Re: [PATCH] powerpc/mm: Remove kvm radix prefetch workaround for Power9 DD2.2

["Oliver O'Halloran" <oohall@gmail.com>]

On Mon, Dec 2, 2019 at 2:08 PM Jordan Niethe <jniethe5@gmail.com> wrote:
>
> Commit a25bd72badfa ("powerpc/mm/radix: Workaround prefetch issue with
> KVM") introduced a number of workarounds as coming out of a guest with
> the mmu enabled would make the cpu would start running in hypervisor
> state with the PID value from the guest. The cpu will then start
> prefetching for the hypervisor with that PID value.
>
> In Power9 DD2.2 the cpu behaviour was modified to fix this. When
> accessing Quadrant 0 in hypervisor mode with LPID != 0 prefetching will
> not be performed. This means that we can get rid of the workarounds for
> Power9 DD2.2 and later revisions.
>
> Signed-off-by: Jordan Niethe <jniethe5@gmail.com>
> ---
>  arch/powerpc/kvm/book3s_hv_rmhandlers.S  |  9 +++++++++
>  arch/powerpc/mm/book3s64/radix_pgtable.c | 18 ++++++++++++------
>  arch/powerpc/mm/book3s64/radix_tlb.c     |  5 +++++
>  3 files changed, 26 insertions(+), 6 deletions(-)
>
> diff --git a/arch/powerpc/kvm/book3s_hv_rmhandlers.S b/arch/powerpc/kvm/book3s_hv_rmhandlers.S
> index faebcbb8c4db..6bbc5fbc7ea9 100644
> --- a/arch/powerpc/kvm/book3s_hv_rmhandlers.S
> +++ b/arch/powerpc/kvm/book3s_hv_rmhandlers.S
> @@ -1793,6 +1793,15 @@ END_FTR_SECTION_IFSET(CPU_FTR_ARCH_300)
>         tlbsync
>         ptesync
>
> +       /* We do not need this work around from POWER9 DD2.2 and onwards */
> +       mfspr   r3, SPRN_PVR
> +       srwi    r6, r3, 16
> +       cmpwi   cr0, r6, PVR_POWER9
> +       bne     cr0, 2f
> +       andi.   r3, r3, 0xfff
> +       cmpwi   cr0, r3, 0x202
> +       bge     cr0, 2f

This seems like the sort of thing we'd want to use a CPU_FTR_SECTION()
for. We probably want to have a CPU_FTR for this anyway so we're not
open-coding the PVR check all over the place.

> +
>         /* Radix: Handle the case where the guest used an illegal PID */
>         LOAD_REG_ADDR(r4, mmu_base_pid)
>         lwz     r3, VCPU_GUEST_PID(r9)
> diff --git a/arch/powerpc/mm/book3s64/radix_pgtable.c b/arch/powerpc/mm/book3s64/radix_pgtable.c
> index 6ee17d09649c..1f280124994e 100644
> --- a/arch/powerpc/mm/book3s64/radix_pgtable.c
> +++ b/arch/powerpc/mm/book3s64/radix_pgtable.c
> @@ -312,6 +312,7 @@ static void __init radix_init_pgtable(void)
>  {
>         unsigned long rts_field;
>         struct memblock_region *reg;
> +       unsigned int pvr;
>
>         /* We don't support slb for radix */
>         mmu_slb_size = 0;
> @@ -336,24 +337,29 @@ static void __init radix_init_pgtable(void)
>         }
>
>         /* Find out how many PID bits are supported */
> +       pvr = mfspr(SPRN_PVR);
>         if (cpu_has_feature(CPU_FTR_HVMODE)) {
>                 if (!mmu_pid_bits)
>                         mmu_pid_bits = 20;
>  #ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
>                 /*
> -                * When KVM is possible, we only use the top half of the
> -                * PID space to avoid collisions between host and guest PIDs
> -                * which can cause problems due to prefetch when exiting the
> -                * guest with AIL=3
> +                * Before Power9 DD2.2, when KVM is possible, we only use the
> +                * top half of the PID space to avoid collisions between host
> +                * and guest PIDs which can cause problems due to prefetch when
> +                * exiting the guest with AIL=3
>                  */
> -               mmu_base_pid = 1 << (mmu_pid_bits - 1);
> +               if (PVR_VER(pvr) == PVR_POWER9 && ((0xfff & pvr) < 0x202))
> +                       mmu_base_pid = 1;
> +               else
> +                       mmu_base_pid = 1 << (mmu_pid_bits - 1);
>  #else
>                 mmu_base_pid = 1;
>  #endif
>         } else {
>                 /* The guest uses the bottom half of the PID space */
>                 if (!mmu_pid_bits)
> -                       mmu_pid_bits = 19;
> +                       mmu_pid_bits = (PVR_VER(pvr) == PVR_POWER9 &&
> +                                       ((0xfff & pvr) < 0x202)) ? 19 : 20;
>                 mmu_base_pid = 1;
>         }
>
> diff --git a/arch/powerpc/mm/book3s64/radix_tlb.c b/arch/powerpc/mm/book3s64/radix_tlb.c
> index 67af871190c6..cc86d8a88b86 100644
> --- a/arch/powerpc/mm/book3s64/radix_tlb.c
> +++ b/arch/powerpc/mm/book3s64/radix_tlb.c
> @@ -1217,10 +1217,15 @@ void radix__flush_tlb_all(void)
>  extern void radix_kvm_prefetch_workaround(struct mm_struct *mm)
>  {
>         unsigned long pid = mm->context.id;
> +       unsigned int pvr;
>
>         if (unlikely(pid == MMU_NO_CONTEXT))
>                 return;
>
> +       pvr = mfspr(SPRN_PVR);
> +       if (PVR_VER(pvr) != PVR_POWER9 || ((0xfff & pvr) >= 0x202))
> +               return;
> +
>         /*
>          * If this context hasn't run on that CPU before and KVM is
>          * around, there's a slim chance that the guest on another
> --
> 2.20.1
>