Re: [Qemu-devel] [PATCH qom-cpu 25/59] cpu: Change qemu_init_vcpu() argument to CPUState

["Andreas Färber" <afaerber@suse.de>]

Am 16.06.2013 10:27, schrieb Andreas Färber:
> 嗨,
> 
> Am 11.06.2013 04:39, schrieb li guang:
>> 在 2013-06-09日的 21:12 +0200，Andreas Färber写道：
>>> This allows to move the call to CPUState's realizefn.
>>> Therefore move the stub into libqemustub.a.
>>>
>>> Signed-off-by: Andreas Färber <afaerber@suse.de>
>> [...] 
>>> diff --git a/target-arm/cpu.c b/target-arm/cpu.c
>>> index a40290c..88ee61b 100644
>>> --- a/target-arm/cpu.c
>>> +++ b/target-arm/cpu.c
>>> @@ -205,7 +205,6 @@ static void arm_cpu_realizefn(DeviceState *dev, Error **errp)
>>>      arm_cpu_register_gdb_regs_for_features(cpu);
>>>  
>>>      cpu_reset(CPU(cpu));
>>
>> can this and following 'cpu_reset' also be moved to
>> 'cpu_common_realizefn' like 'qemu_init_vcpu'?
> 
> I have a branch qom-cpu-reset that tried this. However the culprit was
> that alpha reset was broken and we have not yet fully fixed it.
> I.e., through the memset() up to "breakpoints" some pointers would get
> overridden, so calling it is not a good idea at the moment. But if you
> have some spare time and an alpha guest... ;)

Sorry. While the above is still an issue, the reason for not being able
to make the change yet were rather sparc and ppc, who do not reset on
creation today. For alpha it would simply be a no-op today due to lack
of reset (experimental implementation on qom-cpu-alpha branch).

Andreas

> I chose to leave the cpu_index logging line in place until I can move it
> together with log_cpu_state() (depended on patch 18/59), for which I
> have a patch on qom-cpu-11 branch. :)
> 
> Cheers,
> Andreas
> 
>>> -    qemu_init_vcpu(env);
>>>  
>>>      acc->parent_realize(dev, errp);
>>>  }
> [snip]
> 


-- 
SUSE LINUX Products GmbH, Maxfeldstr. 5, 90409 Nürnberg, Germany
GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer; HRB 16746 AG Nürnberg