Re: [kvm-devel] [PATCH][RFC] kvm-scheduler integration

[Avi Kivity <avi@qumranet.com>]

Rusty Russell wrote:
> On Sun, 2007-07-08 at 15:48 +0200, Ingo Molnar wrote:
>   
>> * Avi Kivity <avi@qumranet.com> wrote:
>>
>>     
>>>>> +#ifdef CONFIG_SCHED_KVM
>>>>> +static __read_mostly struct sched_kvm_hooks kvm_hooks;
>>>>> +#endif
>>>>>           
>>>> please just add a current->put_vcpu() function pointer instead of 
>>>> this hooks thing.
>>>>         
>>> Won't that increase task_struct (16 bytes on 64-bit) unnecessarily?  
>>> The function pointers are common to all virtual machines.
>>>       
>> well, this function pointer could then be reused by other virtual 
>> machines as well, couldnt it? If the task struct overhead is a problem 
>> (it really isnt, and it's dependent on CONFIG_KVM) then we could switch 
>> it around to a notifier-alike mechanism.
>>     
>
> OK, this patch is *ugly*.  Not that there's anything wrong with a patch
> which says "I'm going to preempt you", but making it kvm-specific is
> ugly.  ISTR times past where I wanted such a hook, although none spring
> immediately into my pre-coffee brain.
>
> I think a "struct preempt_ops *" and a "void *preempt_ops_data" inside
> every task struct is a better idea.  Call the config option
> PREEMPT_SCHED_HOOKS and now there's nothing kvm-specific about it...
>   

I considered that, but your proposal does not allow a single task to
have multiple preemption hooks installed (hookers?!).  Since in general
there's no reason to suppose that users would be mutually exclusive, we
need to have a struct hlist of these things.  All in all this seemed to
indicate that the second user should have the honor of figuring out that
stuff.

-- 
Do not meddle in the internals of kernels, for they are subtle and quick to panic.