If a spinner is present, there is a chance that the load of
rwsem_has_spinner() in rwsem_wake() can be reordered with
respect to decrement of rwsem count in __up_write() leading
to wakeup being missed:
spinning writer up_write caller
--------------- -----------------------
[S] osq_unlock() [L] osq
spin_lock(wait_lock)
sem->count=0xFFFFFFFF00000001
+0xFFFFFFFF00000000
count=sem->count
MB
sem->count=0xFFFFFFFE00000001
-0xFFFFFFFF00000001
spin_trylock(wait_lock)
return
rwsem_try_write_lock(count)
spin_unlock(wait_lock)
schedule()
Reordering of atomic_long_sub_return_release() in __up_write()
and rwsem_has_spinner() in rwsem_wake() can cause missing of
wakeup in up_write() context. In spinning writer, sem->count
and local variable count is 0XFFFFFFFE00000001. It would result
in rwsem_try_write_lock() failing to acquire rwsem and spinning
writer going to sleep in rwsem_down_write_failed().
The smp_rmb() will make sure that the spinner state is
consulted after sem->count is updated in up_write context.
Signed-off-by: Prateek Sood <prsood@codeaurora.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: dave@stgolabs.net
Cc: longman@redhat.com
Cc: parri.andrea@gmail.com
Cc: sramana@codeaurora.org
Link: http://lkml.kernel.org/r/1504794658-15397-1-git-send-email-prsood@codeaurora.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
GFP_TEMPORARY was introduced by commit e12ba74d8f ("Group short-lived
and reclaimable kernel allocations") along with __GFP_RECLAIMABLE. It's
primary motivation was to allow users to tell that an allocation is
short lived and so the allocator can try to place such allocations close
together and prevent long term fragmentation. As much as this sounds
like a reasonable semantic it becomes much less clear when to use the
highlevel GFP_TEMPORARY allocation flag. How long is temporary? Can the
context holding that memory sleep? Can it take locks? It seems there is
no good answer for those questions.
The current implementation of GFP_TEMPORARY is basically GFP_KERNEL |
__GFP_RECLAIMABLE which in itself is tricky because basically none of
the existing caller provide a way to reclaim the allocated memory. So
this is rather misleading and hard to evaluate for any benefits.
I have checked some random users and none of them has added the flag
with a specific justification. I suspect most of them just copied from
other existing users and others just thought it might be a good idea to
use without any measuring. This suggests that GFP_TEMPORARY just
motivates for cargo cult usage without any reasoning.
I believe that our gfp flags are quite complex already and especially
those with highlevel semantic should be clearly defined to prevent from
confusion and abuse. Therefore I propose dropping GFP_TEMPORARY and
replace all existing users to simply use GFP_KERNEL. Please note that
SLAB users with shrinkers will still get __GFP_RECLAIMABLE heuristic and
so they will be placed properly for memory fragmentation prevention.
I can see reasons we might want some gfp flag to reflect shorterm
allocations but I propose starting from a clear semantic definition and
only then add users with proper justification.
This was been brought up before LSF this year by Matthew [1] and it
turned out that GFP_TEMPORARY really doesn't have a clear semantic. It
seems to be a heuristic without any measured advantage for most (if not
all) its current users. The follow up discussion has revealed that
opinions on what might be temporary allocation differ a lot between
developers. So rather than trying to tweak existing users into a
semantic which they haven't expected I propose to simply remove the flag
and start from scratch if we really need a semantic for short term
allocations.
[1] http://lkml.kernel.org/r/20170118054945.GD18349@bombadil.infradead.org
[akpm@linux-foundation.org: fix typo]
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: drm/i915: fix up]
Link: http://lkml.kernel.org/r/20170816144703.378d4f4d@canb.auug.org.au
Link: http://lkml.kernel.org/r/20170728091904.14627-1-mhocko@kernel.org
Signed-off-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Acked-by: Mel Gorman <mgorman@suse.de>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Neil Brown <neilb@suse.de>
Cc: "Theodore Ts'o" <tytso@mit.edu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Pull locking updates from Ingo Molnar:
- Add 'cross-release' support to lockdep, which allows APIs like
completions, where it's not the 'owner' who releases the lock, to be
tracked. It's all activated automatically under
CONFIG_PROVE_LOCKING=y.
- Clean up (restructure) the x86 atomics op implementation to be more
readable, in preparation of KASAN annotations. (Dmitry Vyukov)
- Fix static keys (Paolo Bonzini)
- Add killable versions of down_read() et al (Kirill Tkhai)
- Rework and fix jump_label locking (Marc Zyngier, Paolo Bonzini)
- Rework (and fix) tlb_flush_pending() barriers (Peter Zijlstra)
- Remove smp_mb__before_spinlock() and convert its usages, introduce
smp_mb__after_spinlock() (Peter Zijlstra)
* 'locking-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (56 commits)
locking/lockdep/selftests: Fix mixed read-write ABBA tests
sched/completion: Avoid unnecessary stack allocation for COMPLETION_INITIALIZER_ONSTACK()
acpi/nfit: Fix COMPLETION_INITIALIZER_ONSTACK() abuse
locking/pvqspinlock: Relax cmpxchg's to improve performance on some architectures
smp: Avoid using two cache lines for struct call_single_data
locking/lockdep: Untangle xhlock history save/restore from task independence
locking/refcounts, x86/asm: Disable CONFIG_ARCH_HAS_REFCOUNT for the time being
futex: Remove duplicated code and fix undefined behaviour
Documentation/locking/atomic: Finish the document...
locking/lockdep: Fix workqueue crossrelease annotation
workqueue/lockdep: 'Fix' flush_work() annotation
locking/lockdep/selftests: Add mixed read-write ABBA tests
mm, locking/barriers: Clarify tlb_flush_pending() barriers
locking/lockdep: Make CONFIG_LOCKDEP_CROSSRELEASE and CONFIG_LOCKDEP_COMPLETIONS truly non-interactive
locking/lockdep: Explicitly initialize wq_barrier::done::map
locking/lockdep: Rename CONFIG_LOCKDEP_COMPLETE to CONFIG_LOCKDEP_COMPLETIONS
locking/lockdep: Reword title of LOCKDEP_CROSSRELEASE config
locking/lockdep: Make CONFIG_LOCKDEP_CROSSRELEASE part of CONFIG_PROVE_LOCKING
locking/refcounts, x86/asm: Implement fast refcount overflow protection
locking/lockdep: Fix the rollback and overwrite detection logic in crossrelease
...
All the locking related cmpxchg's in the following functions are
replaced with the _acquire variants:
- pv_queued_spin_steal_lock()
- trylock_clear_pending()
This change should help performance on architectures that use LL/SC.
The cmpxchg in pv_kick_node() is replaced with a relaxed version
with explicit memory barrier to make sure that it is fully ordered
in the writing of next->lock and the reading of pn->state whether
the cmpxchg is a success or failure without affecting performance in
non-LL/SC architectures.
On a 2-socket 12-core 96-thread Power8 system with pvqspinlock
explicitly enabled, the performance of a locking microbenchmark
with and without this patch on a 4.13-rc4 kernel with Xinhui's PPC
qspinlock patch were as follows:
# of thread w/o patch with patch % Change
----------- --------- ---------- --------
8 5054.8 Mop/s 5209.4 Mop/s +3.1%
16 3985.0 Mop/s 4015.0 Mop/s +0.8%
32 2378.2 Mop/s 2396.0 Mop/s +0.7%
Suggested-by: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Waiman Long <longman@redhat.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Andrea Parri <parri.andrea@gmail.com>
Cc: Boqun Feng <boqun.feng@gmail.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Pan Xinhui <xinhui@linux.vnet.ibm.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Will Deacon <will.deacon@arm.com>
Link: http://lkml.kernel.org/r/1502741222-24360-1-git-send-email-longman@redhat.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
The new completion/crossrelease annotations interact unfavourable with
the extant flush_work()/flush_workqueue() annotations.
The problem is that when a single work class does:
wait_for_completion(&C)
and
complete(&C)
in different executions, we'll build dependencies like:
lock_map_acquire(W)
complete_acquire(C)
and
lock_map_acquire(W)
complete_release(C)
which results in the dependency chain: W->C->W, which lockdep thinks
spells deadlock, even though there is no deadlock potential since
works are ran concurrently.
One possibility would be to change the work 'lock' to recursive-read,
but that would mean hitting a lockdep limitation on recursive locks.
Also, unconditinoally switching to recursive-read here would fail to
detect the actual deadlock on single-threaded workqueues, which do
have a problem with this.
For now, forcefully disregard these locks for crossrelease.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Tejun Heo <tj@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: boqun.feng@gmail.com
Cc: byungchul.park@lge.com
Cc: david@fromorbit.com
Cc: johannes@sipsolutions.net
Cc: oleg@redhat.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
No acquisition might be in progress on commit of a crosslock. Completion
operations enabling crossrelease are the case like:
CONTEXT X CONTEXT Y
--------- ---------
trigger completion context
complete AX
commit AX
wait_for_complete AX
acquire AX
wait
where AX is a crosslock.
When no acquisition is in progress, we should not perform commit because
the lock does not exist, which might cause incorrect memory access. So
we have to track the number of acquisitions of a crosslock to handle it.
Moreover, in case that more than one acquisition of a crosslock are
overlapped like:
CONTEXT W CONTEXT X CONTEXT Y CONTEXT Z
--------- --------- --------- ---------
acquire AX (gen_id: 1)
acquire A
acquire AX (gen_id: 10)
acquire B
commit AX
acquire C
commit AX
where A, B and C are typical locks and AX is a crosslock.
Current crossrelease code performs commits in Y and Z with gen_id = 10.
However, we can use gen_id = 1 to do it, since not only 'acquire AX in X'
but 'acquire AX in W' also depends on each acquisition in Y and Z until
their commits. So make it use gen_id = 1 instead of 10 on their commits,
which adds an additional dependency 'AX -> A' in the example above.
Signed-off-by: Byungchul Park <byungchul.park@lge.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: akpm@linux-foundation.org
Cc: boqun.feng@gmail.com
Cc: kernel-team@lge.com
Cc: kirill@shutemov.name
Cc: npiggin@gmail.com
Cc: walken@google.com
Cc: willy@infradead.org
Link: http://lkml.kernel.org/r/1502089981-21272-8-git-send-email-byungchul.park@lge.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Two boots + a make defconfig, the first didn't have the redundant bit
in, the second did:
lock-classes: 1168 1169 [max: 8191]
direct dependencies: 7688 5812 [max: 32768]
indirect dependencies: 25492 25937
all direct dependencies: 220113 217512
dependency chains: 9005 9008 [max: 65536]
dependency chain hlocks: 34450 34366 [max: 327680]
in-hardirq chains: 55 51
in-softirq chains: 371 378
in-process chains: 8579 8579
stack-trace entries: 108073 88474 [max: 524288]
combined max dependencies: 178738560 169094640
max locking depth: 15 15
max bfs queue depth: 320 329
cyclic checks: 9123 9190
redundant checks: 5046
redundant links: 1828
find-mask forwards checks: 2564 2599
find-mask backwards checks: 39521 39789
So it saves nearly 2k links and a fair chunk of stack-trace entries, but
as expected, makes no real difference on the indirect dependencies.
At the same time, you see the max BFS depth increase, which is also
expected, although it could easily be boot variance -- these numbers are
not entirely stable between boots.
The down side is that the cycles in the graph become larger and thus
the reports harder to read.
XXX: do we want this as a CONFIG variable, implied by LOCKDEP_SMALL?
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Byungchul Park <byungchul.park@lge.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Nikolay Borisov <nborisov@suse.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: akpm@linux-foundation.org
Cc: boqun.feng@gmail.com
Cc: iamjoonsoo.kim@lge.com
Cc: kernel-team@lge.com
Cc: kirill@shutemov.name
Cc: npiggin@gmail.com
Cc: walken@google.com
Link: http://lkml.kernel.org/r/20170303091338.GH6536@twins.programming.kicks-ass.net
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Fix ordering of link creation between node->prev and prev->next in
osq_lock(). A case in which the status of optimistic spin queue is
CPU6->CPU2 in which CPU6 has acquired the lock.
tail
v
,-. <- ,-.
|6| |2|
`-' -> `-'
At this point if CPU0 comes in to acquire osq_lock, it will update the
tail count.
CPU2 CPU0
----------------------------------
tail
v
,-. <- ,-. ,-.
|6| |2| |0|
`-' -> `-' `-'
After tail count update if CPU2 starts to unqueue itself from
optimistic spin queue, it will find an updated tail count with CPU0 and
update CPU2 node->next to NULL in osq_wait_next().
unqueue-A
tail
v
,-. <- ,-. ,-.
|6| |2| |0|
`-' `-' `-'
unqueue-B
->tail != curr && !node->next
If reordering of following stores happen then prev->next where prev
being CPU2 would be updated to point to CPU0 node:
tail
v
,-. <- ,-. ,-.
|6| |2| |0|
`-' `-' -> `-'
osq_wait_next()
node->next <- 0
xchg(node->next, NULL)
tail
v
,-. <- ,-. ,-.
|6| |2| |0|
`-' `-' `-'
unqueue-C
At this point if next instruction
WRITE_ONCE(next->prev, prev);
in CPU2 path is committed before the update of CPU0 node->prev = prev then
CPU0 node->prev will point to CPU6 node.
tail
v----------. v
,-. <- ,-. ,-.
|6| |2| |0|
`-' `-' `-'
`----------^
At this point if CPU0 path's node->prev = prev is committed resulting
in change of CPU0 prev back to CPU2 node. CPU2 node->next is NULL
currently,
tail
v
,-. <- ,-. <- ,-.
|6| |2| |0|
`-' `-' `-'
`----------^
so if CPU0 gets into unqueue path of osq_lock it will keep spinning
in infinite loop as condition prev->next == node will never be true.
Signed-off-by: Prateek Sood <prsood@codeaurora.org>
[ Added pictures, rewrote comments. ]
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: sramana@codeaurora.org
Link: http://lkml.kernel.org/r/1500040076-27626-1-git-send-email-prsood@codeaurora.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Pull locking fixlet from Ingo Molnar:
"Remove an unnecessary priority adjustment in the rtmutex code"
* 'locking-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
locking/rtmutex: Remove unnecessary priority adjustment
