Commit 65d8fc777f6d ("futex: Remove requirement for lock_page() in
get_futex_key()") removed an unnecessary lock_page() with the side-effect
that page->mapping needed to be treated very carefully. Two defensive
warnings were added in case any assumption was wrong with the first warning
assuming a correct application would not truncate a mapping backing an
active futex key. Since merging, it has not triggered for any unexpected
reason but Mark Rutland reported the following bug;
------------[ cut here ]------------
kernel BUG at kernel/futex.c:679!
Internal error: Oops - BUG: 0 [#1] PREEMPT SMP
Modules linked in:
CPU: 0 PID: 3695 Comm: syz-executor1 Not tainted 4.13.0-rc3-00020-g307fec773ba3 #3
Hardware name: linux,dummy-virt (DT)
task: ffff80001e271780 task.stack: ffff000010908000
PC is at get_futex_key+0x6a4/0xcf0 kernel/futex.c:679
LR is at get_futex_key+0x6a4/0xcf0 kernel/futex.c:679
pc : [<ffff00000821ac14>] lr : [<ffff00000821ac14>] pstate: 80000145
The fact that it's a bug instead of a warning was due to an unrelated issue
but the warning itself triggered because the underlying mapping changed. This
is an application issue but it's recoverable and the warning is unnecessary
so this patch removes the warning. The warning may be triggered with the
following test program from Mark although it may require NR_FUTEX_THREADS
to be less than the number of cores in the system.
#include <linux/futex.h>
#include <pthread.h>
#include <stdio.h>
#include <stdlib.h>
#include <sys/mman.h>
#include <sys/syscall.h>
#include <sys/time.h>
#include <unistd.h>
#define NR_FUTEX_THREADS 16
pthread_t threads[NR_FUTEX_THREADS];
void *mem;
#define MEM_PROT (PROT_READ | PROT_WRITE)
#define MEM_SIZE 65536
static int futex_wrapper(int *uaddr, int op, int val,
const struct timespec *timeout,
int *uaddr2, int val3)
{
syscall(SYS_futex, uaddr, op, val, timeout, uaddr2, val3);
}
void *poll_futex(void *unused)
{
for (;;) {
futex_wrapper(mem, FUTEX_CMP_REQUEUE_PI, 1, NULL, mem + 4, 1);
}
}
int main(int argc, char *argv[])
{
int i;
mem = mmap(NULL, MEM_SIZE, MEM_PROT,
MAP_SHARED | MAP_ANONYMOUS, -1, 0);
printf("Mapping @ %p\n", mem);
printf("Creating futex threads...\n");
for (i = 0; i < NR_FUTEX_THREADS; i++)
pthread_create(&threads[i], NULL, poll_futex, NULL);
printf("Flipping mapping...\n");
for (;;) {
mmap(mem, MEM_SIZE, MEM_PROT,
MAP_FIXED | MAP_SHARED | MAP_ANONYMOUS, -1, 0);
}
return 0;
}
Reported-by: Mark Rutland <[email protected]>
Signed-off-by: Mel Gorman <[email protected]>
Cc: [email protected] # 4.7+
---
kernel/futex.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/kernel/futex.c b/kernel/futex.c
index 16dbe4c93895..d0752185668d 100644
--- a/kernel/futex.c
+++ b/kernel/futex.c
@@ -676,7 +676,7 @@ get_futex_key(u32 __user *uaddr, int fshared, union futex_key *key, int rw)
* cases, therefore a successful atomic_inc return below will
* guarantee that get_futex_key() will still imply smp_mb(); (B).
*/
- if (WARN_ON_ONCE(!atomic_inc_not_zero(&inode->i_count))) {
+ if (!atomic_inc_not_zero(&inode->i_count)) {
rcu_read_unlock();
put_page(page);
On Wed, 09 Aug 2017, Mel Gorman wrote:
>@@ -676,7 +676,7 @@ get_futex_key(u32 __user *uaddr, int fshared, union futex_key *key, int rw)
> * cases, therefore a successful atomic_inc return below will
> * guarantee that get_futex_key() will still imply smp_mb(); (B).
You missed the comment above.
diff --git a/kernel/futex.c b/kernel/futex.c
index 16dbe4c93895..6b4a6a7cad3d 100644
--- a/kernel/futex.c
+++ b/kernel/futex.c
@@ -670,13 +670,14 @@ get_futex_key(u32 __user *uaddr, int fshared, union futex_key *key, int rw)
* this reference was taken by ihold under the page lock
* pinning the inode in place so i_lock was unnecessary. The
* only way for this check to fail is if the inode was
- * truncated in parallel so warn for now if this happens.
+ * truncated in parallel -- which is a bizarre scenario, in
+ * any case, just retry.
*
* We are not calling into get_futex_key_refs() in file-backed
* cases, therefore a successful atomic_inc return below will
* guarantee that get_futex_key() will still imply smp_mb(); (B).
*/
- if (WARN_ON_ONCE(!atomic_inc_not_zero(&inode->i_count))) {
+ if (!atomic_inc_not_zero(&inode->i_count)) {
rcu_read_unlock();
put_page(page);
Thanks,
Davidlohr