[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Qemu-devel] [RFC 04/17] pseries: Make cpu_update during CAS uncondi
From: |
David Gibson |
Subject: |
Re: [Qemu-devel] [RFC 04/17] pseries: Make cpu_update during CAS unconditional |
Date: |
Tue, 8 Nov 2016 14:44:41 +1100 |
User-agent: |
Mutt/1.7.1 (2016-10-04) |
On Fri, Nov 04, 2016 at 11:45:43AM +0100, Thomas Huth wrote:
> On 30.10.2016 12:11, David Gibson wrote:
> > spapr_h_cas_compose_response() includes a cpu_update parameter which
> > controls whether it includes updated information on the CPUs in the device
> > tree fragment returned from the ibm,client-architecture-support (CAS) call.
> >
> > Providing the updated information is essential when CAS has negotiated
> > compatibility options which require different cpu information to be
> > presented to the guest. However, it should be safe to provide in other
> > cases (it will just override the existing data in the device tree with
> > identical data). This simplifies the code by removing the parameter and
> > always providing the cpu update information.
>
> But updating the CPU device tree again and again will also increase the
> QEMU start-up time... Considering that guest start up time is sometimes
> also an issue, do you think that this code simplification really worth
> the effort here?
Given how much it made my brain hurt to try to work the subsequent
changes around that parameter; yes.
If we really have problems with startup time we can revisit this - and
we can probably do better in the context of cleaned up dt building.
--
David Gibson | I'll have my music baroque, and my code
david AT gibson.dropbear.id.au | minimalist, thank you. NOT _the_ _other_
| _way_ _around_!
http://www.ozlabs.org/~dgibson
signature.asc
Description: PGP signature