* [PATCH] mm: softdirty: unmapped addresses between VMAs are clean
@ 2014-09-10 23:24 Peter Feiner
2014-09-10 23:36 ` Andrew Morton
2014-09-15 18:40 ` [PATCH v2] " Peter Feiner
0 siblings, 2 replies; 7+ messages in thread
From: Peter Feiner @ 2014-09-10 23:24 UTC (permalink / raw)
To: linux-mm
Cc: linux-kernel, Peter Feiner, Kirill A. Shutemov, Cyrill Gorcunov,
Pavel Emelyanov, Jamie Liu, Hugh Dickins, Naoya Horiguchi,
Andrew Morton
If a /proc/pid/pagemap read spans a [VMA, an unmapped region, then a
VM_SOFTDIRTY VMA], the virtual pages in the unmapped region are reported
as softdirty. Here's a program to demonstrate the bug:
int main() {
const uint64_t PAGEMAP_SOFTDIRTY = 1ul << 55;
uint64_t pme[3];
int fd = open("/proc/self/pagemap", O_RDONLY);;
char *m = mmap(NULL, 3 * getpagesize(), PROT_READ,
MAP_ANONYMOUS | MAP_SHARED, -1, 0);
munmap(m + getpagesize(), getpagesize());
pread(fd, pme, 24, (unsigned long) m / getpagesize() * 8);
assert(pme[0] & PAGEMAP_SOFTDIRTY); /* passes */
assert(!(pme[1] & PAGEMAP_SOFTDIRTY)); /* fails */
assert(pme[2] & PAGEMAP_SOFTDIRTY); /* passes */
return 0;
}
(Note that all pages in new VMAs are softdirty until cleared).
Tested:
Used the program given above. I'm going to include this code in
a selftest in the future.
Signed-off-by: Peter Feiner <pfeiner@google.com>
---
fs/proc/task_mmu.c | 61 +++++++++++++++++++++++++++++++++++-------------------
1 file changed, 40 insertions(+), 21 deletions(-)
diff --git a/fs/proc/task_mmu.c b/fs/proc/task_mmu.c
index dfc791c..eb92a8c 100644
--- a/fs/proc/task_mmu.c
+++ b/fs/proc/task_mmu.c
@@ -1020,7 +1020,6 @@ static int pagemap_pte_range(pmd_t *pmd, unsigned long addr, unsigned long end,
spinlock_t *ptl;
pte_t *pte;
int err = 0;
- pagemap_entry_t pme = make_pme(PM_NOT_PRESENT(pm->v2));
/* find the first VMA at or above 'addr' */
vma = find_vma(walk->mm, addr);
@@ -1034,6 +1033,7 @@ static int pagemap_pte_range(pmd_t *pmd, unsigned long addr, unsigned long end,
for (; addr != end; addr += PAGE_SIZE) {
unsigned long offset;
+ pagemap_entry_t pme;
offset = (addr & ~PAGEMAP_WALK_MASK) >>
PAGE_SHIFT;
@@ -1048,32 +1048,51 @@ static int pagemap_pte_range(pmd_t *pmd, unsigned long addr, unsigned long end,
if (pmd_trans_unstable(pmd))
return 0;
- for (; addr != end; addr += PAGE_SIZE) {
- int flags2;
-
- /* check to see if we've left 'vma' behind
- * and need a new, higher one */
- if (vma && (addr >= vma->vm_end)) {
- vma = find_vma(walk->mm, addr);
- if (vma && (vma->vm_flags & VM_SOFTDIRTY))
- flags2 = __PM_SOFT_DIRTY;
- else
- flags2 = 0;
- pme = make_pme(PM_NOT_PRESENT(pm->v2) | PM_STATUS2(pm->v2, flags2));
+
+ while (1) {
+ unsigned long vm_start = end;
+ unsigned long vm_end = end;
+ unsigned long vm_flags = 0;
+
+ if (vma) {
+ /*
+ * We can't possibly be in a hugetlb VMA. In general,
+ * for a mm_walk with a pmd_entry and a hugetlb_entry,
+ * the pmd_entry can only be called on addresses in a
+ * hugetlb if the walk starts in a non-hugetlb VMA and
+ * spans a hugepage VMA. Since pagemap_read walks are
+ * PMD-sized and PMD-aligned, this will never be true.
+ */
+ BUG_ON(is_vm_hugetlb_page(vma));
+ vm_start = vma->vm_start;
+ vm_end = min(end, vma->vm_end);
+ vm_flags = vma->vm_flags;
+ }
+
+ /* Addresses before the VMA. */
+ for (; addr < vm_start; addr += PAGE_SIZE) {
+ pagemap_entry_t pme = make_pme(PM_NOT_PRESENT(pm->v2));
+
+ err = add_to_pagemap(addr, &pme, pm);
+ if (err)
+ return err;
}
- /* check that 'vma' actually covers this address,
- * and that it isn't a huge page vma */
- if (vma && (vma->vm_start <= addr) &&
- !is_vm_hugetlb_page(vma)) {
+ /* Addresses in the VMA. */
+ for (; addr < vm_end; addr += PAGE_SIZE) {
+ pagemap_entry_t pme;
pte = pte_offset_map(pmd, addr);
pte_to_pagemap_entry(&pme, pm, vma, addr, *pte);
- /* unmap before userspace copy */
pte_unmap(pte);
+ err = add_to_pagemap(addr, &pme, pm);
+ if (err)
+ return err;
}
- err = add_to_pagemap(addr, &pme, pm);
- if (err)
- return err;
+
+ if (addr == end)
+ break;
+
+ vma = find_vma(walk->mm, addr);
}
cond_resched();
--
2.1.0.rc2.206.gedb03e5
^ permalink raw reply related [flat|nested] 7+ messages in thread
* Re: [PATCH] mm: softdirty: unmapped addresses between VMAs are clean
2014-09-10 23:24 [PATCH] mm: softdirty: unmapped addresses between VMAs are clean Peter Feiner
@ 2014-09-10 23:36 ` Andrew Morton
2014-09-11 5:41 ` Peter Feiner
2014-09-15 18:40 ` [PATCH v2] " Peter Feiner
1 sibling, 1 reply; 7+ messages in thread
From: Andrew Morton @ 2014-09-10 23:36 UTC (permalink / raw)
To: Peter Feiner
Cc: linux-mm, linux-kernel, Kirill A. Shutemov, Cyrill Gorcunov,
Pavel Emelyanov, Jamie Liu, Hugh Dickins, Naoya Horiguchi
On Wed, 10 Sep 2014 16:24:46 -0700 Peter Feiner <pfeiner@google.com> wrote:
> If a /proc/pid/pagemap read spans a [VMA, an unmapped region, then a
> VM_SOFTDIRTY VMA], the virtual pages in the unmapped region are reported
> as softdirty. Here's a program to demonstrate the bug:
>
> int main() {
> const uint64_t PAGEMAP_SOFTDIRTY = 1ul << 55;
> uint64_t pme[3];
> int fd = open("/proc/self/pagemap", O_RDONLY);;
> char *m = mmap(NULL, 3 * getpagesize(), PROT_READ,
> MAP_ANONYMOUS | MAP_SHARED, -1, 0);
> munmap(m + getpagesize(), getpagesize());
> pread(fd, pme, 24, (unsigned long) m / getpagesize() * 8);
> assert(pme[0] & PAGEMAP_SOFTDIRTY); /* passes */
> assert(!(pme[1] & PAGEMAP_SOFTDIRTY)); /* fails */
> assert(pme[2] & PAGEMAP_SOFTDIRTY); /* passes */
> return 0;
> }
>
> (Note that all pages in new VMAs are softdirty until cleared).
>
> Tested:
> Used the program given above. I'm going to include this code in
> a selftest in the future.
>
> ...
>
> --- a/fs/proc/task_mmu.c
> +++ b/fs/proc/task_mmu.c
>
> ...
>
> @@ -1048,32 +1048,51 @@ static int pagemap_pte_range(pmd_t *pmd, unsigned long addr, unsigned long end,
>
> if (pmd_trans_unstable(pmd))
> return 0;
> - for (; addr != end; addr += PAGE_SIZE) {
> - int flags2;
> -
> - /* check to see if we've left 'vma' behind
> - * and need a new, higher one */
> - if (vma && (addr >= vma->vm_end)) {
> - vma = find_vma(walk->mm, addr);
> - if (vma && (vma->vm_flags & VM_SOFTDIRTY))
> - flags2 = __PM_SOFT_DIRTY;
> - else
> - flags2 = 0;
> - pme = make_pme(PM_NOT_PRESENT(pm->v2) | PM_STATUS2(pm->v2, flags2));
> +
> + while (1) {
> + unsigned long vm_start = end;
Did you really mean to do that? If so, perhaps a little comment to
explain how it works?
> + unsigned long vm_end = end;
> + unsigned long vm_flags = 0;
> +
> + if (vma) {
> + /*
> + * We can't possibly be in a hugetlb VMA. In general,
> + * for a mm_walk with a pmd_entry and a hugetlb_entry,
> + * the pmd_entry can only be called on addresses in a
> + * hugetlb if the walk starts in a non-hugetlb VMA and
> + * spans a hugepage VMA. Since pagemap_read walks are
> + * PMD-sized and PMD-aligned, this will never be true.
> + */
> + BUG_ON(is_vm_hugetlb_page(vma));
> + vm_start = vma->vm_start;
> + vm_end = min(end, vma->vm_end);
> + vm_flags = vma->vm_flags;
> + }
> +
> + /* Addresses before the VMA. */
> + for (; addr < vm_start; addr += PAGE_SIZE) {
> + pagemap_entry_t pme = make_pme(PM_NOT_PRESENT(pm->v2));
> +
> + err = add_to_pagemap(addr, &pme, pm);
> + if (err)
> + return err;
>
> ...
>
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH] mm: softdirty: unmapped addresses between VMAs are clean
2014-09-10 23:36 ` Andrew Morton
@ 2014-09-11 5:41 ` Peter Feiner
2014-09-11 19:54 ` Andrew Morton
0 siblings, 1 reply; 7+ messages in thread
From: Peter Feiner @ 2014-09-11 5:41 UTC (permalink / raw)
To: Andrew Morton
Cc: linux-mm, linux-kernel, Kirill A. Shutemov, Cyrill Gorcunov,
Pavel Emelyanov, Jamie Liu, Hugh Dickins, Naoya Horiguchi
On Wed, Sep 10, 2014 at 04:36:28PM -0700, Andrew Morton wrote:
> On Wed, 10 Sep 2014 16:24:46 -0700 Peter Feiner <pfeiner@google.com> wrote:
> > @@ -1048,32 +1048,51 @@ static int pagemap_pte_range(pmd_t *pmd, unsigned long addr, unsigned long end,
> > + while (1) {
> > + unsigned long vm_start = end;
>
> Did you really mean to do that? If so, perhaps a little comment to
> explain how it works?
It's the same idea that I used in the pagemap_pte_hole patch that I submitted
today: if vma is NULL, then we fill in the pagemap from (addr) to (end) with
non-present pagemap entries.
Should I submit a v2 with a comment?
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH] mm: softdirty: unmapped addresses between VMAs are clean
2014-09-11 5:41 ` Peter Feiner
@ 2014-09-11 19:54 ` Andrew Morton
0 siblings, 0 replies; 7+ messages in thread
From: Andrew Morton @ 2014-09-11 19:54 UTC (permalink / raw)
To: Peter Feiner
Cc: linux-mm, linux-kernel, Kirill A. Shutemov, Cyrill Gorcunov,
Pavel Emelyanov, Jamie Liu, Hugh Dickins, Naoya Horiguchi
On Wed, 10 Sep 2014 22:41:04 -0700 Peter Feiner <pfeiner@google.com> wrote:
> On Wed, Sep 10, 2014 at 04:36:28PM -0700, Andrew Morton wrote:
> > On Wed, 10 Sep 2014 16:24:46 -0700 Peter Feiner <pfeiner@google.com> wrote:
> > > @@ -1048,32 +1048,51 @@ static int pagemap_pte_range(pmd_t *pmd, unsigned long addr, unsigned long end,
> > > + while (1) {
> > > + unsigned long vm_start = end;
> >
> > Did you really mean to do that? If so, perhaps a little comment to
> > explain how it works?
>
> It's the same idea that I used in the pagemap_pte_hole patch that I submitted
> today: if vma is NULL, then we fill in the pagemap from (addr) to (end) with
> non-present pagemap entries.
>
> Should I submit a v2 with a comment?
I spent quite some time staring at that code wondering wtf, so anything
you can do to clarify it would be good.
I think a better name would be plain old "start", to communicate that
it's just a local convenience variable. "vm_start" means "start of a
vma" and that isn't accurate in this context; in fact it is misleading.
^ permalink raw reply [flat|nested] 7+ messages in thread
* [PATCH v2] mm: softdirty: unmapped addresses between VMAs are clean
2014-09-10 23:24 [PATCH] mm: softdirty: unmapped addresses between VMAs are clean Peter Feiner
2014-09-10 23:36 ` Andrew Morton
@ 2014-09-15 18:40 ` Peter Feiner
2014-09-26 20:33 ` Naoya Horiguchi
1 sibling, 1 reply; 7+ messages in thread
From: Peter Feiner @ 2014-09-15 18:40 UTC (permalink / raw)
To: linux-mm
Cc: linux-kernel, Peter Feiner, Kirill A. Shutemov, Cyrill Gorcunov,
Pavel Emelyanov, Jamie Liu, Hugh Dickins, Naoya Horiguchi,
Andrew Morton
If a /proc/pid/pagemap read spans a [VMA, an unmapped region, then a
VM_SOFTDIRTY VMA], the virtual pages in the unmapped region are reported
as softdirty. Here's a program to demonstrate the bug:
int main() {
const uint64_t PAGEMAP_SOFTDIRTY = 1ul << 55;
uint64_t pme[3];
int fd = open("/proc/self/pagemap", O_RDONLY);;
char *m = mmap(NULL, 3 * getpagesize(), PROT_READ,
MAP_ANONYMOUS | MAP_SHARED, -1, 0);
munmap(m + getpagesize(), getpagesize());
pread(fd, pme, 24, (unsigned long) m / getpagesize() * 8);
assert(pme[0] & PAGEMAP_SOFTDIRTY); /* passes */
assert(!(pme[1] & PAGEMAP_SOFTDIRTY)); /* fails */
assert(pme[2] & PAGEMAP_SOFTDIRTY); /* passes */
return 0;
}
(Note that all pages in new VMAs are softdirty until cleared).
Tested:
Used the program given above. I'm going to include this code in
a selftest in the future.
Signed-off-by: Peter Feiner <pfeiner@google.com>
---
v1 -> v2:
Restructured patch to make logic more clear.
---
fs/proc/task_mmu.c | 61 +++++++++++++++++++++++++++++++++++-------------------
1 file changed, 40 insertions(+), 21 deletions(-)
diff --git a/fs/proc/task_mmu.c b/fs/proc/task_mmu.c
index dfc791c..2abf37b 100644
--- a/fs/proc/task_mmu.c
+++ b/fs/proc/task_mmu.c
@@ -1020,7 +1020,6 @@ static int pagemap_pte_range(pmd_t *pmd, unsigned long addr, unsigned long end,
spinlock_t *ptl;
pte_t *pte;
int err = 0;
- pagemap_entry_t pme = make_pme(PM_NOT_PRESENT(pm->v2));
/* find the first VMA at or above 'addr' */
vma = find_vma(walk->mm, addr);
@@ -1034,6 +1033,7 @@ static int pagemap_pte_range(pmd_t *pmd, unsigned long addr, unsigned long end,
for (; addr != end; addr += PAGE_SIZE) {
unsigned long offset;
+ pagemap_entry_t pme;
offset = (addr & ~PAGEMAP_WALK_MASK) >>
PAGE_SHIFT;
@@ -1048,32 +1048,51 @@ static int pagemap_pte_range(pmd_t *pmd, unsigned long addr, unsigned long end,
if (pmd_trans_unstable(pmd))
return 0;
- for (; addr != end; addr += PAGE_SIZE) {
- int flags2;
-
- /* check to see if we've left 'vma' behind
- * and need a new, higher one */
- if (vma && (addr >= vma->vm_end)) {
- vma = find_vma(walk->mm, addr);
- if (vma && (vma->vm_flags & VM_SOFTDIRTY))
- flags2 = __PM_SOFT_DIRTY;
- else
- flags2 = 0;
- pme = make_pme(PM_NOT_PRESENT(pm->v2) | PM_STATUS2(pm->v2, flags2));
+
+ while (1) {
+ /* End of address space hole, which we mark as non-present. */
+ unsigned long hole_end;
+
+ if (vma)
+ hole_end = min(end, vma->vm_start);
+ else
+ hole_end = end;
+
+ for (; addr < hole_end; addr += PAGE_SIZE) {
+ pagemap_entry_t pme = make_pme(PM_NOT_PRESENT(pm->v2));
+
+ err = add_to_pagemap(addr, &pme, pm);
+ if (err)
+ return err;
}
- /* check that 'vma' actually covers this address,
- * and that it isn't a huge page vma */
- if (vma && (vma->vm_start <= addr) &&
- !is_vm_hugetlb_page(vma)) {
+ if (!vma)
+ break;
+ /*
+ * We can't possibly be in a hugetlb VMA. In general,
+ * for a mm_walk with a pmd_entry and a hugetlb_entry,
+ * the pmd_entry can only be called on addresses in a
+ * hugetlb if the walk starts in a non-hugetlb VMA and
+ * spans a hugepage VMA. Since pagemap_read walks are
+ * PMD-sized and PMD-aligned, this will never be true.
+ */
+ BUG_ON(is_vm_hugetlb_page(vma));
+
+ /* Addresses in the VMA. */
+ for (; addr < min(end, vma->vm_end); addr += PAGE_SIZE) {
+ pagemap_entry_t pme;
pte = pte_offset_map(pmd, addr);
pte_to_pagemap_entry(&pme, pm, vma, addr, *pte);
- /* unmap before userspace copy */
pte_unmap(pte);
+ err = add_to_pagemap(addr, &pme, pm);
+ if (err)
+ return err;
}
- err = add_to_pagemap(addr, &pme, pm);
- if (err)
- return err;
+
+ if (addr == end)
+ break;
+
+ vma = find_vma(walk->mm, addr);
}
cond_resched();
--
2.1.0.rc2.206.gedb03e5
^ permalink raw reply related [flat|nested] 7+ messages in thread
* Re: [PATCH v2] mm: softdirty: unmapped addresses between VMAs are clean
2014-09-15 18:40 ` [PATCH v2] " Peter Feiner
@ 2014-09-26 20:33 ` Naoya Horiguchi
2014-10-02 0:25 ` Peter Feiner
0 siblings, 1 reply; 7+ messages in thread
From: Naoya Horiguchi @ 2014-09-26 20:33 UTC (permalink / raw)
To: Peter Feiner
Cc: linux-mm, linux-kernel, Kirill A. Shutemov, Cyrill Gorcunov,
Pavel Emelyanov, Jamie Liu, Hugh Dickins, Andrew Morton
On Mon, Sep 15, 2014 at 11:40:38AM -0700, Peter Feiner wrote:
> If a /proc/pid/pagemap read spans a [VMA, an unmapped region, then a
> VM_SOFTDIRTY VMA], the virtual pages in the unmapped region are reported
> as softdirty. Here's a program to demonstrate the bug:
>
> int main() {
> const uint64_t PAGEMAP_SOFTDIRTY = 1ul << 55;
> uint64_t pme[3];
> int fd = open("/proc/self/pagemap", O_RDONLY);;
> char *m = mmap(NULL, 3 * getpagesize(), PROT_READ,
> MAP_ANONYMOUS | MAP_SHARED, -1, 0);
> munmap(m + getpagesize(), getpagesize());
> pread(fd, pme, 24, (unsigned long) m / getpagesize() * 8);
> assert(pme[0] & PAGEMAP_SOFTDIRTY); /* passes */
> assert(!(pme[1] & PAGEMAP_SOFTDIRTY)); /* fails */
> assert(pme[2] & PAGEMAP_SOFTDIRTY); /* passes */
> return 0;
> }
>
> (Note that all pages in new VMAs are softdirty until cleared).
>
> Tested:
> Used the program given above. I'm going to include this code in
> a selftest in the future.
>
> Signed-off-by: Peter Feiner <pfeiner@google.com>
I triggered the BUG_ON(is_vm_hugetlb_page(vma)) introduced by this patch,
when I simply read /proc/pid/pagemap of the process using hugetlb.
This BUG_ON looks right itself, but find_vma() can find vmas beyond
the pmd boundary, so checking the overrun is necessary.
Could you test and merge the following change?
---
From: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Date: Fri, 26 Sep 2014 15:57:39 -0400
Subject: [PATCH] pagemap: prevent pagemap_pte_range() from overrunning
When the vm_end address of the last vma just before vma(VM_HUGETLB)
is not aligned to PMD boundary, the while loop in pagemap_pte_range()
gets vma(VM_HUGETLB) and triggers BUG_ON(is_vm_hugetlb_page(vma)).
This patch fixes it by checking the overrun.
Fixes: 62c98294410d ("mm: softdirty: unmapped addresses between VMAs are clean")
Signed-off-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
---
fs/proc/task_mmu.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/fs/proc/task_mmu.c b/fs/proc/task_mmu.c
index 5674675adeae..f2b15da32a7f 100644
--- a/fs/proc/task_mmu.c
+++ b/fs/proc/task_mmu.c
@@ -1083,7 +1083,7 @@ static int pagemap_pte_range(pmd_t *pmd, unsigned long addr, unsigned long end,
return err;
}
- if (!vma)
+ if (!vma || vma->vm_start >= end)
break;
/*
* We can't possibly be in a hugetlb VMA. In general,
--
1.9.3
Thanks,
Naoya Horiguchi
> ---
>
> v1 -> v2:
> Restructured patch to make logic more clear.
> ---
> fs/proc/task_mmu.c | 61 +++++++++++++++++++++++++++++++++++-------------------
> 1 file changed, 40 insertions(+), 21 deletions(-)
>
> diff --git a/fs/proc/task_mmu.c b/fs/proc/task_mmu.c
> index dfc791c..2abf37b 100644
> --- a/fs/proc/task_mmu.c
> +++ b/fs/proc/task_mmu.c
> @@ -1020,7 +1020,6 @@ static int pagemap_pte_range(pmd_t *pmd, unsigned long addr, unsigned long end,
> spinlock_t *ptl;
> pte_t *pte;
> int err = 0;
> - pagemap_entry_t pme = make_pme(PM_NOT_PRESENT(pm->v2));
>
> /* find the first VMA at or above 'addr' */
> vma = find_vma(walk->mm, addr);
> @@ -1034,6 +1033,7 @@ static int pagemap_pte_range(pmd_t *pmd, unsigned long addr, unsigned long end,
>
> for (; addr != end; addr += PAGE_SIZE) {
> unsigned long offset;
> + pagemap_entry_t pme;
>
> offset = (addr & ~PAGEMAP_WALK_MASK) >>
> PAGE_SHIFT;
> @@ -1048,32 +1048,51 @@ static int pagemap_pte_range(pmd_t *pmd, unsigned long addr, unsigned long end,
>
> if (pmd_trans_unstable(pmd))
> return 0;
> - for (; addr != end; addr += PAGE_SIZE) {
> - int flags2;
> -
> - /* check to see if we've left 'vma' behind
> - * and need a new, higher one */
> - if (vma && (addr >= vma->vm_end)) {
> - vma = find_vma(walk->mm, addr);
> - if (vma && (vma->vm_flags & VM_SOFTDIRTY))
> - flags2 = __PM_SOFT_DIRTY;
> - else
> - flags2 = 0;
> - pme = make_pme(PM_NOT_PRESENT(pm->v2) | PM_STATUS2(pm->v2, flags2));
> +
> + while (1) {
> + /* End of address space hole, which we mark as non-present. */
> + unsigned long hole_end;
> +
> + if (vma)
> + hole_end = min(end, vma->vm_start);
> + else
> + hole_end = end;
> +
> + for (; addr < hole_end; addr += PAGE_SIZE) {
> + pagemap_entry_t pme = make_pme(PM_NOT_PRESENT(pm->v2));
> +
> + err = add_to_pagemap(addr, &pme, pm);
> + if (err)
> + return err;
> }
>
> - /* check that 'vma' actually covers this address,
> - * and that it isn't a huge page vma */
> - if (vma && (vma->vm_start <= addr) &&
> - !is_vm_hugetlb_page(vma)) {
> + if (!vma)
> + break;
> + /*
> + * We can't possibly be in a hugetlb VMA. In general,
> + * for a mm_walk with a pmd_entry and a hugetlb_entry,
> + * the pmd_entry can only be called on addresses in a
> + * hugetlb if the walk starts in a non-hugetlb VMA and
> + * spans a hugepage VMA. Since pagemap_read walks are
> + * PMD-sized and PMD-aligned, this will never be true.
> + */
> + BUG_ON(is_vm_hugetlb_page(vma));
> +
> + /* Addresses in the VMA. */
> + for (; addr < min(end, vma->vm_end); addr += PAGE_SIZE) {
> + pagemap_entry_t pme;
> pte = pte_offset_map(pmd, addr);
> pte_to_pagemap_entry(&pme, pm, vma, addr, *pte);
> - /* unmap before userspace copy */
> pte_unmap(pte);
> + err = add_to_pagemap(addr, &pme, pm);
> + if (err)
> + return err;
> }
> - err = add_to_pagemap(addr, &pme, pm);
> - if (err)
> - return err;
> +
> + if (addr == end)
> + break;
> +
> + vma = find_vma(walk->mm, addr);
> }
>
> cond_resched();
> --
> 2.1.0.rc2.206.gedb03e5
>
> --
> 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] 7+ messages in thread
* Re: [PATCH v2] mm: softdirty: unmapped addresses between VMAs are clean
2014-09-26 20:33 ` Naoya Horiguchi
@ 2014-10-02 0:25 ` Peter Feiner
0 siblings, 0 replies; 7+ messages in thread
From: Peter Feiner @ 2014-10-02 0:25 UTC (permalink / raw)
To: Naoya Horiguchi
Cc: linux-mm, linux-kernel, Kirill A. Shutemov, Cyrill Gorcunov,
Pavel Emelyanov, Jamie Liu, Hugh Dickins, Andrew Morton
On Fri, Sep 26, 2014 at 04:33:26PM -0400, Naoya Horiguchi wrote:
> Could you test and merge the following change?
Many apologies for the late reply! Your email was in my spam folder :-( I see
that Andrew has already merged the patch, so we're in good shape!
Thanks for fixing this bug Naoya!
^ permalink raw reply [flat|nested] 7+ messages in thread
end of thread, other threads:[~2014-10-02 0:25 UTC | newest]
Thread overview: 7+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2014-09-10 23:24 [PATCH] mm: softdirty: unmapped addresses between VMAs are clean Peter Feiner
2014-09-10 23:36 ` Andrew Morton
2014-09-11 5:41 ` Peter Feiner
2014-09-11 19:54 ` Andrew Morton
2014-09-15 18:40 ` [PATCH v2] " Peter Feiner
2014-09-26 20:33 ` Naoya Horiguchi
2014-10-02 0:25 ` Peter Feiner
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).