rcu/tree: Correctly handle single cpu check in rcu_blocking_is_gp
PATCH 1/1 rcu/tree: Correctly handle single cpu check in rcu_blocking_is_gp
Currently, for non-preempt kernels (with CONFIG_PREEMPTION=n), rcu_blocking_is_gp() checks (with preemption disabled), whether there is only one cpu online. It uses num_online_cpus() to decide whether only one cpu is online. If there is only single cpu online, synchronize_rcu() is optimized to return without doing all the work to wait for grace period. However, there are few issues with the num_online_cpus() check used:
1. num_online_cpus() does a atomic_read(&__num_online_cpus). As hotplug locks are not held, this does not ensure that new incoming cpus update of the count is visible. This can result in read side section on new incoming cpu, observe updates which should not be visible beyond the grace period corresponding to synchronize_rcu().
For e.g. below litmus test, where P0 process corresponds to synchronize_rcu() and P1 corresponds to new online cpu, has positive witnesses; confirming the possibility of read side section to extend before and after the grace period, thereby breaking guarantees provided by synchronize_rcu().
{
int x = 0;
atomic_t numonline = ATOMIC_INIT(1);
}
P0(int *x, atomic_t *numonline)
{
int r0;
WRITE_ONCE(*x, 1);
r0 = atomic_read(numonline);
if (r0 == 1) {
smp_mb();
} else {
synchronize_rcu();
}
WRITE_ONCE(*x, 2);
}
P1(int *x, atomic_t *numonline)
{
int r0; int r1; atomic_inc(numonline);
smp_mb();
rcu_read_lock();
r0 = READ_ONCE(*x);
smp_rmb();
r1 = READ_ONCE(*x);
rcu_read_unlock();}
locations [x;numonline;]
exists (1:r0=0 /\ 1:r1=2)
2. Second problem is, the same early exit, from synchronize_rcu() does not provide full ordering, memory barrier, w.r.t. memory accesses after synchronize_rcu() call.
3. Third, more important issue is related to outgoing cpu. Checking only for __num_online_cpus with preemotion disabled isn't sufficient for RCU, as, on completion of CPUHP_TEARDOWN_CPU stop machine (which clears outgoing cpu from __num_online_cpus, the CPU switches to idle task. So, checking only for __num_online_cpus does not consider RCU read side sections in scheduler code (before switching to idle task) and any potential read side sections in idle task, before final RCU-quiesce entry into cpuhp_report_idle_dead() -> rcu_report_dead().
To handle these issues, add a new rcu_state member n_online_cpus, to keep account of the current number of online cpus. The atomic updates to this counter from rcu_report_dead() and rcu_cpu_starting() and the added read/write memory ordering semantics ensure that synchronize_rcu() fast path waits for all read side sections, where incoming/outgoing cpus are considered online, for RCU i.e. after rcu_cpu_starting() and before rcu_report_dead().
---
Below is the reproducer for issue described in point 3; this snippet is based on klitmus generated test, which is modified to sample reads from idle thread:
static void code0(int* x) {WRITE_ONCE(*x, 1);
idle_ctr = 0;
smp_mb();
mdelay(10);
WRITE_ONCE(*x, 1);
idle_ctr = 1;
synchronize_rcu();
WRITE_ONCE(*x, 2);
idle_ctr = 2;
}
static int thread0(void *_p) {
int _j, _i;
ctx_t *_a = (ctx_t *)_p;
smp_mb();
for (_j = 0 ; _j < stride ; _j++) {
for (_i = _j ; _i < size ; _i += stride) {
while (idle_wait1) {
cpu_relax();
cond_resched();
}
code0(&_a->x[_i]);
smp_mb();
get_online_cpus();
idle_wait1 = true;
put_online_cpus();
} }
atomic_inc(&done);
smp_mb();
wake_up(wq);
smp_mb();
do_exit(0);
}
static void code1(int* x,int* out_1_r1,int* out_1_r0) {
int r0; int r1;
r0 = READ_ONCE(idle_ctr_snap1); r1 = READ_ONCE(idle_ctr_snap2);
*out_1_r1 = (int)r1;
*out_1_r0 = (int)r0;
}
static int thread1(void *_p) {
ctx_t *_a = (ctx_t *)_p;
int _j, _i;
smp_mb();
for (_j = 0 ; _j < stride ; _j++) {
for (_i = _j ; _i < size ; _i += stride) {
while (idle_wait2) {
cpu_relax();
cond_resched();
}
get_online_cpus();
code1(&_a->x[_i],&_a->out_1_r1[_i],&_a->out_1_r0[_i]);
smp_mb();
idle_wait2 = true;
put_online_cpus();
} }
atomic_inc(&done);
smp_mb();
wake_up(wq);
smp_mb();
do_exit(0);
}
Idle thread snippet:
if (cpu_is_offline(cpu)) {
smp_mb();
idle_wait1 = false;
mdelay(8);
smp_mb();
rcu_read_lock();
idle_ctr_snap1 = idle_ctr;
mdelay(40);
smp_rmb();
idle_ctr_snap2 = idle_ctr;
rcu_read_unlock();
smp_mb();
idle_wait2 = false;
tick_nohz_idle_stop_tick();
cpuhp_report_idle_dead();
arch_cpu_idle_dead();kernel/rcu/tree.c | 65 +++++++++++++++++++++++++++++++++++++++++++++++++++++++ kernel/rcu/tree.h | 1 + 2 files changed, 66 insertions(+)
kernel/rcu/tree.c +65 -0
@@ -3609,9 +3609,59 @@ static int rcu_blocking_is_gp(void)if (IS_ENABLED (CONFIG_PREEMPTION ))return rcu_scheduler_active ==RCU_SCHEDULER_INACTIVE ;might_sleep ();/* Check for RCU read-side critical section. */ +/* +* a = p +* a = NULL +* synchronize_rcu() +* rcu_blocking_is_gp() +* num_online_cpus() +* atomic_read(&__num_online_cpus) +* kfree(p); +* +* - VS - +* +* __cpu_up() +* set_cpu_online(cpu) +* atomic_inc(&__num_online_cpus) +* rcu_read_lock() +* rcu_dereference(a) (a == p) +* rcu_read_unlock() +* +* rcu_blocking_is_gp() must observe atomic_inc(&__num_online_cpus), +* in order to ensure that, RCU read side critical section on new +* online cpu, either start after synchronize_rcu()'s GP starts or +* it completes before synchronize_rcu() returns. +* +* However, atomic_read(&__num_online_cpus) does not ensure that. +* +* Essentially, below condition exist: +* +* { +* int x = 0; +* atomic_t numonline = ATOMIC_INIT(1); +* } +* +* P0(int *x, atomic_t *numonline) P1(int *x, atomic_t *numonline) +* { { +* int r0; int r0; int r1; +* WRITE_ONCE(*x, 1); atomic_inc(numonline); +* r0 = atomic_read(numonline); rcu_read_lock(); +* if (r0 == 2) { r0 = READ_ONCE(*x); +* synchronize_rcu(); smp_rmb(); +* } r1 = READ_ONCE(*x); +* WRITE_ONCE(*x, 2); rcu_read_unlock(); +* } } +* +* exists (1:r0=0 /\ 1:r1=2) +* +* atomic_add_return(0, &rcu_state.n_online_cpus) and corresponding +* atomic_inc(&rcu_state.n_online_cpus) in rcu_cpu_starting() corrects +* this ordering issue. +*/ preempt_disable (); ret =num_online_cpus () <=1 ;preempt_enable (); + ret = ret && (atomic_add_return (0 , &rcu_state.n_online_cpus) <=1 );return ret; }
@@ -3655,6 +3705,11 @@ void synchronize_rcu(void)lock_is_held (&rcu_sched_lock_map), "Illegal synchronize_rcu() in RCU read-side critical section");if (rcu_blocking_is_gp ()) +/* +* atomic_add_return() in rcu_blocking_is_gp () provides +* full memory barrier ordering with any rcu section after +* synchronize_rcu() call. +*/ return ;if (rcu_gp_is_expedited ())synchronize_rcu_expedited ();
@@ -4086,6 +4141,10 @@ void rcu_cpu_starting(unsigned int cpu) mask = rdp->grpmask;raw_spin_lock_irqsave_rcu_node (rnp, flags);WRITE_ONCE (rnp->qsmaskinitnext, rnp->qsmaskinitnext | mask); +/* Order with access of n_online_cpus in rcu_blocking_is_gp */ +atomic_inc (&rcu_state.n_online_cpus); +/* Order with rcu-side usages after this */ +smp_mb__after_atomic (); newcpu = !(rnp->expmaskinitnext & mask); rnp->expmaskinitnext |= mask;/* Allow lockless access for expedited grace periods. */
@@ -4138,6 +4197,12 @@ void rcu_report_dead(unsigned int cpu)raw_spin_lock_irqsave_rcu_node (rnp, flags); }WRITE_ONCE (rnp->qsmaskinitnext, rnp->qsmaskinitnext & ~mask); +/* +* Order with access of n_online_cpus in rcu_blocking_is_gp(). +* Release semantics ensures that RCU read sections before it +* are observed by rcu_blocking_is_gp(). +*/ +atomic_dec_return_release (&rcu_state.n_online_cpus);raw_spin_unlock_irqrestore_rcu_node (rnp, flags);raw_spin_unlock (&rcu_state.ofl_lock);
kernel/rcu/tree.h +1 -0
@@ -298,6 +298,7 @@ struct rcu_state {/* Hierarchy levels (+1 to */ /* shut bogus gcc warning) */ int ncpus;/* # CPUs seen so far. */ +atomic_t n_online_cpus;/* # CPUs online for RCU. */ /* The following fields are guarded by the root rcu_node's lock. */
Signature · 2 lines
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
Paul E. McKenney
maintainerQuoted 7 lines · Neeraj Upadhyay
Currently, for non-preempt kernels (with
CONFIG_PREEMPTION=n),rcu_blocking_is_gp()checks (with preemption disabled), whether there is only one cpu online. It usesnum_online_cpus()to decide whether only one cpu is online. If there is only single cpu online,synchronize_rcu()is optimized to return without doing all the work to wait for grace period. However, there are few issues with thenum_online_cpus()check used:
Great catch!!!
I do have some questions about your suggested fix, though.
Quoted 223 lines
1.
num_online_cpus()does aatomic_read(&__num_online_cpus). As hotplug locks are not held, this does not ensure that new incoming cpus update of the count is visible. This can result in read side section on new incoming cpu, observe updates which should not be visible beyond the grace period corresponding tosynchronize_rcu().For e.g. below litmus test, where P0 process corresponds to synchronize_rcu() and P1 corresponds to new online cpu, has positive witnesses; confirming the possibility of read side section to extend before and after the grace period, thereby breaking guarantees provided by synchronize_rcu(). { int x = 0; atomic_t numonline = ATOMIC_INIT(1); } P0(int *x, atomic_t *numonline) { int r0; WRITE_ONCE(*x, 1); r0 = atomic_read(numonline); if (r0 == 1) { smp_mb(); } else { synchronize_rcu(); } WRITE_ONCE(*x, 2); } P1(int *x, atomic_t *numonline) { int r0; int r1; atomic_inc(numonline); smp_mb(); rcu_read_lock(); r0 = READ_ONCE(*x); smp_rmb(); r1 = READ_ONCE(*x); rcu_read_unlock(); } locations [x;numonline;] exists (1:r0=0 /\ 1:r1=2)2. Second problem is, the same early exit, from
synchronize_rcu()does not provide full ordering, memory barrier, w.r.t. memory accesses aftersynchronize_rcu()call.3. Third, more important issue is related to outgoing cpu. Checking only for
__num_online_cpuswith preemotion disabled isn't sufficient for RCU, as, on completion ofCPUHP_TEARDOWN_CPUstop machine (which clears outgoing cpu from__num_online_cpus, the CPU switches to idle task. So, checking only for__num_online_cpusdoes not consider RCU read side sections in scheduler code (before switching to idle task) and any potential read side sections in idle task, before final RCU-quiesce entry intocpuhp_report_idle_dead()->rcu_report_dead().To handle these issues, add a new
rcu_statemembern_online_cpus, to keep account of the current number of online cpus. The atomic updates to this counter fromrcu_report_dead()andrcu_cpu_starting()and the added read/write memory ordering semantics ensure thatsynchronize_rcu()fast path waits for all read side sections, where incoming/outgoing cpus are considered online, for RCU i.e. afterrcu_cpu_starting()and beforercu_report_dead().Signed-off-by Neeraj Upadhyay <[email protected]>---
Below is the reproducer for issue described in point 3; this snippet is based on klitmus generated test, which is modified to sample reads from idle thread:
static void
code0(int* x){WRITE_ONCE(*x, 1);
idle_ctr= 0;
smp_mb();
mdelay(10);
WRITE_ONCE(*x, 1);
idle_ctr= 1;
synchronize_rcu();
WRITE_ONCE(*x, 2);
idle_ctr= 2;}
static int
thread0(void *_p){
int _j, _i;
ctx_t*_a = (ctx_t*)_p;
smp_mb();for (_j = 0 ; _j < stride ; _j++) { for (_i = _j ; _i < size ; _i += stride) { while (idle_wait1) { cpu_relax(); cond_resched(); } code0(&_a->x[_i]); smp_mb(); get_online_cpus(); idle_wait1 = true; put_online_cpus(); }}
atomic_inc(&done);
smp_mb();
wake_up(wq);
smp_mb();
do_exit(0);
}static void
code1(int* x,int* out_1_r1,int* out_1_r0){int r0; int r1;
r0 = READ_ONCE(idle_ctr_snap1); r1 = READ_ONCE(idle_ctr_snap2);
*out_1_r1= (int)r1;
*out_1_r0= (int)r0;
}static int
thread1(void *_p){
ctx_t*_a = (ctx_t*)_p;
int _j, _i;
smp_mb();for (_j = 0 ; _j < stride ; _j++) { for (_i = _j ; _i < size ; _i += stride) { while (idle_wait2) { cpu_relax(); cond_resched(); } get_online_cpus(); code1(&_a->x[_i],&_a->out_1_r1[_i],&_a->out_1_r0[_i]); smp_mb(); idle_wait2 = true; put_online_cpus(); }}
atomic_inc(&done);
smp_mb();
wake_up(wq);
smp_mb();
do_exit(0);
}Idle thread snippet:
if (cpu_is_offline(cpu)) { smp_mb(); idle_wait1 = false; mdelay(8); smp_mb(); rcu_read_lock(); idle_ctr_snap1 = idle_ctr; mdelay(40); smp_rmb(); idle_ctr_snap2 = idle_ctr; rcu_read_unlock(); smp_mb(); idle_wait2 = false; tick_nohz_idle_stop_tick(); cpuhp_report_idle_dead(); arch_cpu_idle_dead();kernel/rcu/tree.c | 65 +++++++++++++++++++++++++++++++++++++++++++++++++++++++ kernel/rcu/tree.h | 1 + 2 files changed, 66 insertions(+)kernel/rcu/tree.c +41
-0@@ -3609,9 +3609,59 @@ static int rcu_blocking_is_gp(void)if (IS_ENABLED (CONFIG_PREEMPTION ))return rcu_scheduler_active ==RCU_SCHEDULER_INACTIVE ;might_sleep ();/* Check for RCU read-side critical section. */ +/* +* a = p +* a = NULL +* synchronize_rcu() +* rcu_blocking_is_gp() +* num_online_cpus() +* atomic_read(&__num_online_cpus) +* kfree(p); +* +* - VS - +* +* __cpu_up() +* set_cpu_online(cpu) +* atomic_inc(&__num_online_cpus) +* rcu_read_lock() +* rcu_dereference(a) (a == p) +* rcu_read_unlock() +* +* rcu_blocking_is_gp() must observe atomic_inc(&__num_online_cpus), +* in order to ensure that, RCU read side critical section on new +* online cpu, either start after synchronize_rcu()'s GP starts or +* it completes before synchronize_rcu() returns. +* +* However, atomic_read(&__num_online_cpus) does not ensure that. +* +* Essentially, below condition exist: +* +* { +* int x = 0; +* atomic_t numonline = ATOMIC_INIT(1); +* } +* +* P0(int *x, atomic_t *numonline) P1(int *x, atomic_t *numonline) +* { { +* int r0; int r0; int r1; +* WRITE_ONCE(*x, 1); atomic_inc(numonline); +* r0 = atomic_read(numonline); rcu_read_lock(); +* if (r0 == 2) { r0 = READ_ONCE(*x); +* synchronize_rcu(); smp_rmb(); +* } r1 = READ_ONCE(*x); +* WRITE_ONCE(*x, 2); rcu_read_unlock();
kernel/rcu/tree.c static int rcu_blocking_is_gp(void) · line 3662
Quoted 10 lines
kernel/rcu/tree.c +8
-0+* } } +* +* exists (1:r0=0 /\ 1:r1=2) +* +* atomic_add_return(0, &rcu_state.n_online_cpus) and corresponding +* atomic_inc(&rcu_state.n_online_cpus) in rcu_cpu_starting() corrects +* this ordering issue. +*/ preempt_disable (); ret =num_online_cpus () <=1 ;
Here I assume that rcu_state.n_online_cpus is incremented early in the CPU-hotplug CPU-online process, that is, on one of the CPUs that was running prior to the new CPU coming online. (The problem with the existing code is not the lack of ordering, but rather that the changes to the number of online CPUs happen in places that are not helpful to synchronize_rcu().)
If rcu_state.n_online_cpus is equal to one at any point in this region of code, there is only one CPU, and that CPU sees all prior accesses made by any CPU that was online at the time of its access. Furthermore, if rcu_state.n_online_cpus is equal to one, its value cannot change until after the preempt_enable() below.
Furthermore, if n_online_cpus is equal to one here, all later CPUs (both this one and any that come online later on) are guaranteed to see all accesses by any CPU prior to this point in the code, and without added memory barriers. Those memory barriers have to be present in the CPU-hotplug code or lots of things would break.
On the other hand, if n_online_cpus is greater than one, then we will be using the heavyweight call to synchronize_rcu(), which will guarantee all the ordering we need. (Please refer to the rather lengthy header comment for synchronize_rcu().)
So if you access rcu_state.n_online_cpus with preemption disabled, READ_ONCE() suffices and no memory barriers are required.
kernel/rcu/tree.c static int rcu_blocking_is_gp(void) · line 3663
Quoted 1 line
kernel/rcu/tree.c +0
-0preempt_enable ();
And we only get to this point in the code when CONFIG_PREEMPT_NONE=y, so the preempt_disable() and preempt_enable() are optional. Though they can be argued to be useful documentation. Or maybe not...
kernel/rcu/tree.c void synchronize_rcu(void) · line 3712
Quoted 13 lines
kernel/rcu/tree.c +6
-0+ ret = ret && (atomic_add_return (0 , &rcu_state.n_online_cpus) <=1 );return ret; }@@ -3655,6 +3705,11 @@ void synchronize_rcu(void)lock_is_held (&rcu_sched_lock_map), "Illegal synchronize_rcu() in RCU read-side critical section");if (rcu_blocking_is_gp ()) +/* +* atomic_add_return() in rcu_blocking_is_gp () provides +* full memory barrier ordering with any rcu section after +* synchronize_rcu() call. +*/
Given your fix of having RCU keep its own count of the number of online CPUs, no additional ordering is required. Either synchronize_rcu() provides what is required or we are in single-CPU state, meaning we don't need any ordering.
kernel/rcu/tree.c void rcu_cpu_starting(unsigned int cpu) · line 4147
Quoted 11 lines
kernel/rcu/tree.c +4
-0return ;if (rcu_gp_is_expedited ())synchronize_rcu_expedited ();@@ -4086,6 +4141,10 @@ void rcu_cpu_starting(unsigned int cpu) mask = rdp->grpmask;raw_spin_lock_irqsave_rcu_node (rnp, flags);WRITE_ONCE (rnp->qsmaskinitnext, rnp->qsmaskinitnext | mask); +/* Order with access of n_online_cpus in rcu_blocking_is_gp */ +atomic_inc (&rcu_state.n_online_cpus); +/* Order with rcu-side usages after this */ +smp_mb__after_atomic ();
Ah, here is the problem. Please instead put the increment in rcutree_prepare_cpu(), which in the one-to-two transition will be running on the single CPU in the system, thus avoiding the need for ordering. Yes, this will result in unnecessary calls to synchronize_rcu() during the CPU-online process, but who cares? ;-)
kernel/rcu/tree.c void rcu_report_dead(unsigned int cpu) · line 4205
Quoted 13 lines
kernel/rcu/tree.c +6
-0newcpu = !(rnp->expmaskinitnext & mask); rnp->expmaskinitnext |= mask;/* Allow lockless access for expedited grace periods. */ @@ -4138,6 +4197,12 @@ void rcu_report_dead(unsigned int cpu)raw_spin_lock_irqsave_rcu_node (rnp, flags); }WRITE_ONCE (rnp->qsmaskinitnext, rnp->qsmaskinitnext & ~mask); +/* +* Order with access of n_online_cpus in rcu_blocking_is_gp(). +* Release semantics ensures that RCU read sections before it +* are observed by rcu_blocking_is_gp(). +*/ +atomic_dec_return_release (&rcu_state.n_online_cpus);
Similarly, please put this decrement into rcutree_dead_cpu(), which runs on one of the remaining CPUs after the outgoing CPU is long gone. In the two-to-one transition, this will run on the single remaining CPU in the system, thus avoiding the need for ordering. Again, yes, this will result in unnecessary calls to synchronize_rcu() during the CPU-online process, but again who cares?
kernel/rcu/tree.h struct rcu_state { · line 301
Quoted 12 lines
kernel/rcu/tree.c +0
-0raw_spin_unlock_irqrestore_rcu_node (rnp, flags);raw_spin_unlock (&rcu_state.ofl_lock);kernel/rcu/tree.h +1
-0@@ -298,6 +298,7 @@ struct rcu_state {/* Hierarchy levels (+1 to */ /* shut bogus gcc warning) */ int ncpus;/* # CPUs seen so far. */ +atomic_t n_online_cpus;/* # CPUs online for RCU. */
With those changes in place, this can be just an int. The increments and decrements can use normal C-language loads and WRITE_ONCE() for the stores. The trick is that this value will only change from one to two (and vice versa) when there is only one online CPU.
And the num_online_cpus() can be replaced with a READ_ONCE().
Does this make sense, or am I missing something?
Thanx, Paul
Quoted 6 lines
kernel/rcu/tree.h +0
-1/* The following fields are guarded by the root rcu_node's lock. */ --The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
Hi Paul,
Quoted 271 lines · Paul E. McKenney
Quoted 7 lines · Neeraj Upadhyay
Currently, for non-preempt kernels (with
CONFIG_PREEMPTION=n),rcu_blocking_is_gp()checks (with preemption disabled), whether there is only one cpu online. It usesnum_online_cpus()to decide whether only one cpu is online. If there is only single cpu online,synchronize_rcu()is optimized to return without doing all the work to wait for grace period. However, there are few issues with thenum_online_cpus()check used:Great catch!!!
I do have some questions about your suggested fix, though.
Quoted 223 lines
1.
num_online_cpus()does aatomic_read(&__num_online_cpus). Ashotplug locks are not held, this does not ensure that new incoming cpus update of the count is visible. This can result in read side section on new incoming cpu, observe updates which should not be visible beyond the grace period corresponding to synchronize_rcu(). For e.g. below litmus test, where P0 process corresponds to synchronize_rcu() and P1 corresponds to new online cpu, has positive witnesses; confirming the possibility of read side section to extend before and after the grace period, thereby breaking guarantees provided by synchronize_rcu(). { int x = 0; atomic_t numonline = ATOMIC_INIT(1); } P0(int *x, atomic_t *numonline) { int r0; WRITE_ONCE(*x, 1); r0 = atomic_read(numonline); if (r0 == 1) { smp_mb(); } else { synchronize_rcu(); } WRITE_ONCE(*x, 2); } P1(int *x, atomic_t *numonline) { int r0; int r1; atomic_inc(numonline); smp_mb(); rcu_read_lock(); r0 = READ_ONCE(*x); smp_rmb(); r1 = READ_ONCE(*x); rcu_read_unlock(); } locations [x;numonline;] exists (1:r0=0 /\ 1:r1=2)2. Second problem is, the same early exit, from
synchronize_rcu()does not provide full ordering, memory barrier, w.r.t. memory accesses aftersynchronize_rcu()call.3. Third, more important issue is related to outgoing cpu. Checking only for
__num_online_cpuswith preemotion disabled isn't sufficient for RCU, as, on completion ofCPUHP_TEARDOWN_CPUstop machine (which clears outgoing cpu from__num_online_cpus, the CPU switches to idle task. So, checking only for__num_online_cpusdoes not consider RCU read side sections in scheduler code (before switching to idle task) and any potential read side sections in idle task, before final RCU-quiesce entry intocpuhp_report_idle_dead()->rcu_report_dead().To handle these issues, add a new
rcu_statemembern_online_cpus, to keep account of the current number of online cpus. The atomic updates to this counter fromrcu_report_dead()andrcu_cpu_starting()and the added read/write memory ordering semantics ensure thatsynchronize_rcu()fast path waits for all read side sections, where incoming/outgoing cpus are considered online, for RCU i.e. afterrcu_cpu_starting()and beforercu_report_dead().Signed-off-by Neeraj Upadhyay <[email protected]>---
Below is the reproducer for issue described in point 3; this snippet is based on klitmus generated test, which is modified to sample reads from idle thread:
static void
code0(int* x){WRITE_ONCE(*x, 1); idle_ctr = 0; smp_mb(); mdelay(10); WRITE_ONCE(*x, 1); idle_ctr = 1; synchronize_rcu(); WRITE_ONCE(*x, 2); idle_ctr = 2;}
static int thread0(void *_p) { int _j, _i; ctx_t *_a = (ctx_t *)_p; smp_mb(); for (_j = 0 ; _j < stride ; _j++) { for (_i = _j ; _i < size ; _i += stride) { while (idle_wait1) { cpu_relax(); cond_resched(); } code0(&_a->x[_i]); smp_mb(); get_online_cpus(); idle_wait1 = true; put_online_cpus(); } } atomic_inc(&done); smp_mb(); wake_up(wq); smp_mb(); do_exit(0); }static void
code1(int* x,int* out_1_r1,int* out_1_r0){int r0; int r1; r0 = READ_ONCE(idle_ctr_snap1); r1 = READ_ONCE(idle_ctr_snap2); *out_1_r1 = (int)r1; *out_1_r0 = (int)r0; }static int thread1(void *_p) { ctx_t *_a = (ctx_t *)_p; int _j, _i; smp_mb(); for (_j = 0 ; _j < stride ; _j++) { for (_i = _j ; _i < size ; _i += stride) { while (idle_wait2) { cpu_relax(); cond_resched(); } get_online_cpus(); code1(&_a->x[_i],&_a->out_1_r1[_i],&_a->out_1_r0[_i]); smp_mb(); idle_wait2 = true; put_online_cpus(); } } atomic_inc(&done); smp_mb(); wake_up(wq); smp_mb(); do_exit(0); }Idle thread snippet:
if (cpu_is_offline(cpu)) { smp_mb(); idle_wait1 = false; mdelay(8); smp_mb(); rcu_read_lock(); idle_ctr_snap1 = idle_ctr; mdelay(40); smp_rmb(); idle_ctr_snap2 = idle_ctr; rcu_read_unlock(); smp_mb(); idle_wait2 = false; tick_nohz_idle_stop_tick(); cpuhp_report_idle_dead(); arch_cpu_idle_dead();kernel/rcu/tree.c | 65 +++++++++++++++++++++++++++++++++++++++++++++++++++++++ kernel/rcu/tree.h | 1 + 2 files changed, 66 insertions(+)kernel/rcu/tree.c +41
-0@@ -3609,9 +3609,59 @@ static int rcu_blocking_is_gp(void)if (IS_ENABLED (CONFIG_PREEMPTION ))return rcu_scheduler_active ==RCU_SCHEDULER_INACTIVE ;might_sleep ();/* Check for RCU read-side critical section. */ +/* +* a = p +* a = NULL +* synchronize_rcu() +* rcu_blocking_is_gp() +* num_online_cpus() +* atomic_read(&__num_online_cpus) +* kfree(p); +* +* - VS - +* +* __cpu_up() +* set_cpu_online(cpu) +* atomic_inc(&__num_online_cpus) +* rcu_read_lock() +* rcu_dereference(a) (a == p) +* rcu_read_unlock() +* +* rcu_blocking_is_gp() must observe atomic_inc(&__num_online_cpus), +* in order to ensure that, RCU read side critical section on new +* online cpu, either start after synchronize_rcu()'s GP starts or +* it completes before synchronize_rcu() returns. +* +* However, atomic_read(&__num_online_cpus) does not ensure that. +* +* Essentially, below condition exist: +* +* { +* int x = 0; +* atomic_t numonline = ATOMIC_INIT(1); +* } +* +* P0(int *x, atomic_t *numonline) P1(int *x, atomic_t *numonline) +* { { +* int r0; int r0; int r1; +* WRITE_ONCE(*x, 1); atomic_inc(numonline); +* r0 = atomic_read(numonline); rcu_read_lock(); +* if (r0 == 2) { r0 = READ_ONCE(*x); +* synchronize_rcu(); smp_rmb(); +* } r1 = READ_ONCE(*x); +* WRITE_ONCE(*x, 2); rcu_read_unlock(); kernel/rcu/tree.c static int rcu_blocking_is_gp(void) · line 3662
Quoted 10 lines
kernel/rcu/tree.c +8
-0+* } } +* +* exists (1:r0=0 /\ 1:r1=2) +* +* atomic_add_return(0, &rcu_state.n_online_cpus) and corresponding +* atomic_inc(&rcu_state.n_online_cpus) in rcu_cpu_starting() corrects +* this ordering issue. +*/ preempt_disable (); ret =num_online_cpus () <=1 ;Here I assume that rcu_state.n_online_cpus is incremented early in the CPU-hotplug CPU-online process, that is, on one of the CPUs that was running prior to the new CPU coming online. (The problem with the existing code is not the lack of ordering, but rather that the changes to the number of online CPUs happen in places that are not helpful to
synchronize_rcu().)If rcu_state.n_online_cpus is equal to one at any point in this region of code, there is only one CPU, and that CPU sees all prior accesses made by any CPU that was online at the time of its access. Furthermore, if rcu_state.n_online_cpus is equal to one, its value cannot change until after the
preempt_enable()below.Furthermore, if
n_online_cpusis equal to one here, all later CPUs (both this one and any that come online later on) are guaranteed to see all accesses by any CPU prior to this point in the code, and without added memory barriers. Those memory barriers have to be present in the CPU-hotplug code or lots of things would break.On the other hand, if
n_online_cpusis greater than one, then we will be using the heavyweight call tosynchronize_rcu(), which will guarantee all the ordering we need. (Please refer to the rather lengthy header comment forsynchronize_rcu().)
Quoted 9 lines
So if you access rcu_state.n_online_cpus with preemption disabled,
READ_ONCE()suffices and no memory barriers are required.kernel/rcu/tree.c static int rcu_blocking_is_gp(void) · line 3663
Quoted 1 line
kernel/rcu/tree.c +0
-0preempt_enable ();And we only get to this point in the code when
CONFIG_PREEMPT_NONE=y, so thepreempt_disable()andpreempt_enable()are optional. Though they can be argued to be useful documentation. Or maybe not...
I also noticed it. For CONFIG_PREEMPTION=n, preempt_disable() and preempt_enable() adds barrier(); I thought that was required for the case where num online cpus <= 1?
Quoted 76 lines
kernel/rcu/tree.c void synchronize_rcu(void) · line 3712
Quoted 13 lines
kernel/rcu/tree.c +6
-0+ ret = ret && (atomic_add_return (0 , &rcu_state.n_online_cpus) <=1 );return ret; }@@ -3655,6 +3705,11 @@ void synchronize_rcu(void)lock_is_held (&rcu_sched_lock_map), "Illegal synchronize_rcu() in RCU read-side critical section");if (rcu_blocking_is_gp ()) +/* +* atomic_add_return() in rcu_blocking_is_gp () provides +* full memory barrier ordering with any rcu section after +* synchronize_rcu() call. +*/ Given your fix of having RCU keep its own count of the number of online CPUs, no additional ordering is required. Either
synchronize_rcu()provides what is required or we are in single-CPU state, meaning we don't need any ordering.kernel/rcu/tree.c void rcu_cpu_starting(unsigned int cpu) · line 4147
Quoted 11 lines
kernel/rcu/tree.c +4
-0return ;if (rcu_gp_is_expedited ())synchronize_rcu_expedited ();@@ -4086,6 +4141,10 @@ void rcu_cpu_starting(unsigned int cpu) mask = rdp->grpmask;raw_spin_lock_irqsave_rcu_node (rnp, flags);WRITE_ONCE (rnp->qsmaskinitnext, rnp->qsmaskinitnext | mask); +/* Order with access of n_online_cpus in rcu_blocking_is_gp */ +atomic_inc (&rcu_state.n_online_cpus); +/* Order with rcu-side usages after this */ +smp_mb__after_atomic ();Ah, here is the problem. Please instead put the increment in
rcutree_prepare_cpu(), which in the one-to-two transition will be running on the single CPU in the system, thus avoiding the need for ordering. Yes, this will result in unnecessary calls tosynchronize_rcu()during the CPU-online process, but who cares? ;-)kernel/rcu/tree.c void rcu_report_dead(unsigned int cpu) · line 4205
Quoted 13 lines
kernel/rcu/tree.c +6
-0newcpu = !(rnp->expmaskinitnext & mask); rnp->expmaskinitnext |= mask;/* Allow lockless access for expedited grace periods. */ @@ -4138,6 +4197,12 @@ void rcu_report_dead(unsigned int cpu)raw_spin_lock_irqsave_rcu_node (rnp, flags); }WRITE_ONCE (rnp->qsmaskinitnext, rnp->qsmaskinitnext & ~mask); +/* +* Order with access of n_online_cpus in rcu_blocking_is_gp(). +* Release semantics ensures that RCU read sections before it +* are observed by rcu_blocking_is_gp(). +*/ +atomic_dec_return_release (&rcu_state.n_online_cpus);Similarly, please put this decrement into
rcutree_dead_cpu(), which runs on one of the remaining CPUs after the outgoing CPU is long gone. In the two-to-one transition, this will run on the single remaining CPU in the system, thus avoiding the need for ordering. Again, yes, this will result in unnecessary calls tosynchronize_rcu()during the CPU-online process, but again who cares?kernel/rcu/tree.h struct rcu_state { · line 301
Quoted 12 lines
kernel/rcu/tree.c +0
-0raw_spin_unlock_irqrestore_rcu_node (rnp, flags);raw_spin_unlock (&rcu_state.ofl_lock);kernel/rcu/tree.h +1
-0@@ -298,6 +298,7 @@ struct rcu_state {/* Hierarchy levels (+1 to */ /* shut bogus gcc warning) */ int ncpus;/* # CPUs seen so far. */ +atomic_t n_online_cpus;/* # CPUs online for RCU. */ With those changes in place, this can be just an int. The increments and decrements can use normal C-language loads and
WRITE_ONCE()for the stores. The trick is that this value will only change from one to two (and vice versa) when there is only one online CPU.
Quoted 6 lines
And the
num_online_cpus()can be replaced with aREAD_ONCE().Does this make sense, or am I missing something?
Thanx, Paul
Yes, this makes sense; thanks for the details! Will post v2.
Thanks
Neeraj
Quoted 6 lines
Quoted 6 lines
kernel/rcu/tree.h +0
-1/* The following fields are guarded by the root rcu_node's lock. */ --The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
Signature · 1 line
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, hosted by The Linux Foundation
Quoted 47 lines · Neeraj Upadhyay
Currently, for non-preempt kernels (with
CONFIG_PREEMPTION=n),rcu_blocking_is_gp()checks (with preemption disabled), whether there is only one cpu online. It usesnum_online_cpus()to decide whether only one cpu is online. If there is only single cpu online,synchronize_rcu()is optimized to return without doing all the work to wait for grace period. However, there are few issues with thenum_online_cpus()check used:1.
num_online_cpus()does aatomic_read(&__num_online_cpus). As hotplug locks are not held, this does not ensure that new incoming cpus update of the count is visible. This can result in read side section on new incoming cpu, observe updates which should not be visible beyond the grace period corresponding tosynchronize_rcu().For e.g. below litmus test, where P0 process corresponds to synchronize_rcu() and P1 corresponds to new online cpu, has positive witnesses; confirming the possibility of read side section to extend before and after the grace period, thereby breaking guarantees provided by synchronize_rcu(). { int x = 0; atomic_t numonline = ATOMIC_INIT(1); } P0(int *x, atomic_t *numonline) { int r0; WRITE_ONCE(*x, 1); r0 = atomic_read(numonline); if (r0 == 1) { smp_mb(); } else { synchronize_rcu(); } WRITE_ONCE(*x, 2); } P1(int *x, atomic_t *numonline) { int r0; int r1; atomic_inc(numonline); smp_mb(); rcu_read_lock(); r0 = READ_ONCE(*x);
Quoted 8 lines
smp_rmb();
r1 = READ_ONCE(*x);
rcu_read_unlock();
}
locations [x;numonline;]
exists (1:r0=0 /\ 1:r1=2)<self rambling>
Right so basically:
P1(int *x, atomic_t *numonline)
{
int r0; int r1;
atomic_inc(numonline);
smp_mb();
rcu_read_lock();
r0 = READ_ONCE(*x);Happened before:
P0(int *x, atomic_t *numonline)
{
int r0;
WRITE_ONCE(*x, 1);
r0 = atomic_read(numonline);
if (r0 == 1) {
smp_mb();
...
}However, P0 did not see P1's update of numonline because the atomic_read() does not provide any ordering.
</self rambling>
Btw, its cool how Paul/you removed the need for memory-barriers for the single-CPU case at all, by making the update to num_online_cpus in rcu_state, from the CPU doing the onlining and offlining.
(For the version in -rcu :)).
thanks,
- Joel
Quoted 235 lines
2. Second problem is, the same early exit, from
synchronize_rcu()does not provide full ordering, memory barrier, w.r.t. memory accesses aftersynchronize_rcu()call.3. Third, more important issue is related to outgoing cpu. Checking only for
__num_online_cpuswith preemotion disabled isn't sufficient for RCU, as, on completion ofCPUHP_TEARDOWN_CPUstop machine (which clears outgoing cpu from__num_online_cpus, the CPU switches to idle task. So, checking only for__num_online_cpusdoes not consider RCU read side sections in scheduler code (before switching to idle task) and any potential read side sections in idle task, before final RCU-quiesce entry intocpuhp_report_idle_dead()->rcu_report_dead().To handle these issues, add a new
rcu_statemembern_online_cpus, to keep account of the current number of online cpus. The atomic updates to this counter fromrcu_report_dead()andrcu_cpu_starting()and the added read/write memory ordering semantics ensure thatsynchronize_rcu()fast path waits for all read side sections, where incoming/outgoing cpus are considered online, for RCU i.e. afterrcu_cpu_starting()and beforercu_report_dead().Signed-off-by Neeraj Upadhyay <[email protected]>---
Below is the reproducer for issue described in point 3; this snippet is based on klitmus generated test, which is modified to sample reads from idle thread:
static void
code0(int* x){WRITE_ONCE(*x, 1);
idle_ctr= 0;
smp_mb();
mdelay(10);
WRITE_ONCE(*x, 1);
idle_ctr= 1;
synchronize_rcu();
WRITE_ONCE(*x, 2);
idle_ctr= 2;}
static int
thread0(void *_p){
int _j, _i;
ctx_t*_a = (ctx_t*)_p;
smp_mb();for (_j = 0 ; _j < stride ; _j++) { for (_i = _j ; _i < size ; _i += stride) { while (idle_wait1) { cpu_relax(); cond_resched(); } code0(&_a->x[_i]); smp_mb(); get_online_cpus(); idle_wait1 = true; put_online_cpus(); }}
atomic_inc(&done);
smp_mb();
wake_up(wq);
smp_mb();
do_exit(0);
}static void
code1(int* x,int* out_1_r1,int* out_1_r0){int r0; int r1;
r0 = READ_ONCE(idle_ctr_snap1); r1 = READ_ONCE(idle_ctr_snap2);
*out_1_r1= (int)r1;
*out_1_r0= (int)r0;
}static int
thread1(void *_p){
ctx_t*_a = (ctx_t*)_p;
int _j, _i;
smp_mb();for (_j = 0 ; _j < stride ; _j++) { for (_i = _j ; _i < size ; _i += stride) { while (idle_wait2) { cpu_relax(); cond_resched(); } get_online_cpus(); code1(&_a->x[_i],&_a->out_1_r1[_i],&_a->out_1_r0[_i]); smp_mb(); idle_wait2 = true; put_online_cpus(); }}
atomic_inc(&done);
smp_mb();
wake_up(wq);
smp_mb();
do_exit(0);
}Idle thread snippet:
if (cpu_is_offline(cpu)) { smp_mb(); idle_wait1 = false; mdelay(8); smp_mb(); rcu_read_lock(); idle_ctr_snap1 = idle_ctr; mdelay(40); smp_rmb(); idle_ctr_snap2 = idle_ctr; rcu_read_unlock(); smp_mb(); idle_wait2 = false; tick_nohz_idle_stop_tick(); cpuhp_report_idle_dead(); arch_cpu_idle_dead();kernel/rcu/tree.c | 65 +++++++++++++++++++++++++++++++++++++++++++++++++++++++ kernel/rcu/tree.h | 1 + 2 files changed, 66 insertions(+)kernel/rcu/tree.c +65
-0@@ -3609,9 +3609,59 @@ static int rcu_blocking_is_gp(void)if (IS_ENABLED (CONFIG_PREEMPTION ))return rcu_scheduler_active ==RCU_SCHEDULER_INACTIVE ;might_sleep ();/* Check for RCU read-side critical section. */ +/* +* a = p +* a = NULL +* synchronize_rcu() +* rcu_blocking_is_gp() +* num_online_cpus() +* atomic_read(&__num_online_cpus) +* kfree(p); +* +* - VS - +* +* __cpu_up() +* set_cpu_online(cpu) +* atomic_inc(&__num_online_cpus) +* rcu_read_lock() +* rcu_dereference(a) (a == p) +* rcu_read_unlock() +* +* rcu_blocking_is_gp() must observe atomic_inc(&__num_online_cpus), +* in order to ensure that, RCU read side critical section on new +* online cpu, either start after synchronize_rcu()'s GP starts or +* it completes before synchronize_rcu() returns. +* +* However, atomic_read(&__num_online_cpus) does not ensure that. +* +* Essentially, below condition exist: +* +* { +* int x = 0; +* atomic_t numonline = ATOMIC_INIT(1); +* } +* +* P0(int *x, atomic_t *numonline) P1(int *x, atomic_t *numonline) +* { { +* int r0; int r0; int r1; +* WRITE_ONCE(*x, 1); atomic_inc(numonline); +* r0 = atomic_read(numonline); rcu_read_lock(); +* if (r0 == 2) { r0 = READ_ONCE(*x); +* synchronize_rcu(); smp_rmb(); +* } r1 = READ_ONCE(*x); +* WRITE_ONCE(*x, 2); rcu_read_unlock(); +* } } +* +* exists (1:r0=0 /\ 1:r1=2) +* +* atomic_add_return(0, &rcu_state.n_online_cpus) and corresponding +* atomic_inc(&rcu_state.n_online_cpus) in rcu_cpu_starting() corrects +* this ordering issue. +*/ preempt_disable (); ret =num_online_cpus () <=1 ;preempt_enable (); + ret = ret && (atomic_add_return (0 , &rcu_state.n_online_cpus) <=1 );return ret; }@@ -3655,6 +3705,11 @@ void synchronize_rcu(void)lock_is_held (&rcu_sched_lock_map), "Illegal synchronize_rcu() in RCU read-side critical section");if (rcu_blocking_is_gp ()) +/* +* atomic_add_return() in rcu_blocking_is_gp () provides +* full memory barrier ordering with any rcu section after +* synchronize_rcu() call. +*/ return ;if (rcu_gp_is_expedited ())synchronize_rcu_expedited ();@@ -4086,6 +4141,10 @@ void rcu_cpu_starting(unsigned int cpu) mask = rdp->grpmask;raw_spin_lock_irqsave_rcu_node (rnp, flags);WRITE_ONCE (rnp->qsmaskinitnext, rnp->qsmaskinitnext | mask); +/* Order with access of n_online_cpus in rcu_blocking_is_gp */ +atomic_inc (&rcu_state.n_online_cpus); +/* Order with rcu-side usages after this */ +smp_mb__after_atomic (); newcpu = !(rnp->expmaskinitnext & mask); rnp->expmaskinitnext |= mask;/* Allow lockless access for expedited grace periods. */ @@ -4138,6 +4197,12 @@ void rcu_report_dead(unsigned int cpu)raw_spin_lock_irqsave_rcu_node (rnp, flags); }WRITE_ONCE (rnp->qsmaskinitnext, rnp->qsmaskinitnext & ~mask); +/* +* Order with access of n_online_cpus in rcu_blocking_is_gp(). +* Release semantics ensures that RCU read sections before it +* are observed by rcu_blocking_is_gp(). +*/ +atomic_dec_return_release (&rcu_state.n_online_cpus);raw_spin_unlock_irqrestore_rcu_node (rnp, flags);raw_spin_unlock (&rcu_state.ofl_lock);kernel/rcu/tree.h +1
-0@@ -298,6 +298,7 @@ struct rcu_state {/* Hierarchy levels (+1 to */ /* shut bogus gcc warning) */ int ncpus;/* # CPUs seen so far. */ +atomic_t n_online_cpus;/* # CPUs online for RCU. */