Re: [Xen-devel] [PATCH 5/8] tools/libxc: Rework xc_cpuid_set() to use {get, set}_cpu_policy()

[Andrew Cooper <andrew.cooper3@citrix.com>]

On 12/09/2019 09:19, Jan Beulich wrote:
> On 11.09.2019 22:05, Andrew Cooper wrote:
>> The purpose of this change is to stop using xc_cpuid_do_domctl(), and to stop
>> basing decisions on a local CPUID instruction.  This is not an appropriate way
>> to construct policy information for other domains.
>>
>> Obtain the host and domain-max policies from Xen, and mix the results as
>> before.  Provide rather more error logging than before.
> And this passing through of host values is meant to stay long
> term? Shouldn't this rather be passing through of max-policy
> values, with max-policy long term wider than default-policy? The
> change itself looks good to me, but before giving my R-b I'd
> like to understand this aspect.

There is a very large amount wrong with xc_cpuid_set().

For one, its behaviour is only useful for feature leaves, but it will
operate on any leaves the user requests, and while it is capable of
returning errors, libxl doesn't check the return value and continues
blindly forwards.

Also from the L1TF embargo days is a series of work (or at least, the
start of) which is a total overhaul of how libxl and libxc interact when
it comes CPUID and MSR settings.  Neither xc_cpuid_set() nor
xc_cpuid_apply_policy() will survive to the end.

For now, I've opted with not changing the semantics of the calls while
altering the data-transfer mechanism, to avoid conflating the two areas
of work.

~Andrew

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