Add handeling of R_X86_64_GOTPCRELX and R_X86_64_REX_GOTPCRELX.
They are not that old, so I provided ifdef-guards for them.
According to x86-64 ABI specification paper, they should be
generated instead of R_X86_64_GOTPCREL for cases when relaxation
is possible.
At this moment we can handle them the same way like R_X86_64_GOTPCREL.
[0] https://github.com/hjl-tools/x86-psABI/wiki/x86-64-psABI-r249.pdfFixes: #397
Cc: Cyrill Gorcunov <gorcunov@openvz.org>
Reported-by: Adrian Reber <areber@redhat.com>
Signed-off-by: Dmitry Safonov <dima@arista.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
p + FAST_SIZE > buffer + length
In this sentence p + FAST_SIZE may be bigger than (1<<32),
and we will be in trouble.
$ gdb -c coredump test/zdtm/static/write_read01
(gdb) p p
$3 = (uint8_t *) 0xffffa89e
(gdb) p buffer
$4 = (uint8_t *) 0xfff06780
(gdb) p length
$5 = 1000000
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
The lazy-thp may fail with older kernels if khugepaded will be fast enough
to merge pages in the partially populated areas.
Signed-off-by: Mike Rapoport <rppt@linux.vnet.ibm.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
The CMSG_NXTHDR checks the length of the __next__ cmsg, i.e. the one
that it is about to return for filling. Thus, keeping there anything
but zeroes is unsafe.
Reported-by: Pavel Tikhomirov <snorcht@gmail.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
Most of the pieces has already been described in the previous patches :)
so here's the summary.
* Dump:
When receiving a message, also receive any SCM-s (already there) and when
SCM_RIGHTs one is met -- go ahead and just dump received descriptors using
regular code, but taking current as the victim task.
Few words about file paths resolution -- since we do dump path-ed files
by receiving them from victim's parasite, such files sent via sockets
should still work OK, as we still receive them, just from another socket.
Several problems here:
1. Unix sockets sent via unix sockets form knots. Not supported.
2. Eventpolls sent via unix might themseves poll unix sockets. Knots
again. Not supported either.
* Restore:
On restore we need to make unix socket wait for the soon-to-be-scm-sent
descriptors to get restored, so we need to find them, then put a dependency.
After that, the fake fdinfo entry is attached to the respective file
descs, when sent the respective descriptors are closed.
https://github.com/xemul/criu/issues/251
v2: Addressed comments from Kirill
* Moved prepare_scms before adding fake fles (with comment)
* Add scm-only fles as fake, thus removing close_scm_fds
* Try hard finding any suitable fle to use as scm one when
queuing them for unix socket scm list, only allocate a new
one if really needed
Reviewed-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
In order to make sender of a descriptor (unix socket) be restored
_after_ the descriptor in question, we need to find out those
sockets early (before post-prep calls).
The problem is that current code gives us info about who's the
queuer for who only in post-prep hooks, so the peer resolution
should happen right in collect callback.
At the same time we need to make sure that all peers configured
in the image exist, as well as need to put master/slave flags
for socketpairs. Both these actions can only happen in post-prep.
Said that -- the current peer resolve routine should be split
into two steps.
v2: Handle the socket connected to self (dgram)
Reviewed-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
The sk-queue.img keeps socket IDs as recepient descriptor. For
SCM code we'll need to find the unix sk info who is the sender
for a particular other socket by this ID. There's already the
queuer field in the unix_sk_info, but it needs to be an ID
rather than inode. Fortunately, id-inode pairs are unique.
Reviewed-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
For SCM restore we need to know all queuers, even if they are
socketpair-s, but current code skips this setting for one
half of them. So set this info for every socket out there.
Reviewed-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
In SCM code there will be the need to attach a newly created
FdinfoEntry to a particulat file desc object, then get the
created fdinfo helper object back.
Current code only allows this via two lookup calls -- first
one to attach entry to desc by ID (lookup #1), then get the
fdinfo helper by FD (lookup #2).
Fortunately, the exising code allows simple split that gives
us the optimized routine.
Reviewed-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
In SCM patch we'll need to generate temporary fdinfo entry
for a task that will send descriptors via sockets. Sometimes
the files in questions will NOT have any other fdinfo-s (if
they are SCM-sent and closed), so we need a helper that
would either give us existing fdinfo or say there's no such.
Reviewed-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
An aux code to dump the received file and report back its
ID and type.
Reviewed-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Actually all file-s we dump have unique IDs, regardless of
their types. This fact will be used to reduce complexity
of the SCM code -- instead of keeping TYPE:ID pair it'll
save only the ID.
Siad that -- we will need the way to lookup desc by ID only.
v2: Older images had fifo-s and tty-s having matching IDs
with respective reg-file entries
Reviewed-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
To support SCMs we'll need to receive them into criu task
(see the SCM patch for details), then dump the received
file as if it was in the dumpee. Then the info about received
descriptor will be written into packet entry.
For this we'll need to perform all the regular file dumping
code BUT not write the FdinfoEntry into image, so shuffle
the code for that.
The gist of the patch is in two changes -- one in the
do_dump_gen_file(), the other in dump_task_files_seized().
The rest is just tossing the arguments of the functions
relevant to that change.
Reviewed-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Same thing as for fifo-s.
Reviewed-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
The plan is to have all file entries have unique ID. Fifo
generates a reg file entry to reuse the whole reg-files
c/r-ing engine (ghosts, open-by-path, etc.) and right now
ID for this entry is the same as for fifo entry.
Reviewed-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
SCM-dumping code won't provide the ctl for TTY dump. To
make this work we'll have to peek required tty info from
current (see the SCM patch), but I doubt that anyone
sends TTYs via sockets ... so keep this in TODO list.
Reviewed-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Only the send code is altered, as upon receiving kernel
merges all scm_rights int one. CRIU relies on this merge
and this is to catch situations if the kernel suddenly
stops doing this.
Reviewed-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
This will be used in further to determine whether task's namespace
allows to restore file_desc.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Keep int pid with short-len fields to avoid
alignment gaps in structure.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
The algorhythm of searching file master will be more complicated
in next patches. So, move current functionality in separate func.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
The only user is fifo, but it uses collect_fd()
to do read only actions (the only change is
fifo_info::reg_d population, which may be easily
moved to open_fifo_fd()).
So, move the collect_special_file() functionality
from collect_fifo_fd() to open_fifo_fd() and
kill collect_fd method at all.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Declare it extern and allow to use outside files.c.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
On 01.06.2017 14:34, Kirill Tkhai wrote:
> Everything is prepared: the pair "type and id" is
> uniqul, so find_file_desc() in collect_fd() will
> work as expected and it will find the file_desc
> we just added in autofs_create_pipe()->collect_one_pipe_ops().
>
> So, the not-generic helper may be killed.
>
> Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Returned debug print:
[PATCH 4/5 v2]files: Kill autofs_create_fle() and use generic helper
Everything is prepared: the pair "type and id" is
uniqul, so find_file_desc() in collect_fd() will
work as expected and it will find the file_desc
we just added in autofs_create_pipe()->collect_one_pipe_ops().
So, the not-generic helper may be killed.
v2: Leave debug print on the place.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Acked-by: Stanislav Kinsburskiy <skinsbursky@virtuozzo.com>
Acked-by: Stanislav Kinsburskiy <skinsbursky@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
shmalloc() already returns address with sizeof(unsigned long)
alignment. So, the second alignment is useless.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Allow to mark some fles as "fake" and close them, when they are not needed.
v2: Cache fake file in separate list, when they are restored,
to do not iterate over files list again.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
It has only one user. Also, in next patches will be
used pstree_item_by_virt() function, and I don't
want to add "#include" to "files.h" for it, and to
nest headers.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Fake masters are files, which are not owned by
a process, that is able to open them.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Due to a misprint in an if the 6rd info for sit device
was restored when not present.
https://github.com/xemul/criu/issues/366
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Same here -- prepare the IFLA_INFO_DATA section using the
date from SitEntry.
Issue #11
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Nothing special here, just parse all known NLAs and keep them
on the image.
Issue #11
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
page_ids is a global variable and it is used to dump shared memroy,
process memory and ipc shared memory.
The problem is that an IPC namespace is dumped in a child process,
so page_ids isn't changed in a parent process. This patch is a quick
workaround this issue.
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Currently test_daemon() sends a signal to a parent process,
when a test completed a prepartion stage. But it doesn't work,
if a child changed uid and gid, because it doesn't have enough
permissions in this case.
This patch reworks a synchronization scheme to use a futex
instead of sending signals.
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Wait for thread to finish its creation instead of
calling sleep. Same time call for test_daemon
only when we're ready.
CC: Vitaly Ostrosablin <vostrosablin@virtuozzo.com>
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
For us it doesn't matter what names of users and groups are used,
we want to be sure that we restore right uid-s and gid-s.
If we use these functions, we can't run tests in namespaces and
they use an external unix socket, what can be a problem too.
sk unix: Runaway socket: ino 0x1df1ae4 peer_ino 0x1df1ae3 family 1 type 1 state 1 name /run/dbus/system_bus_socket
Error (criu/sk-unix.c:712): sk unix: External socket is used. Consider using --ext-unix-sk option.
Cc: Vitaly Ostrosablin <vostrosablin@virtuozzo.com>
Cc: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Acked-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
so that .gcda (coverage data) files are created with read-write
permissions for all, otherwise uns tests will not be able to
modify these files and code coverage data will not be collected
for them.
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
