Introduce a new -o argument to piegen to specify generate file name.
Send the debug stream to stdout and force it to /dev/null in the makefile
if V=1 is not specify.
Signed-off-by: Laurent Dufour <ldufour@linux.vnet.ibm.com>
Acked-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>

handle elf should return 0 on success.
Signed-off-by: Laurent Dufour <ldufour@linux.vnet.ibm.com>
Acked-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>

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

To hide @nr_gotpcrel
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>

When building the blob in the generated header file, we may
shrink the output blobk and only copy the sections with the SHF_ALLOC
bit set, the other ones are not needed at runtime.
Signed-off-by: Laurent Dufour <ldufour@linux.vnet.ibm.com>
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>

This cleans the assembly code, removing no more needed trick with the
register 2 (TOC pointer). As a consequence, the __export_restore_task_trampoline()
and __export_unmap_trampoline() are no more needed.

Thus, the changes introduced by the commit de9df910 ("Per architecture restorer
trampolines") in cr-restore.c are no more used but are not impacting
runtime code anyway.
Signed-off-by: Laurent Dufour <ldufour@linux.vnet.ibm.com>
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>

PowerPC linking requires the TOC to be in its own section
and to be aligned.
Signed-off-by: Laurent Dufour <ldufour@linux.vnet.ibm.com>
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>

 - Move relocs application into a separate file which get
   compiled as a regular C file in criu (pie/pie-relocs.[ch])
 - Move types used by piegen into pie/piegen/uapi/types.h
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>

At moment both parasite and restorer do not have
any relocs because we support x86-64 only, but
this will be changed soon so do a call and apply
relocations.
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>

We should use provided @nr_relocs instead of ARRAY_SIZE here.
Didn't spot it earlier simply because at moment on x86-64
there is no relocs at all.

Also when we apply relocation they are to be computed from
virtual base of parasite address, not from local memory
map address, so add @vbase parameter. And fix typo on
addend in gotpcrel.
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>

In case of @gotpcrel relocations we need additional
space to carry pointers.
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>

After this patch one can run ARCH="ia32" make to build
32bit version on CRIU on 64bit host. Note this is only
build procedure which tuned up, the CRIU itself is not
yet ready to make a checkpoint/restore cycle -- a lot
of additional code is needed and here we rather put
stubs simply to make build procedure run.
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>

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

There is no rip addressing in 32bit code but PIE code
require GOT tables and friend which we better escape
for performance sake. So lets use pc relocations it
should do the trick.
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>

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

To support x86-32 mode we will need own syscall table.
Here is it. Note the CRIU itself doesn't support such
mode yet.

Meanwhile put syscall table here just in case if someone
is adding new syscall 32 bit variant should be updated
as well.
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>

Since at the moment we're running only x86-64 not 32 bit tasks,
and our code is not carrying any big statically defined structures
we can use relocatable files directly with all relocation applied
during building.

This gonna be changed soon once we start supporting 32 bit tasks.

IOW even currently we need (which is not yet done but it's safe)

 - check for gotpcrel relocations
 - apply relocations with generated elf_apply_relocs helper

Currently overall scheme looks this way

 - our object files are linked together into parasite.built-in.bin.o file
 - then pie/piegen/piegen tool is called which parses this file and generates
   C source code file with bytestream and all information needed to rellocate
   this bytestream into a new place (see elf_apply_relocs helper)
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>

Here we simply build piegen tool which gonna be used
to generate parasite code safe to rellocate. The tool
is taking object file as an argument, parses it and
generates C file with rellocations encoded in form
suitable for fast appliance.

Currently only x86-32 x86-64 is supported.

v2 (by ldufour@):
 - Filter PIEGEN
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>

On restore we have several arrays of objects that get remapped
into pie area and their number is also passed. Clean and shorten
the remapping code a bit and bing their naming to common format.
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>

systemd executes tests in subgroups:
 9679 ?        Ssl    0:41 /etc/alternatives/java -Dcom.sun.akuma.Daemon=daemonized -Djava.awt.headless=true -DJENKINS_HOME=/var/lib/jenkins -jar /usr/lib/jenkins/jenkins.war -
[root@jenkins ~]# cat /proc/9679/cgroup
10:net_cls:/
9:hugetlb:/
8:cpuset:/
7:blkio:/system.slice
6:freezer:/
5:cpu,cpuacct:/system.slice
4:devices:/system.slice/jenkins.service
3:perf_event:/
2:memory:/system.slice
1:name=systemd:/system.slice/jenkins.service

Reported-by: Mr Jenkins
Signed-off-by: Andrew Vagin <avagin@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>

Signed-off-by: Andrew Vagin <avagin@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>

The criu(8) man-page is generated using asciidoc. The problem with
asciidoc is that, due to its dependencies, it is not available on
all distributions or it is undesired to install all asciidoc
dependencies. The install target was unconditionally installing and thus
building the man-page even if not explicitly specified with 'make docs'.
With the new 'install-criu' target everything besides the man-page is
installed and the target 'install-man' is only called by the target
'install'.
Signed-off-by: Adrian Reber <areber@redhat.com>
Acked-by: Tycho Andersen <tycho.andersen@canonical.com>
Looks-ok-to: Cyril Gorcunov <gorcunov@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>

On restore we will use the peer's name to connect() the
socket back, so if there's no name dump should be aborted.

This situation happens when we create a socketpair(), fork
and dump only one task with one pair end.
Reported-by: Artem Kuzmitskiy <artem.kuzmitskiy@gmail.com>
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>

Use SIG_SETMASK instead of SIG_BLOCKMASK here in case the parent had SIGCHLD
blocked. In this case if one of the criu threads has a problem, since the
SIGCHLD is blocked, the restore simply hangs.
Signed-off-by: Tycho Andersen <tycho.andersen@canonical.com>
Acked-by: Andrew Vagin <avagin@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>

Reported-by: criuport@gmail.com
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>

Signed-off-by: Pavel Emelyanov <xemul@parallels.com>

Signed-off-by: Pavel Emelyanov <xemul@parallels.com>
Reviewed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>

It's typo fix. SIGTERM generates code dump and it's a reason
why this test fails by timeout sometimes.
Signed-off-by: Andrew Vagin <avagin@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>

I've updated Mr Jenkins to fc22 and can't compile zdtm tests:
cc -g -O2 -Wall -Werror -U_FORTIFY_SOURCE -D_FORTIFY_SOURCE=0  -iquote ../../lib/arch/x86/include -I../../lib   ptrace_sig.c ../../lib/libzdtmtst.a   -o ptrace_sig
ptrace_sig.c: In function ‘child’:
ptrace_sig.c:33:5: error: type of ‘fd’ defaults to ‘int’ [-Werror=implicit-int]
 int child(fd)

Reported-by: Mr Jenkins
Signed-off-by: Andrew Vagin <avagin@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>

Signed-off-by: Salvatore Bonaccorso <carnil@debian.org>
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>

Signed-off-by: Salvatore Bonaccorso <carnil@debian.org>
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>

Otherwise we eventually get compiler warnings, ending up with the
build abort.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>

Passing both $DESTDIR and $PREFIX into --prefix option of crit install
results in the bogus $DESTDIR embedded in the installed object.
Instead, pass $DESTDIR to --root and $PREFIX to --prefix options,
respectively.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>

So, we've got quite a lot of new stuff this release :) The biggest
code-merge was POWER arch port. Another big thing that has happened
was criu integration with libcontainer. And a tiny change that will
become big next release -- the "criu show" command is going to be
thrown away in 3 monthes, the CRIT tool should be used instead.
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>

Signed-off-by: Pavel Emelyanov <xemul@parallels.com>
Acked-by: Andrew Vagin <avagin@virtuozzo.com>

Now it isn't clear which lock is not supported.
Signed-off-by: Andrey Vagin <avagin@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>

When we call open_proc(PROC_SELF, ...) the /proc/self descriptor is
cached in criu. If the process fork()-s after than and child goes
open_proc(PROC_SELF, ...) then it will get the parent's proc descriptor.
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>
Acked-by: Andrew Vagin <avagin@virtuozzo.com>

If dump doesn't generate data into some image file the respective file
is not created at all as it used to be in 1.5. This brings potential
problem -- if the file with the image name exists during dump (e.g. a
dump file left from previous, maybe unsuccessful, dump) then restore
would pick this file and read bad/wrong/fake data from it.

To return the proper behavior back the file with the name of empty image
should be removed.
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>

If a test fails, we save the criu directory and expect to have
all logs and images there.
Signed-off-by: Andrey Vagin <avagin@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>

Signed-off-by: Pavel Emelyanov <xemul@parallels.com>