Session 15's(20) leader is in first pidns, one it's process is in
second pidns and one is
in the third. So we create two helpers here for each aditional
pidns.
(It is critical that
Full test now looks like (mind pids here are different(real) from
their id's in source code e.g. 15 is 20 here):
(pid,ppid,sid)
session04(1, 0, 1)───session04(4, 1, 4)───session04(5, 4, 4)───session04(6, 5, 6,pid1)─┬─session04(8, 6, 8)───session04(9, 8, 7)
├─session04(10, 6, 6)───session04(11, 10, 11)
├─session04(13, 6, 13)───session04(14, 13, 11)
├─session04(15, 6, 15)
├─session04(17, 6, 17)─┬─session04(18, 17, 15)
│ └─session04(19, 17, 17,pid2)───session04(22, 19, 20)
├─session04(20, 6, 20)
└─session04(23, 6, 6,pid3)───session04(25, 23, 20)
Signed-off-by: Pavel Tikhomirov <ptikhomirov@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Demand ns_pid, ns_get_userns and ns_get_parent features, else will
have "Can't do ns ioctl" error in criu:set_ns_opt().
v2:remove unused variable i in cleanup
Signed-off-by: Pavel Tikhomirov <ptikhomirov@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Before "pstree: rework init reparent handling for pid namespaces" patch
we would get:
$ ./test/zdtm.py run -t zdtm/static/session01
=== Run 1/1 ================ zdtm/static/session01
======================= Run zdtm/static/session01 in ns ========================
Start test
./session01 --pidfile=session01.pid --outfile=session01.out
Run criu dump
Run criu restore
=[log]=> dump/zdtm/static/session01/31/1/restore.log
------------------------ grep Error ------------------------
(00.001103) 8 was born with sid 4
(00.001105) 7 was born with sid 4
(00.001106) 21 was born with sid 17
(00.001108) 1 was born with sid 17
(00.001109) Error (criu/pstree.c:1005): Can't find a session leader for 17
------------------------ ERROR OVER ------------------------
Corresponding tree before dump:
(combined 'pstree -pS 1' and 'ps axf -o pid,ppid,sid')
session01(1, 0, 1)─┬─session01(3, 1, 1)───session01(4, 3, 4)─┬─session01(5, 4, 5)─┬─session01(23, 5, 5)
│ │ ├─session01(24, 5, 5)
│ │ └─session01(26, 5, 5)
│ ├─session01(6, 4, 4)
│ ├─session01(7, 4, 7)───session01(16, 7, 4)
│ └─session01(8, 4, 8)───session01(15, 8, 15)───session01(20, 15, 4)
├─session01(12, 1, 12)───session01(17, 12, 17)───session01(18, 17, 18)───session01(27, 18, 4)
├─session01(13, 1, 10)
├─session01(14, 1, 4)
└─session01(21, 1, 21)───session01(22, 21, 17)
22 can not restore as it needs session 17, but 17-th's leader is not in
ancestors(21 had been reparented from 17; 12, 13 an 14 from 4).
Signed-off-by: Pavel Tikhomirov <ptikhomirov@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
These checks skip adding helpers and setting ids in case
of nested pid namespaces.
FIXME disable pgid, as it does not work yet
v2: add a comment near the added check for pgid
Signed-off-by: Pavel Tikhomirov <ptikhomirov@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
- Put code into new handle_init_reparent, make it pidns relative
and call it for each pidns.
- Consider the case when process tree branch(subtree) reparented to init
(parent of root of these branch died) riping some session in two
pieces and representative of these session in reparented branch can
not inherit its session if we simply try to fork the tree as is.
Patch adds helper can_inherit_sid to find such "adopted" brunches and
re-reparent them to helpers.
Previousely we had only direct children of init handled.
- We need many helpers for one session as:
1) The leader of session, if it is already dead, can not be recreated as
a helper in arbitrary pidns. But only in pidns ancestor of pidns of
any alive process of these session (sessions processes can't leave
pidns in which the session had been created).
More over session can be created only on proper level: sid array of the
alive process can end with several zerroes, meaning that after creation
of session, processes had entered several more pidnses, so we need to
cut these extra levels before creating the leader.
2) We can not re-reparent branch directly to session leader as the latter
can be in other pidns, thus create additional helper in our init's pidns,
and it's children will reparent to init.
If parents of session processes are in multiple pidnses we will need
helper per each such pidns, to be able to re-reparent them. See test
with setns for an example
- Collect all helper processes in separate list, so that it would
be easier to find them with get_helper_by_sid for other possibly
existing pieces of these sid. Branches re-reparented to such helpers
are temporary out of the tree and also skipped from walk over items
in for_each_pssubtree_item.
- Collect zombies and helpers which will reparent to init of pidns in
collect_child_pids to init of pidns instead of root task.
- The process tree which had only reparents to pidns init process
(no child subreapers reparents) will be restored fine). One tricky case
than we need re-reparent and the session leader is in same pidns with us
and our parent is in lower pid ns will fail - it happens than somebody
enters the pidns does setsid and then does clone(CLONE_PARENT).
v2: handle get_free_pids returns 0 as error
v3: rebase due to patchwork fail - use add_child_task and move_child_task
Signed-off-by: Pavel Tikhomirov <ptikhomirov@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
If process belonging to some session is in different pidns than leader
of these session, it will have zeroes on all aditional levels in sid,
so though levels for these process and leader does not match - sids do.
v2: change to static inline function as there is no more pr_err
Signed-off-by: Pavel Tikhomirov <ptikhomirov@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Need it to lookup adoptive children of pidns init. Also add
skip_descendants flag to be able to skip unneeded subtrees.
Signed-off-by: Pavel Tikhomirov <ptikhomirov@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
v2: handle get_free_pids returns 0 as error, remove unneeded iter var in
get_free_pids
Signed-off-by: Pavel Tikhomirov <ptikhomirov@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Create a child in new pid_ns; then the child creats thread and zombie.
Zombie is in the second created new pid_ns. Then the great parent
setns() to its active pid_ns. So, lets draw the table:
pid_ns vs pid_for_children_ns
great parent: equal
child: not equal
child thread: equal
grand child zombie: zombies don't have pid_for_children_ns
After signal chech that everything remains the same.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Threads may have different pid_for_children ns.
Allow them to set it after they are created:
just get a pid_ns fd from fdstore, and setns()
to it, after thread creation.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Restore it in dependence of thread numbers:
1)single-threaded -- before user_ns assignment
2)multi-threaded -- after thread creation (in next patch).
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
In next patches set_pid_for_children_ns() will be used
without pid, so move pid check out of function.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
No functional changes -- just to improve readability.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
It may differ from group leader's pid_for_children_ns,
so dump it separate.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Handle it as other namespaces, except of case of
just unshared pid ns w/o child reaper (fail if so),
and in case of we don't have it on restore (old
images -- then get pid_for_children ns from pid ns).
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Account the fact, that ns may be found by alternative
name, and use it if so.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
pid_for_children_ns of root_item may differ from its pid_ns.
In this case we don't want mark such pid_for_children_ns
as NS_ROOT in nsid_add().
Also, we don't want to create fake pid ns, if pid_for_children_ns
is a just create pid_ns without child reaper (kernel returns ENOENT
in this case). Better we later add a support to kernel to pick
such namespaces. So, we return UINT_MAX and fail.
We encode both the possibilities using the only "alternative"
parameter, as it's enough for us. If anybody needs to separate
them in the future and introduce separate parameters, it'd be rather
simple.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
It may be read from "/proc/[pid]/ns/pid_for_children_ns",
so add this alias to pid_ns_desc.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
This is need for "/proc/[pid]/ns/pid_for_children_ns",
which is a pid namespace with not standard file name.
Also pass "alternative" parameter to generate_ns_id().
v2: Enlarge buffer size
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Currently it's set in generate_ns_id() and it's equal to:
1)NS_CRIU, if root_ns_mask does not contain CLONE_NEWPID,
2)NS_ROOT, if it contains.
The assignment is not obvious, as it's set NS_CRIU firstly,
and then rewrites in NS_ROOT, if it exists.
Mark top_pid_ns after ns_ids population in more clear way.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Don't call destroy_pid_ns_helpers() twice; add a new case
for that.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Parent waits them by pid in its active pid namespace. So,
it's need to converts them there instead of using vpid().
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Child reaper of a ns have initial user_ns equal to its pid_ns->user_ns.
Keep all ns assignments together.
v2: Delete the assignment from call_clone_fn() and rename patch.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
1)Create a pid namespace and child reaper in it;
2)Set a specific next pid for future created process;
3)Create one more process in the namespace and kill it;
4)Wait for signal
5)Check, that NSpids of dead task remains the same.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Place child reapers of pid namespaces at the beginning
of pstree_item::children list and sort them by nesting
level.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Currently, one feature is supported. Add possibility
for a test to depend on several features.
v2: Delete excess "if" as suggested by Andrey Vagin.
Rename variables to decrise patch size.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Glibc has BUG with process creation:
https://sourceware.org/bugzilla/show_bug.cgi?id=21386
It doesn't behave well when parent and child are from
different pid namespaces and have the same pid.
Use raw syscall without glibc's asserts as workaround.
Also, use raw syscall for getpid() in tests too,
as these two function go in the pair (glibc's getpid()
relies on glibc's fork()).
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
When new rb_root is created for pidns it is initialized with
RB_ROOT, so ns->pid.rb_root.rb_node is NULL at first. Later
then insert first node in lookup_create_pid() to these rb-tree
it will have (NULL & color) in node->rb_parent_color.
So the check "!rb_parent(&found->ns[i].node)" will be true for
the rb-tree's root node, and criu will fail lookup these node.
We haven't hit that yet as to get to these check we need task in
at least two levels of pidns which at the same time is the root
in rb-tree on e.g. level 0.
Signed-off-by: Pavel Tikhomirov <ptikhomirov@virtuozzo.com>
Reviewed-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
This minimize chances to hit problem where files
used for page transfer are trying to use same number
reserved for service fd.
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Will need it to unlimit the files allocation
for service fd reserving and later for parasite code run
(which is implemented in vz7 instance and soon will be
ported into vanilla).
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Add a fake fd type for autofs. This allows functions
like find_file_desc() work as expected, without
having two different file_desc with the same type
and same id.
Also, later, it will allow to delete autofs_create_fle()
and to use generic helper.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
1)Use CLONE_VFORK to create subprocess, as it's safe after patch
"clone_noasan: Allow to create CLONE_VM|CLONE_VFORK processe".
2)add more CLONE_XXX to flags to speedup the syscall.
3)Do not send SIGCHLD, as parent sees child's exit() synchronuos anyway.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Picked from patch "[PATCH RFC] namespaces: use CLONE_VFORK
with CLONE_VM when it is possible" by Andrew Vagin.
Currenly parent touches child's stack, as in moment of clone() call
its stack pointer is above the child's (we allocate char stack[128]
on parent's stack). This prevents to create CLONE_VM|CLONE_VFORK
processes, because the child uses stack addresses occupied by parent.
The patch changes clone_noasan() behaviour and allows to do that
with the same memory consumption. We give a child memory, which
is not used by parent clone(), so parent's and child's stacks
have no tntersection.
This allows to create CLONE_VM|CLONE_VFORK processes.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Move switch_ns() down because __pstree_pid_by_virt()
does not need cleanup.
Add more goto labels and restore ns back in case of fail.
Also delete pr_err(), because the error is already printed
by request_set_next_pid().
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Before this patch we used flock to order task creation,
but this way is not good. It took 5 syscalls to synchronize
a creation of a single child:
1)open()
2)flock(LOCK_EX)
3)flock(LOCK_UN)
4)close() in parent
5)close() in child
The patch introduces more effective way for synchronization,
which executes 2 syscalls only. We use last_pid_mutex,
and the syscalls number sounds definitely better.
v2: Don't use flock() at all
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Group them for 1)error and 2)parent cases. This minimize the code
and will be used in next patches.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
It's impossible to create a task from a pid_ns if its helper
is not created, because we wait in wait_pid_ns_helper_prepared()
for that. So, such situation here is a bug.
Move the wait and convert it to BUG().
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
Get pid_ns fd from INIT_PID task of this namespace and
use switch_ns() and restore_ns().
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
It must sound do_destroy_pid_ns_helpers() with *s*.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
In next patches, root_item will need to have its real pid
to be sure, usernsd already sees it.
Also add a comment, explaning why set real pid in two places.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
We never call this function for root_item.
It's for dropping user ns, which may happen
with the rest of tasks only.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Andrei Vagin <avagin@virtuozzo.com>
