Re: [PATCH v9 08/11] KVM: TDX: Get/put PAMT pages when (un)mapping private memory

On Thu, 2026-08-06 at 07:16 +0000, [email protected] wrote:
> > [Severity: Critical]
> > When tdh_mem_sept_add() returns an error like TDX_OPERAND_BUSY, this path
> > calls tdx_pamt_put(), which frees the page back to the global allocator
> > instead of returning it to the vCPU cache.
> > 
> > Since a busy error returns -EBUSY and triggers a RET_PF_RETRY loop in
> > kvm_tdp_mmu_map_private_pfn():
> > 
> > kvm_tdp_mmu_map_private_pfn() {
> >     ...
> >     do {
> >         ...
> >         r = kvm_tdp_mmu_map(vcpu, &fault);
> >     } while (r == RET_PF_RETRY);
> >     ...
> > }
> > 
> > Does this repeatedly drain the vCPU's memory cache without topping it up?
> > 
> > The loop re-attempts the mapping without calling mmu_topup_memory_caches(),
> > meaning repeated contentions could quickly exhaust the cache capacity.

This is the same false positive we get every time. Existing issue. Sean was
going to send a cleanup. Oh he just did.