[PATCH] ARM64: Kernel managed pages are only flushed

[Catalin Marinas]

On Wed, Mar 12, 2014 at 04:03:40PM +0000, Catalin Marinas wrote:
> On Wed, Mar 12, 2014 at 02:56:53PM +0000, Will Deacon wrote:
> > On Wed, Mar 12, 2014 at 02:41:31PM +0000, Bharat.Bhushan at freescale.com wrote:
> > > Did you get the chance to look into this? What is your take for this patch
> > > or you want to suggest some other solution?
> > 
> > See my reply to Laura here:
> > 
> >   http://lists.infradead.org/pipermail/linux-arm-kernel/2014-March/238510.html
> > 
> > We *really* don't want executable device mappings.
> 
> /dev/mem mapping use pgprot_noncached() but I think we could generalise
> any of the writecombine and dmacoherent mappings to XN like in the patch
> below:
> 
> diff --git a/arch/arm64/include/asm/pgtable.h b/arch/arm64/include/asm/pgtable.h
> index b524dcd17243..2d3cede62709 100644
> --- a/arch/arm64/include/asm/pgtable.h
> +++ b/arch/arm64/include/asm/pgtable.h
> @@ -286,11 +286,11 @@ static inline int has_transparent_hugepage(void)
>   * Mark the prot value as uncacheable and unbufferable.
>   */
>  #define pgprot_noncached(prot) \
> -	__pgprot_modify(prot, PTE_ATTRINDX_MASK, PTE_ATTRINDX(MT_DEVICE_nGnRnE))
> +	__pgprot_modify(prot, PTE_ATTRINDX_MASK, PTE_ATTRINDX(MT_DEVICE_nGnRnE) | PTE_PXN | PTE_UXN)
>  #define pgprot_writecombine(prot) \
> -	__pgprot_modify(prot, PTE_ATTRINDX_MASK, PTE_ATTRINDX(MT_NORMAL_NC))
> +	__pgprot_modify(prot, PTE_ATTRINDX_MASK, PTE_ATTRINDX(MT_NORMAL_NC) | PTE_PXN | PTE_UXN)
>  #define pgprot_dmacoherent(prot) \
> -	__pgprot_modify(prot, PTE_ATTRINDX_MASK, PTE_ATTRINDX(MT_NORMAL_NC))
> +	__pgprot_modify(prot, PTE_ATTRINDX_MASK, PTE_ATTRINDX(MT_NORMAL_NC) | PTE_PXN | PTE_UXN)
>  #define __HAVE_PHYS_MEM_ACCESS_PROT
>  struct file;
>  extern pgprot_t phys_mem_access_prot(struct file *file, unsigned long pfn,

And one more patch as suggested by Steve Capper. Both would be needed.

-------------8<----------------------

>From 31d84855d71778e4a0f615f61ab836be3a70a58b Mon Sep 17 00:00:00 2001
From: Catalin Marinas <catalin.marinas at arm.com>
Date: Wed, 12 Mar 2014 16:28:09 +0000
Subject: [PATCH] arm64: Do not synchronise I and D caches for special ptes

Special pte mappings are not intended to be executable and do not even
have an associated struct page. This patch ensures that we do not call
__sync_icache_dcache() on such ptes.

Signed-off-by: Catalin Marinas <catalin.marinas at arm.com>
Reported-by: Steve Capper <Steve.Capper at arm.com>
---
 arch/arm64/include/asm/pgtable.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/arm64/include/asm/pgtable.h b/arch/arm64/include/asm/pgtable.h
index 2d3cede62709..72c9ac38cdd9 100644
--- a/arch/arm64/include/asm/pgtable.h
+++ b/arch/arm64/include/asm/pgtable.h
@@ -199,7 +199,7 @@ static inline void set_pte_at(struct mm_struct *mm, unsigned long addr,
 			      pte_t *ptep, pte_t pte)
 {
 	if (pte_valid_user(pte)) {
-		if (pte_exec(pte))
+		if (!pte_special(pte) && pte_exec(pte))
 			__sync_icache_dcache(pte, addr);
 		if (pte_dirty(pte) && pte_write(pte))
 			pte_val(pte) &= ~PTE_RDONLY;