perf: Enqueue SIGTRAP always via task_work.

A signal is delivered by raising irq_work() which works from any context
including NMI. irq_work() can be delayed if the architecture does not
provide an interrupt vector. In order not to lose a signal, the signal
is injected via task_work during event_sched_out().

Instead going via irq_work, the signal could be added directly via
task_work. The signal is sent to current and can be enqueued on its
return path to userland.

Queue signal via task_work and consider possible NMI context. Remove
perf_event::pending_sigtrap and and use perf_event::pending_work
instead.

Reported-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Marco Elver <elver@google.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Link: https://lore.kernel.org/r/20240704170424.1466941-4-bigeasy@linutronix.de

Author: Sebastian Andrzej Siewior

include/linux/perf_event.h

@@ -781,7 +781,6 @@ struct perf_event {
 	unsigned int			pending_wakeup;
 	unsigned int			pending_kill;
 	unsigned int			pending_disable;
-	unsigned int			pending_sigtrap;
 	unsigned long			pending_addr;	/* SIGTRAP */
 	struct irq_work			pending_irq;
 	struct callback_head		pending_task;
@@ -963,7 +962,7 @@ struct perf_event_context {
 	struct rcu_head			rcu_head;
 
 	/*
-	 * Sum (event->pending_sigtrap + event->pending_work)
+	 * Sum (event->pending_work + event->pending_work)
 	 *
 	 * The SIGTRAP is targeted at ctx->task, as such it won't do changing
 	 * that until the signal is delivered.

kernel/events/core.c

@@ -2283,17 +2283,6 @@ event_sched_out(struct perf_event *event, struct perf_event_context *ctx)
 		state = PERF_EVENT_STATE_OFF;
 	}
 
-	if (event->pending_sigtrap) {
-		event->pending_sigtrap = 0;
-		if (state != PERF_EVENT_STATE_OFF &&
-		    !event->pending_work &&
-		    !task_work_add(current, &event->pending_task, TWA_RESUME)) {
-			event->pending_work = 1;
-		} else {
-			local_dec(&event->ctx->nr_pending);
-		}
-	}
-
 	perf_event_set_state(event, state);
 
 	if (!is_software_event(event))
@@ -6776,11 +6765,6 @@ static void __perf_pending_irq(struct perf_event *event)
 	 * Yay, we hit home and are in the context of the event.
 	 */
 	if (cpu == smp_processor_id()) {
-		if (event->pending_sigtrap) {
-			event->pending_sigtrap = 0;
-			perf_sigtrap(event);
-			local_dec(&event->ctx->nr_pending);
-		}
 		if (event->pending_disable) {
 			event->pending_disable = 0;
 			perf_event_disable_local(event);
@@ -9721,21 +9705,26 @@ static int __perf_event_overflow(struct perf_event *event,
 		 */
 		bool valid_sample = sample_is_allowed(event, regs);
 		unsigned int pending_id = 1;
+		enum task_work_notify_mode notify_mode;
 
 		if (regs)
 			pending_id = hash32_ptr((void *)instruction_pointer(regs)) ?: 1;
-		if (!event->pending_sigtrap) {
-			event->pending_sigtrap = pending_id;
+
+		notify_mode = in_nmi() ? TWA_NMI_CURRENT : TWA_RESUME;
+
+		if (!event->pending_work &&
+		    !task_work_add(current, &event->pending_task, notify_mode)) {
+			event->pending_work = pending_id;
 			local_inc(&event->ctx->nr_pending);
 
 			event->pending_addr = 0;
 			if (valid_sample && (data->sample_flags & PERF_SAMPLE_ADDR))
 				event->pending_addr = data->addr;
-			irq_work_queue(&event->pending_irq);
+
 		} else if (event->attr.exclude_kernel && valid_sample) {
 			/*
 			 * Should not be able to return to user space without
-			 * consuming pending_sigtrap; with exceptions:
+			 * consuming pending_work; with exceptions:
 			 *
 			 *  1. Where !exclude_kernel, events can overflow again
 			 *     in the kernel without returning to user space.
@@ -9745,7 +9734,7 @@ static int __perf_event_overflow(struct perf_event *event,
 			 *     To approximate progress (with false negatives),
 			 *     check 32-bit hash of the current IP.
 			 */
-			WARN_ON_ONCE(event->pending_sigtrap != pending_id);
+			WARN_ON_ONCE(event->pending_work != pending_id);
 		}
 	}