2019-07-18 13:54:20

by Peter Zijlstra

Subject: [PATCH 1/4] locking/rwsem: Add missing ACQUIRE to read_slowpath exit when queue is empty

From: Jan Stancek <[email protected]>

LTP mtest06 has been observed to occasionally hit "still mapped when
deleted" and following BUG_ON on arm64.

The extra mapcount originated from pagefault handler, which handled
pagefault for vma that has already been detached. vma is detached
under mmap_sem write lock by detach_vmas_to_be_unmapped(), which
also invalidates vmacache.

When the pagefault handler (under mmap_sem read lock) calls
find_vma(), vmacache_valid() wrongly reports vmacache as valid.

After rwsem down_read() returns via 'queue empty' path (as of v5.2),
it does so without an ACQUIRE on sem->count:

down_read()
__down_read()
rwsem_down_read_failed()
__rwsem_down_read_failed_common()
raw_spin_lock_irq(&sem->wait_lock);
if (list_empty(&sem->wait_list)) {
if (atomic_long_read(&sem->count) >= 0) {
raw_spin_unlock_irq(&sem->wait_lock);
return sem;

The problem can be reproduced by running LTP mtest06 in a loop and
building the kernel (-j $NCPUS) in parallel. It does reproduces since
v4.20 on arm64 HPE Apollo 70 (224 CPUs, 256GB RAM, 2 nodes). It
triggers reliably in about an hour.

The patched kernel ran fine for 10+ hours.

Cc: [email protected]
Cc: [email protected]
Fixes: 4b486b535c33 ("locking/rwsem: Exit read lock slowpath if queue empty & no writer")
Signed-off-by: Jan Stancek <[email protected]>
Reviewed-by: Will Deacon <[email protected]>
Acked-by: Waiman Long <[email protected]>
Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
Link: https://lkml.kernel.org/r/50b8914e20d1d62bb2dee42d342836c2c16ebee7.1563438048.git.jstancek@redhat.com
---
kernel/locking/rwsem.c | 2 ++
1 file changed, 2 insertions(+)

--- a/kernel/locking/rwsem.c
+++ b/kernel/locking/rwsem.c
@@ -1032,6 +1032,8 @@ rwsem_down_read_slowpath(struct rw_semap
*/
if (adjustment && !(atomic_long_read(&sem->count) &
(RWSEM_WRITER_MASK | RWSEM_FLAG_HANDOFF))) {
+ /* Provide lock ACQUIRE */
+ smp_acquire__after_ctrl_dep();
raw_spin_unlock_irq(&sem->wait_lock);
rwsem_set_reader_owned(sem);
lockevent_inc(rwsem_rlock_fast);

2019-07-19 08:52:33

by Andrea Parri

Subject: Re: [PATCH 1/4] locking/rwsem: Add missing ACQUIRE to read_slowpath exit when queue is empty

> @@ -1032,6 +1032,8 @@ rwsem_down_read_slowpath(struct rw_semap
> */
> if (adjustment && !(atomic_long_read(&sem->count) &
> (RWSEM_WRITER_MASK | RWSEM_FLAG_HANDOFF))) {
> + /* Provide lock ACQUIRE */
> + smp_acquire__after_ctrl_dep();

Does this also make the lock RCtso? Or maybe RCtso was already
guaranteed (and I'm failing to see why)?

Thanks,
Andrea

2019-07-29 15:26:48

by Peter Zijlstra

Subject: Re: [PATCH 1/4] locking/rwsem: Add missing ACQUIRE to read_slowpath exit when queue is empty

On Fri, Jul 19, 2019 at 10:49:56AM +0200, Andrea Parri wrote:
> > @@ -1032,6 +1032,8 @@ rwsem_down_read_slowpath(struct rw_semap
> > */
> > if (adjustment && !(atomic_long_read(&sem->count) &
> > (RWSEM_WRITER_MASK | RWSEM_FLAG_HANDOFF))) {
> > + /* Provide lock ACQUIRE */
> > + smp_acquire__after_ctrl_dep();
>
> Does this also make the lock RCtso? Or maybe RCtso was already
> guaranteed (and I'm failing to see why)?

I didn't specifically look for that, but I suspect not, esp given the
next patch.