KVM: do not prepare new memslot for KVM_MR_DELETE
Commit Message
kvm_prepare_memory_region() is not useful for KVM_MR_DELETE,
and each kvm_arch_prepare_memory_region() does nothing more than returning
0 for KVM_MR_DELETE.
So, just don't call into kvm_prepare_memory_region() to avoid unnecessary
error handling for KVM_MR_DELETE.
Signed-off-by: Yan Zhao <yan.y.zhao@intel.com>
---
virt/kvm/kvm_main.c | 52 ++++++++++++++++++++++-----------------------
1 file changed, 26 insertions(+), 26 deletions(-)
Comments
On Thu, Nov 10, 2022, Yan Zhao wrote:
> kvm_prepare_memory_region() is not useful for KVM_MR_DELETE,
> and each kvm_arch_prepare_memory_region() does nothing more than returning
> 0 for KVM_MR_DELETE.
This is not true, s390 has an error path that fires in the DELETE case.
/* When we are protected, we should not change the memory slots */
if (kvm_s390_pv_get_handle(kvm))
return -EINVAL;
if (change == KVM_MR_DELETE || change == KVM_MR_FLAGS_ONLY)
return 0;
> So, just don't call into kvm_prepare_memory_region() to avoid unnecessary
> error handling for KVM_MR_DELETE.
Even if the s390 case didn't exit, I would still prefer keeping the code as is.
I agree that a "dummy" call is somewhat confusing, but skipping the arch call
for one operation is equally confusing and could easily lead to future bugs,
e.g. if the order of changes were reversed and an s390 developer wasn't aware
that s390's arch hook would be bypassed on memslot deletion.
On Thu, Nov 10, 2022 at 01:53:47AM +0000, Sean Christopherson wrote:
> On Thu, Nov 10, 2022, Yan Zhao wrote:
> > kvm_prepare_memory_region() is not useful for KVM_MR_DELETE,
> > and each kvm_arch_prepare_memory_region() does nothing more than returning
> > 0 for KVM_MR_DELETE.
>
> This is not true, s390 has an error path that fires in the DELETE case.
>
> /* When we are protected, we should not change the memory slots */
> if (kvm_s390_pv_get_handle(kvm))
> return -EINVAL;
>
> if (change == KVM_MR_DELETE || change == KVM_MR_FLAGS_ONLY)
> return 0;
>
>
> > So, just don't call into kvm_prepare_memory_region() to avoid unnecessary
> > error handling for KVM_MR_DELETE.
>
> Even if the s390 case didn't exit, I would still prefer keeping the code as is.
> I agree that a "dummy" call is somewhat confusing, but skipping the arch call
> for one operation is equally confusing and could easily lead to future bugs,
> e.g. if the order of changes were reversed and an s390 developer wasn't aware
> that s390's arch hook would be bypassed on memslot deletion.
yes, you are right!
@@ -1612,19 +1612,17 @@ static int kvm_prepare_memory_region(struct kvm *kvm,
* new and KVM isn't using a ring buffer, allocate and initialize a
* new bitmap.
*/
- if (change != KVM_MR_DELETE) {
- if (!(new->flags & KVM_MEM_LOG_DIRTY_PAGES))
- new->dirty_bitmap = NULL;
- else if (old && old->dirty_bitmap)
- new->dirty_bitmap = old->dirty_bitmap;
- else if (!kvm->dirty_ring_size) {
- r = kvm_alloc_dirty_bitmap(new);
- if (r)
- return r;
+ if (!(new->flags & KVM_MEM_LOG_DIRTY_PAGES))
+ new->dirty_bitmap = NULL;
+ else if (old && old->dirty_bitmap)
+ new->dirty_bitmap = old->dirty_bitmap;
+ else if (!kvm->dirty_ring_size) {
+ r = kvm_alloc_dirty_bitmap(new);
+ if (r)
+ return r;
- if (kvm_dirty_log_manual_protect_and_init_set(kvm))
- bitmap_set(new->dirty_bitmap, 0, new->npages);
- }
+ if (kvm_dirty_log_manual_protect_and_init_set(kvm))
+ bitmap_set(new->dirty_bitmap, 0, new->npages);
}
r = kvm_arch_prepare_memory_region(kvm, old, new, change);
@@ -1849,21 +1847,23 @@ static int kvm_set_memslot(struct kvm *kvm,
kvm_invalidate_memslot(kvm, old, invalid_slot);
}
- r = kvm_prepare_memory_region(kvm, old, new, change);
- if (r) {
- /*
- * For DELETE/MOVE, revert the above INVALID change. No
- * modifications required since the original slot was preserved
- * in the inactive slots. Changing the active memslots also
- * release slots_arch_lock.
- */
- if (change == KVM_MR_DELETE || change == KVM_MR_MOVE) {
- kvm_activate_memslot(kvm, invalid_slot, old);
- kfree(invalid_slot);
- } else {
- mutex_unlock(&kvm->slots_arch_lock);
+ if (change != KVM_MR_DELETE) {
+ r = kvm_prepare_memory_region(kvm, old, new, change);
+ if (r) {
+ /*
+ * For MOVE, revert the above INVALID change. No
+ * modifications required since the original slot was preserved
+ * in the inactive slots. Changing the active memslots also
+ * release slots_arch_lock.
+ */
+ if (change == KVM_MR_MOVE) {
+ kvm_activate_memslot(kvm, invalid_slot, old);
+ kfree(invalid_slot);
+ } else {
+ mutex_unlock(&kvm->slots_arch_lock);
+ }
+ return r;
}
- return r;
}
/*