Re: [patch] x86: Fix /dev/mem mmap breakage when PAT is disabled

[Ingo Molnar <mingo@elte.hu>]

* Ravikiran G Thirumalai <kiran@scalex86.org> wrote:

> Fix mmap to /dev/mem when CONFIG_X86_PAT is off and CONFIG_STRICT_DEVMEM is
> off
> 
> mmap to /dev/mem on kernel memory has been failing since the
> introduction of PAT (CONFIG_STRICT_DEVMEM=n case).   Seems like
> the check to avoid cache aliasing with PAT is kicking in even
> when PAT is disabled. The bug seems to have crept in 2.6.26.
> 
> This patch makes sure that mmap to regular kernel memory
> succeeds if CONFIG_STRICT_DEVMEM=n and
> PAT is disabled.  The checks to avoid cache aliasing
> still happens if PAT is enabled.
> 
> Signed-off-by: Ravikiran Thirumalai <kiran@scalex86.org>
> Tested-by: Tim Sirianni <tim@scalemp.com>
> 
> Index: git.tip/arch/x86/mm/pat.c
> ===================================================================
> --- git.tip.orig/arch/x86/mm/pat.c	2008-10-21 14:02:57.000000000 -0700
> +++ git.tip/arch/x86/mm/pat.c	2008-10-29 13:19:47.000000000 -0800
> @@ -481,6 +481,7 @@
>  	return 1;
>  }
>  #else
> +/* This check is needed to avoid cache aliasing when PAT is enabled */
>  static inline int range_is_allowed(unsigned long pfn, unsigned long size)
>  {
>  	u64 from = ((u64)pfn) << PAGE_SHIFT;
> @@ -508,7 +509,8 @@
>  	unsigned long flags = -1;
>  	int retval;
>  
> -	if (!range_is_allowed(pfn, size))
> +	/* Ensure cache aliasing does not occur if and only if PAT is on */
> +	if (pat_enabled && !range_is_allowed(pfn, size))
>  		return 0;

I guess it might be cleaner to push this check into the 
range_is_allowed() function?

	Ingo