Hi Ricardo,
On Sat, Nov 12, 2022 at 08:17:03AM +0000, Ricardo Koller wrote:
> stage2_make_pte() throws a warning when used in a non-shared walk, as PTEs
> are not "locked" when walking non-shared. Add a check so it can be used
> non-shared.
>
> Signed-off-by: Ricardo Koller <ricarkol@xxxxxxxxxx>
I would very much prefer to leave this WARN as-is. Correct me if I am
wrong, but I do not believe this warning is firing with the existing
code.
While the locking portion doesn't make a whole lot of sense for a
non-shared walk, it is also a magic value that indicates we've already
done the break side of break-before-make. If the warning fires then that
would suggest our break-before-make implementation isn't working as
expected.
--
Thanks,
Oliver
> ---
> arch/arm64/kvm/hyp/pgtable.c | 3 ++-
> 1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/arch/arm64/kvm/hyp/pgtable.c b/arch/arm64/kvm/hyp/pgtable.c
> index c12462439e70..b16107bf917c 100644
> --- a/arch/arm64/kvm/hyp/pgtable.c
> +++ b/arch/arm64/kvm/hyp/pgtable.c
> @@ -733,7 +733,8 @@ static void stage2_make_pte(const struct kvm_pgtable_visit_ctx *ctx, kvm_pte_t n
> {
> struct kvm_pgtable_mm_ops *mm_ops = ctx->mm_ops;
>
> - WARN_ON(!stage2_pte_is_locked(*ctx->ptep));
> + if (kvm_pgtable_walk_shared(ctx))
> + WARN_ON(!stage2_pte_is_locked(*ctx->ptep));
>
> if (stage2_pte_is_counted(new))
> mm_ops->get_page(ctx->ptep);
> --
> 2.38.1.431.g37b22c650d-goog
>
>
_______________________________________________
kvmarm mailing list
kvmarm@xxxxxxxxxxxxxxxxxxxxx
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
