Skip to contentSkip to conversation

NFSv4.1 backchannel xprt problems.

PATCH In review NeilBrown 5 participants 8 messages Oct 9, 2019 – Oct 16, 2019

Subsystems (3)
lore
Share & tools
lore

NeilBrown

author
#0 lore raw

Hi,
I have a customer with a 4.12-based kernel who is experiencing memory exhaustion.

There are over 100,000 rpc_rqst structures queue on sv_cb_list for handing by the NFSv4 callback, which is idle.

The rpc_rqst.rq_xprt pointer points to freed memory.

I notice that that server code calls svc_xprt_get() on the xprt before storing it in rq_xprt, but the client/backchannel code doesn't.

I'm wondering if the following might be useful.

I plan to explore the code a bit more tomorrow and if this still seems likely I get the customer to test this change, but I thought I would ask here as well incase someone more knowledgeable can give me any pointers.

Thanks,
NeilBrown

net/sunrpc/backchannel_rqst.c +2 -1
index 339e8c077c2d..c95ca39688b6 100644
@@ -61,6 +61,7 @@ static void xprt_free_allocation(struct rpc_rqst *req)
 	free_page((unsigned long)xbufp->head[0].iov_base);
 	xbufp = &req->rq_snd_buf;
 	free_page((unsigned long)xbufp->head[0].iov_base);
+	xprt_put(req->rq_xprt);
 	kfree(req);
 }
 
@@ -85,7 +86,7 @@ struct rpc_rqst *xprt_alloc_bc_req(struct rpc_xprt *xprt, gfp_t gfp_flags)
 	if (req == NULL)
 		return NULL;
 
-	req->rq_xprt = xprt;
+	req->rq_xprt = xprt_get(xprt);
 	INIT_LIST_HEAD(&req->rq_bc_list);
 
 	/* Preallocate one XDR receive buffer */
#1 lore raw
Quoted 11 lines · NeilBrown

I have a customer with a 4.12-based kernel who is experiencing memory exhaustion.

There are over 100,000 rpc_rqst structures queue on sv_cb_list for handing by the NFSv4 callback, which is idle.

The rpc_rqst.rq_xprt pointer points to freed memory.

I notice that that server code calls svc_xprt_get() on the xprt before storing it in rq_xprt, but the client/backchannel code doesn't.

Quoted 6 lines

I'm wondering if the following might be useful.

I plan to explore the code a bit more tomorrow and if this still seems likely I get the customer to test this change, but I thought I would ask here as well incase someone more knowledgeable can give me any pointers.

Looks sensible. But if I ever understood how this works, I've got no memory of it now.... Good luck.

Quoted 25 lines

Thanks,
NeilBrown

net/sunrpc/backchannel_rqst.c +2 -1
index 339e8c077c2d..c95ca39688b6 100644
@@ -61,6 +61,7 @@ static void xprt_free_allocation(struct rpc_rqst *req)
 	free_page((unsigned long)xbufp->head[0].iov_base);
 	xbufp = &req->rq_snd_buf;
 	free_page((unsigned long)xbufp->head[0].iov_base);
+	xprt_put(req->rq_xprt);
 	kfree(req);
 }
 
@@ -85,7 +86,7 @@ struct rpc_rqst *xprt_alloc_bc_req(struct rpc_xprt *xprt, gfp_t gfp_flags)
 	if (req == NULL)
 		return NULL;
 
-	req->rq_xprt = xprt;
+	req->rq_xprt = xprt_get(xprt);
 	INIT_LIST_HEAD(&req->rq_bc_list);
 
 	/* Preallocate one XDR receive buffer */
author
#2 lore raw

PATCH 1/1 SUNRPC: backchannel RPC request must reference XPRT

The backchannel RPC requests - that are queued waiting for the reply to be sent by the "NFSv4 callback" thread - have a pointer to the xprt, but it is not reference counted. It is possible for the xprt to be freed while there are still queued requests.

I think this has been a problem since
Commit fb7a0b9addbd ("nfs41: New backchannel helper routines") when the code was introduced, but I suspect it became more of
a problem after
Commit 80b14d5e61ca ("SUNRPC: Add a structure to track multiple transports") (or there abouts).
Before this second patch, the nfs client would hold a reference to the xprt to keep it alive. After multipath was introduced,
a client holds a reference to a swtich, and the switch can have multiple xprts which can be added and removed.

I'm not sure of all the causal issues, but this patch has fixed a customer problem were an NFSv4.1 client would run out of memory with tens of thousands of backchannel rpc requests queued for an xprt that had been freed. This was a 64K-page machine so each rpc_rqst consumed more than 128K of memory.

Fixes 80b14d5e61ca ("SUNRPC: Add a structure to track multiple transports")
---
 net/sunrpc/backchannel_rqst.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
net/sunrpc/backchannel_rqst.c +2 -1
index 339e8c077c2d..c95ca39688b6 100644
@@ -61,6 +61,7 @@ static void xprt_free_allocation(struct rpc_rqst *req)
 	free_page((unsigned long)xbufp->head[0].iov_base);
 	xbufp = &req->rq_snd_buf;
 	free_page((unsigned long)xbufp->head[0].iov_base);
+	xprt_put(req->rq_xprt);
 	kfree(req);
 }
 
@@ -85,7 +86,7 @@ struct rpc_rqst *xprt_alloc_bc_req(struct rpc_xprt *xprt, gfp_t gfp_flags)
 	if (req == NULL)
 		return NULL;
 
-	req->rq_xprt = xprt;
+	req->rq_xprt = xprt_get(xprt);
 	INIT_LIST_HEAD(&req->rq_bc_list);
 
 	/* Preallocate one XDR receive buffer */
Signature · 1 line
2.23.0
#3 lore raw

Re: [PATCH] SUNRPC: backchannel RPC request must reference XPRT

Hi Neil,

Quoted 39 lines · NeilBrown

The backchannel RPC requests - that are queued waiting for the reply to be sent by the "NFSv4 callback" thread - have a pointer to the xprt, but it is not reference counted. It is possible for the xprt to be freed while there are still queued requests.

I think this has been a problem since
Commit fb7a0b9addbd ("nfs41: New backchannel helper routines") when the code was introduced, but I suspect it became more of a problem after
Commit 80b14d5e61ca ("SUNRPC: Add a structure to track multiple transports")
(or there abouts).
Before this second patch, the nfs client would hold a reference to the xprt to keep it alive. After multipath was introduced, a client holds a reference to a swtich, and the switch can have multiple
xprts which can be added and removed.

I'm not sure of all the causal issues, but this patch has fixed a customer problem were an NFSv4.1 client would run out of memory with tens of thousands of backchannel rpc requests queued for an xprt that had been freed. This was a 64K-page machine so each rpc_rqst consumed more than 128K of memory.

Fixes 80b14d5e61ca ("SUNRPC: Add a structure to track multiple

transports")

---
 net/sunrpc/backchannel_rqst.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
net/sunrpc/backchannel_rqst.c +0 -0
index 339e8c077c2d..c95ca39688b6 100644

net/sunrpc/backchannel_rqst.c static void xprt_free_allocation(struct rpc_rqst · line 67

Quoted 7 lines
net/sunrpc/backchannel_rqst.c +1 -0
@@ -61,6 +61,7 @@ static void xprt_free_allocation(struct rpc_rqst
 *req)
 	free_page((unsigned long)xbufp->head[0].iov_base);
 	xbufp = &req->rq_snd_buf;
 	free_page((unsigned long)xbufp->head[0].iov_base);
+	xprt_put(req->rq_xprt);
 	kfree(req);
 }

Would it perhaps make better sense to move the xprt_get() to xprt_lookup_bc_request() and to release it in xprt_free_bc_rqst()?

Otherwise as far as I can tell, we will have freed slots on the xprt-

Quoted 1 line

bc_pa_list that hold a reference to the transport itself, meaning that

the latter never gets released.

Quoted 11 lines
net/sunrpc/backchannel_rqst.c +1 -1
@@ -85,7 +86,7 @@ struct rpc_rqst *xprt_alloc_bc_req(struct rpc_xprt
 *xprt, gfp_t gfp_flags)
 	if (req == NULL)
 		return NULL;
 
-	req->rq_xprt = xprt;
+	req->rq_xprt = xprt_get(xprt);
 	INIT_LIST_HEAD(&req->rq_bc_list);
 

/* Preallocate one XDR receive buffer */

Signature · 3 lines
Trond Myklebust
Linux NFS client maintainer, Hammerspace
[email protected]
#4 lore raw

Re: [PATCH] SUNRPC: backchannel RPC request must reference XPRT

Quoted 40 lines · Trond Myklebust

Hi Neil,

Quoted 33 lines · NeilBrown

The backchannel RPC requests - that are queued waiting for the reply to be sent by the "NFSv4 callback" thread - have a pointer to the xprt, but it is not reference counted. It is possible for the xprt to be freed while there are still queued requests.

I think this has been a problem since
Commit fb7a0b9addbd ("nfs41: New backchannel helper routines") when the code was introduced, but I suspect it became more of a problem after
Commit 80b14d5e61ca ("SUNRPC: Add a structure to track multiple transports")
(or there abouts).
Before this second patch, the nfs client would hold a reference to the xprt to keep it alive. After multipath was introduced, a client holds a reference to a swtich, and the switch can have multiple
xprts which can be added and removed.

I'm not sure of all the causal issues, but this patch has fixed a customer problem were an NFSv4.1 client would run out of memory with tens of thousands of backchannel rpc requests queued for an xprt that had been freed. This was a 64K-page machine so each rpc_rqst consumed more than 128K of memory.

Fixes 80b14d5e61ca ("SUNRPC: Add a structure to track multiple

transports")

---
net/sunrpc/backchannel_rqst.c | 3 ++-

1 file changed, 2 insertions(+), 1 deletion(-)
Quoted 16 lines

net/sunrpc/backchannel_rqst.c static void xprt_free_allocation(struct rpc_rqst · line 67

Quoted 13 lines
net/sunrpc/backchannel_rqst.c +1 -0
index 339e8c077c2d..c95ca39688b6 100644
@@ -61,6 +61,7 @@ static void xprt_free_allocation(struct rpc_rqst
 *req)
 	free_page((unsigned long)xbufp->head[0].iov_base);
 	xbufp = &req->rq_snd_buf;
 	free_page((unsigned long)xbufp->head[0].iov_base);
+	xprt_put(req->rq_xprt);
 	kfree(req);
 }

Would it perhaps make better sense to move the xprt_get() to xprt_lookup_bc_request() and to release it in xprt_free_bc_rqst()?

/me wonders if the same problem exists for the RPC/RDMA backchannel....

Quoted 10 lines

Otherwise as far as I can tell, we will have freed slots on the xprt-

Quoted 1 line

bc_pa_list that hold a reference to the transport itself, meaning that

the latter never gets released.

Quoted 6 lines
net/sunrpc/backchannel_rqst.c +0 -0
@@ -85,7 +86,7 @@ struct rpc_rqst *xprt_alloc_bc_req(struct rpc_xprt
 *xprt, gfp_t gfp_flags)
 	if (req == NULL)
 		return NULL;
 
Quoted 9 lines
Quoted 5 lines
net/sunrpc/backchannel_rqst.c +1 -1
-	req->rq_xprt = xprt;
+	req->rq_xprt = xprt_get(xprt);
 	INIT_LIST_HEAD(&req->rq_bc_list);
 
	/* Preallocate one XDR receive buffer */
Signature · 3 lines
Trond Myklebust
Linux NFS client maintainer, Hammerspace
[email protected]

--
Chuck Lever

author
#5 lore raw

Re: [PATCH] SUNRPC: backchannel RPC request must reference XPRT

In reply to Trond Myklebust
Quoted 36 lines · Trond Myklebust

Hi Neil,

Quoted 33 lines · NeilBrown

The backchannel RPC requests - that are queued waiting for the reply to be sent by the "NFSv4 callback" thread - have a pointer to the xprt, but it is not reference counted. It is possible for the xprt to be freed while there are still queued requests.

I think this has been a problem since
Commit fb7a0b9addbd ("nfs41: New backchannel helper routines") when the code was introduced, but I suspect it became more of a problem after
Commit 80b14d5e61ca ("SUNRPC: Add a structure to track multiple transports")
(or there abouts).
Before this second patch, the nfs client would hold a reference to the xprt to keep it alive. After multipath was introduced, a client holds a reference to a swtich, and the switch can have multiple
xprts which can be added and removed.

I'm not sure of all the causal issues, but this patch has fixed a customer problem were an NFSv4.1 client would run out of memory with tens of thousands of backchannel rpc requests queued for an xprt that had been freed. This was a 64K-page machine so each rpc_rqst consumed more than 128K of memory.

Fixes 80b14d5e61ca ("SUNRPC: Add a structure to track multiple

transports")

---
 net/sunrpc/backchannel_rqst.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
Quoted 16 lines

net/sunrpc/backchannel_rqst.c static void xprt_free_allocation(struct rpc_rqst · line 67

Quoted 13 lines
net/sunrpc/backchannel_rqst.c +1 -0
index 339e8c077c2d..c95ca39688b6 100644
@@ -61,6 +61,7 @@ static void xprt_free_allocation(struct rpc_rqst
 *req)
 	free_page((unsigned long)xbufp->head[0].iov_base);
 	xbufp = &req->rq_snd_buf;
 	free_page((unsigned long)xbufp->head[0].iov_base);
+	xprt_put(req->rq_xprt);
 	kfree(req);
 }

Would it perhaps make better sense to move the xprt_get() to xprt_lookup_bc_request() and to release it in xprt_free_bc_rqst()?

... maybe ...

Quoted 4 lines

Otherwise as far as I can tell, we will have freed slots on the xprt-

Quoted 1 line

bc_pa_list that hold a reference to the transport itself, meaning that

the latter never gets released.

Apart from cleanup-on-error paths, xprt_free_allocation() is called:
- in xprt_destroy_bc() - at most 'max_reqs' times.
- in xprt_free_bc_rqst(), if the bc_alloc_count >= bc_alloc_max

So when xprt_destroy_bc() is called (from nfs4_destroy_session, via xprt_destroy_backchannel()), bc_alloc_max() is reduced, and possibly the requests are freed and bc_alloc_count is reduced accordingly. If the requests were busy, they won't have been freed then, but will then be freed when xprt_free_bc_reqst is called - because bc_alloc_max has been reduced.

Once nfs4_destroy_session() has been called on all session, and xprt_free_bc_rqst() has been called on all active requests, I think they should be no requests left to be holding a reference to the xprt.

And if there were requests left, and if we change the refcount code (like you suggest) so that they weren't holding a reference, then they would never be freed. - freeing an xprt doesn't clean out the bc_pa_list.

Though ... the connection from a session to an xprt isn't permanent (I guess, based on the rcu_read_lock in nfs4_destroy_session... which itself seems a bit odd as it doesn't inc a refcount while holding the lock).

So maybe the counts can get out of balance.

Conclusion: I'm not sure the ref counts are entirely correct, but I
cannot see a benefit from moving the xprt_get/put requests like you suggest.

Thanks,
NeilBrown

Quoted 16 lines
Quoted 12 lines
net/sunrpc/backchannel_rqst.c +1 -1
@@ -85,7 +86,7 @@ struct rpc_rqst *xprt_alloc_bc_req(struct rpc_xprt
 *xprt, gfp_t gfp_flags)
 	if (req == NULL)
 		return NULL;
 
-	req->rq_xprt = xprt;
+	req->rq_xprt = xprt_get(xprt);
 	INIT_LIST_HEAD(&req->rq_bc_list);
 

/* Preallocate one XDR receive buffer */

Signature · 3 lines
Trond Myklebust
Linux NFS client maintainer, Hammerspace
[email protected]
#6 lore raw

Re: [PATCH] SUNRPC: backchannel RPC request must reference XPRT

Quoted 94 lines · NeilBrown
Quoted 46 lines · Trond Myklebust

Hi Neil,

Quoted 43 lines · NeilBrown

The backchannel RPC requests - that are queued waiting for the reply to be sent by the "NFSv4 callback" thread - have a pointer to the xprt, but it is not reference counted. It is possible for the xprt to be freed while there are still queued requests.

I think this has been a problem since
Commit fb7a0b9addbd ("nfs41: New backchannel helper routines") when the code was introduced, but I suspect it became more of a problem after
Commit 80b14d5e61ca ("SUNRPC: Add a structure to track multiple transports")
(or there abouts).
Before this second patch, the nfs client would hold a reference to
the xprt to keep it alive. After multipath was introduced, a client holds a reference to a swtich, and the switch can have multiple
xprts which can be added and removed.

I'm not sure of all the causal issues, but this patch has fixed a customer problem were an NFSv4.1 client would run out of memory with tens of thousands of backchannel rpc requests queued for an xprt that had been freed. This was a 64K-page machine so each rpc_rqst consumed more than 128K of memory.

Fixes 80b14d5e61ca ("SUNRPC: Add a structure to track multiple

transports")

---
 net/sunrpc/backchannel_rqst.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
net/sunrpc/backchannel_rqst.c +0 -0
index 339e8c077c2d..c95ca39688b6 100644
@@ -61,6 +61,7 @@ static void xprt_free_allocation(struct

rpc_rqst
*req)
free_page((unsigned long)xbufp->head[0].iov_base);

Quoted 8 lines

net/sunrpc/backchannel_rqst.c static void xprt_free_allocation(struct rpc_rqst *req) · line 66

Quoted 5 lines
net/sunrpc/backchannel_rqst.c +1 -0
 	xbufp = &req->rq_snd_buf;
 	free_page((unsigned long)xbufp->head[0].iov_base);
+	xprt_put(req->rq_xprt);
 	kfree(req);
 }

Would it perhaps make better sense to move the xprt_get() to xprt_lookup_bc_request() and to release it in xprt_free_bc_rqst()?

... maybe ...

Quoted 5 lines

Otherwise as far as I can tell, we will have freed slots on the xprt-

Quoted 2 lines

bc_pa_list that hold a reference to the transport itself, meaning that

the latter never gets released.

Apart from cleanup-on-error paths, xprt_free_allocation() is called:
- in xprt_destroy_bc() - at most 'max_reqs' times.
- in xprt_free_bc_rqst(), if the bc_alloc_count >= bc_alloc_max

So when xprt_destroy_bc() is called (from nfs4_destroy_session, via xprt_destroy_backchannel()), bc_alloc_max() is reduced, and possibly the requests are freed and bc_alloc_count is reduced accordingly. If the requests were busy, they won't have been freed then, but will then be freed when xprt_free_bc_reqst is called - because bc_alloc_max
has been reduced.

Once nfs4_destroy_session() has been called on all session, and xprt_free_bc_rqst() has been called on all active requests, I think they should be no requests left to be holding a reference to the xprt.

And if there were requests left, and if we change the refcount code (like you suggest) so that they weren't holding a reference, then they
would never be freed. - freeing an xprt doesn't clean out the bc_pa_list.

Though ... the connection from a session to an xprt isn't permanent (I
guess, based on the rcu_read_lock in nfs4_destroy_session... which itself seems a bit odd as it doesn't inc a refcount while holding the lock).

Quoted 7 lines

So maybe the counts can get out of balance.

Conclusion: I'm not sure the ref counts are entirely correct, but I
cannot see a benefit from moving the xprt_get/put requests like
you
suggest.

I don't buy that conclusion.

Nothing stops me from changing the value of NFS41_BC_MIN_CALLBACKS to zero. Why should that affect the existence or not of the transport by changing the number of references held? That parameter is supposed to determine the number of pre-allocated requests and nothing else.

I do agree with your assessment that destroying the xprt does not currrently destroy the contents of xprt->bc_pa_list if they are non- zero, but that would be a bug, and an easy one to fix.

Signature · 3 lines
Trond Myklebust
Linux NFS client maintainer, Hammerspace
[email protected]
author
#7 lore raw

Re: [PATCH] SUNRPC: backchannel RPC request must reference XPRT

Quoted 103 lines · Trond Myklebust
Quoted 94 lines · NeilBrown
Quoted 46 lines · Trond Myklebust

Hi Neil,

Quoted 43 lines · NeilBrown

The backchannel RPC requests - that are queued waiting for the reply to be sent by the "NFSv4 callback" thread - have a pointer to the xprt, but it is not reference counted. It is possible for the xprt to be freed while there are still queued requests.

I think this has been a problem since
Commit fb7a0b9addbd ("nfs41: New backchannel helper routines") when the code was introduced, but I suspect it became more of a problem after
Commit 80b14d5e61ca ("SUNRPC: Add a structure to track multiple transports")
(or there abouts).
Before this second patch, the nfs client would hold a reference to
the xprt to keep it alive. After multipath was introduced, a client holds a reference to a swtich, and the switch can have multiple
xprts which can be added and removed.

I'm not sure of all the causal issues, but this patch has fixed a customer problem were an NFSv4.1 client would run out of memory with tens of thousands of backchannel rpc requests queued for an xprt that had been freed. This was a 64K-page machine so each rpc_rqst consumed more than 128K of memory.

Fixes 80b14d5e61ca ("SUNRPC: Add a structure to track multiple

transports")

---
 net/sunrpc/backchannel_rqst.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
net/sunrpc/backchannel_rqst.c +0 -0
index 339e8c077c2d..c95ca39688b6 100644
@@ -61,6 +61,7 @@ static void xprt_free_allocation(struct

rpc_rqst
*req)
free_page((unsigned long)xbufp->head[0].iov_base);

Quoted 8 lines

net/sunrpc/backchannel_rqst.c static void xprt_free_allocation(struct rpc_rqst *req) · line 66

Quoted 5 lines
net/sunrpc/backchannel_rqst.c +1 -0
 	xbufp = &req->rq_snd_buf;
 	free_page((unsigned long)xbufp->head[0].iov_base);
+	xprt_put(req->rq_xprt);
 	kfree(req);
 }

Would it perhaps make better sense to move the xprt_get() to xprt_lookup_bc_request() and to release it in xprt_free_bc_rqst()?

... maybe ...

Quoted 5 lines

Otherwise as far as I can tell, we will have freed slots on the xprt-

Quoted 2 lines

bc_pa_list that hold a reference to the transport itself, meaning that

the latter never gets released.

Apart from cleanup-on-error paths, xprt_free_allocation() is called:
- in xprt_destroy_bc() - at most 'max_reqs' times.
- in xprt_free_bc_rqst(), if the bc_alloc_count >= bc_alloc_max

So when xprt_destroy_bc() is called (from nfs4_destroy_session, via xprt_destroy_backchannel()), bc_alloc_max() is reduced, and possibly the requests are freed and bc_alloc_count is reduced accordingly. If the requests were busy, they won't have been freed then, but will then be freed when xprt_free_bc_reqst is called - because bc_alloc_max
has been reduced.

Once nfs4_destroy_session() has been called on all session, and xprt_free_bc_rqst() has been called on all active requests, I think they should be no requests left to be holding a reference to the xprt.

And if there were requests left, and if we change the refcount code (like you suggest) so that they weren't holding a reference, then they
would never be freed. - freeing an xprt doesn't clean out the bc_pa_list.

Though ... the connection from a session to an xprt isn't permanent (I
guess, based on the rcu_read_lock in nfs4_destroy_session... which itself seems a bit odd as it doesn't inc a refcount while holding the lock).

Quoted 7 lines

So maybe the counts can get out of balance.

Conclusion: I'm not sure the ref counts are entirely correct, but I
cannot see a benefit from moving the xprt_get/put requests like
you
suggest.

Quoted 6 lines

I don't buy that conclusion.

Nothing stops me from changing the value of NFS41_BC_MIN_CALLBACKS to zero. Why should that affect the existence or not of the transport by changing the number of references held? That parameter is supposed to determine the number of pre-allocated requests and nothing else.

If you reduce NFS41_BC_MIN_CALLBACKS to zero, bc_alloc_max won't be increased and will remain at zero. So xprt_get_bc_request() will always need to allocate a new request (bc_pa_list will be empty), and xprt_free_bc_rqst() will always free the request - never putting it on the bc_pa_list.

Maybe it would make sense to do exactly that, get rid of bc_pa_list, and possibly use a mempool if there is any concern about delays when allocating.

I agree that it is a little odd for requests on the pa list to hold a reference to the xprt, and I don't object at all to changing that. My only point is that I think my patch as it stands is correct, that it doesn't introduce new problems (and so might be the simplest thing for -stable).

Thanks,
NeilBrown

Quoted 10 lines

I do agree with your assessment that destroying the xprt does not currrently destroy the contents of xprt->bc_pa_list if they are non- zero, but that would be a bug, and an easy one to fix.

Signature · 3 lines
Trond Myklebust
Linux NFS client maintainer, Hammerspace
[email protected]