[PATCH 0/2] x86, mm: Fix PAT bit handling of large pages

[PATCH 0/2] x86, mm: Fix PAT bit handling of large pages

[Toshi Kani]

The PAT bit gets relocated to bit 12 when PUD and PMD mappings are
used.  This bit 12, however, is not covered by PTE_FLAGS_MASK, which
is corrently used for masking the flag bits for all cases. 

Patch 1/2 fixes pud_flags() and pmd_flags() to handle the PAT bit
when PUD and PMD mappings are used.

Patch 2/2 fixes /sys/kernel/debug/kernel_page_tables to show the
PAT bit properly.

Note, the PAT bit is first enabled in 4.2-rc1 with WT mappings.

---
Toshi Kani (2):
  1/2 x86: Fix pXd_flags() to handle _PAGE_PAT_LARGE
  2/2 x86, mm: Fix page table dump to show PAT bit

---
 arch/x86/include/asm/pgtable_types.h | 16 ++++++++++++---
 arch/x86/mm/dump_pagetables.c        | 39 +++++++++++++++++++-----------------
 2 files changed, 34 insertions(+), 21 deletions(-)

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

[PATCH 1/2] x86: Fix pXd_flags() to handle _PAGE_PAT_LARGE

[Toshi Kani]

The PAT bit gets relocated to bit 12 when PUD and PMD mappings are
used.  This bit 12, however, is not covered by PTE_FLAGS_MASK, which
is corrently used for masking the flag bits for all cases.

Fix pud_flags() and pmd_flags() to cover the PAT bit, _PAGE_PAT_LARGE,
when they are used to map a large page with _PAGE_PSE set.

Signed-off-by: Toshi Kani <toshi.kani@hp.com>
Cc: Juergen Gross <jgross@suse.com>
Cc: Konrad Wilk <konrad.wilk@oracle.com>
Cc: Robert Elliott <elliott@hp.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Andrew Morton <akpm@linux-foundation.org>
---
 arch/x86/include/asm/pgtable_types.h |   16 +++++++++++++---
 1 file changed, 13 insertions(+), 3 deletions(-)

diff --git a/arch/x86/include/asm/pgtable_types.h b/arch/x86/include/asm/pgtable_types.h
index 13f310b..caaf45c 100644
--- a/arch/x86/include/asm/pgtable_types.h
+++ b/arch/x86/include/asm/pgtable_types.h
@@ -212,9 +212,13 @@ enum page_cache_mode {
 /* PTE_PFN_MASK extracts the PFN from a (pte|pmd|pud|pgd)val_t */
 #define PTE_PFN_MASK		((pteval_t)PHYSICAL_PAGE_MASK)
 
-/* PTE_FLAGS_MASK extracts the flags from a (pte|pmd|pud|pgd)val_t */
+/* Extracts the flags from a (pte|pmd|pud|pgd)val_t of a 4KB page */
 #define PTE_FLAGS_MASK		(~PTE_PFN_MASK)
 
+/* Extracts the flags from a (pmd|pud)val_t of a (1GB|2MB) page */
+#define PMD_FLAGS_MASK_LARGE	((~PTE_PFN_MASK) | _PAGE_PAT_LARGE)
+#define PUD_FLAGS_MASK_LARGE	((~PTE_PFN_MASK) | _PAGE_PAT_LARGE)
+
 typedef struct pgprot { pgprotval_t pgprot; } pgprot_t;
 
 typedef struct { pgdval_t pgd; } pgd_t;
@@ -278,12 +282,18 @@ static inline pmdval_t native_pmd_val(pmd_t pmd)
 
 static inline pudval_t pud_flags(pud_t pud)
 {
-	return native_pud_val(pud) & PTE_FLAGS_MASK;
+	if (native_pud_val(pud) & _PAGE_PSE)
+		return native_pud_val(pud) & PUD_FLAGS_MASK_LARGE;
+	else
+		return native_pud_val(pud) & PTE_FLAGS_MASK;
 }
 
 static inline pmdval_t pmd_flags(pmd_t pmd)
 {
-	return native_pmd_val(pmd) & PTE_FLAGS_MASK;
+	if (native_pmd_val(pmd) & _PAGE_PSE)
+		return native_pmd_val(pmd) & PMD_FLAGS_MASK_LARGE;
+	else
+		return native_pmd_val(pmd) & PTE_FLAGS_MASK;
 }
 
 static inline pte_t native_make_pte(pteval_t val)

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

[PATCH 2/2] x86, mm: Fix page table dump to show PAT bit

[Toshi Kani]

/sys/kernel/debug/kernel_page_tables does not show the PAT bit
for PUD and PMD mappings.  This is because walk_pud_level(),
walk_pmd_level() and note_page() mask the flags with PTE_FLAGS_MASK,
which does not cover their PAT bit, _PAGE_PAT_LARGE.

Fix it by replacing the use of PTE_FLAGS_MASK with pXd_flags(),
which mask the flags properly.

Change also to show the PAT bit as "PAT" to be consistent with
other bits.

Reported-by: Robert Elliott <elliott@hp.com>
Signed-off-by: Toshi Kani <toshi.kani@hp.com>
Cc: Juergen Gross <jgross@suse.com>
Cc: Konrad Wilk <konrad.wilk@oracle.com>
Cc: Robert Elliott <elliott@hp.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Andrew Morton <akpm@linux-foundation.org>
---
 arch/x86/mm/dump_pagetables.c |   39 +++++++++++++++++++++------------------
 1 file changed, 21 insertions(+), 18 deletions(-)

diff --git a/arch/x86/mm/dump_pagetables.c b/arch/x86/mm/dump_pagetables.c
index f0cedf3..71ab2d7 100644
--- a/arch/x86/mm/dump_pagetables.c
+++ b/arch/x86/mm/dump_pagetables.c
@@ -155,7 +155,7 @@ static void printk_prot(struct seq_file *m, pgprot_t prot, int level, bool dmsg)
 			pt_dump_cont_printf(m, dmsg, "    ");
 		if ((level == 4 && pr & _PAGE_PAT) ||
 		    ((level == 3 || level == 2) && pr & _PAGE_PAT_LARGE))
-			pt_dump_cont_printf(m, dmsg, "pat ");
+			pt_dump_cont_printf(m, dmsg, "PAT ");
 		else
 			pt_dump_cont_printf(m, dmsg, "    ");
 		if (pr & _PAGE_GLOBAL)
@@ -198,8 +198,8 @@ static void note_page(struct seq_file *m, struct pg_state *st,
 	 * we have now. "break" is either changing perms, levels or
 	 * address space marker.
 	 */
-	prot = pgprot_val(new_prot) & PTE_FLAGS_MASK;
-	cur = pgprot_val(st->current_prot) & PTE_FLAGS_MASK;
+	prot = pgprot_val(new_prot);
+	cur = pgprot_val(st->current_prot);
 
 	if (!st->level) {
 		/* First entry */
@@ -269,13 +269,13 @@ static void walk_pte_level(struct seq_file *m, struct pg_state *st, pmd_t addr,
 {
 	int i;
 	pte_t *start;
+	pgprotval_t prot;
 
 	start = (pte_t *) pmd_page_vaddr(addr);
 	for (i = 0; i < PTRS_PER_PTE; i++) {
-		pgprot_t prot = pte_pgprot(*start);
-
+		prot = pte_flags(*start);
 		st->current_address = normalize_addr(P + i * PTE_LEVEL_MULT);
-		note_page(m, st, prot, 4);
+		note_page(m, st, __pgprot(prot), 4);
 		start++;
 	}
 }
@@ -287,18 +287,19 @@ static void walk_pmd_level(struct seq_file *m, struct pg_state *st, pud_t addr,
 {
 	int i;
 	pmd_t *start;
+	pgprotval_t prot;
 
 	start = (pmd_t *) pud_page_vaddr(addr);
 	for (i = 0; i < PTRS_PER_PMD; i++) {
 		st->current_address = normalize_addr(P + i * PMD_LEVEL_MULT);
 		if (!pmd_none(*start)) {
-			pgprotval_t prot = pmd_val(*start) & PTE_FLAGS_MASK;
-
-			if (pmd_large(*start) || !pmd_present(*start))
+			if (pmd_large(*start) || !pmd_present(*start)) {
+				prot = pmd_flags(*start);
 				note_page(m, st, __pgprot(prot), 3);
-			else
+			} else {
 				walk_pte_level(m, st, *start,
 					       P + i * PMD_LEVEL_MULT);
+			}
 		} else
 			note_page(m, st, __pgprot(0), 3);
 		start++;
@@ -318,19 +319,20 @@ static void walk_pud_level(struct seq_file *m, struct pg_state *st, pgd_t addr,
 {
 	int i;
 	pud_t *start;
+	pgprotval_t prot;
 
 	start = (pud_t *) pgd_page_vaddr(addr);
 
 	for (i = 0; i < PTRS_PER_PUD; i++) {
 		st->current_address = normalize_addr(P + i * PUD_LEVEL_MULT);
 		if (!pud_none(*start)) {
-			pgprotval_t prot = pud_val(*start) & PTE_FLAGS_MASK;
-
-			if (pud_large(*start) || !pud_present(*start))
+			if (pud_large(*start) || !pud_present(*start)) {
+				prot = pud_flags(*start);
 				note_page(m, st, __pgprot(prot), 2);
-			else
+			} else {
 				walk_pmd_level(m, st, *start,
 					       P + i * PUD_LEVEL_MULT);
+			}
 		} else
 			note_page(m, st, __pgprot(0), 2);
 
@@ -351,6 +353,7 @@ void ptdump_walk_pgd_level(struct seq_file *m, pgd_t *pgd)
 #else
 	pgd_t *start = swapper_pg_dir;
 #endif
+	pgprotval_t prot;
 	int i;
 	struct pg_state st = {};
 
@@ -362,13 +365,13 @@ void ptdump_walk_pgd_level(struct seq_file *m, pgd_t *pgd)
 	for (i = 0; i < PTRS_PER_PGD; i++) {
 		st.current_address = normalize_addr(i * PGD_LEVEL_MULT);
 		if (!pgd_none(*start)) {
-			pgprotval_t prot = pgd_val(*start) & PTE_FLAGS_MASK;
-
-			if (pgd_large(*start) || !pgd_present(*start))
+			if (pgd_large(*start) || !pgd_present(*start)) {
+				prot = pgd_flags(*start);
 				note_page(m, &st, __pgprot(prot), 1);
-			else
+			} else {
 				walk_pud_level(m, &st, *start,
 					       i * PGD_LEVEL_MULT);
+			}
 		} else
 			note_page(m, &st, __pgprot(0), 1);
 

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: [PATCH 1/2] x86: Fix pXd_flags() to handle _PAGE_PAT_LARGE

[Juergen Gross]

On 07/09/2015 07:03 PM, Toshi Kani wrote:
> The PAT bit gets relocated to bit 12 when PUD and PMD mappings are
> used.  This bit 12, however, is not covered by PTE_FLAGS_MASK, which
> is corrently used for masking the flag bits for all cases.
>
> Fix pud_flags() and pmd_flags() to cover the PAT bit, _PAGE_PAT_LARGE,
> when they are used to map a large page with _PAGE_PSE set.
>
> Signed-off-by: Toshi Kani <toshi.kani@hp.com>
> Cc: Juergen Gross <jgross@suse.com>
> Cc: Konrad Wilk <konrad.wilk@oracle.com>
> Cc: Robert Elliott <elliott@hp.com>
> Cc: Thomas Gleixner <tglx@linutronix.de>
> Cc: H. Peter Anvin <hpa@zytor.com>
> Cc: Ingo Molnar <mingo@redhat.com>
> Cc: Borislav Petkov <bp@alien8.de>
> Cc: Andrew Morton <akpm@linux-foundation.org>
> ---
>   arch/x86/include/asm/pgtable_types.h |   16 +++++++++++++---
>   1 file changed, 13 insertions(+), 3 deletions(-)
>
> diff --git a/arch/x86/include/asm/pgtable_types.h b/arch/x86/include/asm/pgtable_types.h
> index 13f310b..caaf45c 100644
> --- a/arch/x86/include/asm/pgtable_types.h
> +++ b/arch/x86/include/asm/pgtable_types.h
> @@ -212,9 +212,13 @@ enum page_cache_mode {
>   /* PTE_PFN_MASK extracts the PFN from a (pte|pmd|pud|pgd)val_t */
>   #define PTE_PFN_MASK		((pteval_t)PHYSICAL_PAGE_MASK)
>
> -/* PTE_FLAGS_MASK extracts the flags from a (pte|pmd|pud|pgd)val_t */
> +/* Extracts the flags from a (pte|pmd|pud|pgd)val_t of a 4KB page */
>   #define PTE_FLAGS_MASK		(~PTE_PFN_MASK)
>
> +/* Extracts the flags from a (pmd|pud)val_t of a (1GB|2MB) page */
> +#define PMD_FLAGS_MASK_LARGE	((~PTE_PFN_MASK) | _PAGE_PAT_LARGE)
> +#define PUD_FLAGS_MASK_LARGE	((~PTE_PFN_MASK) | _PAGE_PAT_LARGE)
> +
>   typedef struct pgprot { pgprotval_t pgprot; } pgprot_t;
>
>   typedef struct { pgdval_t pgd; } pgd_t;
> @@ -278,12 +282,18 @@ static inline pmdval_t native_pmd_val(pmd_t pmd)
>
>   static inline pudval_t pud_flags(pud_t pud)
>   {
> -	return native_pud_val(pud) & PTE_FLAGS_MASK;
> +	if (native_pud_val(pud) & _PAGE_PSE)
> +		return native_pud_val(pud) & PUD_FLAGS_MASK_LARGE;
> +	else
> +		return native_pud_val(pud) & PTE_FLAGS_MASK;
>   }
>
>   static inline pmdval_t pmd_flags(pmd_t pmd)
>   {
> -	return native_pmd_val(pmd) & PTE_FLAGS_MASK;
> +	if (native_pmd_val(pmd) & _PAGE_PSE)
> +		return native_pmd_val(pmd) & PMD_FLAGS_MASK_LARGE;
> +	else
> +		return native_pmd_val(pmd) & PTE_FLAGS_MASK;
>   }

Hmm, I think this covers only half of the problem. pud_pfn() and
pmd_pfn() will return wrong results for large pages with PAT bit
set as well.

I'd rather use something like:

static inline unsigned long pmd_pfn_mask(pmd_t pmd)
{
	if (pmd_large(pmd))
		return PMD_PAGE_MASK & PHYSICAL_PAGE_MASK;
	else
		return PTE_PFN_MASK;
}

static inline unsigned long pmd_flags_mask(pmd_t pmd)
{
	if (pmd_large(pmd))
		return ~(PMD_PAGE_MASK & PHYSICAL_PAGE_MASK);
	else
		return ~PTE_PFN_MASK;
}

static inline unsigned long pmd_pfn(pmd_t pmd)
{
         return (pmd_val(pmd) & pmd_pfn_mask(pmd)) >> PAGE_SHIFT;
}

static inline pmdval_t pmd_flags(pmd_t pmd)
{
	return native_pmd_val(pmd) & ~pmd_flags_mask(pmd);
}


Juergen

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: [PATCH 1/2] x86: Fix pXd_flags() to handle _PAGE_PAT_LARGE

[Toshi Kani]

On Fri, 2015-07-10 at 05:57 +0200, Juergen Gross wrote:
> On 07/09/2015 07:03 PM, Toshi Kani wrote:
> > The PAT bit gets relocated to bit 12 when PUD and PMD mappings are
> > used.  This bit 12, however, is not covered by PTE_FLAGS_MASK, 
> > which
> > is corrently used for masking the flag bits for all cases.
> > 
> > Fix pud_flags() and pmd_flags() to cover the PAT bit, 
> > _PAGE_PAT_LARGE,
> > when they are used to map a large page with _PAGE_PSE set.
  :
> Hmm, I think this covers only half of the problem. pud_pfn() and
> pmd_pfn() will return wrong results for large pages with PAT bit
> set as well.
> 
> I'd rather use something like:
> 
> static inline unsigned long pmd_pfn_mask(pmd_t pmd)
> {
> 	if (pmd_large(pmd))
> 		return PMD_PAGE_MASK & PHYSICAL_PAGE_MASK;
> 	else
> 		return PTE_PFN_MASK;
> }
> 
> static inline unsigned long pmd_flags_mask(pmd_t pmd)
> {
> 	if (pmd_large(pmd))
> 		return ~(PMD_PAGE_MASK & PHYSICAL_PAGE_MASK);
> 	else
> 		return ~PTE_PFN_MASK;
> }
> 
> static inline unsigned long pmd_pfn(pmd_t pmd)
> {
>          return (pmd_val(pmd) & pmd_pfn_mask(pmd)) >> PAGE_SHIFT;
> }
> 
> static inline pmdval_t pmd_flags(pmd_t pmd)
> {
> 	return native_pmd_val(pmd) & ~pmd_flags_mask(pmd);
> }

Thanks for the suggestion!  I agree that it is cleaner in this way.  I
am updating the patches and found the following changes are needed as
well:

 - Define PGTABLE_LEVELS to 2 in
"arch/x86/entry/vdso/vdso32/vclock_gettime.c".  This file redefines to
X86_32.  Setting to 2 levels (since X86_PAE is not set) allows <asm
-generic/pgtable-nopmd.h> be included to define PMD_SHIFT.

 - Move PUD_PAGE_SIZE & PUD_PAGE_MASK from <asm/page_64_types.h> to
<asm/page_types.h>.  This allows X86_32 to refer the PUD macros.

 - Nit: pmd_large() cannot be used in pmd_xxx_mask() since it calls
pmd_flags().  Use (native_pud_val(pud) & _PAGE_PSE), instead.

Thanks,
-Toshi

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>