On 13.07.2018 12:18, Janosch Frank wrote:
> On 13.07.2018 11:00, David Hildenbrand wrote:
>> On 13.07.2018 08:37, Janosch Frank wrote:
>>> Let's allow huge pmd linking when enabled through the
>>> KVM_CAP_S390_HPAGE_1M capability. Also we can now restrict gmap
>>> invalidation and notification to the cases where the capability has
>>> been activated and save some cycles when that's not the case.
>>>
>>> @@ -435,7 +435,7 @@ pmd_t pmdp_xchg_direct(struct mm_struct *mm, unsigned long addr,
>>> pmd_t old;
>>>
>>> preempt_disable();
>>> - if (mm_has_pgste(mm)) {
>>> + if (mm_has_pgste(mm) && mm->context.allow_gmap_hpage_1m) {
>>> pmdp_clear_skeys(mm, pmdp, new);
>>> pmdp_notify(mm, addr);
>>> }
>>> @@ -452,7 +452,7 @@ pmd_t pmdp_xchg_lazy(struct mm_struct *mm, unsigned long addr,
>>> pmd_t old;
>>>
>>> preempt_disable();
>>> - if (mm_has_pgste(mm)) {
>>> + if (mm_has_pgste(mm) && mm->context.allow_gmap_hpage_1m) {
>>> pmdp_clear_skeys(mm, pmdp, new);
>>> pmdp_notify(mm, addr);
>>> }
>>>
>>
>> You should probably change the order of the last two patches, so you
>> really only expose the capability if support for huge pages is there.
>>
>
> Which would mean I don't have mm->context.allow_gmap_hpage_1m in this
> one and compilation breaks...
>
Well you can easily move introduction also to this patch.
--
Thanks,
David / dhildenb
