[PATCH 0/3] 3 bugfixes for memory cgroup (2.6.38-rc2)

[PATCH 0/3] 3 bugfixes for memory cgroup (2.6.38-rc2)

[KAMEZAWA Hiroyuki]

Hi, these are 3 bugfix patches for 2.6.38-rc2 + 
 mm-memcontrolc-fix-uninitialized-variable-use-in-mem_cgroup_move_parent.patch
in mm tree.

3 patches are independent from each other but [1/3] patch is for
2.6.36-stable, tree.

[1/3] fix account leak at failure of memsw acconting. (for 2.6.36 stable)
[2/3] check mem_cgroup_disabled() at split fixup (for recent 2.6.38-git)
[3/3] fix race at move_parent() (This depends on m-memcontrolc
      fix-uninitialized-variable-use-in-mem_cgroup_move_parent.patch)

I'll send other fixes for THP/memcg but this is an early cut for obvious bugs.

Thanks,
-Kame

[PATCH 0/3] 3 bugfixes for memory cgroup (2.6.38-rc2)

[KAMEZAWA Hiroyuki]

Hi, these are 3 bugfix patches for 2.6.38-rc2 + 
 mm-memcontrolc-fix-uninitialized-variable-use-in-mem_cgroup_move_parent.patch
in mm tree.

3 patches are independent from each other but [1/3] patch is for
2.6.36-stable, tree.

[1/3] fix account leak at failure of memsw acconting. (for 2.6.36 stable)
[2/3] check mem_cgroup_disabled() at split fixup (for recent 2.6.38-git)
[3/3] fix race at move_parent() (This depends on m-memcontrolc
      fix-uninitialized-variable-use-in-mem_cgroup_move_parent.patch)

I'll send other fixes for THP/memcg but this is an early cut for obvious bugs.

Thanks,
-Kame

--
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/ .
Fight unfair telecom policy in Canada: sign http://dissolvethecrtc.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

[PATCH 1/3] memcg: fix account leak at failure of memsw acconting.

[KAMEZAWA Hiroyuki]

This is required for 2.6.36-stable and later.
==
Commit 4b53433468c87794b523e4683fbd4e8e8aca1f63 removes
a cancel of charge at case: 
memory charge-> success.
mem+swap charge-> failure.

This leaks usage of memory. Fix it.

This patch is required for stable tree since 2.6.36.

CC: stable@kernel.org
Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
---
 mm/memcontrol.c |    1 +
 1 file changed, 1 insertion(+)

Index: linux-2.6.38-rc2/mm/memcontrol.c
===================================================================
--- linux-2.6.38-rc2.orig/mm/memcontrol.c
+++ linux-2.6.38-rc2/mm/memcontrol.c
@@ -1832,6 +1832,7 @@ static int __mem_cgroup_do_charge(struct
 		if (likely(!ret))
 			return CHARGE_OK;
 
+		res_counter_uncharge(&mem->res, csize);
 		mem_over_limit = mem_cgroup_from_res_counter(fail_res, memsw);
 		flags |= MEM_CGROUP_RECLAIM_NOSWAP;
 	} else

[PATCH 2/3] memcg: bugfix check mem_cgroup_disabled() at split fixup

[KAMEZAWA Hiroyuki]

From: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>

mem_cgroup_disabled() should be checked at splitting.
If diabled, no heavy work is necesary.

Acked-by: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Reviewed-by: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
---
 mm/memcontrol.c |    2 ++
 1 file changed, 2 insertions(+)

Index: linux-2.6.38-rc2/mm/memcontrol.c
===================================================================
--- linux-2.6.38-rc2.orig/mm/memcontrol.c
+++ linux-2.6.38-rc2/mm/memcontrol.c
@@ -2145,6 +2145,8 @@ void mem_cgroup_split_huge_fixup(struct 
 	struct page_cgroup *tail_pc = lookup_page_cgroup(tail);
 	unsigned long flags;
 
+	if (mem_cgroup_disabled())
+		return;
 	/*
 	 * We have no races with charge/uncharge but will have races with
 	 * page state accounting.

[PATCH 2/3] memcg: bugfix check mem_cgroup_disabled() at split fixup

[KAMEZAWA Hiroyuki]

From: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>

mem_cgroup_disabled() should be checked at splitting.
If diabled, no heavy work is necesary.

Acked-by: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Reviewed-by: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
---
 mm/memcontrol.c |    2 ++
 1 file changed, 2 insertions(+)

Index: linux-2.6.38-rc2/mm/memcontrol.c
===================================================================
--- linux-2.6.38-rc2.orig/mm/memcontrol.c
+++ linux-2.6.38-rc2/mm/memcontrol.c
@@ -2145,6 +2145,8 @@ void mem_cgroup_split_huge_fixup(struct 
 	struct page_cgroup *tail_pc = lookup_page_cgroup(tail);
 	unsigned long flags;
 
+	if (mem_cgroup_disabled())
+		return;
 	/*
 	 * We have no races with charge/uncharge but will have races with
 	 * page state accounting.

--
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/ .
Fight unfair telecom policy in Canada: sign http://dissolvethecrtc.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

[PATCH 3/3] memcg: fix race at move_parent around compound_order()

[KAMEZAWA Hiroyuki]

Based on 
2.6.38-rc2 + 
 mm-memcontrolc-fix-uninitialized-variable-use-in-mem_cgroup_move_parent.patch
==
From: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>

A fix up mem_cgroup_move_parent() which use compound_order() in
asynchrnous manner. This compound_order() may return unknown value
because we don't take lock. Use PageTransHuge() and HPAGE_SIZE instead of
it.

Also clean up for mem_cgroup_move_parent(). 
 - remove unnecessary initialization of local variable.
 - rename charge_size -> page_size
 - remove unnecessary (wrong) comment.
 - added a comment about THP.

Changelog:
 - fixed page size calculation for avoiding race.

Note:
 Current design take compound_page_lock() in caller of move_account().
 This should be revisited when we implement direct move_task of hugepage
 without splitting.

Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
---
 mm/memcontrol.c |   25 ++++++++++++++++---------
 1 file changed, 16 insertions(+), 9 deletions(-)

Index: linux-2.6.38-rc2/mm/memcontrol.c
===================================================================
--- linux-2.6.38-rc2.orig/mm/memcontrol.c
+++ linux-2.6.38-rc2/mm/memcontrol.c
@@ -2234,7 +2234,12 @@ static int mem_cgroup_move_account(struc
 {
 	int ret = -EINVAL;
 	unsigned long flags;
-
+	/*
+	 * The page is isolated from LRU. So, collapse function
+ 	 * will not handle this page. But page splitting can happen.
+ 	 * Do this check under compound_page_lock(). The caller should
+ 	 * hold it.
+ 	 */
 	if ((charge_size > PAGE_SIZE) && !PageTransHuge(pc->page))
 		return -EBUSY;
 
@@ -2266,7 +2271,7 @@ static int mem_cgroup_move_parent(struct
 	struct cgroup *cg = child->css.cgroup;
 	struct cgroup *pcg = cg->parent;
 	struct mem_cgroup *parent;
-	int charge = PAGE_SIZE;
+	int page_size = PAGE_SIZE;
 	unsigned long flags;
 	int ret;
 
@@ -2279,22 +2284,24 @@ static int mem_cgroup_move_parent(struct
 		goto out;
 	if (isolate_lru_page(page))
 		goto put;
-	/* The page is isolated from LRU and we have no race with splitting */
-	charge = PAGE_SIZE << compound_order(page);
+
+	if (PageTransHuge(page))
+		page_size = HPAGE_SIZE;
 
 	parent = mem_cgroup_from_cont(pcg);
-	ret = __mem_cgroup_try_charge(NULL, gfp_mask, &parent, false, charge);
+	ret = __mem_cgroup_try_charge(NULL, gfp_mask,
+				&parent, false, page_size);
 	if (ret || !parent)
 		goto put_back;
 
-	if (charge > PAGE_SIZE)
+	if (page_size > PAGE_SIZE)
 		flags = compound_lock_irqsave(page);
 
-	ret = mem_cgroup_move_account(pc, child, parent, true, charge);
+	ret = mem_cgroup_move_account(pc, child, parent, true, page_size);
 	if (ret)
-		mem_cgroup_cancel_charge(parent, charge);
+		mem_cgroup_cancel_charge(parent, page_size);
 
-	if (charge > PAGE_SIZE)
+	if (page_size > PAGE_SIZE)
 		compound_unlock_irqrestore(page, flags);
 put_back:
 	putback_lru_page(page);

[PATCH 3/3] memcg: fix race at move_parent around compound_order()

[KAMEZAWA Hiroyuki]

Based on 
2.6.38-rc2 + 
 mm-memcontrolc-fix-uninitialized-variable-use-in-mem_cgroup_move_parent.patch
==
From: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>

A fix up mem_cgroup_move_parent() which use compound_order() in
asynchrnous manner. This compound_order() may return unknown value
because we don't take lock. Use PageTransHuge() and HPAGE_SIZE instead of
it.

Also clean up for mem_cgroup_move_parent(). 
 - remove unnecessary initialization of local variable.
 - rename charge_size -> page_size
 - remove unnecessary (wrong) comment.
 - added a comment about THP.

Changelog:
 - fixed page size calculation for avoiding race.

Note:
 Current design take compound_page_lock() in caller of move_account().
 This should be revisited when we implement direct move_task of hugepage
 without splitting.

Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
---
 mm/memcontrol.c |   25 ++++++++++++++++---------
 1 file changed, 16 insertions(+), 9 deletions(-)

Index: linux-2.6.38-rc2/mm/memcontrol.c
===================================================================
--- linux-2.6.38-rc2.orig/mm/memcontrol.c
+++ linux-2.6.38-rc2/mm/memcontrol.c
@@ -2234,7 +2234,12 @@ static int mem_cgroup_move_account(struc
 {
 	int ret = -EINVAL;
 	unsigned long flags;
-
+	/*
+	 * The page is isolated from LRU. So, collapse function
+ 	 * will not handle this page. But page splitting can happen.
+ 	 * Do this check under compound_page_lock(). The caller should
+ 	 * hold it.
+ 	 */
 	if ((charge_size > PAGE_SIZE) && !PageTransHuge(pc->page))
 		return -EBUSY;
 
@@ -2266,7 +2271,7 @@ static int mem_cgroup_move_parent(struct
 	struct cgroup *cg = child->css.cgroup;
 	struct cgroup *pcg = cg->parent;
 	struct mem_cgroup *parent;
-	int charge = PAGE_SIZE;
+	int page_size = PAGE_SIZE;
 	unsigned long flags;
 	int ret;
 
@@ -2279,22 +2284,24 @@ static int mem_cgroup_move_parent(struct
 		goto out;
 	if (isolate_lru_page(page))
 		goto put;
-	/* The page is isolated from LRU and we have no race with splitting */
-	charge = PAGE_SIZE << compound_order(page);
+
+	if (PageTransHuge(page))
+		page_size = HPAGE_SIZE;
 
 	parent = mem_cgroup_from_cont(pcg);
-	ret = __mem_cgroup_try_charge(NULL, gfp_mask, &parent, false, charge);
+	ret = __mem_cgroup_try_charge(NULL, gfp_mask,
+				&parent, false, page_size);
 	if (ret || !parent)
 		goto put_back;
 
-	if (charge > PAGE_SIZE)
+	if (page_size > PAGE_SIZE)
 		flags = compound_lock_irqsave(page);
 
-	ret = mem_cgroup_move_account(pc, child, parent, true, charge);
+	ret = mem_cgroup_move_account(pc, child, parent, true, page_size);
 	if (ret)
-		mem_cgroup_cancel_charge(parent, charge);
+		mem_cgroup_cancel_charge(parent, page_size);
 
-	if (charge > PAGE_SIZE)
+	if (page_size > PAGE_SIZE)
 		compound_unlock_irqrestore(page, flags);
 put_back:
 	putback_lru_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/ .
Fight unfair telecom policy in Canada: sign http://dissolvethecrtc.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: [PATCH 1/3] memcg: fix account leak at failure of memsw acconting.

[Johannes Weiner]

On Tue, Jan 25, 2011 at 03:00:42PM +0900, KAMEZAWA Hiroyuki wrote:
> This is required for 2.6.36-stable and later.
> ==
> Commit 4b53433468c87794b523e4683fbd4e8e8aca1f63 removes
> a cancel of charge at case: 
> memory charge-> success.
> mem+swap charge-> failure.
> 
> This leaks usage of memory. Fix it.
> 
> This patch is required for stable tree since 2.6.36.
> 
> CC: stable@kernel.org
> Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>

Reviewed-by: Johannes Weiner <hannes@cmpxchg.org>

Re: [PATCH 3/3] memcg: fix race at move_parent around compound_order()

[Johannes Weiner]

On Tue, Jan 25, 2011 at 03:05:16PM +0900, KAMEZAWA Hiroyuki wrote:
> Based on 
> 2.6.38-rc2 + 
>  mm-memcontrolc-fix-uninitialized-variable-use-in-mem_cgroup_move_parent.patch
> ==
> From: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
> 
> A fix up mem_cgroup_move_parent() which use compound_order() in
> asynchrnous manner. This compound_order() may return unknown value
> because we don't take lock. Use PageTransHuge() and HPAGE_SIZE instead of
> it.
> 
> Also clean up for mem_cgroup_move_parent(). 
>  - remove unnecessary initialization of local variable.
>  - rename charge_size -> page_size
>  - remove unnecessary (wrong) comment.
>  - added a comment about THP.
> 
> Changelog:
>  - fixed page size calculation for avoiding race.
> 
> Note:
>  Current design take compound_page_lock() in caller of move_account().
>  This should be revisited when we implement direct move_task of hugepage
>  without splitting.
> 
> Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>

Reviewed-by: Johannes Weiner <hannes@cmpxchg.org>

Re: [PATCH 3/3] memcg: fix race at move_parent around compound_order()

[Johannes Weiner]

On Tue, Jan 25, 2011 at 03:05:16PM +0900, KAMEZAWA Hiroyuki wrote:
> Based on 
> 2.6.38-rc2 + 
>  mm-memcontrolc-fix-uninitialized-variable-use-in-mem_cgroup_move_parent.patch
> ==
> From: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
> 
> A fix up mem_cgroup_move_parent() which use compound_order() in
> asynchrnous manner. This compound_order() may return unknown value
> because we don't take lock. Use PageTransHuge() and HPAGE_SIZE instead of
> it.
> 
> Also clean up for mem_cgroup_move_parent(). 
>  - remove unnecessary initialization of local variable.
>  - rename charge_size -> page_size
>  - remove unnecessary (wrong) comment.
>  - added a comment about THP.
> 
> Changelog:
>  - fixed page size calculation for avoiding race.
> 
> Note:
>  Current design take compound_page_lock() in caller of move_account().
>  This should be revisited when we implement direct move_task of hugepage
>  without splitting.
> 
> Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>

Reviewed-by: Johannes Weiner <hannes@cmpxchg.org>

--
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/ .
Fight unfair telecom policy in Canada: sign http://dissolvethecrtc.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: [PATCH 1/3] memcg: fix account leak at failure of memsw acconting.

[Daisuke Nishimura]

On Tue, 25 Jan 2011 15:00:42 +0900
KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> wrote:

> This is required for 2.6.36-stable and later.
> ==
> Commit 4b53433468c87794b523e4683fbd4e8e8aca1f63 removes
> a cancel of charge at case: 
> memory charge-> success.
> mem+swap charge-> failure.
> 
> This leaks usage of memory. Fix it.
> 
> This patch is required for stable tree since 2.6.36.
> 
> CC: stable@kernel.org
> Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>

Nice catch.

Acked-by: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>

Re: [PATCH 3/3] memcg: fix race at move_parent around compound_order()

[Daisuke Nishimura]

On Tue, 25 Jan 2011 15:05:16 +0900
KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> wrote:

> Based on 
> 2.6.38-rc2 + 
>  mm-memcontrolc-fix-uninitialized-variable-use-in-mem_cgroup_move_parent.patch
> ==
> From: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
> 
> A fix up mem_cgroup_move_parent() which use compound_order() in
> asynchrnous manner. This compound_order() may return unknown value
> because we don't take lock. Use PageTransHuge() and HPAGE_SIZE instead of
> it.
> 
> Also clean up for mem_cgroup_move_parent(). 
>  - remove unnecessary initialization of local variable.
>  - rename charge_size -> page_size
>  - remove unnecessary (wrong) comment.
>  - added a comment about THP.
> 
> Changelog:
>  - fixed page size calculation for avoiding race.
> 
> Note:
>  Current design take compound_page_lock() in caller of move_account().
>  This should be revisited when we implement direct move_task of hugepage
>  without splitting.
> 
> Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>

Acked-by: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>

Re: [PATCH 3/3] memcg: fix race at move_parent around compound_order()

[Daisuke Nishimura]

On Tue, 25 Jan 2011 15:05:16 +0900
KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> wrote:

> Based on 
> 2.6.38-rc2 + 
>  mm-memcontrolc-fix-uninitialized-variable-use-in-mem_cgroup_move_parent.patch
> ==
> From: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
> 
> A fix up mem_cgroup_move_parent() which use compound_order() in
> asynchrnous manner. This compound_order() may return unknown value
> because we don't take lock. Use PageTransHuge() and HPAGE_SIZE instead of
> it.
> 
> Also clean up for mem_cgroup_move_parent(). 
>  - remove unnecessary initialization of local variable.
>  - rename charge_size -> page_size
>  - remove unnecessary (wrong) comment.
>  - added a comment about THP.
> 
> Changelog:
>  - fixed page size calculation for avoiding race.
> 
> Note:
>  Current design take compound_page_lock() in caller of move_account().
>  This should be revisited when we implement direct move_task of hugepage
>  without splitting.
> 
> Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>

Acked-by: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>

--
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/ .
Fight unfair telecom policy in Canada: sign http://dissolvethecrtc.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>