[PATCH 0/2] Two more places that need to handle ENETDOWN/ENETUNREACH

[PATCH 0/2] Two more places that need to handle ENETDOWN/ENETUNREACH

[trondmy]

From: Trond Myklebust <trond.myklebust@hammerspace.com>

Omar Sandoval reports seeing layouts that are leaked when a container is
shut down uncleanly.

Trond Myklebust (2):
  NFSv4: Handle fatal ENETDOWN and ENETUNREACH errors
  NFSv4/pnfs: Layoutreturn on close must handle fatal networking errors

 fs/nfs/nfs4proc.c |  9 +++++++++
 fs/nfs/pnfs.c     | 10 +++++++++-
 2 files changed, 18 insertions(+), 1 deletion(-)

-- 
2.49.0

[PATCH 1/2] NFSv4: Handle fatal ENETDOWN and ENETUNREACH errors

[trondmy]

From: Trond Myklebust <trond.myklebust@hammerspace.com>

Ensure that the NFSv4 error handling code recognises the
RPC_TASK_NETUNREACH_FATAL flag, and handles the ENETDOWN and ENETUNREACH
errors accordingly.

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
---
 fs/nfs/nfs4proc.c | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index da97f87ecaa9..f862c862b3a3 100644
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -671,6 +671,15 @@ nfs4_async_handle_exception(struct rpc_task *task, struct nfs_server *server,
 	struct nfs_client *clp = server->nfs_client;
 	int ret;
 
+	if ((task->tk_rpc_status == -ENETDOWN ||
+	     task->tk_rpc_status == -ENETUNREACH) &&
+	    task->tk_flags & RPC_TASK_NETUNREACH_FATAL) {
+		exception->retry = 0;
+		exception->recovering = 0;
+		exception->retry = 0;
+		return -EIO;
+	}
+
 	ret = nfs4_do_handle_exception(server, errorcode, exception);
 	if (exception->delay) {
 		int ret2 = nfs4_exception_should_retrans(server, exception);
-- 
2.49.0

Re: [PATCH 1/2] NFSv4: Handle fatal ENETDOWN and ENETUNREACH errors

[Jeff Layton]

On Sun, 2025-04-06 at 11:11 +0200, trondmy@kernel.org wrote:
> From: Trond Myklebust <trond.myklebust@hammerspace.com>
> 
> Ensure that the NFSv4 error handling code recognises the
> RPC_TASK_NETUNREACH_FATAL flag, and handles the ENETDOWN and ENETUNREACH
> errors accordingly.
> 
> Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
> ---
>  fs/nfs/nfs4proc.c | 9 +++++++++
>  1 file changed, 9 insertions(+)
> 
> diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
> index da97f87ecaa9..f862c862b3a3 100644
> --- a/fs/nfs/nfs4proc.c
> +++ b/fs/nfs/nfs4proc.c
> @@ -671,6 +671,15 @@ nfs4_async_handle_exception(struct rpc_task *task, struct nfs_server *server,
>  	struct nfs_client *clp = server->nfs_client;
>  	int ret;
>  
> +	if ((task->tk_rpc_status == -ENETDOWN ||
> +	     task->tk_rpc_status == -ENETUNREACH) &&
> +	    task->tk_flags & RPC_TASK_NETUNREACH_FATAL) {
> +		exception->retry = 0;
> +		exception->recovering = 0;
> +		exception->retry = 0;

Why set exception->retry twice?

> +		return -EIO;
> +	}
> +
>  	ret = nfs4_do_handle_exception(server, errorcode, exception);
>  	if (exception->delay) {
>  		int ret2 = nfs4_exception_should_retrans(server, exception);

Other than that, this looks sane.

Reviewed-by: Jeff Layton <jlayton@kernel.org>

Re: [PATCH 1/2] NFSv4: Handle fatal ENETDOWN and ENETUNREACH errors

[Trond Myklebust]

On Sun, 2025-04-06 at 09:08 -0400, Jeff Layton wrote:
> On Sun, 2025-04-06 at 11:11 +0200, trondmy@kernel.org wrote:
> > From: Trond Myklebust <trond.myklebust@hammerspace.com>
> > 
> > Ensure that the NFSv4 error handling code recognises the
> > RPC_TASK_NETUNREACH_FATAL flag, and handles the ENETDOWN and
> > ENETUNREACH
> > errors accordingly.
> > 
> > Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
> > ---
> >  fs/nfs/nfs4proc.c | 9 +++++++++
> >  1 file changed, 9 insertions(+)
> > 
> > diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
> > index da97f87ecaa9..f862c862b3a3 100644
> > --- a/fs/nfs/nfs4proc.c
> > +++ b/fs/nfs/nfs4proc.c
> > @@ -671,6 +671,15 @@ nfs4_async_handle_exception(struct rpc_task
> > *task, struct nfs_server *server,
> >  	struct nfs_client *clp = server->nfs_client;
> >  	int ret;
> >  
> > +	if ((task->tk_rpc_status == -ENETDOWN ||
> > +	     task->tk_rpc_status == -ENETUNREACH) &&
> > +	    task->tk_flags & RPC_TASK_NETUNREACH_FATAL) {
> > +		exception->retry = 0;
> > +		exception->recovering = 0;
> > +		exception->retry = 0;
> 
> Why set exception->retry twice?

Oops. That last one is supposed to be exception->delay = 0

Thanks for noticing!

> 
> > +		return -EIO;
> > +	}
> > +
> >  	ret = nfs4_do_handle_exception(server, errorcode,
> > exception);
> >  	if (exception->delay) {
> >  		int ret2 = nfs4_exception_should_retrans(server,
> > exception);
> 
> Other than that, this looks sane.
> 
> Reviewed-by: Jeff Layton <jlayton@kernel.org>

[PATCH 2/2] NFSv4/pnfs: Layoutreturn on close must handle fatal networking errors

[trondmy]

From: Trond Myklebust <trond.myklebust@hammerspace.com>

If we have a fatal ENETDOWN or ENETUNREACH error, then the layoutreturn
on close code should also handle that as fatal, and free the layouts.

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
---
 fs/nfs/pnfs.c | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)

diff --git a/fs/nfs/pnfs.c b/fs/nfs/pnfs.c
index 5f582713bf05..3554046b5bfc 100644
--- a/fs/nfs/pnfs.c
+++ b/fs/nfs/pnfs.c
@@ -1659,8 +1659,16 @@ int pnfs_roc_done(struct rpc_task *task, struct nfs4_layoutreturn_args **argpp,
 		break;
 	case -NFS4ERR_NOMATCHING_LAYOUT:
 		/* Was there an RPC level error? If not, retry */
-		if (task->tk_rpc_status == 0)
+		if (task->tk_rpc_status == 0) {
+			if ((task->tk_rpc_status == -ENETDOWN ||
+			     task->tk_rpc_status == -ENETUNREACH) &&
+			    task->tk_flags & RPC_TASK_NETUNREACH_FATAL) {
+				*ret = 0;
+				(*respp)->lrs_present = 0;
+				retval = -EIO;
+			}
 			break;
+		}
 		/* If the call was not sent, let caller handle it */
 		if (!RPC_WAS_SENT(task))
 			return 0;
-- 
2.49.0

Re: [PATCH 2/2] NFSv4/pnfs: Layoutreturn on close must handle fatal networking errors

[Jeff Layton]

On Sun, 2025-04-06 at 11:11 +0200, trondmy@kernel.org wrote:
> From: Trond Myklebust <trond.myklebust@hammerspace.com>
> 
> If we have a fatal ENETDOWN or ENETUNREACH error, then the layoutreturn
> on close code should also handle that as fatal, and free the layouts.
> 
> Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
> ---
>  fs/nfs/pnfs.c | 10 +++++++++-
>  1 file changed, 9 insertions(+), 1 deletion(-)
> 
> diff --git a/fs/nfs/pnfs.c b/fs/nfs/pnfs.c
> index 5f582713bf05..3554046b5bfc 100644
> --- a/fs/nfs/pnfs.c
> +++ b/fs/nfs/pnfs.c
> @@ -1659,8 +1659,16 @@ int pnfs_roc_done(struct rpc_task *task, struct nfs4_layoutreturn_args **argpp,
>  		break;
>  	case -NFS4ERR_NOMATCHING_LAYOUT:
>  		/* Was there an RPC level error? If not, retry */
> -		if (task->tk_rpc_status == 0)
> +		if (task->tk_rpc_status == 0) {
> +			if ((task->tk_rpc_status == -ENETDOWN ||
> +			     task->tk_rpc_status == -ENETUNREACH) &&
> +			    task->tk_flags & RPC_TASK_NETUNREACH_FATAL) {

You already checked that task->tk_rpc_status is 0, so the inside if
statement will never be true. I think you probably want to get rid if
the outside 

     if (task->tk_rpc_status == 0)

condition?

> +				*ret = 0;
> +				(*respp)->lrs_present = 0;
> +				retval = -EIO;
> +			}
>  			break;
> +		}
>  		/* If the call was not sent, let caller handle it */
>  		if (!RPC_WAS_SENT(task))
>  			return 0;

-- 
Jeff Layton <jlayton@kernel.org>

Re: [PATCH 2/2] NFSv4/pnfs: Layoutreturn on close must handle fatal networking errors

[Trond Myklebust]

On Sun, 2025-04-06 at 09:16 -0400, Jeff Layton wrote:
> On Sun, 2025-04-06 at 11:11 +0200, trondmy@kernel.org wrote:
> > From: Trond Myklebust <trond.myklebust@hammerspace.com>
> > 
> > If we have a fatal ENETDOWN or ENETUNREACH error, then the
> > layoutreturn
> > on close code should also handle that as fatal, and free the
> > layouts.
> > 
> > Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
> > ---
> >  fs/nfs/pnfs.c | 10 +++++++++-
> >  1 file changed, 9 insertions(+), 1 deletion(-)
> > 
> > diff --git a/fs/nfs/pnfs.c b/fs/nfs/pnfs.c
> > index 5f582713bf05..3554046b5bfc 100644
> > --- a/fs/nfs/pnfs.c
> > +++ b/fs/nfs/pnfs.c
> > @@ -1659,8 +1659,16 @@ int pnfs_roc_done(struct rpc_task *task,
> > struct nfs4_layoutreturn_args **argpp,
> >  		break;
> >  	case -NFS4ERR_NOMATCHING_LAYOUT:
> >  		/* Was there an RPC level error? If not, retry */
> > -		if (task->tk_rpc_status == 0)
> > +		if (task->tk_rpc_status == 0) {
> > +			if ((task->tk_rpc_status == -ENETDOWN ||
> > +			     task->tk_rpc_status == -ENETUNREACH)
> > &&
> > +			    task->tk_flags &
> > RPC_TASK_NETUNREACH_FATAL) {
> 
> You already checked that task->tk_rpc_status is 0, so the inside if
> statement will never be true. I think you probably want to get rid if
> the outside 
> 
>      if (task->tk_rpc_status == 0)
> 
> condition?

Sorry... Jet lagged in Switzerland.
Let me move that out of the == 0 and before the !SENT...

> 
> > +				*ret = 0;
> > +				(*respp)->lrs_present = 0;
> > +				retval = -EIO;
> > +			}
> >  			break;
> > +		}
> >  		/* If the call was not sent, let caller handle it
> > */
> >  		if (!RPC_WAS_SENT(task))
> >  			return 0;
>