[PATCH v2 0/2] Two more places that need to handle ENETDOWN/ENETUNREACH

[PATCH v2 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.

---
v2:
 - Fix issues reported by Jeff Layton

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     | 12 ++++++++++++
 2 files changed, 21 insertions(+)

-- 
2.49.0

[PATCH v2 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..01417e3099e3 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->delay = 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 v2 1/2] NFSv4: Handle fatal ENETDOWN and ENETUNREACH errors

[Jeff Layton]

On Sun, 2025-04-06 at 17:45 +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..01417e3099e3 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) {

We're sprinkling the above conditional in quite a few places now. It
would be nice to turn the above if statement into a helper function.
Something like this maybe?

static inline bool netunreach_fatal(struct rpc_task *task)
{
	return (task->tk_rpc_status == -ENETDOWN || task->tk_rpc_status == -ENETUNREACH) && (task->tk_flags & RPC_TASK_NETUNREACH_FATAL);
}

> +		exception->delay = 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);

Patch looks good though.

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

[PATCH v2 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 | 12 ++++++++++++
 1 file changed, 12 insertions(+)

diff --git a/fs/nfs/pnfs.c b/fs/nfs/pnfs.c
index 5f582713bf05..10fdd065a61c 100644
--- a/fs/nfs/pnfs.c
+++ b/fs/nfs/pnfs.c
@@ -1661,6 +1661,18 @@ int pnfs_roc_done(struct rpc_task *task, struct nfs4_layoutreturn_args **argpp,
 		/* Was there an RPC level error? If not, retry */
 		if (task->tk_rpc_status == 0)
 			break;
+		/*
+		 * Is there a fatal network level error?
+		 * If so release the layout, but flag the error.
+		 */
+		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 v2 2/2] NFSv4/pnfs: Layoutreturn on close must handle fatal networking errors

[Jeff Layton]

On Sun, 2025-04-06 at 17:45 +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 | 12 ++++++++++++
>  1 file changed, 12 insertions(+)
> 
> diff --git a/fs/nfs/pnfs.c b/fs/nfs/pnfs.c
> index 5f582713bf05..10fdd065a61c 100644
> --- a/fs/nfs/pnfs.c
> +++ b/fs/nfs/pnfs.c
> @@ -1661,6 +1661,18 @@ int pnfs_roc_done(struct rpc_task *task, struct nfs4_layoutreturn_args **argpp,
>  		/* Was there an RPC level error? If not, retry */
>  		if (task->tk_rpc_status == 0)
>  			break;
> +		/*
> +		 * Is there a fatal network level error?
> +		 * If so release the layout, but flag the error.
> +		 */
> +		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;

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