Re: [PATCH 19/34] x86, pkeys: optimize fault handling in access_error()

From: Thomas Gleixner <tglx@linutronix.de>
To: Dave Hansen <dave@sr71.net>
Cc: linux-kernel@vger.kernel.org, linux-mm@kvack.org, x86@kernel.org,
	dave.hansen@linux.intel.com
Subject: Re: [PATCH 19/34] x86, pkeys: optimize fault handling in access_error()
Date: Tue, 8 Dec 2015 19:14:54 +0100 (CET)	[thread overview]
Message-ID: <alpine.DEB.2.11.1512081913570.3595@nanos> (raw)
In-Reply-To: <20151204011450.A07593D5@viggo.jf.intel.com>

On Thu, 3 Dec 2015, Dave Hansen wrote:
> diff -puN arch/x86/mm/fault.c~pkeys-15-access_error arch/x86/mm/fault.c
> --- a/arch/x86/mm/fault.c~pkeys-15-access_error	2015-12-03 16:21:26.872727820 -0800
> +++ b/arch/x86/mm/fault.c	2015-12-03 16:21:26.876728002 -0800
> @@ -900,10 +900,16 @@ bad_area(struct pt_regs *regs, unsigned
>  static inline bool bad_area_access_from_pkeys(unsigned long error_code,
>  		struct vm_area_struct *vma)
>  {
> +	/* This code is always called on the current mm */
> +	int foreign = 0;

arch_vma_access_permitted takes a bool ....

>  	if (!boot_cpu_has(X86_FEATURE_OSPKE))
>  		return false;
>  	if (error_code & PF_PK)
>  		return true;
> +	/* this checks permission keys on the VMA: */
> +	if (!arch_vma_access_permitted(vma, (error_code & PF_WRITE), foreign))
> +		return true;
>  	return false;
>  }
>  
> @@ -1091,6 +1097,8 @@ int show_unhandled_signals = 1;
>  static inline int
>  access_error(unsigned long error_code, struct vm_area_struct *vma)
>  {
> +	/* This is only called for the current mm, so: */
> +	int foreign = 0;

Ditto.

Other than that: Reviewed-by: Thomas Gleixner <tglx@linutronix.de>