linux-kernel - Re: [PATCH v6 4/4] x86/xen: Add "nopv" support for HVM guest

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <3d7db7c6-cea3-9dce-0519-a1c600b33273@oracle.com>
Date:   Tue, 9 Jul 2019 10:54:32 -0400
From:   Boris Ostrovsky <boris.ostrovsky@...cle.com>
To:     Zhenzhong Duan <zhenzhong.duan@...cle.com>,
        linux-kernel@...r.kernel.org
Cc:     xen-devel@...ts.xenproject.org, jgross@...e.com,
        sstabellini@...nel.org, tglx@...utronix.de, mingo@...hat.com,
        bp@...en8.de
Subject: Re: [PATCH v6 4/4] x86/xen: Add "nopv" support for HVM guest

On 7/9/19 12:20 AM, Zhenzhong Duan wrote:
>
> -const __initconst struct hypervisor_x86 x86_hyper_xen_hvm = {
> +static uint32_t __init xen_platform_hvm(void)
> +{
> +       uint32_t xen_domain = xen_cpuid_base();
> +       struct x86_hyper_init *h = &x86_hyper_xen_hvm.init;
> +
> +       if (xen_pv_domain())
> +               return 0;
> +
> +       if (xen_pvh_domain() && nopv) {
> +               /* Guest booting via the Xen-PVH boot entry goes here */
> +               pr_info("\"nopv\" parameter is ignored in PVH guest\n");
> +               nopv = false;
> +       } else if (nopv && xen_domain) {
> +               /*
> +                * Guest booting via normal boot entry (like via
> grub2) goes
> +                * here.
> +                *
> +                * Use interface functions for bare hardware if nopv,
> +                * xen_hvm_guest_late_init is an exception as we need to
> +                * detect PVH and panic there.
> +                */
> +               memcpy(h, (void *)&x86_init.hyper,
> sizeof(x86_init.hyper));


And this worked? I'd think it would fail since h points to RO section.


> +               memcpy(&x86_hyper_xen_hvm.runtime, (void
> *)&x86_platform.hyper,
> +                      sizeof(x86_platform.hyper));
> +               h->guest_late_init = xen_hvm_guest_late_init;


To me this still doesn't look right --- you are making assumptions about
x86_platform/x86_init.hyper and I don't think you can assume they have
not been set to point to another hypervisor, for example.

Would modifying all x86_hyper_xen_hvm's ops (except, I guess,
xen_hvm_guest_late_init()) to immediately return if nopv is set work?

Also question for Juergen --- is there anything perhaps in zeropage that
indicates we are booting PVH guest from grub?

-boris


> +       }
> +       return xen_domain;
> +}
> +
> +struct hypervisor_x86 x86_hyper_xen_hvm __initdata = {
>         .name                   = "Xen HVM",
>         .detect                 = xen_platform_hvm,
>         .type                   = X86_HYPER_XEN_HVM,
> @@ -268,4 +292,5 @@ static __init void xen_hvm_guest_late_init(void)
>         .init.init_mem_mapping  = xen_hvm_init_mem_mapping,
>         .init.guest_late_init   = xen_hvm_guest_late_init,
>         .runtime.pin_vcpu       = xen_pin_vcpu,
> +       .ignore_nopv            = true,
>  };
>
>
> Thanks
>
> Zhenzhong
>