Skip to contentSkip to conversation

mm/THP: Don't use HPAGE_SHIFT in transparent hugepage code

PATCH In review Aneesh Kumar K.V 2 participants 4 messages May 6, 2013 – May 7, 2013

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 mm/THP: Don't use HPAGE_SHIFT in transparent hugepage code

From: "Aneesh Kumar K.V" <[email protected]>

For architectures like powerpc that support multiple explicit hugepage sizes, HPAGE_SHIFT indicate the default explicit hugepage shift. For THP to work the hugepage size should be same as PMD_SIZE. So use PMD_SHIFT directly. So move the define outside CONFIG_TRANSPARENT_HUGEPAGE #ifdef because we want to use these defines in generic code with if (pmd_trans_huge()) conditional.

---
 include/linux/huge_mm.h | 10 +++-------
 1 file changed, 3 insertions(+), 7 deletions(-)
include/linux/huge_mm.h +3 -7
index 528454c..cc276d2 100644
@@ -58,12 +58,11 @@ extern pmd_t *page_check_address_pmd(struct page *page,
 
 #define HPAGE_PMD_ORDER (HPAGE_PMD_SHIFT-PAGE_SHIFT)
 #define HPAGE_PMD_NR (1<<HPAGE_PMD_ORDER)
+#define HPAGE_PMD_SHIFT PMD_SHIFT
+#define HPAGE_PMD_SIZE	((1UL) << HPAGE_PMD_SHIFT)
+#define HPAGE_PMD_MASK	(~(HPAGE_PMD_SIZE - 1))
 
 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
-#define HPAGE_PMD_SHIFT HPAGE_SHIFT
-#define HPAGE_PMD_MASK HPAGE_MASK
-#define HPAGE_PMD_SIZE HPAGE_SIZE
-
 extern bool is_vma_temporary_stack(struct vm_area_struct *vma);
 
 #define transparent_hugepage_enabled(__vma)				\
@@ -181,9 +180,6 @@ extern int do_huge_pmd_numa_page(struct mm_struct *mm, struct vm_area_struct *vm
 				unsigned long addr, pmd_t pmd, pmd_t *pmdp);
 
 #else /* CONFIG_TRANSPARENT_HUGEPAGE */
-#define HPAGE_PMD_SHIFT ({ BUILD_BUG(); 0; })
-#define HPAGE_PMD_MASK ({ BUILD_BUG(); 0; })
-#define HPAGE_PMD_SIZE ({ BUILD_BUG(); 0; })
 
 #define hpage_nr_pages(x) 1
 
Signature · 7 lines
1.8.1.2

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#1 lore raw
Quoted 8 lines · Aneesh Kumar K.V

From: "Aneesh Kumar K.V" <[email protected]>

For architectures like powerpc that support multiple explicit hugepage sizes, HPAGE_SHIFT indicate the default explicit hugepage shift. For THP to work the hugepage size should be same as PMD_SIZE. So use PMD_SHIFT directly. So move the define outside CONFIG_TRANSPARENT_HUGEPAGE #ifdef because we want to use these defines in generic code with if (pmd_trans_huge()) conditional.

Sorry, I haven't got why you move it outside #ifdef.
If CONFIG_TRANSPARENT_HUGEPAGE disabled pmd_trans_huge() will be 0 in compile time, so BUILD_BUG() will be optimize out by GCC.

The BUILD_BUGs are useful. It's bug if you *really* use the defines with THP disabled.

Signature · 7 lines
 Kirill A. Shutemov

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#2 lore raw
In reply to Aneesh Kumar K.V
Quoted 19 lines · Aneesh Kumar K.V

From: "Aneesh Kumar K.V" <[email protected]>

For architectures like powerpc that support multiple explicit hugepage sizes, HPAGE_SHIFT indicate the default explicit hugepage shift. For THP to work the hugepage size should be same as PMD_SIZE. So use PMD_SHIFT directly. So move the define outside CONFIG_TRANSPARENT_HUGEPAGE #ifdef because we want to use these defines in generic code with if (pmd_trans_huge()) conditional.

---
 include/linux/huge_mm.h | 10 +++-------
 1 file changed, 3 insertions(+), 7 deletions(-)
include/linux/huge_mm.h +0 -0
index 528454c..cc276d2 100644

include/linux/huge_mm.h extern pmd_t *page_check_address_pmd(struct page *page, · line 61

Quoted 4 lines
include/linux/huge_mm.h +1 -0
@@ -58,12 +58,11 @@ extern pmd_t *page_check_address_pmd(struct page *page,
 
 #define HPAGE_PMD_ORDER (HPAGE_PMD_SHIFT-PAGE_SHIFT)
 #define HPAGE_PMD_NR (1<<HPAGE_PMD_ORDER)
+#define HPAGE_PMD_SHIFT PMD_SHIFT

What about:

#ifndef HPAGE_PMD_SHIFT
#define HPAGE_PMD_SHIFT HPAGE_SHIFT
#endif

And define HPAGE_PMD_SHIFT in arch code if HPAGE_SHIFT is not suitable?

Quoted 29 lines
include/linux/huge_mm.h +2 -7
+#define HPAGE_PMD_SIZE	((1UL) << HPAGE_PMD_SHIFT)
+#define HPAGE_PMD_MASK	(~(HPAGE_PMD_SIZE - 1))
 
 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
-#define HPAGE_PMD_SHIFT HPAGE_SHIFT
-#define HPAGE_PMD_MASK HPAGE_MASK
-#define HPAGE_PMD_SIZE HPAGE_SIZE
-
 extern bool is_vma_temporary_stack(struct vm_area_struct *vma);
 
 #define transparent_hugepage_enabled(__vma)				\
@@ -181,9 +180,6 @@ extern int do_huge_pmd_numa_page(struct mm_struct *mm, struct vm_area_struct *vm
 				unsigned long addr, pmd_t pmd, pmd_t *pmdp);
 
 #else /* CONFIG_TRANSPARENT_HUGEPAGE */
-#define HPAGE_PMD_SHIFT ({ BUILD_BUG(); 0; })
-#define HPAGE_PMD_MASK ({ BUILD_BUG(); 0; })
-#define HPAGE_PMD_SIZE ({ BUILD_BUG(); 0; })
 
 #define hpage_nr_pages(x) 1
 
Signature · 7 lines
1.8.1.2

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
Signature · 7 lines
 Kirill A. Shutemov

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#3 lore raw
Quoted 27 lines · Kirill A. Shutemov
Quoted 19 lines · Aneesh Kumar K.V

From: "Aneesh Kumar K.V" <[email protected]>

For architectures like powerpc that support multiple explicit hugepage sizes, HPAGE_SHIFT indicate the default explicit hugepage shift. For THP to work the hugepage size should be same as PMD_SIZE. So use PMD_SHIFT directly. So move the define outside CONFIG_TRANSPARENT_HUGEPAGE #ifdef because we want to use these defines in generic code with if (pmd_trans_huge()) conditional.

---
 include/linux/huge_mm.h | 10 +++-------
 1 file changed, 3 insertions(+), 7 deletions(-)
include/linux/huge_mm.h +0 -0
index 528454c..cc276d2 100644

include/linux/huge_mm.h extern pmd_t *page_check_address_pmd(struct page *page, · line 61

Quoted 4 lines
include/linux/huge_mm.h +1 -0
@@ -58,12 +58,11 @@ extern pmd_t *page_check_address_pmd(struct page *page,
 
 #define HPAGE_PMD_ORDER (HPAGE_PMD_SHIFT-PAGE_SHIFT)
 #define HPAGE_PMD_NR (1<<HPAGE_PMD_ORDER)
+#define HPAGE_PMD_SHIFT PMD_SHIFT

What about:

Quoted 6 lines

#ifndef HPAGE_PMD_SHIFT
#define HPAGE_PMD_SHIFT HPAGE_SHIFT
#endif

And define HPAGE_PMD_SHIFT in arch code if HPAGE_SHIFT is not suitable?

That would work for me provided the BUILD_BUG_ON is also taken care. But is there a reason why we want to do that ? Will any value other than PMD_SHIFT work ?

The below patch shows how we want to use these. To avoid those BUILD_BUG_ON I ended up doing HUGE_PAGE_SIZE and HUGE_PAGE_MASK

https://lists.ozlabs.org/pipermail/linuxppc-dev/2013-April/105631.html

-aneesh

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