These patches required to prevent the following TRACE02 warning when running rcutorture.
I confirmed the patch fixes the following splat:
[ 765.941351] WARNING: CPU: 0 PID: 80 at kernel/rcu/tasks.h:895 trc_read_check_handler+0x61/0xe0 [ 765.949880] Modules linked in: [ 765.953006] CPU: 0 PID: 80 Comm: rcu_torture_rea Not tainted 5.15.86-rc1+ #25 [ 765.959982] Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.15.0-1 04/01/2014 [ 765.967964] RIP: 0010:trc_read_check_handler+0x61/0xe0 [ 765.973050] Code: 01 00 89 c0 48 03 2c c5 80 f8 a5 ae c6 45 00 00 [..] [ 765.991768] RSP: 0000:ffffa64ac0003fb0 EFLAGS: 00010047 [ 765.997042] RAX: ffffffffad4f8610 RBX: ffffa26b41bd3000 RCX: ffffa26b5f4ac8c0 [ 766.004418] RDX: 0000000000000000 RSI: ffffffffae978121 RDI: ffffa26b41bd3000 [ 766.011502] RBP: ffffa26b41bd6000 R08: ffffa26b41bd3000 R09: 0000000000000000 [ 766.018778] R10: 0000000000000000 R11: ffffa64ac0003ff8 R12: 0000000000000000 [ 766.025943] R13: ffffa26b5f4ac8c0 R14: 0000000000000000 R15: 0000000000000000 [ 766.034383] FS: 0000000000000000(0000) GS:ffffa26b5f400000(0000) knlGS:0000000000000000 [ 766.042925] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 766.048775] CR2: 0000000000000000 CR3: 0000000001924000 CR4: 00000000000006f0 [ 766.055991] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 [ 766.063135] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 [ 766.070711] Call Trace: [ 766.073515] <IRQ> [ 766.075807] flush_smp_call_function_queue+0xec/0x1a0 [ 766.081087] __sysvec_call_function_single+0x3e/0x1d0 [ 766.086466] sysvec_call_function_single+0x89/0xc0 [ 766.091431] </IRQ> [ 766.093713] <TASK> [ 766.095930] asm_sysvec_call_function_single+0x16/0x20
From: "Paul E. McKenney" paulmck@kernel.org
[ Upstream commit bdb0cca0d11060fce8a8a44588ac1470c25d62bc ]
There are several ->trc_reader_nesting data races that are too low-probability for KCSAN to notice, but which will happen sooner or later. This commit therefore marks these accesses, and comments one that cannot race.
Cc: stable@vger.kernel.org # 5.10.x Signed-off-by: Paul E. McKenney paulmck@kernel.org Signed-off-by: Joel Fernandes (Google) joel@joelfernandes.org --- kernel/rcu/tasks.h | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-)
diff --git a/kernel/rcu/tasks.h b/kernel/rcu/tasks.h index c66d47685b28..2b16a86163af 100644 --- a/kernel/rcu/tasks.h +++ b/kernel/rcu/tasks.h @@ -848,7 +848,7 @@ static void trc_read_check_handler(void *t_in)
// If the task is not in a read-side critical section, and // if this is the last reader, awaken the grace-period kthread. - if (likely(!t->trc_reader_nesting)) { + if (likely(!READ_ONCE(t->trc_reader_nesting))) { if (WARN_ON_ONCE(atomic_dec_and_test(&trc_n_readers_need_end))) wake_up(&trc_wait); // Mark as checked after decrement to avoid false @@ -857,7 +857,7 @@ static void trc_read_check_handler(void *t_in) goto reset_ipi; } // If we are racing with an rcu_read_unlock_trace(), try again later. - if (unlikely(t->trc_reader_nesting < 0)) { + if (unlikely(READ_ONCE(t->trc_reader_nesting) < 0)) { if (WARN_ON_ONCE(atomic_dec_and_test(&trc_n_readers_need_end))) wake_up(&trc_wait); goto reset_ipi; @@ -904,6 +904,7 @@ static bool trc_inspect_reader(struct task_struct *t, void *arg) n_heavy_reader_ofl_updates++; in_qs = true; } else { + // The task is not running, so C-language access is safe. in_qs = likely(!t->trc_reader_nesting); }
@@ -936,7 +937,7 @@ static void trc_wait_for_one_reader(struct task_struct *t, // The current task had better be in a quiescent state. if (t == current) { t->trc_reader_checked = true; - WARN_ON_ONCE(t->trc_reader_nesting); + WARN_ON_ONCE(READ_ONCE(t->trc_reader_nesting)); return; }
@@ -1046,7 +1047,7 @@ static void show_stalled_task_trace(struct task_struct *t, bool *firstreport) ".I"[READ_ONCE(t->trc_ipi_to_cpu) > 0], ".i"[is_idle_task(t)], ".N"[cpu > 0 && tick_nohz_full_cpu(cpu)], - t->trc_reader_nesting, + READ_ONCE(t->trc_reader_nesting), " N"[!!t->trc_reader_special.b.need_qs], cpu); sched_show_task(t); @@ -1141,7 +1142,7 @@ static void rcu_tasks_trace_postgp(struct rcu_tasks *rtp) static void exit_tasks_rcu_finish_trace(struct task_struct *t) { WRITE_ONCE(t->trc_reader_checked, true); - WARN_ON_ONCE(t->trc_reader_nesting); + WARN_ON_ONCE(READ_ONCE(t->trc_reader_nesting)); WRITE_ONCE(t->trc_reader_nesting, 0); if (WARN_ON_ONCE(READ_ONCE(t->trc_reader_special.b.need_qs))) rcu_read_unlock_trace_special(t, 0);
From: "Paul E. McKenney" paulmck@kernel.org
[ Upstream commit f8ab3fad80dddf3f2cecb53983063c4431058ca1 ]
There are several ->trc_reader_special.b.need_qs data races that are too low-probability for KCSAN to notice, but which will happen sooner or later. This commit therefore marks these accesses.
Cc: stable@vger.kernel.org # 5.10.x Signed-off-by: Paul E. McKenney paulmck@kernel.org Signed-off-by: Joel Fernandes (Google) joel@joelfernandes.org --- kernel/rcu/tasks.h | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/kernel/rcu/tasks.h b/kernel/rcu/tasks.h index 2b16a86163af..aef2e96c3854 100644 --- a/kernel/rcu/tasks.h +++ b/kernel/rcu/tasks.h @@ -801,7 +801,7 @@ static DEFINE_IRQ_WORK(rcu_tasks_trace_iw, rcu_read_unlock_iw); /* If we are the last reader, wake up the grace-period kthread. */ void rcu_read_unlock_trace_special(struct task_struct *t, int nesting) { - int nq = t->trc_reader_special.b.need_qs; + int nq = READ_ONCE(t->trc_reader_special.b.need_qs);
if (IS_ENABLED(CONFIG_TASKS_TRACE_RCU_READ_MB) && t->trc_reader_special.b.need_mb) @@ -867,7 +867,7 @@ static void trc_read_check_handler(void *t_in) // Get here if the task is in a read-side critical section. Set // its state so that it will awaken the grace-period kthread upon // exit from that critical section. - WARN_ON_ONCE(t->trc_reader_special.b.need_qs); + WARN_ON_ONCE(READ_ONCE(t->trc_reader_special.b.need_qs)); WRITE_ONCE(t->trc_reader_special.b.need_qs, true);
reset_ipi: @@ -919,7 +919,7 @@ static bool trc_inspect_reader(struct task_struct *t, void *arg) // state so that it will awaken the grace-period kthread upon exit // from that critical section. atomic_inc(&trc_n_readers_need_end); // One more to wait on. - WARN_ON_ONCE(t->trc_reader_special.b.need_qs); + WARN_ON_ONCE(READ_ONCE(t->trc_reader_special.b.need_qs)); WRITE_ONCE(t->trc_reader_special.b.need_qs, true); return true; } @@ -1048,7 +1048,7 @@ static void show_stalled_task_trace(struct task_struct *t, bool *firstreport) ".i"[is_idle_task(t)], ".N"[cpu > 0 && tick_nohz_full_cpu(cpu)], READ_ONCE(t->trc_reader_nesting), - " N"[!!t->trc_reader_special.b.need_qs], + " N"[!!READ_ONCE(t->trc_reader_special.b.need_qs)], cpu); sched_show_task(t); }
From: "Paul E. McKenney" paulmck@kernel.org
[ Upstream commit 96017bf9039763a2e02dcc6adaa18592cd73a39d ]
Currently, trc_wait_for_one_reader() atomically increments the trc_n_readers_need_end counter before sending the IPI invoking trc_read_check_handler(). All failure paths out of trc_read_check_handler() and also from the smp_call_function_single() within trc_wait_for_one_reader() must carefully atomically decrement this counter. This is more complex than it needs to be.
This commit therefore simplifies things and saves a few lines of code by dispensing with the atomic decrements in favor of having trc_read_check_handler() do the atomic increment only in the success case. In theory, this represents no change in functionality.
Cc: stable@vger.kernel.org # 5.10.x Signed-off-by: Paul E. McKenney paulmck@kernel.org Signed-off-by: Joel Fernandes (Google) joel@joelfernandes.org --- kernel/rcu/tasks.h | 20 +++----------------- 1 file changed, 3 insertions(+), 17 deletions(-)
diff --git a/kernel/rcu/tasks.h b/kernel/rcu/tasks.h index aef2e96c3854..23101ebbbe1e 100644 --- a/kernel/rcu/tasks.h +++ b/kernel/rcu/tasks.h @@ -841,32 +841,24 @@ static void trc_read_check_handler(void *t_in)
// If the task is no longer running on this CPU, leave. if (unlikely(texp != t)) { - if (WARN_ON_ONCE(atomic_dec_and_test(&trc_n_readers_need_end))) - wake_up(&trc_wait); goto reset_ipi; // Already on holdout list, so will check later. }
// If the task is not in a read-side critical section, and // if this is the last reader, awaken the grace-period kthread. if (likely(!READ_ONCE(t->trc_reader_nesting))) { - if (WARN_ON_ONCE(atomic_dec_and_test(&trc_n_readers_need_end))) - wake_up(&trc_wait); - // Mark as checked after decrement to avoid false - // positives on the above WARN_ON_ONCE(). WRITE_ONCE(t->trc_reader_checked, true); goto reset_ipi; } // If we are racing with an rcu_read_unlock_trace(), try again later. - if (unlikely(READ_ONCE(t->trc_reader_nesting) < 0)) { - if (WARN_ON_ONCE(atomic_dec_and_test(&trc_n_readers_need_end))) - wake_up(&trc_wait); + if (unlikely(READ_ONCE(t->trc_reader_nesting) < 0)) goto reset_ipi; - } WRITE_ONCE(t->trc_reader_checked, true);
// Get here if the task is in a read-side critical section. Set // its state so that it will awaken the grace-period kthread upon // exit from that critical section. + atomic_inc(&trc_n_readers_need_end); // One more to wait on. WARN_ON_ONCE(READ_ONCE(t->trc_reader_special.b.need_qs)); WRITE_ONCE(t->trc_reader_special.b.need_qs, true);
@@ -960,21 +952,15 @@ static void trc_wait_for_one_reader(struct task_struct *t, if (per_cpu(trc_ipi_to_cpu, cpu) || t->trc_ipi_to_cpu >= 0) return;
- atomic_inc(&trc_n_readers_need_end); per_cpu(trc_ipi_to_cpu, cpu) = true; t->trc_ipi_to_cpu = cpu; rcu_tasks_trace.n_ipis++; - if (smp_call_function_single(cpu, - trc_read_check_handler, t, 0)) { + if (smp_call_function_single(cpu, trc_read_check_handler, t, 0)) { // Just in case there is some other reason for // failure than the target CPU being offline. rcu_tasks_trace.n_ipis_fails++; per_cpu(trc_ipi_to_cpu, cpu) = false; t->trc_ipi_to_cpu = cpu; - if (atomic_dec_and_test(&trc_n_readers_need_end)) { - WARN_ON_ONCE(1); - wake_up(&trc_wait); - } } } }
linux-stable-mirror@lists.linaro.org