Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
From:   =?utf-8?q?M=C3=A5rten_Lindahl?= <marten.lindahl@axis.com>
Date:   Thu, 15 Jun 2023 12:34:32 +0200
Subject: [PATCH] ARM: kexec: Make smp stop calls asynchronous
MIME-Version: 1.0
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: 8bit
Message-ID: <20230615-arm-smp-call-async-v1-1-58f4afb20f0f@axis.com>
To:     Russell King <linux@armlinux.org.uk>
CC:     <linux-arm-kernel@lists.infradead.org>,
        <linux-kernel@vger.kernel.org>, <kernel@axis.com>,
        =?utf-8?q?M=C3=A5rten_Lindahl?= <marten.lindahl@axis.com>
Precedence: bulk

If a panic is triggered by a hrtimer interrupt all online cpus will be
noitfied and set offline. But as highlighted by commit 19dbdcb8039c
("smp: Warn on function calls from softirq context") this call should
not be made synchronous with disabled interrupts:

  softdog: Initiating panic
  Kernel panic - not syncing: Software Watchdog Timer expired
  WARNING: CPU: 1 PID: 0 at kernel/smp.c:753 smp_call_function_many_cond
    unwind_backtrace:
      show_stack
      dump_stack_lvl
      __warn
      warn_slowpath_fmt
      smp_call_function_many_cond
      smp_call_function
      crash_smp_send_stop.part.0
      machine_crash_shutdown
      __crash_kexec
      panic
      softdog_fire
      __hrtimer_run_queues
      hrtimer_interrupt

Make the smp call for machine_crash_nonpanic_core() asynchronous.

Signed-off-by: Mårten Lindahl <marten.lindahl@axis.com>
---
 arch/arm/kernel/machine_kexec.c | 14 +++++++++++++-
 1 file changed, 13 insertions(+), 1 deletion(-)

diff --git a/arch/arm/kernel/machine_kexec.c b/arch/arm/kernel/machine_kexec.c
index 46364b699cc3..5d07cf9e0044 100644
--- a/arch/arm/kernel/machine_kexec.c
+++ b/arch/arm/kernel/machine_kexec.c
@@ -94,16 +94,28 @@ static void machine_crash_nonpanic_core(void *unused)
 	}
 }
 
+static DEFINE_PER_CPU(call_single_data_t, cpu_stop_csd) =
+	CSD_INIT(machine_crash_nonpanic_core, NULL);
+
 void crash_smp_send_stop(void)
 {
 	static int cpus_stopped;
 	unsigned long msecs;
+	call_single_data_t *csd;
+	int cpu, this_cpu = raw_smp_processor_id();
 
 	if (cpus_stopped)
 		return;
 
 	atomic_set(&waiting_for_crash_ipi, num_online_cpus() - 1);
-	smp_call_function(machine_crash_nonpanic_core, NULL, false);
+	for_each_online_cpu(cpu) {
+		if (cpu == this_cpu)
+			continue;
+
+		csd = &per_cpu(cpu_stop_csd, cpu);
+		smp_call_function_single_async(cpu, csd);
+	}
+
 	msecs = 1000; /* Wait at most a second for the other cpus to stop */
 	while ((atomic_read(&waiting_for_crash_ipi) > 0) && msecs) {
 		mdelay(1);

---
base-commit: 858fd168a95c5b9669aac8db6c14a9aeab446375
change-id: 20230615-arm-smp-call-async-38653ff40952

Best regards,
-- 
Mårten Lindahl <marten.lindahl@axis.com>