[PATCH v2 2/2] RISC-V: KVM: Fix skip of valid pages in kvm_riscv_gstage_unmap_range

[Anup Patel]

On Fri, Jun 5, 2026 at 4:33 AM <atwufei at 163.com> wrote:
>
> From: Wu Fei <wu.fei9 at sanechips.com.cn>
>
> Same as kvm_riscv_gstage_wp_range, the possible valid pages should not
> be skipped if !found_leaf. Different from wp case, which can
> write-protect more than asked, unmap can't do that, no splitting is
> added right now but a warning is logged instead.
>
> Signed-off-by: Wu Fei <wu.fei9 at sanechips.com.cn>

LGTM.

Reviewed-by: Anup Patel <anup at brainfault.org>

Queued this patch for Linux-7.2

Thanks,
Anup


> ---
>  arch/riscv/kvm/gstage.c | 20 ++++++++++++--------
>  1 file changed, 12 insertions(+), 8 deletions(-)
>
> diff --git a/arch/riscv/kvm/gstage.c b/arch/riscv/kvm/gstage.c
> index 2b141a78ecac..d78f5aeb8b40 100644
> --- a/arch/riscv/kvm/gstage.c
> +++ b/arch/riscv/kvm/gstage.c
> @@ -395,15 +395,19 @@ void kvm_riscv_gstage_unmap_range(struct kvm_gstage *gstage,
>                 if (ret)
>                         break;
>
> -               if (!found_leaf)
> -                       goto next;
> -
> -               if (!(addr & (page_size - 1)) && ((end - addr) >= page_size))
> -                       kvm_riscv_gstage_op_pte(gstage, addr, ptep,
> -                                               ptep_level, GSTAGE_OP_CLEAR);
> +               if (!found_leaf) {
> +                       addr = ALIGN(addr + 1, page_size);
> +               } else {
> +                       if (!(addr & (page_size - 1)) && ((end - addr) >= page_size))
> +                               kvm_riscv_gstage_op_pte(gstage, addr, ptep,
> +                                                       ptep_level, GSTAGE_OP_CLEAR);
> +                       else {
> +                               WARN_ONCE(1, "Skip unmap range addr: %#llx, end: %#llx, page_size: %#lx\n",
> +                                               addr, end, page_size);
> +                       }
>
> -next:
> -               addr += page_size;
> +                       addr += page_size;
> +               }
>
>                 /*
>                  * If the range is too large, release the kvm->mmu_lock
> --
> 2.51.0
>