When smp_send_safe_nmi_ipi() indicates that the target CPU has responded to the IPI, skip the remote paca inspection fallback. Otherwise both the sending and target CPUs attempt the backtrace, usually creating a misleading ("didn't respond to backtrace IPI" is wrong) and interleaved mess:
[ 1658.929157][ C7] rcu: Stack dump where RCU GP kthread last ran: [ 1658.929223][ C7] Sending NMI from CPU 7 to CPUs 1: [ 1658.929303][ C1] NMI backtrace for cpu 1 [ 1658.929303][ C7] CPU 1 didn't respond to backtrace IPI, inspecting paca. [ 1658.929362][ C1] CPU: 1 PID: 325 Comm: kworker/1:1H Tainted: G W E 5.13.0-rc2+ #46 [ 1658.929405][ C7] irq_soft_mask: 0x01 in_mce: 0 in_nmi: 0 current: 325 (kworker/1:1H) [ 1658.929465][ C1] Workqueue: events_highpri test_work_fn [test_lockup] [ 1658.929549][ C7] Back trace of paca->saved_r1 (0xc0000000057fb400) (possibly stale): [ 1658.929592][ C1] NIP: c00000000002cf50 LR: c008000000820178 CTR: c00000000002cfa0 Verified using the test_lockup module, e.g. $ echo 5 > /sys/module/rcupdate/parameters/rcu_cpu_stall_timeout $ insmod test_lockup.ko time_secs=1 iterations=10 state=R lock_rcu \ touch_softlockup all_cpus Fixes: 5cc05910f26e ("powerpc/64s: Wire up arch_trigger_cpumask_backtrace()") Signed-off-by: Nathan Lynch <nath...@linux.ibm.com> --- arch/powerpc/kernel/stacktrace.c | 11 +++++++---- 1 file changed, 7 insertions(+), 4 deletions(-) diff --git a/arch/powerpc/kernel/stacktrace.c b/arch/powerpc/kernel/stacktrace.c index 1deb1bf331dd..e0ccc5a46d7e 100644 --- a/arch/powerpc/kernel/stacktrace.c +++ b/arch/powerpc/kernel/stacktrace.c @@ -174,11 +174,14 @@ static void raise_backtrace_ipi(cpumask_t *mask) { unsigned int cpu; + if (cpumask_test_cpu(smp_processor_id(), mask)) { + handle_backtrace_ipi(NULL); + cpumask_clear_cpu(smp_processor_id(), mask); + } + for_each_cpu(cpu, mask) { - if (cpu == smp_processor_id()) - handle_backtrace_ipi(NULL); - else - smp_send_safe_nmi_ipi(cpu, handle_backtrace_ipi, 5 * USEC_PER_SEC); + if (smp_send_safe_nmi_ipi(cpu, handle_backtrace_ipi, 5 * USEC_PER_SEC)) + cpumask_clear_cpu(cpu, mask); } for_each_cpu(cpu, mask) { -- 2.31.1