Now criu create a new pipe buffer, if a previous one has another set of
flags. In this case, a pipe is not full and we can use it for the
next page buffer.
We need 88 pipes to pre-dump the zdtm/static/fork test without this
patch, and we need only 17 pipes with this patch.
Acked-by: Mike Rapoport <rppt@linux.vnet.ibm.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
v2: and move it upper, because it is going to be used in ppb_alloc()
Acked-by: Mike Rapoport <rppt@linux.vnet.ibm.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
vmsplice can't splice more than UIO_MAXIOV, but we can
call it a few times from a parasite.
v2: s/nr/nr_segs/
Acked-by: Mike Rapoport <rppt@linux.vnet.ibm.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
In addition to writing the CRIU version to the log file this adds the
current kernel version to the log file:
(00.000008) Version: 3.5 (gitid v3.5-511-ga8cc6cf)
(00.000303) Running on node01 Linux 3.10.0-513.el7.x86_64 #1 SMP Tue Feb 29 06:78:90 EST 2017 x86_64
v2:
- small changes as suggested by Dmitry (thanks)
Signed-off-by: Adrian Reber <areber@redhat.com>
Reviewed-by: Dmitry Safonov <0x7f454c46@gmail.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
If ipv6 socket has an IPv4-mapped address, it is used to handle ipv4
connection, so we have to use ipv4 iptables rules to block this
connection.
Reported-by: Mr Jenkins
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Currently we use the "map_files/%p-%p" format, but actually it should
be "map_files/%lx-%lx".
The kernel could handle both formats, but recently Alexey Dobriyan fixed
the kernel and it accept only the second format.
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Acked-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
We always ask users what version of criu they use to investigate a problem,
so it better to have it in a log.
Signed-off-by: Andrei Vagin <avagin@openvz.org>
Reviewed-by: Dmitry Safonov <0x7f454c46@gmail.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
criu and soccr has different values for log levels, so
someone has to remap them.
Cc: Cyrill Gorcunov <gorcunov@openvz.org>
Reported-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Andrei Vagin <avagin@openvz.org>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
This can help to investigate logs from Mr Jenkins.
Cc: Mike Rapoport <rppt@linux.vnet.ibm.com>
Signed-off-by: Andrei Vagin <avagin@openvz.org>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
When inotify is laying on uovermounted fs we should walk over
all mountpoints with same s_dev to find openable path.
Note on restore the path is usually already allocated during
dump stage so get_mark_path won't call for open_handle(), in
turn on dump stage the positive return from open_handle()
will cause fsnotify engine to find openable path, thus there
is kind of double work to be optimized in future.
For example we got a container where systemd-udevd inside
opens inotify for /dev/X entry then overmount ./dev path
with slave option and in result irmap engine on predump
can't figure out where the inotify is sitting causing
migrtion to abort.
Signed-off-by: Cyrill Gorcunov <gorcunov@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
config.h is a generated file with "build-features" defines.
We use it for several purposes:
o to check that compiler can do it's job
o to complement user-visible API between distributions
o to add compile-time options from .config global file
It's used in criu and soccr, but compel also needs such thing.
Previously, soccr has a link to config.h in criu includes,
but it would be much cleaner to move it to other headers,
that are shared between sub-projects into include/common.
Reported-by: Adrian Reber <areber@redhat.com>
Signed-off-by: Dmitry Safonov <dsafonov@virtuozzo.com>
Tested-by: Adrian Reber <areber@redhat.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Distributions starts to supply GCC that is configured to compile
-pie and -fPIC code by default due to security reasons.
CONFIG_COMPAT was unfriendy to -pie by the reason of R_X86_64_32S
relocation in call32.S helper:
LINK criu/criu
/usr/bin/ld: criu/arch/x86/crtools.built-in.o: relocation R_X86_64_32S against `.text' can not be used when making a shared object; recompile with -fPIC
/usr/bin/ld: final link failed: Nonrepresentable section on output
collect2: error: ld returned 1 exit status
make[1]: *** [criu/Makefile:92: criu/criu] Error 1
make: *** [Makefile:225: criu] Error 2
Use %rip-relative addressing to avoid ld errors for shared binary linking.
Puff, all needs to be done with bare hands!
Now CONFIG_COMPAT can be used with -pie binaries and all should
also work for debian toolchain (#315).
Signed-off-by: Dmitry Safonov <dsafonov@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
In this case we can wait it and get an exit code.
For example, it will be useful for p.haul where one connection
is used several times, so we need a way how to understand that
page-server exited unexpectedly.
v2: don't write ps_info if a start descriptor isn't set
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
This file is not executable directly, so it should not have the shebang.
Signed-off-by: Avindra Goolcharan <aavindraa@gmail.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
warning: In the GNU C Library, "major" is defined
by <sys/sysmacros.h>. For historical compatibility, it is
currently defined by <sys/types.h> as well, but we plan to
remove this soon. To use "major", include <sys/sysmacros.h>
directly. If you did not intend to use a system-defined macro
"major", you should undefine it after including <sys/types.h>.
if (major(st.st_rdev) != major(st_rtc.st_rdev) ||
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
aio_context_t is 8 byte long so on 32 bit mode it might be
strippped off when unsigned long used instead. Fix this typo.
Signed-off-by: Cyrill Gorcunov <gorcunov@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
v2: defining crit_bin and using it for Popen() // Mike
Acked-by: Mike Rapoport <rppt@linux.vnet.ibm.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
The number of pipes are limited in a system, so it is better to know how
many we use.
Acked-by: Mike Rapoport <rppt@linux.vnet.ibm.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
IPv6 listening sockets can accept both ipv4 and ipv6 connections,
in both cases a family of an accepted socket will be AF_INET6.
But we have to send tcp packets accoding with a connection type.
------------------------ grep Error ------------------------
(00.002320) 53: Debug: Will set rcv_wscale to 7
(00.002325) 53: Debug: Will turn timestamps on
(00.002331) 53: Debug: Will set mss clamp to 65495
(00.002338) 53: Debug: Restoring TCP 1 queue data 2 bytes
(00.002403) 53: Error (soccr/soccr.c:673): Unable to send a fin packet: libnet_write_raw_ipv6(): -1 bytes written (Network is unreachable)
(00.002434) 53: Error (criu/files.c:1191): Unable to open fd=3 id=0x6
(00.002506) Error (criu/cr-restore.c:2171): Restoring FAILED.
------------------------ ERROR OVER ------------------------
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
The SO_REUSEPORT option allows multiple sockets on the same
host to bind to the same port. This option has to ve restored when all
sockets are bound to a port. The same logic is already used to restore
SO_REUSEADDR.
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Regs are present in unsigned format so convert them
into signed first to provide results.
In particular if memfd_create syscall failed we won't
notice -ENOMEM error but rather treat it as unsigned
hex value
| (05.303002) Putting parasite blob into 0x7f1c6ffe0000->0xfffffff4
| (05.303234) Putting tsock into pid 42773
Signed-off-by: Cyrill Gorcunov <gorcunov@virtuozzo.com>
Reviewed-by: Dmitry Safonov <dima@arista.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
In CT, we do:
mkdir -p /a/b/c1
mkdir -p /c2
mount --bind /c2 /a/b/c1
mount --rbind /a/b /a
And after that container is not dumpable with error:
mnt: Unable to handle mounts under 146:./a
Just because overmounts with shared parent group are prohibited,
but I can't see any problem with enabling them.
https://jira.sw.ru/browse/PSBM-69501
Signed-off-by: Pavel Tikhomirov <ptikhomirov@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
What we do before patch:
1) If we are NOT in the same shared group - if we have some parent's
shared group member unmounted, we just wait for it.
2) If we are in the same group - we wait only for members with root
path len shorter than ours.
That is done to make child mount propagate in all shared group,
but I think it is wrong, e.g.:
mkdir -p /dir/a/b/c /d /e /f
mount --bind /dir/a /d
mount --bind /dir/a/b /e
mount --bind /f /e/c
Before c/r we have:
507 114 182:1017985 / / rw,relatime shared:63 master:60 - ext4 /dev/ploop63624p1 rw,data=ordered,balloon_ino=12
144 507 182:1017985 /dir/a /d rw,relatime shared:63 master:60 - ext4 /dev/ploop63624p1 rw,data=ordered,balloon_ino=12
146 507 182:1017985 /dir/a/b /e rw,relatime shared:63 master:60 - ext4 /dev/ploop63624p1 rw,data=ordered,balloon_ino=12
148 146 182:1017985 /f /e/c rw,relatime shared:63 master:60 - ext4 /dev/ploop63624p1 rw,data=ordered,balloon_ino=12
150 507 182:1017985 /f /dir/a/b/c rw,relatime shared:63 master:60 - ext4 /dev/ploop63624p1 rw,data=ordered,balloon_ino=12
149 144 182:1017985 /f /d/b/c rw,relatime shared:63 master:60 - ext4 /dev/ploop63624p1 rw,data=ordered,balloon_ino=12
After c/r we have:
600 132 182:1017985 / / rw,relatime shared:63 master:60 - ext4 /dev/ploop63624p1 rw,data=ordered,balloon_ino=12
602 600 182:1017985 /f /dir/a/b/c rw,relatime shared:63 master:60 - ext4 /dev/ploop63624p1 rw,data=ordered,balloon_ino=12
603 600 182:1017985 /dir/a /d rw,relatime shared:63 master:60 - ext4 /dev/ploop63624p1 rw,data=ordered,balloon_ino=12
604 600 182:1017985 /dir/a/b /e rw,relatime shared:63 master:60 - ext4 /dev/ploop63624p1 rw,data=ordered,balloon_ino=12
There is no propagation as all mounts are in same shared group and
602(150) has shorter root than 603(144) and 604(146).
What we should do:
Wait member of our parent's shared group only if it has our 'sibling'
mount in it. Sibling mount is the one which had propagated to shared
mount of our parent for us when we were mounted. We need to enforce
propagation only for these case.
https://jira.sw.ru/browse/PSBM-69501
Signed-off-by: Pavel Tikhomirov <ptikhomirov@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
The anonymous shared memory are using shmid for image
name encoding which is unsigned long and we already
met scenario where high bits get strippped off thus
the restore failed.
Lets use unsigned long here, and because pagemap code
is shared between plain memory and anon shared memory
use unsigned long every where.
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
The index comes from mnt_id which is signed integer
both in kernel and in userspace, but negative value
is never valid, thus don't use it.
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
For images which are using pid as id for image names
use unsigned format since here is no negative pid
in real system.
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Both the kernel and criu uses unsigned int
for it, make the format appropriate.
| struct mount_info {
| ...
| unsigned int s_dev;
| ...
| }
We didn't see negative number here in real life so
I don't think if such %d to %u convention cause
backward compatibility problem ever.
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
We pass unsigned 4 byte integer here, so
use appropriate format.
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Namespace descriptors are not promised to have
constant short names, so just to be on a safe
side.
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
