Re: [PATCH v19 11/17] RISC-V: KVM: Implement MMU notifiers

From: Anup Patel
Date: Wed Aug 04 2021 - 03:17:12 EST


On Tue, Aug 3, 2021 at 6:49 PM limingwang (A) <limingwang@xxxxxxxxxx> wrote:
>
> > diff --git a/arch/riscv/kvm/mmu.c b/arch/riscv/kvm/mmu.c index
> > fa9a4f9b9542..4b294113c63b 100644
> > --- a/arch/riscv/kvm/mmu.c
> > +++ b/arch/riscv/kvm/mmu.c
> > @@ -300,7 +300,8 @@ static void stage2_op_pte(struct kvm *kvm, gpa_t
> > addr,
> > }
> > }
> >
> > -static void stage2_unmap_range(struct kvm *kvm, gpa_t start, gpa_t size)
> > +static void stage2_unmap_range(struct kvm *kvm, gpa_t start,
> > + gpa_t size, bool may_block)
> > {
> > int ret;
> > pte_t *ptep;
> > @@ -325,6 +326,13 @@ static void stage2_unmap_range(struct kvm *kvm,
> > gpa_t start, gpa_t size)
> >
> > next:
> > addr += page_size;
> > +
> > + /*
> > + * If the range is too large, release the kvm->mmu_lock
> > + * to prevent starvation and lockup detector warnings.
> > + */
> > + if (may_block && addr < end)
> > + cond_resched_lock(&kvm->mmu_lock);
> > }
> > }
> >
> > @@ -405,7 +413,6 @@ static int stage2_ioremap(struct kvm *kvm, gpa_t gpa,
> > phys_addr_t hpa,
> > out:
> > stage2_cache_flush(&pcache);
> > return ret;
> > -
> > }
> >
> > void kvm_arch_mmu_enable_log_dirty_pt_masked(struct kvm *kvm, @@
> > -547,7 +554,7 @@ int kvm_arch_prepare_memory_region(struct kvm *kvm,
> > spin_lock(&kvm->mmu_lock);
> > if (ret)
> > stage2_unmap_range(kvm, mem->guest_phys_addr,
> > - mem->memory_size);
> > + mem->memory_size, false);
> > spin_unlock(&kvm->mmu_lock);
> >
> > out:
> > @@ -555,6 +562,73 @@ int kvm_arch_prepare_memory_region(struct kvm
> > *kvm,
> > return ret;
> > }
> >
> > +bool kvm_unmap_gfn_range(struct kvm *kvm, struct kvm_gfn_range *range)
> > +{
> > + if (!kvm->arch.pgd)
> > + return 0;
> > +
> > + stage2_unmap_range(kvm, range->start << PAGE_SHIFT,
> > + (range->end - range->start) << PAGE_SHIFT,
> > + range->may_block);
> > + return 0;
> > +}
> > +
> > +bool kvm_set_spte_gfn(struct kvm *kvm, struct kvm_gfn_range *range) {
> > + int ret;
> > + kvm_pfn_t pfn = pte_pfn(range->pte);
> > +
> > + if (!kvm->arch.pgd)
> > + return 0;
> > +
> > + WARN_ON(range->end - range->start != 1);
> > +
> > + ret = stage2_map_page(kvm, NULL, range->start << PAGE_SHIFT,
> > + __pfn_to_phys(pfn), PAGE_SIZE, true, true);
> > + if (ret) {
> > + kvm_err("Failed to map stage2 page (error %d)\n", ret);
> > + return 1;
> > + }
>
> Hi, Anup
>
> I think that it is not appropriate to add kvm_err here, because stage2_set_pte function
> may apply for memory based on the pcache parameter. If the value of pcache is NULL,
> stage2_set_pte function considers that there is not enough memory and here an invalid
> error log is generated.
>
> As an example, this error log is printed when a VM is migrating. But finally the VM migration
> is successful. And if the kvm_err is added to the same position in the ARM architecture, the
> same error log is also printed.

Okay, I have converted kvm_err() to kvm_debug(). In the future, we can totally
remove it as well.

Please try riscv_kvm_v20 branch at:
https://github.com/avpatel/linux.git

Regards,
Anup

>
> Mingwang
>
> > + return 0;
> > +}
> > +
>