[prev in list] [next in list] [prev in thread] [next in thread] 

List:       linux-ia64
Subject:    RE: [ACPI] [PATCH] add acpi_interrupt_to_irq
From:       "Nakajima, Jun" <jun.nakajima () intel ! com>
Date:       2004-01-23 17:35:38
Message-ID: 7F740D512C7C1046AB53446D3720017361886E () scsmsx402 ! sc ! intel ! com
[Download RAW message or body]

Not a problem to me :-) So if we remove the #ifdef (i.e. both
CONFIG_IA64 and CONFIG_PCI_USE_VECTOR) there, I think acpi_gsi_to_irq is
reasonable there. For x86 we'll switch to (the old) acpi_irq_to_vector
if CONFIG_PCI_USE_VECTOR is configured, and don't do anything for the
usual case. But I don't expect Len to replace every irq in ACPI with
gsi, though.

BTW, we need to think how we support MSI (and PCI Express) for IPF; we
already posted a PCI Express patch for x86. Today's IRQ abstraction on
IPF might have some advantage for supporting MSI, which is required for
PCI Express.

Jun

> -----Original Message-----
> From: Bjorn Helgaas [mailto:bjorn.helgaas@hp.com]
> Sent: Friday, January 23, 2004 8:36 AM
> To: Nakajima, Jun; acpi-devel@lists.sourceforge.net; linux-
> ia64@vger.kernel.org
> Cc: Brown, Len
> Subject: Re: [ACPI] [PATCH] add acpi_interrupt_to_irq
> 
> On Thursday 22 January 2004 8:36 pm, Nakajima, Jun wrote:
> > > What does "vector" mean?
> > For MSI, vector means the (external) interrupt vector, i.e. the
index in
> > the IDT for MSI on x86. So acpi_irq_to_vector() is correct in that
case.
> > ACPI looks at IRQ or GSI (Global system interrupt) vector (yes, it's
> > confusing). SOMETHING should be irq or gsi, as David suggested.
> >
> > Since MSI does not require IRQ (but external interrupt vector), the
way
> > we did for x86 was to use the vector to unify IRQ and vector. So
> > request_irq() actually gets the interrupt vector number, instead of
irq.
> > That's the reason I preferred acpi_irq_to_vector() in that code with
MSI
> > configured.
> 
> Sorry to drag this out even longer...  I promise I'll shut up
> after this :-)
> 
> But this business about request_irq() getting an MSI interrupt vector
> number, not an irq, is just a detail of the MSI and architecture
> implementation.  Surely ACPI should just use the abstract Linux
> interrupt interface (request_irq(), free_irq(), etc), which uses
> the "irq" terminology, and should remain ignorant of whether MSI
> is even present.
> 
> Bjorn

-
To unsubscribe from this list: send the line "unsubscribe linux-ia64" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
[prev in list] [next in list] [prev in thread] [next in thread]