21 files changed, 302 insertions, 144 deletions

diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c
index 7a501e73d880..1abc7ed2990e 100644
--- a/fs/btrfs/file.c
+++ b/fs/btrfs/file.c
@@ -2019,13 +2019,14 @@ out:
 	else
 		btrfs_delalloc_release_space(inode, data_reserved, page_start,
 					     reserved_space, true);
-	extent_changeset_free(data_reserved);
 out_noreserve:
 	if (only_release_metadata)
 		btrfs_check_nocow_unlock(inode);
 
 	sb_end_pagefault(inode->vfs_inode.i_sb);
 
+	extent_changeset_free(data_reserved);
+
 	if (ret < 0)
 		return vmf_error(ret);
 
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index c4bee47829ed..317db7d10a21 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -256,6 +256,7 @@ static void print_data_reloc_error(const struct btrfs_inode *inode, u64 file_off
 	if (ret < 0) {
 		btrfs_err_rl(fs_info, "failed to lookup extent item for logical %llu: %d",
 			     logical, ret);
+		btrfs_release_path(&path);
 		return;
 	}
 	eb = path.nodes[0];
diff --git a/fs/btrfs/qgroup.c b/fs/btrfs/qgroup.c
index 9e2b53e90dcb..d9d8d9968a58 100644
--- a/fs/btrfs/qgroup.c
+++ b/fs/btrfs/qgroup.c
@@ -1243,14 +1243,7 @@ out:
 		btrfs_end_transaction(trans);
 	else if (trans)
 		ret = btrfs_end_transaction(trans);
-
-	/*
-	 * At this point we either failed at allocating prealloc, or we
-	 * succeeded and passed the ownership to it to add_qgroup_rb(). In any
-	 * case, this needs to be NULL or there is something wrong.
-	 */
-	ASSERT(prealloc == NULL);
-
+	kfree(prealloc);
 	return ret;
 }
 
@@ -1682,12 +1675,7 @@ int btrfs_create_qgroup(struct btrfs_trans_handle *trans, u64 qgroupid)
 	ret = btrfs_sysfs_add_one_qgroup(fs_info, qgroup);
 out:
 	mutex_unlock(&fs_info->qgroup_ioctl_lock);
-	/*
-	 * At this point we either failed at allocating prealloc, or we
-	 * succeeded and passed the ownership to it to add_qgroup_rb(). In any
-	 * case, this needs to be NULL or there is something wrong.
-	 */
-	ASSERT(prealloc == NULL);
+	kfree(prealloc);
 	return ret;
 }
 
@@ -3279,7 +3267,7 @@ int btrfs_qgroup_inherit(struct btrfs_trans_handle *trans, u64 srcid,
 	struct btrfs_root *quota_root;
 	struct btrfs_qgroup *srcgroup;
 	struct btrfs_qgroup *dstgroup;
-	struct btrfs_qgroup *prealloc = NULL;
+	struct btrfs_qgroup *prealloc;
 	struct btrfs_qgroup_list **qlist_prealloc = NULL;
 	bool free_inherit = false;
 	bool need_rescan = false;
@@ -3520,14 +3508,7 @@ out:
 	}
 	if (free_inherit)
 		kfree(inherit);
-
-	/*
-	 * At this point we either failed at allocating prealloc, or we
-	 * succeeded and passed the ownership to it to add_qgroup_rb(). In any
-	 * case, this needs to be NULL or there is something wrong.
-	 */
-	ASSERT(prealloc == NULL);
-
+	kfree(prealloc);
 	return ret;
 }
 
diff --git a/fs/btrfs/tests/qgroup-tests.c b/fs/btrfs/tests/qgroup-tests.c
index 05cfda8af422..e9124605974b 100644
--- a/fs/btrfs/tests/qgroup-tests.c
+++ b/fs/btrfs/tests/qgroup-tests.c
@@ -187,7 +187,6 @@ static int remove_extent_ref(struct btrfs_root *root, u64 bytenr,
 	ret = btrfs_search_slot(&trans, root, &key, path, -1, 1);
 	if (ret) {
 		test_err("couldn't find backref %d", ret);
-		btrfs_free_path(path);
 		return ret;
 	}
 	btrfs_del_item(&trans, root, path);
diff --git a/fs/btrfs/tree-log.c b/fs/btrfs/tree-log.c
index fff37c8d96a4..31edc93a383e 100644
--- a/fs/btrfs/tree-log.c
+++ b/fs/btrfs/tree-log.c
@@ -5865,14 +5865,6 @@ static int log_new_dir_dentries(struct btrfs_trans_handle *trans,
 	struct btrfs_inode *curr_inode = start_inode;
 	int ret = 0;
 
-	/*
-	 * If we are logging a new name, as part of a link or rename operation,
-	 * don't bother logging new dentries, as we just want to log the names
-	 * of an inode and that any new parents exist.
-	 */
-	if (ctx->logging_new_name)
-		return 0;
-
 	path = btrfs_alloc_path();
 	if (!path)
 		return -ENOMEM;
@@ -6051,6 +6043,33 @@ static int conflicting_inode_is_dir(struct btrfs_root *root, u64 ino,
 	return ret;
 }
 
+static bool can_log_conflicting_inode(const struct btrfs_trans_handle *trans,
+				      const struct btrfs_inode *inode)
+{
+	if (!S_ISDIR(inode->vfs_inode.i_mode))
+		return true;
+
+	if (inode->last_unlink_trans < trans->transid)
+		return true;
+
+	/*
+	 * If this is a directory and its unlink_trans is not from a past
+	 * transaction then we must fallback to a transaction commit in order
+	 * to avoid getting a directory with 2 hard links after log replay.
+	 *
+	 * This happens if a directory A is renamed, moved from one parent
+	 * directory to another one, a new file is created in the old parent
+	 * directory with the old name of our directory A, the new file is
+	 * fsynced, then we moved the new file to some other parent directory
+	 * and fsync again the new file. This results in a log tree where we
+	 * logged that directory A existed, with the INODE_REF item for the
+	 * new location but without having logged its old parent inode, so
+	 * that on log replay we add a new link for the new location but the
+	 * old link remains, resulting in a link count of 2.
+	 */
+	return false;
+}
+
 static int add_conflicting_inode(struct btrfs_trans_handle *trans,
 				 struct btrfs_root *root,
 				 struct btrfs_path *path,
@@ -6154,6 +6173,11 @@ static int add_conflicting_inode(struct btrfs_trans_handle *trans,
 		return 0;
 	}
 
+	if (!can_log_conflicting_inode(trans, inode)) {
+		btrfs_add_delayed_iput(inode);
+		return BTRFS_LOG_FORCE_COMMIT;
+	}
+
 	btrfs_add_delayed_iput(inode);
 
 	ino_elem = kmalloc(sizeof(*ino_elem), GFP_NOFS);
@@ -6218,6 +6242,12 @@ static int log_conflicting_inodes(struct btrfs_trans_handle *trans,
 				break;
 			}
 
+			if (!can_log_conflicting_inode(trans, inode)) {
+				btrfs_add_delayed_iput(inode);
+				ret = BTRFS_LOG_FORCE_COMMIT;
+				break;
+			}
+
 			/*
 			 * Always log the directory, we cannot make this
 			 * conditional on need_log_inode() because the directory
diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index ae1742a35e76..13c514684cfb 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -7128,6 +7128,7 @@ static struct btrfs_fs_devices *open_seed_devices(struct btrfs_fs_info *fs_info,
 
 		fs_devices->seeding = true;
 		fs_devices->opened = 1;
+		list_add(&fs_devices->seed_list, &fs_info->fs_devices->seed_list);
 		return fs_devices;
 	}
 
diff --git a/fs/file_attr.c b/fs/file_attr.c
index 4c4916632f11..13cdb31a3e94 100644
--- a/fs/file_attr.c
+++ b/fs/file_attr.c
@@ -2,6 +2,7 @@
 #include <linux/fs.h>
 #include <linux/security.h>
 #include <linux/fscrypt.h>
+#include <linux/fsnotify.h>
 #include <linux/fileattr.h>
 #include <linux/export.h>
 #include <linux/syscalls.h>
@@ -298,6 +299,7 @@ int vfs_fileattr_set(struct mnt_idmap *idmap, struct dentry *dentry,
 		err = inode->i_op->fileattr_set(idmap, dentry, fa);
 		if (err)
 			goto out;
+		fsnotify_xattr(dentry);
 	}
 
 out:
diff --git a/fs/libfs.c b/fs/libfs.c
index 9264523be85c..591eb649ebba 100644
--- a/fs/libfs.c
+++ b/fs/libfs.c
@@ -346,22 +346,22 @@ void simple_offset_remove(struct offset_ctx *octx, struct dentry *dentry)
  * User space expects the directory offset value of the replaced
  * (new) directory entry to be unchanged after a rename.
  *
- * Returns zero on success, a negative errno value on failure.
+ * Caller must have grabbed a slot for new_dentry in the maple_tree
+ * associated with new_dir, even if dentry is negative.
  */
-int simple_offset_rename(struct inode *old_dir, struct dentry *old_dentry,
-			 struct inode *new_dir, struct dentry *new_dentry)
+void simple_offset_rename(struct inode *old_dir, struct dentry *old_dentry,
+			  struct inode *new_dir, struct dentry *new_dentry)
 {
 	struct offset_ctx *old_ctx = old_dir->i_op->get_offset_ctx(old_dir);
 	struct offset_ctx *new_ctx = new_dir->i_op->get_offset_ctx(new_dir);
 	long new_offset = dentry2offset(new_dentry);
 
-	simple_offset_remove(old_ctx, old_dentry);
+	if (WARN_ON(!new_offset))
+		return;
 
-	if (new_offset) {
-		offset_set(new_dentry, 0);
-		return simple_offset_replace(new_ctx, old_dentry, new_offset);
-	}
-	return simple_offset_add(new_ctx, old_dentry);
+	simple_offset_remove(old_ctx, old_dentry);
+	offset_set(new_dentry, 0);
+	WARN_ON(simple_offset_replace(new_ctx, old_dentry, new_offset));
 }
 
 /**
@@ -388,31 +388,23 @@ int simple_offset_rename_exchange(struct inode *old_dir,
 	long new_index = dentry2offset(new_dentry);
 	int ret;
 
-	simple_offset_remove(old_ctx, old_dentry);
-	simple_offset_remove(new_ctx, new_dentry);
+	if (WARN_ON(!old_index || !new_index))
+		return -EINVAL;
 
-	ret = simple_offset_replace(new_ctx, old_dentry, new_index);
-	if (ret)
-		goto out_restore;
+	ret = mtree_store(&new_ctx->mt, new_index, old_dentry, GFP_KERNEL);
+	if (WARN_ON(ret))
+		return ret;
 
-	ret = simple_offset_replace(old_ctx, new_dentry, old_index);
-	if (ret) {
-		simple_offset_remove(new_ctx, old_dentry);
-		goto out_restore;
+	ret = mtree_store(&old_ctx->mt, old_index, new_dentry, GFP_KERNEL);
+	if (WARN_ON(ret)) {
+		mtree_store(&new_ctx->mt, new_index, new_dentry, GFP_KERNEL);
+		return ret;
 	}
 
-	ret = simple_rename_exchange(old_dir, old_dentry, new_dir, new_dentry);
-	if (ret) {
-		simple_offset_remove(new_ctx, old_dentry);
-		simple_offset_remove(old_ctx, new_dentry);
-		goto out_restore;
-	}
+	offset_set(old_dentry, new_index);
+	offset_set(new_dentry, old_index);
+	simple_rename_exchange(old_dir, old_dentry, new_dir, new_dentry);
 	return 0;
-
-out_restore:
-	(void)simple_offset_replace(old_ctx, old_dentry, old_index);
-	(void)simple_offset_replace(new_ctx, new_dentry, new_index);
-	return ret;
 }
 
 /**
diff --git a/fs/notify/fsnotify.c b/fs/notify/fsnotify.c
index d27ff5e5f165..71bd44e5ab6d 100644
--- a/fs/notify/fsnotify.c
+++ b/fs/notify/fsnotify.c
@@ -270,8 +270,15 @@ int __fsnotify_parent(struct dentry *dentry, __u32 mask, const void *data,
 	/*
 	 * Include parent/name in notification either if some notification
 	 * groups require parent info or the parent is interested in this event.
+	 * The parent interest in ACCESS/MODIFY events does not apply to special
+	 * files, where read/write are not on the filesystem of the parent and
+	 * events can provide an undesirable side-channel for information
+	 * exfiltration.
 	 */
-	parent_interested = mask & p_mask & ALL_FSNOTIFY_EVENTS;
+	parent_interested = mask & p_mask & ALL_FSNOTIFY_EVENTS &&
+			    !(data_type == FSNOTIFY_EVENT_PATH &&
+			      d_is_special(dentry) &&
+			      (mask & (FS_ACCESS | FS_MODIFY)));
 	if (parent_needed || parent_interested) {
 		/* When notifying parent, child should be passed as data */
 		WARN_ON_ONCE(inode != fsnotify_data_inode(data, data_type));
diff --git a/fs/smb/client/cifsfs.h b/fs/smb/client/cifsfs.h
index e9534258d1ef..75d372ceb655 100644
--- a/fs/smb/client/cifsfs.h
+++ b/fs/smb/client/cifsfs.h
@@ -145,6 +145,6 @@ extern const struct export_operations cifs_export_ops;
 #endif /* CONFIG_CIFS_NFSD_EXPORT */
 
 /* when changing internal version - update following two lines at same time */
-#define SMB3_PRODUCT_BUILD 57
-#define CIFS_VERSION   "2.57"
+#define SMB3_PRODUCT_BUILD 58
+#define CIFS_VERSION   "2.58"
 #endif				/* _CIFSFS_H */
diff --git a/fs/smb/client/cifspdu.h b/fs/smb/client/cifspdu.h
index eeb4011cb217..fdd84369e7b8 100644
--- a/fs/smb/client/cifspdu.h
+++ b/fs/smb/client/cifspdu.h
@@ -12,7 +12,7 @@
 #include <net/sock.h>
 #include <linux/unaligned.h>
 #include "../common/smbfsctl.h"
-#include "../common/smb2pdu.h"
+#include "../common/smb1pdu.h"
 
 #define CIFS_PROT   0
 #define POSIX_PROT  (CIFS_PROT+1)
diff --git a/fs/smb/common/smb1pdu.h b/fs/smb/common/smb1pdu.h
new file mode 100644
index 000000000000..df6d4e11ae92
--- /dev/null
+++ b/fs/smb/common/smb1pdu.h
@@ -0,0 +1,56 @@
+/* SPDX-License-Identifier: LGPL-2.1 */
+/*
+ *
+ *   Copyright (C) International Business Machines  Corp., 2002,2009
+ *                 2018 Samsung Electronics Co., Ltd.
+ *   Author(s): Steve French <sfrench@us.ibm.com>
+ *              Namjae Jeon <linkinjeon@kernel.org>
+ *
+ */
+
+#ifndef _COMMON_SMB1_PDU_H
+#define _COMMON_SMB1_PDU_H
+
+#define SMB1_PROTO_NUMBER		cpu_to_le32(0x424d53ff)
+
+/*
+ * See MS-CIFS 2.2.3.1
+ *     MS-SMB 2.2.3.1
+ */
+struct smb_hdr {
+	__u8 Protocol[4];
+	__u8 Command;
+	union {
+		struct {
+			__u8 ErrorClass;
+			__u8 Reserved;
+			__le16 Error;
+		} __packed DosError;
+		__le32 CifsError;
+	} __packed Status;
+	__u8 Flags;
+	__le16 Flags2;		/* note: le */
+	__le16 PidHigh;
+	union {
+		struct {
+			__le32 SequenceNumber;  /* le */
+			__u32 Reserved; /* zero */
+		} __packed Sequence;
+		__u8 SecuritySignature[8];	/* le */
+	} __packed Signature;
+	__u8 pad[2];
+	__u16 Tid;
+	__le16 Pid;
+	__u16 Uid;
+	__le16 Mid;
+	__u8 WordCount;
+} __packed;
+
+/* See MS-CIFS 2.2.4.52.1 */
+typedef struct smb_negotiate_req {
+	struct smb_hdr hdr;	/* wct = 0 */
+	__le16 ByteCount;
+	unsigned char DialectsArray[];
+} __packed SMB_NEGOTIATE_REQ;
+
+#endif /* _COMMON_SMB1_PDU_H */
diff --git a/fs/smb/common/smb2pdu.h b/fs/smb/common/smb2pdu.h
index 3c8d8a4e7439..f5ebbe31384a 100644
--- a/fs/smb/common/smb2pdu.h
+++ b/fs/smb/common/smb2pdu.h
@@ -1293,6 +1293,7 @@ struct create_durable_handle_reconnect_v2 {
 	struct create_context_hdr ccontext;
 	__u8   Name[8];
 	struct durable_reconnect_context_v2 dcontext;
+	__u8 Pad[4];
 } __packed;
 
 /* See MS-SMB2 2.2.14.2.12 */
@@ -1985,39 +1986,6 @@ struct smb2_lease_ack {
 	__le64 LeaseDuration;
 } __packed;
 
-/*
- * See MS-CIFS 2.2.3.1
- *     MS-SMB 2.2.3.1
- */
-struct smb_hdr {
-	__u8 Protocol[4];
-	__u8 Command;
-	union {
-		struct {
-			__u8 ErrorClass;
-			__u8 Reserved;
-			__le16 Error;
-		} __packed DosError;
-		__le32 CifsError;
-	} __packed Status;
-	__u8 Flags;
-	__le16 Flags2;		/* note: le */
-	__le16 PidHigh;
-	union {
-		struct {
-			__le32 SequenceNumber;  /* le */
-			__u32 Reserved; /* zero */
-		} __packed Sequence;
-		__u8 SecuritySignature[8];	/* le */
-	} __packed Signature;
-	__u8 pad[2];
-	__u16 Tid;
-	__le16 Pid;
-	__u16 Uid;
-	__le16 Mid;
-	__u8 WordCount;
-} __packed;
-
 #define OP_BREAK_STRUCT_SIZE_20		24
 #define OP_BREAK_STRUCT_SIZE_21		36
 
@@ -2122,11 +2090,4 @@ struct smb_hdr {
 #define SET_MINIMUM_RIGHTS (FILE_READ_EA | FILE_READ_ATTRIBUTES \
 				| READ_CONTROL | SYNCHRONIZE)
 
-/* See MS-CIFS 2.2.4.52.1 */
-typedef struct smb_negotiate_req {
-	struct smb_hdr hdr;	/* wct = 0 */
-	__le16 ByteCount;
-	unsigned char DialectsArray[];
-} __packed SMB_NEGOTIATE_REQ;
-
 #endif				/* _COMMON_SMB2PDU_H */
diff --git a/fs/smb/common/smbdirect/smbdirect_socket.h b/fs/smb/common/smbdirect/smbdirect_socket.h
index 384b19177e1c..ee4c2726771a 100644
--- a/fs/smb/common/smbdirect/smbdirect_socket.h
+++ b/fs/smb/common/smbdirect/smbdirect_socket.h
@@ -133,6 +133,14 @@ struct smbdirect_socket {
 	struct smbdirect_socket_parameters parameters;
 
 	/*
+	 * The state for connect/negotiation
+	 */
+	struct {
+		spinlock_t lock;
+		struct work_struct work;
+	} connect;
+
+	/*
 	 * The state for keepalive and timeout handling
 	 */
 	struct {
@@ -353,6 +361,10 @@ static __always_inline void smbdirect_socket_init(struct smbdirect_socket *sc)
 	INIT_WORK(&sc->disconnect_work, __smbdirect_socket_disabled_work);
 	disable_work_sync(&sc->disconnect_work);
 
+	spin_lock_init(&sc->connect.lock);
+	INIT_WORK(&sc->connect.work, __smbdirect_socket_disabled_work);
+	disable_work_sync(&sc->connect.work);
+
 	INIT_WORK(&sc->idle.immediate_work, __smbdirect_socket_disabled_work);
 	disable_work_sync(&sc->idle.immediate_work);
 	INIT_DELAYED_WORK(&sc->idle.timer_work, __smbdirect_socket_disabled_work);
diff --git a/fs/smb/common/smbglob.h b/fs/smb/common/smbglob.h
index 9562845a5617..4e33d91cdc9d 100644
--- a/fs/smb/common/smbglob.h
+++ b/fs/smb/common/smbglob.h
@@ -11,8 +11,6 @@
 #ifndef _COMMON_SMB_GLOB_H
 #define _COMMON_SMB_GLOB_H
 
-#define SMB1_PROTO_NUMBER		cpu_to_le32(0x424d53ff)
-
 struct smb_version_values {
 	char		*version_string;
 	__u16		protocol_id;
diff --git a/fs/smb/server/mgmt/user_session.c b/fs/smb/server/mgmt/user_session.c
index 1c181ef99929..7d880ff34402 100644
--- a/fs/smb/server/mgmt/user_session.c
+++ b/fs/smb/server/mgmt/user_session.c
@@ -325,8 +325,10 @@ struct ksmbd_session *ksmbd_session_lookup_all(struct ksmbd_conn *conn,
 	sess = ksmbd_session_lookup(conn, id);
 	if (!sess && conn->binding)
 		sess = ksmbd_session_lookup_slowpath(id);
-	if (sess && sess->state != SMB2_SESSION_VALID)
+	if (sess && sess->state != SMB2_SESSION_VALID) {
+		ksmbd_user_session_put(sess);
 		sess = NULL;
+	}
 	return sess;
 }
 
diff --git a/fs/smb/server/smb2pdu.c b/fs/smb/server/smb2pdu.c
index 27f87a13f20a..8aa483800014 100644
--- a/fs/smb/server/smb2pdu.c
+++ b/fs/smb/server/smb2pdu.c
@@ -2363,7 +2363,7 @@ static int smb2_set_ea(struct smb2_ea_info *eabuf, unsigned int buf_len,
 	int rc = 0;
 	unsigned int next = 0;
 
-	if (buf_len < sizeof(struct smb2_ea_info) + eabuf->EaNameLength +
+	if (buf_len < sizeof(struct smb2_ea_info) + eabuf->EaNameLength + 1 +
 			le16_to_cpu(eabuf->EaValueLength))
 		return -EINVAL;
 
@@ -2440,7 +2440,7 @@ next:
 			break;
 		}
 
-		if (buf_len < sizeof(struct smb2_ea_info) + eabuf->EaNameLength +
+		if (buf_len < sizeof(struct smb2_ea_info) + eabuf->EaNameLength + 1 +
 				le16_to_cpu(eabuf->EaValueLength)) {
 			rc = -EINVAL;
 			break;
diff --git a/fs/smb/server/smb_common.h b/fs/smb/server/smb_common.h
index 067b45048c73..95bf1465387b 100644
--- a/fs/smb/server/smb_common.h
+++ b/fs/smb/server/smb_common.h
@@ -10,6 +10,7 @@
 
 #include "glob.h"
 #include "../common/smbglob.h"
+#include "../common/smb1pdu.h"
 #include "../common/smb2pdu.h"
 #include "../common/fscc.h"
 #include "smb2pdu.h"
diff --git a/fs/smb/server/smbacl.c b/fs/smb/server/smbacl.c
index 5aa7a66334d9..05598d994a68 100644
--- a/fs/smb/server/smbacl.c
+++ b/fs/smb/server/smbacl.c
@@ -1307,9 +1307,6 @@ int smb_check_perm_dacl(struct ksmbd_conn *conn, const struct path *path,
 			granted |= le32_to_cpu(ace->access_req);
 			ace = (struct smb_ace *)((char *)ace + le16_to_cpu(ace->size));
 		}
-
-		if (!pdacl->num_aces)
-			granted = GENERIC_ALL_FLAGS;
 	}
 
 	if (!uid)
diff --git a/fs/smb/server/transport_rdma.c b/fs/smb/server/transport_rdma.c
index 4e7ab8d9314f..f585359684d4 100644
--- a/fs/smb/server/transport_rdma.c
+++ b/fs/smb/server/transport_rdma.c
@@ -242,6 +242,7 @@ static void smb_direct_disconnect_rdma_work(struct work_struct *work)
 	 * disable[_delayed]_work_sync()
 	 */
 	disable_work(&sc->disconnect_work);
+	disable_work(&sc->connect.work);
 	disable_work(&sc->recv_io.posted.refill_work);
 	disable_delayed_work(&sc->idle.timer_work);
 	disable_work(&sc->idle.immediate_work);
@@ -297,6 +298,7 @@ smb_direct_disconnect_rdma_connection(struct smbdirect_socket *sc)
 	 * not queued again but here we don't block and avoid
 	 * disable[_delayed]_work_sync()
 	 */
+	disable_work(&sc->connect.work);
 	disable_work(&sc->recv_io.posted.refill_work);
 	disable_work(&sc->idle.immediate_work);
 	disable_delayed_work(&sc->idle.timer_work);
@@ -467,6 +469,7 @@ static void free_transport(struct smb_direct_transport *t)
 	 */
 	smb_direct_disconnect_wake_up_all(sc);
 
+	disable_work_sync(&sc->connect.work);
 	disable_work_sync(&sc->recv_io.posted.refill_work);
 	disable_delayed_work_sync(&sc->idle.timer_work);
 	disable_work_sync(&sc->idle.immediate_work);
@@ -635,28 +638,8 @@ static void recv_done(struct ib_cq *cq, struct ib_wc *wc)
 
 	switch (sc->recv_io.expected) {
 	case SMBDIRECT_EXPECT_NEGOTIATE_REQ:
-		if (wc->byte_len < sizeof(struct smbdirect_negotiate_req)) {
-			put_recvmsg(sc, recvmsg);
-			smb_direct_disconnect_rdma_connection(sc);
-			return;
-		}
-		sc->recv_io.reassembly.full_packet_received = true;
-		/*
-		 * Some drivers (at least mlx5_ib) might post a
-		 * recv completion before RDMA_CM_EVENT_ESTABLISHED,
-		 * we need to adjust our expectation in that case.
-		 */
-		if (!sc->first_error && sc->status == SMBDIRECT_SOCKET_RDMA_CONNECT_RUNNING)
-			sc->status = SMBDIRECT_SOCKET_NEGOTIATE_NEEDED;
-		if (SMBDIRECT_CHECK_STATUS_WARN(sc, SMBDIRECT_SOCKET_NEGOTIATE_NEEDED)) {
-			put_recvmsg(sc, recvmsg);
-			smb_direct_disconnect_rdma_connection(sc);
-			return;
-		}
-		sc->status = SMBDIRECT_SOCKET_NEGOTIATE_RUNNING;
-		enqueue_reassembly(sc, recvmsg, 0);
-		wake_up(&sc->status_wait);
-		return;
+		/* see smb_direct_negotiate_recv_done */
+		break;
 	case SMBDIRECT_EXPECT_DATA_TRANSFER: {
 		struct smbdirect_data_transfer *data_transfer =
 			(struct smbdirect_data_transfer *)recvmsg->packet;
@@ -742,6 +725,126 @@ static void recv_done(struct ib_cq *cq, struct ib_wc *wc)
 	smb_direct_disconnect_rdma_connection(sc);
 }
 
+static void smb_direct_negotiate_recv_work(struct work_struct *work);
+
+static void smb_direct_negotiate_recv_done(struct ib_cq *cq, struct ib_wc *wc)
+{
+	struct smbdirect_recv_io *recv_io =
+		container_of(wc->wr_cqe, struct smbdirect_recv_io, cqe);
+	struct smbdirect_socket *sc = recv_io->socket;
+	unsigned long flags;
+
+	/*
+	 * reset the common recv_done for later reuse.
+	 */
+	recv_io->cqe.done = recv_done;
+
+	if (wc->status != IB_WC_SUCCESS || wc->opcode != IB_WC_RECV) {
+		put_recvmsg(sc, recv_io);
+		if (wc->status != IB_WC_WR_FLUSH_ERR) {
+			pr_err("Negotiate Recv error. status='%s (%d)' opcode=%d\n",
+			       ib_wc_status_msg(wc->status), wc->status,
+			       wc->opcode);
+			smb_direct_disconnect_rdma_connection(sc);
+		}
+		return;
+	}
+
+	ksmbd_debug(RDMA, "Negotiate Recv completed. status='%s (%d)', opcode=%d\n",
+		    ib_wc_status_msg(wc->status), wc->status,
+		    wc->opcode);
+
+	ib_dma_sync_single_for_cpu(sc->ib.dev,
+				   recv_io->sge.addr,
+				   recv_io->sge.length,
+				   DMA_FROM_DEVICE);
+
+	/*
+	 * This is an internal error!
+	 */
+	if (WARN_ON_ONCE(sc->recv_io.expected != SMBDIRECT_EXPECT_NEGOTIATE_REQ)) {
+		put_recvmsg(sc, recv_io);
+		smb_direct_disconnect_rdma_connection(sc);
+		return;
+	}
+
+	/*
+	 * Don't reset timer to the keepalive interval in
+	 * this will be done in smb_direct_negotiate_recv_work.
+	 */
+
+	/*
+	 * Only remember the recv_io if it has enough bytes,
+	 * this gives smb_direct_negotiate_recv_work enough
+	 * information in order to disconnect if it was not
+	 * valid.
+	 */
+	sc->recv_io.reassembly.full_packet_received = true;
+	if (wc->byte_len >= sizeof(struct smbdirect_negotiate_req))
+		enqueue_reassembly(sc, recv_io, 0);
+	else
+		put_recvmsg(sc, recv_io);
+
+	/*
+	 * Some drivers (at least mlx5_ib and irdma in roce mode)
+	 * might post a recv completion before RDMA_CM_EVENT_ESTABLISHED,
+	 * we need to adjust our expectation in that case.
+	 *
+	 * So we defer further processing of the negotiation
+	 * to smb_direct_negotiate_recv_work().
+	 *
+	 * If we are already in SMBDIRECT_SOCKET_NEGOTIATE_NEEDED
+	 * we queue the work directly otherwise
+	 * smb_direct_cm_handler() will do it, when
+	 * RDMA_CM_EVENT_ESTABLISHED arrived.
+	 */
+	spin_lock_irqsave(&sc->connect.lock, flags);
+	if (!sc->first_error) {
+		INIT_WORK(&sc->connect.work, smb_direct_negotiate_recv_work);
+		if (sc->status == SMBDIRECT_SOCKET_NEGOTIATE_NEEDED)
+			queue_work(sc->workqueue, &sc->connect.work);
+	}
+	spin_unlock_irqrestore(&sc->connect.lock, flags);
+}
+
+static void smb_direct_negotiate_recv_work(struct work_struct *work)
+{
+	struct smbdirect_socket *sc =
+		container_of(work, struct smbdirect_socket, connect.work);
+	const struct smbdirect_socket_parameters *sp = &sc->parameters;
+	struct smbdirect_recv_io *recv_io;
+
+	if (sc->first_error)
+		return;
+
+	ksmbd_debug(RDMA, "Negotiate Recv Work running\n");
+
+	/*
+	 * Reset timer to the keepalive interval in
+	 * order to trigger our next keepalive message.
+	 */
+	sc->idle.keepalive = SMBDIRECT_KEEPALIVE_NONE;
+	mod_delayed_work(sc->workqueue, &sc->idle.timer_work,
+			 msecs_to_jiffies(sp->keepalive_interval_msec));
+
+	/*
+	 * If smb_direct_negotiate_recv_done() detected an
+	 * invalid request we want to disconnect.
+	 */
+	recv_io = get_first_reassembly(sc);
+	if (!recv_io) {
+		smb_direct_disconnect_rdma_connection(sc);
+		return;
+	}
+
+	if (SMBDIRECT_CHECK_STATUS_WARN(sc, SMBDIRECT_SOCKET_NEGOTIATE_NEEDED)) {
+		smb_direct_disconnect_rdma_connection(sc);
+		return;
+	}
+	sc->status = SMBDIRECT_SOCKET_NEGOTIATE_RUNNING;
+	wake_up(&sc->status_wait);
+}
+
 static int smb_direct_post_recv(struct smbdirect_socket *sc,
 				struct smbdirect_recv_io *recvmsg)
 {
@@ -758,7 +861,6 @@ static int smb_direct_post_recv(struct smbdirect_socket *sc,
 		return ret;
 	recvmsg->sge.length = sp->max_recv_size;
 	recvmsg->sge.lkey = sc->ib.pd->local_dma_lkey;
-	recvmsg->cqe.done = recv_done;
 
 	wr.wr_cqe = &recvmsg->cqe;
 	wr.next = NULL;
@@ -1732,6 +1834,7 @@ static int smb_direct_cm_handler(struct rdma_cm_id *cm_id,
 				 struct rdma_cm_event *event)
 {
 	struct smbdirect_socket *sc = cm_id->context;
+	unsigned long flags;
 
 	ksmbd_debug(RDMA, "RDMA CM event. cm_id=%p event=%s (%d)\n",
 		    cm_id, rdma_event_msg(event->event), event->event);
@@ -1739,18 +1842,27 @@ static int smb_direct_cm_handler(struct rdma_cm_id *cm_id,
 	switch (event->event) {
 	case RDMA_CM_EVENT_ESTABLISHED: {
 		/*
-		 * Some drivers (at least mlx5_ib) might post a
-		 * recv completion before RDMA_CM_EVENT_ESTABLISHED,
+		 * Some drivers (at least mlx5_ib and irdma in roce mode)
+		 * might post a recv completion before RDMA_CM_EVENT_ESTABLISHED,
 		 * we need to adjust our expectation in that case.
 		 *
-		 * As we already started the negotiation, we just
-		 * ignore RDMA_CM_EVENT_ESTABLISHED here.
+		 * If smb_direct_negotiate_recv_done was called first
+		 * it initialized sc->connect.work only for us to
+		 * start, so that we turned into
+		 * SMBDIRECT_SOCKET_NEGOTIATE_NEEDED, before
+		 * smb_direct_negotiate_recv_work() runs.
+		 *
+		 * If smb_direct_negotiate_recv_done didn't happen
+		 * yet. sc->connect.work is still be disabled and
+		 * queue_work() is a no-op.
 		 */
-		if (!sc->first_error && sc->status > SMBDIRECT_SOCKET_RDMA_CONNECT_RUNNING)
-			break;
 		if (SMBDIRECT_CHECK_STATUS_DISCONNECT(sc, SMBDIRECT_SOCKET_RDMA_CONNECT_RUNNING))
 			break;
 		sc->status = SMBDIRECT_SOCKET_NEGOTIATE_NEEDED;
+		spin_lock_irqsave(&sc->connect.lock, flags);
+		if (!sc->first_error)
+			queue_work(sc->workqueue, &sc->connect.work);
+		spin_unlock_irqrestore(&sc->connect.lock, flags);
 		wake_up(&sc->status_wait);
 		break;
 	}
@@ -1921,6 +2033,7 @@ static int smb_direct_prepare_negotiation(struct smbdirect_socket *sc)
 	recvmsg = get_free_recvmsg(sc);
 	if (!recvmsg)
 		return -ENOMEM;
+	recvmsg->cqe.done = smb_direct_negotiate_recv_done;
 
 	ret = smb_direct_post_recv(sc, recvmsg);
 	if (ret) {
@@ -2339,6 +2452,7 @@ respond:
 
 static int smb_direct_connect(struct smbdirect_socket *sc)
 {
+	struct smbdirect_recv_io *recv_io;
 	int ret;
 
 	ret = smb_direct_init_params(sc);
@@ -2353,6 +2467,9 @@ static int smb_direct_connect(struct smbdirect_socket *sc)
 		return ret;
 	}
 
+	list_for_each_entry(recv_io, &sc->recv_io.free.list, list)
+		recv_io->cqe.done = recv_done;
+
 	ret = smb_direct_create_qpair(sc);
 	if (ret) {
 		pr_err("Can't accept RDMA client: %d\n", ret);
diff --git a/fs/smb/server/vfs.c b/fs/smb/server/vfs.c
index 98b0eb966d91..f891344bd76b 100644
--- a/fs/smb/server/vfs.c
+++ b/fs/smb/server/vfs.c
@@ -702,7 +702,7 @@ retry:
 	rd.old_parent		= NULL;
 	rd.new_parent		= new_path.dentry;
 	rd.flags		= flags;
-	rd.delegated_inode	= NULL,
+	rd.delegated_inode	= NULL;
 	err = start_renaming_dentry(&rd, lookup_flags, old_child, &new_last);
 	if (err)
 		goto out_drop_write;