Merge tag 'nfsd-6.15' of git://git.kernel.org/pub/scm/linux/kernel/git/cel/linux

Pull nfsd updates from Chuck Lever:
 "Neil Brown contributed more scalability improvements to NFSD's open
  file cache, and Jeff Layton contributed a menagerie of repairs to
  NFSD's NFSv4 callback / backchannel implementation.

  Mike Snitzer contributed a change to NFS re-export support that
  disables support for file locking on a re-exported NFSv4 mount. This
  is because NFSv4 state recovery is currently difficult if not
  impossible for re-exported NFS mounts. The change aims to prevent data
  integrity exposures after the re-export server crashes.

  Work continues on the evolving NFSD netlink administrative API.

  Many thanks to the contributors, reviewers, testers, and bug reporters
  who participated during the v6.15 development cycle"

* tag 'nfsd-6.15' of git://git.kernel.org/pub/scm/linux/kernel/git/cel/linux: (45 commits)
  NFSD: Add a Kconfig setting to enable delegated timestamps
  sysctl: Fixes nsm_local_state bounds
  nfsd: use a long for the count in nfsd4_state_shrinker_count()
  nfsd: remove obsolete comment from nfs4_alloc_stid
  nfsd: remove unneeded forward declaration of nfsd4_mark_cb_fault()
  nfsd: reorganize struct nfs4_delegation for better packing
  nfsd: handle errors from rpc_call_async()
  nfsd: move cb_need_restart flag into cb_flags
  nfsd: replace CB_GETATTR_BUSY with NFSD4_CALLBACK_RUNNING
  nfsd: eliminate cl_ra_cblist and NFSD4_CLIENT_CB_RECALL_ANY
  nfsd: prevent callback tasks running concurrently
  nfsd: disallow file locking and delegations for NFSv4 reexport
  nfsd: filecache: drop the list_lru lock during lock gc scans
  nfsd: filecache: don't repeatedly add/remove files on the lru list
  nfsd: filecache: introduce NFSD_FILE_RECENT
  nfsd: filecache: use list_lru_walk_node() in nfsd_file_gc()
  nfsd: filecache: use nfsd_file_dispose_list() in nfsd_file_close_inode_sync()
  NFSD: Re-organize nfsd_file_gc_worker()
  nfsd: filecache: remove race handling.
  fs: nfs: acl: Avoid -Wflex-array-member-not-at-end warning
  ...

Author: torvalds

Documentation/filesystems/nfs/reexport.rst

@@ -26,9 +26,13 @@ Reboot recovery
 ---------------
 
 The NFS protocol's normal reboot recovery mechanisms don't work for the
-case when the reexport server reboots.  Clients will lose any locks
-they held before the reboot, and further IO will result in errors.
-Closing and reopening files should clear the errors.
+case when the reexport server reboots because the source server has not
+rebooted, and so it is not in grace.  Since the source server is not in
+grace, it cannot offer any guarantees that the file won't have been
+changed between the locks getting lost and any attempt to recover them.
+The same applies to delegations and any associated locks.  Clients are
+not allowed to get file locks or delegations from a reexport server, any
+attempts will fail with operation not supported.
 
 Filehandle limits
 -----------------

Documentation/netlink/specs/lockd.yaml

@@ -0,0 +1,45 @@
+# SPDX-License-Identifier: ((GPL-2.0 WITH Linux-syscall-note) OR BSD-3-Clause)
+
+name: lockd
+protocol: genetlink
+uapi-header: linux/lockd_netlink.h
+
+doc: lockd configuration over generic netlink
+
+attribute-sets:
+  -
+    name: server
+    attributes:
+      -
+        name: gracetime
+        type: u32
+      -
+        name: tcp-port
+        type: u16
+      -
+        name: udp-port
+        type: u16
+
+operations:
+  list:
+    -
+      name: server-set
+      doc: set the lockd server parameters
+      attribute-set: server
+      flags: [ admin-perm ]
+      do:
+        request:
+          attributes:
+            - gracetime
+            - tcp-port
+            - udp-port
+    -
+      name: server-get
+      doc: get the lockd server parameters
+      attribute-set: server
+      do:
+        reply:
+          attributes:
+            - gracetime
+            - tcp-port
+            - udp-port

fs/lockd/Makefile

@@ -8,6 +8,6 @@ ccflags-y += -I$(src)			# needed for trace events
 obj-$(CONFIG_LOCKD) += lockd.o
 
 lockd-y := clntlock.o clntproc.o clntxdr.o host.o svc.o svclock.o \
-	   svcshare.o svcproc.o svcsubs.o mon.o trace.o xdr.o
+	   svcshare.o svcproc.o svcsubs.o mon.o trace.o xdr.o netlink.o
 lockd-$(CONFIG_LOCKD_V4) += clnt4xdr.o xdr4.o svc4proc.o
 lockd-$(CONFIG_PROC_FS) += procfs.o

fs/lockd/netlink.c

@@ -0,0 +1,44 @@
+// SPDX-License-Identifier: ((GPL-2.0 WITH Linux-syscall-note) OR BSD-3-Clause)
+/* Do not edit directly, auto-generated from: */
+/*	Documentation/netlink/specs/lockd.yaml */
+/* YNL-GEN kernel source */
+
+#include <net/netlink.h>
+#include <net/genetlink.h>
+
+#include "netlink.h"
+
+#include <uapi/linux/lockd_netlink.h>
+
+/* LOCKD_CMD_SERVER_SET - do */
+static const struct nla_policy lockd_server_set_nl_policy[LOCKD_A_SERVER_UDP_PORT + 1] = {
+	[LOCKD_A_SERVER_GRACETIME] = { .type = NLA_U32, },
+	[LOCKD_A_SERVER_TCP_PORT] = { .type = NLA_U16, },
+	[LOCKD_A_SERVER_UDP_PORT] = { .type = NLA_U16, },
+};
+
+/* Ops table for lockd */
+static const struct genl_split_ops lockd_nl_ops[] = {
+	{
+		.cmd		= LOCKD_CMD_SERVER_SET,
+		.doit		= lockd_nl_server_set_doit,
+		.policy		= lockd_server_set_nl_policy,
+		.maxattr	= LOCKD_A_SERVER_UDP_PORT,
+		.flags		= GENL_ADMIN_PERM | GENL_CMD_CAP_DO,
+	},
+	{
+		.cmd	= LOCKD_CMD_SERVER_GET,
+		.doit	= lockd_nl_server_get_doit,
+		.flags	= GENL_CMD_CAP_DO,
+	},
+};
+
+struct genl_family lockd_nl_family __ro_after_init = {
+	.name		= LOCKD_FAMILY_NAME,
+	.version	= LOCKD_FAMILY_VERSION,
+	.netnsok	= true,
+	.parallel_ops	= true,
+	.module		= THIS_MODULE,
+	.split_ops	= lockd_nl_ops,
+	.n_split_ops	= ARRAY_SIZE(lockd_nl_ops),
+};

fs/lockd/netlink.h

@@ -0,0 +1,19 @@
+/* SPDX-License-Identifier: ((GPL-2.0 WITH Linux-syscall-note) OR BSD-3-Clause) */
+/* Do not edit directly, auto-generated from: */
+/*	Documentation/netlink/specs/lockd.yaml */
+/* YNL-GEN kernel header */
+
+#ifndef _LINUX_LOCKD_GEN_H
+#define _LINUX_LOCKD_GEN_H
+
+#include <net/netlink.h>
+#include <net/genetlink.h>
+
+#include <uapi/linux/lockd_netlink.h>
+
+int lockd_nl_server_set_doit(struct sk_buff *skb, struct genl_info *info);
+int lockd_nl_server_get_doit(struct sk_buff *skb, struct genl_info *info);
+
+extern struct genl_family lockd_nl_family;
+
+#endif /* _LINUX_LOCKD_GEN_H */

fs/lockd/netns.h

@@ -10,6 +10,9 @@ struct lockd_net {
 	unsigned int nlmsvc_users;
 	unsigned long next_gc;
 	unsigned long nrhosts;
+	u32 gracetime;
+	u16 tcp_port;
+	u16 udp_port;
 
 	struct delayed_work grace_period_end;
 	struct lock_manager lockd_manager;

fs/lockd/svc.c

@@ -41,6 +41,7 @@
 
 #include "netns.h"
 #include "procfs.h"
+#include "netlink.h"
 
 #define NLMDBG_FACILITY		NLMDBG_SVC
 #define LOCKD_BUFSIZE		(1024 + NLMSVC_XDRSIZE)
@@ -83,8 +84,14 @@ static const int		nlm_port_min = 0, nlm_port_max = 65535;
 static struct ctl_table_header * nlm_sysctl_table;
 #endif
 
-static unsigned long get_lockd_grace_period(void)
+static unsigned long get_lockd_grace_period(struct net *net)
 {
+	struct lockd_net *ln = net_generic(net, lockd_net_id);
+
+	/* Return the net-ns specific grace period, if there is one */
+	if (ln->gracetime)
+		return ln->gracetime * HZ;
+
 	/* Note: nlm_timeout should always be nonzero */
 	if (nlm_grace_period)
 		return roundup(nlm_grace_period, nlm_timeout) * HZ;
@@ -103,7 +110,7 @@ static void grace_ender(struct work_struct *grace)
 
 static void set_grace_period(struct net *net)
 {
-	unsigned long grace_period = get_lockd_grace_period();
+	unsigned long grace_period = get_lockd_grace_period(net);
 	struct lockd_net *ln = net_generic(net, lockd_net_id);
 
 	locks_start_grace(net, &ln->lockd_manager);
@@ -166,15 +173,16 @@ static int create_lockd_listener(struct svc_serv *serv, const char *name,
 static int create_lockd_family(struct svc_serv *serv, struct net *net,
 			       const int family, const struct cred *cred)
 {
+	struct lockd_net *ln = net_generic(net, lockd_net_id);
 	int err;
 
-	err = create_lockd_listener(serv, "udp", net, family, nlm_udpport,
-			cred);
+	err = create_lockd_listener(serv, "udp", net, family,
+				    ln->udp_port ? ln->udp_port : nlm_udpport, cred);
 	if (err < 0)
 		return err;
 
-	return create_lockd_listener(serv, "tcp", net, family, nlm_tcpport,
-			cred);
+	return create_lockd_listener(serv, "tcp", net, family,
+				     ln->tcp_port ? ln->tcp_port : nlm_tcpport, cred);
 }
 
 /*
@@ -459,9 +467,10 @@ static const struct ctl_table nlm_sysctls[] = {
 	{
 		.procname	= "nsm_local_state",
 		.data		= &nsm_local_state,
-		.maxlen		= sizeof(int),
+		.maxlen		= sizeof(nsm_local_state),
 		.mode		= 0644,
-		.proc_handler	= proc_dointvec,
+		.proc_handler	= proc_douintvec,
+		.extra1		= SYSCTL_ZERO,
 	},
 };
 
@@ -588,13 +597,19 @@ static int __init init_nlm(void)
 	if (err)
 		goto err_pernet;
 
+	err = genl_register_family(&lockd_nl_family);
+	if (err)
+		goto err_netlink;
+
 	err = lockd_create_procfs();
 	if (err)
 		goto err_procfs;
 
 	return 0;
 
 err_procfs:
+	genl_unregister_family(&lockd_nl_family);
+err_netlink:
 	unregister_pernet_subsys(&lockd_net_ops);
 err_pernet:
 #ifdef CONFIG_SYSCTL
@@ -608,6 +623,7 @@ static void __exit exit_nlm(void)
 {
 	/* FIXME: delete all NLM clients */
 	nlm_shutdown_hosts();
+	genl_unregister_family(&lockd_nl_family);
 	lockd_remove_procfs();
 	unregister_pernet_subsys(&lockd_net_ops);
 #ifdef CONFIG_SYSCTL
@@ -710,3 +726,94 @@ static struct svc_program	nlmsvc_program = {
 	.pg_init_request	= svc_generic_init_request,
 	.pg_rpcbind_set		= svc_generic_rpcbind_set,
 };
+
+/**
+ * lockd_nl_server_set_doit - set the lockd server parameters via netlink
+ * @skb: reply buffer
+ * @info: netlink metadata and command arguments
+ *
+ * This updates the per-net values. When updating the values in the init_net
+ * namespace, also update the "legacy" global values.
+ *
+ * Return 0 on success or a negative errno.
+ */
+int lockd_nl_server_set_doit(struct sk_buff *skb, struct genl_info *info)
+{
+	struct net *net = genl_info_net(info);
+	struct lockd_net *ln = net_generic(net, lockd_net_id);
+	const struct nlattr *attr;
+
+	if (GENL_REQ_ATTR_CHECK(info, LOCKD_A_SERVER_GRACETIME))
+		return -EINVAL;
+
+	if (info->attrs[LOCKD_A_SERVER_GRACETIME] ||
+	    info->attrs[LOCKD_A_SERVER_TCP_PORT] ||
+	    info->attrs[LOCKD_A_SERVER_UDP_PORT]) {
+		attr = info->attrs[LOCKD_A_SERVER_GRACETIME];
+		if (attr) {
+			u32 gracetime = nla_get_u32(attr);
+
+			if (gracetime > nlm_grace_period_max)
+				return -EINVAL;
+
+			ln->gracetime = gracetime;
+
+			if (net == &init_net)
+				nlm_grace_period = gracetime;
+		}
+
+		attr = info->attrs[LOCKD_A_SERVER_TCP_PORT];
+		if (attr) {
+			ln->tcp_port = nla_get_u16(attr);
+			if (net == &init_net)
+				nlm_tcpport = ln->tcp_port;
+		}
+
+		attr = info->attrs[LOCKD_A_SERVER_UDP_PORT];
+		if (attr) {
+			ln->udp_port = nla_get_u16(attr);
+			if (net == &init_net)
+				nlm_udpport = ln->udp_port;
+		}
+	}
+	return 0;
+}
+
+/**
+ * lockd_nl_server_get_doit - get lockd server parameters via netlink
+ * @skb: reply buffer
+ * @info: netlink metadata and command arguments
+ *
+ * Return 0 on success or a negative errno.
+ */
+int lockd_nl_server_get_doit(struct sk_buff *skb, struct genl_info *info)
+{
+	struct net *net = genl_info_net(info);
+	struct lockd_net *ln = net_generic(net, lockd_net_id);
+	void *hdr;
+	int err;
+
+	skb = genlmsg_new(GENLMSG_DEFAULT_SIZE, GFP_KERNEL);
+	if (!skb)
+		return -ENOMEM;
+
+	hdr = genlmsg_iput(skb, info);
+	if (!hdr) {
+		err = -EMSGSIZE;
+		goto err_free_msg;
+	}
+
+	err = nla_put_u32(skb, LOCKD_A_SERVER_GRACETIME, ln->gracetime) ||
+	      nla_put_u16(skb, LOCKD_A_SERVER_TCP_PORT, ln->tcp_port) ||
+	      nla_put_u16(skb, LOCKD_A_SERVER_UDP_PORT, ln->udp_port);
+	if (err)
+		goto err_free_msg;
+
+	genlmsg_end(skb, hdr);
+
+	return genlmsg_reply(skb, info);
+err_free_msg:
+	nlmsg_free(skb);
+
+	return err;
+}

fs/nfs/export.c

@@ -154,5 +154,6 @@ const struct export_operations nfs_export_ops = {
 		 EXPORT_OP_CLOSE_BEFORE_UNLINK	|
 		 EXPORT_OP_REMOTE_FS		|
 		 EXPORT_OP_NOATOMIC_ATTR	|
-		 EXPORT_OP_FLUSH_ON_CLOSE,
+		 EXPORT_OP_FLUSH_ON_CLOSE	|
+		 EXPORT_OP_NOLOCKS,
 };

fs/nfs_common/nfsacl.c

@@ -42,7 +42,7 @@ struct nfsacl_encode_desc {
 };
 
 struct nfsacl_simple_acl {
-	struct posix_acl acl;
+	struct posix_acl_hdr acl;
 	struct posix_acl_entry ace[4];
 };
 
@@ -112,7 +112,8 @@ int nfsacl_encode(struct xdr_buf *buf, unsigned int base, struct inode *inode,
 	    xdr_encode_word(buf, base, entries))
 		return -EINVAL;
 	if (encode_entries && acl && acl->a_count == 3) {
-		struct posix_acl *acl2 = &aclbuf.acl;
+		struct posix_acl *acl2 =
+			container_of(&aclbuf.acl, struct posix_acl, hdr);
 
 		/* Avoid the use of posix_acl_alloc().  nfsacl_encode() is
 		 * invoked in contexts where a memory allocation failure is
@@ -177,7 +178,8 @@ bool nfs_stream_encode_acl(struct xdr_stream *xdr, struct inode *inode,
 		return false;
 
 	if (encode_entries && acl && acl->a_count == 3) {
-		struct posix_acl *acl2 = &aclbuf.acl;
+		struct posix_acl *acl2 =
+			container_of(&aclbuf.acl, struct posix_acl, hdr);
 
 		/* Avoid the use of posix_acl_alloc().  nfsacl_encode() is
 		 * invoked in contexts where a memory allocation failure is

fs/nfsd/Kconfig

@@ -172,6 +172,16 @@ config NFSD_LEGACY_CLIENT_TRACKING
 	  recoverydir, or spawn a process directly using a usermodehelper
 	  upcall.
 
-	  These legacy client tracking methods have proven to be probelmatic
+	  These legacy client tracking methods have proven to be problematic
 	  and will be removed in the future. Say Y here if you need support
 	  for them in the interim.
+
+config NFSD_V4_DELEG_TIMESTAMPS
+	bool "Support delegated timestamps"
+	depends on NFSD_V4
+	default n
+	help
+	  NFSD implements delegated timestamps according to
+	  draft-ietf-nfsv4-delstid-08 "Extending the Opening of Files". This
+	  is currently an experimental feature and is therefore left disabled
+	  by default.