[CRIU] [PATCH 13/13] restore: collect signals from zombies
Andrey Vagin
avagin at openvz.org
Thu Dec 20 06:54:21 EST 2012
Each zombie sends SIGCHLD to parent. crtools restores all pending
signals, so all other signals should be collected.
Here is a problems, that signals SIGCHLD can be merged, but crtools
should be sure, that all signals are collected.
For that crtools enumerate all tasks and waits an answer. A parent
answers for its dead children from a signal handler.
This operation should be done between CR_STATE_RESTORE and
CR_STATE_RESTORE_SIGCHLD. Two new states CR_STATE_RESTORE_ZOMBIES
and CR_STATE_RESTORE_PREP_ZOMBIES were added.
CR_STATE_RESTORE_PREP_ZOMBIES is used for waiting when all tasks
installes a special sighandler.
CR_STATE_RESTORE_ZOMBIES is used for waiting all zombies.
Signed-off-by: Andrey Vagin <avagin at openvz.org>
---
cr-restore.c | 38 ++++++++++++++++++++++++++++++++++++--
include/crtools.h | 3 +++
include/restorer.h | 4 ++++
pie/restorer.c | 10 ++++++++++
pstree.c | 10 ++++++++++
5 files changed, 63 insertions(+), 2 deletions(-)
diff --git a/cr-restore.c b/cr-restore.c
index 127164d..1d5224e 100644
--- a/cr-restore.c
+++ b/cr-restore.c
@@ -684,7 +684,8 @@ static int restore_one_zombie(int pid, int exit_code)
if (task_entries != NULL) {
restore_finish_stage(CR_STATE_RESTORE);
zombie_prepare_signals();
- restore_finish_stage(CR_STATE_RESTORE_SIGCHLD);
+ restore_finish_stage(CR_STATE_RESTORE_PREP_ZOMBIES);
+ futex_wait_until(&task_entries->pid, pid);
}
if (exit_code & 0x7f) {
@@ -1119,8 +1120,11 @@ static inline int stage_participants(int next_stage)
case CR_STATE_RESTORE_PGID:
return task_entries->nr_tasks;
case CR_STATE_RESTORE:
- case CR_STATE_RESTORE_SIGCHLD:
+ case CR_STATE_RESTORE_ZOMBIES:
+ case CR_STATE_RESTORE_PREP_ZOMBIES:
return task_entries->nr_threads;
+ case CR_STATE_RESTORE_SIGCHLD:
+ return task_entries->nr_threads - atomic_get(&task_entries->nr_zombies);
}
BUG();
@@ -1146,6 +1150,7 @@ static int restore_root_task(struct pstree_item *init, struct cr_options *opts)
{
int ret;
struct sigaction act, old_act;
+ struct pstree_item *pi;
ret = sigaction(SIGCHLD, NULL, &act);
if (ret < 0) {
@@ -1201,6 +1206,33 @@ static int restore_root_task(struct pstree_item *init, struct cr_options *opts)
if (ret < 0)
goto out;
+ pr_info("Prepare for restoring zombies\n");
+ ret = restore_switch_stage(CR_STATE_RESTORE_PREP_ZOMBIES);
+ if (ret < 0)
+ goto out;
+
+ pr_info("Prepare for restoring zombies\n");
+ ret = restore_switch_stage(CR_STATE_RESTORE_ZOMBIES);
+ if (ret < 0)
+ goto out;
+
+ pr_info("Wait threads %d %d\n", task_entries->nr_threads, task_entries->nr_tasks);
+ futex_wait_while_gt(&task_entries->nr_in_progress, task_entries->nr_tasks);
+ ret = (int) futex_get(&task_entries->nr_in_progress);
+ if(ret < 0)
+ goto out;
+
+ for_each_pstree_item(pi) {
+ if (pi->state == TASK_HELPER)
+ continue;
+ ret = (int) futex_get(&task_entries->nr_in_progress);
+ if (ret < 0)
+ goto out;
+ futex_set_and_wake(&task_entries->pid, pi->pid.virt);
+ pr_info("Wait %d\n", pi->pid.virt);
+ futex_wait_while(&task_entries->nr_in_progress, ret);
+ }
+
pr_info("Wait until all tasks are restored\n");
ret = restore_switch_stage(CR_STATE_RESTORE_SIGCHLD);
if (ret < 0)
@@ -1252,6 +1284,8 @@ static int prepare_task_entries()
task_entries->nr_tasks = 0;
task_entries->nr_helpers = 0;
futex_set(&task_entries->start, CR_STATE_FORKING);
+ atomic_set(&task_entries->nr_zombies, 0);
+ futex_set(&task_entries->pid, 0);
return 0;
}
diff --git a/include/crtools.h b/include/crtools.h
index 702c59e..1681110 100644
--- a/include/crtools.h
+++ b/include/crtools.h
@@ -8,6 +8,7 @@
#include "list.h"
#include "util.h"
#include "image.h"
+#include "lock.h"
#include "../protobuf/vma.pb-c.h"
@@ -233,6 +234,8 @@ struct rst_info {
void *premmapped_addr;
unsigned long premmapped_len;
+
+ futex_t *nr_zombies;
};
static inline int in_vma_area(struct vma_area *vma, unsigned long addr)
diff --git a/include/restorer.h b/include/restorer.h
index f7a762d..d1d75c9 100644
--- a/include/restorer.h
+++ b/include/restorer.h
@@ -254,14 +254,18 @@ enum {
CR_STATE_FORKING,
CR_STATE_RESTORE_PGID,
CR_STATE_RESTORE,
+ CR_STATE_RESTORE_PREP_ZOMBIES,
+ CR_STATE_RESTORE_ZOMBIES,
CR_STATE_RESTORE_SIGCHLD,
CR_STATE_COMPLETE
};
struct task_entries {
int nr_threads, nr_tasks, nr_helpers;
+ atomic_t nr_zombies;
futex_t nr_in_progress;
futex_t start;
+ futex_t pid;
};
static always_inline struct shmem_info *
diff --git a/pie/restorer.c b/pie/restorer.c
index df96416..02ef2f1 100644
--- a/pie/restorer.c
+++ b/pie/restorer.c
@@ -51,6 +51,10 @@ extern void cr_restore_rt (void) asm ("__cr_restore_rt")
static void sigzombie_handler(int signal, siginfo_t *siginfo, void *data)
{
+ pr_debug("%ld: Collect a zombie with (pid %d, %d)\n",
+ sys_getpid(), siginfo->si_pid, (int) futex_get(&task_entries->pid));
+ BUG_ON(siginfo->si_pid != futex_get(&task_entries->pid));
+ futex_dec_and_wake(&task_entries->nr_in_progress);
}
static void sigchld_handler(int signal, siginfo_t *siginfo, void *data)
@@ -252,6 +256,8 @@ long __export_restore_thread(struct thread_restore_args *args)
pr_info("%ld: Restored\n", sys_gettid());
restore_finish_stage(CR_STATE_RESTORE);
+ restore_finish_stage(CR_STATE_RESTORE_PREP_ZOMBIES);
+ restore_finish_stage(CR_STATE_RESTORE_ZOMBIES);
if (restore_signals(args->siginfo, args->siginfo_nr, 0))
goto core_restore_end;
@@ -803,6 +809,10 @@ long __export_restore_task(struct task_restore_core_args *args)
act.rt_sa_restorer = cr_restore_rt;
sys_sigaction(SIGCHLD, &act, NULL, sizeof(rt_sigset_t));
+ restore_finish_stage(CR_STATE_RESTORE_PREP_ZOMBIES);
+ futex_wait_until(&task_entries->pid, my_pid);
+ restore_finish_stage(CR_STATE_RESTORE_ZOMBIES);
+
sys_sigaction(SIGCHLD, &args->sigchld_act, NULL, sizeof(rt_sigset_t));
ret = restore_signals(args->siginfo, args->siginfo_nr, 1);
diff --git a/pstree.c b/pstree.c
index 3ba645e..30e4074 100644
--- a/pstree.c
+++ b/pstree.c
@@ -37,6 +37,16 @@ struct pstree_item *__alloc_pstree_item(bool rst)
if (!item)
return NULL;
+ if (rst) {
+ item->rst->nr_zombies = shmalloc(sizeof(*item->rst->nr_zombies));
+ if (item->rst->nr_zombies == NULL) {
+ xfree(item);
+ return NULL;
+ }
+
+ futex_init(item->rst->nr_zombies);
+ }
+
INIT_LIST_HEAD(&item->children);
INIT_LIST_HEAD(&item->sibling);
--
1.7.11.7
More information about the CRIU
mailing list