Re: [PATCH 20/25] arm64:ilp32: add sys_ilp32.c and a separate table (in entry.S) to use it

[Catalin Marinas]

On Wed, Apr 06, 2016 at 01:08:42AM +0300, Yury Norov wrote:
> --- a/arch/arm64/kernel/entry.S
> +++ b/arch/arm64/kernel/entry.S
> @@ -715,9 +715,13 @@ ENDPROC(ret_from_fork)
>   */
>  	.align	6
>  el0_svc:
> -	adrp	stbl, sys_call_table		// load syscall table pointer
>  	uxtw	scno, w8			// syscall number in w8
>  	mov	sc_nr, #__NR_syscalls
> +#ifdef CONFIG_ARM64_ILP32
> +	ldr	x16, [tsk, #TI_FLAGS]
> +	tbnz	x16, #TIF_32BIT_AARCH64, el0_ilp32_svc // We are using ILP32
> +#endif

There is another ldr x16, [tsk, #TI_FLAGS] load further down in the
el0_svc_naked block. We should rework these a bit to avoid loading the
same location twice unnecessarily. E.g. move the ldr x16 just before
el0_svc_naked and branch one line after in case of the ILP32 syscall.

> +	adrp	stbl, sys_call_table		// load syscall table pointer
>  el0_svc_naked:					// compat entry point
>  	stp	x0, scno, [sp, #S_ORIG_X0]	// save the original x0 and syscall number
>  	enable_dbg_and_irq
> @@ -737,6 +741,12 @@ ni_sys:
>  	b	ret_fast_syscall
>  ENDPROC(el0_svc)
>  
> +#ifdef CONFIG_ARM64_ILP32
> +el0_ilp32_svc:
> +	adrp	stbl, sys_call_ilp32_table // load syscall table pointer
> +	b el0_svc_naked
> +#endif
> +
>  	/*
>  	 * This is the really slow path.  We're going to be doing context
>  	 * switches, and waiting for our parent to respond.

-- 
Catalin