Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20220810152033.946942-1-pgonda@google.com> <20220810152033.946942-11-pgonda@google.com>
In-Reply-To: <20220810152033.946942-11-pgonda@google.com>
From:   Vishal Annapurve <vannapurve@google.com>
Date:   Fri, 19 Aug 2022 12:27:44 -0700
Message-ID: <CAGtprH-emXA_5dwwdb4noOC-cuy3BTGT8UbKRkPD8j2gjBSu+Q@mail.gmail.com>
Subject: Re: [V3 10/11] KVM: selftests: Add ucall pool based implementation
To:     Peter Gonda <pgonda@google.com>
Cc:     kvm list <kvm@vger.kernel.org>,
        LKML <linux-kernel@vger.kernel.org>,
        Marc Orr <marcorr@google.com>,
        Sean Christopherson <seanjc@google.com>,
        Michael Roth <michael.roth@amd.com>,
        Tom Lendacky <thomas.lendacky@amd.com>,
        Joerg Roedel <joro@8bytes.org>,
        Mingwei Zhang <mizhang@google.com>,
        Paolo Bonzini <pbonzini@redhat.com>, andrew.jones@linux.dev
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Wed, Aug 10, 2022 at 8:20 AM Peter Gonda <pgonda@google.com> wrote:
> ...
> +
> +static struct ucall *ucall_alloc(void)
> +{
> +       struct ucall *uc = NULL;
> +       int i;
> +
> +       if (!use_ucall_pool)
> +               goto out;
> +
> +       for (i = 0; i < KVM_MAX_VCPUS; ++i) {
> +               if (!atomic_test_and_set_bit(i, ucall_pool->in_use)) {
> +                       uc = &ucall_pool->ucalls[i];
> +                       memset(uc->args, 0, sizeof(uc->args));
> +                       break;
> +               }
> +       }
> +out:
> +       return uc;
> +}
> +
> +static inline size_t uc_pool_idx(struct ucall *uc)
> +{
> +       return uc->hva - ucall_pool->ucalls;
> +}
> +
> +static void ucall_free(struct ucall *uc)
> +{
> +       if (!use_ucall_pool)
> +               return;
> +
> +       clear_bit(uc_pool_idx(uc), ucall_pool->in_use);
> +}
>
>  void ucall(uint64_t cmd, int nargs, ...)
>  {
> -       struct ucall uc = {};
> +       struct ucall *uc;
> +       struct ucall tmp = {};

This steps seems to result in generating instructions that need SSE
support on x86:
struct ucall tmp = {};
   movaps %xmm0,0x20(%rsp)
   movaps %xmm0,0x30(%rsp)
   movaps %xmm0,0x40(%rsp)
   movaps %xmm0,0x50(%rsp)

This initialization will need proper compilation flags to generate
instructions according to VM configuration.

>         va_list va;
>         int i;
>
> -       WRITE_ONCE(uc.cmd, cmd);
> +       uc = ucall_alloc();
> +       if (!uc)
> +               uc = &tmp;
> +
> +       WRITE_ONCE(uc->cmd, cmd);
>
>         nargs = min(nargs, UCALL_MAX_ARGS);
>
>         va_start(va, nargs);
>         for (i = 0; i < nargs; ++i)
> -               WRITE_ONCE(uc.args[i], va_arg(va, uint64_t));
> +               WRITE_ONCE(uc->args[i], va_arg(va, uint64_t));
>         va_end(va);
>
> -       ucall_arch_do_ucall((vm_vaddr_t)&uc);
> ...