Re: [Qemu-devel] [PATCH 5/5] sclp-s390: Add memory hotplug SCLPs

[Matthew Rosato <mjrosato@linux.vnet.ibm.com>]

On 12/16/2013 06:18 PM, Alexander Graf wrote:
> 
> On 16.12.2013, at 22:42, Alexander Graf <agraf@suse.de> wrote:
> 
>>
>> On 16.12.2013, at 21:51, Matthew Rosato <mjrosato@linux.vnet.ibm.com> wrote:
>>
> 
> [...]
> 
>>
>>> +            }
>>> +        }
>>> +    }
>>> +    sccb->h.response_code = cpu_to_be16(SCLP_RC_NORMAL_COMPLETION);
>>> +}
>>> +
>>> static void sclp_execute(SCCB *sccb, uint64_t code)
>>> {
>>>    S390SCLPDevice *sdev = get_event_facility();
>>> @@ -50,6 +255,22 @@ static void sclp_execute(SCCB *sccb, uint64_t code)
>>>    case SCLP_CMDW_READ_SCP_INFO_FORCED:
>>>        read_SCP_info(sccb);
>>>        break;
>>> +    case SCLP_READ_STORAGE_ELEMENT_INFO:
>>> +        if (code & 0xff00) {
>>> +            read_storage_element1_info(sccb);
>>> +        } else {
>>> +            read_storage_element0_info(sccb);
>>> +        }
>>> +        break;
>>> +    case SCLP_ATTACH_STORAGE_ELEMENT:
>>> +        attach_storage_element(sccb, (code & 0xff00) >> 8);
>>> +        break;
>>> +    case SCLP_ASSIGN_STORAGE:
>>> +        assign_storage(sccb);
>>> +        break;
>>> +    case SCLP_UNASSIGN_STORAGE:
>>> +        unassign_storage(sccb);
>>> +        break;
>>
>> Do you think it'd be possible to model this whole thing as a device that has its own state? That's where you could keep the bitmap for example. You'd only need some callback mechanism to hook into the SCLP calls, but the PPC guys already have something similar with their hypercalls.
> 
> Speaking of state - in the current model the "is standby storage active" bitmap doesn't get migrated, no?
> 

Yes, you are correct, migration support still needs to be added  - But I
wanted to get this much out for review at least.

> 
> Alex
> 
> 
>