On Thu, Aug 14, 2025 at 9:30 AM Jason Gunthorpe [off-list ref] wrote:

On Thu, Aug 07, 2025 at 01:44:14AM +0000, Pasha Tatashin wrote:

quoted

 static int __kho_abort(void)
 {
-     int err = 0;
-     unsigned long order;
-     struct kho_mem_phys *physxa;
-
-     xa_for_each(&kho_out.track.orders, order, physxa) {
-             struct kho_mem_phys_bits *bits;
-             unsigned long phys;
-
-             xa_for_each(&physxa->phys_bits, phys, bits)
-                     kfree(bits);
-
-             xa_destroy(&physxa->phys_bits);
-             kfree(physxa);
-     }
-     xa_destroy(&kho_out.track.orders);

Now nothing ever cleans this up :\

It is solved with stateless KHO. The current implementation is broken,
dropping everything in abort should never happen for stuff that was
independently preserved.

Are you sure the issue isn't in the caller that it shouldn't be
calling kho abort until all the other stuff is cleaned up first?

I feel like this is another case of absuing globals gives an unclear
lifecycle model.

Yes. But, we have a fix for that.

Pasha