On Thu, Mar 07, 2019 at 12:45:16PM +0100, Peter Zijlstra wrote:
> drivers/xen/privcmd.o: warning: objtool: privcmd_ioctl()+0x1414: call to
> hypercall_page() with UACCESS enabled
>
> Xen needs to do HV calls with AC=1 for hysterical raisins. Make the
> warning go away.
>
> XXX: arguably we should rename ANNOTATE_NOSPEC_ALTERNATIVE to
> ANNOTATE_IGNORE_ALTERNATIVE.
>
> Cc: andrew.coop...@citrix.com
> Signed-off-by: Peter Zijlstra (Intel) <pet...@infradead.org>
> ---
> arch/x86/include/asm/xen/hypercall.h | 26 ++++++++++++++++++++++----
> 1 file changed, 22 insertions(+), 4 deletions(-)
>
> --- a/arch/x86/include/asm/xen/hypercall.h
> +++ b/arch/x86/include/asm/xen/hypercall.h
> @@ -214,6 +214,24 @@ xen_single_call(unsigned int call,
> return (long)__res;
> }
>
> +static __always_inline void __xen_stac(void)
> +{
> + /*
> + * This is just about as horrible as this interface; we abuse the
> + * nospec alternative annotation to supress objtool seeing the
> + * STAC/CLAC and getting confused about it calling random code with
> + * AC=1.
> + */
> + asm volatile(ANNOTATE_NOSPEC_ALTERNATIVE
> + ASM_STAC ::: "memory", "flags");
> +}
> +
> +static __always_inline void __xen_clac(void)
> +{
> + asm volatile(ANNOTATE_NOSPEC_ALTERNATIVE
> + ASM_CLAC ::: "memory", "flags");
> +}
> +
Shouldn't these be using SMAP-based alternatives like stac()/clac() do?
--
Josh
