Re: [PATCH v4 RESEND 4/11] x86/asm: Fix pud/pmd interfaces to handle large PAT bit

From: Boris Ostrovsky
Date: Mon Nov 09 2015 - 17:37:46 EST

On 11/09/2015 03:47 PM, Kirill A. Shutemov wrote:
On Mon, Nov 09, 2015 at 02:39:31PM -0500, Boris Ostrovsky wrote:
On 11/09/2015 02:16 PM, Toshi Kani wrote:
On Mon, 2015-11-09 at 13:06 -0500, Boris Ostrovsky wrote:
On 09/17/2015 02:24 PM, Toshi Kani wrote:
Now that we have pud/pmd mask interfaces, which handle pfn & flags
mask properly for the large PAT bit.

Fix pud/pmd pfn & flags interfaces by replacing PTE_PFN_MASK and
PTE_FLAGS_MASK with the pud/pmd mask interfaces.

Suggested-by: Juergen Gross <jgross@xxxxxxxx>
Signed-off-by: Toshi Kani <toshi.kani@xxxxxxx>
Cc: Juergen Gross <jgross@xxxxxxxx>
Cc: Konrad Wilk <konrad.wilk@xxxxxxxxxx>
Cc: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
Cc: H. Peter Anvin <hpa@xxxxxxxxx>
Cc: Ingo Molnar <mingo@xxxxxxxxxx>
Cc: Borislav Petkov <bp@xxxxxxxxx>
arch/x86/include/asm/pgtable.h | 14 ++++++++------
arch/x86/include/asm/pgtable_types.h | 4 ++--
2 files changed, 10 insertions(+), 8 deletions(-)

Looks like this commit is causing this splat for 32-bit kernels. I am
attaching my config file, just in case.
Thanks for the report! I'd like to reproduce the issue since I am not sure how
this change caused it...

I tried to build a kernel with the attached config file, and got the following
error. Not sure what I am missing.

$ make -j24 ARCH=i386
LD drivers/built-in.o
LINK vmlinux
./.config: line 44: $'\r': command not found
I wonder whether my email client added ^Ms to the file that I send. It
shouldn't have.

Makefile:929: recipe for target 'vmlinux' failed
make: *** [vmlinux] Error 127

Do you have steps to reproduce the issue? Or do you see it during boot-time?
This always happens just after system has booted, it may still be going over
init scripts. I am booting with ramdisk, don't know whether it has anything
to do with this problem.

FWIW, it looks like pmd_pfn_mask() inline is causing this. Reverting it
alone makes this crash go away.
Could you check the patch below?

I does fix the problem on baremetal, thanks. My 32-bit Xen guests still fail which I thought was the same issue but now that I looked at it more carefully it has different signature.


diff --git a/arch/x86/include/asm/pgtable_types.h b/arch/x86/include/asm/pgtable_types.h
index dd5b0aa9dd2f..c1e797266ce9 100644
--- a/arch/x86/include/asm/pgtable_types.h
+++ b/arch/x86/include/asm/pgtable_types.h
@@ -279,17 +279,14 @@ static inline pmdval_t native_pmd_val(pmd_t pmd)
static inline pudval_t pud_pfn_mask(pud_t pud)
if (native_pud_val(pud) & _PAGE_PSE)
+ return ~((1ULL << PUD_SHIFT) - 1) & PHYSICAL_PAGE_MASK;
return PTE_PFN_MASK;
static inline pudval_t pud_flags_mask(pud_t pud)
- if (native_pud_val(pud) & _PAGE_PSE)
- return ~(PUD_PAGE_MASK & (pudval_t)PHYSICAL_PAGE_MASK);
- else
- return ~PTE_PFN_MASK;
+ return ~pud_pfn_mask(pud);
static inline pudval_t pud_flags(pud_t pud)
@@ -300,17 +297,14 @@ static inline pudval_t pud_flags(pud_t pud)
static inline pmdval_t pmd_pfn_mask(pmd_t pmd)
if (native_pmd_val(pmd) & _PAGE_PSE)
+ return ~((1ULL << PMD_SHIFT) - 1) & PHYSICAL_PAGE_MASK;
return PTE_PFN_MASK;
static inline pmdval_t pmd_flags_mask(pmd_t pmd)
- if (native_pmd_val(pmd) & _PAGE_PSE)
- return ~(PMD_PAGE_MASK & (pmdval_t)PHYSICAL_PAGE_MASK);
- else
- return ~PTE_PFN_MASK;
+ return ~pmd_pfn_mask(pmd);
static inline pmdval_t pmd_flags(pmd_t pmd)

To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at
Please read the FAQ at