Re: [Qemu-ppc] [PATCH 4/4] Enable kvm emulated watchdog

[Scott Wood]

On 08/02/2012 10:56 AM, Bhushan Bharat-R65777 wrote:
> 
> 
>> -----Original Message-----
>> From: Wood Scott-B07421
>> Sent: Wednesday, August 01, 2012 11:31 PM
>> To: Bhushan Bharat-R65777
>> Cc: Alexander Graf; address@hidden List; address@hidden; qemu-
>> devel qemu-devel; KVM list
>> Subject: Re: [PATCH 4/4] Enable kvm emulated watchdog
>>
>> On 08/01/2012 12:27 PM, Bhushan Bharat-R65777 wrote:
>>>
>>>
>>>> -----Original Message-----
>>>> From: Alexander Graf [mailto:address@hidden
>>>> Sent: Wednesday, August 01, 2012 7:57 AM
>>>> To: Bhushan Bharat-R65777
>>>> Cc: address@hidden List; address@hidden; Bhushan
>>>> Bharat-R65777; qemu-devel qemu-devel; KVM list
>>>> Subject: Re: [PATCH 4/4] Enable kvm emulated watchdog
>>>>
>>>>
>>>> On 20.07.2012, at 07:23, Bharat Bhushan wrote:
>>>>> @@ -384,6 +437,15 @@ int kvm_arch_init_vcpu(CPUPPCState *cenv)
>>>>>         return ret;
>>>>>     }
>>>>>
>>>>> +    if (enable_watchdog_support) {
>>>>> +        ret = kvm_watchdog_enable(cenv);
>>>>
>>>> Do you think this is a good idea? Why would real hardware not
>>>> implement a watchdog just because the user didn't select an action?
>>>
>>> If there is no watchdog action then why we want to run watchdog timer?
>>
>> On real hardware, if software sets WRC to a non-zero value, the watchdog 
>> action
>> is a system reset.  The user doesn't have to do anything special.
> 
> Scott, maybe I misunderstood you comment, did not you commented to
> enable kvm watchdog if there is watchdog action provided by use.

I changed my mind. :-)

The main difference between the user specifically asking for an action
of system reset, and the user not asking for anything, is that only in
the former case should we error out if watchdog functionality isn't
available -- but if it's a pain to distinguish, don't error out in
either case.

-Scott