This change https://review.trustedfirmware.org/c/hafnium/hafnium/+/16649 removed hypervisor S1 PT defrag calls to overcome a latent issue when the memory size to defragment is large.
The issue most probably has always existed but is exhibited when the number of cores grows as the work memory needed by the hypervisor for stacks, PE and vCPU state increases.

The intent of this ticket is to add back the defrag calls by tuning the design to overcome above breakage.