user/sven/linux.git/kernel/locking/qspinlock.c, branch v4.9.235

locking/qspinlock, x86: Provide liveness guarantee

2018-12-21T13:11:34Z

commit 7aa54be2976550f17c11a1c3e3630002dea39303 upstream. On x86 we cannot do fetch_or() with a single instruction and thus end up using a cmpxchg loop, this reduces determinism. Replace the fetch_or() with a composite operation: tas-pending + load. Using two instructions of course opens a window we previously did not have. Consider the scenario: CPU0 CPU1 CPU2 1) lock trylock -> (0,0,1) 2) lock trylock /* fail */ 3) unlock -> (0,0,0) 4) lock trylock -> (0,0,1) 5) tas-pending -> (0,1,1) load-val <- (0,1,0) from 3 6) clear-pending-set-locked -> (0,0,1) FAIL: _2_ owners where 5) is our new composite operation. When we consider each part of the qspinlock state as a separate variable (as we can when _Q_PENDING_BITS == 8) then the above is entirely possible, because tas-pending will only RmW the pending byte, so the later load is able to observe prior tail and lock state (but not earlier than its own trylock, which operates on the whole word, due to coherence). To avoid this we need 2 things: - the load must come after the tas-pending (obviously, otherwise it can trivially observe prior state). - the tas-pending must be a full word RmW instruction, it cannot be an XCHGB for example, such that we cannot observe other state prior to setting pending. On x86 we can realize this by using "LOCK BTS m32, r32" for tas-pending followed by a regular load. Note that observing later state is not a problem: - if we fail to observe a later unlock, we'll simply spin-wait for that store to become visible. - if we observe a later xchg_tail(), there is no difference from that xchg_tail() having taken place before the tas-pending. Suggested-by: Will Deacon Reported-by: Thomas Gleixner Signed-off-by: Peter Zijlstra (Intel) Reviewed-by: Will Deacon Cc: Linus Torvalds Cc: Peter Zijlstra Cc: andrea.parri@amarulasolutions.com Cc: longman@redhat.com Fixes: 59fb586b4a07 ("locking/qspinlock: Remove unbounded cmpxchg() loop from locking slowpath") Link: https://lkml.kernel.org/r/20181003130957.183726335@infradead.org Signed-off-by: Ingo Molnar [bigeasy: GEN_BINARY_RMWcc macro redo] Signed-off-by: Sebastian Andrzej Siewior Signed-off-by: Sasha Levin

locking/qspinlock: Re-order code

2018-12-21T13:11:34Z

commit 53bf57fab7321fb42b703056a4c80fc9d986d170 upstream. Flip the branch condition after atomic_fetch_or_acquire(_Q_PENDING_VAL) such that we loose the indent. This also result in a more natural code flow IMO. Signed-off-by: Peter Zijlstra (Intel) Acked-by: Will Deacon Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Thomas Gleixner Cc: andrea.parri@amarulasolutions.com Cc: longman@redhat.com Link: https://lkml.kernel.org/r/20181003130257.156322446@infradead.org Signed-off-by: Ingo Molnar Signed-off-by: Sebastian Andrzej Siewior Signed-off-by: Sasha Levin

locking/qspinlock: Kill cmpxchg() loop when claiming lock from head of queue

2018-12-21T13:11:34Z

commit c61da58d8a9ba9238250a548f00826eaf44af0f7 upstream. When a queued locker reaches the head of the queue, it claims the lock by setting _Q_LOCKED_VAL in the lockword. If there isn't contention, it must also clear the tail as part of this operation so that subsequent lockers can avoid taking the slowpath altogether. Currently this is expressed as a cmpxchg() loop that practically only runs up to two iterations. This is confusing to the reader and unhelpful to the compiler. Rewrite the cmpxchg() loop without the loop, so that a failed cmpxchg() implies that there is contention and we just need to write to _Q_LOCKED_VAL without considering the rest of the lockword. Signed-off-by: Will Deacon Acked-by: Peter Zijlstra (Intel) Acked-by: Waiman Long Cc: Linus Torvalds Cc: Thomas Gleixner Cc: boqun.feng@gmail.com Cc: linux-arm-kernel@lists.infradead.org Cc: paulmck@linux.vnet.ibm.com Link: http://lkml.kernel.org/r/1524738868-31318-7-git-send-email-will.deacon@arm.com Signed-off-by: Ingo Molnar Signed-off-by: Sebastian Andrzej Siewior Signed-off-by: Sasha Levin

locking/qspinlock: Remove unbounded cmpxchg() loop from locking slowpath

2018-12-21T13:11:33Z

commit 59fb586b4a07b4e1a0ee577140ab4842ba451acd upstream. The qspinlock locking slowpath utilises a "pending" bit as a simple form of an embedded test-and-set lock that can avoid the overhead of explicit queuing in cases where the lock is held but uncontended. This bit is managed using a cmpxchg() loop which tries to transition the uncontended lock word from (0,0,0) -> (0,0,1) or (0,0,1) -> (0,1,1). Unfortunately, the cmpxchg() loop is unbounded and lockers can be starved indefinitely if the lock word is seen to oscillate between unlocked (0,0,0) and locked (0,0,1). This could happen if concurrent lockers are able to take the lock in the cmpxchg() loop without queuing and pass it around amongst themselves. This patch fixes the problem by unconditionally setting _Q_PENDING_VAL using atomic_fetch_or, and then inspecting the old value to see whether we need to spin on the current lock owner, or whether we now effectively hold the lock. The tricky scenario is when concurrent lockers end up queuing on the lock and the lock becomes available, causing us to see a lockword of (n,0,0). With pending now set, simply queuing could lead to deadlock as the head of the queue may not have observed the pending flag being cleared. Conversely, if the head of the queue did observe pending being cleared, then it could transition the lock from (n,0,0) -> (0,0,1) meaning that any attempt to "undo" our setting of the pending bit could race with a concurrent locker trying to set it. We handle this race by preserving the pending bit when taking the lock after reaching the head of the queue and leaving the tail entry intact if we saw pending set, because we know that the tail is going to be updated shortly. Signed-off-by: Will Deacon Acked-by: Peter Zijlstra (Intel) Acked-by: Waiman Long Cc: Linus Torvalds Cc: Thomas Gleixner Cc: boqun.feng@gmail.com Cc: linux-arm-kernel@lists.infradead.org Cc: paulmck@linux.vnet.ibm.com Link: http://lkml.kernel.org/r/1524738868-31318-6-git-send-email-will.deacon@arm.com Signed-off-by: Ingo Molnar Signed-off-by: Sebastian Andrzej Siewior Signed-off-by: Sasha Levin

locking/qspinlock: Merge 'struct __qspinlock' into 'struct qspinlock'

2018-12-21T13:11:33Z

commit 625e88be1f41b53cec55827c984e4a89ea8ee9f9 upstream. 'struct __qspinlock' provides a handy union of fields so that subcomponents of the lockword can be accessed by name, without having to manage shifts and masks explicitly and take endianness into account. This is useful in qspinlock.h and also potentially in arch headers, so move the 'struct __qspinlock' into 'struct qspinlock' and kill the extra definition. Signed-off-by: Will Deacon Acked-by: Peter Zijlstra (Intel) Acked-by: Waiman Long Acked-by: Boqun Feng Cc: Linus Torvalds Cc: Thomas Gleixner Cc: linux-arm-kernel@lists.infradead.org Cc: paulmck@linux.vnet.ibm.com Link: http://lkml.kernel.org/r/1524738868-31318-3-git-send-email-will.deacon@arm.com Signed-off-by: Ingo Molnar Signed-off-by: Sebastian Andrzej Siewior Signed-off-by: Sasha Levin

locking/qspinlock: Bound spinning on pending->locked transition in slowpath

2018-12-21T13:11:33Z

commit 6512276d97b160d90b53285bd06f7f201459a7e3 upstream. If a locker taking the qspinlock slowpath reads a lock value indicating that only the pending bit is set, then it will spin whilst the concurrent pending->locked transition takes effect. Unfortunately, there is no guarantee that such a transition will ever be observed since concurrent lockers could continuously set pending and hand over the lock amongst themselves, leading to starvation. Whilst this would probably resolve in practice, it means that it is not possible to prove liveness properties about the lock and means that lock acquisition time is unbounded. Rather than removing the pending->locked spinning from the slowpath altogether (which has been shown to heavily penalise a 2-threaded locking stress test on x86), this patch replaces the explicit spinning with a call to atomic_cond_read_relaxed and allows the architecture to provide a bound on the number of spins. For architectures that can respond to changes in cacheline state in their smp_cond_load implementation, it should be sufficient to use the default bound of 1. Suggested-by: Waiman Long Signed-off-by: Will Deacon Acked-by: Peter Zijlstra (Intel) Acked-by: Waiman Long Cc: Linus Torvalds Cc: Thomas Gleixner Cc: boqun.feng@gmail.com Cc: linux-arm-kernel@lists.infradead.org Cc: paulmck@linux.vnet.ibm.com Link: http://lkml.kernel.org/r/1524738868-31318-4-git-send-email-will.deacon@arm.com Signed-off-by: Ingo Molnar Signed-off-by: Sebastian Andrzej Siewior Signed-off-by: Sasha Levin

locking/qspinlock: Ensure node is initialised before updating prev->next

2018-12-21T13:11:33Z

commit 95bcade33a8af38755c9b0636e36a36ad3789fe6 upstream. When a locker ends up queuing on the qspinlock locking slowpath, we initialise the relevant mcs node and publish it indirectly by updating the tail portion of the lock word using xchg_tail. If we find that there was a pre-existing locker in the queue, we subsequently update their ->next field to point at our node so that we are notified when it's our turn to take the lock. This can be roughly illustrated as follows: /* Initialise the fields in node and encode a pointer to node in tail */ tail = initialise_node(node); /* * Exchange tail into the lockword using an atomic read-modify-write * operation with release semantics */ old = xchg_tail(lock, tail); /* If there was a pre-existing waiter ... */ if (old & _Q_TAIL_MASK) { prev = decode_tail(old); smp_read_barrier_depends(); /* ... then update their ->next field to point to node. WRITE_ONCE(prev->next, node); } The conditional update of prev->next therefore relies on the address dependency from the result of xchg_tail ensuring order against the prior initialisation of node. However, since the release semantics of the xchg_tail operation apply only to the write portion of the RmW, then this ordering is not guaranteed and it is possible for the CPU to return old before the writes to node have been published, consequently allowing us to point prev->next to an uninitialised node. This patch fixes the problem by making the update of prev->next a RELEASE operation, which also removes the reliance on dependency ordering. Signed-off-by: Will Deacon Acked-by: Peter Zijlstra (Intel) Cc: Linus Torvalds Cc: Thomas Gleixner Link: http://lkml.kernel.org/r/1518528177-19169-2-git-send-email-will.deacon@arm.com Signed-off-by: Ingo Molnar Signed-off-by: Sebastian Andrzej Siewior Signed-off-by: Sasha Levin

locking: Remove smp_read_barrier_depends() from queued_spin_lock_slowpath()

2018-12-21T13:11:33Z

commit 548095dea63ffc016d39c35b32c628d033638aca upstream. Queued spinlocks are not used by DEC Alpha, and furthermore operations such as READ_ONCE() and release/relaxed RMW atomics are being changed to imply smp_read_barrier_depends(). This commit therefore removes the now-redundant smp_read_barrier_depends() from queued_spin_lock_slowpath(), and adjusts the comments accordingly. Signed-off-by: Paul E. McKenney Cc: Peter Zijlstra Cc: Ingo Molnar Signed-off-by: Sebastian Andrzej Siewior Signed-off-by: Sasha Levin

locking/qspinlock: Ensure node->count is updated before initialising node

2018-05-30T05:50:26Z

[ Upstream commit 11dc13224c975efcec96647a4768a6f1bb7a19a8 ] When queuing on the qspinlock, the count field for the current CPU's head node is incremented. This needn't be atomic because locking in e.g. IRQ context is balanced and so an IRQ will return with node->count as it found it. However, the compiler could in theory reorder the initialisation of node[idx] before the increment of the head node->count, causing an IRQ to overwrite the initialised node and potentially corrupt the lock state. Avoid the potential for this harmful compiler reordering by placing a barrier() between the increment of the head node->count and the subsequent node initialisation. Signed-off-by: Will Deacon Acked-by: Peter Zijlstra (Intel) Cc: Linus Torvalds Cc: Thomas Gleixner Link: http://lkml.kernel.org/r/1518528177-19169-3-git-send-email-will.deacon@arm.com Signed-off-by: Ingo Molnar Signed-off-by: Sasha Levin Signed-off-by: Greg Kroah-Hartman

locking/qspinlock: Use __this_cpu_dec() instead of full-blown this_cpu_dec()

2016-06-27T09:37:41Z

queued_spin_lock_slowpath() should not worry about another queued_spin_lock_slowpath() running in interrupt context and changing node->count by accident, because node->count keeps the same value every time we enter/leave queued_spin_lock_slowpath(). On some architectures this_cpu_dec() will save/restore irq flags, which has high overhead. Use the much cheaper __this_cpu_dec() instead. Signed-off-by: Pan Xinhui Signed-off-by: Peter Zijlstra (Intel) Cc: Linus Torvalds Cc: Mike Galbraith Cc: Peter Zijlstra Cc: Thomas Gleixner Cc: Waiman.Long@hpe.com Link: http://lkml.kernel.org/r/1465886247-3773-1-git-send-email-xinhui.pan@linux.vnet.ibm.com [ Rewrote changelog. ] Signed-off-by: Ingo Molnar