Re: [PATCH 2/3] target/ppc: Have gen_pause() actually pause vCPUs

["Philippe Mathieu-Daudé" <philmd@linaro.org>]

(Cc'ing Aaron for commit 9e196938aa1 "target-ppc: gen_pause for
instructions: yield, mdoio, mdoom, miso")

On 30/9/25 00:59, Benjamin Herrenschmidt wrote:
> On Mon, 2025-09-29 at 09:51 +0200, Philippe Mathieu-Daudé wrote:
>>> What will resume it though ? The smt_low() case isn't meant to
>>> *halt*
>>> the CPUs permanently. smt_*() levels are about SMT thread
>>> priorities.
>>> Using a "pause" that just gets out of TCG (and back in), is a way
>>> to
>>> "yield" to another thread, thus enabling more forward progress when
>>> a
>>
>> What you describe can be achieved with a helper doing:
>>
>>     cs->exception_index = EXCP_YIELD;
>>     cpu_loop_exit(cs);
>>
>> Is that what you wanted?
> 
> I suppose so... this was many years ago and I don't have much context
> anymore, so I don't know why I didn't do it that way back then.

I *think* Nick implemented something similar for sPAPR in commit
e8ce0e40ee9 ("spapr: Implement H_CONFER"):

     /*
      * The targeted confer does not do anything special beyond yielding
      * the current vCPU, but even this should be better than nothing.
      * At least for single-threaded tcg, it gives the target a chance to
      * run before we run again. Multi-threaded tcg does not really do
      * anything with EXCP_YIELD yet.
      */

     cs->exception_index = EXCP_YIELD;
     cpu_exit(cs);

Nick, would that make sense to implement smt_low?