Re: [PATCH 2/2] libata: make sure IRQ is cleared after ata_bmdma_freeze()

[Tejun Heo <htejun@gmail.com>]

Mark Lord wrote:
> Tejun Heo wrote:
>>
>> Note that ->freeze() handler is always called under ap->lock held and
>> irq disabled.  Even if CTL manipulation causes stuck IRQ, it's cleared
>> immediately.  This should be safe (enough) even in SMP environment.
>> More correct solution is to mask the IRQ from IRQ controller but that
>> would be an overkill.
> 
> Could this possibly lead to a "nobody cared" message from the IRQ layer,
> followed by the kernel then disabling this IRQ?
> 
> Just wondering, because that's what can happen if we trigger an IRQ
> but then don't report "handled" for it from our interrupt handler.

Nobody cared triggers after a LOT of interrupts are consecutively 
unhandled, so this is safe for UP and safe enough for SMP, IMHO.  Also, 
that's how it has been done in ata_irq_on().  Without this patch, my 
ich7 in native pci mode w/ shared IRQ does cause nobody cared.  That's 
why this patch is written.

> I'm puzzling through a similar situation in sata_qstor right now,
> and the fix will involve manipulating the NIEN bit, with the same risk.

Well, I guess, testing is the only way to find answer.  :-(

-- 
tejun