Re: [PATCH] libahci: ahci interrupt check for disabled port since private_data may be NULL

[Alexander Gordeev <agordeev@redhat.com>]

On Wed, Apr 16, 2014 at 03:14:54PM -0400, Tejun Heo wrote:
> Hello,
> 
> On Wed, Apr 16, 2014 at 08:51:05PM +0200, Alexander Gordeev wrote:
> > The crash occured in ahci_hw_interrupt() which means multiple MSIs
> > were enabled. The fact driver does not request IRQ does not mean
> > the PCI device does not send an MSI interrupt (and it does as we're
> > observing the crash). So my question if the dummy port interrupt
> > does not end up in handle_bad_irq() or some?
> 
> My memory is kinda fuzzy now but several stray interrupts don't
> trigger anything.

Mine too :)
Just thought may be this way would be better:

diff --git a/drivers/ata/ahci.c b/drivers/ata/ahci.c
index e45b18e..9f1169a 100644
--- a/drivers/ata/ahci.c
+++ b/drivers/ata/ahci.c
@@ -1240,6 +1240,8 @@ int ahci_host_activate(struct ata_host *host, int irq, unsigned int n_msis)
 						       irq + i, ahci_hw_interrupt,
 						       ahci_thread_fn, IRQF_SHARED,
 						       pp->irq_desc, host->ports[i]);
+		else
+			disable_irq(irq + i);
 		if (rc)
 			goto out_free_irqs;
 	}

> Thanks.
> 
> -- 
> tejun

-- 
Regards,
Alexander Gordeev
agordeev@redhat.com