Xunlei Pang [Thu, 23 Mar 2017 14:56:09 +0000 (15:56 +0100)]
sched/deadline/rtmutex: Dont miss the dl_runtime/dl_period update
Currently dl tasks will actually return at the very beginning
of rt_mutex_adjust_prio_chain() in !detect_deadlock cases:
if (waiter->prio == task->prio) {
if (!detect_deadlock)
goto out_unlock_pi; // out here
else
requeue = false;
}
As the deadline value of blocked deadline tasks(waiters) without
changing their sched_class(thus prio doesn't change) never changes,
this seems reasonable, but it actually misses the chance of updating
rt_mutex_waiter's "dl_runtime(period)_copy" if a waiter updates its
deadline parameters(dl_runtime, dl_period) or boosted waiter changes
to !deadline class.
Thus, force deadline task not out by adding the !dl_prio() condition.
This is because rt_mutex_enqueue_pi() and rt_mutex_dequeue_pi()
are only protected by pi_lock when operating pi waiters, while
rt_mutex_get_top_task(), will access them with rq lock held but
not holding pi_lock.
In order to tackle it, we introduce new "pi_top_task" pointer
cached in task_struct, and add new rt_mutex_update_top_task()
to update its value, it can be called by rt_mutex_setprio()
which held both owner's pi_lock and rq lock. Thus "pi_top_task"
can be safely accessed by enqueue_task_dl() under rq lock.
Originally-From: Peter Zijlstra <peterz@infradead.org> Signed-off-by: Xunlei Pang <xlpang@redhat.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Acked-by: Steven Rostedt <rostedt@goodmis.org> Reviewed-by: Thomas Gleixner <tglx@linutronix.de> Cc: juri.lelli@arm.com Cc: bigeasy@linutronix.de Cc: mathieu.desnoyers@efficios.com Cc: jdesfossez@efficios.com Cc: bristot@redhat.com Link: http://lkml.kernel.org/r/20170323150216.157682758@infradead.org Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Xunlei Pang [Thu, 23 Mar 2017 14:56:07 +0000 (15:56 +0100)]
rtmutex: Deboost before waking up the top waiter
We should deboost before waking the high-priority task, such that we
don't run two tasks with the same "state" (priority, deadline,
sched_class, etc).
In order to make sure the boosting task doesn't start running between
unlock and deboost (due to 'spurious' wakeup), we move the deboost
under the wait_lock, that way its serialized against the wait loop in
__rt_mutex_slowlock().
Doing the deboost early can however lead to priority-inversion if
current would get preempted after the deboost but before waking our
high-prio task, hence we disable preemption before doing deboost, and
enabling it after the wake up is over.
This gets us the right semantic order, but most importantly however;
this change ensures pointer stability for the next patch, where we
have rt_mutex_setprio() cache a pointer to the top-most waiter task.
If we, as before this change, do the wakeup first and then deboost,
this pointer might point into thin air.
[peterz: Changelog + patch munging] Suggested-by: Peter Zijlstra <peterz@infradead.org> Signed-off-by: Xunlei Pang <xlpang@redhat.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Acked-by: Steven Rostedt <rostedt@goodmis.org> Cc: juri.lelli@arm.com Cc: bigeasy@linutronix.de Cc: mathieu.desnoyers@efficios.com Cc: jdesfossez@efficios.com Cc: bristot@redhat.com Link: http://lkml.kernel.org/r/20170323150216.110065320@infradead.org Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Yuyang Du [Sun, 12 Feb 2017 21:44:23 +0000 (05:44 +0800)]
sched/fair: Optimize ___update_sched_avg()
The main PELT function ___update_load_avg(), which implements the
accumulation and progression of the geometric average series, is
implemented along the following lines for the scenario where the time
delta spans all 3 possible sections (see figure below):
1. add the remainder of the last incomplete period
2. decay old sum
3. accumulate new sum in full periods since last_update_time
4. accumulate the current incomplete period
5. update averages
d1 - is the delta part completing the remainder of the last
incomplete period,
d2 - is the delta part spannind complete periods, and
d3 - is the delta part starting the current incomplete period.
We can simplify the code in two steps; the first step is to separate
the first term into new and old parts like:
The __update_load_avg() function is an __always_inline because its
used with constant propagation to generate different variants of the
code without having to duplicate it (which would be prone to bugs).
Explicitly instantiate the 3 variants.
Note that most of this is called from rather hot paths, so reducing
branches is good.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
Dmitry noted that the new atomic_try_cmpxchg() primitive is broken when
the old pointer doesn't point to the local stack.
He writes:
"Consider a classical lock-free stack push:
node->next = atomic_read(&head);
do {
} while (!atomic_try_cmpxchg(&head, &node->next, node));
This code is broken with the current implementation, the problem is
with unconditional update of *__po.
In case of success it writes the same value back into *__po, but in
case of cmpxchg success we might have lose ownership of some memory
locations and potentially over what __po has pointed to. The same
holds for the re-read of *__po. "
He also points out that this makes it surprisingly different from the
similar C/C++ atomic operation.
After investigating the code-gen differences caused by this patch; and
a number of alternatives (Linus dislikes this interface lots), we
arrived at these results (size x86_64-defconfig/vmlinux):
sched/fair: Prefer sibiling only if local group is under-utilized
If the child domain prefers tasks to go siblings, the local group could
end up pulling tasks to itself even if the local group is almost equally
loaded as the source group.
Lets assume a 4 core,smt==2 machine running 5 thread ebizzy workload.
Everytime, local group has capacity and source group has atleast 2 threads,
local group tries to pull the task. This causes the threads to constantly
move between different cores. This is even more profound if the cores have
more threads, like in Power 8, smt 8 mode.
Fix this by only allowing local group to pull a task, if the source group
has more number of tasks than the local group.
Here are the relevant perf stat numbers of a 22 core,smt 8 Power 8 machine.
Without patch:
Performance counter stats for 'ebizzy -t 22 -S 100' (5 runs):
Peter Zijlstra [Mon, 20 Mar 2017 11:26:55 +0000 (12:26 +0100)]
lockdep: Fix per-cpu static objects
Since commit 383776fa7527 ("locking/lockdep: Handle statically initialized
PER_CPU locks properly") we try to collapse per-cpu locks into a single
class by giving them all the same key. For this key we choose the canonical
address of the per-cpu object, which would be the offset into the per-cpu
area.
This has two problems:
- there is a case where we run !0 lock->key through static_obj() and
expect this to pass; it doesn't for canonical pointers.
- 0 is a valid canonical address.
Cure both issues by redefining the canonical address as the address of the
per-cpu variable on the boot CPU.
Since I didn't want to rely on CPU0 being the boot-cpu, or even existing at
all, track the boot CPU in a variable.
Fixes: 383776fa7527 ("locking/lockdep: Handle statically initialized PER_CPU locks properly") Reported-by: kernel test robot <fengguang.wu@intel.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Tested-by: Borislav Petkov <bp@suse.de> Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de> Cc: linux-mm@kvack.org Cc: wfg@linux.intel.com Cc: kernel test robot <fengguang.wu@intel.com> Cc: LKP <lkp@01.org> Link: http://lkml.kernel.org/r/20170320114108.kbvcsuepem45j5cr@hirez.programming.kicks-ass.net Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Peter Zijlstra [Wed, 22 Mar 2017 10:36:00 +0000 (11:36 +0100)]
futex: Drop hb->lock before enqueueing on the rtmutex
When PREEMPT_RT_FULL does the spinlock -> rt_mutex substitution the PI
chain code will (falsely) report a deadlock and BUG.
The problem is that it hold hb->lock (now an rt_mutex) while doing
task_blocks_on_rt_mutex on the futex's pi_state::rtmutex. This, when
interleaved just right with futex_unlock_pi() leads it to believe to see an
AB-BA deadlock.
Task1 (holds rt_mutex, Task2 (does FUTEX_LOCK_PI)
does FUTEX_UNLOCK_PI)
lock hb->lock
lock rt_mutex (as per start_proxy)
lock hb->lock
Which is a trivial AB-BA.
It is not an actual deadlock, because it won't be holding hb->lock by the
time it actually blocks on the rt_mutex, but the chainwalk code doesn't
know that and it would be a nightmare to handle this gracefully.
To avoid this problem, do the same as in futex_unlock_pi() and drop
hb->lock after acquiring wait_lock. This still fully serializes against
futex_unlock_pi(), since adding to the wait_list does the very same lock
dance, and removing it holds both locks.
Aside of solving the RT problem this makes the lock and unlock mechanism
symetric and reduces the hb->lock held time.
Reported-and-tested-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> Suggested-by: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: juri.lelli@arm.com Cc: xlpang@redhat.com Cc: rostedt@goodmis.org Cc: mathieu.desnoyers@efficios.com Cc: jdesfossez@efficios.com Cc: dvhart@infradead.org Cc: bristot@redhat.com Link: http://lkml.kernel.org/r/20170322104152.161341537@infradead.org Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Peter Zijlstra [Wed, 22 Mar 2017 10:35:59 +0000 (11:35 +0100)]
futex: Futex_unlock_pi() determinism
The problem with returning -EAGAIN when the waiter state mismatches is that
it becomes very hard to proof a bounded execution time on the
operation. And seeing that this is a RT operation, this is somewhat
important.
While in practise; given the previous patch; it will be very unlikely to
ever really take more than one or two rounds, proving so becomes rather
hard.
However, now that modifying wait_list is done while holding both hb->lock
and wait_lock, the scenario can be avoided entirely by acquiring wait_lock
while still holding hb-lock. Doing a hand-over, without leaving a hole.
It does however solve the earlier starvation/live-lock scenario which got
introduced with the -EAGAIN since unlike the before scenario; where the
-EAGAIN happens while futex_unlock_pi() doesn't hold any locks; in the
after scenario it happens while futex_unlock_pi() actually holds a lock,
and then it is serialized on that lock.
With the ultimate goal of keeping rt_mutex wait_list and futex_q waiters
consistent it's necessary to split 'rt_mutex_futex_lock()' into finer
parts, such that only the actual blocking can be done without hb->lock
held.
Split split_mutex_finish_proxy_lock() into two parts, one that does the
blocking and one that does remove_waiter() when the lock acquire failed.
When the rtmutex was acquired successfully the waiter can be removed in the
acquisiton path safely, since there is no concurrency on the lock owner.
This means that, except for futex_lock_pi(), all wait_list modifications
are done with both hb->lock and wait_lock held.
[bigeasy@linutronix.de: fix for futex_requeue_pi_signal_restart]
Peter Zijlstra [Wed, 22 Mar 2017 10:35:55 +0000 (11:35 +0100)]
futex: Pull rt_mutex_futex_unlock() out from under hb->lock
There's a number of 'interesting' problems, all caused by holding
hb->lock while doing the rt_mutex_unlock() equivalient.
Notably:
- a PI inversion on hb->lock; and,
- a SCHED_DEADLINE crash because of pointer instability.
The previous changes:
- changed the locking rules to cover {uval,pi_state} with wait_lock.
- allow to do rt_mutex_futex_unlock() without dropping wait_lock; which in
turn allows to rely on wait_lock atomicity completely.
- simplified the waiter conundrum.
It's now sufficient to hold rtmutex::wait_lock and a reference on the
pi_state to protect the state consistency, so hb->lock can be dropped
before calling rt_mutex_futex_unlock().
Peter Zijlstra [Wed, 22 Mar 2017 10:35:54 +0000 (11:35 +0100)]
futex: Rework inconsistent rt_mutex/futex_q state
There is a weird state in the futex_unlock_pi() path when it interleaves
with a concurrent futex_lock_pi() at the point where it drops hb->lock.
In this case, it can happen that the rt_mutex wait_list and the futex_q
disagree on pending waiters, in particular rt_mutex will find no pending
waiters where futex_q thinks there are. In this case the rt_mutex unlock
code cannot assign an owner.
The futex side fixup code has to cleanup the inconsistencies with quite a
bunch of interesting corner cases.
Simplify all this by changing wake_futex_pi() to return -EAGAIN when this
situation occurs. This then gives the futex_lock_pi() code the opportunity
to continue and the retried futex_unlock_pi() will now observe a coherent
state.
The only problem is that this breaks RT timeliness guarantees. That
is, consider the following scenario:
T1 and T2 are both pinned to CPU0. prio(T2) > prio(T1)
CPU0
T1
lock_pi()
queue_me() <- Waiter is visible
preemption
T2
unlock_pi()
loops with -EAGAIN forever
Which is undesirable for PI primitives. Future patches will rectify
this.
Peter Zijlstra [Wed, 22 Mar 2017 10:35:52 +0000 (11:35 +0100)]
futex: Change locking rules
Currently futex-pi relies on hb->lock to serialize everything. But hb->lock
creates another set of problems, especially priority inversions on RT where
hb->lock becomes a rt_mutex itself.
The rt_mutex::wait_lock is the most obvious protection for keeping the
futex user space value and the kernel internal pi_state in sync.
Rework and document the locking so rt_mutex::wait_lock is held accross all
operations which modify the user space value and the pi state.
This allows to invoke rt_mutex_unlock() (including deboost) without holding
hb->lock as a next step.
Peter Zijlstra [Wed, 22 Mar 2017 10:35:49 +0000 (11:35 +0100)]
futex: Use smp_store_release() in mark_wake_futex()
Since the futex_q can dissapear the instruction after assigning NULL,
this really should be a RELEASE barrier. That stops loads from hitting
dead memory too.
Where the boolean returns the result of the compare; and thus if the
exchange happened; and in case of failure, the new value of *ptr is
returned in *val.
This allows simplification/improvement of loops like:
for (;;) {
new = val $op $imm;
old = cmpxchg(ptr, val, new);
if (old == val)
break;
val = old;
}
into:
do {
} while (!try_cmpxchg(ptr, &val, val $op $imm));
while also generating better code (GCC6 and onwards).
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Andy Lutomirski <luto@kernel.org> Cc: Borislav Petkov <bp@alien8.de> Cc: Brian Gerst <brgerst@gmail.com> Cc: Denys Vlasenko <dvlasenk@redhat.com> Cc: H. Peter Anvin <hpa@zytor.com> Cc: Josh Poimboeuf <jpoimboe@redhat.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
The only part of the patch that can increase the noise is the update
of blocked load of group entity in update_blocked_averages().
We can optimize this call and skip the update of group entity if its load
and utilization are already null and there is no pending propagation of load
in the task group.
This optimization partly restores the noise score. A more agressive
optimization has been tried but has shown worse score.
Reported-by: ying.huang@linux.intel.com Signed-off-by: Vincent Guittot <vincent.guittot@linaro.org> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: dietmar.eggemann@arm.com Cc: ying.huang@intel.com Fixes: 4e5160766fcc ("sched/fair: Propagate asynchrous detach") Link: http://lkml.kernel.org/r/1489758442-2877-1-git-send-email-vincent.guittot@linaro.org
[ Fixed typos, improved layout. ] Signed-off-by: Ingo Molnar <mingo@kernel.org>
Peter Zijlstra [Fri, 3 Mar 2017 12:57:56 +0000 (13:57 +0100)]
locking/ww_mutex: Improve test to cover acquire context changes
Currently each thread starts an acquire context only once, and
performs all its loop iterations under it.
This means that the Wound/Wait relations between threads are fixed.
To make things a little more realistic and cover more of the
functionality with the test, open a new acquire context for each loop.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Acked-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Ingo Molnar <mingo@kernel.org>
then lockdep assigns a seperate key to each lock because the logic for
assigning a key to statically initialized locks is to use the address as
the key. With per CPU locks the address is obvioulsy different on each CPU.
That's wrong, because all locks should have the same key.
To solve this the following modifications are required:
1) Extend the is_kernel/module_percpu_addr() functions to hand back the
canonical address of the per CPU address, i.e. the per CPU address
minus the per CPU offset.
2) Check the lock address with these functions and if the per CPU check
matches use the returned canonical address as the lock key, so all per
CPU locks have the same key.
3) Move the static_obj(key) check into look_up_lock_class() so this check
can be avoided for statically initialized per CPU locks. That's
required because the canonical address fails the static_obj(key) check
for obvious reasons.
Reported-by: Mike Galbraith <efault@gmx.de> Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
[ Merged Dan's fixups for !MODULES and !SMP into this patch. ] Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Dan Murphy <dmurphy@ti.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/20170227143736.pectaimkjkan5kow@linutronix.de Signed-off-by: Ingo Molnar <mingo@kernel.org>
J. R. Okajima [Thu, 2 Feb 2017 16:38:17 +0000 (01:38 +0900)]
locking/lockdep: Add new check to lock_downgrade()
Commit:
f8319483f57f ("locking/lockdep: Provide a type check for lock_is_held")
didn't fully cover rwsems as downgrade_write() was left out.
Introduce lock_downgrade() and use it to add new checks.
See-also: http://marc.info/?l=linux-kernel&m=148581164003149&w=2 Originally-written-by: Peter Zijlstra <peterz@infradead.org> Signed-off-by: J. R. Okajima <hooanon05g@gmail.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Link: http://lkml.kernel.org/r/1486053497-9948-3-git-send-email-hooanon05g@gmail.com
[ Rewrote the changelog. ] Signed-off-by: Ingo Molnar <mingo@kernel.org>
J. R. Okajima [Thu, 2 Feb 2017 16:38:16 +0000 (01:38 +0900)]
locking/lockdep: Factor out the validate_held_lock() helper function
Behaviour should not change.
Signed-off-by: J. R. Okajima <hooanon05g@gmail.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Link: http://lkml.kernel.org/r/1486053497-9948-2-git-send-email-hooanon05g@gmail.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
J. R. Okajima [Thu, 2 Feb 2017 16:38:15 +0000 (01:38 +0900)]
locking/lockdep: Factor out the find_held_lock() helper function
A simple consolidataion to factor out repeated patterns.
The behaviour should not change.
Signed-off-by: J. R. Okajima <hooanon05g@gmail.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Link: http://lkml.kernel.org/r/1486053497-9948-1-git-send-email-hooanon05g@gmail.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
Peter Zijlstra [Tue, 21 Feb 2017 13:40:35 +0000 (14:40 +0100)]
sched/core: Simplify update_rq_clock() in __schedule()
Instead of relying on deactivate_task() to call update_rq_clock() and
handling the case where it didn't happen (task_on_rq_queued),
unconditionally do update_rq_clock() and skip any further updates.
This also avoids a double update on deactivate_task() + ttwu_local().
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Ingo Molnar <mingo@kernel.org>
Peter Zijlstra [Tue, 4 Oct 2016 14:29:45 +0000 (16:29 +0200)]
sched/core: Add {EN,DE}QUEUE_NOCLOCK flags
Currently {en,de}queue_task() do an unconditional update_rq_clock().
However since we want to avoid duplicate updates, so that each
rq->lock section appears atomic in time, we need to be able to skip
these clock updates.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Ingo Molnar <mingo@kernel.org>
Peter Zijlstra [Tue, 4 Oct 2016 14:04:35 +0000 (16:04 +0200)]
sched/core: Add rq->lock wrappers
The missing update_rq_clock() check can work with partial rq->lock
wrappery, since a missing wrapper can cause the warning to not be
emitted when it should have, but cannot cause the warning to trigger
when it should not have.
The duplicate update_rq_clock() check however can cause false warnings
to trigger. Therefore add more comprehensive rq->lock wrappery.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Ingo Molnar <mingo@kernel.org>
sched/rt: Add comments describing the RT IPI pull method
While looking into optimizations for the RT scheduler IPI logic, I realized
that the comments are lacking to describe it efficiently. It deserves a
lengthy description describing its design.
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Clark Williams <williams@redhat.com> Cc: Daniel Bristot de Oliveira <bristot@redhat.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Link: http://lkml.kernel.org/r/20170228155030.30c69068@gandalf.local.home
[ Small typographical edits. ] Signed-off-by: Ingo Molnar <mingo@kernel.org>
sched/deadline: Use deadline instead of period when calculating overflow
I was testing Daniel's changes with his test case, and tweaked it a
little. Instead of having the runtime equal to the deadline, I
increased the deadline ten fold.
The results were rather surprising. The behavior that Daniel's patch
was fixing came back. The task started using much more than .1% of the
CPU. More like 20%.
Looking into this I found that it was due to the dl_entity_overflow()
constantly returning true. That's because it uses the relative period
against relative runtime vs the absolute deadline against absolute
runtime.
runtime / (deadline - t) > dl_runtime / dl_period
There's even a comment mentioning this, and saying that when relative
deadline equals relative period, that the equation is the same as using
deadline instead of period. That comment is backwards! What we really
want is:
We care about if the runtime can make its deadline, not its period. And
then we can say "when the deadline equals the period, the equation is
the same as using dl_period instead of dl_deadline".
After correcting this, now when the task gets enqueued, it can throttle
correctly, and Daniel's fix to the throttling of sleeping deadline
tasks works even when the runtime and deadline are not the same.
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Daniel Bristot de Oliveira <bristot@redhat.com> Cc: Juri Lelli <juri.lelli@arm.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Luca Abeni <luca.abeni@santannapisa.it> Cc: Mike Galbraith <efault@gmx.de> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Romulo Silva de Oliveira <romulo.deoliveira@ufsc.br> Cc: Steven Rostedt <rostedt@goodmis.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Tommaso Cucinotta <tommaso.cucinotta@sssup.it> Link: http://lkml.kernel.org/r/02135a27f1ae3fe5fd032568a5a2f370e190e8d7.1488392936.git.bristot@redhat.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
sched/deadline: Throttle a constrained deadline task activated after the deadline
During the activation, CBS checks if it can reuse the current task's
runtime and period. If the deadline of the task is in the past, CBS
cannot use the runtime, and so it replenishes the task. This rule
works fine for implicit deadline tasks (deadline == period), and the
CBS was designed for implicit deadline tasks. However, a task with
constrained deadline (deadine < period) might be awakened after the
deadline, but before the next period. In this case, replenishing the
task would allow it to run for runtime / deadline. As in this case
deadline < period, CBS enables a task to run for more than the
runtime / period. In a very loaded system, this can cause a domino
effect, making other tasks miss their deadlines.
To avoid this problem, in the activation of a constrained deadline
task after the deadline but before the next period, throttle the
task and set the replenishing timer to the begin of the next period,
unless it is boosted.
Reproducer:
--------------- %< ---------------
int main (int argc, char **argv)
{
int ret;
int flags = 0;
unsigned long l = 0;
struct timespec ts;
struct sched_attr attr;
if (ret < 0) {
perror("sched_setattr");
exit(-1);
}
for(;;) {
/* XXX: you may need to adjust the loop */
for (l = 0; l < 150000; l++);
/*
* The ideia is to go to sleep right before the deadline
* and then wake up before the next period to receive
* a new replenishment.
*/
nanosleep(&ts, NULL);
}
exit(0);
}
--------------- >% ---------------
On my box, this reproducer uses almost 50% of the CPU time, which is
obviously wrong for a task with 2/2000 reservation.
Signed-off-by: Daniel Bristot de Oliveira <bristot@redhat.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Juri Lelli <juri.lelli@arm.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Luca Abeni <luca.abeni@santannapisa.it> Cc: Mike Galbraith <efault@gmx.de> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Romulo Silva de Oliveira <romulo.deoliveira@ufsc.br> Cc: Steven Rostedt <rostedt@goodmis.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Tommaso Cucinotta <tommaso.cucinotta@sssup.it> Link: http://lkml.kernel.org/r/edf58354e01db46bf42df8d2dd32418833f68c89.1488392936.git.bristot@redhat.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
sched/deadline: Make sure the replenishment timer fires in the next period
Currently, the replenishment timer is set to fire at the deadline
of a task. Although that works for implicit deadline tasks because the
deadline is equals to the begin of the next period, that is not correct
for constrained deadline tasks (deadline < period).
For instance:
f.c:
--------------- %< ---------------
int main (void)
{
for(;;);
}
--------------- >% ---------------
The task is throttled after running 492318 ms, as expected:
f-11295 [003] 13322.606094: sched_switch: f:11295 [-1] R ==> watchdog/3:32 [0]
But then, the task is replenished 500719 ms after the first
replenishment:
<idle>-0 [003] 13322.614495: sched_switch: swapper/3:0 [120] R ==> f:11295 [-1]
Running for 490277 ms:
f-11295 [003] 13323.104772: sched_switch: f:11295 [-1] R ==> swapper/3:0 [120]
Hence, in the first period, the task runs 2 * runtime, and that is a bug.
During the first replenishment, the next deadline is set one period away.
So the runtime / period starts to be respected. However, as the second
replenishment took place in the wrong instant, the next replenishment
will also be held in a wrong instant of time. Rather than occurring in
the nth period away from the first activation, it is taking place
in the (nth period - relative deadline).
Signed-off-by: Daniel Bristot de Oliveira <bristot@redhat.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Luca Abeni <luca.abeni@santannapisa.it> Reviewed-by: Steven Rostedt (VMware) <rostedt@goodmis.org> Reviewed-by: Juri Lelli <juri.lelli@arm.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Romulo Silva de Oliveira <romulo.deoliveira@ufsc.br> Cc: Steven Rostedt <rostedt@goodmis.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Tommaso Cucinotta <tommaso.cucinotta@sssup.it> Link: http://lkml.kernel.org/r/ac50d89887c25285b47465638354b63362f8adff.1488392936.git.bristot@redhat.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
Niklas Cassel [Sat, 25 Feb 2017 00:17:53 +0000 (01:17 +0100)]
locking/rwsem: Fix down_write_killable() for CONFIG_RWSEM_GENERIC_SPINLOCK=y
We hang if SIGKILL has been sent, but the task is stuck in down_read()
(after do_exit()), even though no task is doing down_write() on the
rwsem in question:
INFO: task libupnp:21868 blocked for more than 120 seconds.
libupnp D 0 21868 1 0x08100008
...
Call Trace:
__schedule()
schedule()
__down_read()
do_exit()
do_group_exit()
__wake_up_parent()
This bug has already been fixed for CONFIG_RWSEM_XCHGADD_ALGORITHM=y in
the following commit:
Matt Fleming [Fri, 17 Feb 2017 12:07:31 +0000 (12:07 +0000)]
sched/loadavg: Use {READ,WRITE}_ONCE() for sample window
'calc_load_update' is accessed without any kind of locking and there's
a clear assumption in the code that only a single value is read or
written.
Make this explicit by using READ_ONCE() and WRITE_ONCE(), and avoid
unintentionally seeing multiple values, or having the load/stores
split.
Technically the loads in calc_global_*() don't require this since
those are the only functions that update 'calc_load_update', but I've
added the READ_ONCE() for consistency.
Suggested-by: Peter Zijlstra <peterz@infradead.org> Signed-off-by: Matt Fleming <matt@codeblueprint.co.uk> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Mike Galbraith <umgwanakikbuti@gmail.com> Cc: Morten Rasmussen <morten.rasmussen@arm.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Vincent Guittot <vincent.guittot@linaro.org> Link: http://lkml.kernel.org/r/20170217120731.11868-3-matt@codeblueprint.co.uk Signed-off-by: Ingo Molnar <mingo@kernel.org>
Matt Fleming [Fri, 17 Feb 2017 12:07:30 +0000 (12:07 +0000)]
sched/loadavg: Avoid loadavg spikes caused by delayed NO_HZ accounting
If we crossed a sample window while in NO_HZ we will add LOAD_FREQ to
the pending sample window time on exit, setting the next update not
one window into the future, but two.
This has the effect of delaying load average updates for potentially
up to ~9seconds.
This can result in huge spikes in the load average values due to
per-cpu uninterruptible task counts being out of sync when accumulated
across all CPUs.
It's safe to update the per-cpu active count if we wake between sample
windows because any load that we left in 'calc_load_idle' will have
been zero'd when the idle load was folded in calc_global_load().
This issue is easy to reproduce before,
commit 9d89c257dfb9 ("sched/fair: Rewrite runnable load and utilization average tracking")
just by forking short-lived process pipelines built from ps(1) and
grep(1) in a loop. I'm unable to reproduce the spikes after that
commit, but the bug still seems to be present from code review.
Signed-off-by: Matt Fleming <matt@codeblueprint.co.uk> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Mike Galbraith <umgwanakikbuti@gmail.com> Cc: Morten Rasmussen <morten.rasmussen@arm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Vincent Guittot <vincent.guittot@linaro.org> Fixes: commit 5167e8d ("sched/nohz: Rewrite and fix load-avg computation -- again") Link: http://lkml.kernel.org/r/20170217120731.11868-2-matt@codeblueprint.co.uk Signed-off-by: Ingo Molnar <mingo@kernel.org>
The DL task will be migrated to a suitable later deadline rq once the DL
timer fires and currnet rq is offline. The rq clock of the new rq should
be updated. This patch fixes it by updating the rq clock after holding
the new rq's rq lock.
Signed-off-by: Wanpeng Li <wanpeng.li@hotmail.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Matt Fleming <matt@codeblueprint.co.uk> Cc: Juri Lelli <juri.lelli@arm.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Link: http://lkml.kernel.org/r/1488865888-15894-1-git-send-email-wanpeng.li@hotmail.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
Linus Torvalds [Wed, 15 Mar 2017 23:54:58 +0000 (16:54 -0700)]
Merge branch 'for-linus' of git://git.kernel.dk/linux-block
Pull block fixes from Jens Axboe:
"Four small fixes for this cycle:
- followup fix from Neil for a fix that went in before -rc2, ensuring
that we always see the full per-task bio_list.
- fix for blk-mq-sched from me that ensures that we retain similar
direct-to-issue behavior on running the queue.
- fix from Sagi fixing a potential NULL pointer dereference in blk-mq
on spurious CPU unplug.
- a memory leak fix in writeback from Tahsin, fixing a case where
device removal of a mounted device can leak a struct
wb_writeback_work"
* 'for-linus' of git://git.kernel.dk/linux-block:
blk-mq-sched: don't run the queue async from blk_mq_try_issue_directly()
writeback: fix memory leak in wb_queue_work()
blk-mq: Fix tagset reinit in the presence of cpu hot-unplug
blk: Ensure users for current->bio_list can see the full list.
Linus Torvalds [Wed, 15 Mar 2017 17:44:19 +0000 (10:44 -0700)]
Merge tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi
Pull SCSI fixes from James Bottomley:
"This is a rather large set of fixes. The bulk are for lpfc correcting
a lot of issues in the new NVME driver code which just went in in the
merge window.
The others are:
- fix a hang in the vmware paravirt driver caused by incorrect
handling of the new MSI vector allocation
- long standing bug in storvsc, which recent block changes turned
from being a harmless annoyance into a hang
- yet more fallout (in mpt3sas) from the changes to device blocking
The remainder are small fixes and updates"
* tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi: (34 commits)
scsi: lpfc: Add shutdown method for kexec
scsi: storvsc: Workaround for virtual DVD SCSI version
scsi: lpfc: revise version number to 11.2.0.10
scsi: lpfc: code cleanups in NVME initiator discovery
scsi: lpfc: code cleanups in NVME initiator base
scsi: lpfc: correct rdp diag portnames
scsi: lpfc: remove dead sli3 nvme code
scsi: lpfc: correct double print
scsi: lpfc: Rename LPFC_MAX_EQ_DELAY to LPFC_MAX_EQ_DELAY_EQID_CNT
scsi: lpfc: Rework lpfc Kconfig for NVME options
scsi: lpfc: add transport eh_timed_out reference
scsi: lpfc: Fix eh_deadline setting for sli3 adapters.
scsi: lpfc: add NVME exchange aborts
scsi: lpfc: Fix nvme allocation bug on failed nvme_fc_register_localport
scsi: lpfc: Fix IO submission if WQ is full
scsi: lpfc: Fix NVME CMD IU byte swapped word 1 problem
scsi: lpfc: Fix RCTL value on NVME LS request and response
scsi: lpfc: Fix crash during Hardware error recovery on SLI3 adapters
scsi: lpfc: fix missing spin_unlock on sql_list_lock
scsi: lpfc: don't dereference dma_buf->iocbq before null check
...
Linus Torvalds [Wed, 15 Mar 2017 16:33:15 +0000 (09:33 -0700)]
Merge tag 'gfs2-4.11-rc3.fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/gfs2/linux-gfs2
Pull gfs2 fix from Bob Peterson:
"This is an emergency patch for 4.11-rc3
The GFS2 developers uncovered a really nasty problem that can lead to
random corruption and kernel panic, much like the last one. Andreas
Gruenbacher wrote a simple one-line patch to fix the problem."
* tag 'gfs2-4.11-rc3.fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/gfs2/linux-gfs2:
gfs2: Avoid alignment hole in struct lm_lockname
Commit 88ffbf3e03 switches to using rhashtables for glocks, hashing over
the entire struct lm_lockname instead of its individual fields. On some
architectures, struct lm_lockname contains a hole of uninitialized
memory due to alignment rules, which now leads to incorrect hash values.
Get rid of that hole.
Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com> Signed-off-by: Bob Peterson <rpeterso@redhat.com> CC: <stable@vger.kernel.org> #v4.3+
1) Ensure that mtu is at least IPV6_MIN_MTU in ipv6 VTI tunnel driver,
from Steffen Klassert.
2) Fix crashes when user tries to get_next_key on an LPM bpf map, from
Alexei Starovoitov.
3) Fix detection of VLAN fitlering feature for bnx2x VF devices, from
Michal Schmidt.
4) We can get a divide by zero when TCP socket are morphed into
listening state, fix from Eric Dumazet.
5) Fix socket refcounting bugs in skb_complete_wifi_ack() and
skb_complete_tx_timestamp(). From Eric Dumazet.
6) Use after free in dccp_feat_activate_values(), also from Eric
Dumazet.
7) Like bonding team needs to use ETH_MAX_MTU as netdev->max_mtu, from
Jarod Wilson.
8) Fix use after free in vrf_xmit(), from David Ahern.
9) Don't do UDP Fragmentation Offload on IPComp ipsec packets, from
Alexey Kodanev.
10) Properly check napi_complete_done() return value in order to decide
whether to re-enable IRQs or not in amd-xgbe driver, from Thomas
Lendacky.
11) Fix double free of hwmon device in marvell phy driver, from Andrew
Lunn.
12) Don't crash on malformed netlink attributes in act_connmark, from
Etienne Noss.
13) Don't remove routes with a higher metric in ipv6 ECMP route replace,
from Sabrina Dubroca.
14) Don't write into a cloned SKB in ipv6 fragmentation handling, from
Florian Westphal.
15) Fix routing redirect races in dccp and tcp, basically the ICMP
handler can't modify the socket's cached route in it's locked by the
user at this moment. From Jon Maxwell.
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (108 commits)
qed: Enable iSCSI Out-of-Order
qed: Correct out-of-bound access in OOO history
qed: Fix interrupt flags on Rx LL2
qed: Free previous connections when releasing iSCSI
qed: Fix mapping leak on LL2 rx flow
qed: Prevent creation of too-big u32-chains
qed: Align CIDs according to DORQ requirement
mlxsw: reg: Fix SPVMLR max record count
mlxsw: reg: Fix SPVM max record count
net: Resend IGMP memberships upon peer notification.
dccp: fix memory leak during tear-down of unsuccessful connection request
tun: fix premature POLLOUT notification on tun devices
dccp/tcp: fix routing redirect race
ucc/hdlc: fix two little issue
vxlan: fix ovs support
net: use net->count to check whether a netns is alive or not
bridge: drop netfilter fake rtable unconditionally
ipv6: avoid write to a possibly cloned skb
net: wimax/i2400m: fix NULL-deref at probe
isdn/gigaset: fix NULL-deref at probe
...
Linus Torvalds [Tue, 14 Mar 2017 22:00:43 +0000 (15:00 -0700)]
Merge branch 'for-4.11-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/libata
Pull libata fixes from Tejun Heo:
"Three libata fixes:
- fix for a circular reference bug in sysfs code which prevented
pata_legacy devices from being released after probe failure, which
in turn prevented devres from releasing the associated resources.
- drop spurious WARN in the command issue path which can be triggered
by a legitimate passthrough command.
- an ahci_qoriq specific fix"
* 'for-4.11-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/libata:
ahci: qoriq: correct the sata ecc setting error
libata: drop WARN from protocol error in ata_sff_qc_issue()
libata: transport: Remove circular dependency at free time
Linus Torvalds [Tue, 14 Mar 2017 21:52:08 +0000 (14:52 -0700)]
Merge branch 'for-4.11-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/wq
Pull workqueue fix from Tejun Heo:
"If a delayed work is queued with NULL @wq, workqueue code explodes
after the timer expires at which point it's difficult to tell who the
culprit was.
This actually happened and the offender was net/smc this time.
Add an explicit sanity check for it in the queueing path"
* 'for-4.11-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/wq:
workqueue: trigger WARN if queue_delayed_work() is called with NULL @wq
Linus Torvalds [Tue, 14 Mar 2017 21:48:50 +0000 (14:48 -0700)]
Merge branch 'for-4.11-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/percpu
Pull percpu fixes from Tejun Heo:
- the allocation path was updating pcpu_nr_empty_pop_pages without the
required locking which can lead to incorrect handling of empty chunks
(e.g. keeping too many around), which is buggy but shouldn't lead to
critical failures. Fixed by adding the locking
- a trivial patch to drop an unused param from pcpu_get_pages()
* 'for-4.11-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/percpu:
percpu: remove unused chunk_alloc parameter from pcpu_get_pages()
percpu: acquire pcpu_lock when updating pcpu_nr_empty_pop_pages
David S. Miller [Tue, 14 Mar 2017 18:37:06 +0000 (11:37 -0700)]
Merge branch 'qed-fixes'
Yuval Mintz says:
====================
qed: Fixes series
This address several different issues in qed.
The more significant portions:
Patch #1 would cause timeout when qedr utilizes the highest
CIDs availble for it [or when future qede adapters would utilize
queues in some constellations].
Patch #4 fixes a leak of mapped addresses; When iommu is enabled,
offloaded storage protocols might eventually run out of resources
and fail to map additional buffers.
Patches #6,#7 were missing in the initial iSCSI infrastructure
submissions, and would hamper qedi's stability when it reaches
out-of-order scenarios.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Mintz, Yuval [Tue, 14 Mar 2017 13:26:04 +0000 (15:26 +0200)]
qed: Enable iSCSI Out-of-Order
Missing in the initial submission, qed fails to propagate qedi's
request to enable OOO to firmware.
Fixes: fc831825f99e ("qed: Add support for hardware offloaded iSCSI") Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Mintz, Yuval [Tue, 14 Mar 2017 13:26:03 +0000 (15:26 +0200)]
qed: Correct out-of-bound access in OOO history
Need to set the number of entries in database, otherwise the logic
would quickly surpass the array.
Fixes: 1d6cff4fca43 ("qed: Add iSCSI out of order packet handling") Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Ram Amrani [Tue, 14 Mar 2017 13:26:02 +0000 (15:26 +0200)]
qed: Fix interrupt flags on Rx LL2
Before iterating over the the LL2 Rx ring, the ring's
spinlock is taken via spin_lock_irqsave().
The actual processing of the packet [including handling
by the protocol driver] is done without said lock,
so qed releases the spinlock and re-claims it afterwards.
Problem is that the final spin_lock_irqrestore() at the end
of the iteration uses the original flags saved from the
initial irqsave() instead of the flags from the most recent
irqsave(). So it's possible that the interrupt status would
be incorrect at the end of the processing.
Fixes: 0a7fb11c23c0 ("qed: Add Light L2 support"); CC: Ram Amrani <Ram.Amrani@cavium.com> Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Mintz, Yuval [Tue, 14 Mar 2017 13:26:01 +0000 (15:26 +0200)]
qed: Free previous connections when releasing iSCSI
Fixes: fc831825f99e ("qed: Add support for hardware offloaded iSCSI") Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Tomer Tayar [Tue, 14 Mar 2017 13:25:59 +0000 (15:25 +0200)]
qed: Prevent creation of too-big u32-chains
Current Logic would allow the creation of a chain with U32_MAX + 1
elements, when the actual maximum supported by the driver infrastructure
is U32_MAX.
Fixes: a91eb52abb50 ("qed: Revisit chain implementation") Signed-off-by: Tomer Tayar <Tomer.Tayar@cavium.com> Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Ram Amrani [Tue, 14 Mar 2017 13:25:58 +0000 (15:25 +0200)]
qed: Align CIDs according to DORQ requirement
The Doorbell HW block can be configured at a granularity
of 16 x CIDs, so we need to make sure that the actual number
of CIDs configured would be a multiplication of 16.
Today, when RoCE is enabled - given that the number is unaligned,
doorbelling the higher CIDs would fail to reach the firmware and
would eventually timeout.
Fixes: dbb799c39717 ("qed: Initialize hardware for new protocols") Signed-off-by: Ram Amrani <Ram.Amrani@cavium.com> Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Jiri Pirko [Tue, 14 Mar 2017 13:00:01 +0000 (14:00 +0100)]
mlxsw: reg: Fix SPVMLR max record count
The num_rec field is 8 bit, so the maximal count number is 255.
This fixes vlans learning not being enabled for wider ranges than 255.
Fixes: a4feea74cd7a ("mlxsw: reg: Add Switch Port VLAN MAC Learning register definition") Signed-off-by: Jiri Pirko <jiri@mellanox.com> Reviewed-by: Ido Schimmel <idosch@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Jiri Pirko [Tue, 14 Mar 2017 13:00:00 +0000 (14:00 +0100)]
mlxsw: reg: Fix SPVM max record count
The num_rec field is 8 bit, so the maximal count number is 255. This
fixes vlans not being enabled for wider ranges than 255.
Fixes: b2e345f9a454 ("mlxsw: reg: Add Switch Port VID and Switch Port VLAN Membership registers definitions") Signed-off-by: Jiri Pirko <jiri@mellanox.com> Reviewed-by: Ido Schimmel <idosch@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Vlad Yasevich [Tue, 14 Mar 2017 12:58:08 +0000 (08:58 -0400)]
net: Resend IGMP memberships upon peer notification.
When we notify peers of potential changes, it's also good to update
IGMP memberships. For example, during VM migration, updating IGMP
memberships will redirect existing multicast streams to the VM at the
new location.
Signed-off-by: Vladislav Yasevich <vyasevic@redhat.com> Acked-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Jens Axboe [Tue, 14 Mar 2017 17:51:59 +0000 (11:51 -0600)]
blk-mq-sched: don't run the queue async from blk_mq_try_issue_directly()
If we have scheduling enabled, we jump directly to insert-and-run.
That's fine, but we run the queue async and we don't pass in information
on whether we can block from this context or not. Fixup both these
cases.
dccp: fix memory leak during tear-down of unsuccessful connection request
This patch fixes a memory leak, which happens if the connection request
is not fulfilled between parsing the DCCP options and handling the SYN
(because e.g. the backlog is full), because we forgot to free the
list of ack vectors.
Reported-by: Jianwen Ji <jiji@redhat.com> Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org> Signed-off-by: David S. Miller <davem@davemloft.net>
tun: fix premature POLLOUT notification on tun devices
aszlig observed failing ssh tunnels (-w) during initialization since
commit cc9da6cc4f56e0 ("ipv6: addrconf: use stable address generator for
ARPHRD_NONE"). We already had reports that the mentioned commit breaks
Juniper VPN connections. I can't clearly say that the Juniper VPN client
has the same problem, but it is worth a try to hint to this patch.
Because of the early generation of link local addresses, the kernel now
can start asking for routers on the local subnet much earlier than usual.
Those router solicitation packets arrive inside the ssh channels and
should be transmitted to the tun fd before the configuration scripts
might have upped the interface and made it ready for transmission.
ssh polls on the interface and receives back a POLL_OUT. It tries to send
the earily router solicitation packet to the tun interface. Unfortunately
it hasn't been up'ed yet by config scripts, thus failing with -EIO. ssh
doesn't retry again and considers the tun interface broken forever.
Link: https://bugzilla.kernel.org/show_bug.cgi?id=121131 Fixes: cc9da6cc4f56 ("ipv6: addrconf: use stable address generator for ARPHRD_NONE") Cc: Bjørn Mork <bjorn@mork.no> Reported-by: Valdis Kletnieks <Valdis.Kletnieks@vt.edu> Cc: Valdis Kletnieks <Valdis.Kletnieks@vt.edu> Reported-by: Jonas Lippuner <jonas@lippuner.ca> Cc: Jonas Lippuner <jonas@lippuner.ca> Reported-by: aszlig <aszlig@redmoonstudios.org> Cc: aszlig <aszlig@redmoonstudios.org> Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Jon Maxwell [Fri, 10 Mar 2017 05:40:33 +0000 (16:40 +1100)]
dccp/tcp: fix routing redirect race
As Eric Dumazet pointed out this also needs to be fixed in IPv6.
v2: Contains the IPv6 tcp/Ipv6 dccp patches as well.
We have seen a few incidents lately where a dst_enty has been freed
with a dangling TCP socket reference (sk->sk_dst_cache) pointing to that
dst_entry. If the conditions/timings are right a crash then ensues when the
freed dst_entry is referenced later on. A Common crashing back trace is:
But there are other backtraces attributed to the same freed dst_entry in
netfilter code as well.
All the vmcores showed 2 significant clues:
- Remote hosts behind the default gateway had always been redirected to a
different gateway. A rtable/dst_entry will be added for that host. Making
more dst_entrys with lower reference counts. Making this more probable.
- All vmcores showed a postitive LockDroppedIcmps value, e.g:
LockDroppedIcmps 267
A closer look at the tcp_v4_err() handler revealed that do_redirect() will run
regardless of whether user space has the socket locked. This can result in a
race condition where the same dst_entry cached in sk->sk_dst_entry can be
decremented twice for the same socket via:
do_redirect()->__sk_dst_check()-> dst_release().
Which leads to the dst_entry being prematurely freed with another socket
pointing to it via sk->sk_dst_cache and a subsequent crash.
To fix this skip do_redirect() if usespace has the socket locked. Instead let
the redirect take place later when user space does not have the socket
locked.
The dccp/IPv6 code is very similar in this respect, so fixing it there too.
As Eric Garver pointed out the following commit now invalidates routes. Which
can set the dst->obsolete flag so that ipv4_dst_check() returns null and
triggers the dst_release().
Fixes: ceb3320610d6 ("ipv4: Kill routes during PMTU/redirect updates.") Cc: Eric Garver <egarver@redhat.com> Cc: Hannes Sowa <hsowa@redhat.com> Signed-off-by: Jon Maxwell <jmaxwell37@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Zhao Qiang [Tue, 14 Mar 2017 01:38:33 +0000 (09:38 +0800)]
ucc/hdlc: fix two little issue
1. modify bd_status from u32 to u16 in function hdlc_rx_done,
because bd_status register is 16bits
2. write bd_length register before writing bd_status register
Signed-off-by: Zhao Qiang <qiang.zhao@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Linus Torvalds [Tue, 14 Mar 2017 02:48:22 +0000 (19:48 -0700)]
Merge tag 'powerpc-4.11-4' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux
Pull some more powerpc fixes from Michael Ellerman:
"The main item is the addition of the Power9 Machine Check handler.
This was delayed to make sure some details were correct, and is as
minimal as possible.
The rest is small fixes, two for the Power9 PMU, two dealing with
obscure toolchain problems, two for the PowerNV IOMMU code (used by
VFIO), and one to fix a crash on 32-bit machines with macio devices
due to missing dma_ops.
Thanks to:
Alexey Kardashevskiy, Cyril Bur, Larry Finger, Madhavan Srinivasan,
Nicholas Piggin"
* tag 'powerpc-4.11-4' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux:
powerpc/64s: POWER9 machine check handler
powerpc/64s: allow machine check handler to set severity and initiator
powerpc/64s: fix handling of non-synchronous machine checks
powerpc/pmac: Fix crash in dma-mapping.h with NULL dma_ops
powerpc/powernv/ioda2: Update iommu table base on ownership change
powerpc/powernv/ioda2: Gracefully fail if too many TCE levels requested
selftests/powerpc: Replace stxvx and lxvx with stxvd2x/lxvd2x
powerpc/perf: Handle sdar_mode for marked event in power9
powerpc/perf: Fix perf_get_data_addr() for power9 DD1
powerpc/boot: Fix zImage TOC alignment
Nicolas Dichtel [Mon, 13 Mar 2017 15:24:03 +0000 (16:24 +0100)]
vxlan: fix ovs support
The required changes in the function vxlan_dev_create() were missing
in commit 8bcdc4f3a20b.
The vxlan device is not registered anymore after this patch and the error
path causes an stack dump:
WARNING: CPU: 3 PID: 1498 at net/core/dev.c:6713 rollback_registered_many+0x9d/0x3f0
Fixes: 8bcdc4f3a20b ("vxlan: add changelink support") CC: Roopa Prabhu <roopa@cumulusnetworks.com> Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com> Acked-by: Roopa Prabhu <roopa@cumulusnetworks.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Andrey Vagin [Mon, 13 Mar 2017 04:36:18 +0000 (21:36 -0700)]
net: use net->count to check whether a netns is alive or not
The previous idea was to check whether a net namespace is in
net_exit_list or not. It doesn't work, because net->exit_list is used in
__register_pernet_operations and __unregister_pernet_operations where
all namespaces are added to a temporary list to make cleanup in a error
case, so list_empty(&net->exit_list) always returns false.
Reported-by: Mantas Mikulėnas <grawity@gmail.com> Fixes: 002d8a1a6c11 ("net: skip genenerating uevents for network namespaces that are exiting") Signed-off-by: Andrei Vagin <avagin@openvz.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Florian Westphal [Mon, 13 Mar 2017 16:38:17 +0000 (17:38 +0100)]
bridge: drop netfilter fake rtable unconditionally
Andreas reports kernel oops during rmmod of the br_netfilter module.
Hannes debugged the oops down to a NULL rt6info->rt6i_indev.
Problem is that br_netfilter has the nasty concept of adding a fake
rtable to skb->dst; this happens in a br_netfilter prerouting hook.
A second hook (in bridge LOCAL_IN) is supposed to remove these again
before the skb is handed up the stack.
However, on module unload hooks get unregistered which means an
skb could traverse the prerouting hook that attaches the fake_rtable,
while the 'fake rtable remove' hook gets removed from the hooklist
immediately after.
Fixes: 34666d467cbf1e2e3c7 ("netfilter: bridge: move br_netfilter out of the core") Reported-by: Andreas Karis <akaris@redhat.com> Debugged-by: Hannes Frederic Sowa <hannes@stressinduktion.org> Signed-off-by: Florian Westphal <fw@strlen.de> Acked-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Florian Westphal [Mon, 13 Mar 2017 15:24:28 +0000 (16:24 +0100)]
ipv6: avoid write to a possibly cloned skb
ip6_fragment, in case skb has a fraglist, checks if the
skb is cloned. If it is, it will move to the 'slow path' and allocates
new skbs for each fragment.
However, right before entering the slowpath loop, it updates the
nexthdr value of the last ipv6 extension header to NEXTHDR_FRAGMENT,
to account for the fragment header that will be inserted in the new
ipv6-fragment skbs.
In case original skb is cloned this munges nexthdr value of another
skb. Avoid this by doing the nexthdr update for each of the new fragment
skbs separately.
This was observed with tcpdump on a bridge device where netfilter ipv6
reassembly is active: tcpdump shows malformed fragment headers as
the l4 header (icmpv6, tcp, etc). is decoded as a fragment header.
Cc: Hannes Frederic Sowa <hannes@stressinduktion.org> Reported-by: Andreas Karis <akaris@redhat.com> Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
Johan Hovold [Mon, 13 Mar 2017 12:42:03 +0000 (13:42 +0100)]
net: wimax/i2400m: fix NULL-deref at probe
Make sure to check the number of endpoints to avoid dereferencing a
NULL-pointer or accessing memory beyond the endpoint array should a
malicious device lack the expected endpoints.
The endpoints are specifically dereferenced in the i2400m_bootrom_init
path during probe (e.g. in i2400mu_tx_bulk_out).
Fixes: f398e4240fce ("i2400m/USB: probe/disconnect, dev init/shutdown
and reset backends") Cc: Inaky Perez-Gonzalez <inaky@linux.intel.com> Signed-off-by: Johan Hovold <johan@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Johan Hovold [Mon, 13 Mar 2017 12:39:01 +0000 (13:39 +0100)]
isdn/gigaset: fix NULL-deref at probe
Make sure to check the number of endpoints to avoid dereferencing a
NULL-pointer should a malicious device lack endpoints.
Fixes: cf7776dc05b8 ("[PATCH] isdn4linux: Siemens Gigaset drivers -
direct USB connection") Cc: stable <stable@vger.kernel.org> # 2.6.17 Cc: Hansjoerg Lipp <hjlipp@web.de> Signed-off-by: Johan Hovold <johan@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Sabrina Dubroca [Mon, 13 Mar 2017 12:28:09 +0000 (13:28 +0100)]
ipv6: make ECMP route replacement less greedy
Commit 27596472473a ("ipv6: fix ECMP route replacement") introduced a
loop that removes all siblings of an ECMP route that is being
replaced. However, this loop doesn't stop when it has replaced
siblings, and keeps removing other routes with a higher metric.
We also end up triggering the WARN_ON after the loop, because after
this nsiblings < 0.
Instead, stop the loop when we have taken care of all routes with the
same metric as the route being replaced.
Reproducer:
===========
#!/bin/sh
ip netns add ns1
ip netns add ns2
ip -net ns1 link set lo up
for x in 0 1 2 ; do
ip link add veth$x netns ns2 type veth peer name eth$x netns ns1
ip -net ns1 link set eth$x up
ip -net ns2 link set veth$x up
done
ip -net ns1 -6 r a 2000::/64 nexthop via fe80::0 dev eth0 \
nexthop via fe80::1 dev eth1 nexthop via fe80::2 dev eth2
ip -net ns1 -6 r a 2000::/64 via fe80::42 dev eth0 metric 256
ip -net ns1 -6 r a 2000::/64 via fe80::43 dev eth0 metric 2048
echo "before replace, 3 routes"
ip -net ns1 -6 r | grep -v '^fe80\|^ff00'
echo
ip -net ns1 -6 r c 2000::/64 nexthop via fe80::4 dev eth0 \
nexthop via fe80::5 dev eth1 nexthop via fe80::6 dev eth2
echo "after replace, only 2 routes, metric 2048 is gone"
ip -net ns1 -6 r | grep -v '^fe80\|^ff00'
Fixes: 27596472473a ("ipv6: fix ECMP route replacement") Signed-off-by: Sabrina Dubroca <sd@queasysnail.net> Acked-by: Nicolas Dichtel <nicolas.dichtel@6wind.com> Reviewed-by: Xin Long <lucien.xin@gmail.com> Reviewed-by: Michal Kubecek <mkubecek@suse.cz> Signed-off-by: David S. Miller <davem@davemloft.net>
Tahsin Erdogan [Fri, 10 Mar 2017 20:09:49 +0000 (12:09 -0800)]
writeback: fix memory leak in wb_queue_work()
When WB_registered flag is not set, wb_queue_work() skips queuing the
work, but does not perform the necessary clean up. In particular, if
work->auto_free is true, it should free the memory.
The leak condition can be reprouced by following these steps:
mount /dev/sdb /mnt/sdb
/* In qemu console: device_del sdb */
umount /dev/sdb
Above will result in a wb_queue_work() call on an unregistered wb and
thus leak memory.
Reported-by: John Sperbeck <jsperbeck@google.com> Signed-off-by: Tahsin Erdogan <tahsin@google.com> Reviewed-by: Jan Kara <jack@suse.cz> Signed-off-by: Jens Axboe <axboe@fb.com>
Sagi Grimberg [Mon, 13 Mar 2017 14:10:11 +0000 (16:10 +0200)]
blk-mq: Fix tagset reinit in the presence of cpu hot-unplug
In case cpu was unplugged, we need to make sure not to assume
that the tags for that cpu are still allocated. so check
for null tags when reinitializing a tagset.
Daniel Borkmann [Sat, 11 Mar 2017 15:55:49 +0000 (16:55 +0100)]
bpf: improve read-only handling
Improve bpf_{prog,jit_binary}_{un,}lock_ro() by throwing a
one-time warning in case of an error when the image couldn't
be set read-only, and also mark struct bpf_prog as locked when
bpf_prog_lock_ro() was called.
Reason for the latter is that bpf_prog_unlock_ro() is called from
various places including error paths, and we shouldn't mess with
page attributes when really not needed.
For bpf_jit_binary_unlock_ro() this is not needed as jited flag
implicitly indicates this, thus for archs with ARCH_HAS_SET_MEMORY
we're guaranteed to have a previously locked image. Overall, this
should also help us to identify any further potential issues with
set_memory_*() helpers.
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Acked-by: Alexei Starovoitov <ast@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Recent merge of 'linux-kselftest-4.11-rc1' tree broke bpf test build.
None of the tests were building and test_verifier.c had tons of compiler errors.
Fix it and add #ifdef CAP_IS_SUPPORTED to support old versions of libcap.
Tested on centos 6.8 and 7
Signed-off-by: Alexei Starovoitov <ast@kernel.org> Acked-by: Daniel Borkmann <daniel@iogearbox.net> Tested-by: Daniel Borkmann <daniel@iogearbox.net> Signed-off-by: David S. Miller <davem@davemloft.net>
David Ahern [Fri, 10 Mar 2017 22:11:39 +0000 (14:11 -0800)]
mpls: Do not decrement alive counter for unregister events
Multipath routes can be rendered usesless when a device in one of the
paths is deleted. For example:
$ ip -f mpls ro ls
100
nexthop as to 200 via inet 172.16.2.2 dev virt12
nexthop as to 300 via inet 172.16.3.2 dev br0
101
nexthop as to 201 via inet6 2000:2::2 dev virt12
nexthop as to 301 via inet6 2000:3::2 dev br0
$ ip li del br0
When br0 is deleted the other hop is not considered in
mpls_select_multipath because of the alive check -- rt_nhn_alive
is 0.
rt_nhn_alive is decremented once in mpls_ifdown when the device is taken
down (NETDEV_DOWN) and again when it is deleted (NETDEV_UNREGISTER). For
a 2 hop route, deleting one device drops the alive count to 0. Since
devices are taken down before unregistering, the decrement on
NETDEV_UNREGISTER is redundant.
Fixes: c89359a42e2a4 ("mpls: support for dead routes") Signed-off-by: David Ahern <dsa@cumulusnetworks.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Igor Druzhinin [Fri, 10 Mar 2017 21:36:22 +0000 (21:36 +0000)]
xen-netback: fix race condition on XenBus disconnect
In some cases during XenBus disconnect event handling and subsequent
queue resource release there may be some TX handlers active on
other processors. Use RCU in order to synchronize with them.
Signed-off-by: Igor Druzhinin <igor.druzhinin@citrix.com> Signed-off-by: David S. Miller <davem@davemloft.net>
David Ahern [Fri, 10 Mar 2017 17:46:15 +0000 (09:46 -0800)]
mpls: Send route delete notifications when router module is unloaded
When the mpls_router module is unloaded, mpls routes are deleted but
notifications are not sent to userspace leaving userspace caches
out of sync. Add the call to mpls_notify_route in mpls_net_exit as
routes are freed.
Fixes: 0189197f44160 ("mpls: Basic routing support") Signed-off-by: David Ahern <dsa@cumulusnetworks.com> Signed-off-by: David S. Miller <davem@davemloft.net>
projects / karo-tx-linux.git / log