Skip to contentSkip to conversation

Two more places that need to handle ENETDOWN/ENETUNREACH

PATCH Superseded [email protected] 2 participants 7 messages Apr 6, 2025

Subsystems (2)

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore

[email protected]

author maintainer
#0 lore raw

cover Two more places that need to handle ENETDOWN/ENETUNREACH

From: Trond Myklebust <[email protected]>

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(-)
Signature · 1 line
2.49.0
author maintainer
#1 lore raw

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

From: Trond Myklebust <[email protected]>

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

---
 fs/nfs/nfs4proc.c | 9 +++++++++
 1 file changed, 9 insertions(+)
fs/nfs/nfs4proc.c +9 -0
index da97f87ecaa9..f862c862b3a3 100644
@@ -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);
Signature · 1 line
2.49.0
maintainer Reviewed-by on v1 1/2
#2 lore raw

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

Quoted 16 lines · On Sun, 2025-04-06 at 11:11 +0200, [email protected] wrote:

From: Trond Myklebust <[email protected]>

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

---
 fs/nfs/nfs4proc.c | 9 +++++++++
 1 file changed, 9 insertions(+)
fs/nfs/nfs4proc.c +0 -0
index da97f87ecaa9..f862c862b3a3 100644

fs/nfs/nfs4proc.c nfs4_async_handle_exception(struct rpc_task *task, struct nfs_server *server, · line 679

Quoted 9 lines
fs/nfs/nfs4proc.c +6 -0
@@ -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?

fs/nfs/nfs4proc.c nfs4_async_handle_exception(struct rpc_task *task, struct nfs_server *server, · line 685

Quoted 6 lines
fs/nfs/nfs4proc.c +3 -0
+		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.

author maintainer
#3 lore raw

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

Quoted 22 lines · Jeff Layton
Quoted 21 lines · On Sun, 2025-04-06 at 11:11 +0200, [email protected] wrote:

From: Trond Myklebust <[email protected]>

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

---
 fs/nfs/nfs4proc.c | 9 +++++++++

 1 file changed, 9 insertions(+)
fs/nfs/nfs4proc.c +0 -0
index da97f87ecaa9..f862c862b3a3 100644
@@ -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;

Quoted 8 lines

fs/nfs/nfs4proc.c nfs4_async_handle_exception(struct rpc_task *task, struct nfs_server *server, · line 679

Quoted 6 lines
fs/nfs/nfs4proc.c +6 -0
+	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!

Quoted 13 lines
Quoted 9 lines
fs/nfs/nfs4proc.c +3 -0
+		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.

author maintainer
#4 lore raw

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

In reply to [email protected]

From: Trond Myklebust <[email protected]>

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.

---
 fs/nfs/pnfs.c | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)
fs/nfs/pnfs.c +9 -1
index 5f582713bf05..3554046b5bfc 100644
@@ -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;
Signature · 1 line
2.49.0
maintainer
#5 lore raw

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

Quoted 15 lines · On Sun, 2025-04-06 at 11:11 +0200, [email protected] wrote:

From: Trond Myklebust <[email protected]>

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.

---
 fs/nfs/pnfs.c | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)
fs/nfs/pnfs.c +0 -0
index 5f582713bf05..3554046b5bfc 100644

fs/nfs/pnfs.c int pnfs_roc_done(struct rpc_task *task, struct nfs4_layoutreturn_args **argpp, · line 1665

Quoted 8 lines
fs/nfs/pnfs.c +4 -1
@@ -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?

Quoted 9 lines
fs/nfs/pnfs.c +5 -0
+				*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;
Signature · 1 line
Jeff Layton <[email protected]>
author maintainer
#6 lore raw

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

Quoted 34 lines · Jeff Layton
Quoted 27 lines · On Sun, 2025-04-06 at 11:11 +0200, [email protected] wrote:

From: Trond Myklebust <[email protected]>

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.

---
 fs/nfs/pnfs.c | 10 +++++++++-

 1 file changed, 9 insertions(+), 1 deletion(-)
fs/nfs/pnfs.c +0 -0
index 5f582713bf05..3554046b5bfc 100644
@@ -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 */

fs/nfs/pnfs.c +3 -1
-		if (task->tk_rpc_status == 0)
+		if (task->tk_rpc_status == 0) {
+			if ((task->tk_rpc_status == -ENETDOWN ||
+			     task->tk_rpc_status == -ENETUNREACH)

&&

fs/nfs/pnfs.c +1 -0
+			    task->tk_flags &
Quoted 1 line

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

Quoted 3 lines

     if (task->tk_rpc_status == 0)

condition?

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

Quoted 12 lines
Quoted 11 lines
fs/nfs/pnfs.c +4 -0
+				*ret = 0;
+				(*respp)->lrs_present = 0;
+				retval = -EIO;
+			}

  break;

fs/nfs/pnfs.c +1 -0
+		}

  /* If the call was not sent, let caller handle it */
  if (!RPC_WAS_SENT(task))
  return 0;