, otherwise it gets out with
free entries and may hit use-after-free bug.
Signed-off-by: Cyrill Gorcunov <gorcunov@virtuozzo.com>
Acked-by: Tycho Andersen <tycho.andersen@canonical.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

We will still able to distinguish if it's garbage in ax register, but
this is definitely prettier:
(00.036900) 687 was trapped
(00.036903) 687 is going to execute the syscall 158
(00.036928) 687 was trapped
(00.036931) 687 is going to execute the syscall 173

Than that:
(00.024403) 687 was trapped
(00.024407) 687 is going to execute the syscall 9e
(00.024419) 687 was trapped
(00.024421) 687 is going to execute the syscall ad
Signed-off-by: Dmitry Safonov <dsafonov@virtuozzo.com>
Acked-by: Cyrill Gorcunov <gorcunov@openvz.org>
Acked-by: Andrew Vagin <avagin@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

Previously when we've been on early stage we
tried to workaround lack of prlimit libc call
simply providing own implementation,  but it
cause problems on some libc configurations so
simply use rlimit64 and __NR_prlimit64 syscall
directly.

The kernel must support __NR_prlimit64 syscall
and provide rlimit64 structure as well.
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

The option was in use when we didn't have the /proc/pid/ns links
and thus ns ids. Later we used it at restore time to tell which
namespaces to restore the pstree into. Right now ids are always
generated and this option has no effect (and meaning).
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

We've been writing this file since 0.4, hopefully nobody's
using such an oldie :)

v2: Don't BUG_ON, but return error, as this can be triggered
by corrupting images.
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

This helper is used in both PIE and non-PIE code.
For the second we can use normal pr_perror.
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>