Re: [PATCH] testdev: adjust for ISA irq changes

[Jan Kiszka <jan.kiszka@web.de>]

[-- Attachment #1: Type: text/plain, Size: 2228 bytes --]

On 2011-05-29 17:10, Avi Kivity wrote:
> On 05/29/2011 06:05 PM, Jan Kiszka wrote:
>> On 2011-05-29 14:57, Avi Kivity wrote:
>> >  Recent changes killed the ioapic_irq_hack hack, use the
>> isa_get_irq() API
>> >  instead.
>> >
>> >  Signed-off-by: Avi Kivity<avi@redhat.com>
>> >  ---
>> >   hw/isa-bus.c |    2 +-
>> >   hw/testdev.c |    4 +---
>> >   2 files changed, 2 insertions(+), 4 deletions(-)
>> >
>> >  diff --git a/hw/isa-bus.c b/hw/isa-bus.c
>> >  index 2765543..7e06efc 100644
>> >  --- a/hw/isa-bus.c
>> >  +++ b/hw/isa-bus.c
>> >  @@ -66,7 +66,7 @@ void isa_bus_irqs(qemu_irq *irqs)
>> >    */
>> >   qemu_irq isa_get_irq(int isairq)
>> >   {
>> >  -    if (isairq<  0 || isairq>  15) {
>> >  +    if (isairq<  0 || isairq>  23) {
>>
>> That's a fairly evil hack. It will break again when we clean up the kvm
>> irqchips (not to speak of side effects in non-irqchip/non-kvm mode).
>>
>> Why not hook into the kvm irqchip directly? Would keep generic code out
>> of this business until we have real irq pin manipulation in qemu.
> 
> Firstly, I did the minimal patch to get things working, as is my custom
> when fixing up merge/refactoring problems.  Not that I often follow up
> with the clean fix.

The minimal fix would be simply initializing ioapic_isa_hack again. That
was lost while merging with the xen bits.

> 
> Second, I'm not sure it's such a hack.  

The hack is that isa_get_irq is supposed to return ISA IRQs, not GSIs.
Name-wise, there is already a bit mixed up in QEMU, but at least not in
that function.

> Suppose our motherboard wired
> the PCI links to GSI16-19 (or GSI16-23, as we once wanted before we had
> MSI-X)?  We'd need an API to access non-ISA interrupt lines.
> 
> So what's the clean fix here?  gsi_get_irq()?

Maybe. Depends on the requirements of the testdev. If you also want to
address PIC and IOAPIC separately or simulate injection from a specific
device, we need more logic.

We also need a better interface to discover and track legacy IRQ routes
for device assignment. Markus is currently collecting requirements for
qdev enhancements, and I think generic IRQ manipulation and discovery
belongs there.

Jan


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 259 bytes --]