> -----Original Message-----
> From: Jan Beulich <jbeul...@suse.com>
> Sent: 19 December 2019 13:25
> To: Durrant, Paul <pdurr...@amazon.com>
> Cc: Andrew Cooper <andrew.coop...@citrix.com>; xen-
> de...@lists.xenproject.org; Wei Liu <w...@xen.org>; Roger Pau Monné
> <roger....@citrix.com>
> Subject: Re: [PATCH v2] x86/save: reserve HVM save record numbers that
> have been consumed...
>
> On 19.12.2019 14:15, Durrant, Paul wrote:
> >> From: Andrew Cooper <andrew.coop...@citrix.com>
> >> Sent: 19 December 2019 13:08
> >>
> >> On 19/12/2019 12:37, Durrant, Paul wrote:
> >>>> From: Andrew Cooper <andrew.coop...@citrix.com>
> >>>> Sent: 19 December 2019 12:16
> >>>>
> >>>> On 19/12/2019 12:04, Paul Durrant wrote:
> >>>>> --- a/xen/include/public/arch-x86/hvm/save.h
> >>>>> +++ b/xen/include/public/arch-x86/hvm/save.h
> >>>>> @@ -639,6 +639,8 @@ struct hvm_msr {
> >>>>>
> >>>>> #define CPU_MSR_CODE 20
> >>>>>
> >>>>> +/* Range 22 - 40 reserved for Amazon */
> >>>> What about 21 and 22? And where does the actual number stop, because
> >>>> based on v1, its not 40.
> >>>>
> >>> The range is inclusive (maybe that's not obvious?). For some reason 21
> >> was skipped but why do you say the top is not 40? That was what I set
> >> HVM_SAVE_CODE_MAX to in v1.
> >>
> >> You also said that included prospective headroom, which definitely
> isn't
> >> fair to reserve for ABI breakage reasons.
> >>
> >> Which numbers have actually been allocated?
> >>
> >
> > The problem is that I don't yet know for sure. I have definitely seen
> > patches using 22 thru 34. It is *probably* safe to restrict to that but
> > does it really cost that much more to reserve some extra space just in
> > case? I.e. if someone else adds 41 vs. 35 is it going to make much of a
> > difference?
>
> Not _that much_, but still - it's a bodge, so let's try to limit it as
> much as we can.
>
OK, I'll send a v3 using 34 as the limit.
Paul
> Jan
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel
