Merge master.kernel.org:/home/hch/BK/xfs/linux-2.5

into penguin.transmeta.com:/home/penguin/torvalds/repositories/kernel/linux

fs/xfs/linux/xfs_linux.h

@@ -42,6 +42,7 @@
 #include <linux/sched.h>
 #include <linux/bitops.h>
 #include <linux/major.h>
+#include <linux/pagemap.h>
 #include <linux/vfs.h>
 
 #include <asm/page.h>
@@ -169,7 +170,6 @@ typedef struct xfs_dirent {		/* data from readdir() */
 #define DEFAULT_PROJID	0
 #define dfltprid	DEFAULT_PROJID
 
-#define MAXNAMELEN	256
 #define MAXPATHLEN	1024
 
 #define FINVIS		0x0100	/* don't update timestamps - XFS */

fs/xfs/linux/xfs_lrw.c

@@ -872,13 +872,12 @@ XFS_log_write_unmount_ro(bhv_desc_t	*bdp)
 	int error;
 
 	mp = XFS_BHVTOM(bdp);
-	xfs_binval(mp->m_ddev_targp);
+	pagebuf_delwri_flush(mp->m_ddev_targp, PBDF_WAIT, &pincount);
+	xfs_finish_reclaim_all(mp);
 
 	do {
-		xfs_log_force(mp, (xfs_lsn_t)0, XFS_LOG_FORCE | XFS_LOG_SYNC);
 		VFS_SYNC(XFS_MTOVFS(mp), SYNC_ATTR|SYNC_WAIT, NULL, error);
-		pagebuf_delwri_flush(mp->m_ddev_targp,
-				PBDF_WAIT, &pincount);
+		pagebuf_delwri_flush(mp->m_ddev_targp, PBDF_WAIT, &pincount);
 		if (pincount == 0) {delay(50); count++;}
 	}  while (count < 2);
 

fs/xfs/linux/xfs_vnode.h

@@ -671,13 +671,14 @@ extern void	vn_rele(struct vnode *);
 #define VN_HOLD(vp)		((void)vn_hold(vp))
 #define VN_RELE(vp)		(iput(LINVFS_GET_IP(vp)))
 
-#endif	/* ! (defined(CONFIG_XFS_VNODE_TRACING) */
+#endif	/* ! (defined(CONFIG_XFS_VNODE_TRACING)) */
 
 /*
  * Vname handling macros.
  */
 #define VNAME(dentry)		((char *) (dentry)->d_name.name)
 #define VNAMELEN(dentry)	((dentry)->d_name.len)
+#define VNAME_TO_VNODE(dentry)	(LINVFS_GET_VP((dentry)->d_inode))
 
 /*
  * Vnode spinlock manipulation.

fs/xfs/pagebuf/page_buf.c

@@ -1108,7 +1108,7 @@ _pagebuf_wait_unpin(
  *	pagebuf_iodone
  *
  *	pagebuf_iodone marks a buffer for which I/O is in progress
- *	done with respect to that I/O.	The pb_done routine, if
+ *	done with respect to that I/O.	The pb_iodone routine, if
  *	present, will be called as a side-effect.
  */
 void
@@ -1173,7 +1173,7 @@ pagebuf_ioerror(			/* mark/clear buffer error flag */
  *	pagebuf_iostart initiates I/O on a buffer, based on the flags supplied.
  *	If necessary, it will arrange for any disk space allocation required,
  *	and it will break up the request if the block mappings require it.
- *	An pb_iodone routine in the buffer supplied will only be called
+ *	The pb_iodone routine in the buffer supplied will only be called
  *	when all of the subsidiary I/O requests, if any, have been completed.
  *	pagebuf_iostart calls the pagebuf_ioinitiate routine or
  *	pagebuf_iorequest, if the former routine is not defined, to start
@@ -1317,7 +1317,7 @@ pagebuf_iorequest(			/* start real I/O		*/
 
 	/* Set the count to 1 initially, this will stop an I/O
 	 * completion callout which happens before we have started
-	 * all the I/O from calling iodone too early
+	 * all the I/O from calling pagebuf_iodone too early.
 	 */
 	atomic_set(&pb->pb_io_remaining, 1);
 

fs/xfs/support/debug.c

@@ -41,7 +41,10 @@ static char		message[256];	/* keep it off the stack */
 static spinlock_t 	xfs_err_lock = SPIN_LOCK_UNLOCKED;
 
 /* Translate from CE_FOO to KERN_FOO, err_level(CE_FOO) == KERN_FOO */
-static char		*err_level[8] = {KERN_EMERG, KERN_ALERT, KERN_CRIT,
+#define XFS_MAX_ERR_LEVEL	7
+#define XFS_ERR_MASK		((1 << 3) - 1)
+static char		*err_level[XFS_MAX_ERR_LEVEL+1] =
+					{KERN_EMERG, KERN_ALERT, KERN_CRIT,
 					 KERN_ERR, KERN_WARNING, KERN_NOTICE,
 					 KERN_INFO, KERN_DEBUG};
 
@@ -82,15 +85,22 @@ void
 cmn_err(register int level, char *fmt, ...)
 {
 	char	*fp = fmt;
+	int	len;
+	int	flags;
 	va_list	ap;
 
-	spin_lock(&xfs_err_lock);
+	level &= XFS_ERR_MASK;
+	if (level > XFS_MAX_ERR_LEVEL)
+		level = XFS_MAX_ERR_LEVEL;
+	spin_lock_irqsave(&xfs_err_lock,flags);
 	va_start(ap, fmt);
 	if (*fmt == '!') fp++;
-	vsprintf(message, fp, ap);
-	printk("%s%s\n", err_level[level], message);
+	len = vsprintf(message, fp, ap);
+	if (message[len-1] != '\n')
+		strcat(message, "\n");
+	printk("%s%s", err_level[level], message);
 	va_end(ap);
-	spin_unlock(&xfs_err_lock);
+	spin_unlock_irqrestore(&xfs_err_lock,flags);
 
 	if (level == CE_PANIC)
 		BUG();
@@ -100,10 +110,18 @@ cmn_err(register int level, char *fmt, ...)
 void
 icmn_err(register int level, char *fmt, va_list ap)
 {
-	spin_lock(&xfs_err_lock);
-	vsprintf(message, fmt, ap);
-	spin_unlock(&xfs_err_lock);
-	printk("%s%s\n", err_level[level], message);
+	int	len;
+	int	flags;
+
+	level &= XFS_ERR_MASK;
+	if(level > XFS_MAX_ERR_LEVEL)
+		level = XFS_MAX_ERR_LEVEL;
+	spin_lock_irqsave(&xfs_err_lock,flags);
+	len = vsprintf(message, fmt, ap);
+	if (message[len-1] != '\n')
+		strcat(message, "\n");
+	spin_unlock_irqrestore(&xfs_err_lock,flags);
+	printk("%s%s", err_level[level], message);
 	if (level == CE_PANIC)
 		BUG();
 }

fs/xfs/xfs_dir.c

@@ -386,9 +386,6 @@ xfs_dir_lookup(xfs_trans_t *trans, xfs_inode_t *dp, char *name, int namelen,
 	int retval;
 
 	ASSERT((dp->i_d.di_mode & IFMT) == IFDIR);
-	if (namelen >= MAXNAMELEN) {
-		return(XFS_ERROR(EINVAL));
-	}
 
 	XFS_STATS_INC(xfsstats.xs_dir_lookup);
 	/*
@@ -479,9 +476,6 @@ xfs_dir_replace(xfs_trans_t *trans, xfs_inode_t *dp, char *name, int namelen,
 	int retval;
 
 	ASSERT((dp->i_d.di_mode & IFMT) == IFDIR);
-	if (namelen >= MAXNAMELEN) {
-		return(XFS_ERROR(EINVAL));
-	}
 
 	if ((retval = xfs_dir_ino_validate(trans->t_mountp, inum)))
 		return retval;

fs/xfs/xfs_dir2.c

@@ -232,10 +232,8 @@ xfs_dir2_lookup(
 	int		v;		/* type-checking value */
 
 	ASSERT((dp->i_d.di_mode & IFMT) == IFDIR);
-	if (namelen >= MAXNAMELEN) {
-		return XFS_ERROR(EINVAL);
-	}
 	XFS_STATS_INC(xfsstats.xs_dir_lookup);
+
 	/*
 	 * Fill in the arg structure for this request.
 	 */
@@ -395,9 +393,7 @@ xfs_dir2_replace(
 	int		v;		/* type-checking value */
 
 	ASSERT((dp->i_d.di_mode & IFMT) == IFDIR);
-	if (namelen >= MAXNAMELEN) {
-		return XFS_ERROR(EINVAL);
-	}
+
 	if ((rval = xfs_dir_ino_validate(tp->t_mountp, inum))) {
 		return rval;
 	}

fs/xfs/xfs_inode.h

@@ -414,11 +414,9 @@ void xfs_ifork_next_set(xfs_inode_t *ip, int w, int n);
  * max file offset is 2^(31+PAGE_SHIFT) - 1 (due to linux page cache)
  *
  * NOTE: XFS itself can handle 2^63 - 1 (largest positive value of xfs_fsize_t)
- * but Linux can't go above 2^(31+PAGE_SHIFT)-1: the Linux VM uses a 32 bit
- * signed variable to index cache data, so 2^31 * PAGE_SIZE is as big as
- * you can go.
+ * but this is the Linux limit.
  */
-#define XFS_MAX_FILE_OFFSET	((long long)((1ULL<<(31+PAGE_SHIFT))-1ULL))
+#define XFS_MAX_FILE_OFFSET	MAX_LFS_FILESIZE
 
 #if XFS_WANT_FUNCS || (XFS_WANT_SPACE && XFSSO_XFS_ITOV)
 struct vnode *xfs_itov(xfs_inode_t *ip);
@@ -483,6 +481,7 @@ void		xfs_iunlock_map_shared(xfs_inode_t *, uint);
 void		xfs_ifunlock(xfs_inode_t *);
 void		xfs_ireclaim(xfs_inode_t *);
 int		xfs_finish_reclaim(xfs_inode_t *, int, int);
+int		xfs_finish_reclaim_all(struct xfs_mount *);
 
 /*
  * xfs_inode.c prototypes.

fs/xfs/xfs_log_recover.c

@@ -2952,7 +2952,6 @@ xlog_recover_process_iunlinks(xlog_t	*log)
 
 				if (!error) {
 					ASSERT(ip->i_d.di_nlink == 0);
-					ASSERT(ip->i_d.di_mode != 0);
 
 					/* setup for the next pass */
 					agino = INT_GET(dip->di_next_unlinked,
@@ -2970,7 +2969,9 @@ xlog_recover_process_iunlinks(xlog_t	*log)
 					ip->i_d.di_dmevmask = 0;
 
 					/*
-					 * Drop our reference to the
+					 * If this is a new inode, handle
+					 * it specially.  Otherwise,
+					 * just drop our reference to the
 					 * inode.  If there are no
 					 * other references, this will
 					 * send the inode to
@@ -2978,6 +2979,9 @@ xlog_recover_process_iunlinks(xlog_t	*log)
 					 * truncate the file and free
 					 * the inode.
 					 */
+					if (ip->i_d.di_mode == 0)
+						xfs_iput_new(ip, 0);
+					else
 						VN_RELE(XFS_ITOV(ip));
 				} else {
 					/*

fs/xfs/xfs_rename.c

@@ -72,14 +72,13 @@ int xfs_rename_skip, xfs_rename_nskip;
  * We are renaming dp1/name1 to dp2/name2.
  *
  * Return ENOENT if dp1 does not exist, other lookup errors, or 0 for success.
- * Return EAGAIN if the caller needs to try again.
  */
 STATIC int
 xfs_lock_for_rename(
 	xfs_inode_t	*dp1,	/* old (source) directory inode */
 	xfs_inode_t	*dp2,	/* new (target) directory inode */
-	vname_t		*dentry1, /* old entry name */
-	vname_t		*dentry2, /* new entry name */
+	vname_t		*vname1,/* old entry name */
+	vname_t		*vname2,/* new entry name */
 	xfs_inode_t	**ipp1, /* inode of old entry */
 	xfs_inode_t	**ipp2, /* inode of new entry, if it
 				   already exists, NULL otherwise. */
@@ -102,7 +101,7 @@ xfs_lock_for_rename(
 	 * to see if we still have the right inodes, directories, etc.
 	 */
 	lock_mode = xfs_ilock_map_shared(dp1);
-	error = xfs_get_dir_entry(dentry1, &ip1);
+	error = xfs_get_dir_entry(vname1, &ip1);
 	if (error) {
 		xfs_iunlock_map_shared(dp1, lock_mode);
 		return error;
@@ -123,7 +122,7 @@ xfs_lock_for_rename(
 	}
 
 	error = xfs_dir_lookup_int(XFS_ITOBHV(dp2), lock_mode,
-				   dentry2, &inum2, &ip2);
+				   vname2, &inum2, &ip2);
 	if (error == ENOENT) {		/* target does not need to exist. */
 		inum2 = 0;
 	} else if (error) {
@@ -213,20 +212,15 @@ xfs_lock_for_rename(
 
 int rename_which_error_return = 0;
 
-#ifdef DEBUG
-int xfs_rename_agains;
-int xfs_renames;
-#endif
-
 /*
  * xfs_rename
  */
 int
 xfs_rename(
 	bhv_desc_t	*src_dir_bdp,
-	vname_t		*src_dentry,
+	vname_t		*src_vname,
 	vnode_t		*target_dir_vp,
-	vname_t		*target_dentry,
+	vname_t		*target_vname,
 	cred_t		*credp)
 {
 	xfs_trans_t	*tp;
@@ -246,15 +240,11 @@ xfs_rename(
 	int		spaceres;
 	int		target_link_zero = 0;
 	int		num_inodes;
-	char		*src_name = VNAME(src_dentry);
-	char		*target_name = VNAME(target_dentry);
-	int		src_namelen;
-	int		target_namelen;
-#ifdef DEBUG
-	int		retries;
+	char		*src_name = VNAME(src_vname);
+	char		*target_name = VNAME(target_vname);
+	int		src_namelen = VNAMELEN(src_vname);
+	int		target_namelen = VNAMELEN(target_vname);
 
-	xfs_renames++;
-#endif
 	src_dir_vp = BHV_TO_VNODE(src_dir_bdp);
 	vn_trace_entry(src_dir_vp, "xfs_rename", (inst_t *)__return_address);
 	vn_trace_entry(target_dir_vp, "xfs_rename", (inst_t *)__return_address);
@@ -268,14 +258,10 @@ xfs_rename(
 	if (target_dir_bdp == NULL) {
 		return XFS_ERROR(EXDEV);
 	}
-	src_namelen = VNAMELEN(src_dentry);
-	if (src_namelen >= MAXNAMELEN)
-		return XFS_ERROR(ENAMETOOLONG);
-	target_namelen = VNAMELEN(target_dentry);
-	if (target_namelen >= MAXNAMELEN)
-		return XFS_ERROR(ENAMETOOLONG);
+
 	src_dp = XFS_BHVTOI(src_dir_bdp);
 	target_dp = XFS_BHVTOI(target_dir_bdp);
+
 	if (DM_EVENT_ENABLED(src_dir_vp->v_vfsp, src_dp, DM_EVENT_RENAME) ||
 	    DM_EVENT_ENABLED(target_dir_vp->v_vfsp,
 				target_dp, DM_EVENT_RENAME)) {
@@ -290,9 +276,6 @@ xfs_rename(
 	}
 	/* Return through std_return after this point. */
 
-#ifdef DEBUG
-	retries = 0;
-#endif
 	/*
 	 * Lock all the participating inodes. Depending upon whether
 	 * the target_name exists in the target directory, and
@@ -302,15 +285,9 @@ xfs_rename(
 	 * does not exist in the source directory.
 	 */
 	tp = NULL;
-	do {
-		error = xfs_lock_for_rename(src_dp, target_dp, src_dentry,
-				target_dentry, &src_ip, &target_ip, inodes,
+	error = xfs_lock_for_rename(src_dp, target_dp, src_vname,
+			target_vname, &src_ip, &target_ip, inodes,
 			&num_inodes);
-#ifdef DEBUG
-		if (error == EAGAIN)
-			xfs_rename_agains++;
-#endif
-	} while (error == EAGAIN);
 
 	if (error) {
 		rename_which_error_return = __LINE__;

fs/xfs/xfs_utils.c

@@ -47,9 +47,7 @@ xfs_get_dir_entry(
 	vnode_t		*vp;
 	bhv_desc_t	*bdp;
 
-	ASSERT(dentry->d_inode);
-
-	vp = LINVFS_GET_VP(dentry->d_inode);
+	vp = VNAME_TO_VNODE(dentry);
 	bdp = vn_bhv_lookup_unlocked(VN_BHV_HEAD(vp), &xfs_vnodeops);
 	if (!bdp) {
 		*ipp = NULL;

fs/xfs/xfs_vfsops.c

@@ -926,7 +926,7 @@ xfs_syncsub(
 		IPOINTER_CLR; \
 	}
 
-#define PREEMPT_MASK	0x7f
+#define XFS_PREEMPT_MASK	0x7f
 
 	if (bypassed)
 		*bypassed = 0;
@@ -1395,7 +1395,7 @@ xfs_syncsub(
 		 * if we have looped many times without dropping the
 		 * lock.
 		 */
-		if ((++preempt & PREEMPT_MASK) == 0) {
+		if ((++preempt & XFS_PREEMPT_MASK) == 0) {
 			if (mount_locked) {
 				IPOINTER_INSERT(ip, mp);
 			}

fs/xfs/xfs_vnodeops.c

@@ -82,7 +82,7 @@ xfs_open(
 /*
  * xfs_getattr
  */
-int
+STATIC int
 xfs_getattr(
 	bhv_desc_t	*bdp,
 	vattr_t		*vap,
@@ -1935,8 +1935,6 @@ xfs_create(
 
 	dm_di_mode = vap->va_mode|VTTOIF(vap->va_type);
 	namelen = VNAMELEN(dentry);
-	if (namelen >= MAXNAMELEN)
-		return XFS_ERROR(ENAMETOOLONG);
 
 	if (DM_EVENT_ENABLED(dir_vp->v_vfsp, dp, DM_EVENT_CREATE)) {
 		error = dm_send_namesp_event(DM_EVENT_CREATE,
@@ -2433,8 +2431,7 @@ xfs_remove(
 		return XFS_ERROR(EIO);
 
 	namelen = VNAMELEN(dentry);
-	if (namelen >= MAXNAMELEN)
-		return XFS_ERROR(ENAMETOOLONG);
+
 	if (DM_EVENT_ENABLED(dir_vp->v_vfsp, dp, DM_EVENT_REMOVE)) {
 		error = dm_send_namesp_event(DM_EVENT_REMOVE, dir_bdp, DM_RIGHT_NULL,
 					NULL, DM_RIGHT_NULL,
@@ -2668,8 +2665,6 @@ xfs_link(
 	vn_trace_entry(src_vp, __FUNCTION__, (inst_t *)__return_address);
 
 	target_namelen = VNAMELEN(dentry);
-	if (target_namelen >= MAXNAMELEN)
-		return XFS_ERROR(ENAMETOOLONG);
 	if (src_vp->v_type == VDIR)
 		return XFS_ERROR(EPERM);
 
@@ -2857,8 +2852,6 @@ xfs_mkdir(
 		return XFS_ERROR(EIO);
 
 	dir_namelen = VNAMELEN(dentry);
-	if (dir_namelen >= MAXNAMELEN)
-		return XFS_ERROR(ENAMETOOLONG);
 
 	tp = NULL;
 	dp_joined_to_trans = B_FALSE;
@@ -3100,8 +3093,6 @@ xfs_rmdir(
 	if (XFS_FORCED_SHUTDOWN(XFS_BHVTOI(dir_bdp)->i_mount))
 		return XFS_ERROR(EIO);
 	namelen = VNAMELEN(dentry);
-	if (namelen >= MAXNAMELEN)
-		return XFS_ERROR(ENAMETOOLONG);
 
 	if (DM_EVENT_ENABLED(dir_vp->v_vfsp, dp, DM_EVENT_REMOVE)) {
 		error = dm_send_namesp_event(DM_EVENT_REMOVE,
@@ -3415,8 +3406,7 @@ xfs_symlink(
 		return XFS_ERROR(EIO);
 
 	link_namelen = VNAMELEN(dentry);
-	if (link_namelen >= MAXNAMELEN)
-		return XFS_ERROR(ENAMETOOLONG);
+
 	/*
 	 * Check component lengths of the target path name.
 	 */
@@ -4070,6 +4060,49 @@ xfs_finish_reclaim(
 	return 0;
 }
 
+int
+xfs_finish_reclaim_all(xfs_mount_t *mp)
+{
+	int		purged;
+	xfs_inode_t	*ip;
+	vnode_t		*vp;
+	int		done = 0;
+
+	while (!done) {
+		purged = 0;
+		XFS_MOUNT_ILOCK(mp);
+		ip = mp->m_inodes;
+		if (ip == NULL) {
+			break;
+		}
+		do {
+			/* Make sure we skip markers inserted by sync */
+			if (ip->i_mount == NULL) {
+				ip = ip->i_mnext;
+				continue;
+			}
+	
+			/*
+			 * It's up to our caller to purge the root
+			 * and quota vnodes later.
+			 */
+			vp = XFS_ITOV_NULL(ip);
+	
+			if (!vp) {
+				XFS_MOUNT_IUNLOCK(mp);	
+				xfs_finish_reclaim(ip, 0, XFS_IFLUSH_ASYNC);
+				purged = 1;
+				break;
+			}
+		} while (ip != mp->m_inodes);
+	
+		done = !purged;
+	}
+
+	XFS_MOUNT_IUNLOCK(mp);
+	return 0;
+}
+
 /*
  * xfs_alloc_file_space()
  *	This routine allocates disk space for the given file.

include/linux/miscdevice.h

@@ -18,6 +18,7 @@
 #define RTC_MINOR 135
 #define EFI_RTC_MINOR		136	/* EFI Time services */
 #define SUN_OPENPROM_MINOR 139
+#define DMAPI_MINOR		140	/* DMAPI */
 #define NVRAM_MINOR 144
 #define I2O_MINOR 166
 #define MICROCODE_MINOR		184

include/linux/xattr.h

@@ -4,7 +4,7 @@
   Extended attributes handling.
 
   Copyright (C) 2001 by Andreas Gruenbacher <a.gruenbacher@computer.org>
-  Copyright (C) 2001 SGI - Silicon Graphics, Inc <linux-xfs@oss.sgi.com>
+  Copyright (c) 2001-2002 Silicon Graphics, Inc.  All Rights Reserved.
 */
 #ifndef _LINUX_XATTR_H
 #define _LINUX_XATTR_H