Re: [RFC 08/17] x86/asm/64: De-Xen-ify our NMI code

[Juergen Gross]

This is a multi-part message in MIME format.On 06/09/17 23:36, Andy Lutomirski wrote:
> Xen PV is fundamentally incompatible with our fancy NMI code: it
> doesn't use IST at all, and Xen entries clobber two stack slots
> below the hardware frame.
> 
> Drop Xen PV support from our NMI code entirely.
> 
> XXX: Juergen: could you write and test the tiny patch needed to
> make Xen PV have a xen_nmi entry that handles NMIs?  I don't know
> how to test it.

You mean something like the attached one?

Seems to work at least for the "normal" case of a NMI coming in at
a random point in time.

Regarding testing: in case you have a Xen setup you can easily send
a NMI to a domain from dom0:

xl trigger <domain> nmi


Juergen

> 
> Cc: Juergen Gross <jgross@xxxxxxxx>
> Cc: Boris Ostrovsky <boris.ostrovsky@xxxxxxxxxx>
> Signed-off-by: Andy Lutomirski <luto@xxxxxxxxxx>
> ---
>  arch/x86/entry/entry_64.S | 41 +++++++++++++++++------------------------
>  1 file changed, 17 insertions(+), 24 deletions(-)
> 
> diff --git a/arch/x86/entry/entry_64.S b/arch/x86/entry/entry_64.S
> index a9e318f7cc9b..c81e05fb999e 100644
> --- a/arch/x86/entry/entry_64.S
> +++ b/arch/x86/entry/entry_64.S
> @@ -1171,21 +1171,12 @@ ENTRY(error_exit)
>  	jmp	retint_user
>  END(error_exit)
>  
> -/* Runs on exception stack */
> +/*
> + * Runs on exception stack.  Xen PV does not go through this path at all,
> + * so we can use real assembly here.
> + */
>  ENTRY(nmi)
>  	/*
> -	 * Fix up the exception frame if we're on Xen.
> -	 * PARAVIRT_ADJUST_EXCEPTION_FRAME is guaranteed to push at most
> -	 * one value to the stack on native, so it may clobber the rdx
> -	 * scratch slot, but it won't clobber any of the important
> -	 * slots past it.
> -	 *
> -	 * Xen is a different story, because the Xen frame itself overlaps
> -	 * the "NMI executing" variable.
> -	 */
> -	PARAVIRT_ADJUST_EXCEPTION_FRAME
> -
> -	/*
>  	 * We allow breakpoints in NMIs. If a breakpoint occurs, then
>  	 * the iretq it performs will take us out of NMI context.
>  	 * This means that we can have nested NMIs where the next
> @@ -1240,7 +1231,7 @@ ENTRY(nmi)
>  	 * stacks lest we corrupt the "NMI executing" variable.
>  	 */
>  
> -	SWAPGS_UNSAFE_STACK
> +	swapgs
>  	cld
>  	movq	%rsp, %rdx
>  	movq	PER_CPU_VAR(cpu_current_top_of_stack), %rsp
> @@ -1402,7 +1393,7 @@ nested_nmi_out:
>  	popq	%rdx
>  
>  	/* We are returning to kernel mode, so this cannot result in a fault. */
> -	INTERRUPT_RETURN
> +	iretq
>  
>  first_nmi:
>  	/* Restore rdx. */
> @@ -1432,7 +1423,7 @@ first_nmi:
>  	pushfq			/* RFLAGS */
>  	pushq	$__KERNEL_CS	/* CS */
>  	pushq	$1f		/* RIP */
> -	INTERRUPT_RETURN	/* continues at repeat_nmi below */
> +	iretq			/* continues at repeat_nmi below */
>  1:
>  #endif
>  
> @@ -1502,20 +1493,22 @@ nmi_restore:
>  	/*
>  	 * Clear "NMI executing".  Set DF first so that we can easily
>  	 * distinguish the remaining code between here and IRET from
> -	 * the SYSCALL entry and exit paths.  On a native kernel, we
> -	 * could just inspect RIP, but, on paravirt kernels,
> -	 * INTERRUPT_RETURN can translate into a jump into a
> -	 * hypercall page.
> +	 * the SYSCALL entry and exit paths.
> +	 *
> +	 * We arguably should just inspect RIP instead, but I (Andy) wrote
> +	 * this code when I had the misapprehension that Xen PV supported
> +	 * NMIs, and Xen PV would break that approach.
>  	 */
>  	std
>  	movq	$0, 5*8(%rsp)		/* clear "NMI executing" */
>  
>  	/*
> -	 * INTERRUPT_RETURN reads the "iret" frame and exits the NMI
> -	 * stack in a single instruction.  We are returning to kernel
> -	 * mode, so this cannot result in a fault.
> +	 * iretq reads the "iret" frame and exits the NMI stack in a
> +	 * single instruction.  We are returning to kernel mode, so this
> +	 * cannot result in a fault.  Similarly, we don't need to worry
> +	 * about espfix64 on the way back to kernel mode.
>  	 */
> -	INTERRUPT_RETURN
> +	iretq
>  END(nmi)
>  
>  ENTRY(ignore_sysret)
>