The server socket is marked as nonblocking, and if the client doesn't
connect, accept() will fail and set errno to EAGAIN (or EWOULDBLOCK).
Instead, use poll to wait for POLLIN event on the file descriptor.
Suggested-by: Andrei Vagin <avagin@gmail.com>
Signed-off-by: Radostin Stoyanov <rstoyanov1@gmail.com>
Running crit with python2 gives following minimal help message:
$ crit/crit
usage: crit [-h] {decode,encode,info,x,show} ...
crit: error: too few arguments
Using a python3 only system crit shows the following error:
$ crit/crit
Traceback (most recent call last):
File "crit/crit", line 6, in <module>
cli.main()
File "/home/criu/crit/pycriu/cli.py", line 334, in main
opts["func"](opts)
KeyError: 'func'
Using this patch the python3 output changes to:
$ crit/crit
usage: crit [-h] {decode,encode,info,x,show} ...
crit: error: too few arguments
Suggested-by: Andrei Vagin <avagin@gmail.com>
Signed-off-by: Adrian Reber <areber@redhat.com>
When the --ps-socket option is used with page-server, instead of
--address and --port, this message would appear as:
(00.028440) Disconnect from the page server (null):0
Signed-off-by: Radostin Stoyanov <rstoyanov1@gmail.com>
From man inet_pton(3):
inet_pton() returns 1 on success (network address was successfully
converted). 0 is returned if src does not contain a character
string representing a valid network address in the specified
address family. If af does not contain a valid address family,
-1 is returned and errno is set to EAFNOSUPPORT.
We can assume that the return value is 1 or 0 (because af is set to
AF_INET4 or AF_INET6), therefore errno will not be set.
If a user attempts to bind a server using invalid network address the
following error message will be shown:
Bad server address: Success
Which is not very clear, with this change the error message will look
like this:
Invalid server address "localhost". The address must be in IPv4 or IPv6 format.
Signed-off-by: Radostin Stoyanov <rstoyanov1@gmail.com>
* "post-resume" was introduced with commit:
2ab599398d
cr-restore: "post-resume" hook introduced
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.
* "post-setup-namespaces" was introduced with commit:
eec66f3d30
criu [PATCH] post-setup-namespaces
Introduce post-setup-namespaces action script
It needed to have possibility to run cutom script after mount
namespace is configured
* "orphan-pts-master" was introduced with commit:
6afe523d97
tty: notify about orphan tty-s via rpc
Now Docker creates a pty pair from a container devpts to use is as console.
A slave tty is set as a control tty for the init process and bind-mounted
into /dev/console. The master tty is handled externelly.
Now CRIU can handle external resources, but here we have internal resources
which are used externaly.
Signed-off-by: Radostin Stoyanov <rstoyanov1@gmail.com>
Suppress the false positive fail in criu-live-migration job:
https://ci.openvz.org/job/CRIU/job/criu-live-migration/job/criu-dev/1796/
[criu]# ./test/zdtm.py run -t zdtm/static/overmounted_file -f uns --lazy-migrate
=== Run 1/1 ================ zdtm/static/overmounted_file
=================== Run zdtm/static/overmounted_file in uns ====================
Start test
Test is SUID
./overmounted_file --pidfile=overmounted_file.pid --outfile=overmounted_file.out --dirname=overmounted_file.test
Run criu dump
Test zdtm/static/overmounted_file FAIL at criu dump exited with 1 ######
Send the 9 signal to 49
Wait for zdtm/static/overmounted_file(49) to die for 0.100000
Signed-off-by: Pavel Tikhomirov <ptikhomirov@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@gmail.com>
The '-R' is short for '--leave-running', which is a boolean option and
does not require an argument.
From getopt(3) man page:
optstring is a string containing the legitimate option characters. If
such a character is followed by a colon, the option requires an
argument, ...
Signed-off-by: Radostin Stoyanov <rstoyanov1@gmail.com>
This commit removes the functions criu_(local_)set_service_comm().
These functions are not neccessary, because if
set_service_address(), set_service_fd() or
set_service_binary() has been called it is already clear, which
service comm type should be used.
Furhtermore, this commit reduces the number of misuses.
E.g. if set_service_comm() was set to socket, but a binary was given
via set_service_binary().
Signed-off-by: Martin Wührer <martin.wuehrer@artech.at>
This commit checks after each strdup() call if the call was successful.
If not, the function that calls strdup() returns an error.
This requires, that the return value of several functions has to be
changed from void to int.
Signed-off-by: Martin Wührer <martin.wuehrer@artech.at>
Files from such mounts can switch on restore to different files on the
overmounting mount, as we yet don't fully control the mount on which
the file is restored.
v3: make mnt_is_overmounted boolean again
Signed-off-by: Pavel Tikhomirov <ptikhomirov@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@gmail.com>
Test creates two shared mounts, both are readonly, on each mount we have
a different ghost file. Also let these mounts have different mount
flags. That makes us both test do_new_mount hunk for delaying readonly
and do_bind_mount, as one mount will bind from another on restore.
https://jira.sw.ru/browse/PSBM-82991
v2: minor cleanup
Signed-off-by: Pavel Tikhomirov <ptikhomirov@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@gmail.com>
We can have ghost-files on readonly mounts, for them we will need to
recreate the file on restore, and we can't do that if mount is readonly,
so the idea is to remount the mount we want to operate on to be writable,
and later after all ghost-files restored return mounts to their proper
state if needed.
There are three exceptions, where we don't remount:
a) Overmounted mounts can't be easily remounted writable, as their
mountpoints are invisible for us.
b) If the mount has readonly superblock - there can be no ghost-files on
such a mount.
c) When we are in host mntns, we should not remount mounts in it, else
if we face errors in between we'll forget to remount back.
We have 3 places where we need to add these remount:
1) create_ghost()
2) clean_one_remap()
3) rfi_remap()
For (1) and (2) we can just remount the mount writable without
remounting it back as they are called in service mntns (the one we save
in mnt_ns_fd), which will be destroyed with all it's mounts at the end.
We mark such mounts as remounted in service mntns - REMOUNTED_RW_SERVICE.
For (3) we need to remount these mounts back to readonly so we mark them
with REMOUNTED_RW and later in remount_readonly_mounts all such mounts
are re-remounted back.
For (3) we also need to enter proper mntns of tmi before remounting.
These solution v3 is better than v2 as for v2 we added additional
remount for all bind-readonly mounts, now we do remounts only for
those having ghost-files restore operations on them. These should be
quiet a rare thing, so ~3 remounts added for each suitable mount is a
relatively small price.
note: Also I thought and tried to implement the complete remove of the
step of remounting back to readonly, but it requires quiet a tricky
playing with usernsd and only removes one remount (of ~3) for already a
rare case so I don't thing it worth the effort.
v2: minor commit message cleanup and remove warn
v4: don't delay, only remount the mounts we explicitly want to write to
just before operating, rename patch accordingly, reuse
do_restore_task_mnt_ns, optimize inefficient ns_remount_readonly_mounts,
and also add another exception.
v5: simplify child status check, fix log messages and brackets, do not
drop all flags but only the readonly flag
Signed-off-by: Pavel Tikhomirov <ptikhomirov@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@gmail.com>
On restore we change mounts tree topology and paths in find_remap_mounts
thus mns_is_overmounted does not account remaped overmounts. So to know
which mounts will be actually overmounted after restore we need two much
effort, better prepare these info in advance.
That also should improve performance of mnt_is_overmounted on dump,
as we reuse calculations from previous calls.
v3: make mnt_is_overmounted boolean again
Signed-off-by: Pavel Tikhomirov <ptikhomirov@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@gmail.com>
Use realloc() to change the size of the memory block pointed
to by info->source instead of allocating new memory area and
releasing the old one. If info->source is NULL then the call
is equivalent to malloc(size).
Signed-off-by: Radostin Stoyanov <rstoyanov1@gmail.com>
The `show` and `exec` commands have been deprecated for some time,
and it is unlikely to be called by users. It would be better to move
them the end of the sequence of if-statements.
Signed-off-by: Radostin Stoyanov <rstoyanov1@gmail.com>
The function parse_options() should return an integer 0, 1 or 2, where:
0 := Success
1 := Failure
2 := CRIU should display help text
The current behaviour always creates a warning message which shows
the return value of that function. However, when the return value
is different than 0 (Success) CRIU will fail. Since the logging is
not initialised when this message is created, it will be lost, and
therefore it can be removed completely.
Signed-off-by: Radostin Stoyanov <rstoyanov1@gmail.com>
Although non-cooperative userfaultfd events have been merged into the Linux
kernel since 4.11 they were racy until 4.18.
Exclude "dangerous" tests for older kernels.
Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
Somehow in travis tests the mmap of 8K gets "rounded" down and the actual
VMA starts 4K below the returned address. This causes /proc/pid/smaps
parsing to fail.
Increasing the allocation size to 2M seems to resolve the problem and until
the root cause is identified it looks like a viable workaround.
Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
The functions criu_(local_)set_service_address and
criu_(local_)set_service_binary are the only functions that
do not create a copy of the given string arguments.
This may lead to problems, if the original string gets freed
but criu relies on them.
Additionally, the function criu_local_init_opts() assigns the
default service_binary now to opt->service_binary instead of
opt->service_address which is in my opinion the preferred way
since both are types of an anonymous union.
Signed-off-by: Martin Wührer <martin.wuehrer@artech.at>
Verify that enabling / disabling THP with prctl(PR_SET_THP_DISABLE) does
not change VMA flags
Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
Signed-off-by: Andrei Vagin <avagin@gmail.com>
Since asciidoc is based on Phyton 2, we want to move to alternative,
and a promising one is asciidoctor. This patch allows to use
asciidoctor for formatting man pages instead of asiidoc, by passing
a make option, USE_ASCIIDOCTOR=yes.
Although asciidoctor is almost compatible with asciidoc, it can
produce a man page directly from a text file without XML, which is
more efficiently. So in asciidoctor mode, we don't require xmlto.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Andrei Vagin <avagin@gmail.com>
a2x is never used although its presence is checked mandatorily.
Let's remove this superfluous check and the unused entry.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Reviewed-by: Dmitry Safonov <0x7f454c46@gmail.com>
Signed-off-by: Andrei Vagin <avagin@gmail.com>
Split the function parse_config() into smaller parts by introducing
the parse_statement(), and add more descriptive comments to improve
readability. In addition, make sure that the last element of the array
of strings 'configuration' is initialised to NULL, which is necessary
to properly count the number of elements after parsing.
Signed-off-by: Radostin Stoyanov <rstoyanov1@gmail.com>
Signed-off-by: Andrei Vagin <avagin@gmail.com>
As most of the `criu_(local_)*` functions already call `strdup()`,
it is possible, to change the function signature to `const char *`.
As the struct `criu_opts` already contains a `const char *
service_binary`, also the member `service_address` is changed to
`const char`.
Additonally, also the function `criu_local_set_freeze_cgroup()` now
calls `strdup()`.
Signed-off-by: Martin Wührer <martin.wuehrer@artech.at>
Signed-off-by: Andrei Vagin <avagin@gmail.com>
According to https://criu.org/API_compliance, the C-library
doesn't support the pageserver option.
This patch contains the functions
`criu_(local_)set_page_server_address_port()`
that allow to specify on which ip and tcp-port the pageserver
is listening.
This patch affects only the c-lib, as criu-rpc already supports the
pageserver settings.
Signed-off-by: Martin Wührer <martin.wuehrer@artech.at>
Signed-off-by: Andrei Vagin <avagin@gmail.com>
Add run as a phony target. Move .PHONY declarations next to
their targets.
Reviewed-by: Cyrill Gorcunov <gorcunov@gmail.com>
Signed-off-by: Kir Kolyshkin <kolyshkin@gmail.com>
Signed-off-by: Andrei Vagin <avagin@gmail.com>
Here $(Q) is substituted with either @ or nothing, depending on
whether V is set, to either have verbose output or not. So, it does
not make sense to add another @.
Reviewed-by: Cyrill Gorcunov <gorcunov@gmail.com>
Signed-off-by: Kir Kolyshkin <kolyshkin@gmail.com>
Signed-off-by: Andrei Vagin <avagin@gmail.com>
Since this part is inside "define" statement, double $ should be
used to refer to internal functions, to call the patsubst when
a define is called (rather than defined).
This should fix dependency generation for images/
Reviewed-by: Cyrill Gorcunov <gorcunov@gmail.com>
Signed-off-by: Kir Kolyshkin <kolyshkin@gmail.com>
Signed-off-by: Andrei Vagin <avagin@gmail.com>
If `criu_local_init_opts()` is applied on the same opts-object
several times, not all of the allocated memory gets freed.
Therefore, the functions `criu_(local_)free_opts()` were introduced.
These functions ensure, that opts get freed accordingly.
Furthermore, `criu_(local_)free_opts()` gets part of the c-api,
and can therefore be called by external projects too.
Additionally, with this commit `criu_local_init_opts()` now uses
`criu_local_free_opts()`, to free the opts-parameter if it was already
initalized before.
This commit also contains a fix in `send_req_and_recv_resp_sk()` which
lead to a memory leak, if criu-notifications were received.
Signed-off-by: Martin Wührer <martin.wuehrer@artech.at>
Signed-off-by: Andrei Vagin <avagin@gmail.com>
The purpose of this helper function is to perform actions common to
different interfaces, after all CRIU options have been parsed. This can
be used, for instance, to verify that a specific option was passed to
CRIU via CLI, RPC or config file.
Signed-off-by: Radostin Stoyanov <rstoyanov1@gmail.com>
Acked-by: Adrian Reber <areber@redhat.com>
Signed-off-by: Andrei Vagin <avagin@gmail.com>
The header "lsm.h" was added in crtools.c with commit 6af96c8
and it became unused after commit 09c131c.
Signed-off-by: Radostin Stoyanov <rstoyanov1@gmail.com>
Signed-off-by: Andrei Vagin <avagin@gmail.com>
The option --lsm-profile was added with commit:
6af96c8404
lsm: add a --lsm-profile flag
In LXD, we use the container name in the LSM profile. If the container name
is changed on migrate (on the host side), we want to use a different LSM
profile name (a. la. --cgroup-root). This flag adds that support.
A usage example is available in
13389b2963
Signed-off-by: Radostin Stoyanov <rstoyanov1@gmail.com>
Signed-off-by: Andrei Vagin <avagin@gmail.com>
The --tcp-close option was introduced with commit
2c37042821
tcp: Add tcp-close option to restore connected TCP sockets in closed state
This options is applicable only for restore. Therefore, move the
documentation from 'dump' to 'restore'.
Signed-off-by: Radostin Stoyanov <rstoyanov1@gmail.com>
Signed-off-by: Andrei Vagin <avagin@gmail.com>
The --log-pid option was introduced with commit
fc7bedc50a
crtools: make to be able to split messages by pid
This option is applicable only for restore. Therefore, move the
documentation from "Common options" to "restore".
Signed-off-by: Radostin Stoyanov <rstoyanov1@gmail.com>
Signed-off-by: Andrei Vagin <avagin@gmail.com>
The current behaviour of CRIU is to inherit the group leader of the
parent process when migrating a session leader (with --shell-job).
However, it is possible for a process to be group leader without being
a session leader (e.g. a shell process). In this case CRIU should
restore the original group ID of the process and inherit only the
session ID.
Closes#593
Signed-off-by: Radostin Stoyanov <rstoyanov1@gmail.com>
Reviewed-by: Cyrill Gorcunov <gorcunov@gmail.com>
Signed-off-by: Andrei Vagin <avagin@gmail.com>
getsid(0) and getpgid(0) will return session ID and PGID of
the calling process.
Signed-off-by: Radostin Stoyanov <rstoyanov1@gmail.com>
Signed-off-by: Andrei Vagin <avagin@gmail.com>
I received this patch from Jeff Law as a fix for build failures with the
upcoming GCC 9. The following is Jeff Law's description of the patch:
Attached you'll find the fix for criu. You'll see it's just a matter
of dropping the sp/esp clobber from the relevant asm statements. THe
details:
criu has a macro which defines an asm which appears to want to set a new
stack pointer, then directly issue a sigreturn call to the kernel. Some
variants clobber sp (aarch64, arm, x86), others do not (ppc, s390)
While the asm does indeed set a new stack pointer, we never return from
a sigreturn syscall -- at least not in the normal way. We actually
return back to the point where the process was interrupted by the
signal. So describing the affect of the asm on the stack pointer is
pedantically correct, it actually has no real effect and can just be
dropped to avoid the hard error from gcc-9.
Suggested-by: Jeff Law <law@redhat.com>
Reviewed-by: Dmitry Safonov <0x7f454c46@gmail.com>
Signed-off-by: Adrian Reber <areber@redhat.com>
Signed-off-by: Andrei Vagin <avagin@gmail.com>
