* [PATCH 0/2] mm, smaps: MMUPageSize for device-dax
@ 2017-12-08 3:30 ` Dan Williams
0 siblings, 0 replies; 10+ messages in thread
From: Dan Williams @ 2017-12-08 3:30 UTC (permalink / raw)
To: akpm
Cc: Jane Chu, linux-nvdimm, Michael Ellerman, linux-mm,
Paul Mackerras, Benjamin Herrenschmidt
Andrew,
Here is another occasion where we want special-case hugetlbfs enabling
to also apply to device-dax. I started to wonder what other hstate
conversions we might do beyond ->split() and ->pagesize(), but this
appears to be the last of the usages of hstate_vma() in
generic/non-hugetlbfs specific code paths.
This is 4.16 material.
---
Dan Williams (2):
mm, hugetlbfs: introduce ->pagesize() to vm_operations_struct
device-dax: implement ->pagesize() for smaps to report MMUPageSize
arch/powerpc/mm/hugetlbpage.c | 5 +----
drivers/dax/device.c | 10 ++++++++++
include/linux/hugetlb.h | 30 ++++++++++++++++++++++++------
include/linux/mm.h | 1 +
mm/hugetlb.c | 38 ++++++++------------------------------
5 files changed, 44 insertions(+), 40 deletions(-)
_______________________________________________
Linux-nvdimm mailing list
Linux-nvdimm@lists.01.org
https://lists.01.org/mailman/listinfo/linux-nvdimm
^ permalink raw reply [flat|nested] 10+ messages in thread
* [PATCH 0/2] mm, smaps: MMUPageSize for device-dax
@ 2017-12-08 3:30 ` Dan Williams
0 siblings, 0 replies; 10+ messages in thread
From: Dan Williams @ 2017-12-08 3:30 UTC (permalink / raw)
To: akpm
Cc: Jane Chu, linux-nvdimm, Michael Ellerman, linux-mm,
Paul Mackerras, Benjamin Herrenschmidt
Andrew,
Here is another occasion where we want special-case hugetlbfs enabling
to also apply to device-dax. I started to wonder what other hstate
conversions we might do beyond ->split() and ->pagesize(), but this
appears to be the last of the usages of hstate_vma() in
generic/non-hugetlbfs specific code paths.
This is 4.16 material.
---
Dan Williams (2):
mm, hugetlbfs: introduce ->pagesize() to vm_operations_struct
device-dax: implement ->pagesize() for smaps to report MMUPageSize
arch/powerpc/mm/hugetlbpage.c | 5 +----
drivers/dax/device.c | 10 ++++++++++
include/linux/hugetlb.h | 30 ++++++++++++++++++++++++------
include/linux/mm.h | 1 +
mm/hugetlb.c | 38 ++++++++------------------------------
5 files changed, 44 insertions(+), 40 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>
^ permalink raw reply [flat|nested] 10+ messages in thread
* [PATCH 1/2] mm, hugetlbfs: introduce ->pagesize() to vm_operations_struct
2017-12-08 3:30 ` Dan Williams
@ 2017-12-08 3:30 ` Dan Williams
-1 siblings, 0 replies; 10+ messages in thread
From: Dan Williams @ 2017-12-08 3:30 UTC (permalink / raw)
To: akpm
Cc: Jane Chu, linux-nvdimm, Benjamin Herrenschmidt, linux-mm,
Paul Mackerras, Michael Ellerman
When device-dax is operating in huge-page mode we want it to behave like
hugetlbfs and report the MMU page mapping size that is being enforced by
the vma. Similar to commit 31383c6865a5 "mm, hugetlbfs: introduce
->split() to vm_operations_struct" it would be messy to teach
vma_mmu_pagesize() about device-dax page mapping sizes in the same
(hstate) way that hugetlbfs communicates this attribute. Instead, these
patches introduce a new ->pagesize() vm operation.
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Reported-by: Jane Chu <jane.chu@oracle.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
---
arch/powerpc/mm/hugetlbpage.c | 5 +----
include/linux/hugetlb.h | 30 ++++++++++++++++++++++++------
include/linux/mm.h | 1 +
mm/hugetlb.c | 38 ++++++++------------------------------
4 files changed, 34 insertions(+), 40 deletions(-)
diff --git a/arch/powerpc/mm/hugetlbpage.c b/arch/powerpc/mm/hugetlbpage.c
index a9b9083c5e49..c6a2e577e842 100644
--- a/arch/powerpc/mm/hugetlbpage.c
+++ b/arch/powerpc/mm/hugetlbpage.c
@@ -568,10 +568,7 @@ unsigned long vma_mmu_pagesize(struct vm_area_struct *vma)
if (!radix_enabled())
return 1UL << mmu_psize_to_shift(psize);
#endif
- if (!is_vm_hugetlb_page(vma))
- return PAGE_SIZE;
-
- return huge_page_size(hstate_vma(vma));
+ return vma_kernel_pagesize(vma);
}
static inline bool is_power_of_4(unsigned long x)
diff --git a/include/linux/hugetlb.h b/include/linux/hugetlb.h
index 82a25880714a..716ccf14ff7b 100644
--- a/include/linux/hugetlb.h
+++ b/include/linux/hugetlb.h
@@ -254,6 +254,30 @@ enum {
HUGETLB_ANONHUGE_INODE = 2,
};
+/*
+ * Return the size of the pages allocated when backing a VMA. In the majority
+ * cases this will be same size as used by the page table entries.
+ */
+static inline unsigned long vma_kernel_pagesize(struct vm_area_struct *vma)
+{
+ if (vma->vm_ops && vma->vm_ops->pagesize)
+ return vma->vm_ops->pagesize(vma);
+ return PAGE_SIZE;
+}
+
+/*
+ * Return the page size being used by the MMU to back a VMA. In the majority
+ * of cases, the page size used by the kernel matches the MMU size. On
+ * architectures where it differs, an architecture-specific version of this
+ * function is required.
+ */
+#ifndef vma_mmu_pagesize
+static inline unsigned long vma_mmu_pagesize(struct vm_area_struct *vma)
+{
+ return vma_kernel_pagesize(vma);
+}
+#endif
+
#ifdef CONFIG_HUGETLBFS
struct hugetlbfs_sb_info {
long max_inodes; /* inodes allowed */
@@ -395,10 +419,6 @@ static inline unsigned long huge_page_size(struct hstate *h)
return (unsigned long)PAGE_SIZE << h->order;
}
-extern unsigned long vma_kernel_pagesize(struct vm_area_struct *vma);
-
-extern unsigned long vma_mmu_pagesize(struct vm_area_struct *vma);
-
static inline unsigned long huge_page_mask(struct hstate *h)
{
return h->mask;
@@ -533,8 +553,6 @@ struct hstate {};
#define page_hstate(page) NULL
#define huge_page_size(h) PAGE_SIZE
#define huge_page_mask(h) PAGE_MASK
-#define vma_kernel_pagesize(v) PAGE_SIZE
-#define vma_mmu_pagesize(v) PAGE_SIZE
#define huge_page_order(h) 0
#define huge_page_shift(h) PAGE_SHIFT
static inline bool hstate_is_gigantic(struct hstate *h)
diff --git a/include/linux/mm.h b/include/linux/mm.h
index ea818ff739cd..37b9aef91ec7 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -383,6 +383,7 @@ struct vm_operations_struct {
int (*huge_fault)(struct vm_fault *vmf, enum page_entry_size pe_size);
void (*map_pages)(struct vm_fault *vmf,
pgoff_t start_pgoff, pgoff_t end_pgoff);
+ unsigned long (*pagesize)(struct vm_area_struct * area);
/* notification that a previously read-only page is about to become
* writable, if an error is returned it will cause a SIGBUS */
diff --git a/mm/hugetlb.c b/mm/hugetlb.c
index 9a334f5fb730..f45ded1f978a 100644
--- a/mm/hugetlb.c
+++ b/mm/hugetlb.c
@@ -632,36 +632,6 @@ pgoff_t linear_hugepage_index(struct vm_area_struct *vma,
EXPORT_SYMBOL_GPL(linear_hugepage_index);
/*
- * Return the size of the pages allocated when backing a VMA. In the majority
- * cases this will be same size as used by the page table entries.
- */
-unsigned long vma_kernel_pagesize(struct vm_area_struct *vma)
-{
- struct hstate *hstate;
-
- if (!is_vm_hugetlb_page(vma))
- return PAGE_SIZE;
-
- hstate = hstate_vma(vma);
-
- return 1UL << huge_page_shift(hstate);
-}
-EXPORT_SYMBOL_GPL(vma_kernel_pagesize);
-
-/*
- * Return the page size being used by the MMU to back a VMA. In the majority
- * of cases, the page size used by the kernel matches the MMU size. On
- * architectures where it differs, an architecture-specific version of this
- * function is required.
- */
-#ifndef vma_mmu_pagesize
-unsigned long vma_mmu_pagesize(struct vm_area_struct *vma)
-{
- return vma_kernel_pagesize(vma);
-}
-#endif
-
-/*
* Flags for MAP_PRIVATE reservations. These are stored in the bottom
* bits of the reservation map pointer, which are always clear due to
* alignment.
@@ -3132,6 +3102,13 @@ static int hugetlb_vm_op_split(struct vm_area_struct *vma, unsigned long addr)
return 0;
}
+static unsigned long hugetlb_vm_op_pagesize(struct vm_area_struct *vma)
+{
+ struct hstate *hstate = hstate_vma(vma);
+
+ return 1UL << huge_page_shift(hstate);
+}
+
/*
* We cannot handle pagefaults against hugetlb pages at all. They cause
* handle_mm_fault() to try to instantiate regular-sized pages in the
@@ -3149,6 +3126,7 @@ const struct vm_operations_struct hugetlb_vm_ops = {
.open = hugetlb_vm_op_open,
.close = hugetlb_vm_op_close,
.split = hugetlb_vm_op_split,
+ .pagesize = hugetlb_vm_op_pagesize,
};
static pte_t make_huge_pte(struct vm_area_struct *vma, struct page *page,
_______________________________________________
Linux-nvdimm mailing list
Linux-nvdimm@lists.01.org
https://lists.01.org/mailman/listinfo/linux-nvdimm
^ permalink raw reply related [flat|nested] 10+ messages in thread
* [PATCH 1/2] mm, hugetlbfs: introduce ->pagesize() to vm_operations_struct
@ 2017-12-08 3:30 ` Dan Williams
0 siblings, 0 replies; 10+ messages in thread
From: Dan Williams @ 2017-12-08 3:30 UTC (permalink / raw)
To: akpm
Cc: Jane Chu, linux-nvdimm, Benjamin Herrenschmidt, linux-mm,
Paul Mackerras, Michael Ellerman
When device-dax is operating in huge-page mode we want it to behave like
hugetlbfs and report the MMU page mapping size that is being enforced by
the vma. Similar to commit 31383c6865a5 "mm, hugetlbfs: introduce
->split() to vm_operations_struct" it would be messy to teach
vma_mmu_pagesize() about device-dax page mapping sizes in the same
(hstate) way that hugetlbfs communicates this attribute. Instead, these
patches introduce a new ->pagesize() vm operation.
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Reported-by: Jane Chu <jane.chu@oracle.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
---
arch/powerpc/mm/hugetlbpage.c | 5 +----
include/linux/hugetlb.h | 30 ++++++++++++++++++++++++------
include/linux/mm.h | 1 +
mm/hugetlb.c | 38 ++++++++------------------------------
4 files changed, 34 insertions(+), 40 deletions(-)
diff --git a/arch/powerpc/mm/hugetlbpage.c b/arch/powerpc/mm/hugetlbpage.c
index a9b9083c5e49..c6a2e577e842 100644
--- a/arch/powerpc/mm/hugetlbpage.c
+++ b/arch/powerpc/mm/hugetlbpage.c
@@ -568,10 +568,7 @@ unsigned long vma_mmu_pagesize(struct vm_area_struct *vma)
if (!radix_enabled())
return 1UL << mmu_psize_to_shift(psize);
#endif
- if (!is_vm_hugetlb_page(vma))
- return PAGE_SIZE;
-
- return huge_page_size(hstate_vma(vma));
+ return vma_kernel_pagesize(vma);
}
static inline bool is_power_of_4(unsigned long x)
diff --git a/include/linux/hugetlb.h b/include/linux/hugetlb.h
index 82a25880714a..716ccf14ff7b 100644
--- a/include/linux/hugetlb.h
+++ b/include/linux/hugetlb.h
@@ -254,6 +254,30 @@ enum {
HUGETLB_ANONHUGE_INODE = 2,
};
+/*
+ * Return the size of the pages allocated when backing a VMA. In the majority
+ * cases this will be same size as used by the page table entries.
+ */
+static inline unsigned long vma_kernel_pagesize(struct vm_area_struct *vma)
+{
+ if (vma->vm_ops && vma->vm_ops->pagesize)
+ return vma->vm_ops->pagesize(vma);
+ return PAGE_SIZE;
+}
+
+/*
+ * Return the page size being used by the MMU to back a VMA. In the majority
+ * of cases, the page size used by the kernel matches the MMU size. On
+ * architectures where it differs, an architecture-specific version of this
+ * function is required.
+ */
+#ifndef vma_mmu_pagesize
+static inline unsigned long vma_mmu_pagesize(struct vm_area_struct *vma)
+{
+ return vma_kernel_pagesize(vma);
+}
+#endif
+
#ifdef CONFIG_HUGETLBFS
struct hugetlbfs_sb_info {
long max_inodes; /* inodes allowed */
@@ -395,10 +419,6 @@ static inline unsigned long huge_page_size(struct hstate *h)
return (unsigned long)PAGE_SIZE << h->order;
}
-extern unsigned long vma_kernel_pagesize(struct vm_area_struct *vma);
-
-extern unsigned long vma_mmu_pagesize(struct vm_area_struct *vma);
-
static inline unsigned long huge_page_mask(struct hstate *h)
{
return h->mask;
@@ -533,8 +553,6 @@ struct hstate {};
#define page_hstate(page) NULL
#define huge_page_size(h) PAGE_SIZE
#define huge_page_mask(h) PAGE_MASK
-#define vma_kernel_pagesize(v) PAGE_SIZE
-#define vma_mmu_pagesize(v) PAGE_SIZE
#define huge_page_order(h) 0
#define huge_page_shift(h) PAGE_SHIFT
static inline bool hstate_is_gigantic(struct hstate *h)
diff --git a/include/linux/mm.h b/include/linux/mm.h
index ea818ff739cd..37b9aef91ec7 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -383,6 +383,7 @@ struct vm_operations_struct {
int (*huge_fault)(struct vm_fault *vmf, enum page_entry_size pe_size);
void (*map_pages)(struct vm_fault *vmf,
pgoff_t start_pgoff, pgoff_t end_pgoff);
+ unsigned long (*pagesize)(struct vm_area_struct * area);
/* notification that a previously read-only page is about to become
* writable, if an error is returned it will cause a SIGBUS */
diff --git a/mm/hugetlb.c b/mm/hugetlb.c
index 9a334f5fb730..f45ded1f978a 100644
--- a/mm/hugetlb.c
+++ b/mm/hugetlb.c
@@ -632,36 +632,6 @@ pgoff_t linear_hugepage_index(struct vm_area_struct *vma,
EXPORT_SYMBOL_GPL(linear_hugepage_index);
/*
- * Return the size of the pages allocated when backing a VMA. In the majority
- * cases this will be same size as used by the page table entries.
- */
-unsigned long vma_kernel_pagesize(struct vm_area_struct *vma)
-{
- struct hstate *hstate;
-
- if (!is_vm_hugetlb_page(vma))
- return PAGE_SIZE;
-
- hstate = hstate_vma(vma);
-
- return 1UL << huge_page_shift(hstate);
-}
-EXPORT_SYMBOL_GPL(vma_kernel_pagesize);
-
-/*
- * Return the page size being used by the MMU to back a VMA. In the majority
- * of cases, the page size used by the kernel matches the MMU size. On
- * architectures where it differs, an architecture-specific version of this
- * function is required.
- */
-#ifndef vma_mmu_pagesize
-unsigned long vma_mmu_pagesize(struct vm_area_struct *vma)
-{
- return vma_kernel_pagesize(vma);
-}
-#endif
-
-/*
* Flags for MAP_PRIVATE reservations. These are stored in the bottom
* bits of the reservation map pointer, which are always clear due to
* alignment.
@@ -3132,6 +3102,13 @@ static int hugetlb_vm_op_split(struct vm_area_struct *vma, unsigned long addr)
return 0;
}
+static unsigned long hugetlb_vm_op_pagesize(struct vm_area_struct *vma)
+{
+ struct hstate *hstate = hstate_vma(vma);
+
+ return 1UL << huge_page_shift(hstate);
+}
+
/*
* We cannot handle pagefaults against hugetlb pages at all. They cause
* handle_mm_fault() to try to instantiate regular-sized pages in the
@@ -3149,6 +3126,7 @@ const struct vm_operations_struct hugetlb_vm_ops = {
.open = hugetlb_vm_op_open,
.close = hugetlb_vm_op_close,
.split = hugetlb_vm_op_split,
+ .pagesize = hugetlb_vm_op_pagesize,
};
static pte_t make_huge_pte(struct vm_area_struct *vma, struct page *page,
--
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>
^ permalink raw reply related [flat|nested] 10+ messages in thread
* [PATCH 2/2] device-dax: implement ->pagesize() for smaps to report MMUPageSize
2017-12-08 3:30 ` Dan Williams
@ 2017-12-08 3:31 ` Dan Williams
-1 siblings, 0 replies; 10+ messages in thread
From: Dan Williams @ 2017-12-08 3:31 UTC (permalink / raw)
To: akpm; +Cc: Jane Chu, linux-mm, linux-nvdimm
Given that device-dax is making similar page mapping size guarantees as
hugetlbfs, emit the size in smaps and any other kernel path that
requests the mapping size of a vma.
Reported-by: Jane Chu <jane.chu@oracle.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
---
drivers/dax/device.c | 10 ++++++++++
1 file changed, 10 insertions(+)
diff --git a/drivers/dax/device.c b/drivers/dax/device.c
index 7b0bf825c4e7..b57cd5a7b0bd 100644
--- a/drivers/dax/device.c
+++ b/drivers/dax/device.c
@@ -439,10 +439,20 @@ static int dev_dax_split(struct vm_area_struct *vma, unsigned long addr)
return 0;
}
+static unsigned long dev_dax_pagesize(struct vm_area_struct *vma)
+{
+ struct file *filp = vma->vm_file;
+ struct dev_dax *dev_dax = filp->private_data;
+ struct dax_region *dax_region = dev_dax->region;
+
+ return dax_region->align;
+}
+
static const struct vm_operations_struct dax_vm_ops = {
.fault = dev_dax_fault,
.huge_fault = dev_dax_huge_fault,
.split = dev_dax_split,
+ .pagesize = dev_dax_pagesize,
};
static int dax_mmap(struct file *filp, struct vm_area_struct *vma)
_______________________________________________
Linux-nvdimm mailing list
Linux-nvdimm@lists.01.org
https://lists.01.org/mailman/listinfo/linux-nvdimm
^ permalink raw reply related [flat|nested] 10+ messages in thread
* [PATCH 2/2] device-dax: implement ->pagesize() for smaps to report MMUPageSize
@ 2017-12-08 3:31 ` Dan Williams
0 siblings, 0 replies; 10+ messages in thread
From: Dan Williams @ 2017-12-08 3:31 UTC (permalink / raw)
To: akpm; +Cc: Jane Chu, linux-mm, linux-nvdimm
Given that device-dax is making similar page mapping size guarantees as
hugetlbfs, emit the size in smaps and any other kernel path that
requests the mapping size of a vma.
Reported-by: Jane Chu <jane.chu@oracle.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
---
drivers/dax/device.c | 10 ++++++++++
1 file changed, 10 insertions(+)
diff --git a/drivers/dax/device.c b/drivers/dax/device.c
index 7b0bf825c4e7..b57cd5a7b0bd 100644
--- a/drivers/dax/device.c
+++ b/drivers/dax/device.c
@@ -439,10 +439,20 @@ static int dev_dax_split(struct vm_area_struct *vma, unsigned long addr)
return 0;
}
+static unsigned long dev_dax_pagesize(struct vm_area_struct *vma)
+{
+ struct file *filp = vma->vm_file;
+ struct dev_dax *dev_dax = filp->private_data;
+ struct dax_region *dax_region = dev_dax->region;
+
+ return dax_region->align;
+}
+
static const struct vm_operations_struct dax_vm_ops = {
.fault = dev_dax_fault,
.huge_fault = dev_dax_huge_fault,
.split = dev_dax_split,
+ .pagesize = dev_dax_pagesize,
};
static int dax_mmap(struct file *filp, struct vm_area_struct *vma)
--
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>
^ permalink raw reply related [flat|nested] 10+ messages in thread
* Re: [PATCH 1/2] mm, hugetlbfs: introduce ->pagesize() to vm_operations_struct
2017-12-08 3:30 ` Dan Williams
@ 2017-12-10 11:37 ` Michal Hocko
-1 siblings, 0 replies; 10+ messages in thread
From: Michal Hocko @ 2017-12-10 11:37 UTC (permalink / raw)
To: Dan Williams
Cc: Jane Chu, linux-nvdimm, Benjamin Herrenschmidt, linux-mm,
Paul Mackerras, Michael Ellerman, akpm
On Thu 07-12-17 19:30:55, Dan Williams wrote:
> When device-dax is operating in huge-page mode we want it to behave like
> hugetlbfs and report the MMU page mapping size that is being enforced by
> the vma. Similar to commit 31383c6865a5 "mm, hugetlbfs: introduce
> ->split() to vm_operations_struct" it would be messy to teach
> vma_mmu_pagesize() about device-dax page mapping sizes in the same
> (hstate) way that hugetlbfs communicates this attribute. Instead, these
> patches introduce a new ->pagesize() vm operation.
>
> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
> Cc: Paul Mackerras <paulus@samba.org>
> Cc: Michael Ellerman <mpe@ellerman.id.au>
> Reported-by: Jane Chu <jane.chu@oracle.com>
> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
My build battery choked on the following
In file included from drivers/infiniband/core/umem_odp.c:41:0:
./include/linux/hugetlb.h: In function 'vma_kernel_pagesize':
./include/linux/hugetlb.h:262:32: error: dereferencing pointer to incomplete type
if (vma->vm_ops && vma->vm_ops->pagesize)
^
./include/linux/hugetlb.h:263:21: error: dereferencing pointer to incomplete type
return vma->vm_ops->pagesize(vma);
I thought that adding #include <linux/mm.h> into linux/hugetlb.h would
be sufficient but then it failed for powerpc defconfig which overrides
vma_kernel_pagesize
In file included from ./include/linux/hugetlb.h:452:0,
from arch/powerpc/mm/hugetlbpage.c:14:
./arch/powerpc/include/asm/hugetlb.h:131:26: error: redefinition of 'vma_mmu_pagesize'
#define vma_mmu_pagesize vma_mmu_pagesize
^
arch/powerpc/mm/hugetlbpage.c:563:15: note: in expansion of macro 'vma_mmu_pagesize'
unsigned long vma_mmu_pagesize(struct vm_area_struct *vma)
^
In file included from arch/powerpc/mm/hugetlbpage.c:14:0:
./include/linux/hugetlb.h:275:29: note: previous definition of 'vma_mmu_pagesize' was here
static inline unsigned long vma_mmu_pagesize(struct vm_area_struct *vma)
So it looks this needs something more laborous.
--
Michal Hocko
SUSE Labs
_______________________________________________
Linux-nvdimm mailing list
Linux-nvdimm@lists.01.org
https://lists.01.org/mailman/listinfo/linux-nvdimm
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH 1/2] mm, hugetlbfs: introduce ->pagesize() to vm_operations_struct
@ 2017-12-10 11:37 ` Michal Hocko
0 siblings, 0 replies; 10+ messages in thread
From: Michal Hocko @ 2017-12-10 11:37 UTC (permalink / raw)
To: Dan Williams
Cc: akpm, Jane Chu, linux-nvdimm, Benjamin Herrenschmidt, linux-mm,
Paul Mackerras, Michael Ellerman
On Thu 07-12-17 19:30:55, Dan Williams wrote:
> When device-dax is operating in huge-page mode we want it to behave like
> hugetlbfs and report the MMU page mapping size that is being enforced by
> the vma. Similar to commit 31383c6865a5 "mm, hugetlbfs: introduce
> ->split() to vm_operations_struct" it would be messy to teach
> vma_mmu_pagesize() about device-dax page mapping sizes in the same
> (hstate) way that hugetlbfs communicates this attribute. Instead, these
> patches introduce a new ->pagesize() vm operation.
>
> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
> Cc: Paul Mackerras <paulus@samba.org>
> Cc: Michael Ellerman <mpe@ellerman.id.au>
> Reported-by: Jane Chu <jane.chu@oracle.com>
> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
My build battery choked on the following
In file included from drivers/infiniband/core/umem_odp.c:41:0:
./include/linux/hugetlb.h: In function 'vma_kernel_pagesize':
./include/linux/hugetlb.h:262:32: error: dereferencing pointer to incomplete type
if (vma->vm_ops && vma->vm_ops->pagesize)
^
./include/linux/hugetlb.h:263:21: error: dereferencing pointer to incomplete type
return vma->vm_ops->pagesize(vma);
I thought that adding #include <linux/mm.h> into linux/hugetlb.h would
be sufficient but then it failed for powerpc defconfig which overrides
vma_kernel_pagesize
In file included from ./include/linux/hugetlb.h:452:0,
from arch/powerpc/mm/hugetlbpage.c:14:
./arch/powerpc/include/asm/hugetlb.h:131:26: error: redefinition of 'vma_mmu_pagesize'
#define vma_mmu_pagesize vma_mmu_pagesize
^
arch/powerpc/mm/hugetlbpage.c:563:15: note: in expansion of macro 'vma_mmu_pagesize'
unsigned long vma_mmu_pagesize(struct vm_area_struct *vma)
^
In file included from arch/powerpc/mm/hugetlbpage.c:14:0:
./include/linux/hugetlb.h:275:29: note: previous definition of 'vma_mmu_pagesize' was here
static inline unsigned long vma_mmu_pagesize(struct vm_area_struct *vma)
So it looks this needs something more laborous.
--
Michal Hocko
SUSE Labs
--
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>
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH 1/2] mm, hugetlbfs: introduce ->pagesize() to vm_operations_struct
2017-12-10 11:37 ` Michal Hocko
@ 2017-12-13 13:07 ` Michael Ellerman
-1 siblings, 0 replies; 10+ messages in thread
From: Michael Ellerman @ 2017-12-13 13:07 UTC (permalink / raw)
To: Michal Hocko, Dan Williams
Cc: Jane Chu, linux-nvdimm, Benjamin Herrenschmidt, linux-mm,
Paul Mackerras, akpm
Michal Hocko <mhocko@kernel.org> writes:
> On Thu 07-12-17 19:30:55, Dan Williams wrote:
>> When device-dax is operating in huge-page mode we want it to behave like
>> hugetlbfs and report the MMU page mapping size that is being enforced by
>> the vma. Similar to commit 31383c6865a5 "mm, hugetlbfs: introduce
>> ->split() to vm_operations_struct" it would be messy to teach
>> vma_mmu_pagesize() about device-dax page mapping sizes in the same
>> (hstate) way that hugetlbfs communicates this attribute. Instead, these
>> patches introduce a new ->pagesize() vm operation.
>>
>> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
>> Cc: Paul Mackerras <paulus@samba.org>
>> Cc: Michael Ellerman <mpe@ellerman.id.au>
>> Reported-by: Jane Chu <jane.chu@oracle.com>
>> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
>
> My build battery choked on the following
> In file included from drivers/infiniband/core/umem_odp.c:41:0:
> ./include/linux/hugetlb.h: In function 'vma_kernel_pagesize':
> ./include/linux/hugetlb.h:262:32: error: dereferencing pointer to incomplete type
> if (vma->vm_ops && vma->vm_ops->pagesize)
> ^
> ./include/linux/hugetlb.h:263:21: error: dereferencing pointer to incomplete type
> return vma->vm_ops->pagesize(vma);
>
> I thought that adding #include <linux/mm.h> into linux/hugetlb.h would
> be sufficient but then it failed for powerpc defconfig which overrides
> vma_kernel_pagesize
> In file included from ./include/linux/hugetlb.h:452:0,
> from arch/powerpc/mm/hugetlbpage.c:14:
> ./arch/powerpc/include/asm/hugetlb.h:131:26: error: redefinition of 'vma_mmu_pagesize'
> #define vma_mmu_pagesize vma_mmu_pagesize
> ^
> arch/powerpc/mm/hugetlbpage.c:563:15: note: in expansion of macro 'vma_mmu_pagesize'
> unsigned long vma_mmu_pagesize(struct vm_area_struct *vma)
> ^
> In file included from arch/powerpc/mm/hugetlbpage.c:14:0:
> ./include/linux/hugetlb.h:275:29: note: previous definition of 'vma_mmu_pagesize' was here
> static inline unsigned long vma_mmu_pagesize(struct vm_area_struct *vma)
>
> So it looks this needs something more laborous.
This builds for me.
cheers
diff --git a/arch/powerpc/include/asm/hugetlb.h b/arch/powerpc/include/asm/hugetlb.h
index 14c9d44f355b..3cc6ca1bdaf2 100644
--- a/arch/powerpc/include/asm/hugetlb.h
+++ b/arch/powerpc/include/asm/hugetlb.h
@@ -123,6 +123,7 @@ void hugetlb_free_pgd_range(struct mmu_gather *tlb, unsigned long addr,
* to override the version in mm/hugetlb.c
*/
#define vma_mmu_pagesize vma_mmu_pagesize
+unsigned long vma_mmu_pagesize(struct vm_area_struct *vma);
/*
* If the arch doesn't supply something else, assume that hugepage
diff --git a/arch/powerpc/mm/hugetlbpage.c b/arch/powerpc/mm/hugetlbpage.c
index a9b9083c5e49..c6a2e577e842 100644
--- a/arch/powerpc/mm/hugetlbpage.c
+++ b/arch/powerpc/mm/hugetlbpage.c
@@ -568,10 +568,7 @@ unsigned long vma_mmu_pagesize(struct vm_area_struct *vma)
if (!radix_enabled())
return 1UL << mmu_psize_to_shift(psize);
#endif
- if (!is_vm_hugetlb_page(vma))
- return PAGE_SIZE;
-
- return huge_page_size(hstate_vma(vma));
+ return vma_kernel_pagesize(vma);
}
static inline bool is_power_of_4(unsigned long x)
diff --git a/include/linux/hugetlb.h b/include/linux/hugetlb.h
index 6e3696c7b35a..fe7b74325856 100644
--- a/include/linux/hugetlb.h
+++ b/include/linux/hugetlb.h
@@ -394,10 +394,6 @@ static inline unsigned long huge_page_size(struct hstate *h)
return (unsigned long)PAGE_SIZE << h->order;
}
-extern unsigned long vma_kernel_pagesize(struct vm_area_struct *vma);
-
-extern unsigned long vma_mmu_pagesize(struct vm_area_struct *vma);
-
static inline unsigned long huge_page_mask(struct hstate *h)
{
return h->mask;
@@ -430,6 +426,30 @@ static inline unsigned int blocks_per_huge_page(struct hstate *h)
#include <asm/hugetlb.h>
+/*
+ * Return the size of the pages allocated when backing a VMA. In the majority
+ * cases this will be same size as used by the page table entries.
+ */
+static inline unsigned long vma_kernel_pagesize(struct vm_area_struct *vma)
+{
+ if (vma->vm_ops && vma->vm_ops->pagesize)
+ return vma->vm_ops->pagesize(vma);
+ return PAGE_SIZE;
+}
+
+/*
+ * Return the page size being used by the MMU to back a VMA. In the majority
+ * of cases, the page size used by the kernel matches the MMU size. On
+ * architectures where it differs, an architecture-specific version of this
+ * function is required.
+ */
+#ifndef vma_mmu_pagesize
+static inline unsigned long vma_mmu_pagesize(struct vm_area_struct *vma)
+{
+ return vma_kernel_pagesize(vma);
+}
+#endif
+
#ifndef arch_make_huge_pte
static inline pte_t arch_make_huge_pte(pte_t entry, struct vm_area_struct *vma,
struct page *page, int writable)
@@ -533,8 +553,6 @@ struct hstate {};
#define page_hstate(page) NULL
#define huge_page_size(h) PAGE_SIZE
#define huge_page_mask(h) PAGE_MASK
-#define vma_kernel_pagesize(v) PAGE_SIZE
-#define vma_mmu_pagesize(v) PAGE_SIZE
#define huge_page_order(h) 0
#define huge_page_shift(h) PAGE_SHIFT
static inline bool hstate_is_gigantic(struct hstate *h)
diff --git a/include/linux/mm.h b/include/linux/mm.h
index 7661156552d3..1933499f896d 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -383,6 +383,7 @@ struct vm_operations_struct {
int (*huge_fault)(struct vm_fault *vmf, enum page_entry_size pe_size);
void (*map_pages)(struct vm_fault *vmf,
pgoff_t start_pgoff, pgoff_t end_pgoff);
+ unsigned long (*pagesize)(struct vm_area_struct * area);
/* notification that a previously read-only page is about to become
* writable, if an error is returned it will cause a SIGBUS */
diff --git a/mm/hugetlb.c b/mm/hugetlb.c
index 4137fb67cd79..7c1c45bb3d08 100644
--- a/mm/hugetlb.c
+++ b/mm/hugetlb.c
@@ -629,36 +629,6 @@ pgoff_t linear_hugepage_index(struct vm_area_struct *vma,
}
EXPORT_SYMBOL_GPL(linear_hugepage_index);
-/*
- * Return the size of the pages allocated when backing a VMA. In the majority
- * cases this will be same size as used by the page table entries.
- */
-unsigned long vma_kernel_pagesize(struct vm_area_struct *vma)
-{
- struct hstate *hstate;
-
- if (!is_vm_hugetlb_page(vma))
- return PAGE_SIZE;
-
- hstate = hstate_vma(vma);
-
- return 1UL << huge_page_shift(hstate);
-}
-EXPORT_SYMBOL_GPL(vma_kernel_pagesize);
-
-/*
- * Return the page size being used by the MMU to back a VMA. In the majority
- * of cases, the page size used by the kernel matches the MMU size. On
- * architectures where it differs, an architecture-specific version of this
- * function is required.
- */
-#ifndef vma_mmu_pagesize
-unsigned long vma_mmu_pagesize(struct vm_area_struct *vma)
-{
- return vma_kernel_pagesize(vma);
-}
-#endif
-
/*
* Flags for MAP_PRIVATE reservations. These are stored in the bottom
* bits of the reservation map pointer, which are always clear due to
@@ -3142,6 +3112,13 @@ static int hugetlb_vm_op_split(struct vm_area_struct *vma, unsigned long addr)
return 0;
}
+static unsigned long hugetlb_vm_op_pagesize(struct vm_area_struct *vma)
+{
+ struct hstate *hstate = hstate_vma(vma);
+
+ return 1UL << huge_page_shift(hstate);
+}
+
/*
* We cannot handle pagefaults against hugetlb pages at all. They cause
* handle_mm_fault() to try to instantiate regular-sized pages in the
@@ -3159,6 +3136,7 @@ const struct vm_operations_struct hugetlb_vm_ops = {
.open = hugetlb_vm_op_open,
.close = hugetlb_vm_op_close,
.split = hugetlb_vm_op_split,
+ .pagesize = hugetlb_vm_op_pagesize,
};
static pte_t make_huge_pte(struct vm_area_struct *vma, struct page *page,
_______________________________________________
Linux-nvdimm mailing list
Linux-nvdimm@lists.01.org
https://lists.01.org/mailman/listinfo/linux-nvdimm
^ permalink raw reply related [flat|nested] 10+ messages in thread
* Re: [PATCH 1/2] mm, hugetlbfs: introduce ->pagesize() to vm_operations_struct
@ 2017-12-13 13:07 ` Michael Ellerman
0 siblings, 0 replies; 10+ messages in thread
From: Michael Ellerman @ 2017-12-13 13:07 UTC (permalink / raw)
To: Michal Hocko, Dan Williams
Cc: akpm, Jane Chu, linux-nvdimm, Benjamin Herrenschmidt, linux-mm,
Paul Mackerras
Michal Hocko <mhocko@kernel.org> writes:
> On Thu 07-12-17 19:30:55, Dan Williams wrote:
>> When device-dax is operating in huge-page mode we want it to behave like
>> hugetlbfs and report the MMU page mapping size that is being enforced by
>> the vma. Similar to commit 31383c6865a5 "mm, hugetlbfs: introduce
>> ->split() to vm_operations_struct" it would be messy to teach
>> vma_mmu_pagesize() about device-dax page mapping sizes in the same
>> (hstate) way that hugetlbfs communicates this attribute. Instead, these
>> patches introduce a new ->pagesize() vm operation.
>>
>> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
>> Cc: Paul Mackerras <paulus@samba.org>
>> Cc: Michael Ellerman <mpe@ellerman.id.au>
>> Reported-by: Jane Chu <jane.chu@oracle.com>
>> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
>
> My build battery choked on the following
> In file included from drivers/infiniband/core/umem_odp.c:41:0:
> ./include/linux/hugetlb.h: In function 'vma_kernel_pagesize':
> ./include/linux/hugetlb.h:262:32: error: dereferencing pointer to incomplete type
> if (vma->vm_ops && vma->vm_ops->pagesize)
> ^
> ./include/linux/hugetlb.h:263:21: error: dereferencing pointer to incomplete type
> return vma->vm_ops->pagesize(vma);
>
> I thought that adding #include <linux/mm.h> into linux/hugetlb.h would
> be sufficient but then it failed for powerpc defconfig which overrides
> vma_kernel_pagesize
> In file included from ./include/linux/hugetlb.h:452:0,
> from arch/powerpc/mm/hugetlbpage.c:14:
> ./arch/powerpc/include/asm/hugetlb.h:131:26: error: redefinition of 'vma_mmu_pagesize'
> #define vma_mmu_pagesize vma_mmu_pagesize
> ^
> arch/powerpc/mm/hugetlbpage.c:563:15: note: in expansion of macro 'vma_mmu_pagesize'
> unsigned long vma_mmu_pagesize(struct vm_area_struct *vma)
> ^
> In file included from arch/powerpc/mm/hugetlbpage.c:14:0:
> ./include/linux/hugetlb.h:275:29: note: previous definition of 'vma_mmu_pagesize' was here
> static inline unsigned long vma_mmu_pagesize(struct vm_area_struct *vma)
>
> So it looks this needs something more laborous.
This builds for me.
cheers
diff --git a/arch/powerpc/include/asm/hugetlb.h b/arch/powerpc/include/asm/hugetlb.h
index 14c9d44f355b..3cc6ca1bdaf2 100644
--- a/arch/powerpc/include/asm/hugetlb.h
+++ b/arch/powerpc/include/asm/hugetlb.h
@@ -123,6 +123,7 @@ void hugetlb_free_pgd_range(struct mmu_gather *tlb, unsigned long addr,
* to override the version in mm/hugetlb.c
*/
#define vma_mmu_pagesize vma_mmu_pagesize
+unsigned long vma_mmu_pagesize(struct vm_area_struct *vma);
/*
* If the arch doesn't supply something else, assume that hugepage
diff --git a/arch/powerpc/mm/hugetlbpage.c b/arch/powerpc/mm/hugetlbpage.c
index a9b9083c5e49..c6a2e577e842 100644
--- a/arch/powerpc/mm/hugetlbpage.c
+++ b/arch/powerpc/mm/hugetlbpage.c
@@ -568,10 +568,7 @@ unsigned long vma_mmu_pagesize(struct vm_area_struct *vma)
if (!radix_enabled())
return 1UL << mmu_psize_to_shift(psize);
#endif
- if (!is_vm_hugetlb_page(vma))
- return PAGE_SIZE;
-
- return huge_page_size(hstate_vma(vma));
+ return vma_kernel_pagesize(vma);
}
static inline bool is_power_of_4(unsigned long x)
diff --git a/include/linux/hugetlb.h b/include/linux/hugetlb.h
index 6e3696c7b35a..fe7b74325856 100644
--- a/include/linux/hugetlb.h
+++ b/include/linux/hugetlb.h
@@ -394,10 +394,6 @@ static inline unsigned long huge_page_size(struct hstate *h)
return (unsigned long)PAGE_SIZE << h->order;
}
-extern unsigned long vma_kernel_pagesize(struct vm_area_struct *vma);
-
-extern unsigned long vma_mmu_pagesize(struct vm_area_struct *vma);
-
static inline unsigned long huge_page_mask(struct hstate *h)
{
return h->mask;
@@ -430,6 +426,30 @@ static inline unsigned int blocks_per_huge_page(struct hstate *h)
#include <asm/hugetlb.h>
+/*
+ * Return the size of the pages allocated when backing a VMA. In the majority
+ * cases this will be same size as used by the page table entries.
+ */
+static inline unsigned long vma_kernel_pagesize(struct vm_area_struct *vma)
+{
+ if (vma->vm_ops && vma->vm_ops->pagesize)
+ return vma->vm_ops->pagesize(vma);
+ return PAGE_SIZE;
+}
+
+/*
+ * Return the page size being used by the MMU to back a VMA. In the majority
+ * of cases, the page size used by the kernel matches the MMU size. On
+ * architectures where it differs, an architecture-specific version of this
+ * function is required.
+ */
+#ifndef vma_mmu_pagesize
+static inline unsigned long vma_mmu_pagesize(struct vm_area_struct *vma)
+{
+ return vma_kernel_pagesize(vma);
+}
+#endif
+
#ifndef arch_make_huge_pte
static inline pte_t arch_make_huge_pte(pte_t entry, struct vm_area_struct *vma,
struct page *page, int writable)
@@ -533,8 +553,6 @@ struct hstate {};
#define page_hstate(page) NULL
#define huge_page_size(h) PAGE_SIZE
#define huge_page_mask(h) PAGE_MASK
-#define vma_kernel_pagesize(v) PAGE_SIZE
-#define vma_mmu_pagesize(v) PAGE_SIZE
#define huge_page_order(h) 0
#define huge_page_shift(h) PAGE_SHIFT
static inline bool hstate_is_gigantic(struct hstate *h)
diff --git a/include/linux/mm.h b/include/linux/mm.h
index 7661156552d3..1933499f896d 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -383,6 +383,7 @@ struct vm_operations_struct {
int (*huge_fault)(struct vm_fault *vmf, enum page_entry_size pe_size);
void (*map_pages)(struct vm_fault *vmf,
pgoff_t start_pgoff, pgoff_t end_pgoff);
+ unsigned long (*pagesize)(struct vm_area_struct * area);
/* notification that a previously read-only page is about to become
* writable, if an error is returned it will cause a SIGBUS */
diff --git a/mm/hugetlb.c b/mm/hugetlb.c
index 4137fb67cd79..7c1c45bb3d08 100644
--- a/mm/hugetlb.c
+++ b/mm/hugetlb.c
@@ -629,36 +629,6 @@ pgoff_t linear_hugepage_index(struct vm_area_struct *vma,
}
EXPORT_SYMBOL_GPL(linear_hugepage_index);
-/*
- * Return the size of the pages allocated when backing a VMA. In the majority
- * cases this will be same size as used by the page table entries.
- */
-unsigned long vma_kernel_pagesize(struct vm_area_struct *vma)
-{
- struct hstate *hstate;
-
- if (!is_vm_hugetlb_page(vma))
- return PAGE_SIZE;
-
- hstate = hstate_vma(vma);
-
- return 1UL << huge_page_shift(hstate);
-}
-EXPORT_SYMBOL_GPL(vma_kernel_pagesize);
-
-/*
- * Return the page size being used by the MMU to back a VMA. In the majority
- * of cases, the page size used by the kernel matches the MMU size. On
- * architectures where it differs, an architecture-specific version of this
- * function is required.
- */
-#ifndef vma_mmu_pagesize
-unsigned long vma_mmu_pagesize(struct vm_area_struct *vma)
-{
- return vma_kernel_pagesize(vma);
-}
-#endif
-
/*
* Flags for MAP_PRIVATE reservations. These are stored in the bottom
* bits of the reservation map pointer, which are always clear due to
@@ -3142,6 +3112,13 @@ static int hugetlb_vm_op_split(struct vm_area_struct *vma, unsigned long addr)
return 0;
}
+static unsigned long hugetlb_vm_op_pagesize(struct vm_area_struct *vma)
+{
+ struct hstate *hstate = hstate_vma(vma);
+
+ return 1UL << huge_page_shift(hstate);
+}
+
/*
* We cannot handle pagefaults against hugetlb pages at all. They cause
* handle_mm_fault() to try to instantiate regular-sized pages in the
@@ -3159,6 +3136,7 @@ const struct vm_operations_struct hugetlb_vm_ops = {
.open = hugetlb_vm_op_open,
.close = hugetlb_vm_op_close,
.split = hugetlb_vm_op_split,
+ .pagesize = hugetlb_vm_op_pagesize,
};
static pte_t make_huge_pte(struct vm_area_struct *vma, struct page *page,
--
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>
^ permalink raw reply related [flat|nested] 10+ messages in thread
end of thread, other threads:[~2017-12-13 13:07 UTC | newest]
Thread overview: 10+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2017-12-08 3:30 [PATCH 0/2] mm, smaps: MMUPageSize for device-dax Dan Williams
2017-12-08 3:30 ` Dan Williams
2017-12-08 3:30 ` [PATCH 1/2] mm, hugetlbfs: introduce ->pagesize() to vm_operations_struct Dan Williams
2017-12-08 3:30 ` Dan Williams
2017-12-10 11:37 ` Michal Hocko
2017-12-10 11:37 ` Michal Hocko
2017-12-13 13:07 ` Michael Ellerman
2017-12-13 13:07 ` Michael Ellerman
2017-12-08 3:31 ` [PATCH 2/2] device-dax: implement ->pagesize() for smaps to report MMUPageSize Dan Williams
2017-12-08 3:31 ` Dan Williams
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.