Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>

Signed-off-by: Radostin Stoyanov <rstoyanov1@gmail.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>

Fixes cov 191305
Signed-off-by: Mike Rapoport <rppt@linux.vnet.ibm.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>

It fails too often and nobody knows how to fix it.
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>

>>>     CID 164715:    (BUFFER_SIZE_WARNING)
>>>     Calling strncpy with a maximum size argument of 16 bytes on
	destination array "thread_args[i].comm" of size 16 bytes might
	leave the destination string unterminated.
3473    strncpy(thread_args[i].comm, core->tc->comm, TASK_COMM_LEN);
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Acked-by: Cyrill Gorcunov <gorcunov@gmail.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>

When zdtm.py is executed with `list` sub-command the 'criu_bin'
option is not defined and criu.available() fails.

$ python test/zdtm.py list
Traceback (most recent call last):
  File "test/zdtm.py", line 2243, in <module>
    criu.available()
  File "test/zdtm.py", line 1185, in available
    if not os.access(opts['criu_bin'], os.X_OK):
KeyError: u'criu_bin'

However, we don't need to check the existence of criu_bin
unless we use the `run` action.
Signed-off-by: Radostin Stoyanov <rstoyanov1@gmail.com>
Acked-by: Adrian Reber <areber@redhat.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>

Signed-off-by: Radostin Stoyanov <rstoyanov1@gmail.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>

It is especially important when accessing a hash --
there must be no negative indices ever.

https://jira.sw.ru/browse/PSBM-82945Signed-off-by: Cyrill Gorcunov <gorcunov@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>

Most of the typos were found by codespell.
Signed-off-by: Radostin Stoyanov <rstoyanov1@gmail.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>

There are no functional changes.
Signed-off-by: Radostin Stoyanov <rstoyanov1@gmail.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>

Signed-off-by: Adrian Reber <areber@redhat.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>

Signed-off-by: Alice Frosi <alice@linux.vnet.ibm.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>

Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>

In commit 23b76949 ("dump: optimize dead_pid_conflict by searching
in rbtree") we'd replaced for_each_pstree_item and thread walk search
with pstree_pid_by_virt (rb-tree search). The latter already found us
the thread with conflicting pid if it exists, the only thing left is to
skip if the thread is also a main thread of the thread group (as it was
before patch). But some leftover left which checks something wrong: we
index ->threads array with "i", but "i" is not a number of the thread
it is a number of current dead pid.

Not sure if it helps with the initial bug and duplicates, but it might:
https://jira.sw.ru/browse/PSBM-55217Signed-off-by: Pavel Tikhomirov <ptikhomirov@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>

This can happen when running in user namespace with auto-dedup enabled.
Right now this means auto-dedup gets disabled.
Signed-off-by: Pawel Stradomski <pstradomski@google.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>

call variables of type RemapFilePathEntry - "rpe" everywhere, similar as
we already name them in oher places

while on it remove unused second argument of open_remap_linked
Signed-off-by: Pavel Tikhomirov <ptikhomirov@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>

./test//zdtm.py --set inhfd run --all --rpc

Cc: Adrian Reber <areber@redhat.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>

After running criu test, docker on the node becomes unusable, as it is
confused by our leftover cgroups. Surely docker should be fixed to
ignore custom cgroups (https://github.com/moby/moby/issues/37601), but
we should not leave them after test also.

v2: rmdir the holder only if it exists, remove racy wait and remove
wrongly added cleanup method in class criu
v3: bring back missed semicolon
Signed-off-by: Pavel Tikhomirov <ptikhomirov@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>

When checkpointing applications with really big memory slab
(like in our vz7 test with 920G of memory) the int type get
cutted, we should use long int instead, just like we do in
other code pieces.

Otherwise get

 | pie: 756: Daemon waits for command
 | (01.193097) Wait for ack 12 on daemon socket
 | (01.193112) Fetched ack: 12 12 0
 | (01.193164) 988065 fdinfo 0: pos:                0 flags:           100002/0
 | (01.193201) fdinfo: type: 0xb flags: 0100002/0 pos:        0 fd: 0
 | (01.193279) 988065 fdinfo 1: pos:                0 flags:           100002/0
 | (01.193307) fdinfo: type: 0xb flags: 0100002/0 pos:        0 fd: 1
 | (01.193341) 988065 fdinfo 2: pos:                0 flags:           100002/0
 | (01.193365) fdinfo: type: 0xb flags: 0100002/0 pos:        0 fd: 2
 | (01.193375) ----------------------------------------
 | (01.193405) Error (criu/parasite-syscall.c:243): BUG at criu/parasite-syscall.c:243
 | pie: 756: Error (criu/pie/parasite.c:676): Trimmed message received (1>
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>

In case if pipe buffers are full the splicing
may wakeup reader first with incomplete data
transfer. Thus splice data in cycle.
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
Acked-by: Mike Rapoport <rppt@linux.vnet.ibm.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>

05:57:35.640:  1238: FAIL: fork.c:80: Task 16275 didn't exit (errno = 10 (No child processes))

There is no info about the killing signal in logs, add it.

https://jira.sw.ru/browse/PSBM-87579Signed-off-by: Pavel Tikhomirov <ptikhomirov@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>

Integer is too small for big memory slabs.
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>

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

Signed-off-by: Mike Rapoport <rppt@linux.vnet.ibm.com>

Signed-off-by: Mike Rapoport <rppt@linux.vnet.ibm.com>

The stack pointers will be later use by the memory dump to ensure that
current stack pages are not treated as lazy.
Signed-off-by: Mike Rapoport <rppt@linux.vnet.ibm.com>

Currently parasite_thread_ctl for non-leader threads is initialized after we
stop the compel daemon. Moving this initialization earlier will allow us to
make stack pointers of all threads available during memory dump.
Signed-off-by: Mike Rapoport <rppt@linux.vnet.ibm.com>

Signed-off-by: Mike Rapoport <rppt@linux.vnet.ibm.com>

We need to know what are stack pointers of every thread to ensure that the
current stack page will not be treated as lazy.
Signed-off-by: Mike Rapoport <rppt@linux.vnet.ibm.com>
Acked-by: Alice Frosi <alice@linux.vnet.ibm.com>
Reviewed-by: Laurent Dufour <ldufour@linux.vnet.ibm.com>

Signed-off-by: Adrian Reber <areber@redhat.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>

This is obviously not a real check. This only exists, so that CRIU
clients/users can check if this CRIU version supports the external
network namespace feature. Theoretically the CRIU client or user could
also parse the version, but especially for CLI users version comparison
in the shell is not easy.

This feature check does not exist for RPC as RPC has a special version
call which does not require string parsing and the external network
namespace feature is available for all CRIU versions newer than 3.9.
Signed-off-by: Adrian Reber <areber@redhat.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>

Since the processes and fds are restored from images in the increasing
order, and they are stored in sorted lists, CRIU always needed (at least
in my experiments) to traverse the whole list before inserting them.
With this change, lists are traversed in reverse, so the fds should be
inserted immediately.
Acked-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Radoslaw Burny <rburny@google.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>

The latest patches to cr-service.c broke compilation with gcc-8:

criu/cr-service.c: In function ‘setup_opts_from_req’:
criu/cr-service.c:323:3: error: ‘strncpy’ specified bound 4096 equals destination size [-Werror=stringop-truncation]
   strncpy(images_dir_path, opts.imgs_dir, PATH_MAX);
   ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
criu/cr-service.c:343:3: error: ‘strncpy’ specified bound 4096 equals destination size [-Werror=stringop-truncation]
   strncpy(work_dir_path, opts.work_dir, PATH_MAX);
   ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
cc1: all warnings being treated as errors

This fixes the errors by specifying the strncpy() size as 'PATH_MAX - 1'.
Signed-off-by: Adrian Reber <areber@redhat.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>

With this commit it is possible to specify a configuration file via
RPC. In python this would look like this:

 req.opts.config_file = 'path/to/config_file'

With this commit CRIU's configuration file handling works like this:

 * apply_config(global_conf)
 * apply_config(user_conf)
 * apply_config(environment variable)
 * apply_config(config file via CLI)
 * apply_rpc_options() or apply_cli_options()
 * apply_config(rpc_conf) (only for RPC)

This is at least (probably) the third iteration of the RPC configuration
file code and it still is complicated.

Most CRIU options are correctly used by just writing the new values to
the corresponding fields of the opts structure. For the RPC case there
are, however, a few options (output, work_dir, imgs_dir) which need
special handling.

So the RPC configuration file is parsed twice. First time to get output,
work_dir and imgs_dir. Once those are read and correctly used, the RPC
code overwrites all options again by values set by the RPC interface. At
the end the RPC configuration file is read a second time and finally
overwrites the values set via RPC.
Signed-off-by: Adrian Reber <areber@redhat.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>

This test checks the following things:

 * Does configuration file parsing work at all.
 * Does the parser detect wrong options.
 * Does the configuration file work via RPC.
 * Do the configuration file options not overwrite the RPC settings in
   the default setup.
 * Is it possible to tell CRIU to prefer the configuration file via RPC.
Signed-off-by: Adrian Reber <areber@redhat.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>

When CRIU is started in RPC mode ('criu swrk') it used to ignore all
command-line options and configuration files. This moves the jump to RPC
mode after the configuration file parsing to enable configuration.

With this configuration files are now also evaluated in RPC mode and it
is possible to change the behavior of CRIU via the configuration file if
used via RPC.
Signed-off-by: Adrian Reber <areber@redhat.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>

With this it is possible to point the environment variable
CRIU_CONFIG_FILE to a CRIU configuration file.

The order the configuration files are evaluated now is:

 1. global (/etc/criu/default.conf)
 2. user ($HOME/.criu/default.conf)
 3. CRIU_CONFIG_FILE
 4. --config FILENAME
 5. CLI
Signed-off-by: Adrian Reber <areber@redhat.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>

With the previous change to dynamically allocate memory for each
possible configuration source (three different configuration files, CLI,
RPC) the char * options can no longer directly point to the character
strings extracted by getopt() as the memory might be free'd at some
point. This introduces a macro to set the char * options which first
does a xfree() and then a xstrdup().
Signed-off-by: Adrian Reber <areber@redhat.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>

The code to read and parse configuration files was writing the result of
the global configuration file to one variable and the result from the
configuration file in the user's home to another variable:

  char **global_conf = NULL;
  char **user_conf = NULL;

With this change the code now uses dynamic memory allocation to handle
the different configuration files.

It used to be:

  * parse global config
  * parse user config
  * evaluate global config
  * evaluate user config
  * evaluate CLI

And now it is:

  * parse global config
  * evaluate global config
  * parse user config
  * evaluate user config
  * evaluate CLI

This change is in preparation for the upcoming setting of a
configuration file via environment variable and RPC configuration file
usage.
Signed-off-by: Adrian Reber <areber@redhat.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>

This just moves the functions init_opts() and deprecated_ok() also to
config.c as that is where most of the option and configuration setup
and handling is done today.
Signed-off-by: Adrian Reber <areber@redhat.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>