On Wed, 2022-03-02 at 19:36 +0100, Paolo Bonzini wrote:
> On 3/1/22 00:36, David Woodhouse wrote:
> > On Sat, 2022-02-26 at 00:15 +0000, Sean Christopherson wrote:
> > > Extract the code for allocating guest memory via memfd out of
> > > vm_userspace_mem_region_add() and into a new helper, kvm_memfd_alloc().
> > > A future selftest to populate a guest with the maximum amount of guest
> > > memory will abuse KVM's memslots to alias guest memory regions to a
> > > single memfd-backed host region, i.e. needs to back a guest with memfd
> > > memory without a 1:1 association between a memslot and a memfd instance.
> > > 
> > > No functional change intended.
> > > 
> > > Signed-off-by: Sean Christopherson <seanjc@google.com>
> > 
> > While we're at it, please can we make the whole thing go away and just
> > return failure #ifndef MFD_CLOEXEC, instead of breaking the build on
> > older userspace?
> 
> We can just use old school F_SETFD if that's helpful for you.

The system on which I had that problem doesn't have memfd at all. As a
local hack just to build the TSC test case on a system that actually
supports TSC scaling, I just made it all go away thus:

--- a/tools/testing/selftests/kvm/lib/kvm_util.c
+++ b/tools/testing/selftests/kvm/lib/kvm_util.c
@@ -938,6 +938,7 @@ void vm_userspace_mem_region_add(struct kvm_vm *vm,
 
        region->fd = -1;
        if (backing_src_is_shared(src_type)) {
+#ifdef MFD_CLOEXEC
                int memfd_flags = MFD_CLOEXEC;
 
                if (src_type == VM_MEM_SRC_SHARED_HUGETLB)
@@ -954,6 +955,9 @@ void vm_userspace_mem_region_add(struct kvm_vm *vm,
                                FALLOC_FL_PUNCH_HOLE | FALLOC_FL_KEEP_SIZE, 0,
                                region->mmap_size);
                TEST_ASSERT(ret == 0, "fallocate failed, errno: %i", errno);
+#else
+               TEST_ASSERT(false, "memfd support not present");
+#endif
        }
 
        region->mmap_start = mmap(NULL, region->mmap_size,