2019-05-27 14:55:06 +08:00
|
|
|
// SPDX-License-Identifier: GPL-2.0-or-later
|
membarrier: Provide expedited private command
Implement MEMBARRIER_CMD_PRIVATE_EXPEDITED with IPIs using cpumask built
from all runqueues for which current thread's mm is the same as the
thread calling sys_membarrier. It executes faster than the non-expedited
variant (no blocking). It also works on NOHZ_FULL configurations.
Scheduler-wise, it requires a memory barrier before and after context
switching between processes (which have different mm). The memory
barrier before context switch is already present. For the barrier after
context switch:
* Our TSO archs can do RELEASE without being a full barrier. Look at
x86 spin_unlock() being a regular STORE for example. But for those
archs, all atomics imply smp_mb and all of them have atomic ops in
switch_mm() for mm_cpumask(), and on x86 the CR3 load acts as a full
barrier.
* From all weakly ordered machines, only ARM64 and PPC can do RELEASE,
the rest does indeed do smp_mb(), so there the spin_unlock() is a full
barrier and we're good.
* ARM64 has a very heavy barrier in switch_to(), which suffices.
* PPC just removed its barrier from switch_to(), but appears to be
talking about adding something to switch_mm(). So add a
smp_mb__after_unlock_lock() for now, until this is settled on the PPC
side.
Changes since v3:
- Properly document the memory barriers provided by each architecture.
Changes since v2:
- Address comments from Peter Zijlstra,
- Add smp_mb__after_unlock_lock() after finish_lock_switch() in
finish_task_switch() to add the memory barrier we need after storing
to rq->curr. This is much simpler than the previous approach relying
on atomic_dec_and_test() in mmdrop(), which actually added a memory
barrier in the common case of switching between userspace processes.
- Return -EINVAL when MEMBARRIER_CMD_SHARED is used on a nohz_full
kernel, rather than having the whole membarrier system call returning
-ENOSYS. Indeed, CMD_PRIVATE_EXPEDITED is compatible with nohz_full.
Adapt the CMD_QUERY mask accordingly.
Changes since v1:
- move membarrier code under kernel/sched/ because it uses the
scheduler runqueue,
- only add the barrier when we switch from a kernel thread. The case
where we switch from a user-space thread is already handled by
the atomic_dec_and_test() in mmdrop().
- add a comment to mmdrop() documenting the requirement on the implicit
memory barrier.
CC: Peter Zijlstra <peterz@infradead.org>
CC: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
CC: Boqun Feng <boqun.feng@gmail.com>
CC: Andrew Hunter <ahh@google.com>
CC: Maged Michael <maged.michael@gmail.com>
CC: gromer@google.com
CC: Avi Kivity <avi@scylladb.com>
CC: Benjamin Herrenschmidt <benh@kernel.crashing.org>
CC: Paul Mackerras <paulus@samba.org>
CC: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: Dave Watson <davejwatson@fb.com>
2017-07-29 04:40:40 +08:00
|
|
|
/*
|
|
|
|
* Copyright (C) 2010-2017 Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
|
|
|
|
*
|
|
|
|
* membarrier system call
|
|
|
|
*/
|
2018-03-03 19:20:47 +08:00
|
|
|
#include "sched.h"
|
membarrier: Provide expedited private command
Implement MEMBARRIER_CMD_PRIVATE_EXPEDITED with IPIs using cpumask built
from all runqueues for which current thread's mm is the same as the
thread calling sys_membarrier. It executes faster than the non-expedited
variant (no blocking). It also works on NOHZ_FULL configurations.
Scheduler-wise, it requires a memory barrier before and after context
switching between processes (which have different mm). The memory
barrier before context switch is already present. For the barrier after
context switch:
* Our TSO archs can do RELEASE without being a full barrier. Look at
x86 spin_unlock() being a regular STORE for example. But for those
archs, all atomics imply smp_mb and all of them have atomic ops in
switch_mm() for mm_cpumask(), and on x86 the CR3 load acts as a full
barrier.
* From all weakly ordered machines, only ARM64 and PPC can do RELEASE,
the rest does indeed do smp_mb(), so there the spin_unlock() is a full
barrier and we're good.
* ARM64 has a very heavy barrier in switch_to(), which suffices.
* PPC just removed its barrier from switch_to(), but appears to be
talking about adding something to switch_mm(). So add a
smp_mb__after_unlock_lock() for now, until this is settled on the PPC
side.
Changes since v3:
- Properly document the memory barriers provided by each architecture.
Changes since v2:
- Address comments from Peter Zijlstra,
- Add smp_mb__after_unlock_lock() after finish_lock_switch() in
finish_task_switch() to add the memory barrier we need after storing
to rq->curr. This is much simpler than the previous approach relying
on atomic_dec_and_test() in mmdrop(), which actually added a memory
barrier in the common case of switching between userspace processes.
- Return -EINVAL when MEMBARRIER_CMD_SHARED is used on a nohz_full
kernel, rather than having the whole membarrier system call returning
-ENOSYS. Indeed, CMD_PRIVATE_EXPEDITED is compatible with nohz_full.
Adapt the CMD_QUERY mask accordingly.
Changes since v1:
- move membarrier code under kernel/sched/ because it uses the
scheduler runqueue,
- only add the barrier when we switch from a kernel thread. The case
where we switch from a user-space thread is already handled by
the atomic_dec_and_test() in mmdrop().
- add a comment to mmdrop() documenting the requirement on the implicit
memory barrier.
CC: Peter Zijlstra <peterz@infradead.org>
CC: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
CC: Boqun Feng <boqun.feng@gmail.com>
CC: Andrew Hunter <ahh@google.com>
CC: Maged Michael <maged.michael@gmail.com>
CC: gromer@google.com
CC: Avi Kivity <avi@scylladb.com>
CC: Benjamin Herrenschmidt <benh@kernel.crashing.org>
CC: Paul Mackerras <paulus@samba.org>
CC: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: Dave Watson <davejwatson@fb.com>
2017-07-29 04:40:40 +08:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Bitmask made from a "or" of all commands within enum membarrier_cmd,
|
|
|
|
* except MEMBARRIER_CMD_QUERY.
|
|
|
|
*/
|
2018-01-30 04:20:17 +08:00
|
|
|
#ifdef CONFIG_ARCH_HAS_MEMBARRIER_SYNC_CORE
|
sched: Clean up and harmonize the coding style of the scheduler code base
A good number of small style inconsistencies have accumulated
in the scheduler core, so do a pass over them to harmonize
all these details:
- fix speling in comments,
- use curly braces for multi-line statements,
- remove unnecessary parentheses from integer literals,
- capitalize consistently,
- remove stray newlines,
- add comments where necessary,
- remove invalid/unnecessary comments,
- align structure definitions and other data types vertically,
- add missing newlines for increased readability,
- fix vertical tabulation where it's misaligned,
- harmonize preprocessor conditional block labeling
and vertical alignment,
- remove line-breaks where they uglify the code,
- add newline after local variable definitions,
No change in functionality:
md5:
1191fa0a890cfa8132156d2959d7e9e2 built-in.o.before.asm
1191fa0a890cfa8132156d2959d7e9e2 built-in.o.after.asm
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>
2018-03-03 21:01:12 +08:00
|
|
|
#define MEMBARRIER_PRIVATE_EXPEDITED_SYNC_CORE_BITMASK \
|
|
|
|
(MEMBARRIER_CMD_PRIVATE_EXPEDITED_SYNC_CORE \
|
2018-01-30 04:20:17 +08:00
|
|
|
| MEMBARRIER_CMD_REGISTER_PRIVATE_EXPEDITED_SYNC_CORE)
|
|
|
|
#else
|
|
|
|
#define MEMBARRIER_PRIVATE_EXPEDITED_SYNC_CORE_BITMASK 0
|
|
|
|
#endif
|
|
|
|
|
sched: Clean up and harmonize the coding style of the scheduler code base
A good number of small style inconsistencies have accumulated
in the scheduler core, so do a pass over them to harmonize
all these details:
- fix speling in comments,
- use curly braces for multi-line statements,
- remove unnecessary parentheses from integer literals,
- capitalize consistently,
- remove stray newlines,
- add comments where necessary,
- remove invalid/unnecessary comments,
- align structure definitions and other data types vertically,
- add missing newlines for increased readability,
- fix vertical tabulation where it's misaligned,
- harmonize preprocessor conditional block labeling
and vertical alignment,
- remove line-breaks where they uglify the code,
- add newline after local variable definitions,
No change in functionality:
md5:
1191fa0a890cfa8132156d2959d7e9e2 built-in.o.before.asm
1191fa0a890cfa8132156d2959d7e9e2 built-in.o.after.asm
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>
2018-03-03 21:01:12 +08:00
|
|
|
#define MEMBARRIER_CMD_BITMASK \
|
|
|
|
(MEMBARRIER_CMD_GLOBAL | MEMBARRIER_CMD_GLOBAL_EXPEDITED \
|
|
|
|
| MEMBARRIER_CMD_REGISTER_GLOBAL_EXPEDITED \
|
|
|
|
| MEMBARRIER_CMD_PRIVATE_EXPEDITED \
|
|
|
|
| MEMBARRIER_CMD_REGISTER_PRIVATE_EXPEDITED \
|
2018-01-30 04:20:17 +08:00
|
|
|
| MEMBARRIER_PRIVATE_EXPEDITED_SYNC_CORE_BITMASK)
|
membarrier: Provide expedited private command
Implement MEMBARRIER_CMD_PRIVATE_EXPEDITED with IPIs using cpumask built
from all runqueues for which current thread's mm is the same as the
thread calling sys_membarrier. It executes faster than the non-expedited
variant (no blocking). It also works on NOHZ_FULL configurations.
Scheduler-wise, it requires a memory barrier before and after context
switching between processes (which have different mm). The memory
barrier before context switch is already present. For the barrier after
context switch:
* Our TSO archs can do RELEASE without being a full barrier. Look at
x86 spin_unlock() being a regular STORE for example. But for those
archs, all atomics imply smp_mb and all of them have atomic ops in
switch_mm() for mm_cpumask(), and on x86 the CR3 load acts as a full
barrier.
* From all weakly ordered machines, only ARM64 and PPC can do RELEASE,
the rest does indeed do smp_mb(), so there the spin_unlock() is a full
barrier and we're good.
* ARM64 has a very heavy barrier in switch_to(), which suffices.
* PPC just removed its barrier from switch_to(), but appears to be
talking about adding something to switch_mm(). So add a
smp_mb__after_unlock_lock() for now, until this is settled on the PPC
side.
Changes since v3:
- Properly document the memory barriers provided by each architecture.
Changes since v2:
- Address comments from Peter Zijlstra,
- Add smp_mb__after_unlock_lock() after finish_lock_switch() in
finish_task_switch() to add the memory barrier we need after storing
to rq->curr. This is much simpler than the previous approach relying
on atomic_dec_and_test() in mmdrop(), which actually added a memory
barrier in the common case of switching between userspace processes.
- Return -EINVAL when MEMBARRIER_CMD_SHARED is used on a nohz_full
kernel, rather than having the whole membarrier system call returning
-ENOSYS. Indeed, CMD_PRIVATE_EXPEDITED is compatible with nohz_full.
Adapt the CMD_QUERY mask accordingly.
Changes since v1:
- move membarrier code under kernel/sched/ because it uses the
scheduler runqueue,
- only add the barrier when we switch from a kernel thread. The case
where we switch from a user-space thread is already handled by
the atomic_dec_and_test() in mmdrop().
- add a comment to mmdrop() documenting the requirement on the implicit
memory barrier.
CC: Peter Zijlstra <peterz@infradead.org>
CC: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
CC: Boqun Feng <boqun.feng@gmail.com>
CC: Andrew Hunter <ahh@google.com>
CC: Maged Michael <maged.michael@gmail.com>
CC: gromer@google.com
CC: Avi Kivity <avi@scylladb.com>
CC: Benjamin Herrenschmidt <benh@kernel.crashing.org>
CC: Paul Mackerras <paulus@samba.org>
CC: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: Dave Watson <davejwatson@fb.com>
2017-07-29 04:40:40 +08:00
|
|
|
|
|
|
|
static void ipi_mb(void *info)
|
|
|
|
{
|
|
|
|
smp_mb(); /* IPIs should be serializing but paranoid. */
|
|
|
|
}
|
|
|
|
|
2019-09-20 01:37:02 +08:00
|
|
|
static void ipi_sync_rq_state(void *info)
|
|
|
|
{
|
|
|
|
struct mm_struct *mm = (struct mm_struct *) info;
|
|
|
|
|
|
|
|
if (current->mm != mm)
|
|
|
|
return;
|
|
|
|
this_cpu_write(runqueues.membarrier_state,
|
|
|
|
atomic_read(&mm->membarrier_state));
|
|
|
|
/*
|
|
|
|
* Issue a memory barrier after setting
|
|
|
|
* MEMBARRIER_STATE_GLOBAL_EXPEDITED in the current runqueue to
|
|
|
|
* guarantee that no memory access following registration is reordered
|
|
|
|
* before registration.
|
|
|
|
*/
|
|
|
|
smp_mb();
|
|
|
|
}
|
|
|
|
|
|
|
|
void membarrier_exec_mmap(struct mm_struct *mm)
|
|
|
|
{
|
|
|
|
/*
|
|
|
|
* Issue a memory barrier before clearing membarrier_state to
|
|
|
|
* guarantee that no memory access prior to exec is reordered after
|
|
|
|
* clearing this state.
|
|
|
|
*/
|
|
|
|
smp_mb();
|
|
|
|
atomic_set(&mm->membarrier_state, 0);
|
|
|
|
/*
|
|
|
|
* Keep the runqueue membarrier_state in sync with this mm
|
|
|
|
* membarrier_state.
|
|
|
|
*/
|
|
|
|
this_cpu_write(runqueues.membarrier_state, 0);
|
|
|
|
}
|
|
|
|
|
2018-01-30 04:20:13 +08:00
|
|
|
static int membarrier_global_expedited(void)
|
|
|
|
{
|
|
|
|
int cpu;
|
|
|
|
bool fallback = false;
|
|
|
|
cpumask_var_t tmpmask;
|
|
|
|
|
|
|
|
if (num_online_cpus() == 1)
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Matches memory barriers around rq->curr modification in
|
|
|
|
* scheduler.
|
|
|
|
*/
|
|
|
|
smp_mb(); /* system call entry is not a mb. */
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Expedited membarrier commands guarantee that they won't
|
|
|
|
* block, hence the GFP_NOWAIT allocation flag and fallback
|
|
|
|
* implementation.
|
|
|
|
*/
|
|
|
|
if (!zalloc_cpumask_var(&tmpmask, GFP_NOWAIT)) {
|
|
|
|
/* Fallback for OOM. */
|
|
|
|
fallback = true;
|
|
|
|
}
|
|
|
|
|
|
|
|
cpus_read_lock();
|
2019-09-20 01:37:02 +08:00
|
|
|
rcu_read_lock();
|
2018-01-30 04:20:13 +08:00
|
|
|
for_each_online_cpu(cpu) {
|
|
|
|
struct task_struct *p;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Skipping the current CPU is OK even through we can be
|
|
|
|
* migrated at any point. The current CPU, at the point
|
|
|
|
* where we read raw_smp_processor_id(), is ensured to
|
|
|
|
* be in program order with respect to the caller
|
|
|
|
* thread. Therefore, we can skip this CPU from the
|
|
|
|
* iteration.
|
|
|
|
*/
|
|
|
|
if (cpu == raw_smp_processor_id())
|
|
|
|
continue;
|
sched: Clean up and harmonize the coding style of the scheduler code base
A good number of small style inconsistencies have accumulated
in the scheduler core, so do a pass over them to harmonize
all these details:
- fix speling in comments,
- use curly braces for multi-line statements,
- remove unnecessary parentheses from integer literals,
- capitalize consistently,
- remove stray newlines,
- add comments where necessary,
- remove invalid/unnecessary comments,
- align structure definitions and other data types vertically,
- add missing newlines for increased readability,
- fix vertical tabulation where it's misaligned,
- harmonize preprocessor conditional block labeling
and vertical alignment,
- remove line-breaks where they uglify the code,
- add newline after local variable definitions,
No change in functionality:
md5:
1191fa0a890cfa8132156d2959d7e9e2 built-in.o.before.asm
1191fa0a890cfa8132156d2959d7e9e2 built-in.o.after.asm
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>
2018-03-03 21:01:12 +08:00
|
|
|
|
2019-09-20 01:37:02 +08:00
|
|
|
if (!(READ_ONCE(cpu_rq(cpu)->membarrier_state) &
|
|
|
|
MEMBARRIER_STATE_GLOBAL_EXPEDITED))
|
|
|
|
continue;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Skip the CPU if it runs a kernel thread. The scheduler
|
|
|
|
* leaves the prior task mm in place as an optimization when
|
|
|
|
* scheduling a kthread.
|
|
|
|
*/
|
2019-09-14 20:34:30 +08:00
|
|
|
p = rcu_dereference(cpu_rq(cpu)->curr);
|
2019-09-20 01:37:02 +08:00
|
|
|
if (p->flags & PF_KTHREAD)
|
|
|
|
continue;
|
|
|
|
|
|
|
|
if (!fallback)
|
|
|
|
__cpumask_set_cpu(cpu, tmpmask);
|
|
|
|
else
|
|
|
|
smp_call_function_single(cpu, ipi_mb, NULL, 1);
|
2018-01-30 04:20:13 +08:00
|
|
|
}
|
2019-09-20 01:37:02 +08:00
|
|
|
rcu_read_unlock();
|
2018-01-30 04:20:13 +08:00
|
|
|
if (!fallback) {
|
|
|
|
preempt_disable();
|
|
|
|
smp_call_function_many(tmpmask, ipi_mb, NULL, 1);
|
|
|
|
preempt_enable();
|
|
|
|
free_cpumask_var(tmpmask);
|
|
|
|
}
|
|
|
|
cpus_read_unlock();
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Memory barrier on the caller thread _after_ we finished
|
|
|
|
* waiting for the last IPI. Matches memory barriers around
|
|
|
|
* rq->curr modification in scheduler.
|
|
|
|
*/
|
|
|
|
smp_mb(); /* exit from system call is not a mb */
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2018-01-30 04:20:17 +08:00
|
|
|
static int membarrier_private_expedited(int flags)
|
membarrier: Provide expedited private command
Implement MEMBARRIER_CMD_PRIVATE_EXPEDITED with IPIs using cpumask built
from all runqueues for which current thread's mm is the same as the
thread calling sys_membarrier. It executes faster than the non-expedited
variant (no blocking). It also works on NOHZ_FULL configurations.
Scheduler-wise, it requires a memory barrier before and after context
switching between processes (which have different mm). The memory
barrier before context switch is already present. For the barrier after
context switch:
* Our TSO archs can do RELEASE without being a full barrier. Look at
x86 spin_unlock() being a regular STORE for example. But for those
archs, all atomics imply smp_mb and all of them have atomic ops in
switch_mm() for mm_cpumask(), and on x86 the CR3 load acts as a full
barrier.
* From all weakly ordered machines, only ARM64 and PPC can do RELEASE,
the rest does indeed do smp_mb(), so there the spin_unlock() is a full
barrier and we're good.
* ARM64 has a very heavy barrier in switch_to(), which suffices.
* PPC just removed its barrier from switch_to(), but appears to be
talking about adding something to switch_mm(). So add a
smp_mb__after_unlock_lock() for now, until this is settled on the PPC
side.
Changes since v3:
- Properly document the memory barriers provided by each architecture.
Changes since v2:
- Address comments from Peter Zijlstra,
- Add smp_mb__after_unlock_lock() after finish_lock_switch() in
finish_task_switch() to add the memory barrier we need after storing
to rq->curr. This is much simpler than the previous approach relying
on atomic_dec_and_test() in mmdrop(), which actually added a memory
barrier in the common case of switching between userspace processes.
- Return -EINVAL when MEMBARRIER_CMD_SHARED is used on a nohz_full
kernel, rather than having the whole membarrier system call returning
-ENOSYS. Indeed, CMD_PRIVATE_EXPEDITED is compatible with nohz_full.
Adapt the CMD_QUERY mask accordingly.
Changes since v1:
- move membarrier code under kernel/sched/ because it uses the
scheduler runqueue,
- only add the barrier when we switch from a kernel thread. The case
where we switch from a user-space thread is already handled by
the atomic_dec_and_test() in mmdrop().
- add a comment to mmdrop() documenting the requirement on the implicit
memory barrier.
CC: Peter Zijlstra <peterz@infradead.org>
CC: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
CC: Boqun Feng <boqun.feng@gmail.com>
CC: Andrew Hunter <ahh@google.com>
CC: Maged Michael <maged.michael@gmail.com>
CC: gromer@google.com
CC: Avi Kivity <avi@scylladb.com>
CC: Benjamin Herrenschmidt <benh@kernel.crashing.org>
CC: Paul Mackerras <paulus@samba.org>
CC: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: Dave Watson <davejwatson@fb.com>
2017-07-29 04:40:40 +08:00
|
|
|
{
|
|
|
|
int cpu;
|
|
|
|
bool fallback = false;
|
|
|
|
cpumask_var_t tmpmask;
|
2019-09-20 01:37:04 +08:00
|
|
|
struct mm_struct *mm = current->mm;
|
membarrier: Provide expedited private command
Implement MEMBARRIER_CMD_PRIVATE_EXPEDITED with IPIs using cpumask built
from all runqueues for which current thread's mm is the same as the
thread calling sys_membarrier. It executes faster than the non-expedited
variant (no blocking). It also works on NOHZ_FULL configurations.
Scheduler-wise, it requires a memory barrier before and after context
switching between processes (which have different mm). The memory
barrier before context switch is already present. For the barrier after
context switch:
* Our TSO archs can do RELEASE without being a full barrier. Look at
x86 spin_unlock() being a regular STORE for example. But for those
archs, all atomics imply smp_mb and all of them have atomic ops in
switch_mm() for mm_cpumask(), and on x86 the CR3 load acts as a full
barrier.
* From all weakly ordered machines, only ARM64 and PPC can do RELEASE,
the rest does indeed do smp_mb(), so there the spin_unlock() is a full
barrier and we're good.
* ARM64 has a very heavy barrier in switch_to(), which suffices.
* PPC just removed its barrier from switch_to(), but appears to be
talking about adding something to switch_mm(). So add a
smp_mb__after_unlock_lock() for now, until this is settled on the PPC
side.
Changes since v3:
- Properly document the memory barriers provided by each architecture.
Changes since v2:
- Address comments from Peter Zijlstra,
- Add smp_mb__after_unlock_lock() after finish_lock_switch() in
finish_task_switch() to add the memory barrier we need after storing
to rq->curr. This is much simpler than the previous approach relying
on atomic_dec_and_test() in mmdrop(), which actually added a memory
barrier in the common case of switching between userspace processes.
- Return -EINVAL when MEMBARRIER_CMD_SHARED is used on a nohz_full
kernel, rather than having the whole membarrier system call returning
-ENOSYS. Indeed, CMD_PRIVATE_EXPEDITED is compatible with nohz_full.
Adapt the CMD_QUERY mask accordingly.
Changes since v1:
- move membarrier code under kernel/sched/ because it uses the
scheduler runqueue,
- only add the barrier when we switch from a kernel thread. The case
where we switch from a user-space thread is already handled by
the atomic_dec_and_test() in mmdrop().
- add a comment to mmdrop() documenting the requirement on the implicit
memory barrier.
CC: Peter Zijlstra <peterz@infradead.org>
CC: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
CC: Boqun Feng <boqun.feng@gmail.com>
CC: Andrew Hunter <ahh@google.com>
CC: Maged Michael <maged.michael@gmail.com>
CC: gromer@google.com
CC: Avi Kivity <avi@scylladb.com>
CC: Benjamin Herrenschmidt <benh@kernel.crashing.org>
CC: Paul Mackerras <paulus@samba.org>
CC: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: Dave Watson <davejwatson@fb.com>
2017-07-29 04:40:40 +08:00
|
|
|
|
2018-01-30 04:20:17 +08:00
|
|
|
if (flags & MEMBARRIER_FLAG_SYNC_CORE) {
|
|
|
|
if (!IS_ENABLED(CONFIG_ARCH_HAS_MEMBARRIER_SYNC_CORE))
|
|
|
|
return -EINVAL;
|
2019-09-20 01:37:04 +08:00
|
|
|
if (!(atomic_read(&mm->membarrier_state) &
|
2018-01-30 04:20:17 +08:00
|
|
|
MEMBARRIER_STATE_PRIVATE_EXPEDITED_SYNC_CORE_READY))
|
|
|
|
return -EPERM;
|
|
|
|
} else {
|
2019-09-20 01:37:04 +08:00
|
|
|
if (!(atomic_read(&mm->membarrier_state) &
|
2018-01-30 04:20:17 +08:00
|
|
|
MEMBARRIER_STATE_PRIVATE_EXPEDITED_READY))
|
|
|
|
return -EPERM;
|
|
|
|
}
|
2017-10-20 01:30:15 +08:00
|
|
|
|
2019-09-20 01:37:04 +08:00
|
|
|
if (atomic_read(&mm->mm_users) == 1 || num_online_cpus() == 1)
|
2017-10-20 01:30:15 +08:00
|
|
|
return 0;
|
membarrier: Provide expedited private command
Implement MEMBARRIER_CMD_PRIVATE_EXPEDITED with IPIs using cpumask built
from all runqueues for which current thread's mm is the same as the
thread calling sys_membarrier. It executes faster than the non-expedited
variant (no blocking). It also works on NOHZ_FULL configurations.
Scheduler-wise, it requires a memory barrier before and after context
switching between processes (which have different mm). The memory
barrier before context switch is already present. For the barrier after
context switch:
* Our TSO archs can do RELEASE without being a full barrier. Look at
x86 spin_unlock() being a regular STORE for example. But for those
archs, all atomics imply smp_mb and all of them have atomic ops in
switch_mm() for mm_cpumask(), and on x86 the CR3 load acts as a full
barrier.
* From all weakly ordered machines, only ARM64 and PPC can do RELEASE,
the rest does indeed do smp_mb(), so there the spin_unlock() is a full
barrier and we're good.
* ARM64 has a very heavy barrier in switch_to(), which suffices.
* PPC just removed its barrier from switch_to(), but appears to be
talking about adding something to switch_mm(). So add a
smp_mb__after_unlock_lock() for now, until this is settled on the PPC
side.
Changes since v3:
- Properly document the memory barriers provided by each architecture.
Changes since v2:
- Address comments from Peter Zijlstra,
- Add smp_mb__after_unlock_lock() after finish_lock_switch() in
finish_task_switch() to add the memory barrier we need after storing
to rq->curr. This is much simpler than the previous approach relying
on atomic_dec_and_test() in mmdrop(), which actually added a memory
barrier in the common case of switching between userspace processes.
- Return -EINVAL when MEMBARRIER_CMD_SHARED is used on a nohz_full
kernel, rather than having the whole membarrier system call returning
-ENOSYS. Indeed, CMD_PRIVATE_EXPEDITED is compatible with nohz_full.
Adapt the CMD_QUERY mask accordingly.
Changes since v1:
- move membarrier code under kernel/sched/ because it uses the
scheduler runqueue,
- only add the barrier when we switch from a kernel thread. The case
where we switch from a user-space thread is already handled by
the atomic_dec_and_test() in mmdrop().
- add a comment to mmdrop() documenting the requirement on the implicit
memory barrier.
CC: Peter Zijlstra <peterz@infradead.org>
CC: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
CC: Boqun Feng <boqun.feng@gmail.com>
CC: Andrew Hunter <ahh@google.com>
CC: Maged Michael <maged.michael@gmail.com>
CC: gromer@google.com
CC: Avi Kivity <avi@scylladb.com>
CC: Benjamin Herrenschmidt <benh@kernel.crashing.org>
CC: Paul Mackerras <paulus@samba.org>
CC: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: Dave Watson <davejwatson@fb.com>
2017-07-29 04:40:40 +08:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Matches memory barriers around rq->curr modification in
|
|
|
|
* scheduler.
|
|
|
|
*/
|
|
|
|
smp_mb(); /* system call entry is not a mb. */
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Expedited membarrier commands guarantee that they won't
|
|
|
|
* block, hence the GFP_NOWAIT allocation flag and fallback
|
|
|
|
* implementation.
|
|
|
|
*/
|
|
|
|
if (!zalloc_cpumask_var(&tmpmask, GFP_NOWAIT)) {
|
|
|
|
/* Fallback for OOM. */
|
|
|
|
fallback = true;
|
|
|
|
}
|
|
|
|
|
|
|
|
cpus_read_lock();
|
2019-09-20 01:37:02 +08:00
|
|
|
rcu_read_lock();
|
membarrier: Provide expedited private command
Implement MEMBARRIER_CMD_PRIVATE_EXPEDITED with IPIs using cpumask built
from all runqueues for which current thread's mm is the same as the
thread calling sys_membarrier. It executes faster than the non-expedited
variant (no blocking). It also works on NOHZ_FULL configurations.
Scheduler-wise, it requires a memory barrier before and after context
switching between processes (which have different mm). The memory
barrier before context switch is already present. For the barrier after
context switch:
* Our TSO archs can do RELEASE without being a full barrier. Look at
x86 spin_unlock() being a regular STORE for example. But for those
archs, all atomics imply smp_mb and all of them have atomic ops in
switch_mm() for mm_cpumask(), and on x86 the CR3 load acts as a full
barrier.
* From all weakly ordered machines, only ARM64 and PPC can do RELEASE,
the rest does indeed do smp_mb(), so there the spin_unlock() is a full
barrier and we're good.
* ARM64 has a very heavy barrier in switch_to(), which suffices.
* PPC just removed its barrier from switch_to(), but appears to be
talking about adding something to switch_mm(). So add a
smp_mb__after_unlock_lock() for now, until this is settled on the PPC
side.
Changes since v3:
- Properly document the memory barriers provided by each architecture.
Changes since v2:
- Address comments from Peter Zijlstra,
- Add smp_mb__after_unlock_lock() after finish_lock_switch() in
finish_task_switch() to add the memory barrier we need after storing
to rq->curr. This is much simpler than the previous approach relying
on atomic_dec_and_test() in mmdrop(), which actually added a memory
barrier in the common case of switching between userspace processes.
- Return -EINVAL when MEMBARRIER_CMD_SHARED is used on a nohz_full
kernel, rather than having the whole membarrier system call returning
-ENOSYS. Indeed, CMD_PRIVATE_EXPEDITED is compatible with nohz_full.
Adapt the CMD_QUERY mask accordingly.
Changes since v1:
- move membarrier code under kernel/sched/ because it uses the
scheduler runqueue,
- only add the barrier when we switch from a kernel thread. The case
where we switch from a user-space thread is already handled by
the atomic_dec_and_test() in mmdrop().
- add a comment to mmdrop() documenting the requirement on the implicit
memory barrier.
CC: Peter Zijlstra <peterz@infradead.org>
CC: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
CC: Boqun Feng <boqun.feng@gmail.com>
CC: Andrew Hunter <ahh@google.com>
CC: Maged Michael <maged.michael@gmail.com>
CC: gromer@google.com
CC: Avi Kivity <avi@scylladb.com>
CC: Benjamin Herrenschmidt <benh@kernel.crashing.org>
CC: Paul Mackerras <paulus@samba.org>
CC: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: Dave Watson <davejwatson@fb.com>
2017-07-29 04:40:40 +08:00
|
|
|
for_each_online_cpu(cpu) {
|
|
|
|
struct task_struct *p;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Skipping the current CPU is OK even through we can be
|
|
|
|
* migrated at any point. The current CPU, at the point
|
|
|
|
* where we read raw_smp_processor_id(), is ensured to
|
|
|
|
* be in program order with respect to the caller
|
|
|
|
* thread. Therefore, we can skip this CPU from the
|
|
|
|
* iteration.
|
|
|
|
*/
|
|
|
|
if (cpu == raw_smp_processor_id())
|
|
|
|
continue;
|
|
|
|
rcu_read_lock();
|
2019-09-14 20:34:30 +08:00
|
|
|
p = rcu_dereference(cpu_rq(cpu)->curr);
|
2019-09-20 01:37:04 +08:00
|
|
|
if (p && p->mm == mm) {
|
membarrier: Provide expedited private command
Implement MEMBARRIER_CMD_PRIVATE_EXPEDITED with IPIs using cpumask built
from all runqueues for which current thread's mm is the same as the
thread calling sys_membarrier. It executes faster than the non-expedited
variant (no blocking). It also works on NOHZ_FULL configurations.
Scheduler-wise, it requires a memory barrier before and after context
switching between processes (which have different mm). The memory
barrier before context switch is already present. For the barrier after
context switch:
* Our TSO archs can do RELEASE without being a full barrier. Look at
x86 spin_unlock() being a regular STORE for example. But for those
archs, all atomics imply smp_mb and all of them have atomic ops in
switch_mm() for mm_cpumask(), and on x86 the CR3 load acts as a full
barrier.
* From all weakly ordered machines, only ARM64 and PPC can do RELEASE,
the rest does indeed do smp_mb(), so there the spin_unlock() is a full
barrier and we're good.
* ARM64 has a very heavy barrier in switch_to(), which suffices.
* PPC just removed its barrier from switch_to(), but appears to be
talking about adding something to switch_mm(). So add a
smp_mb__after_unlock_lock() for now, until this is settled on the PPC
side.
Changes since v3:
- Properly document the memory barriers provided by each architecture.
Changes since v2:
- Address comments from Peter Zijlstra,
- Add smp_mb__after_unlock_lock() after finish_lock_switch() in
finish_task_switch() to add the memory barrier we need after storing
to rq->curr. This is much simpler than the previous approach relying
on atomic_dec_and_test() in mmdrop(), which actually added a memory
barrier in the common case of switching between userspace processes.
- Return -EINVAL when MEMBARRIER_CMD_SHARED is used on a nohz_full
kernel, rather than having the whole membarrier system call returning
-ENOSYS. Indeed, CMD_PRIVATE_EXPEDITED is compatible with nohz_full.
Adapt the CMD_QUERY mask accordingly.
Changes since v1:
- move membarrier code under kernel/sched/ because it uses the
scheduler runqueue,
- only add the barrier when we switch from a kernel thread. The case
where we switch from a user-space thread is already handled by
the atomic_dec_and_test() in mmdrop().
- add a comment to mmdrop() documenting the requirement on the implicit
memory barrier.
CC: Peter Zijlstra <peterz@infradead.org>
CC: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
CC: Boqun Feng <boqun.feng@gmail.com>
CC: Andrew Hunter <ahh@google.com>
CC: Maged Michael <maged.michael@gmail.com>
CC: gromer@google.com
CC: Avi Kivity <avi@scylladb.com>
CC: Benjamin Herrenschmidt <benh@kernel.crashing.org>
CC: Paul Mackerras <paulus@samba.org>
CC: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: Dave Watson <davejwatson@fb.com>
2017-07-29 04:40:40 +08:00
|
|
|
if (!fallback)
|
|
|
|
__cpumask_set_cpu(cpu, tmpmask);
|
|
|
|
else
|
|
|
|
smp_call_function_single(cpu, ipi_mb, NULL, 1);
|
|
|
|
}
|
|
|
|
}
|
2019-09-20 01:37:02 +08:00
|
|
|
rcu_read_unlock();
|
membarrier: Provide expedited private command
Implement MEMBARRIER_CMD_PRIVATE_EXPEDITED with IPIs using cpumask built
from all runqueues for which current thread's mm is the same as the
thread calling sys_membarrier. It executes faster than the non-expedited
variant (no blocking). It also works on NOHZ_FULL configurations.
Scheduler-wise, it requires a memory barrier before and after context
switching between processes (which have different mm). The memory
barrier before context switch is already present. For the barrier after
context switch:
* Our TSO archs can do RELEASE without being a full barrier. Look at
x86 spin_unlock() being a regular STORE for example. But for those
archs, all atomics imply smp_mb and all of them have atomic ops in
switch_mm() for mm_cpumask(), and on x86 the CR3 load acts as a full
barrier.
* From all weakly ordered machines, only ARM64 and PPC can do RELEASE,
the rest does indeed do smp_mb(), so there the spin_unlock() is a full
barrier and we're good.
* ARM64 has a very heavy barrier in switch_to(), which suffices.
* PPC just removed its barrier from switch_to(), but appears to be
talking about adding something to switch_mm(). So add a
smp_mb__after_unlock_lock() for now, until this is settled on the PPC
side.
Changes since v3:
- Properly document the memory barriers provided by each architecture.
Changes since v2:
- Address comments from Peter Zijlstra,
- Add smp_mb__after_unlock_lock() after finish_lock_switch() in
finish_task_switch() to add the memory barrier we need after storing
to rq->curr. This is much simpler than the previous approach relying
on atomic_dec_and_test() in mmdrop(), which actually added a memory
barrier in the common case of switching between userspace processes.
- Return -EINVAL when MEMBARRIER_CMD_SHARED is used on a nohz_full
kernel, rather than having the whole membarrier system call returning
-ENOSYS. Indeed, CMD_PRIVATE_EXPEDITED is compatible with nohz_full.
Adapt the CMD_QUERY mask accordingly.
Changes since v1:
- move membarrier code under kernel/sched/ because it uses the
scheduler runqueue,
- only add the barrier when we switch from a kernel thread. The case
where we switch from a user-space thread is already handled by
the atomic_dec_and_test() in mmdrop().
- add a comment to mmdrop() documenting the requirement on the implicit
memory barrier.
CC: Peter Zijlstra <peterz@infradead.org>
CC: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
CC: Boqun Feng <boqun.feng@gmail.com>
CC: Andrew Hunter <ahh@google.com>
CC: Maged Michael <maged.michael@gmail.com>
CC: gromer@google.com
CC: Avi Kivity <avi@scylladb.com>
CC: Benjamin Herrenschmidt <benh@kernel.crashing.org>
CC: Paul Mackerras <paulus@samba.org>
CC: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: Dave Watson <davejwatson@fb.com>
2017-07-29 04:40:40 +08:00
|
|
|
if (!fallback) {
|
2017-12-16 03:23:10 +08:00
|
|
|
preempt_disable();
|
membarrier: Provide expedited private command
Implement MEMBARRIER_CMD_PRIVATE_EXPEDITED with IPIs using cpumask built
from all runqueues for which current thread's mm is the same as the
thread calling sys_membarrier. It executes faster than the non-expedited
variant (no blocking). It also works on NOHZ_FULL configurations.
Scheduler-wise, it requires a memory barrier before and after context
switching between processes (which have different mm). The memory
barrier before context switch is already present. For the barrier after
context switch:
* Our TSO archs can do RELEASE without being a full barrier. Look at
x86 spin_unlock() being a regular STORE for example. But for those
archs, all atomics imply smp_mb and all of them have atomic ops in
switch_mm() for mm_cpumask(), and on x86 the CR3 load acts as a full
barrier.
* From all weakly ordered machines, only ARM64 and PPC can do RELEASE,
the rest does indeed do smp_mb(), so there the spin_unlock() is a full
barrier and we're good.
* ARM64 has a very heavy barrier in switch_to(), which suffices.
* PPC just removed its barrier from switch_to(), but appears to be
talking about adding something to switch_mm(). So add a
smp_mb__after_unlock_lock() for now, until this is settled on the PPC
side.
Changes since v3:
- Properly document the memory barriers provided by each architecture.
Changes since v2:
- Address comments from Peter Zijlstra,
- Add smp_mb__after_unlock_lock() after finish_lock_switch() in
finish_task_switch() to add the memory barrier we need after storing
to rq->curr. This is much simpler than the previous approach relying
on atomic_dec_and_test() in mmdrop(), which actually added a memory
barrier in the common case of switching between userspace processes.
- Return -EINVAL when MEMBARRIER_CMD_SHARED is used on a nohz_full
kernel, rather than having the whole membarrier system call returning
-ENOSYS. Indeed, CMD_PRIVATE_EXPEDITED is compatible with nohz_full.
Adapt the CMD_QUERY mask accordingly.
Changes since v1:
- move membarrier code under kernel/sched/ because it uses the
scheduler runqueue,
- only add the barrier when we switch from a kernel thread. The case
where we switch from a user-space thread is already handled by
the atomic_dec_and_test() in mmdrop().
- add a comment to mmdrop() documenting the requirement on the implicit
memory barrier.
CC: Peter Zijlstra <peterz@infradead.org>
CC: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
CC: Boqun Feng <boqun.feng@gmail.com>
CC: Andrew Hunter <ahh@google.com>
CC: Maged Michael <maged.michael@gmail.com>
CC: gromer@google.com
CC: Avi Kivity <avi@scylladb.com>
CC: Benjamin Herrenschmidt <benh@kernel.crashing.org>
CC: Paul Mackerras <paulus@samba.org>
CC: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: Dave Watson <davejwatson@fb.com>
2017-07-29 04:40:40 +08:00
|
|
|
smp_call_function_many(tmpmask, ipi_mb, NULL, 1);
|
2017-12-16 03:23:10 +08:00
|
|
|
preempt_enable();
|
membarrier: Provide expedited private command
Implement MEMBARRIER_CMD_PRIVATE_EXPEDITED with IPIs using cpumask built
from all runqueues for which current thread's mm is the same as the
thread calling sys_membarrier. It executes faster than the non-expedited
variant (no blocking). It also works on NOHZ_FULL configurations.
Scheduler-wise, it requires a memory barrier before and after context
switching between processes (which have different mm). The memory
barrier before context switch is already present. For the barrier after
context switch:
* Our TSO archs can do RELEASE without being a full barrier. Look at
x86 spin_unlock() being a regular STORE for example. But for those
archs, all atomics imply smp_mb and all of them have atomic ops in
switch_mm() for mm_cpumask(), and on x86 the CR3 load acts as a full
barrier.
* From all weakly ordered machines, only ARM64 and PPC can do RELEASE,
the rest does indeed do smp_mb(), so there the spin_unlock() is a full
barrier and we're good.
* ARM64 has a very heavy barrier in switch_to(), which suffices.
* PPC just removed its barrier from switch_to(), but appears to be
talking about adding something to switch_mm(). So add a
smp_mb__after_unlock_lock() for now, until this is settled on the PPC
side.
Changes since v3:
- Properly document the memory barriers provided by each architecture.
Changes since v2:
- Address comments from Peter Zijlstra,
- Add smp_mb__after_unlock_lock() after finish_lock_switch() in
finish_task_switch() to add the memory barrier we need after storing
to rq->curr. This is much simpler than the previous approach relying
on atomic_dec_and_test() in mmdrop(), which actually added a memory
barrier in the common case of switching between userspace processes.
- Return -EINVAL when MEMBARRIER_CMD_SHARED is used on a nohz_full
kernel, rather than having the whole membarrier system call returning
-ENOSYS. Indeed, CMD_PRIVATE_EXPEDITED is compatible with nohz_full.
Adapt the CMD_QUERY mask accordingly.
Changes since v1:
- move membarrier code under kernel/sched/ because it uses the
scheduler runqueue,
- only add the barrier when we switch from a kernel thread. The case
where we switch from a user-space thread is already handled by
the atomic_dec_and_test() in mmdrop().
- add a comment to mmdrop() documenting the requirement on the implicit
memory barrier.
CC: Peter Zijlstra <peterz@infradead.org>
CC: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
CC: Boqun Feng <boqun.feng@gmail.com>
CC: Andrew Hunter <ahh@google.com>
CC: Maged Michael <maged.michael@gmail.com>
CC: gromer@google.com
CC: Avi Kivity <avi@scylladb.com>
CC: Benjamin Herrenschmidt <benh@kernel.crashing.org>
CC: Paul Mackerras <paulus@samba.org>
CC: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: Dave Watson <davejwatson@fb.com>
2017-07-29 04:40:40 +08:00
|
|
|
free_cpumask_var(tmpmask);
|
|
|
|
}
|
|
|
|
cpus_read_unlock();
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Memory barrier on the caller thread _after_ we finished
|
|
|
|
* waiting for the last IPI. Matches memory barriers around
|
|
|
|
* rq->curr modification in scheduler.
|
|
|
|
*/
|
|
|
|
smp_mb(); /* exit from system call is not a mb */
|
sched: Clean up and harmonize the coding style of the scheduler code base
A good number of small style inconsistencies have accumulated
in the scheduler core, so do a pass over them to harmonize
all these details:
- fix speling in comments,
- use curly braces for multi-line statements,
- remove unnecessary parentheses from integer literals,
- capitalize consistently,
- remove stray newlines,
- add comments where necessary,
- remove invalid/unnecessary comments,
- align structure definitions and other data types vertically,
- add missing newlines for increased readability,
- fix vertical tabulation where it's misaligned,
- harmonize preprocessor conditional block labeling
and vertical alignment,
- remove line-breaks where they uglify the code,
- add newline after local variable definitions,
No change in functionality:
md5:
1191fa0a890cfa8132156d2959d7e9e2 built-in.o.before.asm
1191fa0a890cfa8132156d2959d7e9e2 built-in.o.after.asm
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>
2018-03-03 21:01:12 +08:00
|
|
|
|
2017-10-20 01:30:15 +08:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2019-09-20 01:37:02 +08:00
|
|
|
static int sync_runqueues_membarrier_state(struct mm_struct *mm)
|
|
|
|
{
|
|
|
|
int membarrier_state = atomic_read(&mm->membarrier_state);
|
|
|
|
cpumask_var_t tmpmask;
|
|
|
|
int cpu;
|
|
|
|
|
|
|
|
if (atomic_read(&mm->mm_users) == 1 || num_online_cpus() == 1) {
|
|
|
|
this_cpu_write(runqueues.membarrier_state, membarrier_state);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* For single mm user, we can simply issue a memory barrier
|
|
|
|
* after setting MEMBARRIER_STATE_GLOBAL_EXPEDITED in the
|
|
|
|
* mm and in the current runqueue to guarantee that no memory
|
|
|
|
* access following registration is reordered before
|
|
|
|
* registration.
|
|
|
|
*/
|
|
|
|
smp_mb();
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!zalloc_cpumask_var(&tmpmask, GFP_KERNEL))
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* For mm with multiple users, we need to ensure all future
|
|
|
|
* scheduler executions will observe @mm's new membarrier
|
|
|
|
* state.
|
|
|
|
*/
|
|
|
|
synchronize_rcu();
|
|
|
|
|
|
|
|
/*
|
|
|
|
* For each cpu runqueue, if the task's mm match @mm, ensure that all
|
|
|
|
* @mm's membarrier state set bits are also set in in the runqueue's
|
|
|
|
* membarrier state. This ensures that a runqueue scheduling
|
|
|
|
* between threads which are users of @mm has its membarrier state
|
|
|
|
* updated.
|
|
|
|
*/
|
|
|
|
cpus_read_lock();
|
|
|
|
rcu_read_lock();
|
|
|
|
for_each_online_cpu(cpu) {
|
|
|
|
struct rq *rq = cpu_rq(cpu);
|
|
|
|
struct task_struct *p;
|
|
|
|
|
|
|
|
p = rcu_dereference(&rq->curr);
|
|
|
|
if (p && p->mm == mm)
|
|
|
|
__cpumask_set_cpu(cpu, tmpmask);
|
|
|
|
}
|
|
|
|
rcu_read_unlock();
|
|
|
|
|
|
|
|
preempt_disable();
|
|
|
|
smp_call_function_many(tmpmask, ipi_sync_rq_state, mm, 1);
|
|
|
|
preempt_enable();
|
|
|
|
|
|
|
|
free_cpumask_var(tmpmask);
|
|
|
|
cpus_read_unlock();
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2018-01-30 04:20:13 +08:00
|
|
|
static int membarrier_register_global_expedited(void)
|
|
|
|
{
|
|
|
|
struct task_struct *p = current;
|
|
|
|
struct mm_struct *mm = p->mm;
|
2019-09-20 01:37:02 +08:00
|
|
|
int ret;
|
2018-01-30 04:20:13 +08:00
|
|
|
|
|
|
|
if (atomic_read(&mm->membarrier_state) &
|
|
|
|
MEMBARRIER_STATE_GLOBAL_EXPEDITED_READY)
|
|
|
|
return 0;
|
|
|
|
atomic_or(MEMBARRIER_STATE_GLOBAL_EXPEDITED, &mm->membarrier_state);
|
2019-09-20 01:37:02 +08:00
|
|
|
ret = sync_runqueues_membarrier_state(mm);
|
|
|
|
if (ret)
|
|
|
|
return ret;
|
2018-01-30 04:20:13 +08:00
|
|
|
atomic_or(MEMBARRIER_STATE_GLOBAL_EXPEDITED_READY,
|
|
|
|
&mm->membarrier_state);
|
sched: Clean up and harmonize the coding style of the scheduler code base
A good number of small style inconsistencies have accumulated
in the scheduler core, so do a pass over them to harmonize
all these details:
- fix speling in comments,
- use curly braces for multi-line statements,
- remove unnecessary parentheses from integer literals,
- capitalize consistently,
- remove stray newlines,
- add comments where necessary,
- remove invalid/unnecessary comments,
- align structure definitions and other data types vertically,
- add missing newlines for increased readability,
- fix vertical tabulation where it's misaligned,
- harmonize preprocessor conditional block labeling
and vertical alignment,
- remove line-breaks where they uglify the code,
- add newline after local variable definitions,
No change in functionality:
md5:
1191fa0a890cfa8132156d2959d7e9e2 built-in.o.before.asm
1191fa0a890cfa8132156d2959d7e9e2 built-in.o.after.asm
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>
2018-03-03 21:01:12 +08:00
|
|
|
|
2018-01-30 04:20:13 +08:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2018-01-30 04:20:17 +08:00
|
|
|
static int membarrier_register_private_expedited(int flags)
|
2017-10-20 01:30:15 +08:00
|
|
|
{
|
|
|
|
struct task_struct *p = current;
|
|
|
|
struct mm_struct *mm = p->mm;
|
2019-09-20 01:37:02 +08:00
|
|
|
int ready_state = MEMBARRIER_STATE_PRIVATE_EXPEDITED_READY,
|
|
|
|
set_state = MEMBARRIER_STATE_PRIVATE_EXPEDITED,
|
|
|
|
ret;
|
2018-01-30 04:20:17 +08:00
|
|
|
|
|
|
|
if (flags & MEMBARRIER_FLAG_SYNC_CORE) {
|
|
|
|
if (!IS_ENABLED(CONFIG_ARCH_HAS_MEMBARRIER_SYNC_CORE))
|
|
|
|
return -EINVAL;
|
2019-09-20 01:37:02 +08:00
|
|
|
ready_state =
|
|
|
|
MEMBARRIER_STATE_PRIVATE_EXPEDITED_SYNC_CORE_READY;
|
2018-01-30 04:20:17 +08:00
|
|
|
}
|
2017-10-20 01:30:15 +08:00
|
|
|
|
|
|
|
/*
|
|
|
|
* We need to consider threads belonging to different thread
|
|
|
|
* groups, which use the same mm. (CLONE_VM but not
|
|
|
|
* CLONE_THREAD).
|
|
|
|
*/
|
2019-09-20 01:37:02 +08:00
|
|
|
if ((atomic_read(&mm->membarrier_state) & ready_state) == ready_state)
|
2018-01-30 04:20:13 +08:00
|
|
|
return 0;
|
2018-01-30 04:20:17 +08:00
|
|
|
if (flags & MEMBARRIER_FLAG_SYNC_CORE)
|
2019-09-20 01:37:02 +08:00
|
|
|
set_state |= MEMBARRIER_STATE_PRIVATE_EXPEDITED_SYNC_CORE;
|
|
|
|
atomic_or(set_state, &mm->membarrier_state);
|
|
|
|
ret = sync_runqueues_membarrier_state(mm);
|
|
|
|
if (ret)
|
|
|
|
return ret;
|
|
|
|
atomic_or(ready_state, &mm->membarrier_state);
|
sched: Clean up and harmonize the coding style of the scheduler code base
A good number of small style inconsistencies have accumulated
in the scheduler core, so do a pass over them to harmonize
all these details:
- fix speling in comments,
- use curly braces for multi-line statements,
- remove unnecessary parentheses from integer literals,
- capitalize consistently,
- remove stray newlines,
- add comments where necessary,
- remove invalid/unnecessary comments,
- align structure definitions and other data types vertically,
- add missing newlines for increased readability,
- fix vertical tabulation where it's misaligned,
- harmonize preprocessor conditional block labeling
and vertical alignment,
- remove line-breaks where they uglify the code,
- add newline after local variable definitions,
No change in functionality:
md5:
1191fa0a890cfa8132156d2959d7e9e2 built-in.o.before.asm
1191fa0a890cfa8132156d2959d7e9e2 built-in.o.after.asm
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>
2018-03-03 21:01:12 +08:00
|
|
|
|
2018-01-30 04:20:13 +08:00
|
|
|
return 0;
|
membarrier: Provide expedited private command
Implement MEMBARRIER_CMD_PRIVATE_EXPEDITED with IPIs using cpumask built
from all runqueues for which current thread's mm is the same as the
thread calling sys_membarrier. It executes faster than the non-expedited
variant (no blocking). It also works on NOHZ_FULL configurations.
Scheduler-wise, it requires a memory barrier before and after context
switching between processes (which have different mm). The memory
barrier before context switch is already present. For the barrier after
context switch:
* Our TSO archs can do RELEASE without being a full barrier. Look at
x86 spin_unlock() being a regular STORE for example. But for those
archs, all atomics imply smp_mb and all of them have atomic ops in
switch_mm() for mm_cpumask(), and on x86 the CR3 load acts as a full
barrier.
* From all weakly ordered machines, only ARM64 and PPC can do RELEASE,
the rest does indeed do smp_mb(), so there the spin_unlock() is a full
barrier and we're good.
* ARM64 has a very heavy barrier in switch_to(), which suffices.
* PPC just removed its barrier from switch_to(), but appears to be
talking about adding something to switch_mm(). So add a
smp_mb__after_unlock_lock() for now, until this is settled on the PPC
side.
Changes since v3:
- Properly document the memory barriers provided by each architecture.
Changes since v2:
- Address comments from Peter Zijlstra,
- Add smp_mb__after_unlock_lock() after finish_lock_switch() in
finish_task_switch() to add the memory barrier we need after storing
to rq->curr. This is much simpler than the previous approach relying
on atomic_dec_and_test() in mmdrop(), which actually added a memory
barrier in the common case of switching between userspace processes.
- Return -EINVAL when MEMBARRIER_CMD_SHARED is used on a nohz_full
kernel, rather than having the whole membarrier system call returning
-ENOSYS. Indeed, CMD_PRIVATE_EXPEDITED is compatible with nohz_full.
Adapt the CMD_QUERY mask accordingly.
Changes since v1:
- move membarrier code under kernel/sched/ because it uses the
scheduler runqueue,
- only add the barrier when we switch from a kernel thread. The case
where we switch from a user-space thread is already handled by
the atomic_dec_and_test() in mmdrop().
- add a comment to mmdrop() documenting the requirement on the implicit
memory barrier.
CC: Peter Zijlstra <peterz@infradead.org>
CC: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
CC: Boqun Feng <boqun.feng@gmail.com>
CC: Andrew Hunter <ahh@google.com>
CC: Maged Michael <maged.michael@gmail.com>
CC: gromer@google.com
CC: Avi Kivity <avi@scylladb.com>
CC: Benjamin Herrenschmidt <benh@kernel.crashing.org>
CC: Paul Mackerras <paulus@samba.org>
CC: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: Dave Watson <davejwatson@fb.com>
2017-07-29 04:40:40 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* sys_membarrier - issue memory barriers on a set of threads
|
|
|
|
* @cmd: Takes command values defined in enum membarrier_cmd.
|
|
|
|
* @flags: Currently needs to be 0. For future extensions.
|
|
|
|
*
|
|
|
|
* If this system call is not implemented, -ENOSYS is returned. If the
|
|
|
|
* command specified does not exist, not available on the running
|
|
|
|
* kernel, or if the command argument is invalid, this system call
|
|
|
|
* returns -EINVAL. For a given command, with flags argument set to 0,
|
2019-09-20 01:37:02 +08:00
|
|
|
* if this system call returns -ENOSYS or -EINVAL, it is guaranteed to
|
|
|
|
* always return the same value until reboot. In addition, it can return
|
|
|
|
* -ENOMEM if there is not enough memory available to perform the system
|
|
|
|
* call.
|
membarrier: Provide expedited private command
Implement MEMBARRIER_CMD_PRIVATE_EXPEDITED with IPIs using cpumask built
from all runqueues for which current thread's mm is the same as the
thread calling sys_membarrier. It executes faster than the non-expedited
variant (no blocking). It also works on NOHZ_FULL configurations.
Scheduler-wise, it requires a memory barrier before and after context
switching between processes (which have different mm). The memory
barrier before context switch is already present. For the barrier after
context switch:
* Our TSO archs can do RELEASE without being a full barrier. Look at
x86 spin_unlock() being a regular STORE for example. But for those
archs, all atomics imply smp_mb and all of them have atomic ops in
switch_mm() for mm_cpumask(), and on x86 the CR3 load acts as a full
barrier.
* From all weakly ordered machines, only ARM64 and PPC can do RELEASE,
the rest does indeed do smp_mb(), so there the spin_unlock() is a full
barrier and we're good.
* ARM64 has a very heavy barrier in switch_to(), which suffices.
* PPC just removed its barrier from switch_to(), but appears to be
talking about adding something to switch_mm(). So add a
smp_mb__after_unlock_lock() for now, until this is settled on the PPC
side.
Changes since v3:
- Properly document the memory barriers provided by each architecture.
Changes since v2:
- Address comments from Peter Zijlstra,
- Add smp_mb__after_unlock_lock() after finish_lock_switch() in
finish_task_switch() to add the memory barrier we need after storing
to rq->curr. This is much simpler than the previous approach relying
on atomic_dec_and_test() in mmdrop(), which actually added a memory
barrier in the common case of switching between userspace processes.
- Return -EINVAL when MEMBARRIER_CMD_SHARED is used on a nohz_full
kernel, rather than having the whole membarrier system call returning
-ENOSYS. Indeed, CMD_PRIVATE_EXPEDITED is compatible with nohz_full.
Adapt the CMD_QUERY mask accordingly.
Changes since v1:
- move membarrier code under kernel/sched/ because it uses the
scheduler runqueue,
- only add the barrier when we switch from a kernel thread. The case
where we switch from a user-space thread is already handled by
the atomic_dec_and_test() in mmdrop().
- add a comment to mmdrop() documenting the requirement on the implicit
memory barrier.
CC: Peter Zijlstra <peterz@infradead.org>
CC: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
CC: Boqun Feng <boqun.feng@gmail.com>
CC: Andrew Hunter <ahh@google.com>
CC: Maged Michael <maged.michael@gmail.com>
CC: gromer@google.com
CC: Avi Kivity <avi@scylladb.com>
CC: Benjamin Herrenschmidt <benh@kernel.crashing.org>
CC: Paul Mackerras <paulus@samba.org>
CC: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: Dave Watson <davejwatson@fb.com>
2017-07-29 04:40:40 +08:00
|
|
|
*
|
|
|
|
* All memory accesses performed in program order from each targeted thread
|
|
|
|
* is guaranteed to be ordered with respect to sys_membarrier(). If we use
|
|
|
|
* the semantic "barrier()" to represent a compiler barrier forcing memory
|
|
|
|
* accesses to be performed in program order across the barrier, and
|
|
|
|
* smp_mb() to represent explicit memory barriers forcing full memory
|
|
|
|
* ordering across the barrier, we have the following ordering table for
|
|
|
|
* each pair of barrier(), sys_membarrier() and smp_mb():
|
|
|
|
*
|
|
|
|
* The pair ordering is detailed as (O: ordered, X: not ordered):
|
|
|
|
*
|
|
|
|
* barrier() smp_mb() sys_membarrier()
|
|
|
|
* barrier() X X O
|
|
|
|
* smp_mb() X O O
|
|
|
|
* sys_membarrier() O O O
|
|
|
|
*/
|
|
|
|
SYSCALL_DEFINE2(membarrier, int, cmd, int, flags)
|
|
|
|
{
|
|
|
|
if (unlikely(flags))
|
|
|
|
return -EINVAL;
|
|
|
|
switch (cmd) {
|
|
|
|
case MEMBARRIER_CMD_QUERY:
|
|
|
|
{
|
|
|
|
int cmd_mask = MEMBARRIER_CMD_BITMASK;
|
|
|
|
|
|
|
|
if (tick_nohz_full_enabled())
|
2018-01-30 04:20:13 +08:00
|
|
|
cmd_mask &= ~MEMBARRIER_CMD_GLOBAL;
|
membarrier: Provide expedited private command
Implement MEMBARRIER_CMD_PRIVATE_EXPEDITED with IPIs using cpumask built
from all runqueues for which current thread's mm is the same as the
thread calling sys_membarrier. It executes faster than the non-expedited
variant (no blocking). It also works on NOHZ_FULL configurations.
Scheduler-wise, it requires a memory barrier before and after context
switching between processes (which have different mm). The memory
barrier before context switch is already present. For the barrier after
context switch:
* Our TSO archs can do RELEASE without being a full barrier. Look at
x86 spin_unlock() being a regular STORE for example. But for those
archs, all atomics imply smp_mb and all of them have atomic ops in
switch_mm() for mm_cpumask(), and on x86 the CR3 load acts as a full
barrier.
* From all weakly ordered machines, only ARM64 and PPC can do RELEASE,
the rest does indeed do smp_mb(), so there the spin_unlock() is a full
barrier and we're good.
* ARM64 has a very heavy barrier in switch_to(), which suffices.
* PPC just removed its barrier from switch_to(), but appears to be
talking about adding something to switch_mm(). So add a
smp_mb__after_unlock_lock() for now, until this is settled on the PPC
side.
Changes since v3:
- Properly document the memory barriers provided by each architecture.
Changes since v2:
- Address comments from Peter Zijlstra,
- Add smp_mb__after_unlock_lock() after finish_lock_switch() in
finish_task_switch() to add the memory barrier we need after storing
to rq->curr. This is much simpler than the previous approach relying
on atomic_dec_and_test() in mmdrop(), which actually added a memory
barrier in the common case of switching between userspace processes.
- Return -EINVAL when MEMBARRIER_CMD_SHARED is used on a nohz_full
kernel, rather than having the whole membarrier system call returning
-ENOSYS. Indeed, CMD_PRIVATE_EXPEDITED is compatible with nohz_full.
Adapt the CMD_QUERY mask accordingly.
Changes since v1:
- move membarrier code under kernel/sched/ because it uses the
scheduler runqueue,
- only add the barrier when we switch from a kernel thread. The case
where we switch from a user-space thread is already handled by
the atomic_dec_and_test() in mmdrop().
- add a comment to mmdrop() documenting the requirement on the implicit
memory barrier.
CC: Peter Zijlstra <peterz@infradead.org>
CC: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
CC: Boqun Feng <boqun.feng@gmail.com>
CC: Andrew Hunter <ahh@google.com>
CC: Maged Michael <maged.michael@gmail.com>
CC: gromer@google.com
CC: Avi Kivity <avi@scylladb.com>
CC: Benjamin Herrenschmidt <benh@kernel.crashing.org>
CC: Paul Mackerras <paulus@samba.org>
CC: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: Dave Watson <davejwatson@fb.com>
2017-07-29 04:40:40 +08:00
|
|
|
return cmd_mask;
|
|
|
|
}
|
2018-01-30 04:20:13 +08:00
|
|
|
case MEMBARRIER_CMD_GLOBAL:
|
|
|
|
/* MEMBARRIER_CMD_GLOBAL is not compatible with nohz_full. */
|
membarrier: Provide expedited private command
Implement MEMBARRIER_CMD_PRIVATE_EXPEDITED with IPIs using cpumask built
from all runqueues for which current thread's mm is the same as the
thread calling sys_membarrier. It executes faster than the non-expedited
variant (no blocking). It also works on NOHZ_FULL configurations.
Scheduler-wise, it requires a memory barrier before and after context
switching between processes (which have different mm). The memory
barrier before context switch is already present. For the barrier after
context switch:
* Our TSO archs can do RELEASE without being a full barrier. Look at
x86 spin_unlock() being a regular STORE for example. But for those
archs, all atomics imply smp_mb and all of them have atomic ops in
switch_mm() for mm_cpumask(), and on x86 the CR3 load acts as a full
barrier.
* From all weakly ordered machines, only ARM64 and PPC can do RELEASE,
the rest does indeed do smp_mb(), so there the spin_unlock() is a full
barrier and we're good.
* ARM64 has a very heavy barrier in switch_to(), which suffices.
* PPC just removed its barrier from switch_to(), but appears to be
talking about adding something to switch_mm(). So add a
smp_mb__after_unlock_lock() for now, until this is settled on the PPC
side.
Changes since v3:
- Properly document the memory barriers provided by each architecture.
Changes since v2:
- Address comments from Peter Zijlstra,
- Add smp_mb__after_unlock_lock() after finish_lock_switch() in
finish_task_switch() to add the memory barrier we need after storing
to rq->curr. This is much simpler than the previous approach relying
on atomic_dec_and_test() in mmdrop(), which actually added a memory
barrier in the common case of switching between userspace processes.
- Return -EINVAL when MEMBARRIER_CMD_SHARED is used on a nohz_full
kernel, rather than having the whole membarrier system call returning
-ENOSYS. Indeed, CMD_PRIVATE_EXPEDITED is compatible with nohz_full.
Adapt the CMD_QUERY mask accordingly.
Changes since v1:
- move membarrier code under kernel/sched/ because it uses the
scheduler runqueue,
- only add the barrier when we switch from a kernel thread. The case
where we switch from a user-space thread is already handled by
the atomic_dec_and_test() in mmdrop().
- add a comment to mmdrop() documenting the requirement on the implicit
memory barrier.
CC: Peter Zijlstra <peterz@infradead.org>
CC: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
CC: Boqun Feng <boqun.feng@gmail.com>
CC: Andrew Hunter <ahh@google.com>
CC: Maged Michael <maged.michael@gmail.com>
CC: gromer@google.com
CC: Avi Kivity <avi@scylladb.com>
CC: Benjamin Herrenschmidt <benh@kernel.crashing.org>
CC: Paul Mackerras <paulus@samba.org>
CC: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: Dave Watson <davejwatson@fb.com>
2017-07-29 04:40:40 +08:00
|
|
|
if (tick_nohz_full_enabled())
|
|
|
|
return -EINVAL;
|
|
|
|
if (num_online_cpus() > 1)
|
2018-07-12 06:36:43 +08:00
|
|
|
synchronize_rcu();
|
membarrier: Provide expedited private command
Implement MEMBARRIER_CMD_PRIVATE_EXPEDITED with IPIs using cpumask built
from all runqueues for which current thread's mm is the same as the
thread calling sys_membarrier. It executes faster than the non-expedited
variant (no blocking). It also works on NOHZ_FULL configurations.
Scheduler-wise, it requires a memory barrier before and after context
switching between processes (which have different mm). The memory
barrier before context switch is already present. For the barrier after
context switch:
* Our TSO archs can do RELEASE without being a full barrier. Look at
x86 spin_unlock() being a regular STORE for example. But for those
archs, all atomics imply smp_mb and all of them have atomic ops in
switch_mm() for mm_cpumask(), and on x86 the CR3 load acts as a full
barrier.
* From all weakly ordered machines, only ARM64 and PPC can do RELEASE,
the rest does indeed do smp_mb(), so there the spin_unlock() is a full
barrier and we're good.
* ARM64 has a very heavy barrier in switch_to(), which suffices.
* PPC just removed its barrier from switch_to(), but appears to be
talking about adding something to switch_mm(). So add a
smp_mb__after_unlock_lock() for now, until this is settled on the PPC
side.
Changes since v3:
- Properly document the memory barriers provided by each architecture.
Changes since v2:
- Address comments from Peter Zijlstra,
- Add smp_mb__after_unlock_lock() after finish_lock_switch() in
finish_task_switch() to add the memory barrier we need after storing
to rq->curr. This is much simpler than the previous approach relying
on atomic_dec_and_test() in mmdrop(), which actually added a memory
barrier in the common case of switching between userspace processes.
- Return -EINVAL when MEMBARRIER_CMD_SHARED is used on a nohz_full
kernel, rather than having the whole membarrier system call returning
-ENOSYS. Indeed, CMD_PRIVATE_EXPEDITED is compatible with nohz_full.
Adapt the CMD_QUERY mask accordingly.
Changes since v1:
- move membarrier code under kernel/sched/ because it uses the
scheduler runqueue,
- only add the barrier when we switch from a kernel thread. The case
where we switch from a user-space thread is already handled by
the atomic_dec_and_test() in mmdrop().
- add a comment to mmdrop() documenting the requirement on the implicit
memory barrier.
CC: Peter Zijlstra <peterz@infradead.org>
CC: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
CC: Boqun Feng <boqun.feng@gmail.com>
CC: Andrew Hunter <ahh@google.com>
CC: Maged Michael <maged.michael@gmail.com>
CC: gromer@google.com
CC: Avi Kivity <avi@scylladb.com>
CC: Benjamin Herrenschmidt <benh@kernel.crashing.org>
CC: Paul Mackerras <paulus@samba.org>
CC: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: Dave Watson <davejwatson@fb.com>
2017-07-29 04:40:40 +08:00
|
|
|
return 0;
|
2018-01-30 04:20:13 +08:00
|
|
|
case MEMBARRIER_CMD_GLOBAL_EXPEDITED:
|
|
|
|
return membarrier_global_expedited();
|
|
|
|
case MEMBARRIER_CMD_REGISTER_GLOBAL_EXPEDITED:
|
|
|
|
return membarrier_register_global_expedited();
|
membarrier: Provide expedited private command
Implement MEMBARRIER_CMD_PRIVATE_EXPEDITED with IPIs using cpumask built
from all runqueues for which current thread's mm is the same as the
thread calling sys_membarrier. It executes faster than the non-expedited
variant (no blocking). It also works on NOHZ_FULL configurations.
Scheduler-wise, it requires a memory barrier before and after context
switching between processes (which have different mm). The memory
barrier before context switch is already present. For the barrier after
context switch:
* Our TSO archs can do RELEASE without being a full barrier. Look at
x86 spin_unlock() being a regular STORE for example. But for those
archs, all atomics imply smp_mb and all of them have atomic ops in
switch_mm() for mm_cpumask(), and on x86 the CR3 load acts as a full
barrier.
* From all weakly ordered machines, only ARM64 and PPC can do RELEASE,
the rest does indeed do smp_mb(), so there the spin_unlock() is a full
barrier and we're good.
* ARM64 has a very heavy barrier in switch_to(), which suffices.
* PPC just removed its barrier from switch_to(), but appears to be
talking about adding something to switch_mm(). So add a
smp_mb__after_unlock_lock() for now, until this is settled on the PPC
side.
Changes since v3:
- Properly document the memory barriers provided by each architecture.
Changes since v2:
- Address comments from Peter Zijlstra,
- Add smp_mb__after_unlock_lock() after finish_lock_switch() in
finish_task_switch() to add the memory barrier we need after storing
to rq->curr. This is much simpler than the previous approach relying
on atomic_dec_and_test() in mmdrop(), which actually added a memory
barrier in the common case of switching between userspace processes.
- Return -EINVAL when MEMBARRIER_CMD_SHARED is used on a nohz_full
kernel, rather than having the whole membarrier system call returning
-ENOSYS. Indeed, CMD_PRIVATE_EXPEDITED is compatible with nohz_full.
Adapt the CMD_QUERY mask accordingly.
Changes since v1:
- move membarrier code under kernel/sched/ because it uses the
scheduler runqueue,
- only add the barrier when we switch from a kernel thread. The case
where we switch from a user-space thread is already handled by
the atomic_dec_and_test() in mmdrop().
- add a comment to mmdrop() documenting the requirement on the implicit
memory barrier.
CC: Peter Zijlstra <peterz@infradead.org>
CC: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
CC: Boqun Feng <boqun.feng@gmail.com>
CC: Andrew Hunter <ahh@google.com>
CC: Maged Michael <maged.michael@gmail.com>
CC: gromer@google.com
CC: Avi Kivity <avi@scylladb.com>
CC: Benjamin Herrenschmidt <benh@kernel.crashing.org>
CC: Paul Mackerras <paulus@samba.org>
CC: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: Dave Watson <davejwatson@fb.com>
2017-07-29 04:40:40 +08:00
|
|
|
case MEMBARRIER_CMD_PRIVATE_EXPEDITED:
|
2018-01-30 04:20:17 +08:00
|
|
|
return membarrier_private_expedited(0);
|
2017-10-20 01:30:15 +08:00
|
|
|
case MEMBARRIER_CMD_REGISTER_PRIVATE_EXPEDITED:
|
2018-01-30 04:20:17 +08:00
|
|
|
return membarrier_register_private_expedited(0);
|
|
|
|
case MEMBARRIER_CMD_PRIVATE_EXPEDITED_SYNC_CORE:
|
|
|
|
return membarrier_private_expedited(MEMBARRIER_FLAG_SYNC_CORE);
|
|
|
|
case MEMBARRIER_CMD_REGISTER_PRIVATE_EXPEDITED_SYNC_CORE:
|
|
|
|
return membarrier_register_private_expedited(MEMBARRIER_FLAG_SYNC_CORE);
|
membarrier: Provide expedited private command
Implement MEMBARRIER_CMD_PRIVATE_EXPEDITED with IPIs using cpumask built
from all runqueues for which current thread's mm is the same as the
thread calling sys_membarrier. It executes faster than the non-expedited
variant (no blocking). It also works on NOHZ_FULL configurations.
Scheduler-wise, it requires a memory barrier before and after context
switching between processes (which have different mm). The memory
barrier before context switch is already present. For the barrier after
context switch:
* Our TSO archs can do RELEASE without being a full barrier. Look at
x86 spin_unlock() being a regular STORE for example. But for those
archs, all atomics imply smp_mb and all of them have atomic ops in
switch_mm() for mm_cpumask(), and on x86 the CR3 load acts as a full
barrier.
* From all weakly ordered machines, only ARM64 and PPC can do RELEASE,
the rest does indeed do smp_mb(), so there the spin_unlock() is a full
barrier and we're good.
* ARM64 has a very heavy barrier in switch_to(), which suffices.
* PPC just removed its barrier from switch_to(), but appears to be
talking about adding something to switch_mm(). So add a
smp_mb__after_unlock_lock() for now, until this is settled on the PPC
side.
Changes since v3:
- Properly document the memory barriers provided by each architecture.
Changes since v2:
- Address comments from Peter Zijlstra,
- Add smp_mb__after_unlock_lock() after finish_lock_switch() in
finish_task_switch() to add the memory barrier we need after storing
to rq->curr. This is much simpler than the previous approach relying
on atomic_dec_and_test() in mmdrop(), which actually added a memory
barrier in the common case of switching between userspace processes.
- Return -EINVAL when MEMBARRIER_CMD_SHARED is used on a nohz_full
kernel, rather than having the whole membarrier system call returning
-ENOSYS. Indeed, CMD_PRIVATE_EXPEDITED is compatible with nohz_full.
Adapt the CMD_QUERY mask accordingly.
Changes since v1:
- move membarrier code under kernel/sched/ because it uses the
scheduler runqueue,
- only add the barrier when we switch from a kernel thread. The case
where we switch from a user-space thread is already handled by
the atomic_dec_and_test() in mmdrop().
- add a comment to mmdrop() documenting the requirement on the implicit
memory barrier.
CC: Peter Zijlstra <peterz@infradead.org>
CC: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
CC: Boqun Feng <boqun.feng@gmail.com>
CC: Andrew Hunter <ahh@google.com>
CC: Maged Michael <maged.michael@gmail.com>
CC: gromer@google.com
CC: Avi Kivity <avi@scylladb.com>
CC: Benjamin Herrenschmidt <benh@kernel.crashing.org>
CC: Paul Mackerras <paulus@samba.org>
CC: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: Dave Watson <davejwatson@fb.com>
2017-07-29 04:40:40 +08:00
|
|
|
default:
|
|
|
|
return -EINVAL;
|
|
|
|
}
|
|
|
|
}
|