Re: [Qemu-devel] [PATCH RFC 5/9] ich9: Document why cannot_instantiate_with_device_add_yet

[Markus Armbruster]

Paolo Bonzini <address@hidden> writes:

> Il 10/10/2013 18:01, Paolo Bonzini ha scritto:
>> Il 10/10/2013 16:42, address@hidden ha scritto:
>>> From: Markus Armbruster <address@hidden>
>>>
>>> An ICH9 southbridge contains several PCI devices, some of them with
>>> multiple functions.  We model each function as a separate qdev.  Two
>>> of them need some special wiring set up in pc_q35_init() to work: the
>>> LPC controller at 00:1f.0, and the SMBus controller at 00:1f.3.
>>>
>>> Signed-off-by: Markus Armbruster <address@hidden>
>>> ---
>>>  hw/i2c/smbus_ich9.c | 6 +++++-
>>>  hw/isa/lpc_ich9.c   | 7 +++++--
>>>  2 files changed, 10 insertions(+), 3 deletions(-)
>>>
>>> diff --git a/hw/i2c/smbus_ich9.c b/hw/i2c/smbus_ich9.c
>>> index c1ffa34..8d47eaf 100644
>>> --- a/hw/i2c/smbus_ich9.c
>>> +++ b/hw/i2c/smbus_ich9.c
>>> @@ -97,11 +97,15 @@ static void ich9_smb_class_init(ObjectClass *klass, 
>>> void *data)
>>>      k->device_id = PCI_DEVICE_ID_INTEL_ICH9_6;
>>>      k->revision = ICH9_A2_SMB_REVISION;
>>>      k->class_id = PCI_CLASS_SERIAL_SMBUS;
>>> -    dc->cannot_instantiate_with_device_add_yet = true; /* FIXME explain 
>>> why */
>>>      dc->vmsd = &vmstate_ich9_smbus;
>>>      dc->desc = "ICH9 SMBUS Bridge";
>>>      k->init = ich9_smbus_initfn;
>>>      k->config_write = ich9_smbus_write_config;
>>> +    /*
>>> +     * Reason: part of ICH9 southbridge, needs to be wired up by
>>> +     * pc_q35_init()
>>> +     */
>>> +    dc->cannot_instantiate_with_device_add_yet = true;
>>>  }
>>>  
>>>  i2c_bus *ich9_smb_init(PCIBus *bus, int devfn, uint32_t smb_io_base)
>>> diff --git a/hw/isa/lpc_ich9.c b/hw/isa/lpc_ich9.c
>>> index ad841b5..d00d698 100644
>>> --- a/hw/isa/lpc_ich9.c
>>> +++ b/hw/isa/lpc_ich9.c
>>> @@ -604,14 +604,17 @@ static void ich9_lpc_class_init(ObjectClass *klass, 
>>> void *data)
>>>      dc->reset = ich9_lpc_reset;
>>>      k->init = ich9_lpc_initfn;
>>>      dc->vmsd = &vmstate_ich9_lpc;
>>> -    dc->cannot_instantiate_with_device_add_yet = true; /* FIXME explain 
>>> why */
>>>      k->config_write = ich9_lpc_config_write;
>>>      dc->desc = "ICH9 LPC bridge";
>>>      k->vendor_id = PCI_VENDOR_ID_INTEL;
>>>      k->device_id = PCI_DEVICE_ID_INTEL_ICH9_8;
>>>      k->revision = ICH9_A2_LPC_REVISION;
>>>      k->class_id = PCI_CLASS_BRIDGE_ISA;
>>> -
>>> +    /*
>>> +     * Reason: part of ICH9 southbridge, needs to be wired up by
>>> +     * pc_q35_init()
>>> +     */
>>> +    dc->cannot_instantiate_with_device_add_yet = true;
>>>  }
>>>  
>>>  static const TypeInfo ich9_lpc_info = {
>>>
>> 
>> I think SMBus is fine.  Even though the standard configuration of the
>> Q35 SMBus cannot be reproduced with -device, that's a problem of
>> hw/i2c/smbus_eeprom.c, not of the smbus bridge itself.  I think
>> hw/i2c/smbus_eeprom.c should rather be marked as
>> cannot_instantiate_with_device_add_yet (which makes sense, because it
>> has a PROP_PTR).
>> 
>> Alternatively, you can resuscitate the first two patches of
>> http://lists.gnu.org/archive/html/qemu-devel/2012-02/msg00449.html

Getting rid of pointer properties is useful work, but outside the scope
of this series.

> Ah, got it---the SMBus and LPC need to be connected to each other.

Yup, that's the reason.

> Still, the SMBus EEPROM is a candidate for
> cannot_instantiate_with_device_add_yet.

I'll add it in the next version, along with all other devices with
pointer properties.  Thanks!