On 02/06/2015 11:52, Peter Crosthwaite wrote:
> > to devirtualize the functions in the common case?
>
> Yes. Can we make it briefer though as CPU_HOOK? I don't think we need
> to references to MULTI through the core code.
Goes without saying. :)
Paolo
On Mon, Jun 1, 2015 at 12:53 PM, Richard Henderson wrote:
> On 05/30/2015 11:11 PM, Peter Crosthwaite wrote:
>> The tcg_ctx global is tricky. It needs to become part of the cpu state.
>> This is done via a pointer.
>
> Why does it need to be virtualized? We already have one for each arch-obj-y.
>
On Mon, Jun 1, 2015 at 1:11 AM, Paolo Bonzini wrote:
>
>
> On 31/05/2015 08:11, Peter Crosthwaite wrote:
>> }
>> -ret = cpu_exec(cpu);
>> +ret = cpu->cpu_exec(cpu);
>
> Perhaps a
>
> #ifdef TARGET_MULTI
> #define MULTI_CPU_HOOK(cpu, fn) (cpu->fn)
> #else
> #define MULTI_CPU_HOOK(cpu,
On 05/30/2015 11:11 PM, Peter Crosthwaite wrote:
> The tcg_ctx global is tricky. It needs to become part of the cpu state.
> This is done via a pointer.
Why does it need to be virtualized? We already have one for each arch-obj-y.
Is it really accessed from outside arch-obj-y?
r~
On 31/05/2015 08:11, Peter Crosthwaite wrote:
> }
> -ret = cpu_exec(cpu);
> +ret = cpu->cpu_exec(cpu);
Perhaps a
#ifdef TARGET_MULTI
#define MULTI_CPU_HOOK(cpu, fn) (cpu->fn)
#else
#define MULTI_CPU_HOOK(cpu, fn) (fn)
#endif
to devirtualize the functions in the common case?
Paolo
