Merge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6

36 files changed, 968 insertions, 420 deletions

diff --git a/fs/Kconfig b/fs/Kconfig
index 2524629dc83..f9b5842c8d2 100644
--- a/fs/Kconfig
+++ b/fs/Kconfig
@@ -842,6 +842,12 @@ config TMPFS
 config HUGETLBFS
 	bool "HugeTLB file system support"
 	depends X86 || IA64 || PPC64 || SPARC64 || SUPERH || BROKEN
+	help
+	  hugetlbfs is a filesystem backing for HugeTLB pages, based on
+	  ramfs. For architectures that support it, say Y here and read
+	  <file:Documentation/vm/hugetlbpage.txt> for details.
+
+	  If unsure, say N.
 
 config HUGETLB_PAGE
 	def_bool HUGETLBFS
diff --git a/fs/block_dev.c b/fs/block_dev.c
index af88c43043d..f5958f413bd 100644
--- a/fs/block_dev.c
+++ b/fs/block_dev.c
@@ -1104,6 +1104,8 @@ const struct file_operations def_blk_fops = {
 	.readv		= generic_file_readv,
 	.writev		= generic_file_write_nolock,
 	.sendfile	= generic_file_sendfile,
+	.splice_read	= generic_file_splice_read,
+	.splice_write	= generic_file_splice_write,
 };
 
 int ioctl_by_bdev(struct block_device *bdev, unsigned cmd, unsigned long arg)
diff --git a/fs/cifs/CHANGES b/fs/cifs/CHANGES
index 8a2de038882..1a27ecb46c9 100644
--- a/fs/cifs/CHANGES
+++ b/fs/cifs/CHANGES
@@ -1,7 +1,11 @@
 Version 1.42
 ------------
 Fix slow oplock break when mounted to different servers at the same time and
-the tids match and we try to find matching fid on wrong server.
+the tids match and we try to find matching fid on wrong server. Fix read
+looping when signing required by server (2.6.16 kernel only). Fix readdir
+vs. rename race which could cause each to hang. Return . and .. even
+if server does not.  Allow searches to skip first three entries and
+begin at any location. Fix oops in find_writeable_file.
 
 Version 1.41
 ------------
diff --git a/fs/cifs/README b/fs/cifs/README
index b2b4d080376..0355003f4f0 100644
--- a/fs/cifs/README
+++ b/fs/cifs/README
@@ -511,6 +511,14 @@ LinuxExtensionsEnabled	If set to one then the client will attempt to
 			support and want to map the uid and gid fields 
 			to values supplied at mount (rather than the 
 			actual values, then set this to zero. (default 1)
+Experimental            When set to 1 used to enable certain experimental
+			features (currently enables multipage writes
+			when signing is enabled, the multipage write
+			performance enhancement was disabled when
+			signing turned on in case buffer was modified
+			just before it was sent, also this flag will
+			be used to use the new experimental sessionsetup
+			code).
 
 These experimental features and tracing can be enabled by changing flags in 
 /proc/fs/cifs (after the cifs module has been installed or built into the 
diff --git a/fs/cifs/cifsfs.c b/fs/cifs/cifsfs.c
index d4b713e5aff..c262d8874ce 100644
--- a/fs/cifs/cifsfs.c
+++ b/fs/cifs/cifsfs.c
@@ -33,6 +33,7 @@
 #include <linux/vfs.h>
 #include <linux/mempool.h>
 #include <linux/delay.h>
+#include <linux/kthread.h>
 #include "cifsfs.h"
 #include "cifspdu.h"
 #define DECLARE_GLOBALS_HERE
@@ -75,9 +76,6 @@ unsigned int cifs_max_pending = CIFS_MAX_REQ;
 module_param(cifs_max_pending, int, 0);
 MODULE_PARM_DESC(cifs_max_pending,"Simultaneous requests to server. Default: 50 Range: 2 to 256");
 
-static DECLARE_COMPLETION(cifs_oplock_exited);
-static DECLARE_COMPLETION(cifs_dnotify_exited);
-
 extern mempool_t *cifs_sm_req_poolp;
 extern mempool_t *cifs_req_poolp;
 extern mempool_t *cifs_mid_poolp;
@@ -841,10 +839,6 @@ static int cifs_oplock_thread(void * dummyarg)
 	__u16  netfid;
 	int rc;
 
-	daemonize("cifsoplockd");
-	allow_signal(SIGTERM);
-
-	oplockThread = current;
 	do {
 		if (try_to_freeze()) 
 			continue;
@@ -900,9 +894,9 @@ static int cifs_oplock_thread(void * dummyarg)
 			set_current_state(TASK_INTERRUPTIBLE);
 			schedule_timeout(1);  /* yield in case q were corrupt */
 		}
-	} while(!signal_pending(current));
-	oplockThread = NULL;
-	complete_and_exit (&cifs_oplock_exited, 0);
+	} while (!kthread_should_stop());
+
+	return 0;
 }
 
 static int cifs_dnotify_thread(void * dummyarg)
@@ -910,10 +904,6 @@ static int cifs_dnotify_thread(void * dummyarg)
 	struct list_head *tmp;
 	struct cifsSesInfo *ses;
 
-	daemonize("cifsdnotifyd");
-	allow_signal(SIGTERM);
-
-	dnotifyThread = current;
 	do {
 		if(try_to_freeze())
 			continue;
@@ -931,8 +921,9 @@ static int cifs_dnotify_thread(void * dummyarg)
 				wake_up_all(&ses->server->response_q);
 		}
 		read_unlock(&GlobalSMBSeslock);
-	} while(!signal_pending(current));
-	complete_and_exit (&cifs_dnotify_exited, 0);
+	} while (!kthread_should_stop());
+
+	return 0;
 }
 
 static int __init
@@ -982,32 +973,48 @@ init_cifs(void)
 	}
 
 	rc = cifs_init_inodecache();
-	if (!rc) {
-		rc = cifs_init_mids();
-		if (!rc) {
-			rc = cifs_init_request_bufs();
-			if (!rc) {
-				rc = register_filesystem(&cifs_fs_type);
-				if (!rc) {                
-					rc = (int)kernel_thread(cifs_oplock_thread, NULL, 
-						CLONE_FS | CLONE_FILES | CLONE_VM);
-					if(rc > 0) {
-						rc = (int)kernel_thread(cifs_dnotify_thread, NULL,
-							CLONE_FS | CLONE_FILES | CLONE_VM);
-						if(rc > 0)
-							return 0;
-						else
-							cERROR(1,("error %d create dnotify thread", rc));
-					} else {
-						cERROR(1,("error %d create oplock thread",rc));
-					}
-				}
-				cifs_destroy_request_bufs();
-			}
-			cifs_destroy_mids();
-		}
-		cifs_destroy_inodecache();
+	if (rc)
+		goto out_clean_proc;
+
+	rc = cifs_init_mids();
+	if (rc)
+		goto out_destroy_inodecache;
+
+	rc = cifs_init_request_bufs();
+	if (rc)
+		goto out_destroy_mids;
+
+	rc = register_filesystem(&cifs_fs_type);
+	if (rc)
+		goto out_destroy_request_bufs;
+
+	oplockThread = kthread_run(cifs_oplock_thread, NULL, "cifsoplockd");
+	if (IS_ERR(oplockThread)) {
+		rc = PTR_ERR(oplockThread);
+		cERROR(1,("error %d create oplock thread", rc));
+		goto out_unregister_filesystem;
 	}
+
+	dnotifyThread = kthread_run(cifs_dnotify_thread, NULL, "cifsdnotifyd");
+	if (IS_ERR(dnotifyThread)) {
+		rc = PTR_ERR(dnotifyThread);
+		cERROR(1,("error %d create dnotify thread", rc));
+		goto out_stop_oplock_thread;
+	}
+
+	return 0;
+
+ out_stop_oplock_thread:
+	kthread_stop(oplockThread);
+ out_unregister_filesystem:
+	unregister_filesystem(&cifs_fs_type);
+ out_destroy_request_bufs:
+	cifs_destroy_request_bufs();
+ out_destroy_mids:
+	cifs_destroy_mids();
+ out_destroy_inodecache:
+	cifs_destroy_inodecache();
+ out_clean_proc:
 #ifdef CONFIG_PROC_FS
 	cifs_proc_clean();
 #endif
@@ -1025,14 +1032,8 @@ exit_cifs(void)
 	cifs_destroy_inodecache();
 	cifs_destroy_mids();
 	cifs_destroy_request_bufs();
-	if(oplockThread) {
-		send_sig(SIGTERM, oplockThread, 1);
-		wait_for_completion(&cifs_oplock_exited);
-	}
-	if(dnotifyThread) {
-		send_sig(SIGTERM, dnotifyThread, 1);
-		wait_for_completion(&cifs_dnotify_exited);
-	}
+	kthread_stop(oplockThread);
+	kthread_stop(dnotifyThread);
 }
 
 MODULE_AUTHOR("Steve French <sfrench@us.ibm.com>");
diff --git a/fs/cifs/cifssmb.c b/fs/cifs/cifssmb.c
index d705500aa28..fd36892eda5 100644
--- a/fs/cifs/cifssmb.c
+++ b/fs/cifs/cifssmb.c
@@ -3119,7 +3119,7 @@ findFirstRetry:
 				psrch_inf->endOfSearch = FALSE;
 
 			psrch_inf->entries_in_buffer  = le16_to_cpu(parms->SearchCount);
-			psrch_inf->index_of_last_entry = 
+			psrch_inf->index_of_last_entry = 2 /* skip . and .. */ +
 				psrch_inf->entries_in_buffer;
 			*pnetfid = parms->SearchHandle;
 		} else {
diff --git a/fs/cifs/connect.c b/fs/cifs/connect.c
index 0b86d5ca901..d2ec806a4f3 100644
--- a/fs/cifs/connect.c
+++ b/fs/cifs/connect.c
@@ -3447,6 +3447,12 @@ int cifs_setup_session(unsigned int xid, struct cifsSesInfo *pSesInfo,
 			pSesInfo->server->secMode,
 			pSesInfo->server->capabilities,
 			pSesInfo->server->timeZone));
+#ifdef CONFIG_CIFS_EXPERIMENTAL
+		if(experimEnabled > 1)
+			rc = CIFS_SessSetup(xid, pSesInfo, CIFS_NTLM /* type */,
+					    &ntlmv2_flag, nls_info);	
+		else
+#endif
 		if (extended_security
 				&& (pSesInfo->capabilities & CAP_EXTENDED_SECURITY)
 				&& (pSesInfo->server->secType == NTLMSSP)) {
diff --git a/fs/cifs/dir.c b/fs/cifs/dir.c
index 1d0ca3eaaca..82315edc77d 100644
--- a/fs/cifs/dir.c
+++ b/fs/cifs/dir.c
@@ -139,9 +139,7 @@ cifs_create(struct inode *inode, struct dentry *direntry, int mode,
 	cifs_sb = CIFS_SB(inode->i_sb);
 	pTcon = cifs_sb->tcon;
 
-	mutex_lock(&direntry->d_sb->s_vfs_rename_mutex);
 	full_path = build_path_from_dentry(direntry);
-	mutex_unlock(&direntry->d_sb->s_vfs_rename_mutex);
 	if(full_path == NULL) {
 		FreeXid(xid);
 		return -ENOMEM;
@@ -316,9 +314,7 @@ int cifs_mknod(struct inode *inode, struct dentry *direntry, int mode,
 	cifs_sb = CIFS_SB(inode->i_sb);
 	pTcon = cifs_sb->tcon;
 
-	mutex_lock(&direntry->d_sb->s_vfs_rename_mutex);
 	full_path = build_path_from_dentry(direntry);
-	mutex_unlock(&direntry->d_sb->s_vfs_rename_mutex);
 	if(full_path == NULL)
 		rc = -ENOMEM;
 	else if (pTcon->ses->capabilities & CAP_UNIX) {
@@ -440,6 +436,20 @@ cifs_lookup(struct inode *parent_dir_inode, struct dentry *direntry, struct name
 	cifs_sb = CIFS_SB(parent_dir_inode->i_sb);
 	pTcon = cifs_sb->tcon;
 
+	/*
+	 * Don't allow the separator character in a path component.
+	 * The VFS will not allow "/", but "\" is allowed by posix.
+	 */
+	if (!(cifs_sb->mnt_cifs_flags & CIFS_MOUNT_POSIX_PATHS)) {
+		int i;
+		for (i = 0; i < direntry->d_name.len; i++)
+			if (direntry->d_name.name[i] == '\\') {
+				cFYI(1, ("Invalid file name"));
+				FreeXid(xid);
+				return ERR_PTR(-EINVAL);
+			}
+	}
+
 	/* can not grab the rename sem here since it would
 	deadlock in the cases (beginning of sys_rename itself)
 	in which we already have the sb rename sem */
diff --git a/fs/cifs/fcntl.c b/fs/cifs/fcntl.c
index ec4dfe9bf5e..633a9381132 100644
--- a/fs/cifs/fcntl.c
+++ b/fs/cifs/fcntl.c
@@ -86,9 +86,7 @@ int cifs_dir_notify(struct file * file, unsigned long arg)
 	cifs_sb = CIFS_SB(file->f_dentry->d_sb);
 	pTcon = cifs_sb->tcon;
 
-	mutex_lock(&file->f_dentry->d_sb->s_vfs_rename_mutex);
 	full_path = build_path_from_dentry(file->f_dentry);
-	mutex_unlock(&file->f_dentry->d_sb->s_vfs_rename_mutex);
 
 	if(full_path == NULL) {
 		rc = -ENOMEM;
diff --git a/fs/cifs/file.c b/fs/cifs/file.c
index 5c497c52977..e152bf6afa6 100644
--- a/fs/cifs/file.c
+++ b/fs/cifs/file.c
@@ -203,9 +203,7 @@ int cifs_open(struct inode *inode, struct file *file)
 		}
 	}
 
-	mutex_lock(&inode->i_sb->s_vfs_rename_mutex);
 	full_path = build_path_from_dentry(file->f_dentry);
-	mutex_unlock(&inode->i_sb->s_vfs_rename_mutex);
 	if (full_path == NULL) {
 		FreeXid(xid);
 		return -ENOMEM;
@@ -906,8 +904,7 @@ static ssize_t cifs_write(struct file *file, const char *write_data,
 				if (rc != 0)
 					break;
 			}
-			/* BB FIXME We can not sign across two buffers yet */
-			if((pTcon->ses->server->secMode & 
+			if(experimEnabled || (pTcon->ses->server->secMode & 
 			 (SECMODE_SIGN_REQUIRED | SECMODE_SIGN_ENABLED)) == 0) {
 				struct kvec iov[2];
 				unsigned int len;
@@ -923,13 +920,13 @@ static ssize_t cifs_write(struct file *file, const char *write_data,
 						*poffset, &bytes_written,
 						iov, 1, long_op);
 			} else
-			/* BB FIXME fixup indentation of line below */
-			rc = CIFSSMBWrite(xid, pTcon,
-				 open_file->netfid,
-				 min_t(const int, cifs_sb->wsize, 
-				       write_size - total_written),
-				 *poffset, &bytes_written,
-				 write_data + total_written, NULL, long_op);
+				rc = CIFSSMBWrite(xid, pTcon,
+					 open_file->netfid,
+					 min_t(const int, cifs_sb->wsize,
+					       write_size - total_written),
+					 *poffset, &bytes_written,
+					 write_data + total_written,
+					 NULL, long_op);
 		}
 		if (rc || (bytes_written == 0)) {
 			if (total_written)
@@ -968,6 +965,16 @@ struct cifsFileInfo *find_writable_file(struct cifsInodeInfo *cifs_inode)
 	struct cifsFileInfo *open_file;
 	int rc;
 
+	/* Having a null inode here (because mapping->host was set to zero by
+	the VFS or MM) should not happen but we had reports of on oops (due to
+	it being zero) during stress testcases so we need to check for it */
+
+	if(cifs_inode == NULL) {
+		cERROR(1,("Null inode passed to cifs_writeable_file"));
+		dump_stack();
+		return NULL;
+	}
+
 	read_lock(&GlobalSMBSeslock);
 	list_for_each_entry(open_file, &cifs_inode->openFileList, flist) {
 		if (open_file->closePend)
@@ -1093,12 +1100,11 @@ static int cifs_writepages(struct address_space *mapping,
 	if (cifs_sb->wsize < PAGE_CACHE_SIZE)
 		return generic_writepages(mapping, wbc);
 
-	/* BB FIXME we do not have code to sign across multiple buffers yet,
-	   so go to older writepage style write which we can sign if needed */
 	if((cifs_sb->tcon->ses) && (cifs_sb->tcon->ses->server))
 		if(cifs_sb->tcon->ses->server->secMode &
                           (SECMODE_SIGN_REQUIRED | SECMODE_SIGN_ENABLED))
-			return generic_writepages(mapping, wbc);
+			if(!experimEnabled)
+				return generic_writepages(mapping, wbc);
 
 	/*
 	 * BB: Is this meaningful for a non-block-device file system?
diff --git a/fs/cifs/inode.c b/fs/cifs/inode.c
index 957ddd1571c..4093764ef46 100644
--- a/fs/cifs/inode.c
+++ b/fs/cifs/inode.c
@@ -722,9 +722,7 @@ int cifs_mkdir(struct inode *inode, struct dentry *direntry, int mode)
 	cifs_sb = CIFS_SB(inode->i_sb);
 	pTcon = cifs_sb->tcon;
 
-	mutex_lock(&inode->i_sb->s_vfs_rename_mutex);
 	full_path = build_path_from_dentry(direntry);
-	mutex_unlock(&inode->i_sb->s_vfs_rename_mutex);
 	if (full_path == NULL) {
 		FreeXid(xid);
 		return -ENOMEM;
@@ -807,9 +805,7 @@ int cifs_rmdir(struct inode *inode, struct dentry *direntry)
 	cifs_sb = CIFS_SB(inode->i_sb);
 	pTcon = cifs_sb->tcon;
 
-	mutex_lock(&inode->i_sb->s_vfs_rename_mutex);
 	full_path = build_path_from_dentry(direntry);
-	mutex_unlock(&inode->i_sb->s_vfs_rename_mutex);
 	if (full_path == NULL) {
 		FreeXid(xid);
 		return -ENOMEM;
@@ -1141,9 +1137,7 @@ int cifs_setattr(struct dentry *direntry, struct iattr *attrs)
 			rc = 0;
 	}
 		
-	mutex_lock(&direntry->d_sb->s_vfs_rename_mutex);
 	full_path = build_path_from_dentry(direntry);
-	mutex_unlock(&direntry->d_sb->s_vfs_rename_mutex);
 	if (full_path == NULL) {
 		FreeXid(xid);
 		return -ENOMEM;
diff --git a/fs/cifs/link.c b/fs/cifs/link.c
index 9562f5bba65..2ec99f83314 100644
--- a/fs/cifs/link.c
+++ b/fs/cifs/link.c
@@ -48,10 +48,8 @@ cifs_hardlink(struct dentry *old_file, struct inode *inode,
 /* No need to check for cross device links since server will do that
    BB note DFS case in future though (when we may have to check) */
 
-	mutex_lock(&inode->i_sb->s_vfs_rename_mutex);
 	fromName = build_path_from_dentry(old_file);
 	toName = build_path_from_dentry(direntry);
-	mutex_unlock(&inode->i_sb->s_vfs_rename_mutex);
 	if((fromName == NULL) || (toName == NULL)) {
 		rc = -ENOMEM;
 		goto cifs_hl_exit;
@@ -103,9 +101,7 @@ cifs_follow_link(struct dentry *direntry, struct nameidata *nd)
 
 	xid = GetXid();
 
-	mutex_lock(&direntry->d_sb->s_vfs_rename_mutex);
 	full_path = build_path_from_dentry(direntry);
-	mutex_unlock(&direntry->d_sb->s_vfs_rename_mutex);
 
 	if (!full_path)
 		goto out_no_free;
@@ -164,9 +160,7 @@ cifs_symlink(struct inode *inode, struct dentry *direntry, const char *symname)
 	cifs_sb = CIFS_SB(inode->i_sb);
 	pTcon = cifs_sb->tcon;
 
-	mutex_lock(&inode->i_sb->s_vfs_rename_mutex);
 	full_path = build_path_from_dentry(direntry);
-	mutex_unlock(&inode->i_sb->s_vfs_rename_mutex);
 
 	if(full_path == NULL) {
 		FreeXid(xid);
diff --git a/fs/cifs/ntlmssp.c b/fs/cifs/ntlmssp.c
index 78866f92574..115359cc7a3 100644
--- a/fs/cifs/ntlmssp.c
+++ b/fs/cifs/ntlmssp.c
@@ -121,6 +121,20 @@ CIFS_SessSetup(unsigned int xid, struct cifsSesInfo *ses, const int type,
 	}
 
 
+	/* copy session key */
+
+	/* if Unicode, align strings to two byte boundary */
+
+	/* copy user name */ /* BB Do we need to special case null user name? */
+
+	/* copy domain name */
+
+	/* copy Linux version */
+
+	/* copy network operating system name */
+
+	/* update bcc and smb buffer length */
+
 /*	rc = SendReceive2(xid, ses, iov, num_iovecs, &resp_buf_type, 0); */
 	/* SMB request buf freed in SendReceive2 */
 
diff --git a/fs/cifs/readdir.c b/fs/cifs/readdir.c
index 2f6e2825571..b689c503512 100644
--- a/fs/cifs/readdir.c
+++ b/fs/cifs/readdir.c
@@ -404,9 +404,7 @@ static int initiate_cifs_search(const int xid, struct file *file)
 	if(pTcon == NULL)
 		return -EINVAL;
 
-	mutex_lock(&file->f_dentry->d_sb->s_vfs_rename_mutex);
 	full_path = build_path_from_dentry(file->f_dentry);
-	mutex_unlock(&file->f_dentry->d_sb->s_vfs_rename_mutex);
 
 	if(full_path == NULL) {
 		return -ENOMEM;
@@ -592,6 +590,13 @@ static int find_cifs_entry(const int xid, struct cifsTconInfo *pTcon,
 	first_entry_in_buffer = 
 		cifsFile->srch_inf.index_of_last_entry - 
 			cifsFile->srch_inf.entries_in_buffer;
+
+	/* if first entry in buf is zero then is first buffer
+	in search response data which means it is likely . and ..
+	will be in this buffer, although some servers do not return
+	. and .. for the root of a drive and for those we need
+	to start two entries earlier */
+
 /*	dump_cifs_file_struct(file, "In fce ");*/
 	if(((index_to_find < cifsFile->srch_inf.index_of_last_entry) && 
 	     is_dir_changed(file)) || 
@@ -634,23 +639,14 @@ static int find_cifs_entry(const int xid, struct cifsTconInfo *pTcon,
 		char * end_of_smb = cifsFile->srch_inf.ntwrk_buf_start + 
 			smbCalcSize((struct smb_hdr *)
 				cifsFile->srch_inf.ntwrk_buf_start);
+
+		current_entry = cifsFile->srch_inf.srch_entries_start;
 		first_entry_in_buffer = cifsFile->srch_inf.index_of_last_entry
 					- cifsFile->srch_inf.entries_in_buffer;
 		pos_in_buf = index_to_find - first_entry_in_buffer;
 		cFYI(1,("found entry - pos_in_buf %d",pos_in_buf)); 
-		current_entry = cifsFile->srch_inf.srch_entries_start;
 		for(i=0;(i<(pos_in_buf)) && (current_entry != NULL);i++) {
 			/* go entry by entry figuring out which is first */
-			/* if( . or ..)
-				skip */
-			rc = cifs_entry_is_dot(current_entry,cifsFile);
-			if(rc == 1) /* is . or .. so skip */ {
-				cFYI(1,("Entry is .")); /* BB removeme BB */
-				/* continue; */
-			} else if (rc == 2 ) {
-				cFYI(1,("Entry is ..")); /* BB removeme BB */
-				/* continue; */
-			}
 			current_entry = nxt_dir_entry(current_entry,end_of_smb);
 		}
 		if((current_entry == NULL) && (i < pos_in_buf)) {
@@ -770,6 +766,11 @@ static int cifs_filldir(char *pfindEntry, struct file *file,
 	if(file->f_dentry == NULL)
 		return -ENOENT;
 
+	rc = cifs_entry_is_dot(pfindEntry,pCifsF);
+	/* skip . and .. since we added them first */
+	if(rc != 0) 
+		return 0;
+
 	cifs_sb = CIFS_SB(file->f_dentry->d_sb);
 
 	qstring.name = scratch_buf;
@@ -898,22 +899,22 @@ int cifs_readdir(struct file *file, void *direntry, filldir_t filldir)
 
 	switch ((int) file->f_pos) {
 	case 0:
-		/*if (filldir(direntry, ".", 1, file->f_pos,
+		if (filldir(direntry, ".", 1, file->f_pos,
 		     file->f_dentry->d_inode->i_ino, DT_DIR) < 0) {
-			cERROR(1, ("Filldir for current dir failed "));
+			cERROR(1, ("Filldir for current dir failed"));
 			rc = -ENOMEM;
 			break;
 		}
-		file->f_pos++; */
+		file->f_pos++;
 	case 1:
-		/* if (filldir(direntry, "..", 2, file->f_pos,
+		if (filldir(direntry, "..", 2, file->f_pos,
 		     file->f_dentry->d_parent->d_inode->i_ino, DT_DIR) < 0) {
 			cERROR(1, ("Filldir for parent dir failed "));
 			rc = -ENOMEM;
 			break;
 		}
-		file->f_pos++; */
-	case 2:
+		file->f_pos++;
+	default:
 		/* 1) If search is active, 
 			is in current search buffer? 
 			if it before then restart search
@@ -927,7 +928,6 @@ int cifs_readdir(struct file *file, void *direntry, filldir_t filldir)
 				return rc;
 			}
 		}
-	default:
 		if(file->private_data == NULL) {
 			rc = -EINVAL;
 			FreeXid(xid);
@@ -947,8 +947,6 @@ int cifs_readdir(struct file *file, void *direntry, filldir_t filldir)
 		kfree(cifsFile->search_resume_name);
 		cifsFile->search_resume_name = NULL; */
 
-		/* BB account for . and .. in f_pos as special case */
-
 		rc = find_cifs_entry(xid,pTcon, file,
 				&current_entry,&num_to_fill);
 		if(rc) {
@@ -977,7 +975,8 @@ int cifs_readdir(struct file *file, void *direntry, filldir_t filldir)
 					  num_to_fill, i));
 				break;
 			}
-
+			/* if buggy server returns . and .. late do
+			we want to check for that here? */
 			rc = cifs_filldir(current_entry, file, 
 					filldir, direntry,tmp_buf);
 			file->f_pos++;
diff --git a/fs/cifs/xattr.c b/fs/cifs/xattr.c
index 3938444d87b..7754d641775 100644
--- a/fs/cifs/xattr.c
+++ b/fs/cifs/xattr.c
@@ -62,9 +62,7 @@ int cifs_removexattr(struct dentry * direntry, const char * ea_name)
 	cifs_sb = CIFS_SB(sb);
 	pTcon = cifs_sb->tcon;
                                                                                      
-	mutex_lock(&sb->s_vfs_rename_mutex);
 	full_path = build_path_from_dentry(direntry);
-	mutex_unlock(&sb->s_vfs_rename_mutex);
 	if(full_path == NULL) {
 		FreeXid(xid);
 		return -ENOMEM;
@@ -116,9 +114,7 @@ int cifs_setxattr(struct dentry * direntry, const char * ea_name,
 	cifs_sb = CIFS_SB(sb);
 	pTcon = cifs_sb->tcon;
 
-	mutex_lock(&sb->s_vfs_rename_mutex);
 	full_path = build_path_from_dentry(direntry);
-	mutex_unlock(&sb->s_vfs_rename_mutex);
 	if(full_path == NULL) {
 		FreeXid(xid);
 		return -ENOMEM;
@@ -223,9 +219,7 @@ ssize_t cifs_getxattr(struct dentry * direntry, const char * ea_name,
 	cifs_sb = CIFS_SB(sb);
 	pTcon = cifs_sb->tcon;
 
-	mutex_lock(&sb->s_vfs_rename_mutex);
 	full_path = build_path_from_dentry(direntry);
-	mutex_unlock(&sb->s_vfs_rename_mutex);
 	if(full_path == NULL) {
 		FreeXid(xid);
 		return -ENOMEM;
@@ -341,9 +335,7 @@ ssize_t cifs_listxattr(struct dentry * direntry, char * data, size_t buf_size)
 	cifs_sb = CIFS_SB(sb);
 	pTcon = cifs_sb->tcon;
 
-	mutex_lock(&sb->s_vfs_rename_mutex);
 	full_path = build_path_from_dentry(direntry);
-	mutex_unlock(&sb->s_vfs_rename_mutex);
 	if(full_path == NULL) {
 		FreeXid(xid);
 		return -ENOMEM;
diff --git a/fs/compat.c b/fs/compat.c
index 7f8e26ea427..3f3e8f4d43d 100644
--- a/fs/compat.c
+++ b/fs/compat.c
@@ -1217,6 +1217,10 @@ static ssize_t compat_do_readv_writev(int type, struct file *file,
 	if (ret < 0)
 		goto out;
 
+	ret = security_file_permission(file, type == READ ? MAY_READ:MAY_WRITE);
+	if (ret)
+		goto out;
+
 	fnv = NULL;
 	if (type == READ) {
 		fn = file->f_op->read;
@@ -1313,6 +1317,26 @@ out:
 	return ret;
 }
 
+asmlinkage long
+compat_sys_vmsplice(int fd, const struct compat_iovec __user *iov32,
+		    unsigned int nr_segs, unsigned int flags)
+{
+	unsigned i;
+	struct iovec *iov;
+	if (nr_segs >= UIO_MAXIOV)
+		return -EINVAL;
+	iov = compat_alloc_user_space(nr_segs * sizeof(struct iovec));
+	for (i = 0; i < nr_segs; i++) {
+		struct compat_iovec v;
+		if (get_user(v.iov_base, &iov32[i].iov_base) ||
+		    get_user(v.iov_len, &iov32[i].iov_len) ||
+		    put_user(compat_ptr(v.iov_base), &iov[i].iov_base) ||
+		    put_user(v.iov_len, &iov[i].iov_len))
+			return -EFAULT;
+	}
+	return sys_vmsplice(fd, iov, nr_segs, flags);
+}
+
 /*
  * Exactly like fs/open.c:sys_open(), except that it doesn't set the
  * O_LARGEFILE flag.
diff --git a/fs/exec.c b/fs/exec.c
index 4121bb55973..3a79d97ac23 100644
--- a/fs/exec.c
+++ b/fs/exec.c
@@ -712,7 +712,7 @@ static int de_thread(struct task_struct *tsk)
 		attach_pid(current, PIDTYPE_PID,  current->pid);
 		attach_pid(current, PIDTYPE_PGID, current->signal->pgrp);
 		attach_pid(current, PIDTYPE_SID,  current->signal->session);
-		list_add_tail(&current->tasks, &init_task.tasks);
+		list_add_tail_rcu(&current->tasks, &init_task.tasks);
 
 		current->group_leader = current;
 		leader->group_leader = current;
diff --git a/fs/ext3/ioctl.c b/fs/ext3/ioctl.c
index aaf1da17b6d..8c22aa9a7fb 100644
--- a/fs/ext3/ioctl.c
+++ b/fs/ext3/ioctl.c
@@ -48,6 +48,7 @@ int ext3_ioctl (struct inode * inode, struct file * filp, unsigned int cmd,
 		if (!S_ISDIR(inode->i_mode))
 			flags &= ~EXT3_DIRSYNC_FL;
 
+		mutex_lock(&inode->i_mutex);
 		oldflags = ei->i_flags;
 
 		/* The JOURNAL_DATA flag is modifiable only by root */
@@ -60,8 +61,10 @@ int ext3_ioctl (struct inode * inode, struct file * filp, unsigned int cmd,
 		 * This test looks nicer. Thanks to Pauline Middelink
 		 */
 		if ((flags ^ oldflags) & (EXT3_APPEND_FL | EXT3_IMMUTABLE_FL)) {
-			if (!capable(CAP_LINUX_IMMUTABLE))
+			if (!capable(CAP_LINUX_IMMUTABLE)) {
+				mutex_unlock(&inode->i_mutex);
 				return -EPERM;
+			}
 		}
 
 		/*
@@ -69,14 +72,18 @@ int ext3_ioctl (struct inode * inode, struct file * filp, unsigned int cmd,
 		 * the relevant capability.
 		 */
 		if ((jflag ^ oldflags) & (EXT3_JOURNAL_DATA_FL)) {
-			if (!capable(CAP_SYS_RESOURCE))
+			if (!capable(CAP_SYS_RESOURCE)) {
+				mutex_unlock(&inode->i_mutex);
 				return -EPERM;
+			}
 		}
 
 
 		handle = ext3_journal_start(inode, 1);
-		if (IS_ERR(handle))
+		if (IS_ERR(handle)) {
+			mutex_unlock(&inode->i_mutex);
 			return PTR_ERR(handle);
+		}
 		if (IS_SYNC(inode))
 			handle->h_sync = 1;
 		err = ext3_reserve_inode_write(handle, inode, &iloc);
@@ -93,11 +100,14 @@ int ext3_ioctl (struct inode * inode, struct file * filp, unsigned int cmd,
 		err = ext3_mark_iloc_dirty(handle, inode, &iloc);
 flags_err:
 		ext3_journal_stop(handle);
-		if (err)
+		if (err) {
+			mutex_unlock(&inode->i_mutex);
 			return err;
+		}
 
 		if ((jflag ^ oldflags) & (EXT3_JOURNAL_DATA_FL))
 			err = ext3_change_inode_journal_flag(inode, jflag);
+		mutex_unlock(&inode->i_mutex);
 		return err;
 	}
 	case EXT3_IOC_GETVERSION:
diff --git a/fs/ext3/resize.c b/fs/ext3/resize.c
index 14f5f6ea3e7..8aac5334680 100644
--- a/fs/ext3/resize.c
+++ b/fs/ext3/resize.c
@@ -213,7 +213,7 @@ static int setup_new_group_blocks(struct super_block *sb,
 			goto exit_bh;
 		}
 		lock_buffer(bh);
-		memcpy(gdb->b_data, sbi->s_group_desc[i], bh->b_size);
+		memcpy(gdb->b_data, sbi->s_group_desc[i]->b_data, bh->b_size);
 		set_buffer_uptodate(gdb);
 		unlock_buffer(bh);
 		ext3_journal_dirty_metadata(handle, gdb);
@@ -767,6 +767,7 @@ int ext3_group_add(struct super_block *sb, struct ext3_new_group_data *input)
 	if (input->group != sbi->s_groups_count) {
 		ext3_warning(sb, __FUNCTION__,
 			     "multiple resizers run on filesystem!");
+		unlock_super(sb);
 		err = -EBUSY;
 		goto exit_journal;
 	}
diff --git a/fs/fuse/dev.c b/fs/fuse/dev.c
index cc750c68fe7..104a62dadb9 100644
--- a/fs/fuse/dev.c
+++ b/fs/fuse/dev.c
@@ -128,14 +128,24 @@ void fuse_put_request(struct fuse_conn *fc, struct fuse_req *req)
 	}
 }
 
-void fuse_remove_background(struct fuse_conn *fc, struct fuse_req *req)
+/*
+ * Called with sbput_sem held for read (request_end) or write
+ * (fuse_put_super).  By the time fuse_put_super() is finished, all
+ * inodes belonging to background requests must be released, so the
+ * iputs have to be done within the locked region.
+ */
+void fuse_release_background(struct fuse_conn *fc, struct fuse_req *req)
 {
-	list_del_init(&req->bg_entry);
+	iput(req->inode);
+	iput(req->inode2);
+	spin_lock(&fc->lock);
+	list_del(&req->bg_entry);
 	if (fc->num_background == FUSE_MAX_BACKGROUND) {
 		fc->blocked = 0;
 		wake_up_all(&fc->blocked_waitq);
 	}
 	fc->num_background--;
+	spin_unlock(&fc->lock);
 }
 
 /*
@@ -165,27 +175,22 @@ static void request_end(struct fuse_conn *fc, struct fuse_req *req)
 		wake_up(&req->waitq);
 		fuse_put_request(fc, req);
 	} else {
-		struct inode *inode = req->inode;
-		struct inode *inode2 = req->inode2;
-		struct file *file = req->file;
 		void (*end) (struct fuse_conn *, struct fuse_req *) = req->end;
 		req->end = NULL;
-		req->inode = NULL;
-		req->inode2 = NULL;
-		req->file = NULL;
-		if (!list_empty(&req->bg_entry))
-			fuse_remove_background(fc, req);
 		spin_unlock(&fc->lock);
+		down_read(&fc->sbput_sem);
+		if (fc->mounted)
+			fuse_release_background(fc, req);
+		up_read(&fc->sbput_sem);
+
+		/* fput must go outside sbput_sem, otherwise it can deadlock */
+		if (req->file)
+			fput(req->file);
 
 		if (end)
 			end(fc, req);
 		else
 			fuse_put_request(fc, req);
-
-		if (file)
-			fput(file);
-		iput(inode);
-		iput(inode2);
 	}
 }
 
diff --git a/fs/fuse/fuse_i.h b/fs/fuse/fuse_i.h
index 59661c481d9..0474202cb5d 100644
--- a/fs/fuse/fuse_i.h
+++ b/fs/fuse/fuse_i.h
@@ -258,9 +258,15 @@ struct fuse_conn {
 	/** waitq for blocked connection */
 	wait_queue_head_t blocked_waitq;
 
+	/** RW semaphore for exclusion with fuse_put_super() */
+	struct rw_semaphore sbput_sem;
+
 	/** The next unique request id */
 	u64 reqctr;
 
+	/** Mount is active */
+	unsigned mounted;
+
 	/** Connection established, cleared on umount, connection
 	    abort and device release */
 	unsigned connected;
@@ -471,11 +477,11 @@ void request_send_noreply(struct fuse_conn *fc, struct fuse_req *req);
 void request_send_background(struct fuse_conn *fc, struct fuse_req *req);
 
 /**
- * Remove request from the the background list
+ * Release inodes and file associated with background request
  */
-void fuse_remove_background(struct fuse_conn *fc, struct fuse_req *req);
+void fuse_release_background(struct fuse_conn *fc, struct fuse_req *req);
 
-/** Abort all requests */
+/* Abort all requests */
 void fuse_abort_conn(struct fuse_conn *fc);
 
 /**
diff --git a/fs/fuse/inode.c b/fs/fuse/inode.c
index 43a6fc0db8a..7627022446b 100644
--- a/fs/fuse/inode.c
+++ b/fs/fuse/inode.c
@@ -204,26 +204,17 @@ static void fuse_put_super(struct super_block *sb)
 {
 	struct fuse_conn *fc = get_fuse_conn_super(sb);
 
+	down_write(&fc->sbput_sem);
+	while (!list_empty(&fc->background))
+		fuse_release_background(fc,
+					list_entry(fc->background.next,
+						   struct fuse_req, bg_entry));
+
 	spin_lock(&fc->lock);
+	fc->mounted = 0;
 	fc->connected = 0;
-	while (!list_empty(&fc->background)) {
-		struct fuse_req *req = list_entry(fc->background.next,
-						  struct fuse_req, bg_entry);
-		struct inode *inode = req->inode;
-		struct inode *inode2 = req->inode2;
-
-		/* File would hold a reference to vfsmount */
-		BUG_ON(req->file);
-		req->inode = NULL;
-		req->inode2 = NULL;
-		fuse_remove_background(fc, req);
-
-		spin_unlock(&fc->lock);
-		iput(inode);
-		iput(inode2);
-		spin_lock(&fc->lock);
-	}
 	spin_unlock(&fc->lock);
+	up_write(&fc->sbput_sem);
 	/* Flush all readers on this fs */
 	kill_fasync(&fc->fasync, SIGIO, POLL_IN);
 	wake_up_all(&fc->waitq);
@@ -395,6 +386,7 @@ static struct fuse_conn *new_conn(void)
 		INIT_LIST_HEAD(&fc->processing);
 		INIT_LIST_HEAD(&fc->io);
 		INIT_LIST_HEAD(&fc->background);
+		init_rwsem(&fc->sbput_sem);
 		kobj_set_kset_s(fc, connections_subsys);
 		kobject_init(&fc->kobj);
 		atomic_set(&fc->num_waiting, 0);
@@ -508,11 +500,6 @@ static int fuse_fill_super(struct super_block *sb, void *data, int silent)
 	if (file->f_op != &fuse_dev_operations)
 		return -EINVAL;
 
-	/* Setting file->private_data can't race with other mount()
-	   instances, since BKL is held for ->get_sb() */
-	if (file->private_data)
-		return -EINVAL;
-
 	fc = new_conn();
 	if (!fc)
 		return -ENOMEM;
@@ -548,7 +535,14 @@ static int fuse_fill_super(struct super_block *sb, void *data, int silent)
 	if (err)
 		goto err_free_req;
 
+	/* Setting file->private_data can't race with other mount()
+	   instances, since BKL is held for ->get_sb() */
+	err = -EINVAL;
+	if (file->private_data)
+		goto err_kobject_del;
+
 	sb->s_root = root_dentry;
+	fc->mounted = 1;
 	fc->connected = 1;
 	kobject_get(&fc->kobj);
 	file->private_data = fc;
@@ -563,6 +557,8 @@ static int fuse_fill_super(struct super_block *sb, void *data, int silent)
 
 	return 0;
 
+ err_kobject_del:
+	kobject_del(&fc->kobj);
  err_free_req:
 	fuse_request_free(init_req);
  err_put_root:
diff --git a/fs/lockd/svclock.c b/fs/lockd/svclock.c
index d2b66bad7d5..3ef739120df 100644
--- a/fs/lockd/svclock.c
+++ b/fs/lockd/svclock.c
@@ -650,7 +650,7 @@ static void nlmsvc_grant_callback(struct rpc_task *task, void *data)
 	svc_wake_up(block->b_daemon);
 }
 
-void nlmsvc_grant_release(void *data)
+static void nlmsvc_grant_release(void *data)
 {
 	struct nlm_rqst		*call = data;
 
diff --git a/fs/locks.c b/fs/locks.c
index dda83d6cd48..efad798824d 100644
--- a/fs/locks.c
+++ b/fs/locks.c
@@ -2230,7 +2230,12 @@ void steal_locks(fl_owner_t from)
 
 	lock_kernel();
 	j = 0;
-	rcu_read_lock();
+
+	/*
+	 * We are not taking a ref to the file structures, so
+	 * we need to acquire ->file_lock.
+	 */
+	spin_lock(&files->file_lock);
 	fdt = files_fdtable(files);
 	for (;;) {
 		unsigned long set;
@@ -2248,7 +2253,7 @@ void steal_locks(fl_owner_t from)
 			set >>= 1;
 		}
 	}
-	rcu_read_unlock();
+	spin_unlock(&files->file_lock);
 	unlock_kernel();
 }
 EXPORT_SYMBOL(steal_locks);
diff --git a/fs/nfs/dir.c b/fs/nfs/dir.c
index a23f3489416..cae74dd4c7f 100644
--- a/fs/nfs/dir.c
+++ b/fs/nfs/dir.c
@@ -128,15 +128,14 @@ struct inode_operations nfs4_dir_inode_operations = {
 static int
 nfs_opendir(struct inode *inode, struct file *filp)
 {
-	int res = 0;
+	int res;
 
 	dfprintk(VFS, "NFS: opendir(%s/%ld)\n",
 			inode->i_sb->s_id, inode->i_ino);
 
 	lock_kernel();
 	/* Call generic open code in order to cache credentials */
-	if (!res)
-		res = nfs_open(inode, filp);
+	res = nfs_open(inode, filp);
 	unlock_kernel();
 	return res;
 }
diff --git a/fs/nfs/direct.c b/fs/nfs/direct.c
index 0f583cb16dd..3c72b0c0728 100644
--- a/fs/nfs/direct.c
+++ b/fs/nfs/direct.c
@@ -112,10 +112,9 @@ static void nfs_direct_write_complete(struct nfs_direct_req *dreq, struct inode
  */
 ssize_t nfs_direct_IO(int rw, struct kiocb *iocb, const struct iovec *iov, loff_t pos, unsigned long nr_segs)
 {
-	struct dentry *dentry = iocb->ki_filp->f_dentry;
-
 	dprintk("NFS: nfs_direct_IO (%s) off/no(%Ld/%lu) EINVAL\n",
-			dentry->d_name.name, (long long) pos, nr_segs);
+			iocb->ki_filp->f_dentry->d_name.name,
+			(long long) pos, nr_segs);
 
 	return -EINVAL;
 }
@@ -468,7 +467,6 @@ static const struct rpc_call_ops nfs_commit_direct_ops = {
 static void nfs_direct_commit_schedule(struct nfs_direct_req *dreq)
 {
 	struct nfs_write_data *data = dreq->commit_data;
-	struct rpc_task *task = &data->task;
 
 	data->inode = dreq->inode;
 	data->cred = dreq->ctx->cred;
@@ -489,7 +487,7 @@ static void nfs_direct_commit_schedule(struct nfs_direct_req *dreq)
 	/* Note: task.tk_ops->rpc_release will free dreq->commit_data */
 	dreq->commit_data = NULL;
 
-	dprintk("NFS: %5u initiated commit call\n", task->tk_pid);
+	dprintk("NFS: %5u initiated commit call\n", data->task.tk_pid);
 
 	lock_kernel();
 	rpc_execute(&data->task);
diff --git a/fs/nfs/file.c b/fs/nfs/file.c
index f1df2c8d925..fade02c15e6 100644
--- a/fs/nfs/file.c
+++ b/fs/nfs/file.c
@@ -534,10 +534,9 @@ static int nfs_lock(struct file *filp, int cmd, struct file_lock *fl)
  */
 static int nfs_flock(struct file *filp, int cmd, struct file_lock *fl)
 {
-	struct inode * inode = filp->f_mapping->host;
-
 	dprintk("NFS: nfs_flock(f=%s/%ld, t=%x, fl=%x)\n",
-			inode->i_sb->s_id, inode->i_ino,
+			filp->f_dentry->d_inode->i_sb->s_id,
+			filp->f_dentry->d_inode->i_ino,
 			fl->fl_type, fl->fl_flags);
 
 	/*
diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c
index 2f7656b911b..d0b991a9232 100644
--- a/fs/nfs/inode.c
+++ b/fs/nfs/inode.c
@@ -700,12 +700,9 @@ static int nfs_show_stats(struct seq_file *m, struct vfsmount *mnt)
 	/*
 	 * Display superblock I/O counters
 	 */
-	for (cpu = 0; cpu < NR_CPUS; cpu++) {
+	for_each_possible_cpu(cpu) {
 		struct nfs_iostats *stats;
 
-		if (!cpu_possible(cpu))
-			continue;
-
 		preempt_disable();
 		stats = per_cpu_ptr(nfss->io_stats, cpu);
 
diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index 47ece1dd3c6..d86c0db7b1e 100644
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -1218,7 +1218,7 @@ out:
 	return status;
 }
 
-static void nfs4_intent_set_file(struct nameidata *nd, struct dentry *dentry, struct nfs4_state *state)
+static int nfs4_intent_set_file(struct nameidata *nd, struct dentry *dentry, struct nfs4_state *state)
 {
 	struct file *filp;
 
@@ -1227,8 +1227,10 @@ static void nfs4_intent_set_file(struct nameidata *nd, struct dentry *dentry, st
 		struct nfs_open_context *ctx;
 		ctx = (struct nfs_open_context *)filp->private_data;
 		ctx->state = state;
-	} else
-		nfs4_close_state(state, nd->intent.open.flags);
+		return 0;
+	}
+	nfs4_close_state(state, nd->intent.open.flags);
+	return PTR_ERR(filp);
 }
 
 struct dentry *
@@ -1835,7 +1837,7 @@ nfs4_proc_create(struct inode *dir, struct dentry *dentry, struct iattr *sattr,
 			nfs_setattr_update_inode(state->inode, sattr);
 	}
 	if (status == 0 && nd != NULL && (nd->flags & LOOKUP_OPEN))
-		nfs4_intent_set_file(nd, dentry, state);
+		status = nfs4_intent_set_file(nd, dentry, state);
 	else
 		nfs4_close_state(state, flags);
 out:
diff --git a/fs/open.c b/fs/open.c
index c32c89d6d8d..53ec28c3677 100644
--- a/fs/open.c
+++ b/fs/open.c
@@ -331,7 +331,10 @@ out:
 
 asmlinkage long sys_ftruncate(unsigned int fd, unsigned long length)
 {
-	return do_sys_ftruncate(fd, length, 1);
+	long ret = do_sys_ftruncate(fd, length, 1);
+	/* avoid REGPARM breakage on x86: */
+	prevent_tail_call(ret);
+	return ret;
 }
 
 /* LFS versions of truncate are only needed on 32 bit machines */
@@ -343,7 +346,10 @@ asmlinkage long sys_truncate64(const char __user * path, loff_t length)
 
 asmlinkage long sys_ftruncate64(unsigned int fd, loff_t length)
 {
-	return do_sys_ftruncate(fd, length, 0);
+	long ret = do_sys_ftruncate(fd, length, 0);
+	/* avoid REGPARM breakage on x86: */
+	prevent_tail_call(ret);
+	return ret;
 }
 #endif
 
@@ -1093,20 +1099,30 @@ long do_sys_open(int dfd, const char __user *filename, int flags, int mode)
 
 asmlinkage long sys_open(const char __user *filename, int flags, int mode)
 {
+	long ret;
+
 	if (force_o_largefile())
 		flags |= O_LARGEFILE;
 
-	return do_sys_open(AT_FDCWD, filename, flags, mode);
+	ret = do_sys_open(AT_FDCWD, filename, flags, mode);
+	/* avoid REGPARM breakage on x86: */
+	prevent_tail_call(ret);
+	return ret;
 }
 EXPORT_SYMBOL_GPL(sys_open);
 
 asmlinkage long sys_openat(int dfd, const char __user *filename, int flags,
 			   int mode)
 {
+	long ret;
+
 	if (force_o_largefile())
 		flags |= O_LARGEFILE;
 
-	return do_sys_open(dfd, filename, flags, mode);
+	ret = do_sys_open(dfd, filename, flags, mode);
+	/* avoid REGPARM breakage on x86: */
+	prevent_tail_call(ret);
+	return ret;
 }
 EXPORT_SYMBOL_GPL(sys_openat);
 
diff --git a/fs/partitions/check.c b/fs/partitions/check.c
index f3b6af07172..45ae7dd3c65 100644
--- a/fs/partitions/check.c
+++ b/fs/partitions/check.c
@@ -372,6 +372,7 @@ static char *make_block_name(struct gendisk *disk)
 	char *name;
 	static char *block_str = "block:";
 	int size;
+	char *s;
 
 	size = strlen(block_str) + strlen(disk->disk_name) + 1;
 	name = kmalloc(size, GFP_KERNEL);
@@ -379,6 +380,10 @@ static char *make_block_name(struct gendisk *disk)
 		return NULL;
 	strcpy(name, block_str);
 	strcat(name, disk->disk_name);
+	/* ewww... some of these buggers have / in name... */
+	s = strchr(name, '/');
+	if (s)
+		*s = '!';
 	return name;
 }
 
diff --git a/fs/pipe.c b/fs/pipe.c
index 7fefb10db8d..5acd8954aaa 100644
--- a/fs/pipe.c
+++ b/fs/pipe.c
@@ -55,7 +55,8 @@ void pipe_wait(struct pipe_inode_info *pipe)
 }
 
 static int
-pipe_iov_copy_from_user(void *to, struct iovec *iov, unsigned long len)
+pipe_iov_copy_from_user(void *to, struct iovec *iov, unsigned long len,
+			int atomic)
 {
 	unsigned long copy;
 
@@ -64,8 +65,13 @@ pipe_iov_copy_from_user(void *to, struct iovec *iov, unsigned long len)
 			iov++;
 		copy = min_t(unsigned long, len, iov->iov_len);
 
-		if (copy_from_user(to, iov->iov_base, copy))
-			return -EFAULT;
+		if (atomic) {
+			if (__copy_from_user_inatomic(to, iov->iov_base, copy))
+				return -EFAULT;
+		} else {
+			if (copy_from_user(to, iov->iov_base, copy))
+				return -EFAULT;
+		}
 		to += copy;
 		len -= copy;
 		iov->iov_base += copy;
@@ -75,7 +81,8 @@ pipe_iov_copy_from_user(void *to, struct iovec *iov, unsigned long len)
 }
 
 static int
-pipe_iov_copy_to_user(struct iovec *iov, const void *from, unsigned long len)
+pipe_iov_copy_to_user(struct iovec *iov, const void *from, unsigned long len,
+		      int atomic)
 {
 	unsigned long copy;
 
@@ -84,8 +91,13 @@ pipe_iov_copy_to_user(struct iovec *iov, const void *from, unsigned long len)
 			iov++;
 		copy = min_t(unsigned long, len, iov->iov_len);
 
-		if (copy_to_user(iov->iov_base, from, copy))
-			return -EFAULT;
+		if (atomic) {
+			if (__copy_to_user_inatomic(iov->iov_base, from, copy))
+				return -EFAULT;
+		} else {
+			if (copy_to_user(iov->iov_base, from, copy))
+				return -EFAULT;
+		}
 		from += copy;
 		len -= copy;
 		iov->iov_base += copy;
@@ -94,13 +106,52 @@ pipe_iov_copy_to_user(struct iovec *iov, const void *from, unsigned long len)
 	return 0;
 }
 
+/*
+ * Attempt to pre-fault in the user memory, so we can use atomic copies.
+ * Returns the number of bytes not faulted in.
+ */
+static int iov_fault_in_pages_write(struct iovec *iov, unsigned long len)
+{
+	while (!iov->iov_len)
+		iov++;
+
+	while (len > 0) {
+		unsigned long this_len;
+
+		this_len = min_t(unsigned long, len, iov->iov_len);
+		if (fault_in_pages_writeable(iov->iov_base, this_len))
+			break;
+
+		len -= this_len;
+		iov++;
+	}
+
+	return len;
+}
+
+/*
+ * Pre-fault in the user memory, so we can use atomic copies.
+ */
+static void iov_fault_in_pages_read(struct iovec *iov, unsigned long len)
+{
+	while (!iov->iov_len)
+		iov++;
+
+	while (len > 0) {
+		unsigned long this_len;
+
+		this_len = min_t(unsigned long, len, iov->iov_len);
+		fault_in_pages_readable(iov->iov_base, this_len);
+		len -= this_len;
+		iov++;
+	}
+}
+
 static void anon_pipe_buf_release(struct pipe_inode_info *pipe,
 				  struct pipe_buffer *buf)
 {
 	struct page *page = buf->page;
 
-	buf->flags &= ~PIPE_BUF_FLAG_STOLEN;
-
 	/*
 	 * If nobody else uses this page, and we don't already have a
 	 * temporary page, let's keep track of it as a one-deep
@@ -112,38 +163,58 @@ static void anon_pipe_buf_release(struct pipe_inode_info *pipe,
 		page_cache_release(page);
 }
 
-static void * anon_pipe_buf_map(struct file *file, struct pipe_inode_info *pipe,
-				struct pipe_buffer *buf)
+void *generic_pipe_buf_map(struct pipe_inode_info *pipe,
+			   struct pipe_buffer *buf, int atomic)
 {
+	if (atomic) {
+		buf->flags |= PIPE_BUF_FLAG_ATOMIC;
+		return kmap_atomic(buf->page, KM_USER0);
+	}
+
 	return kmap(buf->page);
 }
 
-static void anon_pipe_buf_unmap(struct pipe_inode_info *pipe,
-				struct pipe_buffer *buf)
+void generic_pipe_buf_unmap(struct pipe_inode_info *pipe,
+			    struct pipe_buffer *buf, void *map_data)
 {
-	kunmap(buf->page);
+	if (buf->flags & PIPE_BUF_FLAG_ATOMIC) {
+		buf->flags &= ~PIPE_BUF_FLAG_ATOMIC;
+		kunmap_atomic(map_data, KM_USER0);
+	} else
+		kunmap(buf->page);
 }
 
-static int anon_pipe_buf_steal(struct pipe_inode_info *pipe,
-			       struct pipe_buffer *buf)
+int generic_pipe_buf_steal(struct pipe_inode_info *pipe,
+			   struct pipe_buffer *buf)
 {
-	buf->flags |= PIPE_BUF_FLAG_STOLEN;
-	return 0;
+	struct page *page = buf->page;
+
+	if (page_count(page) == 1) {
+		lock_page(page);
+		return 0;
+	}
+
+	return 1;
 }
 
-static void anon_pipe_buf_get(struct pipe_inode_info *info,
-			      struct pipe_buffer *buf)
+void generic_pipe_buf_get(struct pipe_inode_info *info, struct pipe_buffer *buf)
 {
 	page_cache_get(buf->page);
 }
 
+int generic_pipe_buf_pin(struct pipe_inode_info *info, struct pipe_buffer *buf)
+{
+	return 0;
+}
+
 static struct pipe_buf_operations anon_pipe_buf_ops = {
 	.can_merge = 1,
-	.map = anon_pipe_buf_map,
-	.unmap = anon_pipe_buf_unmap,
+	.map = generic_pipe_buf_map,
+	.unmap = generic_pipe_buf_unmap,
+	.pin = generic_pipe_buf_pin,
 	.release = anon_pipe_buf_release,
-	.steal = anon_pipe_buf_steal,
-	.get = anon_pipe_buf_get,
+	.steal = generic_pipe_buf_steal,
+	.get = generic_pipe_buf_get,
 };
 
 static ssize_t
@@ -174,22 +245,33 @@ pipe_readv(struct file *filp, const struct iovec *_iov,
 			struct pipe_buf_operations *ops = buf->ops;
 			void *addr;
 			size_t chars = buf->len;
-			int error;
+			int error, atomic;
 
 			if (chars > total_len)
 				chars = total_len;
 
-			addr = ops->map(filp, pipe, buf);
-			if (IS_ERR(addr)) {
+			error = ops->pin(pipe, buf);
+			if (error) {
 				if (!ret)
-					ret = PTR_ERR(addr);
+					error = ret;
 				break;
 			}
-			error = pipe_iov_copy_to_user(iov, addr + buf->offset, chars);
-			ops->unmap(pipe, buf);
+
+			atomic = !iov_fault_in_pages_write(iov, chars);
+redo:
+			addr = ops->map(pipe, buf, atomic);
+			error = pipe_iov_copy_to_user(iov, addr + buf->offset, chars, atomic);
+			ops->unmap(pipe, buf, addr);
 			if (unlikely(error)) {
+				/*
+				 * Just retry with the slow path if we failed.
+				 */
+				if (atomic) {
+					atomic = 0;
+					goto redo;
+				}
 				if (!ret)
-					ret = -EFAULT;
+					ret = error;
 				break;
 			}
 			ret += chars;
@@ -293,21 +375,28 @@ pipe_writev(struct file *filp, const struct iovec *_iov,
 		int offset = buf->offset + buf->len;
 
 		if (ops->can_merge && offset + chars <= PAGE_SIZE) {
+			int error, atomic = 1;
 			void *addr;
-			int error;
 
-			addr = ops->map(filp, pipe, buf);
-			if (IS_ERR(addr)) {
-				error = PTR_ERR(addr);
+			error = ops->pin(pipe, buf);
+			if (error)
 				goto out;
-			}
+
+			iov_fault_in_pages_read(iov, chars);
+redo1:
+			addr = ops->map(pipe, buf, atomic);
 			error = pipe_iov_copy_from_user(offset + addr, iov,
-							chars);
-			ops->unmap(pipe, buf);
+							chars, atomic);
+			ops->unmap(pipe, buf, addr);
 			ret = error;
 			do_wakeup = 1;
-			if (error)
+			if (error) {
+				if (atomic) {
+					atomic = 0;
+					goto redo1;
+				}
 				goto out;
+			}
 			buf->len += chars;
 			total_len -= chars;
 			ret = chars;
@@ -330,7 +419,8 @@ pipe_writev(struct file *filp, const struct iovec *_iov,
 			int newbuf = (pipe->curbuf + bufs) & (PIPE_BUFFERS-1);
 			struct pipe_buffer *buf = pipe->bufs + newbuf;
 			struct page *page = pipe->tmp_page;
-			int error;
+			char *src;
+			int error, atomic = 1;
 
 			if (!page) {
 				page = alloc_page(GFP_HIGHUSER);
@@ -350,11 +440,27 @@ pipe_writev(struct file *filp, const struct iovec *_iov,
 			if (chars > total_len)
 				chars = total_len;
 
-			error = pipe_iov_copy_from_user(kmap(page), iov, chars);
-			kunmap(page);
+			iov_fault_in_pages_read(iov, chars);
+redo2:
+			if (atomic)
+				src = kmap_atomic(page, KM_USER0);
+			else
+				src = kmap(page);
+
+			error = pipe_iov_copy_from_user(src, iov, chars,
+							atomic);
+			if (atomic)
+				kunmap_atomic(src, KM_USER0);
+			else
+				kunmap(page);
+
 			if (unlikely(error)) {
+				if (atomic) {
+					atomic = 0;
+					goto redo2;
+				}
 				if (!ret)
-					ret = -EFAULT;
+					ret = error;
 				break;
 			}
 			ret += chars;
diff --git a/fs/proc/base.c b/fs/proc/base.c
index a3a3eecef68..6cc77dc3f3f 100644
--- a/fs/proc/base.c
+++ b/fs/proc/base.c
@@ -297,16 +297,20 @@ static int proc_fd_link(struct inode *inode, struct dentry **dentry, struct vfsm
 
 	files = get_files_struct(task);
 	if (files) {
-		rcu_read_lock();
+		/*
+		 * We are not taking a ref to the file structure, so we must
+		 * hold ->file_lock.
+		 */
+		spin_lock(&files->file_lock);
 		file = fcheck_files(files, fd);
 		if (file) {
 			*mnt = mntget(file->f_vfsmnt);
 			*dentry = dget(file->f_dentry);
-			rcu_read_unlock();
+			spin_unlock(&files->file_lock);
 			put_files_struct(files);
 			return 0;
 		}
-		rcu_read_unlock();
+		spin_unlock(&files->file_lock);
 		put_files_struct(files);
 	}
 	return -ENOENT;
@@ -1523,7 +1527,12 @@ static struct dentry *proc_lookupfd(struct inode * dir, struct dentry * dentry,
 	if (!files)
 		goto out_unlock;
 	inode->i_mode = S_IFLNK;
-	rcu_read_lock();
+
+	/*
+	 * We are not taking a ref to the file structure, so we must
+	 * hold ->file_lock.
+	 */
+	spin_lock(&files->file_lock);
 	file = fcheck_files(files, fd);
 	if (!file)
 		goto out_unlock2;
@@ -1531,7 +1540,7 @@ static struct dentry *proc_lookupfd(struct inode * dir, struct dentry * dentry,
 		inode->i_mode |= S_IRUSR | S_IXUSR;
 	if (file->f_mode & 2)
 		inode->i_mode |= S_IWUSR | S_IXUSR;
-	rcu_read_unlock();
+	spin_unlock(&files->file_lock);
 	put_files_struct(files);
 	inode->i_op = &proc_pid_link_inode_operations;
 	inode->i_size = 64;
@@ -1541,7 +1550,7 @@ static struct dentry *proc_lookupfd(struct inode * dir, struct dentry * dentry,
 	return NULL;
 
 out_unlock2:
-	rcu_read_unlock();
+	spin_unlock(&files->file_lock);
 	put_files_struct(files);
 out_unlock:
 	iput(inode);
diff --git a/fs/reiserfs/xattr_acl.c b/fs/reiserfs/xattr_acl.c
index 58c418fbca2..97ae1b92bc4 100644
--- a/fs/reiserfs/xattr_acl.c
+++ b/fs/reiserfs/xattr_acl.c
@@ -408,8 +408,9 @@ int reiserfs_cache_default_acl(struct inode *inode)
 		acl = reiserfs_get_acl(inode, ACL_TYPE_DEFAULT);
 		reiserfs_read_unlock_xattrs(inode->i_sb);
 		reiserfs_read_unlock_xattr_i(inode);
-		ret = acl ? 1 : 0;
-		posix_acl_release(acl);
+		ret = (acl && !IS_ERR(acl));
+		if (ret)
+			posix_acl_release(acl);
 	}
 
 	return ret;
diff --git a/fs/splice.c b/fs/splice.c
index 8d57e89924a..7fb04970c72 100644
--- a/fs/splice.c
+++ b/fs/splice.c
@@ -27,15 +27,22 @@
 #include <linux/buffer_head.h>
 #include <linux/module.h>
 #include <linux/syscalls.h>
+#include <linux/uio.h>
+
+struct partial_page {
+	unsigned int offset;
+	unsigned int len;
+};
 
 /*
- * Passed to the actors
+ * Passed to splice_to_pipe
  */
-struct splice_desc {
-	unsigned int len, total_len;	/* current and remaining length */
+struct splice_pipe_desc {
+	struct page **pages;		/* page map */
+	struct partial_page *partial;	/* pages[] may not be contig */
+	int nr_pages;			/* number of pages in map */
 	unsigned int flags;		/* splice flags */
-	struct file *file;		/* file to read/write */
-	loff_t pos;			/* file position */
+	struct pipe_buf_operations *ops;/* ops associated with output pipe */
 };
 
 /*
@@ -50,7 +57,8 @@ static int page_cache_pipe_buf_steal(struct pipe_inode_info *info,
 	struct page *page = buf->page;
 	struct address_space *mapping = page_mapping(page);
 
-	WARN_ON(!PageLocked(page));
+	lock_page(page);
+
 	WARN_ON(!PageUptodate(page));
 
 	/*
@@ -65,10 +73,11 @@ static int page_cache_pipe_buf_steal(struct pipe_inode_info *info,
 	if (PagePrivate(page))
 		try_to_release_page(page, mapping_gfp_mask(mapping));
 
-	if (!remove_mapping(mapping, page))
+	if (!remove_mapping(mapping, page)) {
+		unlock_page(page);
 		return 1;
+	}
 
-	buf->flags |= PIPE_BUF_FLAG_STOLEN | PIPE_BUF_FLAG_LRU;
 	return 0;
 }
 
@@ -76,13 +85,10 @@ static void page_cache_pipe_buf_release(struct pipe_inode_info *info,
 					struct pipe_buffer *buf)
 {
 	page_cache_release(buf->page);
-	buf->page = NULL;
-	buf->flags &= ~(PIPE_BUF_FLAG_STOLEN | PIPE_BUF_FLAG_LRU);
 }
 
-static void *page_cache_pipe_buf_map(struct file *file,
-				     struct pipe_inode_info *info,
-				     struct pipe_buffer *buf)
+static int page_cache_pipe_buf_pin(struct pipe_inode_info *info,
+				   struct pipe_buffer *buf)
 {
 	struct page *page = buf->page;
 	int err;
@@ -108,51 +114,58 @@ static void *page_cache_pipe_buf_map(struct file *file,
 		}
 
 		/*
-		 * Page is ok afterall, fall through to mapping.
+		 * Page is ok afterall, we are done.
 		 */
 		unlock_page(page);
 	}
 
-	return kmap(page);
+	return 0;
 error:
 	unlock_page(page);
-	return ERR_PTR(err);
+	return err;
 }
 
-static void page_cache_pipe_buf_unmap(struct pipe_inode_info *info,
-				      struct pipe_buffer *buf)
-{
-	kunmap(buf->page);
-}
+static struct pipe_buf_operations page_cache_pipe_buf_ops = {
+	.can_merge = 0,
+	.map = generic_pipe_buf_map,
+	.unmap = generic_pipe_buf_unmap,
+	.pin = page_cache_pipe_buf_pin,
+	.release = page_cache_pipe_buf_release,
+	.steal = page_cache_pipe_buf_steal,
+	.get = generic_pipe_buf_get,
+};
 
-static void page_cache_pipe_buf_get(struct pipe_inode_info *info,
+static int user_page_pipe_buf_steal(struct pipe_inode_info *pipe,
 				    struct pipe_buffer *buf)
 {
-	page_cache_get(buf->page);
+	if (!(buf->flags & PIPE_BUF_FLAG_GIFT))
+		return 1;
+
+	return generic_pipe_buf_steal(pipe, buf);
 }
 
-static struct pipe_buf_operations page_cache_pipe_buf_ops = {
+static struct pipe_buf_operations user_page_pipe_buf_ops = {
 	.can_merge = 0,
-	.map = page_cache_pipe_buf_map,
-	.unmap = page_cache_pipe_buf_unmap,
+	.map = generic_pipe_buf_map,
+	.unmap = generic_pipe_buf_unmap,
+	.pin = generic_pipe_buf_pin,
 	.release = page_cache_pipe_buf_release,
-	.steal = page_cache_pipe_buf_steal,
-	.get = page_cache_pipe_buf_get,
+	.steal = user_page_pipe_buf_steal,
+	.get = generic_pipe_buf_get,
 };
 
 /*
  * Pipe output worker. This sets up our pipe format with the page cache
  * pipe buffer operations. Otherwise very similar to the regular pipe_writev().
  */
-static ssize_t move_to_pipe(struct pipe_inode_info *pipe, struct page **pages,
-			    int nr_pages, unsigned long offset,
-			    unsigned long len, unsigned int flags)
+static ssize_t splice_to_pipe(struct pipe_inode_info *pipe,
+			      struct splice_pipe_desc *spd)
 {
-	int ret, do_wakeup, i;
+	int ret, do_wakeup, page_nr;
 
 	ret = 0;
 	do_wakeup = 0;
-	i = 0;
+	page_nr = 0;
 
 	if (pipe->inode)
 		mutex_lock(&pipe->inode->i_mutex);
@@ -168,27 +181,22 @@ static ssize_t move_to_pipe(struct pipe_inode_info *pipe, struct page **pages,
 		if (pipe->nrbufs < PIPE_BUFFERS) {
 			int newbuf = (pipe->curbuf + pipe->nrbufs) & (PIPE_BUFFERS - 1);
 			struct pipe_buffer *buf = pipe->bufs + newbuf;
-			struct page *page = pages[i++];
-			unsigned long this_len;
 
-			this_len = PAGE_CACHE_SIZE - offset;
-			if (this_len > len)
-				this_len = len;
+			buf->page = spd->pages[page_nr];
+			buf->offset = spd->partial[page_nr].offset;
+			buf->len = spd->partial[page_nr].len;
+			buf->ops = spd->ops;
+			if (spd->flags & SPLICE_F_GIFT)
+				buf->flags |= PIPE_BUF_FLAG_GIFT;
 
-			buf->page = page;
-			buf->offset = offset;
-			buf->len = this_len;
-			buf->ops = &page_cache_pipe_buf_ops;
 			pipe->nrbufs++;
+			page_nr++;
+			ret += buf->len;
+
 			if (pipe->inode)
 				do_wakeup = 1;
 
-			ret += this_len;
-			len -= this_len;
-			offset = 0;
-			if (!--nr_pages)
-				break;
-			if (!len)
+			if (!--spd->nr_pages)
 				break;
 			if (pipe->nrbufs < PIPE_BUFFERS)
 				continue;
@@ -196,7 +204,7 @@ static ssize_t move_to_pipe(struct pipe_inode_info *pipe, struct page **pages,
 			break;
 		}
 
-		if (flags & SPLICE_F_NONBLOCK) {
+		if (spd->flags & SPLICE_F_NONBLOCK) {
 			if (!ret)
 				ret = -EAGAIN;
 			break;
@@ -231,8 +239,8 @@ static ssize_t move_to_pipe(struct pipe_inode_info *pipe, struct page **pages,
 		kill_fasync(&pipe->fasync_readers, SIGIO, POLL_IN);
 	}
 
-	while (i < nr_pages)
-		page_cache_release(pages[i++]);
+	while (page_nr < spd->nr_pages)
+		page_cache_release(spd->pages[page_nr++]);
 
 	return ret;
 }
@@ -243,15 +251,24 @@ __generic_file_splice_read(struct file *in, loff_t *ppos,
 			   unsigned int flags)
 {
 	struct address_space *mapping = in->f_mapping;
-	unsigned int offset, nr_pages;
+	unsigned int loff, nr_pages;
 	struct page *pages[PIPE_BUFFERS];
+	struct partial_page partial[PIPE_BUFFERS];
 	struct page *page;
-	pgoff_t index;
-	int i, error;
+	pgoff_t index, end_index;
+	loff_t isize;
+	size_t total_len;
+	int error, page_nr;
+	struct splice_pipe_desc spd = {
+		.pages = pages,
+		.partial = partial,
+		.flags = flags,
+		.ops = &page_cache_pipe_buf_ops,
+	};
 
 	index = *ppos >> PAGE_CACHE_SHIFT;
-	offset = *ppos & ~PAGE_CACHE_MASK;
-	nr_pages = (len + offset + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
+	loff = *ppos & ~PAGE_CACHE_MASK;
+	nr_pages = (len + loff + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
 
 	if (nr_pages > PIPE_BUFFERS)
 		nr_pages = PIPE_BUFFERS;
@@ -261,49 +278,92 @@ __generic_file_splice_read(struct file *in, loff_t *ppos,
 	 * read-ahead if this is a non-zero offset (we are likely doing small
 	 * chunk splice and the page is already there) for a single page.
 	 */
-	if (!offset || nr_pages > 1)
-		do_page_cache_readahead(mapping, in, index, nr_pages);
+	if (!loff || nr_pages > 1)
+		page_cache_readahead(mapping, &in->f_ra, in, index, nr_pages);
 
 	/*
 	 * Now fill in the holes:
 	 */
 	error = 0;
-	for (i = 0; i < nr_pages; i++, index++) {
-find_page:
+	total_len = 0;
+
+	/*
+	 * Lookup the (hopefully) full range of pages we need.
+	 */
+	spd.nr_pages = find_get_pages_contig(mapping, index, nr_pages, pages);
+
+	/*
+	 * If find_get_pages_contig() returned fewer pages than we needed,
+	 * allocate the rest.
+	 */
+	index += spd.nr_pages;
+	while (spd.nr_pages < nr_pages) {
 		/*
-		 * lookup the page for this index
+		 * Page could be there, find_get_pages_contig() breaks on
+		 * the first hole.
 		 */
 		page = find_get_page(mapping, index);
 		if (!page) {
 			/*
-			 * If in nonblock mode then dont block on
-			 * readpage (we've kicked readahead so there
-			 * will be asynchronous progress):
+			 * Make sure the read-ahead engine is notified
+			 * about this failure.
 			 */
-			if (flags & SPLICE_F_NONBLOCK)
-				break;
+			handle_ra_miss(mapping, &in->f_ra, index);
 
 			/*
-			 * page didn't exist, allocate one
+			 * page didn't exist, allocate one.
 			 */
 			page = page_cache_alloc_cold(mapping);
 			if (!page)
 				break;
 
 			error = add_to_page_cache_lru(page, mapping, index,
-						mapping_gfp_mask(mapping));
+					      mapping_gfp_mask(mapping));
 			if (unlikely(error)) {
 				page_cache_release(page);
 				break;
 			}
-
-			goto readpage;
+			/*
+			 * add_to_page_cache() locks the page, unlock it
+			 * to avoid convoluting the logic below even more.
+			 */
+			unlock_page(page);
 		}
 
+		pages[spd.nr_pages++] = page;
+		index++;
+	}
+
+	/*
+	 * Now loop over the map and see if we need to start IO on any
+	 * pages, fill in the partial map, etc.
+	 */
+	index = *ppos >> PAGE_CACHE_SHIFT;
+	nr_pages = spd.nr_pages;
+	spd.nr_pages = 0;
+	for (page_nr = 0; page_nr < nr_pages; page_nr++) {
+		unsigned int this_len;
+
+		if (!len)
+			break;
+
+		/*
+		 * this_len is the max we'll use from this page
+		 */
+		this_len = min_t(unsigned long, len, PAGE_CACHE_SIZE - loff);
+		page = pages[page_nr];
+
 		/*
 		 * If the page isn't uptodate, we may need to start io on it
 		 */
 		if (!PageUptodate(page)) {
+			/*
+			 * If in nonblock mode then dont block on waiting
+			 * for an in-flight io page
+			 */
+			if (flags & SPLICE_F_NONBLOCK)
+				break;
+
 			lock_page(page);
 
 			/*
@@ -313,7 +373,6 @@ find_page:
 			 */
 			if (!page->mapping) {
 				unlock_page(page);
-				page_cache_release(page);
 				break;
 			}
 			/*
@@ -324,25 +383,66 @@ find_page:
 				goto fill_it;
 			}
 
-readpage:
 			/*
 			 * need to read in the page
 			 */
 			error = mapping->a_ops->readpage(in, page);
-
 			if (unlikely(error)) {
-				page_cache_release(page);
+				/*
+				 * We really should re-lookup the page here,
+				 * but it complicates things a lot. Instead
+				 * lets just do what we already stored, and
+				 * we'll get it the next time we are called.
+				 */
 				if (error == AOP_TRUNCATED_PAGE)
-					goto find_page;
+					error = 0;
+
 				break;
 			}
+
+			/*
+			 * i_size must be checked after ->readpage().
+			 */
+			isize = i_size_read(mapping->host);
+			end_index = (isize - 1) >> PAGE_CACHE_SHIFT;
+			if (unlikely(!isize || index > end_index))
+				break;
+
+			/*
+			 * if this is the last page, see if we need to shrink
+			 * the length and stop
+			 */
+			if (end_index == index) {
+				loff = PAGE_CACHE_SIZE - (isize & ~PAGE_CACHE_MASK);
+				if (total_len + loff > isize)
+					break;
+				/*
+				 * force quit after adding this page
+				 */
+				len = this_len;
+				this_len = min(this_len, loff);
+				loff = 0;
+			}
 		}
 fill_it:
-		pages[i] = page;
+		partial[page_nr].offset = loff;
+		partial[page_nr].len = this_len;
+		len -= this_len;
+		total_len += this_len;
+		loff = 0;
+		spd.nr_pages++;
+		index++;
 	}
 
-	if (i)
-		return move_to_pipe(pipe, pages, i, offset, len, flags);
+	/*
+	 * Release any pages at the end, if we quit early. 'i' is how far
+	 * we got, 'nr_pages' is how many pages are in the map.
+	 */
+	while (page_nr < nr_pages)
+		page_cache_release(pages[page_nr++]);
+
+	if (spd.nr_pages)
+		return splice_to_pipe(pipe, &spd);
 
 	return error;
 }
@@ -369,17 +469,20 @@ ssize_t generic_file_splice_read(struct file *in, loff_t *ppos,
 	while (len) {
 		ret = __generic_file_splice_read(in, ppos, pipe, len, flags);
 
-		if (ret <= 0)
+		if (ret < 0)
 			break;
+		else if (!ret) {
+			if (spliced)
+				break;
+			if (flags & SPLICE_F_NONBLOCK) {
+				ret = -EAGAIN;
+				break;
+			}
+		}
 
 		*ppos += ret;
 		len -= ret;
 		spliced += ret;
-
-		if (!(flags & SPLICE_F_NONBLOCK))
-			continue;
-		ret = -EAGAIN;
-		break;
 	}
 
 	if (spliced)
@@ -392,38 +495,24 @@ EXPORT_SYMBOL(generic_file_splice_read);
 
 /*
  * Send 'sd->len' bytes to socket from 'sd->file' at position 'sd->pos'
- * using sendpage().
+ * using sendpage(). Return the number of bytes sent.
  */
 static int pipe_to_sendpage(struct pipe_inode_info *info,
 			    struct pipe_buffer *buf, struct splice_desc *sd)
 {
 	struct file *file = sd->file;
 	loff_t pos = sd->pos;
-	unsigned int offset;
-	ssize_t ret;
-	void *ptr;
-	int more;
-
-	/*
-	 * Sub-optimal, but we are limited by the pipe ->map. We don't
-	 * need a kmap'ed buffer here, we just want to make sure we
-	 * have the page pinned if the pipe page originates from the
-	 * page cache.
-	 */
-	ptr = buf->ops->map(file, info, buf);
-	if (IS_ERR(ptr))
-		return PTR_ERR(ptr);
+	int ret, more;
 
-	offset = pos & ~PAGE_CACHE_MASK;
-	more = (sd->flags & SPLICE_F_MORE) || sd->len < sd->total_len;
+	ret = buf->ops->pin(info, buf);
+	if (!ret) {
+		more = (sd->flags & SPLICE_F_MORE) || sd->len < sd->total_len;
 
-	ret = file->f_op->sendpage(file, buf->page, offset, sd->len, &pos,more);
-
-	buf->ops->unmap(info, buf);
-	if (ret == sd->len)
-		return 0;
+		ret = file->f_op->sendpage(file, buf->page, buf->offset,
+					   sd->len, &pos, more);
+	}
 
-	return -EIO;
+	return ret;
 }
 
 /*
@@ -452,56 +541,88 @@ static int pipe_to_file(struct pipe_inode_info *info, struct pipe_buffer *buf,
 	struct file *file = sd->file;
 	struct address_space *mapping = file->f_mapping;
 	gfp_t gfp_mask = mapping_gfp_mask(mapping);
-	unsigned int offset;
+	unsigned int offset, this_len;
 	struct page *page;
 	pgoff_t index;
-	char *src;
 	int ret;
 
 	/*
 	 * make sure the data in this buffer is uptodate
 	 */
-	src = buf->ops->map(file, info, buf);
-	if (IS_ERR(src))
-		return PTR_ERR(src);
+	ret = buf->ops->pin(info, buf);
+	if (unlikely(ret))
+		return ret;
 
 	index = sd->pos >> PAGE_CACHE_SHIFT;
 	offset = sd->pos & ~PAGE_CACHE_MASK;
 
+	this_len = sd->len;
+	if (this_len + offset > PAGE_CACHE_SIZE)
+		this_len = PAGE_CACHE_SIZE - offset;
+
 	/*
-	 * Reuse buf page, if SPLICE_F_MOVE is set.
+	 * Reuse buf page, if SPLICE_F_MOVE is set and we are doing a full
+	 * page.
 	 */
-	if (sd->flags & SPLICE_F_MOVE) {
+	if ((sd->flags & SPLICE_F_MOVE) && this_len == PAGE_CACHE_SIZE) {
 		/*
 		 * If steal succeeds, buf->page is now pruned from the vm
-		 * side (LRU and page cache) and we can reuse it.
+		 * side (page cache) and we can reuse it. The page will also
+		 * be locked on successful return.
 		 */
 		if (buf->ops->steal(info, buf))
 			goto find_page;
 
+		page = buf->page;
+		page_cache_get(page);
+
 		/*
-		 * this will also set the page locked
+		 * page must be on the LRU for adding to the pagecache.
+		 * Check this without grabbing the zone lock, if it isn't
+		 * the do grab the zone lock, recheck, and add if necessary.
 		 */
-		page = buf->page;
-		if (add_to_page_cache(page, mapping, index, gfp_mask))
-			goto find_page;
+		if (!PageLRU(page)) {
+			struct zone *zone = page_zone(page);
+
+			spin_lock_irq(&zone->lru_lock);
+			if (!PageLRU(page)) {
+				SetPageLRU(page);
+				add_page_to_inactive_list(zone, page);
+			}
+			spin_unlock_irq(&zone->lru_lock);
+		}
 
-		if (!(buf->flags & PIPE_BUF_FLAG_LRU))
-			lru_cache_add(page);
+		if (add_to_page_cache(page, mapping, index, gfp_mask)) {
+			page_cache_release(page);
+			unlock_page(page);
+			goto find_page;
+		}
 	} else {
 find_page:
-		ret = -ENOMEM;
-		page = find_or_create_page(mapping, index, gfp_mask);
-		if (!page)
-			goto out_nomem;
+		page = find_lock_page(mapping, index);
+		if (!page) {
+			ret = -ENOMEM;
+			page = page_cache_alloc_cold(mapping);
+			if (unlikely(!page))
+				goto out_nomem;
+
+			/*
+			 * This will also lock the page
+			 */
+			ret = add_to_page_cache_lru(page, mapping, index,
+						    gfp_mask);
+			if (unlikely(ret))
+				goto out;
+		}
 
 		/*
-		 * If the page is uptodate, it is also locked. If it isn't
-		 * uptodate, we can mark it uptodate if we are filling the
-		 * full page. Otherwise we need to read it in first...
+		 * We get here with the page locked. If the page is also
+		 * uptodate, we don't need to do more. If it isn't, we
+		 * may need to bring it in if we are not going to overwrite
+		 * the full page.
 		 */
 		if (!PageUptodate(page)) {
-			if (sd->len < PAGE_CACHE_SIZE) {
+			if (this_len < PAGE_CACHE_SIZE) {
 				ret = mapping->a_ops->readpage(file, page);
 				if (unlikely(ret))
 					goto out;
@@ -520,58 +641,59 @@ find_page:
 					ret = -EIO;
 					goto out;
 				}
-			} else {
-				WARN_ON(!PageLocked(page));
+			} else
 				SetPageUptodate(page);
-			}
 		}
 	}
 
-	ret = mapping->a_ops->prepare_write(file, page, 0, sd->len);
+	ret = mapping->a_ops->prepare_write(file, page, offset, offset+this_len);
 	if (ret == AOP_TRUNCATED_PAGE) {
 		page_cache_release(page);
 		goto find_page;
 	} else if (ret)
 		goto out;
 
-	if (!(buf->flags & PIPE_BUF_FLAG_STOLEN)) {
-		char *dst = kmap_atomic(page, KM_USER0);
+	if (buf->page != page) {
+		/*
+		 * Careful, ->map() uses KM_USER0!
+		 */
+		char *src = buf->ops->map(info, buf, 1);
+		char *dst = kmap_atomic(page, KM_USER1);
 
-		memcpy(dst + offset, src + buf->offset, sd->len);
+		memcpy(dst + offset, src + buf->offset, this_len);
 		flush_dcache_page(page);
-		kunmap_atomic(dst, KM_USER0);
+		kunmap_atomic(dst, KM_USER1);
+		buf->ops->unmap(info, buf, src);
 	}
 
-	ret = mapping->a_ops->commit_write(file, page, 0, sd->len);
-	if (ret == AOP_TRUNCATED_PAGE) {
+	ret = mapping->a_ops->commit_write(file, page, offset, offset+this_len);
+	if (!ret) {
+		/*
+		 * Return the number of bytes written and mark page as
+		 * accessed, we are now done!
+		 */
+		ret = this_len;
+		mark_page_accessed(page);
+		balance_dirty_pages_ratelimited(mapping);
+	} else if (ret == AOP_TRUNCATED_PAGE) {
 		page_cache_release(page);
 		goto find_page;
-	} else if (ret)
-		goto out;
-
-	mark_page_accessed(page);
-	balance_dirty_pages_ratelimited(mapping);
-out:
-	if (!(buf->flags & PIPE_BUF_FLAG_STOLEN)) {
-		page_cache_release(page);
-		unlock_page(page);
 	}
+out:
+	page_cache_release(page);
+	unlock_page(page);
 out_nomem:
-	buf->ops->unmap(info, buf);
 	return ret;
 }
 
-typedef int (splice_actor)(struct pipe_inode_info *, struct pipe_buffer *,
-			   struct splice_desc *);
-
 /*
  * Pipe input worker. Most of this logic works like a regular pipe, the
  * key here is the 'actor' worker passed in that actually moves the data
  * to the wanted destination. See pipe_to_file/pipe_to_sendpage above.
  */
-static ssize_t move_from_pipe(struct pipe_inode_info *pipe, struct file *out,
-			      loff_t *ppos, size_t len, unsigned int flags,
-			      splice_actor *actor)
+ssize_t splice_from_pipe(struct pipe_inode_info *pipe, struct file *out,
+			 loff_t *ppos, size_t len, unsigned int flags,
+			 splice_actor *actor)
 {
 	int ret, do_wakeup, err;
 	struct splice_desc sd;
@@ -597,16 +719,22 @@ static ssize_t move_from_pipe(struct pipe_inode_info *pipe, struct file *out,
 				sd.len = sd.total_len;
 
 			err = actor(pipe, buf, &sd);
-			if (err) {
+			if (err <= 0) {
 				if (!ret && err != -ENODATA)
 					ret = err;
 
 				break;
 			}
 
-			ret += sd.len;
-			buf->offset += sd.len;
-			buf->len -= sd.len;
+			ret += err;
+			buf->offset += err;
+			buf->len -= err;
+
+			sd.len -= err;
+			sd.pos += err;
+			sd.total_len -= err;
+			if (sd.len)
+				continue;
 
 			if (!buf->len) {
 				buf->ops = NULL;
@@ -617,8 +745,6 @@ static ssize_t move_from_pipe(struct pipe_inode_info *pipe, struct file *out,
 					do_wakeup = 1;
 			}
 
-			sd.pos += sd.len;
-			sd.total_len -= sd.len;
 			if (!sd.total_len)
 				break;
 		}
@@ -686,23 +812,27 @@ generic_file_splice_write(struct pipe_inode_info *pipe, struct file *out,
 	struct address_space *mapping = out->f_mapping;
 	ssize_t ret;
 
-	ret = move_from_pipe(pipe, out, ppos, len, flags, pipe_to_file);
-
-	/*
-	 * If file or inode is SYNC and we actually wrote some data, sync it.
-	 */
-	if (unlikely((out->f_flags & O_SYNC) || IS_SYNC(mapping->host))
-	    && ret > 0) {
+	ret = splice_from_pipe(pipe, out, ppos, len, flags, pipe_to_file);
+	if (ret > 0) {
 		struct inode *inode = mapping->host;
-		int err;
 
-		mutex_lock(&inode->i_mutex);
-		err = generic_osync_inode(mapping->host, mapping,
-					  OSYNC_METADATA|OSYNC_DATA);
-		mutex_unlock(&inode->i_mutex);
+		*ppos += ret;
 
-		if (err)
-			ret = err;
+		/*
+		 * If file or inode is SYNC and we actually wrote some data,
+		 * sync it.
+		 */
+		if (unlikely((out->f_flags & O_SYNC) || IS_SYNC(inode))) {
+			int err;
+
+			mutex_lock(&inode->i_mutex);
+			err = generic_osync_inode(inode, mapping,
+						  OSYNC_METADATA|OSYNC_DATA);
+			mutex_unlock(&inode->i_mutex);
+
+			if (err)
+				ret = err;
+		}
 	}
 
 	return ret;
@@ -724,7 +854,7 @@ EXPORT_SYMBOL(generic_file_splice_write);
 ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe, struct file *out,
 				loff_t *ppos, size_t len, unsigned int flags)
 {
-	return move_from_pipe(pipe, out, ppos, len, flags, pipe_to_sendpage);
+	return splice_from_pipe(pipe, out, ppos, len, flags, pipe_to_sendpage);
 }
 
 EXPORT_SYMBOL(generic_splice_sendpage);
@@ -811,7 +941,7 @@ long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
 
 		/*
 		 * We don't have an immediate reader, but we'll read the stuff
-		 * out of the pipe right after the move_to_pipe(). So set
+		 * out of the pipe right after the splice_to_pipe(). So set
 		 * PIPE_READERS appropriately.
 		 */
 		pipe->readers = 1;
@@ -904,6 +1034,7 @@ static long do_splice(struct file *in, loff_t __user *off_in,
 {
 	struct pipe_inode_info *pipe;
 	loff_t offset, *off;
+	long ret;
 
 	pipe = in->f_dentry->d_inode->i_pipe;
 	if (pipe) {
@@ -918,7 +1049,12 @@ static long do_splice(struct file *in, loff_t __user *off_in,
 		} else
 			off = &out->f_pos;
 
-		return do_splice_from(pipe, out, off, len, flags);
+		ret = do_splice_from(pipe, out, off, len, flags);
+
+		if (off_out && copy_to_user(off_out, off, sizeof(loff_t)))
+			ret = -EFAULT;
+
+		return ret;
 	}
 
 	pipe = out->f_dentry->d_inode->i_pipe;
@@ -934,12 +1070,195 @@ static long do_splice(struct file *in, loff_t __user *off_in,
 		} else
 			off = &in->f_pos;
 
-		return do_splice_to(in, off, pipe, len, flags);
+		ret = do_splice_to(in, off, pipe, len, flags);
+
+		if (off_in && copy_to_user(off_in, off, sizeof(loff_t)))
+			ret = -EFAULT;
+
+		return ret;
 	}
 
 	return -EINVAL;
 }
 
+/*
+ * Map an iov into an array of pages and offset/length tupples. With the
+ * partial_page structure, we can map several non-contiguous ranges into
+ * our ones pages[] map instead of splitting that operation into pieces.
+ * Could easily be exported as a generic helper for other users, in which
+ * case one would probably want to add a 'max_nr_pages' parameter as well.
+ */
+static int get_iovec_page_array(const struct iovec __user *iov,
+				unsigned int nr_vecs, struct page **pages,
+				struct partial_page *partial, int aligned)
+{
+	int buffers = 0, error = 0;
+
+	/*
+	 * It's ok to take the mmap_sem for reading, even
+	 * across a "get_user()".
+	 */
+	down_read(&current->mm->mmap_sem);
+
+	while (nr_vecs) {
+		unsigned long off, npages;
+		void __user *base;
+		size_t len;
+		int i;
+
+		/*
+		 * Get user address base and length for this iovec.
+		 */
+		error = get_user(base, &iov->iov_base);
+		if (unlikely(error))
+			break;
+		error = get_user(len, &iov->iov_len);
+		if (unlikely(error))
+			break;
+
+		/*
+		 * Sanity check this iovec. 0 read succeeds.
+		 */
+		if (unlikely(!len))
+			break;
+		error = -EFAULT;
+		if (unlikely(!base))
+			break;
+
+		/*
+		 * Get this base offset and number of pages, then map
+		 * in the user pages.
+		 */
+		off = (unsigned long) base & ~PAGE_MASK;
+
+		/*
+		 * If asked for alignment, the offset must be zero and the
+		 * length a multiple of the PAGE_SIZE.
+		 */
+		error = -EINVAL;
+		if (aligned && (off || len & ~PAGE_MASK))
+			break;
+
+		npages = (off + len + PAGE_SIZE - 1) >> PAGE_SHIFT;
+		if (npages > PIPE_BUFFERS - buffers)
+			npages = PIPE_BUFFERS - buffers;
+
+		error = get_user_pages(current, current->mm,
+				       (unsigned long) base, npages, 0, 0,
+				       &pages[buffers], NULL);
+
+		if (unlikely(error <= 0))
+			break;
+
+		/*
+		 * Fill this contiguous range into the partial page map.
+		 */
+		for (i = 0; i < error; i++) {
+			const int plen = min_t(size_t, len, PAGE_SIZE - off);
+
+			partial[buffers].offset = off;
+			partial[buffers].len = plen;
+
+			off = 0;
+			len -= plen;
+			buffers++;
+		}
+
+		/*
+		 * We didn't complete this iov, stop here since it probably
+		 * means we have to move some of this into a pipe to
+		 * be able to continue.
+		 */
+		if (len)
+			break;
+
+		/*
+		 * Don't continue if we mapped fewer pages than we asked for,
+		 * or if we mapped the max number of pages that we have
+		 * room for.
+		 */
+		if (error < npages || buffers == PIPE_BUFFERS)
+			break;
+
+		nr_vecs--;
+		iov++;
+	}
+
+	up_read(&current->mm->mmap_sem);
+
+	if (buffers)
+		return buffers;
+
+	return error;
+}
+
+/*
+ * vmsplice splices a user address range into a pipe. It can be thought of
+ * as splice-from-memory, where the regular splice is splice-from-file (or
+ * to file). In both cases the output is a pipe, naturally.
+ *
+ * Note that vmsplice only supports splicing _from_ user memory to a pipe,
+ * not the other way around. Splicing from user memory is a simple operation
+ * that can be supported without any funky alignment restrictions or nasty
+ * vm tricks. We simply map in the user memory and fill them into a pipe.
+ * The reverse isn't quite as easy, though. There are two possible solutions
+ * for that:
+ *
+ *	- memcpy() the data internally, at which point we might as well just
+ *	  do a regular read() on the buffer anyway.
+ *	- Lots of nasty vm tricks, that are neither fast nor flexible (it
+ *	  has restriction limitations on both ends of the pipe).
+ *
+ * Alas, it isn't here.
+ *
+ */
+static long do_vmsplice(struct file *file, const struct iovec __user *iov,
+			unsigned long nr_segs, unsigned int flags)
+{
+	struct pipe_inode_info *pipe = file->f_dentry->d_inode->i_pipe;
+	struct page *pages[PIPE_BUFFERS];
+	struct partial_page partial[PIPE_BUFFERS];
+	struct splice_pipe_desc spd = {
+		.pages = pages,
+		.partial = partial,
+		.flags = flags,
+		.ops = &user_page_pipe_buf_ops,
+	};
+
+	if (unlikely(!pipe))
+		return -EBADF;
+	if (unlikely(nr_segs > UIO_MAXIOV))
+		return -EINVAL;
+	else if (unlikely(!nr_segs))
+		return 0;
+
+	spd.nr_pages = get_iovec_page_array(iov, nr_segs, pages, partial,
+					    flags & SPLICE_F_GIFT);
+	if (spd.nr_pages <= 0)
+		return spd.nr_pages;
+
+	return splice_to_pipe(pipe, &spd);
+}
+
+asmlinkage long sys_vmsplice(int fd, const struct iovec __user *iov,
+			     unsigned long nr_segs, unsigned int flags)
+{
+	struct file *file;
+	long error;
+	int fput;
+
+	error = -EBADF;
+	file = fget_light(fd, &fput);
+	if (file) {
+		if (file->f_mode & FMODE_WRITE)
+			error = do_vmsplice(file, iov, nr_segs, flags);
+
+		fput_light(file, fput);
+	}
+
+	return error;
+}
+
 asmlinkage long sys_splice(int fd_in, loff_t __user *off_in,
 			   int fd_out, loff_t __user *off_out,
 			   size_t len, unsigned int flags)
@@ -979,7 +1298,9 @@ static int link_pipe(struct pipe_inode_info *ipipe,
 		     size_t len, unsigned int flags)
 {
 	struct pipe_buffer *ibuf, *obuf;
-	int ret = 0, do_wakeup = 0, i;
+	int ret, do_wakeup, i, ipipe_first;
+
+	ret = do_wakeup = ipipe_first = 0;
 
 	/*
 	 * Potential ABBA deadlock, work around it by ordering lock
@@ -987,6 +1308,7 @@ static int link_pipe(struct pipe_inode_info *ipipe,
 	 * could deadlock (one doing tee from A -> B, the other from B -> A).
 	 */
 	if (ipipe->inode < opipe->inode) {
+		ipipe_first = 1;
 		mutex_lock(&ipipe->inode->i_mutex);
 		mutex_lock(&opipe->inode->i_mutex);
 	} else {
@@ -1019,6 +1341,12 @@ static int link_pipe(struct pipe_inode_info *ipipe,
 				obuf = opipe->bufs + nbuf;
 				*obuf = *ibuf;
 
+				/*
+				 * Don't inherit the gift flag, we need to
+				 * prevent multiple steals of this page.
+				 */
+				obuf->flags &= ~PIPE_BUF_FLAG_GIFT;
+
 				if (obuf->len > len)
 					obuf->len = len;
 
@@ -1035,9 +1363,11 @@ static int link_pipe(struct pipe_inode_info *ipipe,
 
 			/*
 			 * We have input available, but no output room.
-			 * If we already copied data, return that.
+			 * If we already copied data, return that. If we
+			 * need to drop the opipe lock, it must be ordered
+			 * last to avoid deadlocks.
 			 */
-			if (flags & SPLICE_F_NONBLOCK) {
+			if ((flags & SPLICE_F_NONBLOCK) || !ipipe_first) {
 				if (!ret)
 					ret = -EAGAIN;
 				break;
@@ -1071,7 +1401,12 @@ static int link_pipe(struct pipe_inode_info *ipipe,
 			if (ret)
 				break;
 		}
-		if (flags & SPLICE_F_NONBLOCK) {
+		/*
+		 * pipe_wait() drops the ipipe mutex. To avoid deadlocks
+		 * with another process, we can only safely do that if
+		 * the ipipe lock is ordered last.
+		 */
+		if ((flags & SPLICE_F_NONBLOCK) || ipipe_first) {
 			if (!ret)
 				ret = -EAGAIN;
 			break;
diff --git a/fs/stat.c b/fs/stat.c
index 9948cc1685a..0f282face32 100644
--- a/fs/stat.c
+++ b/fs/stat.c
@@ -261,7 +261,7 @@ asmlinkage long sys_newlstat(char __user *filename, struct stat __user *statbuf)
 	return error;
 }
 
-#ifndef __ARCH_WANT_STAT64
+#if !defined(__ARCH_WANT_STAT64) || defined(__ARCH_WANT_SYS_NEWFSTATAT)
 asmlinkage long sys_newfstatat(int dfd, char __user *filename,
 				struct stat __user *statbuf, int flag)
 {