The former expands to a single (memory accessing) insn, which the latter does not guarantee. Yet we'd prefer to read consistent PTEs rather than risking a split read racing with an update done elsewhere.
Signed-off-by: Jan Beulich <jbeul...@suse.com> --- a/xen/arch/x86/pv/mm.c +++ b/xen/arch/x86/pv/mm.c @@ -41,9 +41,7 @@ l1_pgentry_t *map_guest_l1e(unsigned lon return NULL; /* Find this l1e and its enclosing l1mfn in the linear map. */ - if ( copy_from_unsafe(&l2e, - &__linear_l2_table[l2_linear_offset(linear)], - sizeof(l2_pgentry_t)) ) + if ( get_unsafe(l2e, &__linear_l2_table[l2_linear_offset(linear)]) ) return NULL; /* Check flags that it will be safe to read the l1e. */ --- a/xen/arch/x86/pv/mm.h +++ b/xen/arch/x86/pv/mm.h @@ -22,9 +22,7 @@ static inline l1_pgentry_t guest_get_eff toggle_guest_pt(curr); if ( unlikely(!__addr_ok(linear)) || - copy_from_unsafe(&l1e, - &__linear_l1_table[l1_linear_offset(linear)], - sizeof(l1_pgentry_t)) ) + get_unsafe(l1e, &__linear_l1_table[l1_linear_offset(linear)]) ) l1e = l1e_empty(); if ( user_mode )