Re: [Xen-devel] [PATCH 01/11] public / x86: introduce hvmctl hypercall

[Andrew Cooper <andrew.cooper3@xxxxxxxxxx>]

On 23/06/16 16:10, Jan Beulich wrote:
>>>> On 23.06.16 at 16:55, <andrew.cooper3@xxxxxxxxxx> wrote:
>> On 20/06/16 13:52, Jan Beulich wrote:
>>> +/*
>>> + * Note that this value is effectively part of the ABI, even if we don't 
>> need
>>> + * to make it a formal part of it.  Hence this value may only be changed if
>>> + * accompanied by a suitable interface version increase.
>>> + */
>>> +#define HVMCTL_iter_shift 8
>>> +#define HVMCTL_iter_mask  ((1U << HVMCTL_iter_shift) - 1)
>>> +#define HVMCTL_iter_max   (1U << (16 + HVMCTL_iter_shift))
>> This (mis)use of the cmd parameter is surely no longer necessary, given
>> that there is space in xen_hvmctl_t to encode continuation information?
> There's no misuse of cmd anymore. This is just use to make the 16-bit
> continuation value (the opaque structure member) cover a more useful
> range, and at once avoid doing the preemption check on every
> iteration.

Ah ok, but it does leave the minimum iteration at 256, which could
easily be too large, depending on the underlying operation.

In this case, I think it would be far better to bump the cmd field to 32
bits, and opaque to 64bits, which affords us far more flexibility.

~Andrew

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