Re: RT Xen on ARM - R-Car series

[Andrii Anisov <andrii.anisov@gmail.com>]

Hello Julien,

On 14.02.19 19:29, Julien Grall wrote:
> I am not sure why you are speaking about the current implementation
> when my point was about the new implementation.
> 
> I guess your point stick even if we decide to use guest physical
> address.
For sure. The type of guest address used by hypervisor to reach runstate area and having that area mapped are quite orthogonal questions. But, IMHO, tightly coupled and might be solved together.

> Although, I am still unconvinced of the benefits to keep it
> mapped.
My point was reducing context switch time, but those controversial numbers left me confused.

>> I've measured the raw `update_runstate_area()` execution time. With runstate mapped - its execution time is less than my timer tick (120ns), with runstate not mapped - I've seen
> its execution time as 4 to 8 ticks (480-960ns).
> 
> Please provide the code you use to measure it. How often do you call it?
The code to see that is as simple as following:

diff --git a/xen/arch/arm/domain.c b/xen/arch/arm/domain.c
index 0a2e997..d673d00 100644
--- a/xen/arch/arm/domain.c
+++ b/xen/arch/arm/domain.c
@@ -314,8 +314,16 @@ static void schedule_tail(struct vcpu *prev)
      context_saved(prev);
  
      if ( prev != current )
+    {
+        s_time_t t = 0;
+        if (current->domain->domain_id == 1)
+            t = NOW();
          update_runstate_area(current);
  
+        if (current->domain->domain_id == 1)
+            printk("cp = %"PRI_stime"\n", NOW()-t);
+    }
+
      /* Ensure that the vcpu has an up-to-date time base. */
      update_vcpu_system_time(current);
  }

>> But using TBM, I encountered that making runstate mapped with Roger's patch increases the IRQ latency from ~7000ns to ~7900ns. It is opposite to my expectations and to the raw decrease of `runstate_update_area()` execution time.
> 
> Raw benchmarks should be taken with a grain of salt. The more if you
> only benchmark a single function as the context switch may introduce
> latency/cache eviction.
> 
> Although, I would have expected the numbers to be the same. What is
> your configuration here? Do you have others guests running? How many
> context switch do you have?

The configuration is the same as here [1].


> Also, what are the modifications you made in TBM to use runstate?

I've just registered the runstate area with following [2].

[1] https://lists.xenproject.org/archives/html/xen-devel/2018-12/msg00881.html
[2] https://github.com/aanisov/tbm/commit/806e8a7e6e6c72aef10f1c8a579424252aca2635


-- 
Sincerely,
Andrii Anisov.

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel