Re: Question related to patch: x86/irq: Remove x86_io_apic_ops.set_affinity and related interfaces

[Thomas Gleixner <tglx@linutronix.de>]

On Tue, 8 Sep 2015, Mika Westerberg wrote:
> It says that there are no users for x86_io_apic_ops.set_affinity but then
> it does this:
> 
> -               x86_io_apic_ops.set_affinity(idata, mask, false);
> +               irq_set_affinity(irq, mask);
> 
> The difference is that x86_io_apic_ops.set_affinity() programs affinity
> directly to the hardware (if I understand it right) but irq_set_affinity()
> calls irqd_set_move_pending() which defers programming the hardware later.
> 
> Now when an interrupt triggers we end up calling irq_move_masked_irq() with
> unlocked descriptor.
> 
> Without the above change we never do that and the crash does not happen.

Right. Patch below should fix that issue.

Thanks,

	tglx

diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c
index 38a76f826530..d2ea50c5e936 100644
--- a/arch/x86/kernel/apic/io_apic.c
+++ b/arch/x86/kernel/apic/io_apic.c
@@ -2522,6 +2522,7 @@ void __init setup_ioapic_dest(void)
 	int pin, ioapic, irq, irq_entry;
 	const struct cpumask *mask;
 	struct irq_data *idata;
+	struct irq_chip *chip;
 
 	if (skip_ioapic_setup == 1)
 		return;
@@ -2545,9 +2546,9 @@ void __init setup_ioapic_dest(void)
 		else
 			mask = apic->target_cpus();
 
-		irq_set_affinity(irq, mask);
+		chip = irq_data_get_chip(idata);
+		chip->irq_set_affinity(idata, mask, false);
 	}
-
 }
 #endif