On 09/12/16 13:28, Jan Beulich wrote:
>>>> On 09.12.16 at 14:00, <andrew.coop...@citrix.com> wrote:
>> On 09/12/16 11:54, Jan Beulich wrote:
>>> Commit dc88221c97 ("x86: rename XMM* features to SSE*") pointlessly
>>> added them - these features are always available on 64-bit CPUs. (Let's
>>> not assume this for MMX though in at least the insn emulator.)
>>>
>>> Signed-off-by: Jan Beulich <jbeul...@suse.com>
>> This isn't necessarily true when compiled for 32bit in the userspace
>> harness.
> In the test harness vcpu_has_* == cpu_has_*, as also
> demonstrated by
>
> #define host_and_vcpu_must_have(feat) vcpu_must_have(feat)
>
> . The change (as its title is trying to say) really only affects the
> hypervisor.
Right, but this change is still contrary to the written requirement.
/*
* Note the difference between vcpu_must_have_<feature>()
and
* host_and_vcpu_must_have(<feature>): The latter needs to be used
when
* emulation code is using the same instruction class for carrying
out
* the actual
operation.
*/
We are using SSE and SSE2 instructions for carrying out that emulation,
so should still be using the host_and_vcpu check.
Swapping the hypervisor defines to being 1 will cause the
generate_exception_if() clause to become dead and get dropped, turning
host_and_vcpu_must_have() into just vcpu_must_have_##feat
~Andrew
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
https://lists.xen.org/xen-devel
