[PATCH] powerpc: offline CPU in stop_this_cpu

[PATCH] powerpc: offline CPU in stop_this_cpu

[Nicholas Piggin]

printk_safe_flush_on_panic() has special lock breaking code for the
case where we panic()ed with the console lock held. It relies on
panic IPI causing other CPUs to mark themselves offline.

Do as most other architectures do.

This effectively reverts commit de6e5d38417e ("powerpc: smp_send_stop do
not offline stopped CPUs"), unfortunately it may result in some false
positive warnings, but the alternative is more situations where we can
crash without getting messages out.

Fixes: de6e5d38417e ("powerpc: smp_send_stop do not offline stopped CPUs")
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
---
 arch/powerpc/kernel/smp.c | 11 +++++++++++
 1 file changed, 11 insertions(+)

diff --git a/arch/powerpc/kernel/smp.c b/arch/powerpc/kernel/smp.c
index 2e05c783440a..bf12cca86d70 100644
--- a/arch/powerpc/kernel/smp.c
+++ b/arch/powerpc/kernel/smp.c
@@ -619,6 +619,8 @@ static void nmi_stop_this_cpu(struct pt_regs *regs)
 	/*
 	 * IRQs are already hard disabled by the smp_handle_nmi_ipi.
 	 */
+	set_cpu_online(smp_processor_id(), false);
+
 	spin_begin();
 	while (1)
 		spin_cpu_relax();
@@ -634,6 +636,15 @@ void smp_send_stop(void)
 static void stop_this_cpu(void *dummy)
 {
 	hard_irq_disable();
+
+	/*
+	 * Offlining CPUs in stop_this_cpu can result in scheduler warnings,
+	 * (see commit de6e5d38417e), but printk_safe_flush_on_panic() wants
+	 * to know other CPUs are offline before it breaks locks to flush
+	 * printk buffers, in case we panic()ed while holding the lock.
+	 */
+	set_cpu_online(smp_processor_id(), false);
+
 	spin_begin();
 	while (1)
 		spin_cpu_relax();
-- 
2.23.0

Re: [PATCH] powerpc: offline CPU in stop_this_cpu

[Michael Ellerman]

On Wed, 23 Jun 2021 14:12:45 +1000, Nicholas Piggin wrote:
> printk_safe_flush_on_panic() has special lock breaking code for the
> case where we panic()ed with the console lock held. It relies on
> panic IPI causing other CPUs to mark themselves offline.
> 
> Do as most other architectures do.
> 
> This effectively reverts commit de6e5d38417e ("powerpc: smp_send_stop do
> not offline stopped CPUs"), unfortunately it may result in some false
> positive warnings, but the alternative is more situations where we can
> crash without getting messages out.

Applied to powerpc/next.

[1/1] powerpc: offline CPU in stop_this_cpu
      https://git.kernel.org/powerpc/c/bab26238bbd44d5a4687c0a64fd2c7f2755ea937

cheers