On 30.01.2021 16:22, Julien Grall wrote:
> @@ -1442,13 +1447,6 @@ long do_memory_op(unsigned long cmd,
> XEN_GUEST_HANDLE_PARAM(void) arg)
> if ( d == NULL )
> return -ESRCH;
>
> - rc = xatp_permission_check(d, xatpb.space);
> - if ( rc )
> - {
> - rcu_unlock_domain(d);
> - return rc;
> - }
> -
> rc = xenmem_add_to_physmap_batch(d, &xatpb, start_extent);
>
> rcu_unlock_domain(d);
I'd be okay with the code movement if you did so consistently,
i.e. also for the other invocation. I realize this would have
an effect on the dm-op call of the function, but I wonder
whether this wouldn't even be a good thing. If not, I think
duplicating xenmem_add_to_physmap()'s early ASSERT() into
xenmem_add_to_physmap_batch() would be the better course of
action.
Jan
