matthias
/
iproute2
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

ip/netns: use flock when setting up /run/netns 

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>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
This commit is contained in:
Luca Boccassi 2020-11-27 18:06:51 +00:00 committed by Stephen Hemminger
parent fb054cb336
commit 975c4944e8
1 changed files with 35 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

35
ip/ipnetns.c
View File

@ -1,5 +1,6 @@
/* SPDX-License-Identifier: GPL-2.0 */ /* SPDX-License-Identifier: GPL-2.0 */
#define _ATFILE_SOURCE #define _ATFILE_SOURCE
#include <sys/file.h>
#include <sys/types.h> #include <sys/types.h>
#include <sys/stat.h> #include <sys/stat.h>
#include <sys/wait.h> #include <sys/wait.h>
@ -801,6 +802,7 @@ static int netns_add(int argc, char **argv, bool create)
const char *name; const char *name;
pid_t pid; pid_t pid;
int fd; int fd;
int lock;
int made_netns_run_dir_mount = 0; int made_netns_run_dir_mount = 0;
if (create) { 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 * namespace file in one namespace will unmount the network namespace
* file in all namespaces allowing the network namespace to be freed * file in all namespaces allowing the network namespace to be freed
* sooner. * 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)) { while (mount("", NETNS_RUN_DIR, "none", MS_SHARED | MS_REC, NULL)) {
/* Fail unless we need to make the mount point */ /* Fail unless we need to make the mount point */
if (errno != EINVAL || made_netns_run_dir_mount) { if (errno != EINVAL || made_netns_run_dir_mount) {
fprintf(stderr, "mount --make-shared %s failed: %s\n", fprintf(stderr, "mount --make-shared %s failed: %s\n",
NETNS_RUN_DIR, strerror(errno)); NETNS_RUN_DIR, strerror(errno));
if (lock != -1) {
flock(lock, LOCK_UN);
close(lock);
}
return -1; 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)) { if (mount(NETNS_RUN_DIR, NETNS_RUN_DIR, "none", MS_BIND | MS_REC, NULL)) {
fprintf(stderr, "mount --bind %s %s failed: %s\n", fprintf(stderr, "mount --bind %s %s failed: %s\n",
NETNS_RUN_DIR, NETNS_RUN_DIR, strerror(errno)); NETNS_RUN_DIR, NETNS_RUN_DIR, strerror(errno));
if (lock != -1) {
flock(lock, LOCK_UN);
close(lock);
}
return -1; return -1;
} }
made_netns_run_dir_mount = 1; made_netns_run_dir_mount = 1;
} }
if (lock != -1) {
flock(lock, LOCK_UN);
close(lock);
}
/* Create the filesystem state */ /* Create the filesystem state */
fd = open(netns_path, O_RDONLY|O_CREAT|O_EXCL, 0); fd = open(netns_path, O_RDONLY|O_CREAT|O_EXCL, 0);