@@ -1,5 +1,6 @@
/* SPDX-License-Identifier: GPL-2.0 */
#define _ATFILE_SOURCE
+#include <sys/file.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <sys/wait.h>
@@ -801,6 +802,7 @@ static int netns_add(int argc, char **argv, bool create)
const char *name;
pid_t pid;
int fd;
+ int lock;
int made_netns_run_dir_mount = 0;
if (create) {
@@ -831,12 +833,37 @@ static int netns_add(int argc, char **argv, bool create)
* namespace file in one namespace will unmount the network namespace
* file in all namespaces allowing the network namespace to be freed
* sooner.
+ * These setup steps need to happen only once, as if multiple ip processes
+ * try to attempt the same operation at the same time, the mountpoints will
+ * be recursively created multiple times, eventually causing the system
+ * to lock up. For example, this has been observed when multiple netns
+ * namespaces are created in parallel at boot. See:
+ * https://bugs.debian.org/949235
+ * Try to take an exclusive file lock on the top level directory to ensure
+ * this cannot happen, but proceed nonetheless if it cannot happen for any
+ * reason.
*/
+ lock = open(NETNS_RUN_DIR, O_RDONLY|O_DIRECTORY, 0);
+ if (lock < 0) {
+ fprintf(stderr, "Cannot open netns runtime directory \"%s\": %s\n",
+ NETNS_RUN_DIR, strerror(errno));
+ return -1;
+ }
+ if (flock(lock, LOCK_EX) < 0) {
+ fprintf(stderr, "Warning: could not flock netns runtime directory \"%s\": %s\n",
+ NETNS_RUN_DIR, strerror(errno));
+ close(lock);
+ lock = -1;
+ }
while (mount("", NETNS_RUN_DIR, "none", MS_SHARED | MS_REC, NULL)) {
/* Fail unless we need to make the mount point */
if (errno != EINVAL || made_netns_run_dir_mount) {
fprintf(stderr, "mount --make-shared %s failed: %s\n",
NETNS_RUN_DIR, strerror(errno));
+ if (lock != -1) {
+ flock(lock, LOCK_UN);
+ close(lock);
+ }
return -1;
}
@@ -844,10 +871,18 @@ static int netns_add(int argc, char **argv, bool create)
if (mount(NETNS_RUN_DIR, NETNS_RUN_DIR, "none", MS_BIND | MS_REC, NULL)) {
fprintf(stderr, "mount --bind %s %s failed: %s\n",
NETNS_RUN_DIR, NETNS_RUN_DIR, strerror(errno));
+ if (lock != -1) {
+ flock(lock, LOCK_UN);
+ close(lock);
+ }
return -1;
}
made_netns_run_dir_mount = 1;
}
+ if (lock != -1) {
+ flock(lock, LOCK_UN);
+ close(lock);
+ }
/* Create the filesystem state */
fd = open(netns_path, O_RDONLY|O_CREAT|O_EXCL, 0);
If multiple ip processes are ran at the same time to set up separate network namespaces, and it is the first time so /run/netns has to be set up first, and they end up doing it at the same time, the processes might enter a recursive loop creating thousands of mount points, which might crash the system depending on resources available. Try to take a flock on /run/netns before doing the mount() dance, to ensure this cannot happen. But do not try too hard, and if it fails continue after printing a warning, to avoid introducing regressions. First reported on Debian: https://bugs.debian.org/949235 To reproduce (WARNING: run in a VM to avoid system lockups): for i in {0..9} do strace -e trace=mount -e inject=mount:delay_exit=1000000 ip \ netns add "testnetns$i" 2>&1 | tee "$i.log" & done wait The strace is to ensure the problem always reproduces, to add an artificial synchronization point after the first mount(). Reported-by: Etienne Dechamps <etienne@edechamps.fr> Signed-off-by: Luca Boccassi <bluca@debian.org> --- ip/ipnetns.c | 35 +++++++++++++++++++++++++++++++++++ 1 file changed, 35 insertions(+)