Re: [PATCH] x86: Stop relying on magic jmp behavior for early_idt_handlers

["H. Peter Anvin" <hpa@zytor.com>]

On 05/21/2015 05:17 PM, Andy Lutomirski wrote:
> 
> diff --git a/arch/x86/include/asm/segment.h b/arch/x86/include/asm/segment.h
> index 5a9856eb12ba..4bbe0eb24d7e 100644
> --- a/arch/x86/include/asm/segment.h
> +++ b/arch/x86/include/asm/segment.h
> @@ -231,12 +231,21 @@
>  #define TLS_SIZE			(GDT_ENTRY_TLS_ENTRIES* 8)
>  
>  #ifdef __KERNEL__
> -#ifndef __ASSEMBLY__
>  
> -extern const char early_idt_handlers[NUM_EXCEPTION_VECTORS][2+2+5];
> +/*
> + * early_idt_handlers is an array of entry points.  For simplicity, it's
> + * a real array.  We allocate nine bytes for each entry: two one-byte
> + * push instructions and a five-byte jump in the worst case.
> + */
> +#define EARLY_IDT_HANDLER_STRIDE 9
> +#ifndef __ASSEMBLY__
> +extern const char early_idt_handlers[NUM_EXCEPTION_VECTORS][EARLY_IDT_HANDLER_STRIDE];
>  #ifdef CONFIG_TRACING
>  # define trace_early_idt_handlers early_idt_handlers
>  #endif
> +#endif
> +
> +#ifndef __ASSEMBLY__
>  

Please get rid of the unnecessary #ifndef/#endif pair.

> +/* Build the early_idt_handlers array */
>  	__INIT
>  	.globl early_idt_handlers

Let's use ENTRY(early_idt_handlers) for consistency.

Otherwise it looks good.  I like the use of .fill better than moving .,
it seems more robust to me.

Conditionally-Acked-by: H. Peter Anvin <hpa@linux.intel.com>