Re: [Xen-devel] [RFC 1/6] xen/arm: Re-enable interrupt later in the trap path

[Andrii Anisov <andrii.anisov@xxxxxxxxx>]

Hello Julien,

On 30.07.19 23:10, Julien Grall wrote:

> > In this series I think I need interrupts locked until I start time accounting 
> > for hypervisor. Time accounting is started by `tacc_head()` function. I prefer 
> > to have it called from C, because it is more convenient and obvious for those 
> > who are less familiar with the ARM code.
Here is the question to you: what is the best place (and way) to start 
hypervisor time tracking?

> > > Resending the patch without things addressed is only going to make it worst.
> > I'm still convinced the patch would improve interrupt latency for high 
> > interrupt rate use cases.
> > But I understand that I have no experiment to prove the effect, so I'm not 
> > willing to push through the patch.
> The only thing I ask is justification in your commit message rather than 
> throwing things and expecting the reviewer to understand why you do that. I 
> would recommend to refresh yourself how to submit a patch series [1].
I'll follow you recommendation.

> > Also, I have a question to you about another aspect of this patch. In the 
> > function `enter_hypervisor_head()` there is a check for a disabled workaround 
> > and turning it on. If we have the interrupts enabled until there, we have good 
> > chances to go through the interrupt processing `do_IRQ()` before WA enabled. Is 
> > it still OK?
> Hmmm, that's correct.
Sorry I did not get your point. What exactly is correct? My observation of the 
scenario where we can go through the big piece of the hypervisor without WA 
enabled? Or processing IRQs without WA enabled is the correct way to do?

--
Sincerely,
Andrii Anisov.

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