Re: [Xen-devel] [PATCH 1/4] docs: Clarify the expected behaviour of zero length records

[Ian Jackson <ian.jackson@xxxxxxxxxxxxx>]

Andrew Cooper writes ("[PATCH 1/4] docs: Clarify the expected behaviour of zero 
length records"):
> The sending side shouldn't send any variable sized records which end
> up having zero content, but the receiving side will need to tolerate
> such records for compatibility purposes.
...
> +Some records have an exactly specified size.  Some records have
> +variable size depending on their content.  A record with variable
> +size which ends up being zero should be omitted entirely from the
> +stream by the sending side.

I'm not sure this is best stated as a general rule.

Consider a record type which specifies `explicit wombat layout' and
consists of a list of `wombat locations'.

Absence of the record would be semantically different from an empty
record.  In the absence of the record, the receiver would do automatic
choice of number and layout of wombats.  With an empty record, the
receiver would create zero wombats (which might be different to the
default).


I think the underlying spec bug here is that the specification for
`X86_PV_VCPU_BASIC, EXTENDED, XSAVE, MSRS' is too vague.  It simply
specifies
  context          Binary data for this VCPU.
and `The format of these records are identical.  They are all binary
blobs of data which are accessed using specific pairs of domctl
hypercalls.'.


I wasn't able to immediately find the corresponding hypercalls (which
is another docs bug).

But if the hypercalls generate and consume 0-length blocks, then I
think it is fine for the migration stream to contain them.


Have I misunderstood ?

Ian.

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
https://lists.xen.org/xen-devel