kernel: irq: Call irq_chip->irq_mask in irq_disable
PATCH 1/1 kernel: irq: Call irq_chip->irq_mask in irq_disable
From: sreenivasulu velpula <[email protected]>
If interrupt driver has implemented the irq_mask/irq_unmask callbacks for disable/enable interrupt then call these APIs from irq_disable/irq_enable.
Currently, on irq_disable(), it just look for the callback irq_disable() implemented by chip interrupt driver not the irq_mask().
Add check for the valid callback of irq_mask() and if it is there then call it from irq_disable.
This is based on change done by:
sreenivasulu velpula <[email protected]>
--- kernel/irq/chip.c | 3 +++ 1 files changed, 3 insertions(+), 0 deletions(-)
kernel/irq/chip.c +3 -0
@@ -232,6 +232,9 @@ void irq_disable(struct irq_desc *desc)if (desc->irq_data.chip->irq_disable) { desc->irq_data.chip->irq_disable (&desc->irq_data);irq_state_set_masked (desc); + }else if (desc->irq_data.chip->irq_mask) { + desc->irq_data.chip->irq_mask (&desc->irq_data); +irq_state_set_masked (desc); } }
Signature · 1 line
1.7.1.1
Quoted 6 lines · Laxman Dewangan
From: sreenivasulu velpula <[email protected]>
If interrupt driver has implemented the irq_mask/irq_unmask callbacks for disable/enable interrupt then call these APIs from irq_disable/irq_enable.
Quoted 6 lines
Currently, on
irq_disable(), it just look for the callbackirq_disable()implemented by chip interrupt driver not theirq_mask().Add check for the valid callback of
irq_mask()and if it is there then call it fromirq_disable.
No. That changes the semantics of lazy irq disable and breaks suspend/resume because we rely on NOT masking interrupts to detect a wakeup.
The irq_disable() merily marks the interrupt disabled and only touches the hardware when the interrupt controller provides an explicit irq_disable() callback.
Thanks,
tglx