with newer versions of protobuf, I get:
./crit
Traceback (most recent call last):
File "./crit", line 7, in <module>
import pycriu
File "/home/ubuntu/criu/crit/pycriu/__init__.py", line 2, in <module>
import images
File "/home/ubuntu/criu/crit/pycriu/images/__init__.py", line 2, in <module>
from images import *
File "/home/ubuntu/criu/crit/pycriu/images/images.py", line 52, in <module>
from pb import *
File "/home/ubuntu/criu/crit/pycriu/images/pb.py", line 35, in <module>
from packet_sock_pb2 import *
File "/home/ubuntu/criu/crit/pycriu/images/packet_sock_pb2.py", line 18, in <module>
import sk_opts_pb2 as sk__opts__pb2
File "/home/ubuntu/criu/crit/pycriu/images/sk_opts_pb2.py", line 23, in <module>
serialized_pb=_b('\n\rsk-opts.proto\"\xe2\x02\n\rsk_opts_entry\x12\x11\n\tso_sndbuf\x18\x01 \x02(\r\x12\x11\n\tso_rcvbuf\x18\x02 \x02(\r\x12\x16\n\x0eso_snd_tmo_sec\x18\x03 \x02(\x04\x12\x17\n\x0fso_snd_tmo_usec\x18\x04 \x02(\x04\x12\x16\n\x0eso_rcv_tmo_sec\x18\x05 \x02(\x04\x12\x17\n\x0fso_rcv_tmo_usec\x18\x06 \x02(\x04\x12\x11\n\treuseaddr\x18\x07 \x01(\x08\x12\x13\n\x0bso_priority\x18\x08 \x01(\r\x12\x13\n\x0bso_rcvlowat\x18\t \x01(\r\x12\x0f\n\x07so_mark\x18\n \x01(\r\x12\x13\n\x0bso_passcred\x18\x0b \x01(\x08\x12\x12\n\nso_passsec\x18\x0c \x01(\x08\x12\x14\n\x0cso_dontroute\x18\r \x01(\x08\x12\x13\n\x0bso_no_check\x18\x0e \x01(\x08\x12\x14\n\x0cso_bound_dev\x18\x0f \x01(\t\x12\x11\n\tso_filter\x18\x10 \x03(\x06*6\n\x0bsk_shutdown\x12\x08\n\x04NONE\x10\x00\x12\x08\n\x04READ\x10\x01\x12\t\n\x05WRITE\x10\x02\x12\x08\n\x04\x42OTH\x10\x03')
File "/usr/lib/python2.7/dist-packages/google/protobuf/descriptor.py", line 827, in __new__
return _message.default_pool.AddSerializedFile(serialized_pb)
TypeError: Couldn't build proto file into descriptor pool!
Invalid proto descriptor for file "sk-opts.proto":
NONE: "NONE" is already defined in file "rpc.proto".
NONE: Note that enum values use C++ scoping rules, meaning that enum values are siblings of their type, not children of it. Therefore, "NONE" must be unique within the global scope, not just within "sk_shutdown".
this fixes that.
Signed-off-by: Tycho Andersen <tycho.andersen@canonical.com>
CC: Laurent Dufour <ldufour@linux.vnet.ibm.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
avoids:
[libprotobuf WARNING google/protobuf/compiler/parser.cc:547] No syntax specified for the proto file: autofs.proto. Please use 'syntax = "proto2";' or 'syntax = "proto3";' to specify a syntax version. (Defaulted to proto2 syntax.)
Signed-off-by: Tycho Andersen <tycho.andersen@canonical.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
The Power 8 introduces the transactional memory (TM) operations (see
Power ISA 3.0 for details).
The support for the transactional memory operation during the
checkpoint and restart requires extended ptrace API provided by the
kernel 4.8.
When checkpointing a thread while a transactional memory operation is
in progress, the TM checkpointed state is checkpointed through the new
ptrace API. If these new APIs are not available, the checkpoint is
aborted and an explicit error is reported.
At restart time, the TM state is pushed on the stack frame to be
reloaded by the kernel when reading the stack frame.
Only suspended TM operation could be checkpointed since active one
will be aborted once a system call is made. Suspended operation will
be aborted as well, and the checkpointed thread is expected to handle
the TM failure as usual (retrying is a good option).
Signed-off-by: Laurent Dufour <ldufour@linux.vnet.ibm.com>
Reviewed-by: Dmitry Safonov <dsafonov@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
No more use C++ comment style
Signed-off-by: Laurent Dufour <ldufour@linux.vnet.ibm.com>
Reviewed-by: Dmitry Safonov <dsafonov@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
Currently when we create peers we don't consider
if they were owned by someone else. Lest carry
uid/gid into image and restore then.
https://github.com/xemul/criu/issues/198
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Reviewed-by: Dmitry Safonov <dsafonov@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
When running criu in swrk mode the client typically wants to know
the reason of failure. Right now criu reports back NOTHING but the
fact that dump/restore/etc fails. We've tried to address this by
introducing the cr-errno engine, but it doesn't seem to be informative
enough and is hard to maintain -- adding new errno-s is boring :(
I propose to report back the first message with ERROR level upon
failrure as __typically__ the very first error message indicates
that proceeding is impossible and criu rolls back (generating more
error messages, so it's crucial to know the very first one).
If we ever meet the situation that the first pr_err/pr_perror doesn't
cause criu to exit, this printing should be fixed to be pr_warn.
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
Acked-by: Andrei Vagin <avagin@virtuozzo.com>
These are only three left in ipc_kern_table which we haven't
checkpointed yet, I'm not sure if somebody really uses them
except criu, but to be consistent, beter not to change them
while c/r.
v3: do one sysctl_op for all xxx_next_id(as sysctl_op is quiet slow)
v4: do only one sysctl_op in ipc_sysctl_req
v5: do msg*_default only if have /proc/sys/fs/mqueue same as other
ones from fs/mqueue
Signed-off-by: Pavel Tikhomirov <ptikhomirov@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
We found that sometimes a restored tcp socket doesn't work.
A reason of this bug is incorrect window parameters and in this case
tcp_acceptable_seq() returns tcp_wnd_end(tp) instead of tp->snd_nxt. The
other side drops packets with this seq, because seq is less than
tp->rcv_nxt ( tcp_sequence() ).
We need to restore window parameters to avoid such side effects.
https://github.com/xemul/criu/issues/168
Signed-off-by: Andrew Vagin <avagin@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
we already have 3/5 fs.mqueue.xxx sysctls so why not have all:
/proc/sys/fs/mqueue/msg_default is a read/write file for
setting/getting the default number of messages in a queue value
if attr parameter of mq_open(2) is NULL. If it exceed msg_max,
the default value is initialized msg_max.
/proc/sys/fs/mqueue/msgsize_default is a read/write file for
setting/getting the default message size value if attr parameter
of mq_open(2) is NULL. If it exceed msgsize_max, the default
value is initialized msgsize_max.
v2: remove check if kernel has sysctls as we do not support such an
old kernel(v3.5)
Signed-off-by: Pavel Tikhomirov <ptikhomirov@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
For standalone unix sockets, listen() will fail if we haven't called bind()
with an actual address. If we remove the name on dump, we won't call
bind(), and thus sockets in this state will fail to restore.
v2: temporarily rename a unix socket out of the way if necessary in order
to bind() correctly and then delete it (e.g. when there are two unix
sockets bound "on top" of each other)
v3: remove extra unlink(), do the real unlink() in bind_unix_sk() so we
only need to do it once
Signed-off-by: Tycho Andersen <tycho.andersen@canonical.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
This entry will be used to carry all the autofs parameters, required to
restore mount point.
Signed-off-by: Stanislav Kinsburskiy <skinsbursky@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
For the previously added option to skip in-flight connections this adds
that option to the RPC interface. The skip in-flight connections is also
described in criu.txt.
Signed-off-by: Adrian Reber <areber@redhat.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
We will need it for NFS. While we're saving it
into image, the test for mode change must be
done in account with NFS. So put fixme there.
CC: Pavel Emelyanov <xemul@virtuozzo.com>
CC: Stanislav Kinsburskiy <skinsbursky@virtuozzo.com>
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
These entries will be used for NFS support and
we already test partial support in vz7 version
of criu. So to make images compatible lets reserve
some of them in advance.
CC: Pavel Emelyanov <xemul@virtuozzo.com>
CC: Stanislav Kinsburskiy <skinsbursky@virtuozzo.com>
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
Needed for container migration, where arguments are
set via p.haul as rpc request.
Signed-off-by: Nikita Spiridonov <nspiridonov@virtuozzo.com>
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
For handling --cgroup-props, --cgroup-props-file and
--cgroup-dump-controller from RPC interface.
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
this patch add join-ns option to criu CLI and RPC.
This opt can be used in this fomat:
--join-ns NS:PID|NS_FILE
for example --join-ns net:12345 or --join-ns net:/foo/bar.
pid namespaces is not supported yet. As fork() is needed to make
new pid-namespace work. That makes it hard for criu to track the
child-process through pid because another child process has been
created after fork().
Signed-off-by: Deng Guangxing <dengguangxing@huawei.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
use native SYSCTL_TYPE__CTL_32 and SYSCTL_TYPE__CTL_STR
v2: add BUILD_BUG_ONS to check SysctlType constants are equal to
__CTL_STR and CTL_32, change __CTL_STR to CTL_STR in SysctlType enum.
Signed-off-by: Pavel Tikhomirov <ptikhomirov@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
Updated version attached.
>From 6c0e1522e01e01aa89861862fbdf039a0892b89b Mon Sep 17 00:00:00 2001
From: Cyrill Gorcunov <gorcunov@openvz.org>
Date: Tue, 12 Apr 2016 20:00:24 +0300
Subject: [PATCH 1/2] tty: Write unread pty buffers on post dump stage
When unread data present on peers we currently simply ignore it but
actually we can try to fetch it in non(that)destructive way.
For this sake at the end of dump procedure (because fetching
queued data may go wrong and we will have to write it back,
which is heavy, and we need all ttys under our hands)
we walk over all collected TTYs and link PTYs peers which
indices are matching. Note to not overload tty_dump_info we
reuse @list member for new @all_ptys list.
Once link established we literally read queued data and flush
it into new tty-data.img. If something go wrong at this moment,
we stop reading queued data but walk back over already queued
ones and write them back to restore former state. Same applies
if the dump has been requested to leave task alive.
On restore we link peers back and write queued data once
peer back to live.
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
See the comment for details, but basically tracefs is automounted by the
kernel, so we can just mount debugfs with MS_REC and get the right result.
v2: rebase on criu-dev
v3: don't use a new fstype->flags, just always set MS_REC in debugfs'
->parse
Signed-off-by: Tycho Andersen <tycho.andersen@canonical.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
Signed-off-by: Pavel Tikhomirov <ptikhomirov@virtuozzo.com>
Reviewed-by: Andrew Vagin <avagin@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
int32 with boolean value in protobuf has the same size with bool,
many sysctls are boolean but we don't lose anything by storing them
in int32, so add only int32 and string fields
will need string field for stable_secret ipv6 sysctl
also such fromat allows us to easily handle non-present int sysctls
we can check if we have it using has_*arg
v3: rebase images/Makefile to criu-dev branch
v4: use enum for type
Signed-off-by: Pavel Tikhomirov <ptikhomirov@virtuozzo.com>
Reviewed-by: Andrew Vagin <avagin@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
Fixes restoring on AMBA serial line. That is /dev/ttyAMA0,
which is the default serial port to use on qemu-system-aarch64.
(at least in those articles, which I meet it's in boot cmdline)
Also should add support for C/R /dev/ttyUSB*
One may check major numbers on
https://www.kernel.org/doc/Documentation/devices.txt
Also by grepping on kernel repository.
Cc: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Dmitry Safonov <dsafonov@virtuozzo.com>
Acked-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
In commit 9469dae61c we've dropped implicit cleanups as
a pattern but in images dir we use additional custom
rules so simply drop files built.
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
Each time I type `make clean` and find that my tags are cleaned
and generated headers/c-files are deleted, I'm getting annoyed.
Mostly it's about protobuf images definitions.
(And then I need to run `make` and `make tags` again which
is painful).
I think, we may separate clean process on two parts, just
like it's done in the kernel (omitting distclean):
o Cleaning all binaries/objects, but leave enough to navigate
o Clean everything compiled/generated/etc.
In this patch pie blobs are still deleted on clean stage - I may
fix it in the later patches.
(they _should_ be regenerated on `make clean && make`)
Cc: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Dmitry Safonov <dsafonov@virtuozzo.com>
Acked-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
cgroup namespaces are imminent to be merged into the kernel (indeed, they
went into and out of 4.5 for minor issues), and will be carried as a
patchset in the ubuntu 16.04 kernel. Here's an attempt at c/r.
There are essentially three key steps:
* on dump, in parse_task_cgroup, we should ask the task what cgroups it
thinks it is in (unless it has the same cgroup ns id as its parent, then we
should just take the prefixes from the parent's set), and set the prefix on
the cg set
* add a new restore step, prepare_cgroup_namespace(), which happens in
prepare_task_cgroup() that does an unshare() if necessary
* when restoring, in move_in_cgroup, if we're going to restore via usernsd,
leave the full path. if not, use (cgset->path + len(cgset->cgns_prefix) as
the path, since we will have already moved into the cgns_prefix and unshared.
Another observation here is that we can support nesting, since these are
restored heirarchically by nature.
v2: * store cgns prefix length instead of full prefix in images
* set has_cgroup_ns_id conditionally
* drop unused argument to move_in_cgroup
* add extra comments about what is happening when unsharing() on
restore
* add extra comments about what is happening when computing the actual
cgns prefix
Signed-off-by: Tycho Andersen <tycho.andersen@canonical.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
This reverts commit a98014f306.
As per Saied Kazemi, actually dump works without seccomp support
from the kernel on non-seccomped tasks. The only problem was with
criu check, but this would be addressed separately.
Reverting the commit not to burden the API with (yet) unneeded stuff.
Conflicts:
lib/c/criu.h
How to restore an external file? The problem is that a file
can be opened with different flags (O_APPNED, O_RDWR, O_RONLY,
etc). These flags can't be changed for a file descriptor, so
we can't just duplicate an inherit fd, when we meet an external
file.
A file is external, when we can't access them directly.
CRIU is able to restore a file descriptor, if it knows
how to open it, so we need to provide a way to open
an external file.
The idea of this patch is that we can open an external file
via /proc/self/fd/X where X is an inherit fd.
This approach works for files and fifo.
An example how it can be used:
criu -o dump.log -D dump/fifo.py/6/1 -v4 -t 6 --external file[72:a3e7]
criu -o restore.log -D dump/fifo.py/6/1 -v4 --restore-sibling --inherit-fd fd[4]:file[72:a3e7] --restore-detached
https://github.com/xemul/criu/issues/104
Cc: Saied Kazemi <saied@google.com>
Signed-off-by: Andrew Vagin <avagin@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
Sometimes we may want to use CRIU on older kernels which don't support
dumping seccomp state where we don't actually care about the seccomp state.
Of course this is unsafe, but it does allow for c/r of things using
seccomp on these older kernels in some cases. When the task is in
SECCOMP_MODE_STRICT or SECCOMP_MODE_FILTER with filters that block the
syscalls criu's parasite code needs, the dump will still fail.
Note that we disable seccomp by simply feigning that we are in mode 0. This
is a little hacky, but avoids distributing ifs throughout the code and
keeps them in this one place.
Signed-off-by: Tycho Andersen <tycho.andersen@canonical.com>
CC: Saied Kazemi <saied@google.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
Docker requested an option, when network devices and routes are not
dumped and not restored. Instead of this Docker will call libnetwork
hook to tune netns from the setup-namespaces action.
Cc: Saied Kazemi <saied@google.com>
Tested-by: Ross Boucher <boucher@gmail.com>
Signed-off-by: Andrew Vagin <avagin@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
But keep @protobuf as a symlink: we have
this path encoded in sources. Gonna be
removed with time.
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
