On Wed, May 01, 2013 at 12:42:38PM +0100, Will Deacon wrote:

Hi Steve,

On Tue, Apr 30, 2013 at 05:30:46PM +0100, Steve Capper wrote:

quoted

Add huge page support to ARM64, different huge page sizes are
supported depending on the size of normal pages:

PAGE_SIZE is 4K:
   2MB - (pmds) these can be allocated at any time.
1024MB - (puds) usually allocated on bootup with the command line
         with something like: hugepagesz=1G hugepages=6

PAGE_SIZE is 64K:
 512MB - (pmds), usually allocated on bootup via command line.

[...]

quoted

diff --git a/arch/arm64/include/asm/pgtable-hwdef.h b/arch/arm64/include/asm/pgtable-hwdef.h
index 75fd13d..c3cac68 100644
--- a/arch/arm64/include/asm/pgtable-hwdef.h
+++ b/arch/arm64/include/asm/pgtable-hwdef.h

@@ -53,6 +53,7 @@
 #define PTE_TYPE_MASK		(_AT(pteval_t, 3) << 0)
 #define PTE_TYPE_FAULT		(_AT(pteval_t, 0) << 0)
 #define PTE_TYPE_PAGE		(_AT(pteval_t, 3) << 0)
+#define PTE_TYPE_HUGEPAGE	(_AT(pmdval_t, 1) << 0)

This breaks PROT_NONE mappings, where you get:

	pte = pte_mkhuge(pte_modify(pte, newprot));

The pte_modify will clear the valid bit and set the prot_none bit (in order
to create a present, faulting entry) but then your pte_mkhuge will come in
and clobber that with a valid block entry.

Thanks Will, I'll re-work the pte_mkhuge/pte_huge logic and get a PROT_NONE
test case coded up.

Cheers,
-- 
Steve

Will