Re: [PATCH v6 28/33] hw/isa/piix3: Merge hw/isa/piix4.c

[Bernhard Beschow]

Am 12. Januar 2023 16:36:30 UTC schrieb "Philippe Mathieu-Daudé" 
<philmd@linaro.org>:
>On 9/1/23 18:23, Bernhard Beschow wrote:
>> Now that the PIIX3 and PIIX4 device models are sufficiently consolidated,
>> their implementations can be merged into one file for further
>> consolidation.
>> 
>> Signed-off-by: Bernhard Beschow <shentey@gmail.com>
>> Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
>> Message-Id: <20221022150508.26830-37-shentey@gmail.com>
>> ---
>>   hw/isa/{piix3.c => piix.c} | 158 ++++++++++++++++++++
>>   hw/isa/piix4.c             | 285 -------------------------------------
>>   MAINTAINERS                |   6 +-
>>   hw/i386/Kconfig            |   2 +-
>>   hw/isa/Kconfig             |  12 +-
>>   hw/isa/meson.build         |   3 +-
>>   hw/mips/Kconfig            |   2 +-
>>   7 files changed, 165 insertions(+), 303 deletions(-)
>>   rename hw/isa/{piix3.c => piix.c} (75%)
>>   delete mode 100644 hw/isa/piix4.c
>
>
>> @@ -489,11 +534,124 @@ static const TypeInfo piix3_xen_info = {
>>       .class_init    = piix3_xen_class_init,
>>   };
>>   +static void piix4_realize(PCIDevice *dev, Error **errp)
>> +{
>
>> +    /* initialize pit */
>> +    i8254_pit_init(isa_bus, 0x40, 0, NULL);
>Pre-existing, why there is no equivalent PIT creation in the
>PIIX3 variant? Due to in-kernel PIT in KVM?

Correct, that's one reason. The other reason is for interrupt wiring in case an 
HPET is present (see our discussion about the intercept_irq for the rtc).

I would like to create and wire up the PIT in PIIX3 as well. Since we'd have to 
take KVM into account here we may need a similar solution as for I8259. This is 
another open question for PIIX4 to become a drop-in replacement for PIIX3. Any 
ideas?

Best regards,
Bernhard