Currently the kernel doesn't report sequence numbers for contracks
and tries to restore them from first packets.
When we are restoring a tcp connection, we send a window probe and
set seq - 1 in it to get ack immediatly.

        /* Use a previous sequence.  This should cause the other
         * end to send an ack.  Don't queue or clone SKB, just
         * send it.
         */
        tcp_init_nondata_skb(skb, tp->snd_una - !urgent, TCPHDR_ACK);

But conntrack doesn't like this, because then we get ack, which is greater than seq.
It looks like we try to ack data which we haven't received yet.

[  735.528073] td_maxwin == 0
               seq=1081132048 ack=2965916432+(0) sack=2965916432+(0) win=342 end=1081132048
               tcp_in_window: sender end=0 maxend=0 maxwin=0 scale=0 receiver end=0 maxend=0 maxwin=0 scale=0
[  735.533409] log_invalid:
               seq=2965916431 ack=1081132049+(0) sack=1081132049+(0) win=342 end=2965916431
               tcp_in_window: sender end=2965916431 maxend=2965916773 maxwin=342 scale=0 receiver end=1081132048 maxend=1081132390 maxwin=342 scale=0
[  735.537651] nf_ct_tcp: ACK is over the upper bound (ACKed data not seen yet)

The kernel sets IP_CT_TCP_FLAG_BE_LIBERAL for new conntracks,
if we are in the middle of a connection.
Signed-off-by: Andrew Vagin <avagin@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

We request all contracks via netlink and save netlink messages which
describe them in an image file, then we send these netlink messages back on restore.

https://github.com/xemul/criu/issues/54Signed-off-by: Andrew Vagin <avagin@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

rst_mem_alloc() returns alligned slices. When we want to
allocate an array element, we don't expect to get a hole between
a new element and an old one.
Signed-off-by: Andrew Vagin <avagin@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

Stas found that if we don't align a pointer,
futex and atomic operations can fail.

v2: don't hard-code the size of void *
v3: add a function to allocate memory without gaps with
    a privious slice. It's used to allocate arrays.
v4: don't change rst_mem_cpos

Cc: Stanislav Kinsburskiy <skinsbursky@virtuozzo.com>
Signed-off-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>

As of 4.4,
https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=f8e529ed9
c/r of seccomp filters is supported, so let's check for it in --ms.
Signed-off-by: Tycho Andersen <tycho.andersen@canonical.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

As of 4.3,
https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=13c4a9011
PTRACE_O_SUSPEND_SECCOMP is in the mainline kernel, so we should check for
it in --ms.
Signed-off-by: Tycho Andersen <tycho.andersen@canonical.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

CID 157804 (#1 of 1): Missing varargs init or cleanup (VARARGS)
16. missing_va_end: va_end was not called for tmp.

v2: typo fix

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

CID 157800 (#1 of 1): Missing break in switch (MISSING_BREAK)
unterminated_case: The case for value 1071 is not terminated by a 'break' statement.
Signed-off-by: Andrew Vagin <avagin@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

CID 157801 (#1 of 1): Argument cannot be negative (NEGATIVE_RETURNS)
19. negative_returns: fd is passed to a parameter that cannot be negative.
Signed-off-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>

crtools binary is linked with the C library and could rely on all the
services this library is providing, including system calls.

Thus it doesn't need to be linked with the builtin system calls code
made for the parasite/restorer binaries.

This patch does:
 - remove the inclusion of syscall.h
 - replace all call to sys_<syscall>() by C library <syscall>()
 - replace unwrapped system calls by syscall(SYS_<syscall>,...)
 - fix the generated compiler's issues.

There should not be any functional changes. The only 'code' changes is
appearing in locks.h when futex is called through the C library, the
errno value is fetched from errno variable instead of the return
value.
Signed-off-by: Laurent Dufour <ldufour@linux.vnet.ibm.com>
Reviewed-by: Christopher Covington <cov@codeaurora.org>
Reviewed-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

The CRIU internal define of CLONE_SUBNS should not be put in
syscall-types.h since this define is not part of a system call.

This move is required to prepare the removal of syscall.h from the
component of crtools binary.
Signed-off-by: Laurent Dufour <ldufour@linux.vnet.ibm.com>
Reviewed-by: Christopher Covington <cov@codeaurora.org>
Reviewed-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

The file include/prctl.h should define the struct prctl_mm_map only if
it is not already defined in the system include file linux/prctl.h.

The definition should be part of the '#ifndef PR_SET_MM_MAP' block
since this structure is not defined in that case.
Signed-off-by: Laurent Dufour <ldufour@linux.vnet.ibm.com>
Reviewed-by: Christopher Covington <cov@codeaurora.org>
Reviewed-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

It is missged in first place and may cause
problem on exiting via alarm hanling.
Reported-by: Igor Sukhih <igor@virtuozzo.com>
Signed-off-by: Cyrill Gorcunov <gorcunov@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

Signed-off-by: Tycho Andersen <tycho.andersen@canonical.com>
Acked-by: Andrew Vagin <avagin@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

All processes should exit if an user presses ctrl-c.
Signed-off-by: Andrew Vagin <avagin@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

Skip it till remap-links are supported.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Acked-by: Andrew Vagin <avagin@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

Test checks that a deleted file content in a deleted parent directory
restores right. Initially directory is created in a separate mount,
to check the ghost file is created in right mount and link() in
rfi_remap() is working.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

This patch makes ghost files to restore with the right path.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

Since a deleted file may belong to a deleted directory
(which, in turn, may belong to another deleted, etc),
it's necessary to recreate all missing dentries in the path.

Doing this in mkreg_ghost() is not a good idea, because
another ghost file may need some of the dentries in the path,
and this requires to do refcouting of recreated directories.

To avoid this, we restore a ghost file in the first existing
parent directory of the patch. This guarantees, the ghost file
will be in the same mount with target file (rfi_remap() needs
that, because it uses link()).
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

The issue here is that ARCH is not overwritten if the caller Makefile
already define it, and the upper Makefile is not overwriting the ARCH
define to "arm", when it is "armv7l" for instance, so the SRCARCH is
set to "armv7l" which is wrong.

With this patch, SRCARCH is using the inherinted define if called from
the top level Makefile.

This is required to build test on armv7l architecture (qemu).
Signed-off-by: Laurent Dufour <ldufour@linux.vnet.ibm.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

Sometimes I get:

sudo ./zdtm.py run -t zdtm/live/static/cgroup03
umount: /home/tycho/packages/criu/test/zdtm.qPwsoO: target is busy
        (In some cases useful info about processes that
         use the device is found by lsof(8) or fuser(1).)
Traceback (most recent call last):
  File "./zdtm.py", line 1401, in <module>
    tst.available()
  File "./zdtm.py", line 435, in available
    subprocess.check_call(["flock", "zdtm_mount_cgroups", "./zdtm_mount_cgroups"])
  File "/usr/lib/python2.7/subprocess.py", line 540, in check_call
    raise CalledProcessError(retcode, cmd)
subprocess.CalledProcessError: Command '['flock', 'zdtm_mount_cgroups', './zdtm_mount_cgroups']' returned non-zero exit status 1

let's use a lazy unmount so we avoid this.
Signed-off-by: Tycho Andersen <tycho.andersen@canonical.com>
Acked-by: Andrew Vagin <avagin@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

cgroup.sane_behavior is read only (and in fact, the value is always 0), so
we don't need to c/r this value.
Reported-by: Andrey Vagin <avagin@openvz.org>
Signed-off-by: Tycho Andersen <tycho.andersen@canonical.com>
Acked-by: Andrew Vagin <avagin@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

We have two separate cases 1) mount is external - has mi->external set,
2) mount is internal - mi->external == 0. For those two cases we need
separate decesions on making moutn private.

In both cases we need private remount if:
a)mount does not have master_id and shared_id
or
b)if mount has different shared_id from bind source.

But do not private remount if has master_id or it will be lost. As after
setting private the mount is alone in its shared group so setting master
will silently fail as kernel can not find any for our mount.
(see do_make_slave)

https://jira.sw.ru/browse/PSBM-42829

v2: rework patches 2/4 and 4/4 together and no need then in 3/4
Signed-off-by: Pavel Tikhomirov <ptikhomirov@virtuozzo.com>
Acked-by: Andrew Vagin <avagin@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

AutoFS will need to create write pipe end file descriptor, if it was closed.
Thus, pipe_info structure have to be exported.
Signed-off-by: Stanislav Kinsburskiy <skinsbursky@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

This is a precursor patch for AutoFS mount restore.
To restore AutoFS indirect mount points, we have to create mount point dentries.
This can do only the process with pgrp, configured to mount. Process, which
does AutoFS mount call (init) at that moment is not.
Restoring actual pgrp before sid breaks sid restoring.
Thus, restoring of sid is required to make process group leader.
Signed-off-by: Stanislav Kinsburskiy <skinsbursky@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

This cleanup makes create_ghost() smaller.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

This cleanup will be used in next patch.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

In the most cases, file descriptor is not need in this function.
It's need in the case of creating a regular file only. So we can
use chown() instead of fchown(), and operate with a path directly.

The patch moves copying content of a regular file to new function
mkreg_ghost(). This will be used in next patches. Also, parameter
gf will be used.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

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

Signed-off-by: Tycho Andersen <tycho.andersen@canonical.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

Signed-off-by: Tycho Andersen <tycho.andersen@canonical.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

Although we don't want to restore the values on these files, we definitely
do want to restore the permissions, as certain container engines (lxc) make
use of this.
Signed-off-by: Tycho Andersen <tycho.andersen@canonical.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

Signed-off-by: Tycho Andersen <tycho.andersen@canonical.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

These are present in every cgroup, so let's dump them.
Signed-off-by: Tycho Andersen <tycho.andersen@canonical.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

Signed-off-by: Tycho Andersen <tycho.andersen@canonical.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

The root mount is always external and master_id is external too.
It's required to validate mounts. We already do this on restore,
but we need to do this on dump too. Otherwise we will get the error:

Mount %d %s (master_id: %d shared_id: %d) has unreachable sharing. Try --enable-external-masters.

https://jira.sw.ru/browse/PSBM-43260Signed-off-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>

If for some reason ptrace_poke_area return error
we might left dumpee with memfd descriptor opened.

Later in code we remove out injected blob making
dumpee to look untouched but descriptor will hang there.

lsof from container output:

 | systemd-u   48       root    6u      REG                0,4        0  53855 /memfd:CRIUMFD (deleted)

Thus lets close it immediately.

https://jira.sw.ru/browse/PSBM-43199Signed-off-by: Cyrill Gorcunov <gorcunov@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>