Sean Christopherson <[email protected]> writes: > > [...snip...] > > But, Ackerley and Fuad want give kvm_vm_release() the same treatment[*], at > which > point there's no good reason not to be paranoid. I want to do that in a > dedicated > patch though, and harden "everything" in one shot. I'll send something like > the > below. >
Thanks! > [*] https://lore.kernel.org/all/[email protected] > > diff --git tools/testing/selftests/kvm/lib/kvm_util.c > tools/testing/selftests/kvm/lib/kvm_util.c > index 2a76eca7029d..2476167252a1 100644 > --- tools/testing/selftests/kvm/lib/kvm_util.c > +++ tools/testing/selftests/kvm/lib/kvm_util.c > @@ -737,6 +737,12 @@ userspace_mem_region_find(struct kvm_vm *vm, u64 start, > u64 end) > return NULL; > } > > +static void kvm_free_fd(int *fd) Not where the line is drawn between "free" vs "release" in the selftests, just wanted to draw your attention to the two terms we can pick the correct term. > +{ > + kvm_close(*fd); > + *fd = -1; > +} > + > > [...snip...] >
