Re: [RFC PATCH] x86: Make sure verify_cpu has a good stack

[Brian Gerst]

On Wed, Mar 2, 2016 at 6:20 AM, Borislav Petkov <bp@xxxxxxxxx> wrote:
> From: Borislav Petkov <bp@xxxxxxx>
>
> 04633df0c43d ("x86/cpu: Call verify_cpu() after having entered long mode too")
> added the call to verify_cpu() for sanitizing CPU configuration.
>
> The latter uses the stack minimally and it can happen that we land in
> startup_64() directly from a 64-bit bootloader. Then we want to use our
> own, known good stack.
>
> Do that.
>
> APs don't need this as the trampoline sets up a stack for them.
>
> Reported-by: Tom Lendacky <thomas.lendacky@xxxxxxx>
> Signed-off-by: Borislav Petkov <bp@xxxxxxx>
> ---
>  arch/x86/kernel/head_64.S | 4 ++++
>  1 file changed, 4 insertions(+)
>
> diff --git a/arch/x86/kernel/head_64.S b/arch/x86/kernel/head_64.S
> index 22fbf9df61bb..d60a044c2fdc 100644
> --- a/arch/x86/kernel/head_64.S
> +++ b/arch/x86/kernel/head_64.S
> @@ -64,6 +64,10 @@ startup_64:
>          * tables and then reload them.
>          */
>
> +       /* Setup a stack for verify_cpu */
> +       movq    stack_start - __START_KERNEL_map, %rsp

This should be: movq stack_start(%rip), %rsp

> +       subq    $__START_KERNEL_map, %rsp

It would be better to add the offset to the initializer for
stack_start instead of adjusting it at runtime.  That would require
moving the existing load of stack_start from the common path to the
secondary startup, which probably isn't a bad thing as it wouldn't
depend on the trampoline stack anymore.

--
Brian Gerst