linux_dsm_epyc7002/kernel/locking
Ali Saidi 82fa9ced35 locking/qrwlock: Fix ordering in queued_write_lock_slowpath()
[ Upstream commit 84a24bf8c52e66b7ac89ada5e3cfbe72d65c1896 ]

While this code is executed with the wait_lock held, a reader can
acquire the lock without holding wait_lock.  The writer side loops
checking the value with the atomic_cond_read_acquire(), but only truly
acquires the lock when the compare-and-exchange is completed
successfully which isn’t ordered. This exposes the window between the
acquire and the cmpxchg to an A-B-A problem which allows reads
following the lock acquisition to observe values speculatively before
the write lock is truly acquired.

We've seen a problem in epoll where the reader does a xchg while
holding the read lock, but the writer can see a value change out from
under it.

  Writer                                | Reader
  --------------------------------------------------------------------------------
  ep_scan_ready_list()                  |
  |- write_lock_irq()                   |
      |- queued_write_lock_slowpath()   |
	|- atomic_cond_read_acquire()   |
				        | read_lock_irqsave(&ep->lock, flags);
     --> (observes value before unlock) |  chain_epi_lockless()
     |                                  |    epi->next = xchg(&ep->ovflist, epi);
     |                                  | read_unlock_irqrestore(&ep->lock, flags);
     |                                  |
     |     atomic_cmpxchg_relaxed()     |
     |-- READ_ONCE(ep->ovflist);        |

A core can order the read of the ovflist ahead of the
atomic_cmpxchg_relaxed(). Switching the cmpxchg to use acquire
semantics addresses this issue at which point the atomic_cond_read can
be switched to use relaxed semantics.

Fixes: b519b56e37 ("locking/qrwlock: Use atomic_cond_read_acquire() when spinning in qrwlock")
Signed-off-by: Ali Saidi <alisaidi@amazon.com>
[peterz: use try_cmpxchg()]
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Steve Capper <steve.capper@arm.com>
Acked-by: Will Deacon <will@kernel.org>
Acked-by: Waiman Long <longman@redhat.com>
Tested-by: Steve Capper <steve.capper@arm.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2021-04-28 13:40:00 +02:00
..
lock_events_list.h locking/rwsem: Adaptive disabling of reader optimistic spinning 2019-06-17 12:28:09 +02:00
lock_events.c locking/lock_events: Don't show pvqspinlock events on bare metal 2019-04-10 10:56:05 +02:00
lock_events.h locking/lock_events: Use raw_cpu_{add,inc}() for stats 2019-06-03 12:32:56 +02:00
lockdep_internals.h lockdep: Fix usage_traceoverflow 2020-10-09 08:53:08 +02:00
lockdep_proc.c locking/lockdep: Fix overflow in presentation of average lock-time 2020-07-25 21:47:42 +02:00
lockdep_states.h locking/lockdep: Rework FS_RECLAIM annotation 2017-08-10 12:29:03 +02:00
lockdep.c lockdep: Add a missing initialization hint to the "INFO: Trying to register non-static key" message 2021-04-21 13:00:51 +02:00
locktorture.c locktorture: Make function torture_percpu_rwsem_init() static 2020-08-24 18:45:32 -07:00
Makefile kcsan: Make KCSAN compatible with lockdep 2020-03-21 09:41:16 +01:00
mcs_spinlock.h locking/mcs: Use smp_cond_load_acquire() in MCS spin loop 2018-04-27 09:48:49 +02:00
mutex-debug.c lockdep: Introduce wait-type checks 2020-03-21 16:00:24 +01:00
mutex-debug.h License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
mutex.c locking/ww_mutex: Simplify use_ww_ctx & ww_ctx handling 2021-04-07 15:00:06 +02:00
mutex.h mutex: Fix up mutex_waiter usage 2019-08-08 09:09:25 +02:00
osq_lock.c locking/osq_lock: Annotate a data race in osq_lock 2020-06-29 12:04:48 -07:00
percpu-rwsem.c locking/percpu-rwsem: Use this_cpu_{inc,dec}() for read_count 2020-09-16 16:26:56 +02:00
qrwlock.c locking/qrwlock: Fix ordering in queued_write_lock_slowpath() 2021-04-28 13:40:00 +02:00
qspinlock_paravirt.h Revert "locking/pvqspinlock: Don't wait if vCPU is preempted" 2019-09-25 10:22:37 +02:00
qspinlock_stat.h treewide: Replace GPLv2 boilerplate/reference with SPDX - rule 157 2019-05-30 11:26:37 -07:00
qspinlock.c x86/kvm: Add "nopvspin" parameter to disable PV spinlocks 2020-07-08 16:21:57 -04:00
rtmutex_common.h rtmutex: Remove unused argument from rt_mutex_proxy_unlock() 2021-01-30 13:55:17 +01:00
rtmutex-debug.c kernel: rename show_stack_loglvl() => show_stack() 2020-06-09 09:39:13 -07:00
rtmutex-debug.h License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
rtmutex.c rtmutex: Remove unused argument from rt_mutex_proxy_unlock() 2021-01-30 13:55:17 +01:00
rtmutex.h License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
rwsem.c rwsem: Implement down_read_interruptible 2021-01-09 13:46:24 +01:00
rwsem.h locking/percpu-rwsem: Remove the embedded rwsem 2020-02-11 13:10:56 +01:00
semaphore.c treewide: Replace GPLv2 boilerplate/reference with SPDX - rule 436 2019-06-05 17:37:17 +02:00
spinlock_debug.c lockdep: Introduce wait-type checks 2020-03-21 16:00:24 +01:00
spinlock.c asm-generic/mmiowb: Add generic implementation of mmiowb() tracking 2019-04-08 11:59:39 +01:00
test-ww_mutex.c treewide: Replace GPLv2 boilerplate/reference with SPDX - rule 9 2019-05-21 11:28:40 +02:00