The patch titled
Subject: mm, x86: get_user_pages() for dax mappings
has been added to the -mm tree. Its filename is
mm-x86-get_user_pages-for-dax-mappings-v5.patch
This patch should soon appear at
http://ozlabs.org/~akpm/mmots/broken-out/mm-x86-get_user_pages-for-dax-mappings-v5.patch
and later at
http://ozlabs.org/~akpm/mmotm/broken-out/mm-x86-get_user_pages-for-dax-mappings-v5.patch
Before you just go and hit "reply", please:
a) Consider who else should be cc'ed
b) Prefer to cc a suitable mailing list as well
c) Ideally: find the original patch on the mailing list and do a
reply-to-all to that, adding suitable additional cc's
*** Remember to use Documentation/SubmitChecklist when testing your code ***
The -mm tree is included into linux-next and is updated
there every 3-4 working days
------------------------------------------------------
From: Dan Williams <dan.j.williams@xxxxxxxxx>
Subject: mm, x86: get_user_pages() for dax mappings
Changes since v4:
1/ Fix put_page() to drop the zone_device dev_pagemap reference, otherwise
we hang at driver unload due to busy references.
2/ Fix follow_page_pte() to properly handle the dev_pagemap reference
count.
Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx>
Tested-by: Logan Gunthorpe <logang@xxxxxxxxxxxx>
Cc: Dave Hansen <dave@xxxxxxxx>
Cc: Mel Gorman <mgorman@xxxxxxx>
Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
Cc: Andrea Arcangeli <aarcange@xxxxxxxxxx>
Cc: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
Cc: Ingo Molnar <mingo@xxxxxxxxxx>
Cc: "H. Peter Anvin" <hpa@xxxxxxxxx>
Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
---
include/linux/mm.h | 23 +++++++++++++----------
kernel/memremap.c | 12 ++----------
mm/gup.c | 13 +++++++++++--
mm/swap.c | 2 --
4 files changed, 26 insertions(+), 24 deletions(-)
diff -puN include/linux/mm.h~mm-x86-get_user_pages-for-dax-mappings-v5 include/linux/mm.h
--- a/include/linux/mm.h~mm-x86-get_user_pages-for-dax-mappings-v5
+++ a/include/linux/mm.h
@@ -477,13 +477,6 @@ static inline void init_page_count(struc
void __put_page(struct page *page);
-static inline void put_page(struct page *page)
-{
- page = compound_head(page);
- if (put_page_testzero(page))
- __put_page(page);
-}
-
void put_pages_list(struct list_head *pages);
void split_page(struct page *page, unsigned int order);
@@ -666,7 +659,7 @@ static inline enum zone_type page_zonenu
#ifdef CONFIG_ZONE_DEVICE
void get_zone_device_page(struct page *page);
-int release_zone_device_page(struct page *page);
+void put_zone_device_page(struct page *page);
static inline bool is_zone_device_page(const struct page *page)
{
return page_zonenum(page) == ZONE_DEVICE;
@@ -675,9 +668,8 @@ static inline bool is_zone_device_page(c
static inline void get_zone_device_page(struct page *page)
{
}
-static inline int release_zone_device_page(struct page *page)
+static inline void put_zone_device_page(struct page *page)
{
- return 0;
}
static inline bool is_zone_device_page(const struct page *page)
{
@@ -699,6 +691,17 @@ static inline void get_page(struct page
get_zone_device_page(page);
}
+static inline void put_page(struct page *page)
+{
+ page = compound_head(page);
+
+ if (put_page_testzero(page))
+ __put_page(page);
+
+ if (unlikely(is_zone_device_page(page)))
+ put_zone_device_page(page);
+}
+
#if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
#define SECTION_IN_PAGE_FLAGS
#endif
diff -puN kernel/memremap.c~mm-x86-get_user_pages-for-dax-mappings-v5 kernel/memremap.c
--- a/kernel/memremap.c~mm-x86-get_user_pages-for-dax-mappings-v5
+++ a/kernel/memremap.c
@@ -175,19 +175,11 @@ void get_zone_device_page(struct page *p
}
EXPORT_SYMBOL(get_zone_device_page);
-int release_zone_device_page(struct page *page)
+void put_zone_device_page(struct page *page)
{
- /*
- * ZONE_DEVICE pages are never "onlined" so their reference
- * counts never reach zero. They are always owned by a device
- * driver, not the mm core. I.e. the page is 'idle' when the
- * count is 1.
- */
- VM_BUG_ON_PAGE(atomic_read(&page->_count) == 1, page);
put_dev_pagemap(page->pgmap);
- return atomic_dec_return(&page->_count) == 1;
}
-EXPORT_SYMBOL(release_zone_device_page);
+EXPORT_SYMBOL(put_zone_device_page);
static void pgmap_radix_release(struct resource *res)
{
diff -puN mm/gup.c~mm-x86-get_user_pages-for-dax-mappings-v5 mm/gup.c
--- a/mm/gup.c~mm-x86-get_user_pages-for-dax-mappings-v5
+++ a/mm/gup.c
@@ -63,6 +63,7 @@ static struct page *follow_page_pte(stru
unsigned long address, pmd_t *pmd, unsigned int flags)
{
struct mm_struct *mm = vma->vm_mm;
+ struct dev_pagemap *pgmap = NULL;
struct page *page;
spinlock_t *ptl;
pte_t *ptep, pte;
@@ -104,7 +105,8 @@ retry:
* Only return device mapping pages in the FOLL_GET case since
* they are only valid while holding the pgmap reference.
*/
- if (get_dev_pagemap(pte_pfn(pte), NULL))
+ pgmap = get_dev_pagemap(pte_pfn(pte), NULL);
+ if (pgmap)
page = pte_page(pte);
else
goto no_page;
@@ -139,8 +141,15 @@ retry:
goto retry;
}
- if (flags & FOLL_GET)
+ if (flags & FOLL_GET) {
get_page(page);
+
+ /* drop the pgmap reference now that we hold the page */
+ if (pgmap) {
+ put_dev_pagemap(pgmap);
+ pgmap = NULL;
+ }
+ }
if (flags & FOLL_TOUCH) {
if ((flags & FOLL_WRITE) &&
!pte_dirty(pte) && !PageDirty(page))
diff -puN mm/swap.c~mm-x86-get_user_pages-for-dax-mappings-v5 mm/swap.c
--- a/mm/swap.c~mm-x86-get_user_pages-for-dax-mappings-v5
+++ a/mm/swap.c
@@ -95,8 +95,6 @@ void __put_page(struct page *page)
{
if (unlikely(PageCompound(page)))
__put_compound_page(page);
- else if (unlikely(is_zone_device_page(page)))
- release_zone_device_page(page);
else
__put_single_page(page);
}
_
Patches currently in -mm which might be from dan.j.williams@xxxxxxxxx are
pmem-dax-clean-up-clear_pmem.patch
dax-increase-granularity-of-dax_clear_blocks-operations.patch
dax-guarantee-page-aligned-results-from-bdev_direct_access.patch
dax-fix-lifetime-of-in-kernel-dax-mappings-with-dax_map_atomic.patch
dax-fix-lifetime-of-in-kernel-dax-mappings-with-dax_map_atomic-v3.patch
um-kill-pfn_t.patch
kvm-rename-pfn_t-to-kvm_pfn_t.patch
mm-dax-pmem-introduce-pfn_t.patch
mm-skip-memory-block-registration-for-zone_device.patch
mm-introduce-find_dev_pagemap.patch
x86-mm-introduce-vmem_altmap-to-augment-vmemmap_populate.patch
libnvdimm-pfn-pmem-allocate-memmap-array-in-persistent-memory.patch
avr32-convert-to-asm-generic-memory_modelh.patch
hugetlb-fix-compile-error-on-tile.patch
frv-fix-compiler-warning-from-definition-of-__pmd.patch
x86-mm-introduce-_page_devmap.patch
mm-dax-gpu-convert-vm_insert_mixed-to-pfn_t.patch
mm-dax-convert-vmf_insert_pfn_pmd-to-pfn_t.patch
libnvdimm-pmem-move-request_queue-allocation-earlier-in-probe.patch
mm-dax-pmem-introduce-getput_dev_pagemap-for-dax-gup.patch
mm-dax-dax-pmd-vs-thp-pmd-vs-hugetlbfs-pmd.patch
mm-dax-dax-pmd-vs-thp-pmd-vs-hugetlbfs-pmd-v5.patch
mm-x86-get_user_pages-for-dax-mappings.patch
mm-x86-get_user_pages-for-dax-mappings-v5.patch
dax-provide-diagnostics-for-pmd-mapping-failures.patch
dax-re-enable-dax-pmd-mappings.patch
--
To unsubscribe from this list: send the line "unsubscribe mm-commits" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
