[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAMj1kXGQhj4RK=Ks_WD59hG1FzX=vsEkugDEW1tmzJzjahrB4g@mail.gmail.com>
Date: Thu, 26 Sep 2024 12:55:15 +0200
From: Ard Biesheuvel <ardb@...nel.org>
To: Ard Biesheuvel <ardb+git@...gle.com>
Cc: linux-kernel@...r.kernel.org, Jason Andryuk <jason.andryuk@....com>,
Juergen Gross <jgross@...e.com>, Boris Ostrovsky <boris.ostrovsky@...cle.com>, x86@...nel.org,
xen-devel@...ts.xenproject.org
Subject: Re: [PATCH 1/5] x86/pvh: Call C code via the kernel virtual mapping
On Thu, 26 Sept 2024 at 12:41, Ard Biesheuvel <ardb+git@...gle.com> wrote:
>
> From: Ard Biesheuvel <ardb@...nel.org>
>
> Calling C code via a different mapping than it was linked at is
> problematic, because the compiler assumes that RIP-relative and absolute
> symbol references are interchangeable. GCC in particular may use
> RIP-relative per-CPU variable references even when not using -fpic.
>
> So call xen_prepare_pvh() via its kernel virtual mapping on x86_64, so
> that those RIP-relative references produce the correct values. This
> matches the pre-existing behavior for i386, which also invokes
> xen_prepare_pvh() via the kernel virtual mapping before invoking
> startup_32 with paging disabled again.
>
> Fixes: 7243b93345f7 ("xen/pvh: Bootstrap PVH guest")
> Tested-by: Jason Andryuk <jason.andryuk@....com>
> Reviewed-by: Jason Andryuk <jason.andryuk@....com>
> Signed-off-by: Ard Biesheuvel <ardb@...nel.org>
> ---
> arch/x86/platform/pvh/head.S | 8 +++++++-
> 1 file changed, 7 insertions(+), 1 deletion(-)
>
> diff --git a/arch/x86/platform/pvh/head.S b/arch/x86/platform/pvh/head.S
> index 64fca49cd88f..98ddd552885a 100644
> --- a/arch/x86/platform/pvh/head.S
> +++ b/arch/x86/platform/pvh/head.S
> @@ -172,7 +172,13 @@ SYM_CODE_START_LOCAL(pvh_start_xen)
> movq %rbp, %rbx
> subq $_pa(pvh_start_xen), %rbx
> movq %rbx, phys_base(%rip)
> - call xen_prepare_pvh
> +
> + /* Call xen_prepare_pvh() via the kernel virtual mapping */
> + leaq xen_prepare_pvh(%rip), %rax
Just realized that we probably need
+ subq phys_base(%rip), %rax
here (or grab phys_base from %rbx, but the above is more obvious and
less likely to get broken in the future)
> + addq $__START_KERNEL_map, %rax
> + ANNOTATE_RETPOLINE_SAFE
> + call *%rax
> +
> /*
> * Clear phys_base. __startup_64 will *add* to its value,
> * so reset to 0.
> --
> 2.46.0.792.g87dc391469-goog
>
Powered by blists - more mailing lists