On Wed, Feb 17, 2021 at 11:38:07AM +1100, David Gibson wrote:
> On Mon, Feb 15, 2021 at 12:05:41PM +0530, Bharata B Rao wrote:
> > Implement H_RPT_INVALIDATE hcall and add KVM capability
> > KVM_CAP_PPC_RPT_INVALIDATE to indicate the support for the same.
> >
> > This hcall does two types of TLB invalidations:
> >
> > 1. Process-scoped invalidations for guests with LPCR[GTSE]=0.
> > This is currently not used in KVM as GTSE is not usually
> > disabled in KVM.
> > 2. Partition-scoped invalidations that an L1 hypervisor does on
> > behalf of an L2 guest. This replaces the uses of the existing
> > hcall H_TLB_INVALIDATE.
> >
> > In order to handle process scoped invalidations of L2, we
> > intercept the nested exit handling code in L0 only to handle
> > H_TLB_INVALIDATE hcall.
> >
> > Signed-off-by: Bharata B Rao <bharata@xxxxxxxxxxxxx>
> > ---
> > Documentation/virt/kvm/api.rst | 17 +++++
> > arch/powerpc/include/asm/kvm_book3s.h | 3 +
> > arch/powerpc/include/asm/mmu_context.h | 11 +++
> > arch/powerpc/kvm/book3s_hv.c | 91 ++++++++++++++++++++++++
> > arch/powerpc/kvm/book3s_hv_nested.c | 96 ++++++++++++++++++++++++++
> > arch/powerpc/kvm/powerpc.c | 3 +
> > arch/powerpc/mm/book3s64/radix_tlb.c | 25 +++++++
> > include/uapi/linux/kvm.h | 1 +
> > 8 files changed, 247 insertions(+)
> >
> > diff --git a/Documentation/virt/kvm/api.rst b/Documentation/virt/kvm/api.rst
> > index 99ceb978c8b0..416c36aa35d4 100644
> > --- a/Documentation/virt/kvm/api.rst
> > +++ b/Documentation/virt/kvm/api.rst
> > @@ -6038,6 +6038,23 @@ KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR exit notifications which user space
> > can then handle to implement model specific MSR handling and/or user notifications
> > to inform a user that an MSR was not handled.
> >
> > +7.22 KVM_CAP_PPC_RPT_INVALIDATE
> > +------------------------------
> > +
> > +:Capability: KVM_CAP_PPC_RPT_INVALIDATE
> > +:Architectures: ppc
> > +:Type: vm
> > +
> > +This capability indicates that the kernel is capable of handling
> > +H_RPT_INVALIDATE hcall.
> > +
> > +In order to enable the use of H_RPT_INVALIDATE in the guest,
> > +user space might have to advertise it for the guest. For example,
> > +IBM pSeries (sPAPR) guest starts using it if "hcall-rpt-invalidate" is
> > +present in the "ibm,hypertas-functions" device-tree property.
> > +
> > +This capability is always enabled.
>
> I guess that means it's always enabled when it's available - I'm
> pretty sure it won't be enabled on POWER8 or on PR KVM.
Correct, will reword this and restrict this to POWER9, radix etc
>
> > +
> > 8. Other capabilities.
> > ======================
> >
> > diff --git a/arch/powerpc/include/asm/kvm_book3s.h b/arch/powerpc/include/asm/kvm_book3s.h
> > index d32ec9ae73bd..0f1c5fa6e8ce 100644
> > --- a/arch/powerpc/include/asm/kvm_book3s.h
> > +++ b/arch/powerpc/include/asm/kvm_book3s.h
> > @@ -298,6 +298,9 @@ void kvmhv_set_ptbl_entry(unsigned int lpid, u64 dw0, u64 dw1);
> > void kvmhv_release_all_nested(struct kvm *kvm);
> > long kvmhv_enter_nested_guest(struct kvm_vcpu *vcpu);
> > long kvmhv_do_nested_tlbie(struct kvm_vcpu *vcpu);
> > +long kvmhv_h_rpti_nested(struct kvm_vcpu *vcpu, unsigned long lpid,
> > + unsigned long type, unsigned long pg_sizes,
> > + unsigned long start, unsigned long end);
> > int kvmhv_run_single_vcpu(struct kvm_vcpu *vcpu,
> > u64 time_limit, unsigned long lpcr);
> > void kvmhv_save_hv_regs(struct kvm_vcpu *vcpu, struct hv_guest_state *hr);
> > diff --git a/arch/powerpc/include/asm/mmu_context.h b/arch/powerpc/include/asm/mmu_context.h
> > index d5821834dba9..fbf3b5b45fe9 100644
> > --- a/arch/powerpc/include/asm/mmu_context.h
> > +++ b/arch/powerpc/include/asm/mmu_context.h
> > @@ -124,8 +124,19 @@ static inline bool need_extra_context(struct mm_struct *mm, unsigned long ea)
> >
> > #if defined(CONFIG_KVM_BOOK3S_HV_POSSIBLE) && defined(CONFIG_PPC_RADIX_MMU)
> > extern void radix_kvm_prefetch_workaround(struct mm_struct *mm);
> > +void do_h_rpt_invalidate(unsigned long pid, unsigned long lpid,
> > + unsigned long type, unsigned long page_size,
> > + unsigned long psize, unsigned long start,
> > + unsigned long end);
> > #else
> > static inline void radix_kvm_prefetch_workaround(struct mm_struct *mm) { }
> > +static inline void do_h_rpt_invalidate(unsigned long pid,
> > + unsigned long lpid,
> > + unsigned long type,
> > + unsigned long page_size,
> > + unsigned long psize,
> > + unsigned long start,
> > + unsigned long end) { }
> > #endif
> >
> > extern void switch_cop(struct mm_struct *next);
> > diff --git a/arch/powerpc/kvm/book3s_hv.c b/arch/powerpc/kvm/book3s_hv.c
> > index 6f612d240392..802cb77c39cc 100644
> > --- a/arch/powerpc/kvm/book3s_hv.c
> > +++ b/arch/powerpc/kvm/book3s_hv.c
> > @@ -904,6 +904,64 @@ static int kvmppc_get_yield_count(struct kvm_vcpu *vcpu)
> > return yield_count;
> > }
> >
> > +static void do_h_rpt_invalidate_prs(unsigned long pid, unsigned long lpid,
> > + unsigned long type, unsigned long pg_sizes,
> > + unsigned long start, unsigned long end)
> > +{
> > + unsigned long psize;
> > +
> > + if (pg_sizes & H_RPTI_PAGE_64K) {
> > + psize = rpti_pgsize_to_psize(pg_sizes & H_RPTI_PAGE_64K);
> > + do_h_rpt_invalidate(pid, lpid, type, (1UL << 16), psize,
> > + start, end);
> > + }
> > +
> > + if (pg_sizes & H_RPTI_PAGE_2M) {
> > + psize = rpti_pgsize_to_psize(pg_sizes & H_RPTI_PAGE_2M);
> > + do_h_rpt_invalidate(pid, lpid, type, (1UL << 21), psize,
> > + start, end);
> > + }
> > +
> > + if (pg_sizes & H_RPTI_PAGE_1G) {
> > + psize = rpti_pgsize_to_psize(pg_sizes & H_RPTI_PAGE_1G);
> > + do_h_rpt_invalidate(pid, lpid, type, (1UL << 30), psize,
> > + start, end);
> > + }
>
> Hrm. Here you're stepping through the hcall defined pagesizes, then
> mapping each one to the Linux internal page size defs.
>
> It might be more elegant to step through mmu_psize_defs table, and
> conditionally performan an invalidate on that pagesize if the
> corresponding bit in pg_sizes is set (as noted earlier you could
> easily add the H_RPTI_PAGE bit to the table). That way it's a direct
> table lookup rather than a bunch of ifs or switches.
Yes, let me give this a try.
>
> > +}
> > +
> > +static long kvmppc_h_rpt_invalidate(struct kvm_vcpu *vcpu,
> > + unsigned long pid, unsigned long target,
> > + unsigned long type, unsigned long pg_sizes,
> > + unsigned long start, unsigned long end)
> > +{
> > + if (!kvm_is_radix(vcpu->kvm))
> > + return H_UNSUPPORTED;
> > +
> > + if (kvmhv_on_pseries())
> > + return H_UNSUPPORTED;
>
> This doesn't seem quite right. If you have multiply nested guests,
> won't the L2 be issueing H_RPT_INVALIDATE hcalls to the L1 on behalf
> of the L3? The L1 would have to implement them by calling the L0, but
> the L1 can't just reject them, no?
>
> Likewise for the !H_RPTI_TYPE_NESTED case, but on what happens to be a
> nested guest in any case, couldn't this case legitimately arise and
> need to be handled?
The approach is to handle this hcall on behalf of all the nested
guests in L0 only. I am intercepting the nested exit path precisely
for this as shown in the below hunk.
> > @@ -1573,6 +1640,30 @@ static int kvmppc_handle_nested_exit(struct kvm_vcpu *vcpu)
> > if (!xics_on_xive())
> > kvmppc_xics_rm_complete(vcpu, 0);
> > break;
> > + case BOOK3S_INTERRUPT_SYSCALL:
> > + {
> > + unsigned long req = kvmppc_get_gpr(vcpu, 3);
> > +
> > + if (req != H_RPT_INVALIDATE) {
> > + r = RESUME_HOST;
> > + break;
> > + }
> > +
> > + /*
> > + * The H_RPT_INVALIDATE hcalls issued by nested
> > + * guest for process scoped invalidations when
> > + * GTSE=0 are handled here.
> > + */
> > + do_h_rpt_invalidate_prs(kvmppc_get_gpr(vcpu, 4),
> > + vcpu->arch.nested->shadow_lpid,
> > + kvmppc_get_gpr(vcpu, 5),
> > + kvmppc_get_gpr(vcpu, 6),
> > + kvmppc_get_gpr(vcpu, 7),
> > + kvmppc_get_gpr(vcpu, 8));
> > + kvmppc_set_gpr(vcpu, 3, H_SUCCESS);
> > + r = RESUME_GUEST;
> > + break;
> > + }
> > default:
> > r = RESUME_HOST;
> > break;
Thanks for your review.
Regards,
Bharata.
