linux-next: manual merge of the akpm tree with the nfs tree

linux-next: manual merge of the akpm tree with the nfs tree

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 3141 bytes --]

Hi Andrew,

Today's linux-next merge of the akpm tree got a conflict in
net/sunrpc/xprtsock.c between commit 5cf02d09b50b ("nfs: skip commit in
releasepage if we're freeing memory for fs-related reasons") from the nfs
tree and commit "nfs: enable swap on NFS" from the akpm tree.

Just context changes?  I fixed it up (I think - see below) and can carry
the fix as necessary.
-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au

diff --cc net/sunrpc/xprtsock.c
index 9266794,83bb0eb..0000000
--- a/net/sunrpc/xprtsock.c
+++ b/net/sunrpc/xprtsock.c
@@@ -1927,9 -1925,47 +1927,48 @@@ static void xs_local_setup_socket(struc
  out:
  	xprt_clear_connecting(xprt);
  	xprt_wake_pending_tasks(xprt, status);
 +	current->flags &= ~PF_FSTRANS;
  }
  
+ #ifdef CONFIG_SUNRPC_SWAP
+ static void xs_set_memalloc(struct rpc_xprt *xprt)
+ {
+ 	struct sock_xprt *transport = container_of(xprt, struct sock_xprt,
+ 			xprt);
+ 
+ 	if (xprt->swapper)
+ 		sk_set_memalloc(transport->inet);
+ }
+ 
+ /**
+  * xs_swapper - Tag this transport as being used for swap.
+  * @xprt: transport to tag
+  * @enable: enable/disable
+  *
+  */
+ int xs_swapper(struct rpc_xprt *xprt, int enable)
+ {
+ 	struct sock_xprt *transport = container_of(xprt, struct sock_xprt,
+ 			xprt);
+ 	int err = 0;
+ 
+ 	if (enable) {
+ 		xprt->swapper++;
+ 		xs_set_memalloc(xprt);
+ 	} else if (xprt->swapper) {
+ 		xprt->swapper--;
+ 		sk_clear_memalloc(transport->inet);
+ 	}
+ 
+ 	return err;
+ }
+ EXPORT_SYMBOL_GPL(xs_swapper);
+ #else
+ static void xs_set_memalloc(struct rpc_xprt *xprt)
+ {
+ }
+ #endif
+ 
  static void xs_udp_finish_connecting(struct rpc_xprt *xprt, struct socket *sock)
  {
  	struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt);
@@@ -1970,7 -2009,8 +2012,9 @@@ static void xs_udp_setup_socket(struct 
  	if (xprt->shutdown)
  		goto out;
  
 +	current->flags |= PF_FSTRANS;
+ 	if (xprt->swapper)
+ 		current->flags |= PF_MEMALLOC;
  
  	/* Start by resetting any existing state */
  	xs_reset_transport(transport);
@@@ -1990,7 -2030,7 +2034,8 @@@
  out:
  	xprt_clear_connecting(xprt);
  	xprt_wake_pending_tasks(xprt, status);
+ 	tsk_restore_flags(current, pflags, PF_MEMALLOC);
 +	current->flags &= ~PF_FSTRANS;
  }
  
  /*
@@@ -2116,7 -2159,8 +2164,9 @@@ static void xs_tcp_setup_socket(struct 
  	if (xprt->shutdown)
  		goto out;
  
 +	current->flags |= PF_FSTRANS;
+ 	if (xprt->swapper)
+ 		current->flags |= PF_MEMALLOC;
  
  	if (!sock) {
  		clear_bit(XPRT_CONNECTION_ABORT, &xprt->state);
@@@ -2167,7 -2211,7 +2217,8 @@@
  	case -EINPROGRESS:
  	case -EALREADY:
  		xprt_clear_connecting(xprt);
+ 		tsk_restore_flags(current, pflags, PF_MEMALLOC);
 +		current->flags &= ~PF_FSTRANS;
  		return;
  	case -EINVAL:
  		/* Happens, for instance, if the user specified a link
@@@ -2180,7 -2224,7 +2231,8 @@@ out_eagain
  out:
  	xprt_clear_connecting(xprt);
  	xprt_wake_pending_tasks(xprt, status);
+ 	tsk_restore_flags(current, pflags, PF_MEMALLOC);
 +	current->flags &= ~PF_FSTRANS;
  }
  
  /**

[-- Attachment #2: Type: application/pgp-signature, Size: 836 bytes --]

Re: linux-next: manual merge of the akpm tree with the nfs tree

[Mel Gorman]

On Tue, Jul 31, 2012 at 02:24:41PM +1000, Stephen Rothwell wrote:
> Hi Andrew,
> 
> Today's linux-next merge of the akpm tree got a conflict in
> net/sunrpc/xprtsock.c between commit 5cf02d09b50b ("nfs: skip commit in
> releasepage if we're freeing memory for fs-related reasons") from the nfs
> tree and commit "nfs: enable swap on NFS" from the akpm tree.
> 
> Just context changes?  I fixed it up (I think - see below) and can carry
> the fix as necessary.

Functionally it looks fine. As you say, it all looks like context
changes. Arguably code like this

current->flags &= ~PF_FSTRANS

could use tsk_restore_flags instead() even though it should never be
necessary as PF_FSTRANS would not be set on function entry. However,
it would set up a depedency between the patch sets that is undesirable.
If both sets get merged then it might make sense as a cleanup to use
tsk_restore_flags() but not until then.

Thanks Stephen.

-- 
Mel Gorman
SUSE Labs

Re: linux-next: manual merge of the akpm tree with the nfs tree

[Myklebust, Trond]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset="utf-8", Size: 1776 bytes --]

On Tue, 2012-07-31 at 11:33 +0100, Mel Gorman wrote:
> On Tue, Jul 31, 2012 at 02:24:41PM +1000, Stephen Rothwell wrote:
> > Hi Andrew,
> > 
> > Today's linux-next merge of the akpm tree got a conflict in
> > net/sunrpc/xprtsock.c between commit 5cf02d09b50b ("nfs: skip commit in
> > releasepage if we're freeing memory for fs-related reasons") from the nfs
> > tree and commit "nfs: enable swap on NFS" from the akpm tree.
> > 
> > Just context changes?  I fixed it up (I think - see below) and can carry
> > the fix as necessary.
> 
> Functionally it looks fine. As you say, it all looks like context
> changes. Arguably code like this
> 
> current->flags &= ~PF_FSTRANS
> 
> could use tsk_restore_flags instead() even though it should never be
> necessary as PF_FSTRANS would not be set on function entry. However,
> it would set up a depedency between the patch sets that is undesirable.
> If both sets get merged then it might make sense as a cleanup to use
> tsk_restore_flags() but not until then.
> 
> Thanks Stephen.
> 

Do we really need to set both PF_FSTRANS and PF_MEMALLOC here? The
reason why I merged the PF_FSTRANS patch is that we have the deadlock
problem when allocating a new socket even before we add swap-over-nfs.
Adding PF_FSTRANS to disallow entry into the NFS layer by the memory
allocator fixes that issue.
What value does PF_MEMALLOC add? Is that in order to prevent recursion
into other areas of the swap code (say, if you mix swap-over-nfs with
ordinary swap-to-disk)?

Cheers
  Trond
-- 
Trond Myklebust
Linux NFS client maintainer

NetApp
Trond.Myklebust@netapp.com
www.netapp.com

ÿôèº{.nÇ+‰·Ÿ®‰­†+%ŠËÿ±éݶ\x17¥Šwÿº{.nÇ+‰·¥Š{±þG«éÿŠ{ayº\x1dʇڙë,j\a­¢f£¢·hšïêÿ‘êçz_è®\x03(­éšŽŠÝ¢j"ú\x1a¶^[m§ÿÿ¾\a«þG«éÿ¢¸?™¨è­Ú&£ø§~á¶iO•æ¬z·švØ^\x14\x04\x1a¶^[m§ÿÿÃ\fÿ¶ìÿ¢¸?–I¥

Re: linux-next: manual merge of the akpm tree with the nfs tree

[Mel Gorman]

On Tue, Jul 31, 2012 at 02:37:24PM +0000, Myklebust, Trond wrote:
> On Tue, 2012-07-31 at 11:33 +0100, Mel Gorman wrote:
> > On Tue, Jul 31, 2012 at 02:24:41PM +1000, Stephen Rothwell wrote:
> > > Hi Andrew,
> > > 
> > > Today's linux-next merge of the akpm tree got a conflict in
> > > net/sunrpc/xprtsock.c between commit 5cf02d09b50b ("nfs: skip commit in
> > > releasepage if we're freeing memory for fs-related reasons") from the nfs
> > > tree and commit "nfs: enable swap on NFS" from the akpm tree.
> > > 
> > > Just context changes?  I fixed it up (I think - see below) and can carry
> > > the fix as necessary.
> > 
> > Functionally it looks fine. As you say, it all looks like context
> > changes. Arguably code like this
> > 
> > current->flags &= ~PF_FSTRANS
> > 
> > could use tsk_restore_flags instead() even though it should never be
> > necessary as PF_FSTRANS would not be set on function entry. However,
> > it would set up a depedency between the patch sets that is undesirable.
> > If both sets get merged then it might make sense as a cleanup to use
> > tsk_restore_flags() but not until then.
> > 
> > Thanks Stephen.
> > 
> 
> Do we really need to set both PF_FSTRANS and PF_MEMALLOC here? The
> reason why I merged the PF_FSTRANS patch is that we have the deadlock
> problem when allocating a new socket even before we add swap-over-nfs.
> Adding PF_FSTRANS to disallow entry into the NFS layer by the memory
> allocator fixes that issue.

PF_FSTRANS is to prevent recursion into NFS and is set whether swap-over-NFS
is used or not and for all requests.

> What value does PF_MEMALLOC add? Is that in order to prevent recursion
> into other areas of the swap code (say, if you mix swap-over-nfs with
> ordinary swap-to-disk)?
> 

PF_MEMALLOC is normally to prevent the page reclaim recursing into
itself. Page reclaim can call the page allocator and that cannot re-enter
page reclaim.

In the case of swap-over-NFS, PF_MEMALLOC is set only if the socket is
being used for swapping. In softirq context, the allocation request is
allowed to use PFMEMALLOC reserves to avoid deadlock.

I do not see an obvious way to collapse the two flags together.
PF_FSTRANS should not mean the PFMEMALLOC reserves can be used and
PFMEMALLOC is not set for all requests.

-- 
Mel Gorman
SUSE Labs

Re: linux-next: manual merge of the akpm tree with the nfs tree

[Myklebust, Trond]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset="utf-8", Size: 3038 bytes --]

On Tue, 2012-07-31 at 16:19 +0100, Mel Gorman wrote:
> On Tue, Jul 31, 2012 at 02:37:24PM +0000, Myklebust, Trond wrote:
> > On Tue, 2012-07-31 at 11:33 +0100, Mel Gorman wrote:
> > > On Tue, Jul 31, 2012 at 02:24:41PM +1000, Stephen Rothwell wrote:
> > > > Hi Andrew,
> > > > 
> > > > Today's linux-next merge of the akpm tree got a conflict in
> > > > net/sunrpc/xprtsock.c between commit 5cf02d09b50b ("nfs: skip commit in
> > > > releasepage if we're freeing memory for fs-related reasons") from the nfs
> > > > tree and commit "nfs: enable swap on NFS" from the akpm tree.
> > > > 
> > > > Just context changes?  I fixed it up (I think - see below) and can carry
> > > > the fix as necessary.
> > > 
> > > Functionally it looks fine. As you say, it all looks like context
> > > changes. Arguably code like this
> > > 
> > > current->flags &= ~PF_FSTRANS
> > > 
> > > could use tsk_restore_flags instead() even though it should never be
> > > necessary as PF_FSTRANS would not be set on function entry. However,
> > > it would set up a depedency between the patch sets that is undesirable.
> > > If both sets get merged then it might make sense as a cleanup to use
> > > tsk_restore_flags() but not until then.
> > > 
> > > Thanks Stephen.
> > > 
> > 
> > Do we really need to set both PF_FSTRANS and PF_MEMALLOC here? The
> > reason why I merged the PF_FSTRANS patch is that we have the deadlock
> > problem when allocating a new socket even before we add swap-over-nfs.
> > Adding PF_FSTRANS to disallow entry into the NFS layer by the memory
> > allocator fixes that issue.
> 
> PF_FSTRANS is to prevent recursion into NFS and is set whether swap-over-NFS
> is used or not and for all requests.
> 
> > What value does PF_MEMALLOC add? Is that in order to prevent recursion
> > into other areas of the swap code (say, if you mix swap-over-nfs with
> > ordinary swap-to-disk)?
> > 
> 
> PF_MEMALLOC is normally to prevent the page reclaim recursing into
> itself. Page reclaim can call the page allocator and that cannot re-enter
> page reclaim.
> 
> In the case of swap-over-NFS, PF_MEMALLOC is set only if the socket is
> being used for swapping. In softirq context, the allocation request is
> allowed to use PFMEMALLOC reserves to avoid deadlock.
> 
> I do not see an obvious way to collapse the two flags together.
> PF_FSTRANS should not mean the PFMEMALLOC reserves can be used and
> PFMEMALLOC is not set for all requests.

Right, but in this case, we're talking about a GFP_KERNEL allocation
that always happens in an rpciod workqueue process context, so we still
won't be able to access the PFMEMALLOC reserves if I understand you
correctly?

I understand the value of preventing the page reclaim recursing into
itself, but in this case, we're talking about a separate process that is
operating on behalf of the allocator (much like kswapd does).

Cheers
  Trond
ÿôèº{.nÇ+‰·Ÿ®‰­†+%ŠËÿ±éݶ\x17¥Šwÿº{.nÇ+‰·¥Š{±þG«éÿŠ{ayº\x1dʇڙë,j\a­¢f£¢·hšïêÿ‘êçz_è®\x03(­éšŽŠÝ¢j"ú\x1a¶^[m§ÿÿ¾\a«þG«éÿ¢¸?™¨è­Ú&£ø§~á¶iO•æ¬z·švØ^\x14\x04\x1a¶^[m§ÿÿÃ\fÿ¶ìÿ¢¸?–I¥

Re: linux-next: manual merge of the akpm tree with the nfs tree

[Mel Gorman]

On Tue, Jul 31, 2012 at 03:35:07PM +0000, Myklebust, Trond wrote:
> On Tue, 2012-07-31 at 16:19 +0100, Mel Gorman wrote:
> > On Tue, Jul 31, 2012 at 02:37:24PM +0000, Myklebust, Trond wrote:
> > > On Tue, 2012-07-31 at 11:33 +0100, Mel Gorman wrote:
> > > > On Tue, Jul 31, 2012 at 02:24:41PM +1000, Stephen Rothwell wrote:
> > > > > Hi Andrew,
> > > > > 
> > > > > Today's linux-next merge of the akpm tree got a conflict in
> > > > > net/sunrpc/xprtsock.c between commit 5cf02d09b50b ("nfs: skip commit in
> > > > > releasepage if we're freeing memory for fs-related reasons") from the nfs
> > > > > tree and commit "nfs: enable swap on NFS" from the akpm tree.
> > > > > 
> > > > > Just context changes?  I fixed it up (I think - see below) and can carry
> > > > > the fix as necessary.
> > > > 
> > > > Functionally it looks fine. As you say, it all looks like context
> > > > changes. Arguably code like this
> > > > 
> > > > current->flags &= ~PF_FSTRANS
> > > > 
> > > > could use tsk_restore_flags instead() even though it should never be
> > > > necessary as PF_FSTRANS would not be set on function entry. However,
> > > > it would set up a depedency between the patch sets that is undesirable.
> > > > If both sets get merged then it might make sense as a cleanup to use
> > > > tsk_restore_flags() but not until then.
> > > > 
> > > > Thanks Stephen.
> > > > 
> > > 
> > > Do we really need to set both PF_FSTRANS and PF_MEMALLOC here? The
> > > reason why I merged the PF_FSTRANS patch is that we have the deadlock
> > > problem when allocating a new socket even before we add swap-over-nfs.
> > > Adding PF_FSTRANS to disallow entry into the NFS layer by the memory
> > > allocator fixes that issue.
> > 
> > PF_FSTRANS is to prevent recursion into NFS and is set whether swap-over-NFS
> > is used or not and for all requests.
> > 
> > > What value does PF_MEMALLOC add? Is that in order to prevent recursion
> > > into other areas of the swap code (say, if you mix swap-over-nfs with
> > > ordinary swap-to-disk)?
> > > 
> > 
> > PF_MEMALLOC is normally to prevent the page reclaim recursing into
> > itself. Page reclaim can call the page allocator and that cannot re-enter
> > page reclaim.
> > 
> > In the case of swap-over-NFS, PF_MEMALLOC is set only if the socket is
> > being used for swapping. In softirq context, the allocation request is
> > allowed to use PFMEMALLOC reserves to avoid deadlock.
> > 
> > I do not see an obvious way to collapse the two flags together.
> > PF_FSTRANS should not mean the PFMEMALLOC reserves can be used and
> > PFMEMALLOC is not set for all requests.
> 
> Right, but in this case, we're talking about a GFP_KERNEL allocation
> that always happens in an rpciod workqueue process context, so we still
> won't be able to access the PFMEMALLOC reserves if I understand you
> correctly?
> 

Ah, I understand you now. The PFMEMALLOC flag only uses the reserves if
running from softirq context. rpciod is never running in that context and
PF_MEMALLOC in that path is counter-productive. Thanks for catching that.

The actual way to resolve this conflict is to alter the "nfs: enable swap on
NFS" patch in Andrew's tree. Andrew, can you apply this patch and collapse
it with "nfs: enable swap on NFS" please?

---8<---
buildfix: nfs: enable swap on NFS

Stephen Rothwell reported a merge conflict between a MM patch "nfs:
enable swap on NFS" and an NFS patch "nfs: skip commit in
releasepage if we're freeing memory for fs-related reasons".

Trond pointed out that at the points of the conflict the running context
is rpciod and not a softirq context. This patch stops PF_MEMALLOC being
set but rpc_malloc still uses __GFP_MEMALLOC where necessary to allocate
from the reserves. When merged with the "nfs: enable swap on NFS", the
conflict with the NFS tree should disappear.

Signed-off-by: Mel Gorman <mgorman@suse.de>
---
 net/sunrpc/xprtsock.c |   11 -----------
 1 file changed, 11 deletions(-)

diff --git a/net/sunrpc/xprtsock.c b/net/sunrpc/xprtsock.c
index 83bb0eb..bd59d01 100644
--- a/net/sunrpc/xprtsock.c
+++ b/net/sunrpc/xprtsock.c
@@ -2003,15 +2003,11 @@ static void xs_udp_setup_socket(struct work_struct *work)
 		container_of(work, struct sock_xprt, connect_worker.work);
 	struct rpc_xprt *xprt = &transport->xprt;
 	struct socket *sock = transport->sock;
-	unsigned long pflags = current->flags;
 	int status = -EIO;
 
 	if (xprt->shutdown)
 		goto out;
 
-	if (xprt->swapper)
-		current->flags |= PF_MEMALLOC;
-
 	/* Start by resetting any existing state */
 	xs_reset_transport(transport);
 	sock = xs_create_sock(xprt, transport,
@@ -2030,7 +2026,6 @@ static void xs_udp_setup_socket(struct work_struct *work)
 out:
 	xprt_clear_connecting(xprt);
 	xprt_wake_pending_tasks(xprt, status);
-	tsk_restore_flags(current, pflags, PF_MEMALLOC);
 }
 
 /*
@@ -2153,15 +2148,11 @@ static void xs_tcp_setup_socket(struct work_struct *work)
 		container_of(work, struct sock_xprt, connect_worker.work);
 	struct socket *sock = transport->sock;
 	struct rpc_xprt *xprt = &transport->xprt;
-	unsigned long pflags = current->flags;
 	int status = -EIO;
 
 	if (xprt->shutdown)
 		goto out;
 
-	if (xprt->swapper)
-		current->flags |= PF_MEMALLOC;
-
 	if (!sock) {
 		clear_bit(XPRT_CONNECTION_ABORT, &xprt->state);
 		sock = xs_create_sock(xprt, transport,
@@ -2211,7 +2202,6 @@ static void xs_tcp_setup_socket(struct work_struct *work)
 	case -EINPROGRESS:
 	case -EALREADY:
 		xprt_clear_connecting(xprt);
-		tsk_restore_flags(current, pflags, PF_MEMALLOC);
 		return;
 	case -EINVAL:
 		/* Happens, for instance, if the user specified a link
@@ -2224,7 +2214,6 @@ out_eagain:
 out:
 	xprt_clear_connecting(xprt);
 	xprt_wake_pending_tasks(xprt, status);
-	tsk_restore_flags(current, pflags, PF_MEMALLOC);
 }
 
 /**

Re: linux-next: manual merge of the akpm tree with the nfs tree

[Andrew Morton]

On Tue, 31 Jul 2012 18:50:22 +0100
Mel Gorman <mgorman@suse.de> wrote:

> Stephen Rothwell reported a merge conflict between a MM patch "nfs:
> enable swap on NFS" and an NFS patch "nfs: skip commit in
> releasepage if we're freeing memory for fs-related reasons".

grumble.  This happened becase new stuff was added to -next right in
the middle of the merge window.  Please don't dothat.

Re: linux-next: manual merge of the akpm tree with the nfs tree

[Myklebust, Trond]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset="utf-8", Size: 879 bytes --]

On Tue, 2012-07-31 at 11:44 -0700, Andrew Morton wrote:
> On Tue, 31 Jul 2012 18:50:22 +0100
> Mel Gorman <mgorman@suse.de> wrote:
> 
> > Stephen Rothwell reported a merge conflict between a MM patch "nfs:
> > enable swap on NFS" and an NFS patch "nfs: skip commit in
> > releasepage if we're freeing memory for fs-related reasons".
> 
> grumble.  This happened becase new stuff was added to -next right in
> the middle of the merge window.  Please don't dothat.
> 
Normally, I would have added it when Jeff asked me to, but I've been on
vacation for 2 weeks, and this is a bugfix that needs to go into
stable...

Trond
-- 
Trond Myklebust
Linux NFS client maintainer

NetApp
Trond.Myklebust@netapp.com
www.netapp.com

ÿôèº{.nÇ+‰·Ÿ®‰­†+%ŠËÿ±éݶ\x17¥Šwÿº{.nÇ+‰·¥Š{±þG«éÿŠ{ayº\x1dʇڙë,j\a­¢f£¢·hšïêÿ‘êçz_è®\x03(­éšŽŠÝ¢j"ú\x1a¶^[m§ÿÿ¾\a«þG«éÿ¢¸?™¨è­Ú&£ø§~á¶iO•æ¬z·švØ^\x14\x04\x1a¶^[m§ÿÿÃ\fÿ¶ìÿ¢¸?–I¥