[PATCH 0/2] nfs41 Fix session reset deadlocks

[PATCH 0/2] nfs41 Fix session reset deadlocks

[andros]

Fix session reset deadlocks. These patches apply to tronds nfs-2.6 bugfix
branch (2.6.32-rc6 on top of these two already posted patches:
nfs41-fix-multiple-free-slot-calls.patch
nfs41-fix-race-on-session-reset.patch

Fix two problems reported by Trond:

0001-nfs41-remove-nfs4_recover_session.patch
0002-nfs41-fix-state-manager-deadlock-in-session-reset.patch

Testing:

CONFIG_NFS_V4_1
v41 mount: Connectathon tests passed. PyNFS testclient.py SESSIONRESET tests
and the new INJECT_ERROR_CTHON test where NFS4ERR_BADSESSION was returned
every 50th SEQUENCE operation durring a Connectathon basic test run. This passed
with the session being reset 390 times from a variety of client compounds
(READ/WRITE/COMMIT/OPEN/CLOSE/ to name a few).

v4 mount: Connectathon tests passed.

no CONFIG_NFS_V4_1
v4 mount: Connectathon tests passed.

-->Andy

[PATCH 1/2] nfs41: remove nfs4_recover_session

[andros]

From: Andy Adamson <andros@netapp.com>

nfs4_recover_session can put rpciod to sleep. Just use nfs4_schedule_recovery.

Reported-by: Trond Myklebust <trond.myklebust@netapp.com>
Signed-off-by: Andy Adamson <andros@netapp.com>
---
 fs/nfs/nfs4proc.c |   26 +++-----------------------
 1 files changed, 3 insertions(+), 23 deletions(-)

diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index 0f1afde..44d1c14 100644
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -427,24 +427,6 @@ out:
 	return ret_id;
 }
 
-static int nfs4_recover_session(struct nfs4_session *session)
-{
-	struct nfs_client *clp = session->clp;
-	unsigned int loop;
-	int ret;
-
-	for (loop = NFS4_MAX_LOOP_ON_RECOVER; loop != 0; loop--) {
-		ret = nfs4_wait_clnt_recover(clp);
-		if (ret != 0)
-			break;
-		if (!test_bit(NFS4CLNT_SESSION_SETUP, &clp->cl_state))
-			break;
-		nfs4_schedule_state_manager(clp);
-		ret = -EIO;
-	}
-	return ret;
-}
-
 static int nfs41_setup_sequence(struct nfs4_session *session,
 				struct nfs4_sequence_args *args,
 				struct nfs4_sequence_res *res,
@@ -453,7 +435,6 @@ static int nfs41_setup_sequence(struct nfs4_session *session,
 {
 	struct nfs4_slot *slot;
 	struct nfs4_slot_table *tbl;
-	int status = 0;
 	u8 slotid;
 
 	dprintk("--> %s\n", __func__);
@@ -476,11 +457,10 @@ static int nfs41_setup_sequence(struct nfs4_session *session,
 
 		/* The slot table is empty; start the reset thread */
 		dprintk("%s Session Reset\n", __func__);
+		rpc_sleep_on(&tbl->slot_tbl_waitq, task, NULL);
+		nfs4_schedule_state_manager(session->clp);
 		spin_unlock(&tbl->slot_tbl_lock);
-		status = nfs4_recover_session(session);
-		if (status)
-			return status;
-		spin_lock(&tbl->slot_tbl_lock);
+		return -EAGAIN;
 	}
 
 	slotid = nfs4_find_slot(tbl, task);
-- 
1.6.0.6

[PATCH 2/2] nfs41: fix state manager deadlock in session reset

[andros]

From: Andy Adamson <andros@netapp.com>

If the session is reset during state recovery, the state manager thread can
sleep on the slot_tbl_waitq causing a deadlock.

Add a completion framework to the session.  Have the state manager thread set
a new session state (NFS4CLNT_SESSION_DRAINING) and wait for the session slot
table to drain.

Signal the state manager thread in nfs41_sequence_free_slot when the
NFS4CLNT_SESSION_DRAINING bit is set and the session is drained.

Reported-by: Trond Myklebust <trond@netapp.com>
Signed-off-by: Andy Adamson <andros@netapp.com>
---
 fs/nfs/nfs4_fs.h          |    1 +
 fs/nfs/nfs4proc.c         |   25 ++++++++++++++++---------
 fs/nfs/nfs4state.c        |   15 +++++++++++++++
 include/linux/nfs_fs_sb.h |    1 +
 4 files changed, 33 insertions(+), 9 deletions(-)

diff --git a/fs/nfs/nfs4_fs.h b/fs/nfs/nfs4_fs.h
index 6ea07a3..e8604af 100644
--- a/fs/nfs/nfs4_fs.h
+++ b/fs/nfs/nfs4_fs.h
@@ -45,6 +45,7 @@ enum nfs4_client_state {
 	NFS4CLNT_RECLAIM_NOGRACE,
 	NFS4CLNT_DELEGRETURN,
 	NFS4CLNT_SESSION_SETUP,
+	NFS4CLNT_SESSION_DRAINING,
 };
 
 /*
diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index 44d1c14..ebde948 100644
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -355,6 +355,16 @@ void nfs41_sequence_free_slot(const struct nfs_client *clp,
 	}
 	nfs4_free_slot(tbl, res->sr_slotid);
 	res->sr_slotid = NFS4_MAX_SLOT_TABLE;
+
+	/* Signal state manager thread if session is drained */
+	if (test_bit(NFS4CLNT_SESSION_DRAINING, &clp->cl_state)) {
+		spin_lock(&tbl->slot_tbl_lock);
+		if (tbl->highest_used_slotid == -1) {
+			dprintk("%s COMPLETE: Session Drained\n", __func__);
+			complete(&clp->cl_session->complete);
+		}
+		spin_unlock(&tbl->slot_tbl_lock);
+	}
 }
 
 static void nfs41_sequence_done(struct nfs_client *clp,
@@ -448,15 +458,11 @@ static int nfs41_setup_sequence(struct nfs4_session *session,
 
 	spin_lock(&tbl->slot_tbl_lock);
 	if (test_bit(NFS4CLNT_SESSION_SETUP, &session->clp->cl_state)) {
-		if (tbl->highest_used_slotid != -1) {
-			rpc_sleep_on(&tbl->slot_tbl_waitq, task, NULL);
-			spin_unlock(&tbl->slot_tbl_lock);
-			dprintk("<-- %s: Session reset: draining\n", __func__);
-			return -EAGAIN;
-		}
-
-		/* The slot table is empty; start the reset thread */
-		dprintk("%s Session Reset\n", __func__);
+		/*
+		 * The state manager will wait until the slot table is empty.
+		 * Schedule the reset thread
+		 */
+		dprintk("%s Schedule Session Reset\n", __func__);
 		rpc_sleep_on(&tbl->slot_tbl_waitq, task, NULL);
 		nfs4_schedule_state_manager(session->clp);
 		spin_unlock(&tbl->slot_tbl_lock);
@@ -4583,6 +4589,7 @@ struct nfs4_session *nfs4_alloc_session(struct nfs_client *clp)
 	 * nfs_client struct
 	 */
 	clp->cl_cons_state = NFS_CS_SESSION_INITING;
+	init_completion(&session->complete);
 
 	tbl = &session->fc_slot_table;
 	spin_lock_init(&tbl->slot_tbl_lock);
diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c
index 50a1b04..f08abe8 100644
--- a/fs/nfs/nfs4state.c
+++ b/fs/nfs/nfs4state.c
@@ -1162,8 +1162,23 @@ static void nfs4_session_recovery_handle_error(struct nfs_client *clp, int err)
 
 static int nfs4_reset_session(struct nfs_client *clp)
 {
+	struct nfs4_session *ses = clp->cl_session;
+	struct nfs4_slot_table *tbl = &ses->fc_slot_table;
 	int status;
 
+	spin_lock(&tbl->slot_tbl_lock);
+	if (tbl->highest_used_slotid != -1) {
+		set_bit(NFS4CLNT_SESSION_DRAINING, &clp->cl_state);
+		spin_unlock(&tbl->slot_tbl_lock);
+		status = wait_for_completion_interruptible(&ses->complete);
+		clear_bit(NFS4CLNT_SESSION_DRAINING, &clp->cl_state);
+		if (status) /* -ERESTARTSYS */
+			goto out;
+		INIT_COMPLETION(ses->complete);
+	} else {
+		spin_unlock(&tbl->slot_tbl_lock);
+	}
+
 	status = nfs4_proc_destroy_session(clp->cl_session);
 	if (status && status != -NFS4ERR_BADSESSION &&
 	    status != -NFS4ERR_DEADSESSION) {
diff --git a/include/linux/nfs_fs_sb.h b/include/linux/nfs_fs_sb.h
index 320569e..34fc6be 100644
--- a/include/linux/nfs_fs_sb.h
+++ b/include/linux/nfs_fs_sb.h
@@ -209,6 +209,7 @@ struct nfs4_session {
 	unsigned long			session_state;
 	u32				hash_alg;
 	u32				ssv_len;
+	struct completion		complete;
 
 	/* The fore and back channel */
 	struct nfs4_channel_attrs	fc_attrs;
-- 
1.6.0.6

Re: [PATCH 2/2] nfs41: fix state manager deadlock in session reset

[William A. (Andy) Adamson]

On Wed, Nov 11, 2009 at 2:20 PM,  <andros@netapp.com> wrote:
> From: Andy Adamson <andros@netapp.com>
>
> If the session is reset during state recovery, the state manager thre=
ad can
> sleep on the slot_tbl_waitq causing a deadlock.
>
> Add a completion framework to the session. =A0Have the state manager =
thread set
> a new session state (NFS4CLNT_SESSION_DRAINING) and wait for the sess=
ion slot
> table to drain.
>
> Signal the state manager thread in nfs41_sequence_free_slot when the
> NFS4CLNT_SESSION_DRAINING bit is set and the session is drained.
>
> Reported-by: Trond Myklebust <trond@netapp.com>
> Signed-off-by: Andy Adamson <andros@netapp.com>
> ---
> =A0fs/nfs/nfs4_fs.h =A0 =A0 =A0 =A0 =A0| =A0 =A01 +
> =A0fs/nfs/nfs4proc.c =A0 =A0 =A0 =A0 | =A0 25 ++++++++++++++++-------=
--
> =A0fs/nfs/nfs4state.c =A0 =A0 =A0 =A0| =A0 15 +++++++++++++++
> =A0include/linux/nfs_fs_sb.h | =A0 =A01 +
> =A04 files changed, 33 insertions(+), 9 deletions(-)
>
> diff --git a/fs/nfs/nfs4_fs.h b/fs/nfs/nfs4_fs.h
> index 6ea07a3..e8604af 100644
> --- a/fs/nfs/nfs4_fs.h
> +++ b/fs/nfs/nfs4_fs.h
> @@ -45,6 +45,7 @@ enum nfs4_client_state {
> =A0 =A0 =A0 =A0NFS4CLNT_RECLAIM_NOGRACE,
> =A0 =A0 =A0 =A0NFS4CLNT_DELEGRETURN,
> =A0 =A0 =A0 =A0NFS4CLNT_SESSION_SETUP,
> + =A0 =A0 =A0 NFS4CLNT_SESSION_DRAINING,
> =A0};
>
> =A0/*
> diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
> index 44d1c14..ebde948 100644
> --- a/fs/nfs/nfs4proc.c
> +++ b/fs/nfs/nfs4proc.c
> @@ -355,6 +355,16 @@ void nfs41_sequence_free_slot(const struct nfs_c=
lient *clp,
> =A0 =A0 =A0 =A0}
> =A0 =A0 =A0 =A0nfs4_free_slot(tbl, res->sr_slotid);
> =A0 =A0 =A0 =A0res->sr_slotid =3D NFS4_MAX_SLOT_TABLE;
> +
> + =A0 =A0 =A0 /* Signal state manager thread if session is drained */
> + =A0 =A0 =A0 if (test_bit(NFS4CLNT_SESSION_DRAINING, &clp->cl_state)=
) {
> + =A0 =A0 =A0 =A0 =A0 =A0 =A0 spin_lock(&tbl->slot_tbl_lock);
> + =A0 =A0 =A0 =A0 =A0 =A0 =A0 if (tbl->highest_used_slotid =3D=3D -1)=
 {
> + =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 dprintk("%s COMPLETE: S=
ession Drained\n", __func__);
> + =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 complete(&clp->cl_sessi=
on->complete);
> + =A0 =A0 =A0 =A0 =A0 =A0 =A0 }
> + =A0 =A0 =A0 =A0 =A0 =A0 =A0 spin_unlock(&tbl->slot_tbl_lock);
> + =A0 =A0 =A0 }
> =A0}
>
> =A0static void nfs41_sequence_done(struct nfs_client *clp,
> @@ -448,15 +458,11 @@ static int nfs41_setup_sequence(struct nfs4_ses=
sion *session,
>
> =A0 =A0 =A0 =A0spin_lock(&tbl->slot_tbl_lock);
> =A0 =A0 =A0 =A0if (test_bit(NFS4CLNT_SESSION_SETUP, &session->clp->cl=
_state)) {
> - =A0 =A0 =A0 =A0 =A0 =A0 =A0 if (tbl->highest_used_slotid !=3D -1) {
> - =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 rpc_sleep_on(&tbl->slot=
_tbl_waitq, task, NULL);
> - =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 spin_unlock(&tbl->slot_=
tbl_lock);
> - =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 dprintk("<-- %s: Sessio=
n reset: draining\n", __func__);
> - =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 return -EAGAIN;
> - =A0 =A0 =A0 =A0 =A0 =A0 =A0 }
> -
> - =A0 =A0 =A0 =A0 =A0 =A0 =A0 /* The slot table is empty; start the r=
eset thread */
> - =A0 =A0 =A0 =A0 =A0 =A0 =A0 dprintk("%s Session Reset\n", __func__)=
;
> + =A0 =A0 =A0 =A0 =A0 =A0 =A0 /*
> + =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0* The state manager will wait until =
the slot table is empty.
> + =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0* Schedule the reset thread
> + =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0*/
> + =A0 =A0 =A0 =A0 =A0 =A0 =A0 dprintk("%s Schedule Session Reset\n", =
__func__);
> =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0rpc_sleep_on(&tbl->slot_tbl_waitq, tas=
k, NULL);
> =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0nfs4_schedule_state_manager(session->c=
lp);
> =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0spin_unlock(&tbl->slot_tbl_lock);
> @@ -4583,6 +4589,7 @@ struct nfs4_session *nfs4_alloc_session(struct =
nfs_client *clp)
> =A0 =A0 =A0 =A0 * nfs_client struct
> =A0 =A0 =A0 =A0 */
> =A0 =A0 =A0 =A0clp->cl_cons_state =3D NFS_CS_SESSION_INITING;
> + =A0 =A0 =A0 init_completion(&session->complete);
>
> =A0 =A0 =A0 =A0tbl =3D &session->fc_slot_table;
> =A0 =A0 =A0 =A0spin_lock_init(&tbl->slot_tbl_lock);
> diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c
> index 50a1b04..f08abe8 100644
> --- a/fs/nfs/nfs4state.c
> +++ b/fs/nfs/nfs4state.c
> @@ -1162,8 +1162,23 @@ static void nfs4_session_recovery_handle_error=
(struct nfs_client *clp, int err)
>
> =A0static int nfs4_reset_session(struct nfs_client *clp)
> =A0{
> + =A0 =A0 =A0 struct nfs4_session *ses =3D clp->cl_session;
> + =A0 =A0 =A0 struct nfs4_slot_table *tbl =3D &ses->fc_slot_table;
> =A0 =A0 =A0 =A0int status;
>
> + =A0 =A0 =A0 spin_lock(&tbl->slot_tbl_lock);
> + =A0 =A0 =A0 if (tbl->highest_used_slotid !=3D -1) {
> + =A0 =A0 =A0 =A0 =A0 =A0 =A0 set_bit(NFS4CLNT_SESSION_DRAINING, &clp=
->cl_state);
> + =A0 =A0 =A0 =A0 =A0 =A0 =A0 spin_unlock(&tbl->slot_tbl_lock);
> + =A0 =A0 =A0 =A0 =A0 =A0 =A0 status =3D wait_for_completion_interrup=
tible(&ses->complete);
> + =A0 =A0 =A0 =A0 =A0 =A0 =A0 clear_bit(NFS4CLNT_SESSION_DRAINING, &c=
lp->cl_state);
> + =A0 =A0 =A0 =A0 =A0 =A0 =A0 if (status) /* -ERESTARTSYS */
> + =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 goto out;
> + =A0 =A0 =A0 =A0 =A0 =A0 =A0 INIT_COMPLETION(ses->complete);


Looking at my own code - I need to move the INIT_COMPLETION after the
wait_for_completion_interruptible and before the status check so that
it gets called even on -ERESTARTSYS.

I'll wait for other comments before sending a fix.

-->Andy


> + =A0 =A0 =A0 } else {
> + =A0 =A0 =A0 =A0 =A0 =A0 =A0 spin_unlock(&tbl->slot_tbl_lock);
> + =A0 =A0 =A0 }
> +
> =A0 =A0 =A0 =A0status =3D nfs4_proc_destroy_session(clp->cl_session);
> =A0 =A0 =A0 =A0if (status && status !=3D -NFS4ERR_BADSESSION &&
> =A0 =A0 =A0 =A0 =A0 =A0status !=3D -NFS4ERR_DEADSESSION) {
> diff --git a/include/linux/nfs_fs_sb.h b/include/linux/nfs_fs_sb.h
> index 320569e..34fc6be 100644
> --- a/include/linux/nfs_fs_sb.h
> +++ b/include/linux/nfs_fs_sb.h
> @@ -209,6 +209,7 @@ struct nfs4_session {
> =A0 =A0 =A0 =A0unsigned long =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 sess=
ion_state;
> =A0 =A0 =A0 =A0u32 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0=
 =A0 hash_alg;
> =A0 =A0 =A0 =A0u32 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0=
 =A0 ssv_len;
> + =A0 =A0 =A0 struct completion =A0 =A0 =A0 =A0 =A0 =A0 =A0 complete;
>
> =A0 =A0 =A0 =A0/* The fore and back channel */
> =A0 =A0 =A0 =A0struct nfs4_channel_attrs =A0 =A0 =A0 fc_attrs;
> --
> 1.6.0.6
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-nfs" =
in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at =A0http://vger.kernel.org/majordomo-info.html
>

Re: [pnfs] [PATCH 2/2] nfs41: fix state manager deadlock in session reset

[Benny Halevy]

On Nov. 11, 2009, 21:51 +0200, "William A. (Andy) Adamson" <androsadamson@gmail.com> wrote:
> On Wed, Nov 11, 2009 at 2:20 PM,  <andros@netapp.com> wrote:
>> From: Andy Adamson <andros@netapp.com>
>>
>> If the session is reset during state recovery, the state manager thread can
>> sleep on the slot_tbl_waitq causing a deadlock.
>>
>> Add a completion framework to the session.  Have the state manager thread set
>> a new session state (NFS4CLNT_SESSION_DRAINING) and wait for the session slot
>> table to drain.
>>
>> Signal the state manager thread in nfs41_sequence_free_slot when the
>> NFS4CLNT_SESSION_DRAINING bit is set and the session is drained.
>>
>> Reported-by: Trond Myklebust <trond@netapp.com>
>> Signed-off-by: Andy Adamson <andros@netapp.com>
>> ---
>>  fs/nfs/nfs4_fs.h          |    1 +
>>  fs/nfs/nfs4proc.c         |   25 ++++++++++++++++---------
>>  fs/nfs/nfs4state.c        |   15 +++++++++++++++
>>  include/linux/nfs_fs_sb.h |    1 +
>>  4 files changed, 33 insertions(+), 9 deletions(-)
>>
>> diff --git a/fs/nfs/nfs4_fs.h b/fs/nfs/nfs4_fs.h
>> index 6ea07a3..e8604af 100644
>> --- a/fs/nfs/nfs4_fs.h
>> +++ b/fs/nfs/nfs4_fs.h
>> @@ -45,6 +45,7 @@ enum nfs4_client_state {
>>        NFS4CLNT_RECLAIM_NOGRACE,
>>        NFS4CLNT_DELEGRETURN,
>>        NFS4CLNT_SESSION_SETUP,
>> +       NFS4CLNT_SESSION_DRAINING,
>>  };
>>
>>  /*
>> diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
>> index 44d1c14..ebde948 100644
>> --- a/fs/nfs/nfs4proc.c
>> +++ b/fs/nfs/nfs4proc.c
>> @@ -355,6 +355,16 @@ void nfs41_sequence_free_slot(const struct nfs_client *clp,
>>        }
>>        nfs4_free_slot(tbl, res->sr_slotid);
>>        res->sr_slotid = NFS4_MAX_SLOT_TABLE;
>> +
>> +       /* Signal state manager thread if session is drained */
>> +       if (test_bit(NFS4CLNT_SESSION_DRAINING, &clp->cl_state)) {
>> +               spin_lock(&tbl->slot_tbl_lock);
>> +               if (tbl->highest_used_slotid == -1) {
>> +                       dprintk("%s COMPLETE: Session Drained\n", __func__);
>> +                       complete(&clp->cl_session->complete);
>> +               }
>> +               spin_unlock(&tbl->slot_tbl_lock);
>> +       }
>>  }
>>
>>  static void nfs41_sequence_done(struct nfs_client *clp,
>> @@ -448,15 +458,11 @@ static int nfs41_setup_sequence(struct nfs4_session *session,
>>
>>        spin_lock(&tbl->slot_tbl_lock);
>>        if (test_bit(NFS4CLNT_SESSION_SETUP, &session->clp->cl_state)) {
>> -               if (tbl->highest_used_slotid != -1) {
>> -                       rpc_sleep_on(&tbl->slot_tbl_waitq, task, NULL);
>> -                       spin_unlock(&tbl->slot_tbl_lock);
>> -                       dprintk("<-- %s: Session reset: draining\n", __func__);
>> -                       return -EAGAIN;
>> -               }
>> -
>> -               /* The slot table is empty; start the reset thread */
>> -               dprintk("%s Session Reset\n", __func__);
>> +               /*
>> +                * The state manager will wait until the slot table is empty.
>> +                * Schedule the reset thread
>> +                */
>> +               dprintk("%s Schedule Session Reset\n", __func__);
>>                rpc_sleep_on(&tbl->slot_tbl_waitq, task, NULL);
>>                nfs4_schedule_state_manager(session->clp);
>>                spin_unlock(&tbl->slot_tbl_lock);
>> @@ -4583,6 +4589,7 @@ struct nfs4_session *nfs4_alloc_session(struct nfs_client *clp)
>>         * nfs_client struct
>>         */
>>        clp->cl_cons_state = NFS_CS_SESSION_INITING;
>> +       init_completion(&session->complete);
>>
>>        tbl = &session->fc_slot_table;
>>        spin_lock_init(&tbl->slot_tbl_lock);
>> diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c
>> index 50a1b04..f08abe8 100644
>> --- a/fs/nfs/nfs4state.c
>> +++ b/fs/nfs/nfs4state.c
>> @@ -1162,8 +1162,23 @@ static void nfs4_session_recovery_handle_error(struct nfs_client *clp, int err)
>>
>>  static int nfs4_reset_session(struct nfs_client *clp)
>>  {
>> +       struct nfs4_session *ses = clp->cl_session;
>> +       struct nfs4_slot_table *tbl = &ses->fc_slot_table;
>>        int status;
>>
>> +       spin_lock(&tbl->slot_tbl_lock);
>> +       if (tbl->highest_used_slotid != -1) {
>> +               set_bit(NFS4CLNT_SESSION_DRAINING, &clp->cl_state);
>> +               spin_unlock(&tbl->slot_tbl_lock);
>> +               status = wait_for_completion_interruptible(&ses->complete);
>> +               clear_bit(NFS4CLNT_SESSION_DRAINING, &clp->cl_state);
>> +               if (status) /* -ERESTARTSYS */
>> +                       goto out;
>> +               INIT_COMPLETION(ses->complete);
> 
> 
> Looking at my own code - I need to move the INIT_COMPLETION after the
> wait_for_completion_interruptible and before the status check so that
> it gets called even on -ERESTARTSYS.

Might there be a race with nfs41_sequence_free_slot?
Can't you just call INIT_COMPLETION under the slot_tbl_lock
before calling wait_for_completion_interruptible?

I.e, in nfs41_sequence_free_slot:
 	}
 	nfs4_free_slot(tbl, res->sr_slotid);
 	res->sr_slotid = NFS4_MAX_SLOT_TABLE;
+
+	/* Signal state manager thread if session is drained */
+	if (test_bit(NFS4CLNT_SESSION_DRAINING, &clp->cl_state)) {
+		spin_lock(&tbl->slot_tbl_lock);
+		if (tbl->highest_used_slotid == -1 &&
+		    test_bit(NFS4CLNT_SESSION_DRAINING, &clp->cl_state)) {
+			dprintk("%s COMPLETE: Session Drained\n", __func__);
+			complete(&clp->cl_session->complete);
+		}
+		spin_unlock(&tbl->slot_tbl_lock);
+	}
 }

and here:

 static int nfs4_reset_session(struct nfs_client *clp)
 {
+	struct nfs4_session *ses = clp->cl_session;
+	struct nfs4_slot_table *tbl = &ses->fc_slot_table;
 	int status;
 
+	spin_lock(&tbl->slot_tbl_lock);
+	if (tbl->highest_used_slotid != -1) {
+		set_bit(NFS4CLNT_SESSION_DRAINING, &clp->cl_state);
+		INIT_COMPLETION(ses->complete);
+		spin_unlock(&tbl->slot_tbl_lock);
+		status = wait_for_completion_interruptible(&ses->complete);
+		clear_bit(NFS4CLNT_SESSION_DRAINING, &clp->cl_state);
+		if (status) /* -ERESTARTSYS */
+			goto out;

Benny

> 
> I'll wait for other comments before sending a fix.
> 
> -->Andy
> 
> 
>> +       } else {
>> +               spin_unlock(&tbl->slot_tbl_lock);
>> +       }
>> +
>>        status = nfs4_proc_destroy_session(clp->cl_session);
>>        if (status && status != -NFS4ERR_BADSESSION &&
>>            status != -NFS4ERR_DEADSESSION) {
>> diff --git a/include/linux/nfs_fs_sb.h b/include/linux/nfs_fs_sb.h
>> index 320569e..34fc6be 100644
>> --- a/include/linux/nfs_fs_sb.h
>> +++ b/include/linux/nfs_fs_sb.h
>> @@ -209,6 +209,7 @@ struct nfs4_session {
>>        unsigned long                   session_state;
>>        u32                             hash_alg;
>>        u32                             ssv_len;
>> +       struct completion               complete;
>>
>>        /* The fore and back channel */
>>        struct nfs4_channel_attrs       fc_attrs;
>> --
>> 1.6.0.6
>>
>> --
>> To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
>> the body of a message to majordomo@vger.kernel.org
>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>>
> _______________________________________________
> pNFS mailing list
> pNFS@linux-nfs.org
> http://linux-nfs.org/cgi-bin/mailman/listinfo/pnfs

Re: [pnfs] [PATCH 2/2] nfs41: fix state manager deadlock in session reset

[Trond Myklebust]

On Thu, 2009-11-12 at 15:29 +0200, Benny Halevy wrote:
> On Nov. 11, 2009, 21:51 +0200, "William A. (Andy) Adamson" <androsadamson@gmail.com> wrote:
> > On Wed, Nov 11, 2009 at 2:20 PM,  <andros@netapp.com> wrote:
> >> From: Andy Adamson <andros@netapp.com>
> >>
> >> If the session is reset during state recovery, the state manager thread can
> >> sleep on the slot_tbl_waitq causing a deadlock.
> >>
> >> Add a completion framework to the session.  Have the state manager thread set
> >> a new session state (NFS4CLNT_SESSION_DRAINING) and wait for the session slot
> >> table to drain.
> >>
> >> Signal the state manager thread in nfs41_sequence_free_slot when the
> >> NFS4CLNT_SESSION_DRAINING bit is set and the session is drained.
> >>
> >> Reported-by: Trond Myklebust <trond@netapp.com>
> >> Signed-off-by: Andy Adamson <andros@netapp.com>
> >> ---
> >>  fs/nfs/nfs4_fs.h          |    1 +
> >>  fs/nfs/nfs4proc.c         |   25 ++++++++++++++++---------
> >>  fs/nfs/nfs4state.c        |   15 +++++++++++++++
> >>  include/linux/nfs_fs_sb.h |    1 +
> >>  4 files changed, 33 insertions(+), 9 deletions(-)
> >>
> >> diff --git a/fs/nfs/nfs4_fs.h b/fs/nfs/nfs4_fs.h
> >> index 6ea07a3..e8604af 100644
> >> --- a/fs/nfs/nfs4_fs.h
> >> +++ b/fs/nfs/nfs4_fs.h
> >> @@ -45,6 +45,7 @@ enum nfs4_client_state {
> >>        NFS4CLNT_RECLAIM_NOGRACE,
> >>        NFS4CLNT_DELEGRETURN,
> >>        NFS4CLNT_SESSION_SETUP,
> >> +       NFS4CLNT_SESSION_DRAINING,
> >>  };
> >>
> >>  /*
> >> diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
> >> index 44d1c14..ebde948 100644
> >> --- a/fs/nfs/nfs4proc.c
> >> +++ b/fs/nfs/nfs4proc.c
> >> @@ -355,6 +355,16 @@ void nfs41_sequence_free_slot(const struct nfs_client *clp,
> >>        }
> >>        nfs4_free_slot(tbl, res->sr_slotid);
> >>        res->sr_slotid = NFS4_MAX_SLOT_TABLE;
> >> +
> >> +       /* Signal state manager thread if session is drained */
> >> +       if (test_bit(NFS4CLNT_SESSION_DRAINING, &clp->cl_state)) {
> >> +               spin_lock(&tbl->slot_tbl_lock);
> >> +               if (tbl->highest_used_slotid == -1) {
> >> +                       dprintk("%s COMPLETE: Session Drained\n", __func__);
> >> +                       complete(&clp->cl_session->complete);
> >> +               }
> >> +               spin_unlock(&tbl->slot_tbl_lock);
> >> +       }
> >>  }
> >>
> >>  static void nfs41_sequence_done(struct nfs_client *clp,
> >> @@ -448,15 +458,11 @@ static int nfs41_setup_sequence(struct nfs4_session *session,
> >>
> >>        spin_lock(&tbl->slot_tbl_lock);
> >>        if (test_bit(NFS4CLNT_SESSION_SETUP, &session->clp->cl_state)) {
> >> -               if (tbl->highest_used_slotid != -1) {
> >> -                       rpc_sleep_on(&tbl->slot_tbl_waitq, task, NULL);
> >> -                       spin_unlock(&tbl->slot_tbl_lock);
> >> -                       dprintk("<-- %s: Session reset: draining\n", __func__);
> >> -                       return -EAGAIN;
> >> -               }
> >> -
> >> -               /* The slot table is empty; start the reset thread */
> >> -               dprintk("%s Session Reset\n", __func__);
> >> +               /*
> >> +                * The state manager will wait until the slot table is empty.
> >> +                * Schedule the reset thread
> >> +                */
> >> +               dprintk("%s Schedule Session Reset\n", __func__);
> >>                rpc_sleep_on(&tbl->slot_tbl_waitq, task, NULL);
> >>                nfs4_schedule_state_manager(session->clp);
> >>                spin_unlock(&tbl->slot_tbl_lock);
> >> @@ -4583,6 +4589,7 @@ struct nfs4_session *nfs4_alloc_session(struct nfs_client *clp)
> >>         * nfs_client struct
> >>         */
> >>        clp->cl_cons_state = NFS_CS_SESSION_INITING;
> >> +       init_completion(&session->complete);
> >>
> >>        tbl = &session->fc_slot_table;
> >>        spin_lock_init(&tbl->slot_tbl_lock);
> >> diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c
> >> index 50a1b04..f08abe8 100644
> >> --- a/fs/nfs/nfs4state.c
> >> +++ b/fs/nfs/nfs4state.c
> >> @@ -1162,8 +1162,23 @@ static void nfs4_session_recovery_handle_error(struct nfs_client *clp, int err)
> >>
> >>  static int nfs4_reset_session(struct nfs_client *clp)
> >>  {
> >> +       struct nfs4_session *ses = clp->cl_session;
> >> +       struct nfs4_slot_table *tbl = &ses->fc_slot_table;
> >>        int status;
> >>
> >> +       spin_lock(&tbl->slot_tbl_lock);
> >> +       if (tbl->highest_used_slotid != -1) {
> >> +               set_bit(NFS4CLNT_SESSION_DRAINING, &clp->cl_state);
> >> +               spin_unlock(&tbl->slot_tbl_lock);
> >> +               status = wait_for_completion_interruptible(&ses->complete);
> >> +               clear_bit(NFS4CLNT_SESSION_DRAINING, &clp->cl_state);
> >> +               if (status) /* -ERESTARTSYS */
> >> +                       goto out;
> >> +               INIT_COMPLETION(ses->complete);
> > 
> > 
> > Looking at my own code - I need to move the INIT_COMPLETION after the
> > wait_for_completion_interruptible and before the status check so that
> > it gets called even on -ERESTARTSYS.
> 
> Might there be a race with nfs41_sequence_free_slot?
> Can't you just call INIT_COMPLETION under the slot_tbl_lock
> before calling wait_for_completion_interruptible?
> 
> I.e, in nfs41_sequence_free_slot:
>  	}
>  	nfs4_free_slot(tbl, res->sr_slotid);
>  	res->sr_slotid = NFS4_MAX_SLOT_TABLE;
> +
> +	/* Signal state manager thread if session is drained */
> +	if (test_bit(NFS4CLNT_SESSION_DRAINING, &clp->cl_state)) {
> +		spin_lock(&tbl->slot_tbl_lock);
> +		if (tbl->highest_used_slotid == -1 &&
> +		    test_bit(NFS4CLNT_SESSION_DRAINING, &clp->cl_state)) {
> +			dprintk("%s COMPLETE: Session Drained\n", __func__);
> +			complete(&clp->cl_session->complete);
> +		}
> +		spin_unlock(&tbl->slot_tbl_lock);
> +	}
>  }
> 
> and here:
> 
>  static int nfs4_reset_session(struct nfs_client *clp)
>  {
> +	struct nfs4_session *ses = clp->cl_session;
> +	struct nfs4_slot_table *tbl = &ses->fc_slot_table;
>  	int status;
>  
> +	spin_lock(&tbl->slot_tbl_lock);
> +	if (tbl->highest_used_slotid != -1) {
> +		set_bit(NFS4CLNT_SESSION_DRAINING, &clp->cl_state);
> +		INIT_COMPLETION(ses->complete);
> +		spin_unlock(&tbl->slot_tbl_lock);
> +		status = wait_for_completion_interruptible(&ses->complete);
> +		clear_bit(NFS4CLNT_SESSION_DRAINING, &clp->cl_state);
> +		if (status) /* -ERESTARTSYS */
> +			goto out;
> 
> Benny

You shouldn't need to call it under the spinlock. Just remember to call
INIT_COMPLETION before you set NFS4CLNT_SESSION_DRAINING.

IOW:

	INIT_COMPLETION(ses->complete);
	spin_lock(&tbl->slot_tbl_lock);
	if (tbl->highest_used_slotid != -1) {
		set_bit(NFS4CLNT_SESSION_DRAINING, &clp->cl_state);
	....

Trond

-- 
Trond Myklebust
Linux NFS client maintainer

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

Re: [pnfs] [PATCH 2/2] nfs41: fix state manager deadlock in session reset

[William A. (Andy) Adamson]

On Thu, Nov 12, 2009 at 8:46 AM, Trond Myklebust
<Trond.Myklebust@netapp.com> wrote:
> On Thu, 2009-11-12 at 15:29 +0200, Benny Halevy wrote:
>> On Nov. 11, 2009, 21:51 +0200, "William A. (Andy) Adamson" <androsadamson@gmail.com> wrote:
>> > On Wed, Nov 11, 2009 at 2:20 PM,  <andros@netapp.com> wrote:
>> >> From: Andy Adamson <andros@netapp.com>
>> >>
>> >> If the session is reset during state recovery, the state manager thread can
>> >> sleep on the slot_tbl_waitq causing a deadlock.
>> >>
>> >> Add a completion framework to the session.  Have the state manager thread set
>> >> a new session state (NFS4CLNT_SESSION_DRAINING) and wait for the session slot
>> >> table to drain.
>> >>
>> >> Signal the state manager thread in nfs41_sequence_free_slot when the
>> >> NFS4CLNT_SESSION_DRAINING bit is set and the session is drained.
>> >>
>> >> Reported-by: Trond Myklebust <trond@netapp.com>
>> >> Signed-off-by: Andy Adamson <andros@netapp.com>
>> >> ---
>> >>  fs/nfs/nfs4_fs.h          |    1 +
>> >>  fs/nfs/nfs4proc.c         |   25 ++++++++++++++++---------
>> >>  fs/nfs/nfs4state.c        |   15 +++++++++++++++
>> >>  include/linux/nfs_fs_sb.h |    1 +
>> >>  4 files changed, 33 insertions(+), 9 deletions(-)
>> >>
>> >> diff --git a/fs/nfs/nfs4_fs.h b/fs/nfs/nfs4_fs.h
>> >> index 6ea07a3..e8604af 100644
>> >> --- a/fs/nfs/nfs4_fs.h
>> >> +++ b/fs/nfs/nfs4_fs.h
>> >> @@ -45,6 +45,7 @@ enum nfs4_client_state {
>> >>        NFS4CLNT_RECLAIM_NOGRACE,
>> >>        NFS4CLNT_DELEGRETURN,
>> >>        NFS4CLNT_SESSION_SETUP,
>> >> +       NFS4CLNT_SESSION_DRAINING,
>> >>  };
>> >>
>> >>  /*
>> >> diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
>> >> index 44d1c14..ebde948 100644
>> >> --- a/fs/nfs/nfs4proc.c
>> >> +++ b/fs/nfs/nfs4proc.c
>> >> @@ -355,6 +355,16 @@ void nfs41_sequence_free_slot(const struct nfs_client *clp,
>> >>        }
>> >>        nfs4_free_slot(tbl, res->sr_slotid);
>> >>        res->sr_slotid = NFS4_MAX_SLOT_TABLE;
>> >> +
>> >> +       /* Signal state manager thread if session is drained */
>> >> +       if (test_bit(NFS4CLNT_SESSION_DRAINING, &clp->cl_state)) {
>> >> +               spin_lock(&tbl->slot_tbl_lock);
>> >> +               if (tbl->highest_used_slotid == -1) {
>> >> +                       dprintk("%s COMPLETE: Session Drained\n", __func__);
>> >> +                       complete(&clp->cl_session->complete);
>> >> +               }
>> >> +               spin_unlock(&tbl->slot_tbl_lock);
>> >> +       }
>> >>  }
>> >>
>> >>  static void nfs41_sequence_done(struct nfs_client *clp,
>> >> @@ -448,15 +458,11 @@ static int nfs41_setup_sequence(struct nfs4_session *session,
>> >>
>> >>        spin_lock(&tbl->slot_tbl_lock);
>> >>        if (test_bit(NFS4CLNT_SESSION_SETUP, &session->clp->cl_state)) {
>> >> -               if (tbl->highest_used_slotid != -1) {
>> >> -                       rpc_sleep_on(&tbl->slot_tbl_waitq, task, NULL);
>> >> -                       spin_unlock(&tbl->slot_tbl_lock);
>> >> -                       dprintk("<-- %s: Session reset: draining\n", __func__);
>> >> -                       return -EAGAIN;
>> >> -               }
>> >> -
>> >> -               /* The slot table is empty; start the reset thread */
>> >> -               dprintk("%s Session Reset\n", __func__);
>> >> +               /*
>> >> +                * The state manager will wait until the slot table is empty.
>> >> +                * Schedule the reset thread
>> >> +                */
>> >> +               dprintk("%s Schedule Session Reset\n", __func__);
>> >>                rpc_sleep_on(&tbl->slot_tbl_waitq, task, NULL);
>> >>                nfs4_schedule_state_manager(session->clp);
>> >>                spin_unlock(&tbl->slot_tbl_lock);
>> >> @@ -4583,6 +4589,7 @@ struct nfs4_session *nfs4_alloc_session(struct nfs_client *clp)
>> >>         * nfs_client struct
>> >>         */
>> >>        clp->cl_cons_state = NFS_CS_SESSION_INITING;
>> >> +       init_completion(&session->complete);
>> >>
>> >>        tbl = &session->fc_slot_table;
>> >>        spin_lock_init(&tbl->slot_tbl_lock);
>> >> diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c
>> >> index 50a1b04..f08abe8 100644
>> >> --- a/fs/nfs/nfs4state.c
>> >> +++ b/fs/nfs/nfs4state.c
>> >> @@ -1162,8 +1162,23 @@ static void nfs4_session_recovery_handle_error(struct nfs_client *clp, int err)
>> >>
>> >>  static int nfs4_reset_session(struct nfs_client *clp)
>> >>  {
>> >> +       struct nfs4_session *ses = clp->cl_session;
>> >> +       struct nfs4_slot_table *tbl = &ses->fc_slot_table;
>> >>        int status;
>> >>
>> >> +       spin_lock(&tbl->slot_tbl_lock);
>> >> +       if (tbl->highest_used_slotid != -1) {
>> >> +               set_bit(NFS4CLNT_SESSION_DRAINING, &clp->cl_state);
>> >> +               spin_unlock(&tbl->slot_tbl_lock);
>> >> +               status = wait_for_completion_interruptible(&ses->complete);
>> >> +               clear_bit(NFS4CLNT_SESSION_DRAINING, &clp->cl_state);
>> >> +               if (status) /* -ERESTARTSYS */
>> >> +                       goto out;
>> >> +               INIT_COMPLETION(ses->complete);
>> >
>> >
>> > Looking at my own code - I need to move the INIT_COMPLETION after the
>> > wait_for_completion_interruptible and before the status check so that
>> > it gets called even on -ERESTARTSYS.
>>
>> Might there be a race with nfs41_sequence_free_slot?
>> Can't you just call INIT_COMPLETION under the slot_tbl_lock
>> before calling wait_for_completion_interruptible?
>>
>> I.e, in nfs41_sequence_free_slot:
>>       }
>>       nfs4_free_slot(tbl, res->sr_slotid);
>>       res->sr_slotid = NFS4_MAX_SLOT_TABLE;
>> +
>> +     /* Signal state manager thread if session is drained */
>> +     if (test_bit(NFS4CLNT_SESSION_DRAINING, &clp->cl_state)) {
>> +             spin_lock(&tbl->slot_tbl_lock);
>> +             if (tbl->highest_used_slotid == -1 &&
>> +                 test_bit(NFS4CLNT_SESSION_DRAINING, &clp->cl_state)) {
>> +                     dprintk("%s COMPLETE: Session Drained\n", __func__);
>> +                     complete(&clp->cl_session->complete);
>> +             }
>> +             spin_unlock(&tbl->slot_tbl_lock);
>> +     }
>>  }
>>
>> and here:
>>
>>  static int nfs4_reset_session(struct nfs_client *clp)
>>  {
>> +     struct nfs4_session *ses = clp->cl_session;
>> +     struct nfs4_slot_table *tbl = &ses->fc_slot_table;
>>       int status;
>>
>> +     spin_lock(&tbl->slot_tbl_lock);
>> +     if (tbl->highest_used_slotid != -1) {
>> +             set_bit(NFS4CLNT_SESSION_DRAINING, &clp->cl_state);
>> +             INIT_COMPLETION(ses->complete);
>> +             spin_unlock(&tbl->slot_tbl_lock);
>> +             status = wait_for_completion_interruptible(&ses->complete);
>> +             clear_bit(NFS4CLNT_SESSION_DRAINING, &clp->cl_state);
>> +             if (status) /* -ERESTARTSYS */
>> +                     goto out;
>>
>> Benny
>
> You shouldn't need to call it under the spinlock. Just remember to call
> INIT_COMPLETION before you set NFS4CLNT_SESSION_DRAINING.
>
> IOW:
>
>        INIT_COMPLETION(ses->complete);
>        spin_lock(&tbl->slot_tbl_lock);
>        if (tbl->highest_used_slotid != -1) {
>                set_bit(NFS4CLNT_SESSION_DRAINING, &clp->cl_state);
>        ....
>


Yes - I see now. INIT_COMPLETION shoud be called  before
wait_for_completion_XXX, not after.


-->Andy

> Trond
>
> --
> Trond Myklebust
> Linux NFS client maintainer
>
> NetApp
> Trond.Myklebust@netapp.com
> www.netapp.com
>