RE: [PATCH V1 11/16] xen/ioreq: Introduce hvm_domain_has_ioreq_server()

[Paul Durrant <xadimgnik@xxxxxxxxx>]

> -----Original Message-----
> From: Julien Grall <julien@xxxxxxx>
> Sent: 16 September 2020 09:39
> To: paul@xxxxxxx; 'Jan Beulich' <jbeulich@xxxxxxxx>; 'Oleksandr Tyshchenko' 
> <olekstysh@xxxxxxxxx>
> Cc: xen-devel@xxxxxxxxxxxxxxxxxxxx; 'Oleksandr Tyshchenko' 
> <oleksandr_tyshchenko@xxxxxxxx>; 'Stefano
> Stabellini' <sstabellini@xxxxxxxxxx>; 'Volodymyr Babchuk' 
> <Volodymyr_Babchuk@xxxxxxxx>; 'Andrew
> Cooper' <andrew.cooper3@xxxxxxxxxx>; 'Wei Liu' <wl@xxxxxxx>; 'Roger Pau 
> Monné' <roger.pau@xxxxxxxxxx>;
> 'Julien Grall' <julien.grall@xxxxxxx>
> Subject: Re: [PATCH V1 11/16] xen/ioreq: Introduce 
> hvm_domain_has_ioreq_server()
> 
> 
> 
> On 16/09/2020 09:13, Paul Durrant wrote:
> >> -----Original Message-----
> >> From: Jan Beulich <jbeulich@xxxxxxxx>
> >> Sent: 16 September 2020 09:05
> >> To: Oleksandr Tyshchenko <olekstysh@xxxxxxxxx>; Paul Durrant <paul@xxxxxxx>
> >> Cc: xen-devel@xxxxxxxxxxxxxxxxxxxx; Oleksandr Tyshchenko 
> >> <oleksandr_tyshchenko@xxxxxxxx>; Stefano
> >> Stabellini <sstabellini@xxxxxxxxxx>; Julien Grall <julien@xxxxxxx>; 
> >> Volodymyr Babchuk
> >> <Volodymyr_Babchuk@xxxxxxxx>; Andrew Cooper <andrew.cooper3@xxxxxxxxxx>; 
> >> Wei Liu <wl@xxxxxxx>;
> Roger
> >> Pau Monné <roger.pau@xxxxxxxxxx>; Julien Grall <julien.grall@xxxxxxx>
> >> Subject: Re: [PATCH V1 11/16] xen/ioreq: Introduce 
> >> hvm_domain_has_ioreq_server()
> >>
> >> On 10.09.2020 22:22, Oleksandr Tyshchenko wrote:
> >>> From: Oleksandr Tyshchenko <oleksandr_tyshchenko@xxxxxxxx>
> >>>
> >>> This patch introduces a helper the main purpose of which is to check
> >>> if a domain is using IOREQ server(s).
> >>>
> >>> On Arm the benefit is to avoid calling handle_hvm_io_completion()
> >>> (which implies iterating over all possible IOREQ servers anyway)
> >>> on every return in leave_hypervisor_to_guest() if there is no active
> >>> servers for the particular domain.
> >>>
> >
> > Is this really worth it? The limit on the number of ioreq serves is 
> > small... just 8.
> 
> When I suggested this, I failed to realize there was only 8 IOREQ
> servers available. However, I would not be surprised if this increase
> long term as we want to use

If that happens then we'll probably want to move (back to) a list rather than 
an array...

> 
> > I doubt you'd be able measure the difference.
> Bear in mind that entry/exit to the hypervisor is pretty "cheap" on Arm
> compare to x86. So we want to avoid doing extra work if it is not necessary.
> 

... which will seamlessly deal with this issue.

  Paul

> Cheers,
> 
> --
> Julien Grall