Merge git://oss.sgi.com:8090/xfs/xfs-2.6

* git://oss.sgi.com:8090/xfs/xfs-2.6: (49 commits)
  [XFS] Remove v1 dir trace macro - missed in a past commit.
  [XFS] 955947: Infinite loop in xfs_bulkstat() on formatter() error
  [XFS] pv 956241, author: nathans, rv: vapo - make ino validation checks
  [XFS] pv 956240, author: nathans, rv: vapo - Minor fixes in
  [XFS] Really fix use after free in xfs_iunpin.
  [XFS] Collapse sv_init and init_sv into just the one interface.
  [XFS] standardize on one sema init macro
  [XFS] Reduce endian flipping in alloc_btree, same as was done for
  [XFS] Minor cleanup from dio locking fix, remove an extra conditional.
  [XFS] Fix kmem_zalloc_greedy warnings on 64 bit platforms.
  [XFS] pv 955157, rv bnaujok - break the loop on EFAULT formatter() error
  [XFS] pv 955157, rv bnaujok - break the loop on formatter() error
  [XFS] Fixes the leak in reservation space because we weren't ungranting
  [XFS] Add lock annotations to xfs_trans_update_ail and
  [XFS] Fix a porting botch on the realtime subvol growfs code path.
  [XFS] Minor code rearranging and cleanup to prevent some coverity false
  [XFS] Remove a no-longer-correct debug assert from dio completion
  [XFS] Add a greedy allocation interface, allocating within a min/max size
  [XFS] Improve error handling for the zero-fsblock extent detection code.
  [XFS] Be more defensive with page flags (error/private) for metadata
  ...

fs/xfs/Makefile-linux-2.6

@@ -30,7 +30,6 @@ ifeq ($(CONFIG_XFS_TRACE),y)
 	EXTRA_CFLAGS += -DXFS_BLI_TRACE
 	EXTRA_CFLAGS += -DXFS_BMAP_TRACE
 	EXTRA_CFLAGS += -DXFS_BMBT_TRACE
-	EXTRA_CFLAGS += -DXFS_DIR_TRACE
 	EXTRA_CFLAGS += -DXFS_DIR2_TRACE
 	EXTRA_CFLAGS += -DXFS_DQUOT_TRACE
 	EXTRA_CFLAGS += -DXFS_ILOCK_TRACE

fs/xfs/linux-2.6/kmem.c

@@ -34,6 +34,14 @@ kmem_alloc(size_t size, unsigned int __nocast flags)
 	gfp_t	lflags = kmem_flags_convert(flags);
 	void	*ptr;
 
+#ifdef DEBUG
+	if (unlikely(!(flags & KM_LARGE) && (size > PAGE_SIZE))) {
+		printk(KERN_WARNING "Large %s attempt, size=%ld\n",
+			__FUNCTION__, (long)size);
+		dump_stack();
+	}
+#endif
+
 	do {
 		if (size < MAX_SLAB_SIZE || retries > MAX_VMALLOCS)
 			ptr = kmalloc(size, lflags);
@@ -60,6 +68,27 @@ kmem_zalloc(size_t size, unsigned int __nocast flags)
 	return ptr;
 }
 
+void *
+kmem_zalloc_greedy(size_t *size, size_t minsize, size_t maxsize,
+		   unsigned int __nocast flags)
+{
+	void		*ptr;
+	size_t		kmsize = maxsize;
+	unsigned int	kmflags = (flags & ~KM_SLEEP) | KM_NOSLEEP;
+
+	while (!(ptr = kmem_zalloc(kmsize, kmflags))) {
+		if ((kmsize <= minsize) && (flags & KM_NOSLEEP))
+			break;
+		if ((kmsize >>= 1) <= minsize) {
+			kmsize = minsize;
+			kmflags = flags;
+		}
+	}
+	if (ptr)
+		*size = kmsize;
+	return ptr;
+}
+
 void
 kmem_free(void *ptr, size_t size)
 {

fs/xfs/linux-2.6/kmem.h

@@ -30,6 +30,7 @@
 #define KM_NOSLEEP	0x0002u
 #define KM_NOFS		0x0004u
 #define KM_MAYFAIL	0x0008u
+#define KM_LARGE	0x0010u
 
 /*
  * We use a special process flag to avoid recursive callbacks into
@@ -41,7 +42,7 @@ kmem_flags_convert(unsigned int __nocast flags)
 {
 	gfp_t	lflags;
 
-	BUG_ON(flags & ~(KM_SLEEP|KM_NOSLEEP|KM_NOFS|KM_MAYFAIL));
+	BUG_ON(flags & ~(KM_SLEEP|KM_NOSLEEP|KM_NOFS|KM_MAYFAIL|KM_LARGE));
 
 	if (flags & KM_NOSLEEP) {
 		lflags = GFP_ATOMIC | __GFP_NOWARN;
@@ -54,8 +55,9 @@ kmem_flags_convert(unsigned int __nocast flags)
 }
 
 extern void *kmem_alloc(size_t, unsigned int __nocast);
-extern void *kmem_realloc(void *, size_t, size_t, unsigned int __nocast);
 extern void *kmem_zalloc(size_t, unsigned int __nocast);
+extern void *kmem_zalloc_greedy(size_t *, size_t, size_t, unsigned int __nocast);
+extern void *kmem_realloc(void *, size_t, size_t, unsigned int __nocast);
 extern void  kmem_free(void *, size_t);
 
 /*

fs/xfs/linux-2.6/sema.h

@@ -29,8 +29,6 @@
 
 typedef struct semaphore sema_t;
 
-#define init_sema(sp, val, c, d)	sema_init(sp, val)
-#define initsema(sp, val)		sema_init(sp, val)
 #define initnsema(sp, val, name)	sema_init(sp, val)
 #define psema(sp, b)			down(sp)
 #define vsema(sp)			up(sp)

fs/xfs/linux-2.6/sv.h

@@ -53,8 +53,6 @@ static inline void _sv_wait(sv_t *sv, spinlock_t *lock, int state,
 	remove_wait_queue(&sv->waiters, &wait);
 }
 
-#define init_sv(sv,type,name,flag) \
-	init_waitqueue_head(&(sv)->waiters)
 #define sv_init(sv,flag,name) \
 	init_waitqueue_head(&(sv)->waiters)
 #define sv_destroy(sv) \

fs/xfs/linux-2.6/xfs_aops.c

@@ -71,7 +71,7 @@ xfs_page_trace(
 	int		tag,
 	struct inode	*inode,
 	struct page	*page,
-	int		mask)
+	unsigned long	pgoff)
 {
 	xfs_inode_t	*ip;
 	bhv_vnode_t	*vp = vn_from_inode(inode);
@@ -91,7 +91,7 @@ xfs_page_trace(
 		(void *)ip,
 		(void *)inode,
 		(void *)page,
-		(void *)((unsigned long)mask),
+		(void *)pgoff,
 		(void *)((unsigned long)((ip->i_d.di_size >> 32) & 0xffffffff)),
 		(void *)((unsigned long)(ip->i_d.di_size & 0xffffffff)),
 		(void *)((unsigned long)((isize >> 32) & 0xffffffff)),
@@ -105,7 +105,7 @@ xfs_page_trace(
 		(void *)NULL);
 }
 #else
-#define xfs_page_trace(tag, inode, page, mask)
+#define xfs_page_trace(tag, inode, page, pgoff)
 #endif
 
 /*
@@ -1197,7 +1197,7 @@ xfs_vm_releasepage(
 		.nr_to_write = 1,
 	};
 
-	xfs_page_trace(XFS_RELEASEPAGE_ENTER, inode, page, gfp_mask);
+	xfs_page_trace(XFS_RELEASEPAGE_ENTER, inode, page, 0);
 
 	if (!page_has_buffers(page))
 		return 0;
@@ -1356,7 +1356,6 @@ xfs_end_io_direct(
 		ioend->io_size = size;
 		xfs_finish_ioend(ioend);
 	} else {
-		ASSERT(size >= 0);
 		xfs_destroy_ioend(ioend);
 	}
 

fs/xfs/linux-2.6/xfs_buf.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * Copyright (c) 2000-2006 Silicon Graphics, Inc.
  * All Rights Reserved.
  *
  * This program is free software; you can redistribute it and/or
@@ -318,8 +318,12 @@ xfs_buf_free(
 		if ((bp->b_flags & XBF_MAPPED) && (bp->b_page_count > 1))
 			free_address(bp->b_addr - bp->b_offset);
 
-		for (i = 0; i < bp->b_page_count; i++)
-			page_cache_release(bp->b_pages[i]);
+		for (i = 0; i < bp->b_page_count; i++) {
+			struct page	*page = bp->b_pages[i];
+
+			ASSERT(!PagePrivate(page));
+			page_cache_release(page);
+		}
 		_xfs_buf_free_pages(bp);
 	} else if (bp->b_flags & _XBF_KMEM_ALLOC) {
 		 /*
@@ -400,6 +404,7 @@ _xfs_buf_lookup_pages(
 		nbytes = min_t(size_t, size, PAGE_CACHE_SIZE - offset);
 		size -= nbytes;
 
+		ASSERT(!PagePrivate(page));
 		if (!PageUptodate(page)) {
 			page_count--;
 			if (blocksize >= PAGE_CACHE_SIZE) {
@@ -768,7 +773,7 @@ xfs_buf_get_noaddr(
 	_xfs_buf_initialize(bp, target, 0, len, 0);
 
  try_again:
-	data = kmem_alloc(malloc_len, KM_SLEEP | KM_MAYFAIL);
+	data = kmem_alloc(malloc_len, KM_SLEEP | KM_MAYFAIL | KM_LARGE);
 	if (unlikely(data == NULL))
 		goto fail_free_buf;
 
@@ -1117,10 +1122,10 @@ xfs_buf_bio_end_io(
 	do {
 		struct page	*page = bvec->bv_page;
 
+		ASSERT(!PagePrivate(page));
 		if (unlikely(bp->b_error)) {
 			if (bp->b_flags & XBF_READ)
 				ClearPageUptodate(page);
-			SetPageError(page);
 		} else if (blocksize >= PAGE_CACHE_SIZE) {
 			SetPageUptodate(page);
 		} else if (!PagePrivate(page) &&
@@ -1156,16 +1161,16 @@ _xfs_buf_ioapply(
 	total_nr_pages = bp->b_page_count;
 	map_i = 0;
 
-	if (bp->b_flags & _XBF_RUN_QUEUES) {
-		bp->b_flags &= ~_XBF_RUN_QUEUES;
-		rw = (bp->b_flags & XBF_READ) ? READ_SYNC : WRITE_SYNC;
-	} else {
-		rw = (bp->b_flags & XBF_READ) ? READ : WRITE;
-	}
-
 	if (bp->b_flags & XBF_ORDERED) {
 		ASSERT(!(bp->b_flags & XBF_READ));
 		rw = WRITE_BARRIER;
+	} else if (bp->b_flags & _XBF_RUN_QUEUES) {
+		ASSERT(!(bp->b_flags & XBF_READ_AHEAD));
+		bp->b_flags &= ~_XBF_RUN_QUEUES;
+		rw = (bp->b_flags & XBF_WRITE) ? WRITE_SYNC : READ_SYNC;
+	} else {
+		rw = (bp->b_flags & XBF_WRITE) ? WRITE :
+		     (bp->b_flags & XBF_READ_AHEAD) ? READA : READ;
 	}
 
 	/* Special code path for reading a sub page size buffer in --
@@ -1681,6 +1686,7 @@ xfsbufd(
 	xfs_buf_t		*bp, *n;
 	struct list_head	*dwq = &target->bt_delwrite_queue;
 	spinlock_t		*dwlk = &target->bt_delwrite_lock;
+	int			count;
 
 	current->flags |= PF_MEMALLOC;
 
@@ -1696,6 +1702,7 @@ xfsbufd(
 		schedule_timeout_interruptible(
 			xfs_buf_timer_centisecs * msecs_to_jiffies(10));
 
+		count = 0;
 		age = xfs_buf_age_centisecs * msecs_to_jiffies(10);
 		spin_lock(dwlk);
 		list_for_each_entry_safe(bp, n, dwq, b_list) {
@@ -1711,9 +1718,11 @@ xfsbufd(
 					break;
 				}
 
-				bp->b_flags &= ~(XBF_DELWRI|_XBF_DELWRI_Q);
+				bp->b_flags &= ~(XBF_DELWRI|_XBF_DELWRI_Q|
+						 _XBF_RUN_QUEUES);
 				bp->b_flags |= XBF_WRITE;
-				list_move(&bp->b_list, &tmp);
+				list_move_tail(&bp->b_list, &tmp);
+				count++;
 			}
 		}
 		spin_unlock(dwlk);
@@ -1724,12 +1733,12 @@ xfsbufd(
 
 			list_del_init(&bp->b_list);
 			xfs_buf_iostrategy(bp);
-
-			blk_run_address_space(target->bt_mapping);
 		}
 
 		if (as_list_len > 0)
 			purge_addresses();
+		if (count)
+			blk_run_address_space(target->bt_mapping);
 
 		clear_bit(XBT_FORCE_FLUSH, &target->bt_flags);
 	} while (!kthread_should_stop());
@@ -1767,7 +1776,7 @@ xfs_flush_buftarg(
 			continue;
 		}
 
-		list_move(&bp->b_list, &tmp);
+		list_move_tail(&bp->b_list, &tmp);
 	}
 	spin_unlock(dwlk);
 
@@ -1776,7 +1785,7 @@ xfs_flush_buftarg(
 	 */
 	list_for_each_entry_safe(bp, n, &tmp, b_list) {
 		xfs_buf_lock(bp);
-		bp->b_flags &= ~(XBF_DELWRI|_XBF_DELWRI_Q);
+		bp->b_flags &= ~(XBF_DELWRI|_XBF_DELWRI_Q|_XBF_RUN_QUEUES);
 		bp->b_flags |= XBF_WRITE;
 		if (wait)
 			bp->b_flags &= ~XBF_ASYNC;
@@ -1786,6 +1795,9 @@ xfs_flush_buftarg(
 		xfs_buf_iostrategy(bp);
 	}
 
+	if (wait)
+		blk_run_address_space(target->bt_mapping);
+
 	/*
 	 * Remaining list items must be flushed before returning
 	 */
@@ -1797,9 +1809,6 @@ xfs_flush_buftarg(
 		xfs_buf_relse(bp);
 	}
 
-	if (wait)
-		blk_run_address_space(target->bt_mapping);
-
 	return pincount;
 }
 

fs/xfs/linux-2.6/xfs_buf.h

@@ -298,11 +298,6 @@ extern void xfs_buf_trace(xfs_buf_t *, char *, void *, void *);
 #define XFS_BUF_UNWRITE(bp)	((bp)->b_flags &= ~XBF_WRITE)
 #define XFS_BUF_ISWRITE(bp)	((bp)->b_flags & XBF_WRITE)
 
-#define XFS_BUF_ISUNINITIAL(bp)	(0)
-#define XFS_BUF_UNUNINITIAL(bp)	(0)
-
-#define XFS_BUF_BP_ISMAPPED(bp)	(1)
-
 #define XFS_BUF_IODONE_FUNC(bp)			((bp)->b_iodone)
 #define XFS_BUF_SET_IODONE_FUNC(bp, func)	((bp)->b_iodone = (func))
 #define XFS_BUF_CLR_IODONE_FUNC(bp)		((bp)->b_iodone = NULL)
@@ -393,8 +388,6 @@ static inline int XFS_bwrite(xfs_buf_t *bp)
 	return error;
 }
 
-#define XFS_bdwrite(bp)		xfs_buf_iostart(bp, XBF_DELWRI | XBF_ASYNC)
-
 static inline int xfs_bdwrite(void *mp, xfs_buf_t *bp)
 {
 	bp->b_strat = xfs_bdstrat_cb;

fs/xfs/linux-2.6/xfs_globals.c

@@ -34,7 +34,7 @@ xfs_param_t xfs_params = {
 	.restrict_chown	= {	0,		1,		1	},
 	.sgid_inherit	= {	0,		0,		1	},
 	.symlink_mode	= {	0,		0,		1	},
-	.panic_mask	= {	0,		0,		127	},
+	.panic_mask	= {	0,		0,		255	},
 	.error_level	= {	0,		3,		11	},
 	.syncd_timer	= {	1*100,		30*100,		7200*100},
 	.stats_clear	= {	0,		0,		1	},

fs/xfs/linux-2.6/xfs_ioctl.c

@@ -653,7 +653,7 @@ xfs_attrmulti_by_handle(
 STATIC int
 xfs_ioc_space(
 	bhv_desc_t		*bdp,
-	bhv_vnode_t		*vp,
+	struct inode		*inode,
 	struct file		*filp,
 	int			flags,
 	unsigned int		cmd,
@@ -735,7 +735,7 @@ xfs_ioctl(
 		    !capable(CAP_SYS_ADMIN))
 			return -EPERM;
 
-		return xfs_ioc_space(bdp, vp, filp, ioflags, cmd, arg);
+		return xfs_ioc_space(bdp, inode, filp, ioflags, cmd, arg);
 
 	case XFS_IOC_DIOINFO: {
 		struct dioattr	da;
@@ -763,6 +763,8 @@ xfs_ioctl(
 		return xfs_ioc_fsgeometry(mp, arg);
 
 	case XFS_IOC_GETVERSION:
+		return put_user(inode->i_generation, (int __user *)arg);
+
 	case XFS_IOC_GETXFLAGS:
 	case XFS_IOC_SETXFLAGS:
 	case XFS_IOC_FSGETXATTR:
@@ -957,7 +959,7 @@ xfs_ioctl(
 STATIC int
 xfs_ioc_space(
 	bhv_desc_t		*bdp,
-	bhv_vnode_t		*vp,
+	struct inode		*inode,
 	struct file		*filp,
 	int			ioflags,
 	unsigned int		cmd,
@@ -967,13 +969,13 @@ xfs_ioc_space(
 	int			attr_flags = 0;
 	int			error;
 
-	if (vp->v_inode.i_flags & (S_IMMUTABLE|S_APPEND))
+	if (inode->i_flags & (S_IMMUTABLE|S_APPEND))
 		return -XFS_ERROR(EPERM);
 
 	if (!(filp->f_mode & FMODE_WRITE))
 		return -XFS_ERROR(EBADF);
 
-	if (!VN_ISREG(vp))
+	if (!S_ISREG(inode->i_mode))
 		return -XFS_ERROR(EINVAL);
 
 	if (copy_from_user(&bf, arg, sizeof(bf)))
@@ -1264,13 +1266,6 @@ xfs_ioc_xattr(
 		break;
 	}
 
-	case XFS_IOC_GETVERSION: {
-		flags = vn_to_inode(vp)->i_generation;
-		if (copy_to_user(arg, &flags, sizeof(flags)))
-			error = -EFAULT;
-		break;
-	}
-
 	default:
 		error = -ENOTTY;
 		break;

fs/xfs/linux-2.6/xfs_iops.c

@@ -623,12 +623,27 @@ xfs_vn_getattr(
 {
 	struct inode	*inode = dentry->d_inode;
 	bhv_vnode_t	*vp = vn_from_inode(inode);
-	int		error = 0;
+	bhv_vattr_t	vattr = { .va_mask = XFS_AT_STAT };
+	int		error;
 
-	if (unlikely(vp->v_flag & VMODIFIED))
-		error = vn_revalidate(vp);
-	if (!error)
-		generic_fillattr(inode, stat);
+	error = bhv_vop_getattr(vp, &vattr, ATTR_LAZY, NULL);
+	if (likely(!error)) {
+		stat->size = i_size_read(inode);
+		stat->dev = inode->i_sb->s_dev;
+		stat->rdev = (vattr.va_rdev == 0) ? 0 :
+				MKDEV(sysv_major(vattr.va_rdev) & 0x1ff,
+				      sysv_minor(vattr.va_rdev));
+		stat->mode = vattr.va_mode;
+		stat->nlink = vattr.va_nlink;
+		stat->uid = vattr.va_uid;
+		stat->gid = vattr.va_gid;
+		stat->ino = vattr.va_nodeid;
+		stat->atime = vattr.va_atime;
+		stat->mtime = vattr.va_mtime;
+		stat->ctime = vattr.va_ctime;
+		stat->blocks = vattr.va_nblocks;
+		stat->blksize = vattr.va_blocksize;
+	}
 	return -error;
 }
 

fs/xfs/linux-2.6/xfs_linux.h

@@ -148,11 +148,7 @@ BUFFER_FNS(PrivateStart, unwritten);
 		(current->flags = ((current->flags & ~(f)) | (*(sp) & (f))))
 
 #define NBPP		PAGE_SIZE
-#define DPPSHFT		(PAGE_SHIFT - 9)
 #define NDPP		(1 << (PAGE_SHIFT - 9))
-#define dtop(DD)	(((DD) + NDPP - 1) >> DPPSHFT)
-#define dtopt(DD)	((DD) >> DPPSHFT)
-#define dpoff(DD)	((DD) & (NDPP-1))
 
 #define NBBY		8		/* number of bits per byte */
 #define	NBPC		PAGE_SIZE	/* Number of bytes per click */
@@ -172,8 +168,6 @@ BUFFER_FNS(PrivateStart, unwritten);
 #define	btoct(x)	((__psunsigned_t)(x)>>BPCSHIFT)
 #define	btoc64(x)	(((__uint64_t)(x)+(NBPC-1))>>BPCSHIFT)
 #define	btoct64(x)	((__uint64_t)(x)>>BPCSHIFT)
-#define	io_btoc(x)	(((__psunsigned_t)(x)+(IO_NBPC-1))>>IO_BPCSHIFT)
-#define	io_btoct(x)	((__psunsigned_t)(x)>>IO_BPCSHIFT)
 
 /* off_t bytes to clicks */
 #define offtoc(x)       (((__uint64_t)(x)+(NBPC-1))>>BPCSHIFT)
@@ -186,7 +180,6 @@ BUFFER_FNS(PrivateStart, unwritten);
 #define	ctob(x)		((__psunsigned_t)(x)<<BPCSHIFT)
 #define btoct(x)        ((__psunsigned_t)(x)>>BPCSHIFT)
 #define	ctob64(x)	((__uint64_t)(x)<<BPCSHIFT)
-#define	io_ctob(x)	((__psunsigned_t)(x)<<IO_BPCSHIFT)
 
 /* bytes to clicks */
 #define btoc(x)         (((__psunsigned_t)(x)+(NBPC-1))>>BPCSHIFT)
@@ -339,4 +332,11 @@ static inline __uint64_t roundup_64(__uint64_t x, __uint32_t y)
 	return(x * y);
 }
 
+static inline __uint64_t howmany_64(__uint64_t x, __uint32_t y)
+{
+	x += y - 1;
+	do_div(x, y);
+	return x;
+}
+
 #endif /* __XFS_LINUX__ */

fs/xfs/linux-2.6/xfs_lrw.c

@@ -270,12 +270,12 @@ xfs_read(
 		}
 	}
 
-	if (unlikely((ioflags & IO_ISDIRECT) && VN_CACHED(vp)))
-		bhv_vop_flushinval_pages(vp, ctooff(offtoct(*offset)),
-						-1, FI_REMAPF_LOCKED);
-
-	if (unlikely(ioflags & IO_ISDIRECT))
+	if (unlikely(ioflags & IO_ISDIRECT)) {
+		if (VN_CACHED(vp))
+			bhv_vop_flushinval_pages(vp, ctooff(offtoct(*offset)),
+						 -1, FI_REMAPF_LOCKED);
 		mutex_unlock(&inode->i_mutex);
+	}
 
 	xfs_rw_enter_trace(XFS_READ_ENTER, &ip->i_iocore,
 				(void *)iovp, segs, *offset, ioflags);

fs/xfs/linux-2.6/xfs_super.c

@@ -227,7 +227,9 @@ xfs_initialize_vnode(
 		xfs_revalidate_inode(XFS_BHVTOM(bdp), vp, ip);
 		xfs_set_inodeops(inode);
 
+		spin_lock(&ip->i_flags_lock);
 		ip->i_flags &= ~XFS_INEW;
+		spin_unlock(&ip->i_flags_lock);
 		barrier();
 
 		unlock_new_inode(inode);

fs/xfs/linux-2.6/xfs_vfs.h

@@ -79,7 +79,7 @@ typedef enum {
 #define VFS_RDONLY		0x0001	/* read-only vfs */
 #define VFS_GRPID		0x0002	/* group-ID assigned from directory */
 #define VFS_DMI			0x0004	/* filesystem has the DMI enabled */
-#define VFS_UMOUNT		0x0008	/* unmount in progress */
+/* ---- VFS_UMOUNT ----		0x0008	-- unneeded, fixed via kthread APIs */
 #define VFS_32BITINODES		0x0010	/* do not use inums above 32 bits */
 #define VFS_END			0x0010	/* max flag */
 

fs/xfs/linux-2.6/xfs_vnode.h

@@ -85,8 +85,6 @@ typedef enum {
 #define VN_BHV_HEAD(vp)			((bhv_head_t *)(&((vp)->v_bh)))
 #define vn_bhv_head_init(bhp,name)	bhv_head_init(bhp,name)
 #define vn_bhv_remove(bhp,bdp)		bhv_remove(bhp,bdp)
-#define vn_bhv_lookup(bhp,ops)		bhv_lookup(bhp,ops)
-#define vn_bhv_lookup_unlocked(bhp,ops) bhv_lookup_unlocked(bhp,ops)
 
 /*
  * Vnode to Linux inode mapping.

fs/xfs/quota/xfs_dquot_item.c

@@ -381,18 +381,6 @@ xfs_qm_dquot_logitem_unlock(
 }
 
 
-/*
- * The transaction with the dquot locked has aborted.  The dquot
- * must not be dirty within the transaction.  We simply unlock just
- * as if the transaction had been cancelled.
- */
-STATIC void
-xfs_qm_dquot_logitem_abort(
-	xfs_dq_logitem_t    *ql)
-{
-	xfs_qm_dquot_logitem_unlock(ql);
-}
-
 /*
  * this needs to stamp an lsn into the dquot, I think.
  * rpc's that look at user dquot's would then have to
@@ -426,7 +414,6 @@ STATIC struct xfs_item_ops xfs_dquot_item_ops = {
 	.iop_committed	= (xfs_lsn_t(*)(xfs_log_item_t*, xfs_lsn_t))
 					xfs_qm_dquot_logitem_committed,
 	.iop_push	= (void(*)(xfs_log_item_t*))xfs_qm_dquot_logitem_push,
-	.iop_abort	= (void(*)(xfs_log_item_t*))xfs_qm_dquot_logitem_abort,
 	.iop_pushbuf	= (void(*)(xfs_log_item_t*))
 					xfs_qm_dquot_logitem_pushbuf,
 	.iop_committing = (void(*)(xfs_log_item_t*, xfs_lsn_t))
@@ -558,17 +545,6 @@ xfs_qm_qoff_logitem_committed(xfs_qoff_logitem_t *qf, xfs_lsn_t lsn)
 	return (lsn);
 }
 
-/*
- * The transaction of which this QUOTAOFF is a part has been aborted.
- * Just clean up after ourselves.
- * Shouldn't this never happen in the case of qoffend logitems? XXX
- */
-STATIC void
-xfs_qm_qoff_logitem_abort(xfs_qoff_logitem_t *qf)
-{
-	kmem_free(qf, sizeof(xfs_qoff_logitem_t));
-}
-
 /*
  * There isn't much you can do to push on an quotaoff item.  It is simply
  * stuck waiting for the log to be flushed to disk.
@@ -644,7 +620,6 @@ STATIC struct xfs_item_ops xfs_qm_qoffend_logitem_ops = {
 	.iop_committed	= (xfs_lsn_t(*)(xfs_log_item_t*, xfs_lsn_t))
 					xfs_qm_qoffend_logitem_committed,
 	.iop_push	= (void(*)(xfs_log_item_t*))xfs_qm_qoff_logitem_push,
-	.iop_abort	= (void(*)(xfs_log_item_t*))xfs_qm_qoff_logitem_abort,
 	.iop_pushbuf	= NULL,
 	.iop_committing = (void(*)(xfs_log_item_t*, xfs_lsn_t))
 					xfs_qm_qoffend_logitem_committing
@@ -667,7 +642,6 @@ STATIC struct xfs_item_ops xfs_qm_qoff_logitem_ops = {
 	.iop_committed	= (xfs_lsn_t(*)(xfs_log_item_t*, xfs_lsn_t))
 					xfs_qm_qoff_logitem_committed,
 	.iop_push	= (void(*)(xfs_log_item_t*))xfs_qm_qoff_logitem_push,
-	.iop_abort	= (void(*)(xfs_log_item_t*))xfs_qm_qoff_logitem_abort,
 	.iop_pushbuf	= NULL,
 	.iop_committing = (void(*)(xfs_log_item_t*, xfs_lsn_t))
 					xfs_qm_qoff_logitem_committing

fs/xfs/quota/xfs_qm.c

@@ -112,17 +112,17 @@ xfs_Gqm_init(void)
 {
 	xfs_dqhash_t	*udqhash, *gdqhash;
 	xfs_qm_t	*xqm;
-	uint		i, hsize, flags = KM_SLEEP | KM_MAYFAIL;
+	size_t		hsize;
+	uint		i;
 
 	/*
 	 * Initialize the dquot hash tables.
 	 */
-	hsize = XFS_QM_HASHSIZE_HIGH;
-	while (!(udqhash = kmem_zalloc(hsize * sizeof(xfs_dqhash_t), flags))) {
-		if ((hsize >>= 1) <= XFS_QM_HASHSIZE_LOW)
-			flags = KM_SLEEP;
-	}
-	gdqhash = kmem_zalloc(hsize * sizeof(xfs_dqhash_t), KM_SLEEP);
+	udqhash = kmem_zalloc_greedy(&hsize,
+				     XFS_QM_HASHSIZE_LOW, XFS_QM_HASHSIZE_HIGH,
+				     KM_SLEEP | KM_MAYFAIL | KM_LARGE);
+	gdqhash = kmem_zalloc(hsize, KM_SLEEP | KM_LARGE);
+	hsize /= sizeof(xfs_dqhash_t);
 	ndquot = hsize << 8;
 
 	xqm = kmem_zalloc(sizeof(xfs_qm_t), KM_SLEEP);

fs/xfs/quota/xfs_qm.h

@@ -55,12 +55,6 @@ extern kmem_zone_t	*qm_dqtrxzone;
 #define XFS_QM_HASHSIZE_LOW		(NBPP / sizeof(xfs_dqhash_t))
 #define XFS_QM_HASHSIZE_HIGH		((NBPP * 4) / sizeof(xfs_dqhash_t))
 
-/*
- * We output a cmn_err when quotachecking a quota file with more than
- * this many fsbs.
- */
-#define XFS_QM_BIG_QCHECK_NBLKS		500
-
 /*
  * This defines the unit of allocation of dquots.
  * Currently, it is just one file system block, and a 4K blk contains 30

fs/xfs/quota/xfs_quota_priv.h

@@ -75,7 +75,6 @@ static inline int XQMISLCKD(struct xfs_dqhash *h)
 
 #define xfs_qm_freelist_lock(qm)	XQMLCK(&((qm)->qm_dqfreelist))
 #define xfs_qm_freelist_unlock(qm)	XQMUNLCK(&((qm)->qm_dqfreelist))
-#define XFS_QM_IS_FREELIST_LOCKED(qm)	XQMISLCKD(&((qm)->qm_dqfreelist))
 
 /*
  * Hash into a bucket in the dquot hash table, based on <mp, id>.
@@ -170,6 +169,5 @@ for ((dqp) = (qlist)->qh_next; (dqp) != (xfs_dquot_t *)(qlist); \
 #define DQFLAGTO_TYPESTR(d)	(((d)->dq_flags & XFS_DQ_USER) ? "USR" : \
 				 (((d)->dq_flags & XFS_DQ_GROUP) ? "GRP" : \
 				 (((d)->dq_flags & XFS_DQ_PROJ) ? "PRJ":"???")))
-#define DQFLAGTO_DIRTYSTR(d)	(XFS_DQ_IS_DIRTY(d) ? "DIRTY" : "NOTDIRTY")
 
 #endif	/* __XFS_QUOTA_PRIV_H__ */

fs/xfs/support/ktrace.c

@@ -75,7 +75,7 @@ ktrace_alloc(int nentries, unsigned int __nocast sleep)
 							    sleep);
 	} else {
 		ktep = (ktrace_entry_t*)kmem_zalloc((nentries * sizeof(*ktep)),
-							    sleep);
+							    sleep | KM_LARGE);
 	}
 
 	if (ktep == NULL) {

fs/xfs/xfs_ag.h

@@ -150,7 +150,7 @@ typedef struct xfs_agi {
 #define	XFS_BUF_TO_AGFL(bp)	((xfs_agfl_t *)XFS_BUF_PTR(bp))
 
 typedef struct xfs_agfl {
-	xfs_agblock_t	agfl_bno[1];	/* actually XFS_AGFL_SIZE(mp) */
+	__be32		agfl_bno[1];	/* actually XFS_AGFL_SIZE(mp) */
 } xfs_agfl_t;
 
 /*

fs/xfs/xfs_alloc.c

@@ -1477,8 +1477,10 @@ xfs_alloc_ag_vextent_small(
 	/*
 	 * Can't allocate from the freelist for some reason.
 	 */
-	else
+	else {
+		fbno = NULLAGBLOCK;
 		flen = 0;
+	}
 	/*
 	 * Can't do the allocation, give up.
 	 */
@@ -2021,7 +2023,7 @@ xfs_alloc_get_freelist(
 	/*
 	 * Get the block number and update the data structures.
 	 */
-	bno = INT_GET(agfl->agfl_bno[be32_to_cpu(agf->agf_flfirst)], ARCH_CONVERT);
+	bno = be32_to_cpu(agfl->agfl_bno[be32_to_cpu(agf->agf_flfirst)]);
 	be32_add(&agf->agf_flfirst, 1);
 	xfs_trans_brelse(tp, agflbp);
 	if (be32_to_cpu(agf->agf_flfirst) == XFS_AGFL_SIZE(mp))
@@ -2108,7 +2110,7 @@ xfs_alloc_put_freelist(
 {
 	xfs_agf_t		*agf;	/* a.g. freespace structure */
 	xfs_agfl_t		*agfl;	/* a.g. free block array */
-	xfs_agblock_t		*blockp;/* pointer to array entry */
+	__be32			*blockp;/* pointer to array entry */
 	int			error;
 #ifdef XFS_ALLOC_TRACE
 	static char		fname[] = "xfs_alloc_put_freelist";
@@ -2132,7 +2134,7 @@ xfs_alloc_put_freelist(
 	pag->pagf_flcount++;
 	ASSERT(be32_to_cpu(agf->agf_flcount) <= XFS_AGFL_SIZE(mp));
 	blockp = &agfl->agfl_bno[be32_to_cpu(agf->agf_fllast)];
-	INT_SET(*blockp, ARCH_CONVERT, bno);
+	*blockp = cpu_to_be32(bno);
 	TRACE_MODAGF(NULL, agf, XFS_AGF_FLLAST | XFS_AGF_FLCOUNT);
 	xfs_alloc_log_agf(tp, agbp, XFS_AGF_FLLAST | XFS_AGF_FLCOUNT);
 	xfs_trans_log_buf(tp, agflbp,

fs/xfs/xfs_alloc_btree.c

@@ -92,6 +92,7 @@ xfs_alloc_delrec(
 	xfs_alloc_key_t		*rkp;	/* right block key pointer */
 	xfs_alloc_ptr_t		*rpp;	/* right block address pointer */
 	int			rrecs=0;	/* number of records in right block */
+	int			numrecs;
 	xfs_alloc_rec_t		*rrp;	/* right block record pointer */
 	xfs_btree_cur_t		*tcur;	/* temporary btree cursor */
 
@@ -115,7 +116,8 @@ xfs_alloc_delrec(
 	/*
 	 * Fail if we're off the end of the block.
 	 */
-	if (ptr > be16_to_cpu(block->bb_numrecs)) {
+	numrecs = be16_to_cpu(block->bb_numrecs);
+	if (ptr > numrecs) {
 		*stat = 0;
 		return 0;
 	}
@@ -129,18 +131,18 @@ xfs_alloc_delrec(
 		lkp = XFS_ALLOC_KEY_ADDR(block, 1, cur);
 		lpp = XFS_ALLOC_PTR_ADDR(block, 1, cur);
 #ifdef DEBUG
-		for (i = ptr; i < be16_to_cpu(block->bb_numrecs); i++) {
+		for (i = ptr; i < numrecs; i++) {
 			if ((error = xfs_btree_check_sptr(cur, be32_to_cpu(lpp[i]), level)))
 				return error;
 		}
 #endif
-		if (ptr < be16_to_cpu(block->bb_numrecs)) {
+		if (ptr < numrecs) {
 			memmove(&lkp[ptr - 1], &lkp[ptr],
-				(be16_to_cpu(block->bb_numrecs) - ptr) * sizeof(*lkp));
+				(numrecs - ptr) * sizeof(*lkp));
 			memmove(&lpp[ptr - 1], &lpp[ptr],
-				(be16_to_cpu(block->bb_numrecs) - ptr) * sizeof(*lpp));
-			xfs_alloc_log_ptrs(cur, bp, ptr, be16_to_cpu(block->bb_numrecs) - 1);
-			xfs_alloc_log_keys(cur, bp, ptr, be16_to_cpu(block->bb_numrecs) - 1);
+				(numrecs - ptr) * sizeof(*lpp));
+			xfs_alloc_log_ptrs(cur, bp, ptr, numrecs - 1);
+			xfs_alloc_log_keys(cur, bp, ptr, numrecs - 1);
 		}
 	}
 	/*
@@ -149,10 +151,10 @@ xfs_alloc_delrec(
 	 */
 	else {
 		lrp = XFS_ALLOC_REC_ADDR(block, 1, cur);
-		if (ptr < be16_to_cpu(block->bb_numrecs)) {
+		if (ptr < numrecs) {
 			memmove(&lrp[ptr - 1], &lrp[ptr],
-				(be16_to_cpu(block->bb_numrecs) - ptr) * sizeof(*lrp));
-			xfs_alloc_log_recs(cur, bp, ptr, be16_to_cpu(block->bb_numrecs) - 1);
+				(numrecs - ptr) * sizeof(*lrp));
+			xfs_alloc_log_recs(cur, bp, ptr, numrecs - 1);
 		}
 		/*
 		 * If it's the first record in the block, we'll need a key
@@ -167,7 +169,8 @@ xfs_alloc_delrec(
 	/*
 	 * Decrement and log the number of entries in the block.
 	 */
-	be16_add(&block->bb_numrecs, -1);
+	numrecs--;
+	block->bb_numrecs = cpu_to_be16(numrecs);
 	xfs_alloc_log_block(cur->bc_tp, bp, XFS_BB_NUMRECS);
 	/*
 	 * See if the longest free extent in the allocation group was
@@ -181,14 +184,14 @@ xfs_alloc_delrec(
 	if (level == 0 &&
 	    cur->bc_btnum == XFS_BTNUM_CNT &&
 	    be32_to_cpu(block->bb_rightsib) == NULLAGBLOCK &&
-	    ptr > be16_to_cpu(block->bb_numrecs)) {
-		ASSERT(ptr == be16_to_cpu(block->bb_numrecs) + 1);
+	    ptr > numrecs) {
+		ASSERT(ptr == numrecs + 1);
 		/*
 		 * There are still records in the block.  Grab the size
 		 * from the last one.
 		 */
-		if (be16_to_cpu(block->bb_numrecs)) {
-			rrp = XFS_ALLOC_REC_ADDR(block, be16_to_cpu(block->bb_numrecs), cur);
+		if (numrecs) {
+			rrp = XFS_ALLOC_REC_ADDR(block, numrecs, cur);
 			agf->agf_longest = rrp->ar_blockcount;
 		}
 		/*
@@ -211,7 +214,7 @@ xfs_alloc_delrec(
 		 * and it's NOT the leaf level,
 		 * then we can get rid of this level.
 		 */
-		if (be16_to_cpu(block->bb_numrecs) == 1 && level > 0) {
+		if (numrecs == 1 && level > 0) {
 			/*
 			 * lpp is still set to the first pointer in the block.
 			 * Make it the new root of the btree.
@@ -267,7 +270,7 @@ xfs_alloc_delrec(
 	 * If the number of records remaining in the block is at least
 	 * the minimum, we're done.
 	 */
-	if (be16_to_cpu(block->bb_numrecs) >= XFS_ALLOC_BLOCK_MINRECS(level, cur)) {
+	if (numrecs >= XFS_ALLOC_BLOCK_MINRECS(level, cur)) {
 		if (level > 0 && (error = xfs_alloc_decrement(cur, level, &i)))
 			return error;
 		*stat = 1;
@@ -419,19 +422,21 @@ xfs_alloc_delrec(
 	 * See if we can join with the left neighbor block.
 	 */
 	if (lbno != NULLAGBLOCK &&
-	    lrecs + be16_to_cpu(block->bb_numrecs) <= XFS_ALLOC_BLOCK_MAXRECS(level, cur)) {
+	    lrecs + numrecs <= XFS_ALLOC_BLOCK_MAXRECS(level, cur)) {
 		/*
 		 * Set "right" to be the starting block,
 		 * "left" to be the left neighbor.
 		 */
 		rbno = bno;
 		right = block;
+		rrecs = be16_to_cpu(right->bb_numrecs);
 		rbp = bp;
 		if ((error = xfs_btree_read_bufs(mp, cur->bc_tp,
 				cur->bc_private.a.agno, lbno, 0, &lbp,
 				XFS_ALLOC_BTREE_REF)))
 			return error;
 		left = XFS_BUF_TO_ALLOC_BLOCK(lbp);
+		lrecs = be16_to_cpu(left->bb_numrecs);
 		if ((error = xfs_btree_check_sblock(cur, left, level, lbp)))
 			return error;
 	}
@@ -439,20 +444,21 @@ xfs_alloc_delrec(
 	 * If that won't work, see if we can join with the right neighbor block.
 	 */
 	else if (rbno != NULLAGBLOCK &&
-		 rrecs + be16_to_cpu(block->bb_numrecs) <=
-		  XFS_ALLOC_BLOCK_MAXRECS(level, cur)) {
+		 rrecs + numrecs <= XFS_ALLOC_BLOCK_MAXRECS(level, cur)) {
 		/*
 		 * Set "left" to be the starting block,
 		 * "right" to be the right neighbor.
 		 */
 		lbno = bno;
 		left = block;
+		lrecs = be16_to_cpu(left->bb_numrecs);
 		lbp = bp;
 		if ((error = xfs_btree_read_bufs(mp, cur->bc_tp,
 				cur->bc_private.a.agno, rbno, 0, &rbp,
 				XFS_ALLOC_BTREE_REF)))
 			return error;
 		right = XFS_BUF_TO_ALLOC_BLOCK(rbp);
+		rrecs = be16_to_cpu(right->bb_numrecs);
 		if ((error = xfs_btree_check_sblock(cur, right, level, rbp)))
 			return error;
 	}
@@ -474,34 +480,28 @@ xfs_alloc_delrec(
 		/*
 		 * It's a non-leaf.  Move keys and pointers.
 		 */
-		lkp = XFS_ALLOC_KEY_ADDR(left, be16_to_cpu(left->bb_numrecs) + 1, cur);
-		lpp = XFS_ALLOC_PTR_ADDR(left, be16_to_cpu(left->bb_numrecs) + 1, cur);
+		lkp = XFS_ALLOC_KEY_ADDR(left, lrecs + 1, cur);
+		lpp = XFS_ALLOC_PTR_ADDR(left, lrecs + 1, cur);
 		rkp = XFS_ALLOC_KEY_ADDR(right, 1, cur);
 		rpp = XFS_ALLOC_PTR_ADDR(right, 1, cur);
 #ifdef DEBUG
-		for (i = 0; i < be16_to_cpu(right->bb_numrecs); i++) {
+		for (i = 0; i < rrecs; i++) {
 			if ((error = xfs_btree_check_sptr(cur, be32_to_cpu(rpp[i]), level)))
 				return error;
 		}
 #endif
-		memcpy(lkp, rkp, be16_to_cpu(right->bb_numrecs) * sizeof(*lkp));
-		memcpy(lpp, rpp, be16_to_cpu(right->bb_numrecs) * sizeof(*lpp));
-		xfs_alloc_log_keys(cur, lbp, be16_to_cpu(left->bb_numrecs) + 1,
-				   be16_to_cpu(left->bb_numrecs) +
-				   be16_to_cpu(right->bb_numrecs));
-		xfs_alloc_log_ptrs(cur, lbp, be16_to_cpu(left->bb_numrecs) + 1,
-				   be16_to_cpu(left->bb_numrecs) +
-				   be16_to_cpu(right->bb_numrecs));
+		memcpy(lkp, rkp, rrecs * sizeof(*lkp));
+		memcpy(lpp, rpp, rrecs * sizeof(*lpp));
+		xfs_alloc_log_keys(cur, lbp, lrecs + 1, lrecs + rrecs);
+		xfs_alloc_log_ptrs(cur, lbp, lrecs + 1, lrecs + rrecs);
 	} else {
 		/*
 		 * It's a leaf.  Move records.
 		 */
-		lrp = XFS_ALLOC_REC_ADDR(left, be16_to_cpu(left->bb_numrecs) + 1, cur);
+		lrp = XFS_ALLOC_REC_ADDR(left, lrecs + 1, cur);
 		rrp = XFS_ALLOC_REC_ADDR(right, 1, cur);
-		memcpy(lrp, rrp, be16_to_cpu(right->bb_numrecs) * sizeof(*lrp));
-		xfs_alloc_log_recs(cur, lbp, be16_to_cpu(left->bb_numrecs) + 1,
-				   be16_to_cpu(left->bb_numrecs) +
-				   be16_to_cpu(right->bb_numrecs));
+		memcpy(lrp, rrp, rrecs * sizeof(*lrp));
+		xfs_alloc_log_recs(cur, lbp, lrecs + 1, lrecs + rrecs);
 	}
 	/*
 	 * If we joined with the left neighbor, set the buffer in the
@@ -509,7 +509,7 @@ xfs_alloc_delrec(
 	 */
 	if (bp != lbp) {
 		xfs_btree_setbuf(cur, level, lbp);
-		cur->bc_ptrs[level] += be16_to_cpu(left->bb_numrecs);
+		cur->bc_ptrs[level] += lrecs;
 	}
 	/*
 	 * If we joined with the right neighbor and there's a level above
@@ -521,7 +521,8 @@ xfs_alloc_delrec(
 	/*
 	 * Fix up the number of records in the surviving block.
 	 */
-	be16_add(&left->bb_numrecs, be16_to_cpu(right->bb_numrecs));
+	lrecs += rrecs;
+	left->bb_numrecs = cpu_to_be16(lrecs);
 	/*
 	 * Fix up the right block pointer in the surviving block, and log it.
 	 */
@@ -608,6 +609,7 @@ xfs_alloc_insrec(
 	xfs_btree_cur_t		*ncur;	/* new cursor to be used at next lvl */
 	xfs_alloc_key_t		nkey;	/* new key value, from split */
 	xfs_alloc_rec_t		nrec;	/* new record value, for caller */
+	int			numrecs;
 	int			optr;	/* old ptr value */
 	xfs_alloc_ptr_t		*pp;	/* pointer to btree addresses */
 	int			ptr;	/* index in btree block for this rec */
@@ -653,13 +655,14 @@ xfs_alloc_insrec(
 	 */
 	bp = cur->bc_bufs[level];
 	block = XFS_BUF_TO_ALLOC_BLOCK(bp);
+	numrecs = be16_to_cpu(block->bb_numrecs);
 #ifdef DEBUG
 	if ((error = xfs_btree_check_sblock(cur, block, level, bp)))
 		return error;
 	/*
 	 * Check that the new entry is being inserted in the right place.
 	 */
-	if (ptr <= be16_to_cpu(block->bb_numrecs)) {
+	if (ptr <= numrecs) {
 		if (level == 0) {
 			rp = XFS_ALLOC_REC_ADDR(block, ptr, cur);
 			xfs_btree_check_rec(cur->bc_btnum, recp, rp);
@@ -670,12 +673,12 @@ xfs_alloc_insrec(
 	}
 #endif
 	nbno = NULLAGBLOCK;
-	ncur = (xfs_btree_cur_t *)0;
+	ncur = NULL;
 	/*
 	 * If the block is full, we can't insert the new entry until we
 	 * make the block un-full.
 	 */
-	if (be16_to_cpu(block->bb_numrecs) == XFS_ALLOC_BLOCK_MAXRECS(level, cur)) {
+	if (numrecs == XFS_ALLOC_BLOCK_MAXRECS(level, cur)) {
 		/*
 		 * First, try shifting an entry to the right neighbor.
 		 */
@@ -729,6 +732,7 @@ xfs_alloc_insrec(
 	 * At this point we know there's room for our new entry in the block
 	 * we're pointing at.
 	 */
+	numrecs = be16_to_cpu(block->bb_numrecs);
 	if (level > 0) {
 		/*
 		 * It's a non-leaf entry.  Make a hole for the new data
@@ -737,15 +741,15 @@ xfs_alloc_insrec(
 		kp = XFS_ALLOC_KEY_ADDR(block, 1, cur);
 		pp = XFS_ALLOC_PTR_ADDR(block, 1, cur);
 #ifdef DEBUG
-		for (i = be16_to_cpu(block->bb_numrecs); i >= ptr; i--) {
+		for (i = numrecs; i >= ptr; i--) {
 			if ((error = xfs_btree_check_sptr(cur, be32_to_cpu(pp[i - 1]), level)))
 				return error;
 		}
 #endif
 		memmove(&kp[ptr], &kp[ptr - 1],
-			(be16_to_cpu(block->bb_numrecs) - ptr + 1) * sizeof(*kp));
+			(numrecs - ptr + 1) * sizeof(*kp));
 		memmove(&pp[ptr], &pp[ptr - 1],
-			(be16_to_cpu(block->bb_numrecs) - ptr + 1) * sizeof(*pp));
+			(numrecs - ptr + 1) * sizeof(*pp));
 #ifdef DEBUG
 		if ((error = xfs_btree_check_sptr(cur, *bnop, level)))
 			return error;
@@ -755,11 +759,12 @@ xfs_alloc_insrec(
 		 */
 		kp[ptr - 1] = key;
 		pp[ptr - 1] = cpu_to_be32(*bnop);
-		be16_add(&block->bb_numrecs, 1);
-		xfs_alloc_log_keys(cur, bp, ptr, be16_to_cpu(block->bb_numrecs));
-		xfs_alloc_log_ptrs(cur, bp, ptr, be16_to_cpu(block->bb_numrecs));
+		numrecs++;
+		block->bb_numrecs = cpu_to_be16(numrecs);
+		xfs_alloc_log_keys(cur, bp, ptr, numrecs);
+		xfs_alloc_log_ptrs(cur, bp, ptr, numrecs);
 #ifdef DEBUG
-		if (ptr < be16_to_cpu(block->bb_numrecs))
+		if (ptr < numrecs)
 			xfs_btree_check_key(cur->bc_btnum, kp + ptr - 1,
 				kp + ptr);
 #endif
@@ -769,16 +774,17 @@ xfs_alloc_insrec(
 		 */
 		rp = XFS_ALLOC_REC_ADDR(block, 1, cur);
 		memmove(&rp[ptr], &rp[ptr - 1],
-			(be16_to_cpu(block->bb_numrecs) - ptr + 1) * sizeof(*rp));
+			(numrecs - ptr + 1) * sizeof(*rp));
 		/*
 		 * Now stuff the new record in, bump numrecs
 		 * and log the new data.
 		 */
-		rp[ptr - 1] = *recp; /* INT_: struct copy */
-		be16_add(&block->bb_numrecs, 1);
-		xfs_alloc_log_recs(cur, bp, ptr, be16_to_cpu(block->bb_numrecs));
+		rp[ptr - 1] = *recp;
+		numrecs++;
+		block->bb_numrecs = cpu_to_be16(numrecs);
+		xfs_alloc_log_recs(cur, bp, ptr, numrecs);
 #ifdef DEBUG
-		if (ptr < be16_to_cpu(block->bb_numrecs))
+		if (ptr < numrecs)
 			xfs_btree_check_rec(cur->bc_btnum, rp + ptr - 1,
 				rp + ptr);
 #endif
@@ -819,8 +825,8 @@ xfs_alloc_insrec(
 	 */
 	*bnop = nbno;
 	if (nbno != NULLAGBLOCK) {
-		*recp = nrec; /* INT_: struct copy */
-		*curp = ncur; /* INT_: struct copy */
+		*recp = nrec;
+		*curp = ncur;
 	}
 	*stat = 1;
 	return 0;
@@ -981,7 +987,7 @@ xfs_alloc_lookup(
 		 */
 		bp = cur->bc_bufs[level];
 		if (bp && XFS_BUF_ADDR(bp) != d)
-			bp = (xfs_buf_t *)0;
+			bp = NULL;
 		if (!bp) {
 			/*
 			 * Need to get a new buffer.  Read it, then
@@ -1229,7 +1235,7 @@ xfs_alloc_lshift(
 		if ((error = xfs_btree_check_sptr(cur, be32_to_cpu(*rpp), level)))
 			return error;
 #endif
-		*lpp = *rpp; /* INT_: copy */
+		*lpp = *rpp;
 		xfs_alloc_log_ptrs(cur, lbp, nrec, nrec);
 		xfs_btree_check_key(cur->bc_btnum, lkp - 1, lkp);
 	}
@@ -1406,8 +1412,8 @@ xfs_alloc_newroot(
 
 		kp = XFS_ALLOC_KEY_ADDR(new, 1, cur);
 		if (be16_to_cpu(left->bb_level) > 0) {
-			kp[0] = *XFS_ALLOC_KEY_ADDR(left, 1, cur); /* INT_: structure copy */
-			kp[1] = *XFS_ALLOC_KEY_ADDR(right, 1, cur);/* INT_: structure copy */
+			kp[0] = *XFS_ALLOC_KEY_ADDR(left, 1, cur);
+			kp[1] = *XFS_ALLOC_KEY_ADDR(right, 1, cur);
 		} else {
 			xfs_alloc_rec_t	*rp;	/* btree record pointer */
 
@@ -1527,8 +1533,8 @@ xfs_alloc_rshift(
 		if ((error = xfs_btree_check_sptr(cur, be32_to_cpu(*lpp), level)))
 			return error;
 #endif
-		*rkp = *lkp; /* INT_: copy */
-		*rpp = *lpp; /* INT_: copy */
+		*rkp = *lkp;
+		*rpp = *lpp;
 		xfs_alloc_log_keys(cur, rbp, 1, be16_to_cpu(right->bb_numrecs) + 1);
 		xfs_alloc_log_ptrs(cur, rbp, 1, be16_to_cpu(right->bb_numrecs) + 1);
 		xfs_btree_check_key(cur->bc_btnum, rkp, rkp + 1);
@@ -2044,7 +2050,7 @@ xfs_alloc_insert(
 	nbno = NULLAGBLOCK;
 	nrec.ar_startblock = cpu_to_be32(cur->bc_rec.a.ar_startblock);
 	nrec.ar_blockcount = cpu_to_be32(cur->bc_rec.a.ar_blockcount);
-	ncur = (xfs_btree_cur_t *)0;
+	ncur = NULL;
 	pcur = cur;
 	/*
 	 * Loop going up the tree, starting at the leaf level.
@@ -2076,7 +2082,7 @@ xfs_alloc_insert(
 		 */
 		if (ncur) {
 			pcur = ncur;
-			ncur = (xfs_btree_cur_t *)0;
+			ncur = NULL;
 		}
 	} while (nbno != NULLAGBLOCK);
 	*stat = i;

fs/xfs/xfs_attr.c

@@ -91,7 +91,6 @@ STATIC int xfs_attr_refillstate(xfs_da_state_t *state);
 /*
  * Routines to manipulate out-of-line attribute values.
  */
-STATIC int xfs_attr_rmtval_get(xfs_da_args_t *args);
 STATIC int xfs_attr_rmtval_set(xfs_da_args_t *args);
 STATIC int xfs_attr_rmtval_remove(xfs_da_args_t *args);
 
@@ -180,7 +179,7 @@ xfs_attr_get(bhv_desc_t *bdp, const char *name, char *value, int *valuelenp,
 	return(error);
 }
 
-STATIC int
+int
 xfs_attr_set_int(xfs_inode_t *dp, const char *name, int namelen,
 		 char *value, int valuelen, int flags)
 {
@@ -440,7 +439,7 @@ xfs_attr_set(bhv_desc_t *bdp, const char *name, char *value, int valuelen, int f
  * Generic handler routine to remove a name from an attribute list.
  * Transitions attribute list from Btree to shortform as necessary.
  */
-STATIC int
+int
 xfs_attr_remove_int(xfs_inode_t *dp, const char *name, int namelen, int flags)
 {
 	xfs_da_args_t	args;
@@ -591,6 +590,110 @@ xfs_attr_remove(bhv_desc_t *bdp, const char *name, int flags, struct cred *cred)
 	return xfs_attr_remove_int(dp, name, namelen, flags);
 }
 
+int								/* error */
+xfs_attr_list_int(xfs_attr_list_context_t *context)
+{
+	int error;
+	xfs_inode_t *dp = context->dp;
+
+	/*
+	 * Decide on what work routines to call based on the inode size.
+	 */
+	if (XFS_IFORK_Q(dp) == 0 ||
+	    (dp->i_d.di_aformat == XFS_DINODE_FMT_EXTENTS &&
+	     dp->i_d.di_anextents == 0)) {
+		error = 0;
+	} else if (dp->i_d.di_aformat == XFS_DINODE_FMT_LOCAL) {
+		error = xfs_attr_shortform_list(context);
+	} else if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) {
+		error = xfs_attr_leaf_list(context);
+	} else {
+		error = xfs_attr_node_list(context);
+	}
+	return error;
+}
+
+#define	ATTR_ENTBASESIZE		/* minimum bytes used by an attr */ \
+	(((struct attrlist_ent *) 0)->a_name - (char *) 0)
+#define	ATTR_ENTSIZE(namelen)		/* actual bytes used by an attr */ \
+	((ATTR_ENTBASESIZE + (namelen) + 1 + sizeof(u_int32_t)-1) \
+	 & ~(sizeof(u_int32_t)-1))
+
+/*
+ * Format an attribute and copy it out to the user's buffer.
+ * Take care to check values and protect against them changing later,
+ * we may be reading them directly out of a user buffer.
+ */
+/*ARGSUSED*/
+STATIC int
+xfs_attr_put_listent(xfs_attr_list_context_t *context, attrnames_t *namesp,
+		     char *name, int namelen,
+		     int valuelen, char *value)
+{
+	attrlist_ent_t *aep;
+	int arraytop;
+
+	ASSERT(!(context->flags & ATTR_KERNOVAL));
+	ASSERT(context->count >= 0);
+	ASSERT(context->count < (ATTR_MAX_VALUELEN/8));
+	ASSERT(context->firstu >= sizeof(*context->alist));
+	ASSERT(context->firstu <= context->bufsize);
+
+	arraytop = sizeof(*context->alist) +
+			context->count * sizeof(context->alist->al_offset[0]);
+	context->firstu -= ATTR_ENTSIZE(namelen);
+	if (context->firstu < arraytop) {
+		xfs_attr_trace_l_c("buffer full", context);
+		context->alist->al_more = 1;
+		context->seen_enough = 1;
+		return 1;
+	}
+
+	aep = (attrlist_ent_t *)&(((char *)context->alist)[ context->firstu ]);
+	aep->a_valuelen = valuelen;
+	memcpy(aep->a_name, name, namelen);
+	aep->a_name[ namelen ] = 0;
+	context->alist->al_offset[ context->count++ ] = context->firstu;
+	context->alist->al_count = context->count;
+	xfs_attr_trace_l_c("add", context);
+	return 0;
+}
+
+STATIC int
+xfs_attr_kern_list(xfs_attr_list_context_t *context, attrnames_t *namesp,
+		     char *name, int namelen,
+		     int valuelen, char *value)
+{
+	char *offset;
+	int arraytop;
+
+	ASSERT(context->count >= 0);
+
+	arraytop = context->count + namesp->attr_namelen + namelen + 1;
+	if (arraytop > context->firstu) {
+		context->count = -1;	/* insufficient space */
+		return 1;
+	}
+	offset = (char *)context->alist + context->count;
+	strncpy(offset, namesp->attr_name, namesp->attr_namelen);
+	offset += namesp->attr_namelen;
+	strncpy(offset, name, namelen);			/* real name */
+	offset += namelen;
+	*offset = '\0';
+	context->count += namesp->attr_namelen + namelen + 1;
+	return 0;
+}
+
+/*ARGSUSED*/
+STATIC int
+xfs_attr_kern_list_sizes(xfs_attr_list_context_t *context, attrnames_t *namesp,
+		     char *name, int namelen,
+		     int valuelen, char *value)
+{
+	context->count += namesp->attr_namelen + namelen + 1;
+	return 0;
+}
+
 /*
  * Generate a list of extended attribute names and optionally
  * also value lengths.  Positive return value follows the XFS
@@ -615,13 +718,13 @@ xfs_attr_list(bhv_desc_t *bdp, char *buffer, int bufsize, int flags,
 		return(XFS_ERROR(EINVAL));
 	if ((cursor->initted == 0) &&
 	    (cursor->hashval || cursor->blkno || cursor->offset))
-		return(XFS_ERROR(EINVAL));
+		return XFS_ERROR(EINVAL);
 
 	/*
 	 * Check for a properly aligned buffer.
 	 */
 	if (((long)buffer) & (sizeof(int)-1))
-		return(XFS_ERROR(EFAULT));
+		return XFS_ERROR(EFAULT);
 	if (flags & ATTR_KERNOVAL)
 		bufsize = 0;
 
@@ -634,53 +737,47 @@ xfs_attr_list(bhv_desc_t *bdp, char *buffer, int bufsize, int flags,
 	context.dupcnt = 0;
 	context.resynch = 1;
 	context.flags = flags;
-	if (!(flags & ATTR_KERNAMELS)) {
+	context.seen_enough = 0;
+	context.alist = (attrlist_t *)buffer;
+	context.put_value = 0;
+
+	if (flags & ATTR_KERNAMELS) {
+		context.bufsize = bufsize;
+		context.firstu = context.bufsize;
+		if (flags & ATTR_KERNOVAL)
+			context.put_listent = xfs_attr_kern_list_sizes;
+		else
+			context.put_listent = xfs_attr_kern_list;
+	} else {
 		context.bufsize = (bufsize & ~(sizeof(int)-1));  /* align */
 		context.firstu = context.bufsize;
-		context.alist = (attrlist_t *)buffer;
 		context.alist->al_count = 0;
 		context.alist->al_more = 0;
 		context.alist->al_offset[0] = context.bufsize;
-	}
-	else {
-		context.bufsize = bufsize;
-		context.firstu = context.bufsize;
-		context.alist = (attrlist_t *)buffer;
+		context.put_listent = xfs_attr_put_listent;
 	}
 
 	if (XFS_FORCED_SHUTDOWN(dp->i_mount))
-		return (EIO);
+		return EIO;
 
 	xfs_ilock(dp, XFS_ILOCK_SHARED);
-	/*
-	 * Decide on what work routines to call based on the inode size.
-	 */
 	xfs_attr_trace_l_c("syscall start", &context);
-	if (XFS_IFORK_Q(dp) == 0 ||
-	    (dp->i_d.di_aformat == XFS_DINODE_FMT_EXTENTS &&
-	     dp->i_d.di_anextents == 0)) {
-		error = 0;
-	} else if (dp->i_d.di_aformat == XFS_DINODE_FMT_LOCAL) {
-		error = xfs_attr_shortform_list(&context);
-	} else if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) {
-		error = xfs_attr_leaf_list(&context);
-	} else {
-		error = xfs_attr_node_list(&context);
-	}
+
+	error = xfs_attr_list_int(&context);
+
 	xfs_iunlock(dp, XFS_ILOCK_SHARED);
 	xfs_attr_trace_l_c("syscall end", &context);
 
-	if (!(context.flags & (ATTR_KERNOVAL|ATTR_KERNAMELS))) {
-		ASSERT(error >= 0);
-	}
-	else {	/* must return negated buffer size or the error */
+	if (context.flags & (ATTR_KERNOVAL|ATTR_KERNAMELS)) {
+		/* must return negated buffer size or the error */
 		if (context.count < 0)
 			error = XFS_ERROR(ERANGE);
 		else
 			error = -context.count;
-	}
+	} else
+		ASSERT(error >= 0);
 
-	return(error);
+	return error;
 }
 
 int								/* error */
@@ -1122,19 +1219,19 @@ xfs_attr_leaf_list(xfs_attr_list_context_t *context)
 	context->cursor->blkno = 0;
 	error = xfs_da_read_buf(NULL, context->dp, 0, -1, &bp, XFS_ATTR_FORK);
 	if (error)
-		return(error);
+		return XFS_ERROR(error);
 	ASSERT(bp != NULL);
 	leaf = bp->data;
 	if (unlikely(be16_to_cpu(leaf->hdr.info.magic) != XFS_ATTR_LEAF_MAGIC)) {
 		XFS_CORRUPTION_ERROR("xfs_attr_leaf_list", XFS_ERRLEVEL_LOW,
 				     context->dp->i_mount, leaf);
 		xfs_da_brelse(NULL, bp);
-		return(XFS_ERROR(EFSCORRUPTED));
+		return XFS_ERROR(EFSCORRUPTED);
 	}
 
-	(void)xfs_attr_leaf_list_int(bp, context);
+	error = xfs_attr_leaf_list_int(bp, context);
 	xfs_da_brelse(NULL, bp);
-	return(0);
+	return XFS_ERROR(error);
 }
 
 
@@ -1858,8 +1955,12 @@ xfs_attr_node_list(xfs_attr_list_context_t *context)
 			return(XFS_ERROR(EFSCORRUPTED));
 		}
 		error = xfs_attr_leaf_list_int(bp, context);
-		if (error || !leaf->hdr.info.forw)
-			break;	/* not really an error, buffer full or EOF */
+		if (error) {
+			xfs_da_brelse(NULL, bp);
+			return error;
+		}
+		if (context->seen_enough || leaf->hdr.info.forw == 0)
+			break;
 		cursor->blkno = be32_to_cpu(leaf->hdr.info.forw);
 		xfs_da_brelse(NULL, bp);
 		error = xfs_da_read_buf(NULL, context->dp, cursor->blkno, -1,
@@ -1886,7 +1987,7 @@ xfs_attr_node_list(xfs_attr_list_context_t *context)
  * Read the value associated with an attribute from the out-of-line buffer
  * that we stored it in.
  */
-STATIC int
+int
 xfs_attr_rmtval_get(xfs_da_args_t *args)
 {
 	xfs_bmbt_irec_t map[ATTR_RMTVALUE_MAPSIZE];

fs/xfs/xfs_attr.h

@@ -37,6 +37,7 @@
 
 struct cred;
 struct bhv_vnode;
+struct xfs_attr_list_context;
 
 typedef int (*attrset_t)(struct bhv_vnode *, char *, void *, size_t, int);
 typedef int (*attrget_t)(struct bhv_vnode *, char *, void *, size_t, int);
@@ -160,13 +161,16 @@ struct xfs_da_args;
  */
 int xfs_attr_get(bhv_desc_t *, const char *, char *, int *, int, struct cred *);
 int xfs_attr_set(bhv_desc_t *, const char *, char *, int, int, struct cred *);
+int xfs_attr_set_int(struct xfs_inode *, const char *, int, char *, int, int);
 int xfs_attr_remove(bhv_desc_t *, const char *, int, struct cred *);
-int xfs_attr_list(bhv_desc_t *, char *, int, int,
-			 struct attrlist_cursor_kern *, struct cred *);
+int xfs_attr_remove_int(struct xfs_inode *, const char *, int, int);
+int xfs_attr_list(bhv_desc_t *, char *, int, int, struct attrlist_cursor_kern *, struct cred *);
+int xfs_attr_list_int(struct xfs_attr_list_context *);
 int xfs_attr_inactive(struct xfs_inode *dp);
 
 int xfs_attr_shortform_getvalue(struct xfs_da_args *);
 int xfs_attr_fetch(struct xfs_inode *, const char *, int,
 			char *, int *, int, struct cred *);
+int xfs_attr_rmtval_get(struct xfs_da_args *args);
 
 #endif	/* __XFS_ATTR_H__ */

fs/xfs/xfs_attr_leaf.c

@@ -89,9 +89,46 @@ STATIC void xfs_attr_leaf_moveents(xfs_attr_leafblock_t *src_leaf,
 					 int dst_start, int move_count,
 					 xfs_mount_t *mp);
 STATIC int xfs_attr_leaf_entsize(xfs_attr_leafblock_t *leaf, int index);
-STATIC int xfs_attr_put_listent(xfs_attr_list_context_t *context,
-			     attrnames_t *, char *name, int namelen,
-			     int valuelen);
+
+/*========================================================================
+ * Namespace helper routines
+ *========================================================================*/
+
+STATIC inline attrnames_t *
+xfs_attr_flags_namesp(int flags)
+{
+	return ((flags & XFS_ATTR_SECURE) ? &attr_secure:
+		  ((flags & XFS_ATTR_ROOT) ? &attr_trusted : &attr_user));
+}
+
+/*
+ * If namespace bits don't match return 0.
+ * If all match then return 1.
+ */
+STATIC inline int
+xfs_attr_namesp_match(int arg_flags, int ondisk_flags)
+{
+	return XFS_ATTR_NSP_ONDISK(ondisk_flags) == XFS_ATTR_NSP_ARGS_TO_ONDISK(arg_flags);
+}
+
+/*
+ * If namespace bits don't match and we don't have an override for it
+ * then return 0.
+ * If all match or are overridable then return 1.
+ */
+STATIC inline int
+xfs_attr_namesp_match_overrides(int arg_flags, int ondisk_flags)
+{
+	if (((arg_flags & ATTR_SECURE) == 0) !=
+	    ((ondisk_flags & XFS_ATTR_SECURE) == 0) &&
+	    !(arg_flags & ATTR_KERNORMALS))
+		return 0;
+	if (((arg_flags & ATTR_ROOT) == 0) !=
+	    ((ondisk_flags & XFS_ATTR_ROOT) == 0) &&
+	    !(arg_flags & ATTR_KERNROOTLS))
+		return 0;
+	return 1;
+}
 
 
 /*========================================================================
@@ -228,11 +265,7 @@ xfs_attr_shortform_add(xfs_da_args_t *args, int forkoff)
 			continue;
 		if (memcmp(args->name, sfe->nameval, args->namelen) != 0)
 			continue;
-		if (((args->flags & ATTR_SECURE) != 0) !=
-		    ((sfe->flags & XFS_ATTR_SECURE) != 0))
-			continue;
-		if (((args->flags & ATTR_ROOT) != 0) !=
-		    ((sfe->flags & XFS_ATTR_ROOT) != 0))
+		if (!xfs_attr_namesp_match(args->flags, sfe->flags))
 			continue;
 		ASSERT(0);
 #endif
@@ -246,8 +279,7 @@ xfs_attr_shortform_add(xfs_da_args_t *args, int forkoff)
 
 	sfe->namelen = args->namelen;
 	sfe->valuelen = args->valuelen;
-	sfe->flags = (args->flags & ATTR_SECURE) ? XFS_ATTR_SECURE :
-			((args->flags & ATTR_ROOT) ? XFS_ATTR_ROOT : 0);
+	sfe->flags = XFS_ATTR_NSP_ARGS_TO_ONDISK(args->flags);
 	memcpy(sfe->nameval, args->name, args->namelen);
 	memcpy(&sfe->nameval[args->namelen], args->value, args->valuelen);
 	sf->hdr.count++;
@@ -282,11 +314,7 @@ xfs_attr_shortform_remove(xfs_da_args_t *args)
 			continue;
 		if (memcmp(sfe->nameval, args->name, args->namelen) != 0)
 			continue;
-		if (((args->flags & ATTR_SECURE) != 0) !=
-		    ((sfe->flags & XFS_ATTR_SECURE) != 0))
-			continue;
-		if (((args->flags & ATTR_ROOT) != 0) !=
-		    ((sfe->flags & XFS_ATTR_ROOT) != 0))
+		if (!xfs_attr_namesp_match(args->flags, sfe->flags))
 			continue;
 		break;
 	}
@@ -363,11 +391,7 @@ xfs_attr_shortform_lookup(xfs_da_args_t *args)
 			continue;
 		if (memcmp(args->name, sfe->nameval, args->namelen) != 0)
 			continue;
-		if (((args->flags & ATTR_SECURE) != 0) !=
-		    ((sfe->flags & XFS_ATTR_SECURE) != 0))
-			continue;
-		if (((args->flags & ATTR_ROOT) != 0) !=
-		    ((sfe->flags & XFS_ATTR_ROOT) != 0))
+		if (!xfs_attr_namesp_match(args->flags, sfe->flags))
 			continue;
 		return(XFS_ERROR(EEXIST));
 	}
@@ -394,11 +418,7 @@ xfs_attr_shortform_getvalue(xfs_da_args_t *args)
 			continue;
 		if (memcmp(args->name, sfe->nameval, args->namelen) != 0)
 			continue;
-		if (((args->flags & ATTR_SECURE) != 0) !=
-		    ((sfe->flags & XFS_ATTR_SECURE) != 0))
-			continue;
-		if (((args->flags & ATTR_ROOT) != 0) !=
-		    ((sfe->flags & XFS_ATTR_ROOT) != 0))
+		if (!xfs_attr_namesp_match(args->flags, sfe->flags))
 			continue;
 		if (args->flags & ATTR_KERNOVAL) {
 			args->valuelen = sfe->valuelen;
@@ -485,8 +505,7 @@ xfs_attr_shortform_to_leaf(xfs_da_args_t *args)
 		nargs.valuelen = sfe->valuelen;
 		nargs.hashval = xfs_da_hashname((char *)sfe->nameval,
 						sfe->namelen);
-		nargs.flags = (sfe->flags & XFS_ATTR_SECURE) ? ATTR_SECURE :
-				((sfe->flags & XFS_ATTR_ROOT) ? ATTR_ROOT : 0);
+		nargs.flags = XFS_ATTR_NSP_ONDISK_TO_ARGS(sfe->flags);
 		error = xfs_attr_leaf_lookup_int(bp, &nargs); /* set a->index */
 		ASSERT(error == ENOATTR);
 		error = xfs_attr_leaf_add(bp, &nargs);
@@ -520,6 +539,10 @@ xfs_attr_shortform_compare(const void *a, const void *b)
 	}
 }
 
+
+#define XFS_ISRESET_CURSOR(cursor) \
+	(!((cursor)->initted) && !((cursor)->hashval) && \
+	 !((cursor)->blkno) && !((cursor)->offset))
 /*
  * Copy out entries of shortform attribute lists for attr_list().
  * Shortform attribute lists are not stored in hashval sorted order.
@@ -537,6 +560,7 @@ xfs_attr_shortform_list(xfs_attr_list_context_t *context)
 	xfs_attr_sf_entry_t *sfe;
 	xfs_inode_t *dp;
 	int sbsize, nsbuf, count, i;
+	int error;
 
 	ASSERT(context != NULL);
 	dp = context->dp;
@@ -552,46 +576,51 @@ xfs_attr_shortform_list(xfs_attr_list_context_t *context)
 	xfs_attr_trace_l_c("sf start", context);
 
 	/*
-	 * If the buffer is large enough, do not bother with sorting.
+	 * If the buffer is large enough and the cursor is at the start,
+	 * do not bother with sorting since we will return everything in
+	 * one buffer and another call using the cursor won't need to be
+	 * made.
 	 * Note the generous fudge factor of 16 overhead bytes per entry.
+	 * If bufsize is zero then put_listent must be a search function
+	 * and can just scan through what we have.
 	 */
-	if ((dp->i_afp->if_bytes + sf->hdr.count * 16) < context->bufsize) {
+	if (context->bufsize == 0 ||
+	    (XFS_ISRESET_CURSOR(cursor) &&
+             (dp->i_afp->if_bytes + sf->hdr.count * 16) < context->bufsize)) {
 		for (i = 0, sfe = &sf->list[0]; i < sf->hdr.count; i++) {
 			attrnames_t	*namesp;
 
-			if (((context->flags & ATTR_SECURE) != 0) !=
-			    ((sfe->flags & XFS_ATTR_SECURE) != 0) &&
-			    !(context->flags & ATTR_KERNORMALS)) {
-				sfe = XFS_ATTR_SF_NEXTENTRY(sfe);
-				continue;
-			}
-			if (((context->flags & ATTR_ROOT) != 0) !=
-			    ((sfe->flags & XFS_ATTR_ROOT) != 0) &&
-			    !(context->flags & ATTR_KERNROOTLS)) {
+			if (!xfs_attr_namesp_match_overrides(context->flags, sfe->flags)) {
 				sfe = XFS_ATTR_SF_NEXTENTRY(sfe);
 				continue;
 			}
-			namesp = (sfe->flags & XFS_ATTR_SECURE) ? &attr_secure:
-				((sfe->flags & XFS_ATTR_ROOT) ? &attr_trusted :
-				  &attr_user);
-			if (context->flags & ATTR_KERNOVAL) {
-				ASSERT(context->flags & ATTR_KERNAMELS);
-				context->count += namesp->attr_namelen +
-					sfe->namelen + 1;
-			}
-			else {
-				if (xfs_attr_put_listent(context, namesp,
-						   (char *)sfe->nameval,
-						   (int)sfe->namelen,
-						   (int)sfe->valuelen))
-					break;
-			}
+			namesp = xfs_attr_flags_namesp(sfe->flags);
+			error = context->put_listent(context,
+					   namesp,
+					   (char *)sfe->nameval,
+					   (int)sfe->namelen,
+					   (int)sfe->valuelen,
+					   (char*)&sfe->nameval[sfe->namelen]);
+
+			/*
+			 * Either search callback finished early or
+			 * didn't fit it all in the buffer after all.
+			 */
+			if (context->seen_enough)
+				break;
+
+			if (error)
+				return error;
 			sfe = XFS_ATTR_SF_NEXTENTRY(sfe);
 		}
 		xfs_attr_trace_l_c("sf big-gulp", context);
 		return(0);
 	}
 
+	/* do no more for a search callback */
+	if (context->bufsize == 0)
+		return 0;
+
 	/*
 	 * It didn't all fit, so we have to sort everything on hashval.
 	 */
@@ -614,15 +643,7 @@ xfs_attr_shortform_list(xfs_attr_list_context_t *context)
 			kmem_free(sbuf, sbsize);
 			return XFS_ERROR(EFSCORRUPTED);
 		}
-		if (((context->flags & ATTR_SECURE) != 0) !=
-		    ((sfe->flags & XFS_ATTR_SECURE) != 0) &&
-		    !(context->flags & ATTR_KERNORMALS)) {
-			sfe = XFS_ATTR_SF_NEXTENTRY(sfe);
-			continue;
-		}
-		if (((context->flags & ATTR_ROOT) != 0) !=
-		    ((sfe->flags & XFS_ATTR_ROOT) != 0) &&
-		    !(context->flags & ATTR_KERNROOTLS)) {
+		if (!xfs_attr_namesp_match_overrides(context->flags, sfe->flags)) {
 			sfe = XFS_ATTR_SF_NEXTENTRY(sfe);
 			continue;
 		}
@@ -671,24 +692,22 @@ xfs_attr_shortform_list(xfs_attr_list_context_t *context)
 	for ( ; i < nsbuf; i++, sbp++) {
 		attrnames_t	*namesp;
 
-		namesp = (sbp->flags & XFS_ATTR_SECURE) ? &attr_secure :
-			((sbp->flags & XFS_ATTR_ROOT) ? &attr_trusted :
-			  &attr_user);
+		namesp = xfs_attr_flags_namesp(sbp->flags);
 
 		if (cursor->hashval != sbp->hash) {
 			cursor->hashval = sbp->hash;
 			cursor->offset = 0;
 		}
-		if (context->flags & ATTR_KERNOVAL) {
-			ASSERT(context->flags & ATTR_KERNAMELS);
-			context->count += namesp->attr_namelen +
-						sbp->namelen + 1;
-		} else {
-			if (xfs_attr_put_listent(context, namesp,
-					sbp->name, sbp->namelen,
-					sbp->valuelen))
-				break;
-		}
+		error = context->put_listent(context,
+					namesp,
+					sbp->name,
+					sbp->namelen,
+					sbp->valuelen,
+					&sbp->name[sbp->namelen]);
+		if (error)
+			return error;
+		if (context->seen_enough)
+			break;
 		cursor->offset++;
 	}
 
@@ -810,8 +829,7 @@ xfs_attr_leaf_to_shortform(xfs_dabuf_t *bp, xfs_da_args_t *args, int forkoff)
 		nargs.value = (char *)&name_loc->nameval[nargs.namelen];
 		nargs.valuelen = be16_to_cpu(name_loc->valuelen);
 		nargs.hashval = be32_to_cpu(entry->hashval);
-		nargs.flags = (entry->flags & XFS_ATTR_SECURE) ? ATTR_SECURE :
-			      ((entry->flags & XFS_ATTR_ROOT) ? ATTR_ROOT : 0);
+		nargs.flags = XFS_ATTR_NSP_ONDISK_TO_ARGS(entry->flags);
 		xfs_attr_shortform_add(&nargs, forkoff);
 	}
 	error = 0;
@@ -1098,8 +1116,7 @@ xfs_attr_leaf_add_work(xfs_dabuf_t *bp, xfs_da_args_t *args, int mapindex)
 				     be16_to_cpu(map->size));
 	entry->hashval = cpu_to_be32(args->hashval);
 	entry->flags = tmp ? XFS_ATTR_LOCAL : 0;
-	entry->flags |= (args->flags & ATTR_SECURE) ? XFS_ATTR_SECURE :
-			((args->flags & ATTR_ROOT) ? XFS_ATTR_ROOT : 0);
+	entry->flags |= XFS_ATTR_NSP_ARGS_TO_ONDISK(args->flags);
 	if (args->rename) {
 		entry->flags |= XFS_ATTR_INCOMPLETE;
 		if ((args->blkno2 == args->blkno) &&
@@ -1926,7 +1943,7 @@ xfs_attr_leaf_lookup_int(xfs_dabuf_t *bp, xfs_da_args_t *args)
 		else
 			break;
 	}
-	ASSERT((probe >= 0) && 
+	ASSERT((probe >= 0) &&
 	       (!leaf->hdr.count
 	       || (probe < be16_to_cpu(leaf->hdr.count))));
 	ASSERT((span <= 4) || (be32_to_cpu(entry->hashval) == hashval));
@@ -1971,14 +1988,9 @@ xfs_attr_leaf_lookup_int(xfs_dabuf_t *bp, xfs_da_args_t *args)
 			name_loc = XFS_ATTR_LEAF_NAME_LOCAL(leaf, probe);
 			if (name_loc->namelen != args->namelen)
 				continue;
-			if (memcmp(args->name, (char *)name_loc->nameval,
-					     args->namelen) != 0)
+			if (memcmp(args->name, (char *)name_loc->nameval, args->namelen) != 0)
 				continue;
-			if (((args->flags & ATTR_SECURE) != 0) !=
-			    ((entry->flags & XFS_ATTR_SECURE) != 0))
-				continue;
-			if (((args->flags & ATTR_ROOT) != 0) !=
-			    ((entry->flags & XFS_ATTR_ROOT) != 0))
+			if (!xfs_attr_namesp_match(args->flags, entry->flags))
 				continue;
 			args->index = probe;
 			return(XFS_ERROR(EEXIST));
@@ -1989,11 +2001,7 @@ xfs_attr_leaf_lookup_int(xfs_dabuf_t *bp, xfs_da_args_t *args)
 			if (memcmp(args->name, (char *)name_rmt->name,
 					     args->namelen) != 0)
 				continue;
-			if (((args->flags & ATTR_SECURE) != 0) !=
-			    ((entry->flags & XFS_ATTR_SECURE) != 0))
-				continue;
-			if (((args->flags & ATTR_ROOT) != 0) !=
-			    ((entry->flags & XFS_ATTR_ROOT) != 0))
+			if (!xfs_attr_namesp_match(args->flags, entry->flags))
 				continue;
 			args->index = probe;
 			args->rmtblkno = be32_to_cpu(name_rmt->valueblk);
@@ -2312,8 +2320,6 @@ xfs_attr_leaf_list_int(xfs_dabuf_t *bp, xfs_attr_list_context_t *context)
 	attrlist_cursor_kern_t *cursor;
 	xfs_attr_leafblock_t *leaf;
 	xfs_attr_leaf_entry_t *entry;
-	xfs_attr_leaf_name_local_t *name_loc;
-	xfs_attr_leaf_name_remote_t *name_rmt;
 	int retval, i;
 
 	ASSERT(bp != NULL);
@@ -2355,9 +2361,8 @@ xfs_attr_leaf_list_int(xfs_dabuf_t *bp, xfs_attr_list_context_t *context)
 	 * We have found our place, start copying out the new attributes.
 	 */
 	retval = 0;
-	for (  ; (i < be16_to_cpu(leaf->hdr.count))
-	     && (retval == 0); entry++, i++) {
-		attrnames_t	*namesp;
+	for (  ; (i < be16_to_cpu(leaf->hdr.count)); entry++, i++) {
+		attrnames_t *namesp;
 
 		if (be32_to_cpu(entry->hashval) != cursor->hashval) {
 			cursor->hashval = be32_to_cpu(entry->hashval);
@@ -2366,115 +2371,69 @@ xfs_attr_leaf_list_int(xfs_dabuf_t *bp, xfs_attr_list_context_t *context)
 
 		if (entry->flags & XFS_ATTR_INCOMPLETE)
 			continue;		/* skip incomplete entries */
-		if (((context->flags & ATTR_SECURE) != 0) !=
-		    ((entry->flags & XFS_ATTR_SECURE) != 0) &&
-		    !(context->flags & ATTR_KERNORMALS))
-			continue;		/* skip non-matching entries */
-		if (((context->flags & ATTR_ROOT) != 0) !=
-		    ((entry->flags & XFS_ATTR_ROOT) != 0) &&
-		    !(context->flags & ATTR_KERNROOTLS))
-			continue;		/* skip non-matching entries */
-
-		namesp = (entry->flags & XFS_ATTR_SECURE) ? &attr_secure :
-			((entry->flags & XFS_ATTR_ROOT) ? &attr_trusted :
-			  &attr_user);
+		if (!xfs_attr_namesp_match_overrides(context->flags, entry->flags))
+			continue;
+
+		namesp = xfs_attr_flags_namesp(entry->flags);
 
 		if (entry->flags & XFS_ATTR_LOCAL) {
-			name_loc = XFS_ATTR_LEAF_NAME_LOCAL(leaf, i);
-			if (context->flags & ATTR_KERNOVAL) {
-				ASSERT(context->flags & ATTR_KERNAMELS);
-				context->count += namesp->attr_namelen +
-						(int)name_loc->namelen + 1;
-			} else {
-				retval = xfs_attr_put_listent(context, namesp,
-					(char *)name_loc->nameval,
-					(int)name_loc->namelen,
-					be16_to_cpu(name_loc->valuelen));
-			}
+			xfs_attr_leaf_name_local_t *name_loc =
+				XFS_ATTR_LEAF_NAME_LOCAL(leaf, i);
+
+			retval = context->put_listent(context,
+						namesp,
+						(char *)name_loc->nameval,
+						(int)name_loc->namelen,
+						be16_to_cpu(name_loc->valuelen),
+						(char *)&name_loc->nameval[name_loc->namelen]);
+			if (retval)
+				return retval;
 		} else {
-			name_rmt = XFS_ATTR_LEAF_NAME_REMOTE(leaf, i);
-			if (context->flags & ATTR_KERNOVAL) {
-				ASSERT(context->flags & ATTR_KERNAMELS);
-				context->count += namesp->attr_namelen +
-						(int)name_rmt->namelen + 1;
-			} else {
-				retval = xfs_attr_put_listent(context, namesp,
-					(char *)name_rmt->name,
-					(int)name_rmt->namelen,
-					be32_to_cpu(name_rmt->valuelen));
+			xfs_attr_leaf_name_remote_t *name_rmt =
+				XFS_ATTR_LEAF_NAME_REMOTE(leaf, i);
+
+			int valuelen = be32_to_cpu(name_rmt->valuelen);
+
+			if (context->put_value) {
+				xfs_da_args_t args;
+
+				memset((char *)&args, 0, sizeof(args));
+				args.dp = context->dp;
+				args.whichfork = XFS_ATTR_FORK;
+				args.valuelen = valuelen;
+				args.value = kmem_alloc(valuelen, KM_SLEEP);
+				args.rmtblkno = be32_to_cpu(name_rmt->valueblk);
+				args.rmtblkcnt = XFS_B_TO_FSB(args.dp->i_mount, valuelen);
+				retval = xfs_attr_rmtval_get(&args);
+				if (retval)
+					return retval;
+				retval = context->put_listent(context,
+						namesp,
+						(char *)name_rmt->name,
+						(int)name_rmt->namelen,
+						valuelen,
+						(char*)args.value);
+				kmem_free(args.value, valuelen);
 			}
+			else {
+				retval = context->put_listent(context,
+						namesp,
+						(char *)name_rmt->name,
+						(int)name_rmt->namelen,
+						valuelen,
+						NULL);
+			}
+			if (retval)
+				return retval;
 		}
-		if (retval == 0) {
-			cursor->offset++;
-		}
+		if (context->seen_enough)
+			break;
+		cursor->offset++;
 	}
 	xfs_attr_trace_l_cl("blk end", context, leaf);
 	return(retval);
 }
 
-#define	ATTR_ENTBASESIZE		/* minimum bytes used by an attr */ \
-	(((struct attrlist_ent *) 0)->a_name - (char *) 0)
-#define	ATTR_ENTSIZE(namelen)		/* actual bytes used by an attr */ \
-	((ATTR_ENTBASESIZE + (namelen) + 1 + sizeof(u_int32_t)-1) \
-	 & ~(sizeof(u_int32_t)-1))
-
-/*
- * Format an attribute and copy it out to the user's buffer.
- * Take care to check values and protect against them changing later,
- * we may be reading them directly out of a user buffer.
- */
-/*ARGSUSED*/
-STATIC int
-xfs_attr_put_listent(xfs_attr_list_context_t *context,
-		     attrnames_t *namesp, char *name, int namelen, int valuelen)
-{
-	attrlist_ent_t *aep;
-	int arraytop;
-
-	ASSERT(!(context->flags & ATTR_KERNOVAL));
-	if (context->flags & ATTR_KERNAMELS) {
-		char *offset;
-
-		ASSERT(context->count >= 0);
-
-		arraytop = context->count + namesp->attr_namelen + namelen + 1;
-		if (arraytop > context->firstu) {
-			context->count = -1;	/* insufficient space */
-			return(1);
-		}
-		offset = (char *)context->alist + context->count;
-		strncpy(offset, namesp->attr_name, namesp->attr_namelen);
-		offset += namesp->attr_namelen;
-		strncpy(offset, name, namelen);			/* real name */
-		offset += namelen;
-		*offset = '\0';
-		context->count += namesp->attr_namelen + namelen + 1;
-		return(0);
-	}
-
-	ASSERT(context->count >= 0);
-	ASSERT(context->count < (ATTR_MAX_VALUELEN/8));
-	ASSERT(context->firstu >= sizeof(*context->alist));
-	ASSERT(context->firstu <= context->bufsize);
-
-	arraytop = sizeof(*context->alist) +
-			context->count * sizeof(context->alist->al_offset[0]);
-	context->firstu -= ATTR_ENTSIZE(namelen);
-	if (context->firstu < arraytop) {
-		xfs_attr_trace_l_c("buffer full", context);
-		context->alist->al_more = 1;
-		return(1);
-	}
-
-	aep = (attrlist_ent_t *)&(((char *)context->alist)[ context->firstu ]);
-	aep->a_valuelen = valuelen;
-	memcpy(aep->a_name, name, namelen);
-	aep->a_name[ namelen ] = 0;
-	context->alist->al_offset[ context->count++ ] = context->firstu;
-	context->alist->al_count = context->count;
-	xfs_attr_trace_l_c("add", context);
-	return(0);
-}
 
 /*========================================================================
  * Manage the INCOMPLETE flag in a leaf entry

fs/xfs/xfs_attr_leaf.h

@@ -129,6 +129,19 @@ typedef struct xfs_attr_leafblock {
 #define XFS_ATTR_SECURE		(1 << XFS_ATTR_SECURE_BIT)
 #define XFS_ATTR_INCOMPLETE	(1 << XFS_ATTR_INCOMPLETE_BIT)
 
+/*
+ * Conversion macros for converting namespace bits from argument flags
+ * to ondisk flags.
+ */
+#define XFS_ATTR_NSP_ARGS_MASK		(ATTR_ROOT | ATTR_SECURE)
+#define XFS_ATTR_NSP_ONDISK_MASK	(XFS_ATTR_ROOT | XFS_ATTR_SECURE)
+#define XFS_ATTR_NSP_ONDISK(flags)	((flags) & XFS_ATTR_NSP_ONDISK_MASK)
+#define XFS_ATTR_NSP_ARGS(flags)	((flags) & XFS_ATTR_NSP_ARGS_MASK)
+#define XFS_ATTR_NSP_ARGS_TO_ONDISK(x)	(((x) & ATTR_ROOT ? XFS_ATTR_ROOT : 0) |\
+					 ((x) & ATTR_SECURE ? XFS_ATTR_SECURE : 0))
+#define XFS_ATTR_NSP_ONDISK_TO_ARGS(x)	(((x) & XFS_ATTR_ROOT ? ATTR_ROOT : 0) |\
+					 ((x) & XFS_ATTR_SECURE ? ATTR_SECURE : 0))
+
 /*
  * Alignment for namelist and valuelist entries (since they are mixed
  * there can be only one alignment value)
@@ -196,16 +209,26 @@ static inline int xfs_attr_leaf_entsize_local_max(int bsize)
  * Structure used to pass context around among the routines.
  *========================================================================*/
 
+
+struct xfs_attr_list_context;
+
+typedef int (*put_listent_func_t)(struct xfs_attr_list_context *, struct attrnames *,
+				      char *, int, int, char *);
+
 typedef struct xfs_attr_list_context {
-	struct xfs_inode		*dp;	/* inode */
-	struct attrlist_cursor_kern	*cursor;/* position in list */
-	struct attrlist			*alist;	/* output buffer */
-	int				count;	/* num used entries */
-	int				dupcnt;	/* count dup hashvals seen */
-	int				bufsize;/* total buffer size */
-	int				firstu;	/* first used byte in buffer */
-	int				flags;	/* from VOP call */
-	int				resynch;/* T/F: resynch with cursor */
+	struct xfs_inode		*dp;		/* inode */
+	struct attrlist_cursor_kern	*cursor;	/* position in list */
+	struct attrlist			*alist;		/* output buffer */
+	int				seen_enough;	/* T/F: seen enough of list? */
+	int				count;		/* num used entries */
+	int				dupcnt;		/* count dup hashvals seen */
+	int				bufsize;	/* total buffer size */
+	int				firstu;		/* first used byte in buffer */
+	int				flags;		/* from VOP call */
+	int				resynch;	/* T/F: resynch with cursor */
+	int				put_value;	/* T/F: need value for listent */
+	put_listent_func_t		put_listent;	/* list output fmt function */
+	int				index;		/* index into output buffer */
 } xfs_attr_list_context_t;
 
 /*

fs/xfs/xfs_behavior.c

@@ -109,26 +109,6 @@ bhv_remove_not_first(bhv_head_t *bhp, bhv_desc_t *bdp)
 	prev->bd_next = bdp->bd_next;	/* remove from after prev */
 }
 
-/*
- * Look for a specific ops vector on the specified behavior chain.
- * Return the associated behavior descriptor.  Or NULL, if not found.
- */
-bhv_desc_t *
-bhv_lookup(bhv_head_t *bhp, void *ops)
-{
-	bhv_desc_t	*curdesc;
-
-	for (curdesc = bhp->bh_first;
-	     curdesc != NULL;
-	     curdesc = curdesc->bd_next) {
-
-		if (curdesc->bd_ops == ops)
-			return curdesc;
-	}
-
-	return NULL;
-}
-
 /*
  * Looks for the first behavior within a specified range of positions.
  * Return the associated behavior descriptor.  Or NULL, if none found.

fs/xfs/xfs_behavior.h

@@ -176,12 +176,10 @@ extern void bhv_insert_initial(bhv_head_t *, bhv_desc_t *);
  * Behavior module prototypes.
  */
 extern void		bhv_remove_not_first(bhv_head_t *bhp, bhv_desc_t *bdp);
-extern bhv_desc_t *	bhv_lookup(bhv_head_t *bhp, void *ops);
 extern bhv_desc_t *	bhv_lookup_range(bhv_head_t *bhp, int low, int high);
 extern bhv_desc_t *	bhv_base(bhv_head_t *bhp);
 
 /* No bhv locking on Linux */
-#define bhv_lookup_unlocked	bhv_lookup
 #define bhv_base_unlocked	bhv_base
 
 #endif /* __XFS_BEHAVIOR_H__ */

fs/xfs/xfs_bmap.c

@@ -2999,7 +2999,7 @@ xfs_bmap_btree_to_extents(
 	int			error;	/* error return value */
 	xfs_ifork_t		*ifp;	/* inode fork data */
 	xfs_mount_t		*mp;	/* mount point structure */
-	xfs_bmbt_ptr_t		*pp;	/* ptr to block address */
+	__be64			*pp;	/* ptr to block address */
 	xfs_bmbt_block_t	*rblock;/* root btree block */
 
 	ifp = XFS_IFORK_PTR(ip, whichfork);
@@ -3011,12 +3011,12 @@ xfs_bmap_btree_to_extents(
 	ASSERT(XFS_BMAP_BROOT_MAXRECS(ifp->if_broot_bytes) == 1);
 	mp = ip->i_mount;
 	pp = XFS_BMAP_BROOT_PTR_ADDR(rblock, 1, ifp->if_broot_bytes);
+	cbno = be64_to_cpu(*pp);
 	*logflagsp = 0;
 #ifdef DEBUG
-	if ((error = xfs_btree_check_lptr(cur, INT_GET(*pp, ARCH_CONVERT), 1)))
+	if ((error = xfs_btree_check_lptr(cur, cbno, 1)))
 		return error;
 #endif
-	cbno = INT_GET(*pp, ARCH_CONVERT);
 	if ((error = xfs_btree_read_bufl(mp, tp, cbno, 0, &cbp,
 			XFS_BMAP_BTREE_REF)))
 		return error;
@@ -3512,9 +3512,9 @@ xfs_bmap_extents_to_btree(
 	 */
 	kp = XFS_BMAP_KEY_IADDR(block, 1, cur);
 	arp = XFS_BMAP_REC_IADDR(ablock, 1, cur);
-	INT_SET(kp->br_startoff, ARCH_CONVERT, xfs_bmbt_disk_get_startoff(arp));
+	kp->br_startoff = cpu_to_be64(xfs_bmbt_disk_get_startoff(arp));
 	pp = XFS_BMAP_PTR_IADDR(block, 1, cur);
-	INT_SET(*pp, ARCH_CONVERT, args.fsbno);
+	*pp = cpu_to_be64(args.fsbno);
 	/*
 	 * Do all this logging at the end so that
 	 * the root is at the right level.
@@ -3705,7 +3705,7 @@ STATIC xfs_bmbt_rec_t *                 /* pointer to found extent entry */
 xfs_bmap_search_extents(
 	xfs_inode_t     *ip,            /* incore inode pointer */
 	xfs_fileoff_t   bno,            /* block number searched for */
-	int             whichfork,      /* data or attr fork */
+	int             fork,      	/* data or attr fork */
 	int             *eofp,          /* out: end of file found */
 	xfs_extnum_t    *lastxp,        /* out: last extent index */
 	xfs_bmbt_irec_t *gotp,          /* out: extent entry found */
@@ -3713,25 +3713,28 @@ xfs_bmap_search_extents(
 {
 	xfs_ifork_t	*ifp;		/* inode fork pointer */
 	xfs_bmbt_rec_t  *ep;            /* extent record pointer */
-	int		rt;		/* realtime flag    */
 
 	XFS_STATS_INC(xs_look_exlist);
-	ifp = XFS_IFORK_PTR(ip, whichfork);
+	ifp = XFS_IFORK_PTR(ip, fork);
 
 	ep = xfs_bmap_search_multi_extents(ifp, bno, eofp, lastxp, gotp, prevp);
 
-	rt = (whichfork == XFS_DATA_FORK) && XFS_IS_REALTIME_INODE(ip);
-	if (unlikely(!rt && !gotp->br_startblock && (*lastxp != NULLEXTNUM))) {
-                cmn_err(CE_PANIC,"Access to block zero: fs: <%s> inode: %lld "
-			"start_block : %llx start_off : %llx blkcnt : %llx "
-			"extent-state : %x \n",
-			(ip->i_mount)->m_fsname, (long long)ip->i_ino,
+	if (unlikely(!(gotp->br_startblock) && (*lastxp != NULLEXTNUM) &&
+		     !(XFS_IS_REALTIME_INODE(ip) && fork == XFS_DATA_FORK))) {
+		xfs_cmn_err(XFS_PTAG_FSBLOCK_ZERO, CE_ALERT, ip->i_mount,
+				"Access to block zero in inode %llu "
+				"start_block: %llx start_off: %llx "
+				"blkcnt: %llx extent-state: %x lastx: %x\n",
+			(unsigned long long)ip->i_ino,
 			(unsigned long long)gotp->br_startblock,
 			(unsigned long long)gotp->br_startoff,
 			(unsigned long long)gotp->br_blockcount,
-			gotp->br_state);
-        }
-        return ep;
+			gotp->br_state, *lastxp);
+		*lastxp = NULLEXTNUM;
+		*eofp = 1;
+		return NULL;
+	}
+	return ep;
 }
 
 
@@ -4494,7 +4497,7 @@ xfs_bmap_read_extents(
 	xfs_ifork_t		*ifp;	/* fork structure */
 	int			level;	/* btree level, for checking */
 	xfs_mount_t		*mp;	/* file system mount structure */
-	xfs_bmbt_ptr_t		*pp;	/* pointer to block address */
+	__be64			*pp;	/* pointer to block address */
 	/* REFERENCED */
 	xfs_extnum_t		room;	/* number of entries there's room for */
 
@@ -4510,10 +4513,10 @@ xfs_bmap_read_extents(
 	level = be16_to_cpu(block->bb_level);
 	ASSERT(level > 0);
 	pp = XFS_BMAP_BROOT_PTR_ADDR(block, 1, ifp->if_broot_bytes);
-	ASSERT(INT_GET(*pp, ARCH_CONVERT) != NULLDFSBNO);
-	ASSERT(XFS_FSB_TO_AGNO(mp, INT_GET(*pp, ARCH_CONVERT)) < mp->m_sb.sb_agcount);
-	ASSERT(XFS_FSB_TO_AGBNO(mp, INT_GET(*pp, ARCH_CONVERT)) < mp->m_sb.sb_agblocks);
-	bno = INT_GET(*pp, ARCH_CONVERT);
+	bno = be64_to_cpu(*pp);
+	ASSERT(bno != NULLDFSBNO);
+	ASSERT(XFS_FSB_TO_AGNO(mp, bno) < mp->m_sb.sb_agcount);
+	ASSERT(XFS_FSB_TO_AGBNO(mp, bno) < mp->m_sb.sb_agblocks);
 	/*
 	 * Go down the tree until leaf level is reached, following the first
 	 * pointer (leftmost) at each level.
@@ -4530,10 +4533,8 @@ xfs_bmap_read_extents(
 			break;
 		pp = XFS_BTREE_PTR_ADDR(mp->m_sb.sb_blocksize, xfs_bmbt, block,
 			1, mp->m_bmap_dmxr[1]);
-		XFS_WANT_CORRUPTED_GOTO(
-			XFS_FSB_SANITY_CHECK(mp, INT_GET(*pp, ARCH_CONVERT)),
-			error0);
-		bno = INT_GET(*pp, ARCH_CONVERT);
+		bno = be64_to_cpu(*pp);
+		XFS_WANT_CORRUPTED_GOTO(XFS_FSB_SANITY_CHECK(mp, bno), error0);
 		xfs_trans_brelse(tp, bp);
 	}
 	/*
@@ -6141,7 +6142,7 @@ xfs_check_block(
 	short			sz)
 {
 	int			i, j, dmxr;
-	xfs_bmbt_ptr_t		*pp, *thispa;	/* pointer to block address */
+	__be64			*pp, *thispa;	/* pointer to block address */
 	xfs_bmbt_key_t		*prevp, *keyp;
 
 	ASSERT(be16_to_cpu(block->bb_level) > 0);
@@ -6179,11 +6180,10 @@ xfs_check_block(
 				thispa = XFS_BTREE_PTR_ADDR(mp->m_sb.sb_blocksize,
 					xfs_bmbt, block, j, dmxr);
 			}
-			if (INT_GET(*thispa, ARCH_CONVERT) ==
-			    INT_GET(*pp, ARCH_CONVERT)) {
+			if (*thispa == *pp) {
 				cmn_err(CE_WARN, "%s: thispa(%d) == pp(%d) %Ld",
 					__FUNCTION__, j, i,
-					INT_GET(*thispa, ARCH_CONVERT));
+					(unsigned long long)be64_to_cpu(*thispa));
 				panic("%s: ptrs are equal in node\n",
 					__FUNCTION__);
 			}
@@ -6210,7 +6210,7 @@ xfs_bmap_check_leaf_extents(
 	xfs_ifork_t		*ifp;	/* fork structure */
 	int			level;	/* btree level, for checking */
 	xfs_mount_t		*mp;	/* file system mount structure */
-	xfs_bmbt_ptr_t		*pp;	/* pointer to block address */
+	__be64			*pp;	/* pointer to block address */
 	xfs_bmbt_rec_t		*ep;	/* pointer to current extent */
 	xfs_bmbt_rec_t		*lastp; /* pointer to previous extent */
 	xfs_bmbt_rec_t		*nextp;	/* pointer to next extent */
@@ -6231,10 +6231,12 @@ xfs_bmap_check_leaf_extents(
 	ASSERT(level > 0);
 	xfs_check_block(block, mp, 1, ifp->if_broot_bytes);
 	pp = XFS_BMAP_BROOT_PTR_ADDR(block, 1, ifp->if_broot_bytes);
-	ASSERT(INT_GET(*pp, ARCH_CONVERT) != NULLDFSBNO);
-	ASSERT(XFS_FSB_TO_AGNO(mp, INT_GET(*pp, ARCH_CONVERT)) < mp->m_sb.sb_agcount);
-	ASSERT(XFS_FSB_TO_AGBNO(mp, INT_GET(*pp, ARCH_CONVERT)) < mp->m_sb.sb_agblocks);
-	bno = INT_GET(*pp, ARCH_CONVERT);
+	bno = be64_to_cpu(*pp);
+
+	ASSERT(bno != NULLDFSBNO);
+	ASSERT(XFS_FSB_TO_AGNO(mp, bno) < mp->m_sb.sb_agcount);
+	ASSERT(XFS_FSB_TO_AGBNO(mp, bno) < mp->m_sb.sb_agblocks);
+
 	/*
 	 * Go down the tree until leaf level is reached, following the first
 	 * pointer (leftmost) at each level.
@@ -6265,8 +6267,8 @@ xfs_bmap_check_leaf_extents(
 		xfs_check_block(block, mp, 0, 0);
 		pp = XFS_BTREE_PTR_ADDR(mp->m_sb.sb_blocksize, xfs_bmbt, block,
 			1, mp->m_bmap_dmxr[1]);
-		XFS_WANT_CORRUPTED_GOTO(XFS_FSB_SANITY_CHECK(mp, INT_GET(*pp, ARCH_CONVERT)), error0);
-		bno = INT_GET(*pp, ARCH_CONVERT);
+		bno = be64_to_cpu(*pp);
+		XFS_WANT_CORRUPTED_GOTO(XFS_FSB_SANITY_CHECK(mp, bno), error0);
 		if (bp_release) {
 			bp_release = 0;
 			xfs_trans_brelse(NULL, bp);
@@ -6372,7 +6374,7 @@ xfs_bmap_count_blocks(
 	xfs_ifork_t		*ifp;	/* fork structure */
 	int			level;	/* btree level, for checking */
 	xfs_mount_t		*mp;	/* file system mount structure */
-	xfs_bmbt_ptr_t		*pp;	/* pointer to block address */
+	__be64			*pp;	/* pointer to block address */
 
 	bno = NULLFSBLOCK;
 	mp = ip->i_mount;
@@ -6395,10 +6397,10 @@ xfs_bmap_count_blocks(
 	level = be16_to_cpu(block->bb_level);
 	ASSERT(level > 0);
 	pp = XFS_BMAP_BROOT_PTR_ADDR(block, 1, ifp->if_broot_bytes);
-	ASSERT(INT_GET(*pp, ARCH_CONVERT) != NULLDFSBNO);
-	ASSERT(XFS_FSB_TO_AGNO(mp, INT_GET(*pp, ARCH_CONVERT)) < mp->m_sb.sb_agcount);
-	ASSERT(XFS_FSB_TO_AGBNO(mp, INT_GET(*pp, ARCH_CONVERT)) < mp->m_sb.sb_agblocks);
-	bno = INT_GET(*pp, ARCH_CONVERT);
+	bno = be64_to_cpu(*pp);
+	ASSERT(bno != NULLDFSBNO);
+	ASSERT(XFS_FSB_TO_AGNO(mp, bno) < mp->m_sb.sb_agcount);
+	ASSERT(XFS_FSB_TO_AGBNO(mp, bno) < mp->m_sb.sb_agblocks);
 
 	if (unlikely(xfs_bmap_count_tree(mp, tp, ifp, bno, level, count) < 0)) {
 		XFS_ERROR_REPORT("xfs_bmap_count_blocks(2)", XFS_ERRLEVEL_LOW,
@@ -6425,7 +6427,7 @@ xfs_bmap_count_tree(
 	int			error;
 	xfs_buf_t		*bp, *nbp;
 	int			level = levelin;
-	xfs_bmbt_ptr_t          *pp;
+	__be64			*pp;
 	xfs_fsblock_t           bno = blockno;
 	xfs_fsblock_t		nextbno;
 	xfs_bmbt_block_t        *block, *nextblock;
@@ -6452,7 +6454,7 @@ xfs_bmap_count_tree(
 		/* Dive to the next level */
 		pp = XFS_BTREE_PTR_ADDR(mp->m_sb.sb_blocksize,
 			xfs_bmbt, block, 1, mp->m_bmap_dmxr[1]);
-		bno = INT_GET(*pp, ARCH_CONVERT);
+		bno = be64_to_cpu(*pp);
 		if (unlikely((error =
 		     xfs_bmap_count_tree(mp, tp, ifp, bno, level, count)) < 0)) {
 			xfs_trans_brelse(tp, bp);

fs/xfs/xfs_bmap_btree.c

@@ -58,7 +58,7 @@ STATIC void xfs_bmbt_log_ptrs(xfs_btree_cur_t *, xfs_buf_t *, int, int);
 STATIC int xfs_bmbt_lshift(xfs_btree_cur_t *, int, int *);
 STATIC int xfs_bmbt_rshift(xfs_btree_cur_t *, int, int *);
 STATIC int xfs_bmbt_split(xfs_btree_cur_t *, int, xfs_fsblock_t *,
-		xfs_bmbt_key_t *, xfs_btree_cur_t **, int *);
+		__uint64_t *, xfs_btree_cur_t **, int *);
 STATIC int xfs_bmbt_updkey(xfs_btree_cur_t *, xfs_bmbt_key_t *, int);
 
 
@@ -192,16 +192,11 @@ xfs_bmbt_trace_argifk(
 	xfs_btree_cur_t		*cur,
 	int			i,
 	xfs_fsblock_t		f,
-	xfs_bmbt_key_t		*k,
+	xfs_dfiloff_t		o,
 	int			line)
 {
-	xfs_dfsbno_t		d;
-	xfs_dfiloff_t		o;
-
-	d = (xfs_dfsbno_t)f;
-	o = INT_GET(k->br_startoff, ARCH_CONVERT);
 	xfs_bmbt_trace_enter(func, cur, ARGS, XFS_BMBT_KTRACE_ARGIFK, line,
-		i, d >> 32, (int)d, o >> 32,
+		i, (xfs_dfsbno_t)f >> 32, (int)f, o >> 32,
 		(int)o, 0, 0, 0,
 		0, 0, 0);
 }
@@ -248,7 +243,7 @@ xfs_bmbt_trace_argik(
 {
 	xfs_dfiloff_t		o;
 
-	o = INT_GET(k->br_startoff, ARCH_CONVERT);
+	o = be64_to_cpu(k->br_startoff);
 	xfs_bmbt_trace_enter(func, cur, ARGS, XFS_BMBT_KTRACE_ARGIFK, line,
 		i, o >> 32, (int)o, 0,
 		0, 0, 0, 0,
@@ -286,8 +281,8 @@ xfs_bmbt_trace_cursor(
 	xfs_bmbt_trace_argfffi(fname, c, o, b, i, j, __LINE__)
 #define	XFS_BMBT_TRACE_ARGI(c,i)	\
 	xfs_bmbt_trace_argi(fname, c, i, __LINE__)
-#define	XFS_BMBT_TRACE_ARGIFK(c,i,f,k)	\
-	xfs_bmbt_trace_argifk(fname, c, i, f, k, __LINE__)
+#define	XFS_BMBT_TRACE_ARGIFK(c,i,f,s)	\
+	xfs_bmbt_trace_argifk(fname, c, i, f, s, __LINE__)
 #define	XFS_BMBT_TRACE_ARGIFR(c,i,f,r)	\
 	xfs_bmbt_trace_argifr(fname, c, i, f, r, __LINE__)
 #define	XFS_BMBT_TRACE_ARGIK(c,i,k)	\
@@ -299,7 +294,7 @@ xfs_bmbt_trace_cursor(
 #define	XFS_BMBT_TRACE_ARGBII(c,b,i,j)
 #define	XFS_BMBT_TRACE_ARGFFFI(c,o,b,i,j)
 #define	XFS_BMBT_TRACE_ARGI(c,i)
-#define	XFS_BMBT_TRACE_ARGIFK(c,i,f,k)
+#define	XFS_BMBT_TRACE_ARGIFK(c,i,f,s)
 #define	XFS_BMBT_TRACE_ARGIFR(c,i,f,r)
 #define	XFS_BMBT_TRACE_ARGIK(c,i,k)
 #define	XFS_BMBT_TRACE_CURSOR(c,s)
@@ -357,7 +352,7 @@ xfs_bmbt_delrec(
 	XFS_BMBT_TRACE_CURSOR(cur, ENTRY);
 	XFS_BMBT_TRACE_ARGI(cur, level);
 	ptr = cur->bc_ptrs[level];
-	tcur = (xfs_btree_cur_t *)0;
+	tcur = NULL;
 	if (ptr == 0) {
 		XFS_BMBT_TRACE_CURSOR(cur, EXIT);
 		*stat = 0;
@@ -382,7 +377,7 @@ xfs_bmbt_delrec(
 		pp = XFS_BMAP_PTR_IADDR(block, 1, cur);
 #ifdef DEBUG
 		for (i = ptr; i < numrecs; i++) {
-			if ((error = xfs_btree_check_lptr(cur, INT_GET(pp[i], ARCH_CONVERT), level))) {
+			if ((error = xfs_btree_check_lptr_disk(cur, pp[i], level))) {
 				XFS_BMBT_TRACE_CURSOR(cur, ERROR);
 				goto error0;
 			}
@@ -404,7 +399,8 @@ xfs_bmbt_delrec(
 			xfs_bmbt_log_recs(cur, bp, ptr, numrecs - 1);
 		}
 		if (ptr == 1) {
-			INT_SET(key.br_startoff, ARCH_CONVERT, xfs_bmbt_disk_get_startoff(rp));
+			key.br_startoff =
+				cpu_to_be64(xfs_bmbt_disk_get_startoff(rp));
 			kp = &key;
 		}
 	}
@@ -621,7 +617,7 @@ xfs_bmbt_delrec(
 		rpp = XFS_BMAP_PTR_IADDR(right, 1, cur);
 #ifdef DEBUG
 		for (i = 0; i < numrrecs; i++) {
-			if ((error = xfs_btree_check_lptr(cur, INT_GET(rpp[i], ARCH_CONVERT), level))) {
+			if ((error = xfs_btree_check_lptr_disk(cur, rpp[i], level))) {
 				XFS_BMBT_TRACE_CURSOR(cur, ERROR);
 				goto error0;
 			}
@@ -748,7 +744,7 @@ xfs_bmbt_insrec(
 	int			logflags;	/* inode logging flags */
 	xfs_fsblock_t		nbno;		/* new block number */
 	struct xfs_btree_cur	*ncur;		/* new btree cursor */
-	xfs_bmbt_key_t		nkey;		/* new btree key value */
+	__uint64_t		startoff;	/* new btree key value */
 	xfs_bmbt_rec_t		nrec;		/* new record count */
 	int			optr;		/* old key/record index */
 	xfs_bmbt_ptr_t		*pp;		/* pointer to bmap block addr */
@@ -759,9 +755,8 @@ xfs_bmbt_insrec(
 	ASSERT(level < cur->bc_nlevels);
 	XFS_BMBT_TRACE_CURSOR(cur, ENTRY);
 	XFS_BMBT_TRACE_ARGIFR(cur, level, *bnop, recp);
-	ncur = (xfs_btree_cur_t *)0;
-	INT_SET(key.br_startoff, ARCH_CONVERT,
-		xfs_bmbt_disk_get_startoff(recp));
+	ncur = NULL;
+	key.br_startoff = cpu_to_be64(xfs_bmbt_disk_get_startoff(recp));
 	optr = ptr = cur->bc_ptrs[level];
 	if (ptr == 0) {
 		XFS_BMBT_TRACE_CURSOR(cur, EXIT);
@@ -820,7 +815,7 @@ xfs_bmbt_insrec(
 					optr = ptr = cur->bc_ptrs[level];
 				} else {
 					if ((error = xfs_bmbt_split(cur, level,
-							&nbno, &nkey, &ncur,
+							&nbno, &startoff, &ncur,
 							&i))) {
 						XFS_BMBT_TRACE_CURSOR(cur,
 							ERROR);
@@ -840,7 +835,7 @@ xfs_bmbt_insrec(
 #endif
 						ptr = cur->bc_ptrs[level];
 						xfs_bmbt_disk_set_allf(&nrec,
-							nkey.br_startoff, 0, 0,
+							startoff, 0, 0,
 							XFS_EXT_NORM);
 					} else {
 						XFS_BMBT_TRACE_CURSOR(cur,
@@ -858,7 +853,7 @@ xfs_bmbt_insrec(
 		pp = XFS_BMAP_PTR_IADDR(block, 1, cur);
 #ifdef DEBUG
 		for (i = numrecs; i >= ptr; i--) {
-			if ((error = xfs_btree_check_lptr(cur, INT_GET(pp[i - 1], ARCH_CONVERT),
+			if ((error = xfs_btree_check_lptr_disk(cur, pp[i - 1],
 					level))) {
 				XFS_BMBT_TRACE_CURSOR(cur, ERROR);
 				return error;
@@ -870,14 +865,13 @@ xfs_bmbt_insrec(
 		memmove(&pp[ptr], &pp[ptr - 1], /* INT_: direct copy */
 			(numrecs - ptr + 1) * sizeof(*pp));
 #ifdef DEBUG
-		if ((error = xfs_btree_check_lptr(cur, (xfs_bmbt_ptr_t)*bnop,
-				level))) {
+		if ((error = xfs_btree_check_lptr(cur, *bnop, level))) {
 			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
 			return error;
 		}
 #endif
 		kp[ptr - 1] = key;
-		INT_SET(pp[ptr - 1], ARCH_CONVERT, *bnop);
+		pp[ptr - 1] = cpu_to_be64(*bnop);
 		numrecs++;
 		block->bb_numrecs = cpu_to_be16(numrecs);
 		xfs_bmbt_log_keys(cur, bp, ptr, numrecs);
@@ -988,7 +982,7 @@ xfs_bmbt_killroot(
 	cpp = XFS_BMAP_PTR_IADDR(cblock, 1, cur);
 #ifdef DEBUG
 	for (i = 0; i < be16_to_cpu(cblock->bb_numrecs); i++) {
-		if ((error = xfs_btree_check_lptr(cur, INT_GET(cpp[i], ARCH_CONVERT), level - 1))) {
+		if ((error = xfs_btree_check_lptr_disk(cur, cpp[i], level - 1))) {
 			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
 			return error;
 		}
@@ -1132,7 +1126,7 @@ xfs_bmbt_lookup(
 			d = XFS_FSB_TO_DADDR(mp, fsbno);
 			bp = cur->bc_bufs[level];
 			if (bp && XFS_BUF_ADDR(bp) != d)
-				bp = (xfs_buf_t *)0;
+				bp = NULL;
 			if (!bp) {
 				if ((error = xfs_btree_read_bufl(mp, tp, fsbno,
 						0, &bp, XFS_BMAP_BTREE_REF))) {
@@ -1170,7 +1164,7 @@ xfs_bmbt_lookup(
 				keyno = (low + high) >> 1;
 				if (level > 0) {
 					kkp = kkbase + keyno - 1;
-					startoff = INT_GET(kkp->br_startoff, ARCH_CONVERT);
+					startoff = be64_to_cpu(kkp->br_startoff);
 				} else {
 					krp = krbase + keyno - 1;
 					startoff = xfs_bmbt_disk_get_startoff(krp);
@@ -1189,13 +1183,13 @@ xfs_bmbt_lookup(
 			if (diff > 0 && --keyno < 1)
 				keyno = 1;
 			pp = XFS_BMAP_PTR_IADDR(block, keyno, cur);
+			fsbno = be64_to_cpu(*pp);
 #ifdef DEBUG
-			if ((error = xfs_btree_check_lptr(cur, INT_GET(*pp, ARCH_CONVERT), level))) {
+			if ((error = xfs_btree_check_lptr(cur, fsbno, level))) {
 				XFS_BMBT_TRACE_CURSOR(cur, ERROR);
 				return error;
 			}
 #endif
-			fsbno = INT_GET(*pp, ARCH_CONVERT);
 			cur->bc_ptrs[level] = keyno;
 		}
 	}
@@ -1313,7 +1307,7 @@ xfs_bmbt_lshift(
 		lpp = XFS_BMAP_PTR_IADDR(left, lrecs, cur);
 		rpp = XFS_BMAP_PTR_IADDR(right, 1, cur);
 #ifdef DEBUG
-		if ((error = xfs_btree_check_lptr(cur, INT_GET(*rpp, ARCH_CONVERT), level))) {
+		if ((error = xfs_btree_check_lptr_disk(cur, *rpp, level))) {
 			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
 			return error;
 		}
@@ -1340,7 +1334,7 @@ xfs_bmbt_lshift(
 	if (level > 0) {
 #ifdef DEBUG
 		for (i = 0; i < rrecs; i++) {
-			if ((error = xfs_btree_check_lptr(cur, INT_GET(rpp[i + 1], ARCH_CONVERT),
+			if ((error = xfs_btree_check_lptr_disk(cur, rpp[i + 1],
 					level))) {
 				XFS_BMBT_TRACE_CURSOR(cur, ERROR);
 				return error;
@@ -1354,8 +1348,7 @@ xfs_bmbt_lshift(
 	} else {
 		memmove(rrp, rrp + 1, rrecs * sizeof(*rrp));
 		xfs_bmbt_log_recs(cur, rbp, 1, rrecs);
-		INT_SET(key.br_startoff, ARCH_CONVERT,
-			xfs_bmbt_disk_get_startoff(rrp));
+		key.br_startoff = cpu_to_be64(xfs_bmbt_disk_get_startoff(rrp));
 		rkp = &key;
 	}
 	if ((error = xfs_bmbt_updkey(cur, rkp, level + 1))) {
@@ -1445,7 +1438,7 @@ xfs_bmbt_rshift(
 		rpp = XFS_BMAP_PTR_IADDR(right, 1, cur);
 #ifdef DEBUG
 		for (i = be16_to_cpu(right->bb_numrecs) - 1; i >= 0; i--) {
-			if ((error = xfs_btree_check_lptr(cur, INT_GET(rpp[i], ARCH_CONVERT), level))) {
+			if ((error = xfs_btree_check_lptr_disk(cur, rpp[i], level))) {
 				XFS_BMBT_TRACE_CURSOR(cur, ERROR);
 				return error;
 			}
@@ -1454,7 +1447,7 @@ xfs_bmbt_rshift(
 		memmove(rkp + 1, rkp, be16_to_cpu(right->bb_numrecs) * sizeof(*rkp));
 		memmove(rpp + 1, rpp, be16_to_cpu(right->bb_numrecs) * sizeof(*rpp));
 #ifdef DEBUG
-		if ((error = xfs_btree_check_lptr(cur, INT_GET(*lpp, ARCH_CONVERT), level))) {
+		if ((error = xfs_btree_check_lptr_disk(cur, *lpp, level))) {
 			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
 			return error;
 		}
@@ -1469,8 +1462,7 @@ xfs_bmbt_rshift(
 		memmove(rrp + 1, rrp, be16_to_cpu(right->bb_numrecs) * sizeof(*rrp));
 		*rrp = *lrp;
 		xfs_bmbt_log_recs(cur, rbp, 1, be16_to_cpu(right->bb_numrecs) + 1);
-		INT_SET(key.br_startoff, ARCH_CONVERT,
-			xfs_bmbt_disk_get_startoff(rrp));
+		key.br_startoff = cpu_to_be64(xfs_bmbt_disk_get_startoff(rrp));
 		rkp = &key;
 	}
 	be16_add(&left->bb_numrecs, -1);
@@ -1535,7 +1527,7 @@ xfs_bmbt_split(
 	xfs_btree_cur_t		*cur,
 	int			level,
 	xfs_fsblock_t		*bnop,
-	xfs_bmbt_key_t		*keyp,
+	__uint64_t		*startoff,
 	xfs_btree_cur_t		**curp,
 	int			*stat)		/* success/failure */
 {
@@ -1560,7 +1552,7 @@ xfs_bmbt_split(
 	xfs_bmbt_rec_t		*rrp;		/* right record pointer */
 
 	XFS_BMBT_TRACE_CURSOR(cur, ENTRY);
-	XFS_BMBT_TRACE_ARGIFK(cur, level, *bnop, keyp);
+	XFS_BMBT_TRACE_ARGIFK(cur, level, *bnop, *startoff);
 	args.tp = cur->bc_tp;
 	args.mp = cur->bc_mp;
 	lbp = cur->bc_bufs[level];
@@ -1619,7 +1611,7 @@ xfs_bmbt_split(
 		rpp = XFS_BMAP_PTR_IADDR(right, 1, cur);
 #ifdef DEBUG
 		for (i = 0; i < be16_to_cpu(right->bb_numrecs); i++) {
-			if ((error = xfs_btree_check_lptr(cur, INT_GET(lpp[i], ARCH_CONVERT), level))) {
+			if ((error = xfs_btree_check_lptr_disk(cur, lpp[i], level))) {
 				XFS_BMBT_TRACE_CURSOR(cur, ERROR);
 				return error;
 			}
@@ -1629,13 +1621,13 @@ xfs_bmbt_split(
 		memcpy(rpp, lpp, be16_to_cpu(right->bb_numrecs) * sizeof(*rpp));
 		xfs_bmbt_log_keys(cur, rbp, 1, be16_to_cpu(right->bb_numrecs));
 		xfs_bmbt_log_ptrs(cur, rbp, 1, be16_to_cpu(right->bb_numrecs));
-		keyp->br_startoff = INT_GET(rkp->br_startoff, ARCH_CONVERT);
+		*startoff = be64_to_cpu(rkp->br_startoff);
 	} else {
 		lrp = XFS_BMAP_REC_IADDR(left, i, cur);
 		rrp = XFS_BMAP_REC_IADDR(right, 1, cur);
 		memcpy(rrp, lrp, be16_to_cpu(right->bb_numrecs) * sizeof(*rrp));
 		xfs_bmbt_log_recs(cur, rbp, 1, be16_to_cpu(right->bb_numrecs));
-		keyp->br_startoff = xfs_bmbt_disk_get_startoff(rrp);
+		*startoff = xfs_bmbt_disk_get_startoff(rrp);
 	}
 	be16_add(&left->bb_numrecs, -(be16_to_cpu(right->bb_numrecs)));
 	right->bb_rightsib = left->bb_rightsib;
@@ -1728,9 +1720,9 @@ xfs_bmdr_to_bmbt(
 {
 	int			dmxr;
 	xfs_bmbt_key_t		*fkp;
-	xfs_bmbt_ptr_t		*fpp;
+	__be64			*fpp;
 	xfs_bmbt_key_t		*tkp;
-	xfs_bmbt_ptr_t		*tpp;
+	__be64			*tpp;
 
 	rblock->bb_magic = cpu_to_be32(XFS_BMAP_MAGIC);
 	rblock->bb_level = dblock->bb_level;
@@ -1745,7 +1737,7 @@ xfs_bmdr_to_bmbt(
 	tpp = XFS_BMAP_BROOT_PTR_ADDR(rblock, 1, rblocklen);
 	dmxr = be16_to_cpu(dblock->bb_numrecs);
 	memcpy(tkp, fkp, sizeof(*fkp) * dmxr);
-	memcpy(tpp, fpp, sizeof(*fpp) * dmxr); /* INT_: direct copy */
+	memcpy(tpp, fpp, sizeof(*fpp) * dmxr);
 }
 
 /*
@@ -1805,7 +1797,7 @@ xfs_bmbt_decrement(
 	tp = cur->bc_tp;
 	mp = cur->bc_mp;
 	for (block = xfs_bmbt_get_block(cur, lev, &bp); lev > level; ) {
-		fsbno = INT_GET(*XFS_BMAP_PTR_IADDR(block, cur->bc_ptrs[lev], cur), ARCH_CONVERT);
+		fsbno = be64_to_cpu(*XFS_BMAP_PTR_IADDR(block, cur->bc_ptrs[lev], cur));
 		if ((error = xfs_btree_read_bufl(mp, tp, fsbno, 0, &bp,
 				XFS_BMAP_BTREE_REF))) {
 			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
@@ -2135,7 +2127,7 @@ xfs_bmbt_increment(
 	tp = cur->bc_tp;
 	mp = cur->bc_mp;
 	for (block = xfs_bmbt_get_block(cur, lev, &bp); lev > level; ) {
-		fsbno = INT_GET(*XFS_BMAP_PTR_IADDR(block, cur->bc_ptrs[lev], cur), ARCH_CONVERT);
+		fsbno = be64_to_cpu(*XFS_BMAP_PTR_IADDR(block, cur->bc_ptrs[lev], cur));
 		if ((error = xfs_btree_read_bufl(mp, tp, fsbno, 0, &bp,
 				XFS_BMAP_BTREE_REF))) {
 			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
@@ -2178,7 +2170,7 @@ xfs_bmbt_insert(
 	level = 0;
 	nbno = NULLFSBLOCK;
 	xfs_bmbt_disk_set_all(&nrec, &cur->bc_rec.b);
-	ncur = (xfs_btree_cur_t *)0;
+	ncur = NULL;
 	pcur = cur;
 	do {
 		if ((error = xfs_bmbt_insrec(pcur, level++, &nbno, &nrec, &ncur,
@@ -2205,7 +2197,7 @@ xfs_bmbt_insert(
 		}
 		if (ncur) {
 			pcur = ncur;
-			ncur = (xfs_btree_cur_t *)0;
+			ncur = NULL;
 		}
 	} while (nbno != NULLFSBLOCK);
 	XFS_BMBT_TRACE_CURSOR(cur, EXIT);
@@ -2356,12 +2348,12 @@ xfs_bmbt_newroot(
 	args.firstblock = args.fsbno;
 	if (args.fsbno == NULLFSBLOCK) {
 #ifdef DEBUG
-		if ((error = xfs_btree_check_lptr(cur, INT_GET(*pp, ARCH_CONVERT), level))) {
+		if ((error = xfs_btree_check_lptr_disk(cur, *pp, level))) {
 			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
 			return error;
 		}
 #endif
-		args.fsbno = INT_GET(*pp, ARCH_CONVERT);
+		args.fsbno = be64_to_cpu(*pp);
 		args.type = XFS_ALLOCTYPE_START_BNO;
 	} else
 		args.type = XFS_ALLOCTYPE_NEAR_BNO;
@@ -2393,7 +2385,7 @@ xfs_bmbt_newroot(
 	cpp = XFS_BMAP_PTR_IADDR(cblock, 1, cur);
 #ifdef DEBUG
 	for (i = 0; i < be16_to_cpu(cblock->bb_numrecs); i++) {
-		if ((error = xfs_btree_check_lptr(cur, INT_GET(pp[i], ARCH_CONVERT), level))) {
+		if ((error = xfs_btree_check_lptr_disk(cur, pp[i], level))) {
 			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
 			return error;
 		}
@@ -2401,13 +2393,12 @@ xfs_bmbt_newroot(
 #endif
 	memcpy(cpp, pp, be16_to_cpu(cblock->bb_numrecs) * sizeof(*pp));
 #ifdef DEBUG
-	if ((error = xfs_btree_check_lptr(cur, (xfs_bmbt_ptr_t)args.fsbno,
-			level))) {
+	if ((error = xfs_btree_check_lptr(cur, args.fsbno, level))) {
 		XFS_BMBT_TRACE_CURSOR(cur, ERROR);
 		return error;
 	}
 #endif
-	INT_SET(*pp, ARCH_CONVERT, args.fsbno);
+	*pp = cpu_to_be64(args.fsbno);
 	xfs_iroot_realloc(cur->bc_private.b.ip, 1 - be16_to_cpu(cblock->bb_numrecs),
 		cur->bc_private.b.whichfork);
 	xfs_btree_setbuf(cur, level, bp);
@@ -2681,9 +2672,9 @@ xfs_bmbt_to_bmdr(
 {
 	int			dmxr;
 	xfs_bmbt_key_t		*fkp;
-	xfs_bmbt_ptr_t		*fpp;
+	__be64			*fpp;
 	xfs_bmbt_key_t		*tkp;
-	xfs_bmbt_ptr_t		*tpp;
+	__be64			*tpp;
 
 	ASSERT(be32_to_cpu(rblock->bb_magic) == XFS_BMAP_MAGIC);
 	ASSERT(be64_to_cpu(rblock->bb_leftsib) == NULLDFSBNO);
@@ -2698,7 +2689,7 @@ xfs_bmbt_to_bmdr(
 	tpp = XFS_BTREE_PTR_ADDR(dblocklen, xfs_bmdr, dblock, 1, dmxr);
 	dmxr = be16_to_cpu(dblock->bb_numrecs);
 	memcpy(tkp, fkp, sizeof(*fkp) * dmxr);
-	memcpy(tpp, fpp, sizeof(*fpp) * dmxr); /* INT_: direct copy */
+	memcpy(tpp, fpp, sizeof(*fpp) * dmxr);
 }
 
 /*
@@ -2740,7 +2731,7 @@ xfs_bmbt_update(
 		XFS_BMBT_TRACE_CURSOR(cur, EXIT);
 		return 0;
 	}
-	INT_SET(key.br_startoff, ARCH_CONVERT, off);
+	key.br_startoff = cpu_to_be64(off);
 	if ((error = xfs_bmbt_updkey(cur, &key, 1))) {
 		XFS_BMBT_TRACE_CURSOR(cur, ERROR);
 		return error;

fs/xfs/xfs_bmap_btree.h

@@ -163,13 +163,14 @@ typedef struct xfs_bmbt_irec
 /*
  * Key structure for non-leaf levels of the tree.
  */
-typedef struct xfs_bmbt_key
-{
-	xfs_dfiloff_t	br_startoff;	/* starting file offset */
+typedef struct xfs_bmbt_key {
+	__be64		br_startoff;	/* starting file offset */
 } xfs_bmbt_key_t, xfs_bmdr_key_t;
 
-typedef xfs_dfsbno_t xfs_bmbt_ptr_t, xfs_bmdr_ptr_t;	/* btree pointer type */
-					/* btree block header type */
+/* btree pointer type */
+typedef __be64 xfs_bmbt_ptr_t, xfs_bmdr_ptr_t;
+
+/* btree block header type */
 typedef struct xfs_btree_lblock xfs_bmbt_block_t;
 
 #define XFS_BUF_TO_BMBT_BLOCK(bp)	((xfs_bmbt_block_t *)XFS_BUF_PTR(bp))

fs/xfs/xfs_btree.c

@@ -161,7 +161,7 @@ xfs_btree_check_key(
 
 		k1 = ak1;
 		k2 = ak2;
-		ASSERT(INT_GET(k1->br_startoff, ARCH_CONVERT) < INT_GET(k2->br_startoff, ARCH_CONVERT));
+		ASSERT(be64_to_cpu(k1->br_startoff) < be64_to_cpu(k2->br_startoff));
 		break;
 	    }
 	case XFS_BTNUM_INO: {
@@ -170,7 +170,7 @@ xfs_btree_check_key(
 
 		k1 = ak1;
 		k2 = ak2;
-		ASSERT(INT_GET(k1->ir_startino, ARCH_CONVERT) < INT_GET(k2->ir_startino, ARCH_CONVERT));
+		ASSERT(be32_to_cpu(k1->ir_startino) < be32_to_cpu(k2->ir_startino));
 		break;
 	    }
 	default:
@@ -285,8 +285,8 @@ xfs_btree_check_rec(
 
 		r1 = ar1;
 		r2 = ar2;
-		ASSERT(INT_GET(r1->ir_startino, ARCH_CONVERT) + XFS_INODES_PER_CHUNK <=
-		       INT_GET(r2->ir_startino, ARCH_CONVERT));
+		ASSERT(be32_to_cpu(r1->ir_startino) + XFS_INODES_PER_CHUNK <=
+		       be32_to_cpu(r2->ir_startino));
 		break;
 	    }
 	default:

fs/xfs/xfs_btree.h

@@ -145,7 +145,7 @@ typedef struct xfs_btree_cur
 	union {
 		xfs_alloc_rec_incore_t	a;
 		xfs_bmbt_irec_t		b;
-		xfs_inobt_rec_t		i;
+		xfs_inobt_rec_incore_t	i;
 	}		bc_rec;		/* current insert/search record value */
 	struct xfs_buf	*bc_bufs[XFS_BTREE_MAXLEVELS];	/* buf ptr per level */
 	int		bc_ptrs[XFS_BTREE_MAXLEVELS];	/* key/record # */
@@ -243,6 +243,9 @@ xfs_btree_check_lptr(
 	xfs_dfsbno_t		ptr,	/* btree block disk address */
 	int			level);	/* btree block level */
 
+#define xfs_btree_check_lptr_disk(cur, ptr, level) \
+	xfs_btree_check_lptr(cur, be64_to_cpu(ptr), level)
+
 /*
  * Checking routine: check that short form block header is ok.
  */

fs/xfs/xfs_buf_item.c

@@ -234,7 +234,6 @@ xfs_buf_item_format(
 	ASSERT((bip->bli_flags & XFS_BLI_LOGGED) ||
 	       (bip->bli_flags & XFS_BLI_STALE));
 	bp = bip->bli_buf;
-	ASSERT(XFS_BUF_BP_ISMAPPED(bp));
 	vecp = log_vector;
 
 	/*
@@ -627,25 +626,6 @@ xfs_buf_item_committed(
 	return (lsn);
 }
 
-/*
- * This is called when the transaction holding the buffer is aborted.
- * Just behave as if the transaction had been cancelled. If we're shutting down
- * and have aborted this transaction, we'll trap this buffer when it tries to
- * get written out.
- */
-STATIC void
-xfs_buf_item_abort(
-	xfs_buf_log_item_t	*bip)
-{
-	xfs_buf_t	*bp;
-
-	bp = bip->bli_buf;
-	xfs_buftrace("XFS_ABORT", bp);
-	XFS_BUF_SUPER_STALE(bp);
-	xfs_buf_item_unlock(bip);
-	return;
-}
-
 /*
  * This is called to asynchronously write the buffer associated with this
  * buf log item out to disk. The buffer will already have been locked by
@@ -693,7 +673,6 @@ STATIC struct xfs_item_ops xfs_buf_item_ops = {
 	.iop_committed	= (xfs_lsn_t(*)(xfs_log_item_t*, xfs_lsn_t))
 					xfs_buf_item_committed,
 	.iop_push	= (void(*)(xfs_log_item_t*))xfs_buf_item_push,
-	.iop_abort	= (void(*)(xfs_log_item_t*))xfs_buf_item_abort,
 	.iop_pushbuf	= NULL,
 	.iop_committing = (void(*)(xfs_log_item_t*, xfs_lsn_t))
 					xfs_buf_item_committing
@@ -901,7 +880,6 @@ xfs_buf_item_relse(
 	XFS_BUF_SET_FSPRIVATE(bp, bip->bli_item.li_bio_list);
 	if ((XFS_BUF_FSPRIVATE(bp, void *) == NULL) &&
 	    (XFS_BUF_IODONE_FUNC(bp) != NULL)) {
-		ASSERT((XFS_BUF_ISUNINITIAL(bp)) == 0);
 		XFS_BUF_CLR_IODONE_FUNC(bp);
 	}
 

fs/xfs/xfs_da_btree.c

@@ -1054,7 +1054,7 @@ xfs_da_node_lookup_int(xfs_da_state_t *state, int *result)
 	xfs_da_node_entry_t *btree;
 	xfs_dablk_t blkno;
 	int probe, span, max, error, retval;
-	xfs_dahash_t hashval;
+	xfs_dahash_t hashval, btreehashval;
 	xfs_da_args_t *args;
 
 	args = state->args;
@@ -1079,30 +1079,32 @@ xfs_da_node_lookup_int(xfs_da_state_t *state, int *result)
 			return(error);
 		}
 		curr = blk->bp->data;
-		ASSERT(be16_to_cpu(curr->magic) == XFS_DA_NODE_MAGIC ||
-		       be16_to_cpu(curr->magic) == XFS_DIR2_LEAFN_MAGIC ||
-		       be16_to_cpu(curr->magic) == XFS_ATTR_LEAF_MAGIC);
+		blk->magic = be16_to_cpu(curr->magic);
+		ASSERT(blk->magic == XFS_DA_NODE_MAGIC ||
+		       blk->magic == XFS_DIR2_LEAFN_MAGIC ||
+		       blk->magic == XFS_ATTR_LEAF_MAGIC);
 
 		/*
 		 * Search an intermediate node for a match.
 		 */
-		blk->magic = be16_to_cpu(curr->magic);
 		if (blk->magic == XFS_DA_NODE_MAGIC) {
 			node = blk->bp->data;
-			blk->hashval = be32_to_cpu(node->btree[be16_to_cpu(node->hdr.count)-1].hashval);
+			max = be16_to_cpu(node->hdr.count);
+			btreehashval = node->btree[max-1].hashval;
+			blk->hashval = be32_to_cpu(btreehashval);
 
 			/*
 			 * Binary search.  (note: small blocks will skip loop)
 			 */
-			max = be16_to_cpu(node->hdr.count);
 			probe = span = max / 2;
 			hashval = args->hashval;
 			for (btree = &node->btree[probe]; span > 4;
 				   btree = &node->btree[probe]) {
 				span /= 2;
-				if (be32_to_cpu(btree->hashval) < hashval)
+				btreehashval = be32_to_cpu(btree->hashval);
+				if (btreehashval < hashval)
 					probe += span;
-				else if (be32_to_cpu(btree->hashval) > hashval)
+				else if (btreehashval > hashval)
 					probe -= span;
 				else
 					break;
@@ -1133,10 +1135,10 @@ xfs_da_node_lookup_int(xfs_da_state_t *state, int *result)
 				blk->index = probe;
 				blkno = be32_to_cpu(btree->before);
 			}
-		} else if (be16_to_cpu(curr->magic) == XFS_ATTR_LEAF_MAGIC) {
+		} else if (blk->magic == XFS_ATTR_LEAF_MAGIC) {
 			blk->hashval = xfs_attr_leaf_lasthash(blk->bp, NULL);
 			break;
-		} else if (be16_to_cpu(curr->magic) == XFS_DIR2_LEAFN_MAGIC) {
+		} else if (blk->magic == XFS_DIR2_LEAFN_MAGIC) {
 			blk->hashval = xfs_dir2_leafn_lasthash(blk->bp, NULL);
 			break;
 		}
@@ -1152,11 +1154,13 @@ xfs_da_node_lookup_int(xfs_da_state_t *state, int *result)
 		if (blk->magic == XFS_DIR2_LEAFN_MAGIC) {
 			retval = xfs_dir2_leafn_lookup_int(blk->bp, args,
 							&blk->index, state);
-		}
-		else if (blk->magic == XFS_ATTR_LEAF_MAGIC) {
+		} else if (blk->magic == XFS_ATTR_LEAF_MAGIC) {
 			retval = xfs_attr_leaf_lookup_int(blk->bp, args);
 			blk->index = args->index;
 			args->blkno = blk->blkno;
+		} else {
+			ASSERT(0);
+			return XFS_ERROR(EFSCORRUPTED);
 		}
 		if (((retval == ENOENT) || (retval == ENOATTR)) &&
 		    (blk->hashval == args->hashval)) {
@@ -1166,8 +1170,7 @@ xfs_da_node_lookup_int(xfs_da_state_t *state, int *result)
 				return(error);
 			if (retval == 0) {
 				continue;
-			}
-			else if (blk->magic == XFS_ATTR_LEAF_MAGIC) {
+			} else if (blk->magic == XFS_ATTR_LEAF_MAGIC) {
 				/* path_shift() gives ENOENT */
 				retval = XFS_ERROR(ENOATTR);
 			}

fs/xfs/xfs_error.h

@@ -18,14 +18,6 @@
 #ifndef	__XFS_ERROR_H__
 #define	__XFS_ERROR_H__
 
-#define XFS_ERECOVER	1	/* Failure to recover log */
-#define XFS_ELOGSTAT	2	/* Failure to stat log in user space */
-#define XFS_ENOLOGSPACE	3	/* Reservation too large */
-#define XFS_ENOTSUP	4	/* Operation not supported */
-#define	XFS_ENOLSN	5	/* Can't find the lsn you asked for */
-#define XFS_ENOTFOUND	6
-#define XFS_ENOTXFS	7	/* Not XFS filesystem */
-
 #ifdef DEBUG
 #define	XFS_ERROR_NTRAP	10
 extern int	xfs_etrap[XFS_ERROR_NTRAP];
@@ -175,6 +167,7 @@ extern int xfs_errortag_clearall_umount(int64_t fsid, char *fsname, int loud);
 #define		XFS_PTAG_SHUTDOWN_CORRUPT	0x00000010
 #define		XFS_PTAG_SHUTDOWN_IOERROR	0x00000020
 #define		XFS_PTAG_SHUTDOWN_LOGERROR	0x00000040
+#define		XFS_PTAG_FSBLOCK_ZERO		0x00000080
 
 struct xfs_mount;
 /* PRINTFLIKE4 */

fs/xfs/xfs_extfree_item.c

@@ -33,9 +33,6 @@ kmem_zone_t	*xfs_efi_zone;
 kmem_zone_t	*xfs_efd_zone;
 
 STATIC void	xfs_efi_item_unlock(xfs_efi_log_item_t *);
-STATIC void	xfs_efi_item_abort(xfs_efi_log_item_t *);
-STATIC void	xfs_efd_item_abort(xfs_efd_log_item_t *);
-
 
 void
 xfs_efi_item_free(xfs_efi_log_item_t *efip)
@@ -184,7 +181,7 @@ STATIC void
 xfs_efi_item_unlock(xfs_efi_log_item_t *efip)
 {
 	if (efip->efi_item.li_flags & XFS_LI_ABORTED)
-		xfs_efi_item_abort(efip);
+		xfs_efi_item_free(efip);
 	return;
 }
 
@@ -201,18 +198,6 @@ xfs_efi_item_committed(xfs_efi_log_item_t *efip, xfs_lsn_t lsn)
 	return lsn;
 }
 
-/*
- * This is called when the transaction logging the EFI is aborted.
- * Free up the EFI and return.  No need to clean up the slot for
- * the item in the transaction.  That was done by the unpin code
- * which is called prior to this routine in the abort/fs-shutdown path.
- */
-STATIC void
-xfs_efi_item_abort(xfs_efi_log_item_t *efip)
-{
-	xfs_efi_item_free(efip);
-}
-
 /*
  * There isn't much you can do to push on an efi item.  It is simply
  * stuck waiting for all of its corresponding efd items to be
@@ -255,7 +240,6 @@ STATIC struct xfs_item_ops xfs_efi_item_ops = {
 	.iop_committed	= (xfs_lsn_t(*)(xfs_log_item_t*, xfs_lsn_t))
 					xfs_efi_item_committed,
 	.iop_push	= (void(*)(xfs_log_item_t*))xfs_efi_item_push,
-	.iop_abort	= (void(*)(xfs_log_item_t*))xfs_efi_item_abort,
 	.iop_pushbuf	= NULL,
 	.iop_committing = (void(*)(xfs_log_item_t*, xfs_lsn_t))
 					xfs_efi_item_committing
@@ -386,33 +370,6 @@ xfs_efi_release(xfs_efi_log_item_t	*efip,
 	}
 }
 
-/*
- * This is called when the transaction that should be committing the
- * EFD corresponding to the given EFI is aborted.  The committed and
- * canceled flags are used to coordinate the freeing of the EFI and
- * the references by the transaction that committed it.
- */
-STATIC void
-xfs_efi_cancel(
-	xfs_efi_log_item_t	*efip)
-{
-	xfs_mount_t	*mp;
-	SPLDECL(s);
-
-	mp = efip->efi_item.li_mountp;
-	AIL_LOCK(mp, s);
-	if (efip->efi_flags & XFS_EFI_COMMITTED) {
-		/*
-		 * xfs_trans_delete_ail() drops the AIL lock.
-		 */
-		xfs_trans_delete_ail(mp, (xfs_log_item_t *)efip, s);
-		xfs_efi_item_free(efip);
-	} else {
-		efip->efi_flags |= XFS_EFI_CANCELED;
-		AIL_UNLOCK(mp, s);
-	}
-}
-
 STATIC void
 xfs_efd_item_free(xfs_efd_log_item_t *efdp)
 {
@@ -514,7 +471,7 @@ STATIC void
 xfs_efd_item_unlock(xfs_efd_log_item_t *efdp)
 {
 	if (efdp->efd_item.li_flags & XFS_LI_ABORTED)
-		xfs_efd_item_abort(efdp);
+		xfs_efd_item_free(efdp);
 	return;
 }
 
@@ -540,27 +497,6 @@ xfs_efd_item_committed(xfs_efd_log_item_t *efdp, xfs_lsn_t lsn)
 	return (xfs_lsn_t)-1;
 }
 
-/*
- * The transaction of which this EFD is a part has been aborted.
- * Inform its companion EFI of this fact and then clean up after
- * ourselves.  No need to clean up the slot for the item in the
- * transaction.  That was done by the unpin code which is called
- * prior to this routine in the abort/fs-shutdown path.
- */
-STATIC void
-xfs_efd_item_abort(xfs_efd_log_item_t *efdp)
-{
-	/*
-	 * If we got a log I/O error, it's always the case that the LR with the
-	 * EFI got unpinned and freed before the EFD got aborted. So don't
-	 * reference the EFI at all in that case.
-	 */
-	if ((efdp->efd_item.li_flags & XFS_LI_ABORTED) == 0)
-		xfs_efi_cancel(efdp->efd_efip);
-
-	xfs_efd_item_free(efdp);
-}
-
 /*
  * There isn't much you can do to push on an efd item.  It is simply
  * stuck waiting for the log to be flushed to disk.
@@ -602,7 +538,6 @@ STATIC struct xfs_item_ops xfs_efd_item_ops = {
 	.iop_committed	= (xfs_lsn_t(*)(xfs_log_item_t*, xfs_lsn_t))
 					xfs_efd_item_committed,
 	.iop_push	= (void(*)(xfs_log_item_t*))xfs_efd_item_push,
-	.iop_abort	= (void(*)(xfs_log_item_t*))xfs_efd_item_abort,
 	.iop_pushbuf	= NULL,
 	.iop_committing = (void(*)(xfs_log_item_t*, xfs_lsn_t))
 					xfs_efd_item_committing

fs/xfs/xfs_extfree_item.h

@@ -33,14 +33,16 @@ typedef struct xfs_extent {
  * conversion routine.
  */
 
+#ifndef HAVE_FORMAT32
 typedef struct xfs_extent_32 {
-	xfs_dfsbno_t	ext_start;
-	xfs_extlen_t	ext_len;
+	__uint64_t	ext_start;
+	__uint32_t	ext_len;
 } __attribute__((packed)) xfs_extent_32_t;
+#endif
 
 typedef struct xfs_extent_64 {
-	xfs_dfsbno_t	ext_start;
-	xfs_extlen_t	ext_len;
+	__uint64_t	ext_start;
+	__uint32_t	ext_len;
 	__uint32_t	ext_pad;
 } xfs_extent_64_t;
 
@@ -50,25 +52,27 @@ typedef struct xfs_extent_64 {
  * size is given by efi_nextents.
  */
 typedef struct xfs_efi_log_format {
-	unsigned short		efi_type;	/* efi log item type */
-	unsigned short		efi_size;	/* size of this item */
-	uint			efi_nextents;	/* # extents to free */
+	__uint16_t		efi_type;	/* efi log item type */
+	__uint16_t		efi_size;	/* size of this item */
+	__uint32_t		efi_nextents;	/* # extents to free */
 	__uint64_t		efi_id;		/* efi identifier */
 	xfs_extent_t		efi_extents[1];	/* array of extents to free */
 } xfs_efi_log_format_t;
 
+#ifndef HAVE_FORMAT32
 typedef struct xfs_efi_log_format_32 {
-	unsigned short		efi_type;	/* efi log item type */
-	unsigned short		efi_size;	/* size of this item */
-	uint			efi_nextents;	/* # extents to free */
+	__uint16_t		efi_type;	/* efi log item type */
+	__uint16_t		efi_size;	/* size of this item */
+	__uint32_t		efi_nextents;	/* # extents to free */
 	__uint64_t		efi_id;		/* efi identifier */
 	xfs_extent_32_t		efi_extents[1];	/* array of extents to free */
 } __attribute__((packed)) xfs_efi_log_format_32_t;
+#endif
 
 typedef struct xfs_efi_log_format_64 {
-	unsigned short		efi_type;	/* efi log item type */
-	unsigned short		efi_size;	/* size of this item */
-	uint			efi_nextents;	/* # extents to free */
+	__uint16_t		efi_type;	/* efi log item type */
+	__uint16_t		efi_size;	/* size of this item */
+	__uint32_t		efi_nextents;	/* # extents to free */
 	__uint64_t		efi_id;		/* efi identifier */
 	xfs_extent_64_t		efi_extents[1];	/* array of extents to free */
 } xfs_efi_log_format_64_t;
@@ -79,25 +83,27 @@ typedef struct xfs_efi_log_format_64 {
  * size is given by efd_nextents;
  */
 typedef struct xfs_efd_log_format {
-	unsigned short		efd_type;	/* efd log item type */
-	unsigned short		efd_size;	/* size of this item */
-	uint			efd_nextents;	/* # of extents freed */
+	__uint16_t		efd_type;	/* efd log item type */
+	__uint16_t		efd_size;	/* size of this item */
+	__uint32_t		efd_nextents;	/* # of extents freed */
 	__uint64_t		efd_efi_id;	/* id of corresponding efi */
 	xfs_extent_t		efd_extents[1];	/* array of extents freed */
 } xfs_efd_log_format_t;
 
+#ifndef HAVE_FORMAT32
 typedef struct xfs_efd_log_format_32 {
-	unsigned short		efd_type;	/* efd log item type */
-	unsigned short		efd_size;	/* size of this item */
-	uint			efd_nextents;	/* # of extents freed */
+	__uint16_t		efd_type;	/* efd log item type */
+	__uint16_t		efd_size;	/* size of this item */
+	__uint32_t		efd_nextents;	/* # of extents freed */
 	__uint64_t		efd_efi_id;	/* id of corresponding efi */
 	xfs_extent_32_t		efd_extents[1];	/* array of extents freed */
 } __attribute__((packed)) xfs_efd_log_format_32_t;
+#endif
 
 typedef struct xfs_efd_log_format_64 {
-	unsigned short		efd_type;	/* efd log item type */
-	unsigned short		efd_size;	/* size of this item */
-	uint			efd_nextents;	/* # of extents freed */
+	__uint16_t		efd_type;	/* efd log item type */
+	__uint16_t		efd_size;	/* size of this item */
+	__uint32_t		efd_nextents;	/* # of extents freed */
 	__uint64_t		efd_efi_id;	/* id of corresponding efi */
 	xfs_extent_64_t		efd_extents[1];	/* array of extents freed */
 } xfs_efd_log_format_64_t;

fs/xfs/xfs_fs.h

@@ -22,8 +22,6 @@
  * SGI's XFS filesystem's major stuff (constants, structures)
  */
 
-#define XFS_NAME	"xfs"
-
 /*
  * Direct I/O attribute record used with XFS_IOC_DIOINFO
  * d_miniosz is the min xfer size, xfer size multiple and file seek offset
@@ -426,11 +424,7 @@ typedef struct xfs_handle {
 				 - (char *) &(handle))			  \
 				 + (handle).ha_fid.xfs_fid_len)
 
-#define XFS_HANDLE_CMP(h1, h2)	memcmp(h1, h2, sizeof(xfs_handle_t))
-
-#define FSHSIZE		sizeof(fsid_t)
-
-/* 
+/*
  * Flags for going down operation
  */
 #define XFS_FSOP_GOING_FLAGS_DEFAULT		0x0	/* going down */

fs/xfs/xfs_ialloc.c

@@ -458,7 +458,7 @@ nextag:
 		 */
 		if (XFS_FORCED_SHUTDOWN(mp)) {
 			up_read(&mp->m_peraglock);
-			return (xfs_buf_t *)0;
+			return NULL;
 		}
 		agno++;
 		if (agno >= agcount)
@@ -466,7 +466,7 @@ nextag:
 		if (agno == pagno) {
 			if (flags == 0) {
 				up_read(&mp->m_peraglock);
-				return (xfs_buf_t *)0;
+				return NULL;
 			}
 			flags = 0;
 		}
@@ -529,10 +529,10 @@ xfs_dialloc(
 	int		offset;		/* index of inode in chunk */
 	xfs_agino_t	pagino;		/* parent's a.g. relative inode # */
 	xfs_agnumber_t	pagno;		/* parent's allocation group number */
-	xfs_inobt_rec_t	rec;		/* inode allocation record */
+	xfs_inobt_rec_incore_t rec;	/* inode allocation record */
 	xfs_agnumber_t	tagno;		/* testing allocation group number */
 	xfs_btree_cur_t	*tcur;		/* temp cursor */
-	xfs_inobt_rec_t	trec;		/* temp inode allocation record */
+	xfs_inobt_rec_incore_t trec;	/* temp inode allocation record */
 
 
 	if (*IO_agbp == NULL) {
@@ -945,7 +945,7 @@ xfs_difree(
 	int		ilen;	/* inodes in an inode cluster */
 	xfs_mount_t	*mp;	/* mount structure for filesystem */
 	int		off;	/* offset of inode in inode chunk */
-	xfs_inobt_rec_t	rec;	/* btree record */
+	xfs_inobt_rec_incore_t rec;	/* btree record */
 
 	mp = tp->t_mountp;
 
@@ -1195,6 +1195,7 @@ xfs_dilocate(
 					"(0x%llx)",
 					ino, XFS_AGINO_TO_INO(mp, agno, agino));
 		}
+		xfs_stack_trace();
 #endif /* DEBUG */
 		return XFS_ERROR(EINVAL);
 	}

fs/xfs/xfs_ialloc_btree.c

@@ -568,7 +568,7 @@ xfs_inobt_insrec(
 	/*
 	 * Make a key out of the record data to be inserted, and save it.
 	 */
-	key.ir_startino = recp->ir_startino; /* INT_: direct copy */
+	key.ir_startino = recp->ir_startino;
 	optr = ptr = cur->bc_ptrs[level];
 	/*
 	 * If we're off the left edge, return failure.
@@ -600,7 +600,7 @@ xfs_inobt_insrec(
 	}
 #endif
 	nbno = NULLAGBLOCK;
-	ncur = (xfs_btree_cur_t *)0;
+	ncur = NULL;
 	/*
 	 * If the block is full, we can't insert the new entry until we
 	 * make the block un-full.
@@ -641,7 +641,7 @@ xfs_inobt_insrec(
 						return error;
 #endif
 					ptr = cur->bc_ptrs[level];
-					nrec.ir_startino = nkey.ir_startino; /* INT_: direct copy */
+					nrec.ir_startino = nkey.ir_startino;
 				} else {
 					/*
 					 * Otherwise the insert fails.
@@ -681,7 +681,7 @@ xfs_inobt_insrec(
 		if ((error = xfs_btree_check_sptr(cur, *bnop, level)))
 			return error;
 #endif
-		kp[ptr - 1] = key; /* INT_: struct copy */
+		kp[ptr - 1] = key;
 		pp[ptr - 1] = cpu_to_be32(*bnop);
 		numrecs++;
 		block->bb_numrecs = cpu_to_be16(numrecs);
@@ -698,7 +698,7 @@ xfs_inobt_insrec(
 		 * Now stuff the new record in, bump numrecs
 		 * and log the new data.
 		 */
-		rp[ptr - 1] = *recp; /* INT_: struct copy */
+		rp[ptr - 1] = *recp;
 		numrecs++;
 		block->bb_numrecs = cpu_to_be16(numrecs);
 		xfs_inobt_log_recs(cur, bp, ptr, numrecs);
@@ -731,7 +731,7 @@ xfs_inobt_insrec(
 	 */
 	*bnop = nbno;
 	if (nbno != NULLAGBLOCK) {
-		*recp = nrec; /* INT_: struct copy */
+		*recp = nrec;
 		*curp = ncur;
 	}
 	*stat = 1;
@@ -878,7 +878,7 @@ xfs_inobt_lookup(
 		 */
 		bp = cur->bc_bufs[level];
 		if (bp && XFS_BUF_ADDR(bp) != d)
-			bp = (xfs_buf_t *)0;
+			bp = NULL;
 		if (!bp) {
 			/*
 			 * Need to get a new buffer.  Read it, then
@@ -950,12 +950,12 @@ xfs_inobt_lookup(
 					xfs_inobt_key_t	*kkp;
 
 					kkp = kkbase + keyno - 1;
-					startino = INT_GET(kkp->ir_startino, ARCH_CONVERT);
+					startino = be32_to_cpu(kkp->ir_startino);
 				} else {
 					xfs_inobt_rec_t	*krp;
 
 					krp = krbase + keyno - 1;
-					startino = INT_GET(krp->ir_startino, ARCH_CONVERT);
+					startino = be32_to_cpu(krp->ir_startino);
 				}
 				/*
 				 * Compute difference to get next direction.
@@ -1117,7 +1117,7 @@ xfs_inobt_lshift(
 		if ((error = xfs_btree_check_sptr(cur, be32_to_cpu(*rpp), level)))
 			return error;
 #endif
-		*lpp = *rpp; /* INT_: no-change copy */
+		*lpp = *rpp;
 		xfs_inobt_log_ptrs(cur, lbp, nrec, nrec);
 	}
 	/*
@@ -1160,7 +1160,7 @@ xfs_inobt_lshift(
 	} else {
 		memmove(rrp, rrp + 1, be16_to_cpu(right->bb_numrecs) * sizeof(*rrp));
 		xfs_inobt_log_recs(cur, rbp, 1, be16_to_cpu(right->bb_numrecs));
-		key.ir_startino = rrp->ir_startino; /* INT_: direct copy */
+		key.ir_startino = rrp->ir_startino;
 		rkp = &key;
 	}
 	/*
@@ -1297,13 +1297,13 @@ xfs_inobt_newroot(
 	 */
 	kp = XFS_INOBT_KEY_ADDR(new, 1, cur);
 	if (be16_to_cpu(left->bb_level) > 0) {
-		kp[0] = *XFS_INOBT_KEY_ADDR(left, 1, cur); /* INT_: struct copy */
-		kp[1] = *XFS_INOBT_KEY_ADDR(right, 1, cur); /* INT_: struct copy */
+		kp[0] = *XFS_INOBT_KEY_ADDR(left, 1, cur);
+		kp[1] = *XFS_INOBT_KEY_ADDR(right, 1, cur);
 	} else {
 		rp = XFS_INOBT_REC_ADDR(left, 1, cur);
-		INT_COPY(kp[0].ir_startino, rp->ir_startino, ARCH_CONVERT);
+		kp[0].ir_startino = rp->ir_startino;
 		rp = XFS_INOBT_REC_ADDR(right, 1, cur);
-		INT_COPY(kp[1].ir_startino, rp->ir_startino, ARCH_CONVERT);
+		kp[1].ir_startino = rp->ir_startino;
 	}
 	xfs_inobt_log_keys(cur, nbp, 1, 2);
 	/*
@@ -1410,8 +1410,8 @@ xfs_inobt_rshift(
 		if ((error = xfs_btree_check_sptr(cur, be32_to_cpu(*lpp), level)))
 			return error;
 #endif
-		*rkp = *lkp; /* INT_: no change copy */
-		*rpp = *lpp; /* INT_: no change copy */
+		*rkp = *lkp;
+		*rpp = *lpp;
 		xfs_inobt_log_keys(cur, rbp, 1, be16_to_cpu(right->bb_numrecs) + 1);
 		xfs_inobt_log_ptrs(cur, rbp, 1, be16_to_cpu(right->bb_numrecs) + 1);
 	} else {
@@ -1420,7 +1420,7 @@ xfs_inobt_rshift(
 		memmove(rrp + 1, rrp, be16_to_cpu(right->bb_numrecs) * sizeof(*rrp));
 		*rrp = *lrp;
 		xfs_inobt_log_recs(cur, rbp, 1, be16_to_cpu(right->bb_numrecs) + 1);
-		key.ir_startino = rrp->ir_startino; /* INT_: direct copy */
+		key.ir_startino = rrp->ir_startino;
 		rkp = &key;
 	}
 	/*
@@ -1559,7 +1559,7 @@ xfs_inobt_split(
 		rrp = XFS_INOBT_REC_ADDR(right, 1, cur);
 		memcpy(rrp, lrp, be16_to_cpu(right->bb_numrecs) * sizeof(*rrp));
 		xfs_inobt_log_recs(cur, rbp, 1, be16_to_cpu(right->bb_numrecs));
-		keyp->ir_startino = rrp->ir_startino; /* INT_: direct copy */
+		keyp->ir_startino = rrp->ir_startino;
 	}
 	/*
 	 * Find the left block number by looking in the buffer.
@@ -1813,9 +1813,9 @@ xfs_inobt_get_rec(
 	 * Point to the record and extract its data.
 	 */
 	rec = XFS_INOBT_REC_ADDR(block, ptr, cur);
-	*ino = INT_GET(rec->ir_startino, ARCH_CONVERT);
-	*fcnt = INT_GET(rec->ir_freecount, ARCH_CONVERT);
-	*free = INT_GET(rec->ir_free, ARCH_CONVERT);
+	*ino = be32_to_cpu(rec->ir_startino);
+	*fcnt = be32_to_cpu(rec->ir_freecount);
+	*free = be64_to_cpu(rec->ir_free);
 	*stat = 1;
 	return 0;
 }
@@ -1930,10 +1930,10 @@ xfs_inobt_insert(
 
 	level = 0;
 	nbno = NULLAGBLOCK;
-	INT_SET(nrec.ir_startino, ARCH_CONVERT, cur->bc_rec.i.ir_startino);
-	INT_SET(nrec.ir_freecount, ARCH_CONVERT, cur->bc_rec.i.ir_freecount);
-	INT_SET(nrec.ir_free, ARCH_CONVERT, cur->bc_rec.i.ir_free);
-	ncur = (xfs_btree_cur_t *)0;
+	nrec.ir_startino = cpu_to_be32(cur->bc_rec.i.ir_startino);
+	nrec.ir_freecount = cpu_to_be32(cur->bc_rec.i.ir_freecount);
+	nrec.ir_free = cpu_to_be64(cur->bc_rec.i.ir_free);
+	ncur = NULL;
 	pcur = cur;
 	/*
 	 * Loop going up the tree, starting at the leaf level.
@@ -1965,7 +1965,7 @@ xfs_inobt_insert(
 		 */
 		if (ncur) {
 			pcur = ncur;
-			ncur = (xfs_btree_cur_t *)0;
+			ncur = NULL;
 		}
 	} while (nbno != NULLAGBLOCK);
 	*stat = i;
@@ -2060,9 +2060,9 @@ xfs_inobt_update(
 	/*
 	 * Fill in the new contents and log them.
 	 */
-	INT_SET(rp->ir_startino, ARCH_CONVERT, ino);
-	INT_SET(rp->ir_freecount, ARCH_CONVERT, fcnt);
-	INT_SET(rp->ir_free, ARCH_CONVERT, free);
+	rp->ir_startino = cpu_to_be32(ino);
+	rp->ir_freecount = cpu_to_be32(fcnt);
+	rp->ir_free = cpu_to_be64(free);
 	xfs_inobt_log_recs(cur, bp, ptr, ptr);
 	/*
 	 * Updating first record in leaf. Pass new key value up to our parent.
@@ -2070,7 +2070,7 @@ xfs_inobt_update(
 	if (ptr == 1) {
 		xfs_inobt_key_t	key;	/* key containing [ino] */
 
-		INT_SET(key.ir_startino, ARCH_CONVERT, ino);
+		key.ir_startino = cpu_to_be32(ino);
 		if ((error = xfs_inobt_updkey(cur, &key, 1)))
 			return error;
 	}

fs/xfs/xfs_ialloc_btree.h

@@ -47,19 +47,24 @@ static inline xfs_inofree_t xfs_inobt_maskn(int i, int n)
 /*
  * Data record structure
  */
-typedef struct xfs_inobt_rec
-{
+typedef struct xfs_inobt_rec {
+	__be32		ir_startino;	/* starting inode number */
+	__be32		ir_freecount;	/* count of free inodes (set bits) */
+	__be64		ir_free;	/* free inode mask */
+} xfs_inobt_rec_t;
+
+typedef struct xfs_inobt_rec_incore {
 	xfs_agino_t	ir_startino;	/* starting inode number */
 	__int32_t	ir_freecount;	/* count of free inodes (set bits) */
 	xfs_inofree_t	ir_free;	/* free inode mask */
-} xfs_inobt_rec_t;
+} xfs_inobt_rec_incore_t;
+
 
 /*
  * Key structure
  */
-typedef struct xfs_inobt_key
-{
-	xfs_agino_t	ir_startino;	/* starting inode number */
+typedef struct xfs_inobt_key {
+	__be32		ir_startino;	/* starting inode number */
 } xfs_inobt_key_t;
 
 /* btree pointer type */
@@ -77,7 +82,7 @@ typedef	struct xfs_btree_sblock xfs_inobt_block_t;
 #define	XFS_INOBT_IS_FREE(rp,i)		\
 		(((rp)->ir_free & XFS_INOBT_MASK(i)) != 0)
 #define	XFS_INOBT_IS_FREE_DISK(rp,i)	\
-		((INT_GET((rp)->ir_free,ARCH_CONVERT) & XFS_INOBT_MASK(i)) != 0)
+		((be64_to_cpu((rp)->ir_free) & XFS_INOBT_MASK(i)) != 0)
 #define	XFS_INOBT_SET_FREE(rp,i)	((rp)->ir_free |= XFS_INOBT_MASK(i))
 #define	XFS_INOBT_CLR_FREE(rp,i)	((rp)->ir_free &= ~XFS_INOBT_MASK(i))
 

fs/xfs/xfs_iget.c

@@ -50,7 +50,7 @@ void
 xfs_ihash_init(xfs_mount_t *mp)
 {
 	__uint64_t	icount;
-	uint		i, flags = KM_SLEEP | KM_MAYFAIL;
+	uint		i;
 
 	if (!mp->m_ihsize) {
 		icount = mp->m_maxicount ? mp->m_maxicount :
@@ -61,14 +61,13 @@ xfs_ihash_init(xfs_mount_t *mp)
 					(64 * NBPP) / sizeof(xfs_ihash_t));
 	}
 
-	while (!(mp->m_ihash = (xfs_ihash_t *)kmem_zalloc(mp->m_ihsize *
-						sizeof(xfs_ihash_t), flags))) {
-		if ((mp->m_ihsize >>= 1) <= NBPP)
-			flags = KM_SLEEP;
-	}
-	for (i = 0; i < mp->m_ihsize; i++) {
+	mp->m_ihash = kmem_zalloc_greedy(&mp->m_ihsize,
+					 NBPC * sizeof(xfs_ihash_t),
+					 mp->m_ihsize * sizeof(xfs_ihash_t),
+					 KM_SLEEP | KM_MAYFAIL | KM_LARGE);
+	mp->m_ihsize /= sizeof(xfs_ihash_t);
+	for (i = 0; i < mp->m_ihsize; i++)
 		rwlock_init(&(mp->m_ihash[i].ih_lock));
-	}
 }
 
 /*
@@ -77,7 +76,7 @@ xfs_ihash_init(xfs_mount_t *mp)
 void
 xfs_ihash_free(xfs_mount_t *mp)
 {
-	kmem_free(mp->m_ihash, mp->m_ihsize*sizeof(xfs_ihash_t));
+	kmem_free(mp->m_ihash, mp->m_ihsize * sizeof(xfs_ihash_t));
 	mp->m_ihash = NULL;
 }
 
@@ -95,7 +94,7 @@ xfs_chash_init(xfs_mount_t *mp)
 	mp->m_chsize = min_t(uint, mp->m_chsize, mp->m_ihsize);
 	mp->m_chash = (xfs_chash_t *)kmem_zalloc(mp->m_chsize
 						 * sizeof(xfs_chash_t),
-						 KM_SLEEP);
+						 KM_SLEEP | KM_LARGE);
 	for (i = 0; i < mp->m_chsize; i++) {
 		spinlock_init(&mp->m_chash[i].ch_lock,"xfshash");
 	}
@@ -244,7 +243,9 @@ again:
 
 				XFS_STATS_INC(xs_ig_found);
 
+				spin_lock(&ip->i_flags_lock);
 				ip->i_flags &= ~XFS_IRECLAIMABLE;
+				spin_unlock(&ip->i_flags_lock);
 				version = ih->ih_version;
 				read_unlock(&ih->ih_lock);
 				xfs_ihash_promote(ih, ip, version);
@@ -290,15 +291,17 @@ again:
 
 finish_inode:
 			if (ip->i_d.di_mode == 0) {
-				if (!(flags & IGET_CREATE))
+				if (!(flags & XFS_IGET_CREATE))
 					return ENOENT;
 				xfs_iocore_inode_reinit(ip);
 			}
-	
+
 			if (lock_flags != 0)
 				xfs_ilock(ip, lock_flags);
 
+			spin_lock(&ip->i_flags_lock);
 			ip->i_flags &= ~XFS_ISTALE;
+			spin_unlock(&ip->i_flags_lock);
 
 			vn_trace_exit(vp, "xfs_iget.found",
 						(inst_t *)__return_address);
@@ -320,21 +323,20 @@ finish_inode:
 	 * Read the disk inode attributes into a new inode structure and get
 	 * a new vnode for it. This should also initialize i_ino and i_mount.
 	 */
-	error = xfs_iread(mp, tp, ino, &ip, bno);
-	if (error) {
+	error = xfs_iread(mp, tp, ino, &ip, bno,
+			  (flags & XFS_IGET_BULKSTAT) ? XFS_IMAP_BULKSTAT : 0);
+	if (error)
 		return error;
-	}
 
 	vn_trace_exit(vp, "xfs_iget.alloc", (inst_t *)__return_address);
 
 	xfs_inode_lock_init(ip, vp);
 	xfs_iocore_inode_init(ip);
 
-	if (lock_flags != 0) {
+	if (lock_flags)
 		xfs_ilock(ip, lock_flags);
-	}
-		
-	if ((ip->i_d.di_mode == 0) && !(flags & IGET_CREATE)) {
+
+	if ((ip->i_d.di_mode == 0) && !(flags & XFS_IGET_CREATE)) {
 		xfs_idestroy(ip);
 		return ENOENT;
 	}
@@ -369,7 +371,9 @@ finish_inode:
 	ih->ih_next = ip;
 	ip->i_udquot = ip->i_gdquot = NULL;
 	ih->ih_version++;
+	spin_lock(&ip->i_flags_lock);
 	ip->i_flags |= XFS_INEW;
+	spin_unlock(&ip->i_flags_lock);
 
 	write_unlock(&ih->ih_lock);
 
@@ -548,7 +552,7 @@ xfs_inode_lock_init(
 	mrlock_init(&ip->i_iolock, MRLOCK_BARRIER, "xfsio", vp->v_number);
 	init_waitqueue_head(&ip->i_ipin_wait);
 	atomic_set(&ip->i_pincount, 0);
-	init_sema(&ip->i_flock, 1, "xfsfino", vp->v_number);
+	initnsema(&ip->i_flock, 1, "xfsfino");
 }
 
 /*

fs/xfs/xfs_inode.c

@@ -854,7 +854,8 @@ xfs_iread(
 	xfs_trans_t	*tp,
 	xfs_ino_t	ino,
 	xfs_inode_t	**ipp,
-	xfs_daddr_t	bno)
+	xfs_daddr_t	bno,
+	uint		imap_flags)
 {
 	xfs_buf_t	*bp;
 	xfs_dinode_t	*dip;
@@ -866,6 +867,7 @@ xfs_iread(
 	ip = kmem_zone_zalloc(xfs_inode_zone, KM_SLEEP);
 	ip->i_ino = ino;
 	ip->i_mount = mp;
+	spin_lock_init(&ip->i_flags_lock);
 
 	/*
 	 * Get pointer's to the on-disk inode and the buffer containing it.
@@ -874,7 +876,7 @@ xfs_iread(
 	 * return NULL as well.  Set i_blkno to 0 so that xfs_itobp() will
 	 * know that this is a new incore inode.
 	 */
-	error = xfs_itobp(mp, tp, ip, &dip, &bp, bno, 0);
+	error = xfs_itobp(mp, tp, ip, &dip, &bp, bno, imap_flags);
 	if (error) {
 		kmem_zone_free(xfs_inode_zone, ip);
 		return error;
@@ -1113,7 +1115,7 @@ xfs_ialloc(
 	 * to prevent others from looking at until we're done.
 	 */
 	error = xfs_trans_iget(tp->t_mountp, tp, ino,
-			IGET_CREATE, XFS_ILOCK_EXCL, &ip);
+				XFS_IGET_CREATE, XFS_ILOCK_EXCL, &ip);
 	if (error != 0) {
 		return error;
 	}
@@ -2213,7 +2215,9 @@ xfs_ifree_cluster(
 
 			if (ip == free_ip) {
 				if (xfs_iflock_nowait(ip)) {
+					spin_lock(&ip->i_flags_lock);
 					ip->i_flags |= XFS_ISTALE;
+					spin_unlock(&ip->i_flags_lock);
 
 					if (xfs_inode_clean(ip)) {
 						xfs_ifunlock(ip);
@@ -2227,7 +2231,9 @@ xfs_ifree_cluster(
 
 			if (xfs_ilock_nowait(ip, XFS_ILOCK_EXCL)) {
 				if (xfs_iflock_nowait(ip)) {
+					spin_lock(&ip->i_flags_lock);
 					ip->i_flags |= XFS_ISTALE;
+					spin_unlock(&ip->i_flags_lock);
 
 					if (xfs_inode_clean(ip)) {
 						xfs_ifunlock(ip);
@@ -2257,7 +2263,9 @@ xfs_ifree_cluster(
 				AIL_LOCK(mp,s);
 				iip->ili_flush_lsn = iip->ili_item.li_lsn;
 				AIL_UNLOCK(mp, s);
+				spin_lock(&iip->ili_inode->i_flags_lock);
 				iip->ili_inode->i_flags |= XFS_ISTALE;
+				spin_unlock(&iip->ili_inode->i_flags_lock);
 				pre_flushed++;
 			}
 			lip = lip->li_bio_list;
@@ -2753,19 +2761,29 @@ xfs_iunpin(
 		 * call as the inode reclaim may be blocked waiting for
 		 * the inode to become unpinned.
 		 */
+		struct inode *inode = NULL;
+
+		spin_lock(&ip->i_flags_lock);
 		if (!(ip->i_flags & (XFS_IRECLAIM|XFS_IRECLAIMABLE))) {
 			bhv_vnode_t	*vp = XFS_ITOV_NULL(ip);
 
 			/* make sync come back and flush this inode */
 			if (vp) {
-				struct inode	*inode = vn_to_inode(vp);
+				inode = vn_to_inode(vp);
 
 				if (!(inode->i_state &
-						(I_NEW|I_FREEING|I_CLEAR)))
-					mark_inode_dirty_sync(inode);
+						(I_NEW|I_FREEING|I_CLEAR))) {
+					inode = igrab(inode);
+					if (inode)
+						mark_inode_dirty_sync(inode);
+				} else
+					inode = NULL;
 			}
 		}
+		spin_unlock(&ip->i_flags_lock);
 		wake_up(&ip->i_ipin_wait);
+		if (inode)
+			iput(inode);
 	}
 }
 

fs/xfs/xfs_inode.h

@@ -267,6 +267,7 @@ typedef struct xfs_inode {
 	sema_t			i_flock;	/* inode flush lock */
 	atomic_t		i_pincount;	/* inode pin count */
 	wait_queue_head_t	i_ipin_wait;	/* inode pinning wait queue */
+	spinlock_t		i_flags_lock;	/* inode i_flags lock */
 #ifdef HAVE_REFCACHE
 	struct xfs_inode	**i_refcache;	/* ptr to entry in ref cache */
 	struct xfs_inode	*i_release;	/* inode to unref */
@@ -389,11 +390,14 @@ typedef struct xfs_inode {
 	(((vfsp)->vfs_flag & VFS_GRPID) || ((pip)->i_d.di_mode & S_ISGID))
 
 /*
- * xfs_iget.c prototypes.
+ * Flags for xfs_iget()
  */
+#define XFS_IGET_CREATE		0x1
+#define XFS_IGET_BULKSTAT	0x2
 
-#define IGET_CREATE	1
-
+/*
+ * xfs_iget.c prototypes.
+ */
 void		xfs_ihash_init(struct xfs_mount *);
 void		xfs_ihash_free(struct xfs_mount *);
 void		xfs_chash_init(struct xfs_mount *);
@@ -425,7 +429,7 @@ int		xfs_itobp(struct xfs_mount *, struct xfs_trans *,
 			  xfs_inode_t *, xfs_dinode_t **, struct xfs_buf **,
 			  xfs_daddr_t, uint);
 int		xfs_iread(struct xfs_mount *, struct xfs_trans *, xfs_ino_t,
-			  xfs_inode_t **, xfs_daddr_t);
+			  xfs_inode_t **, xfs_daddr_t, uint);
 int		xfs_iread_extents(struct xfs_trans *, xfs_inode_t *, int);
 int		xfs_ialloc(struct xfs_trans *, xfs_inode_t *, mode_t,
 			   xfs_nlink_t, xfs_dev_t, struct cred *, xfs_prid_t,

fs/xfs/xfs_inode_item.c

@@ -742,21 +742,6 @@ xfs_inode_item_committed(
 	return (lsn);
 }
 
-/*
- * The transaction with the inode locked has aborted.  The inode
- * must not be dirty within the transaction (unless we're forcibly
- * shutting down).  We simply unlock just as if the transaction
- * had been cancelled.
- */
-STATIC void
-xfs_inode_item_abort(
-	xfs_inode_log_item_t	*iip)
-{
-	xfs_inode_item_unlock(iip);
-	return;
-}
-
-
 /*
  * This gets called by xfs_trans_push_ail(), when IOP_TRYLOCK
  * failed to get the inode flush lock but did get the inode locked SHARED.
@@ -915,7 +900,6 @@ STATIC struct xfs_item_ops xfs_inode_item_ops = {
 	.iop_committed	= (xfs_lsn_t(*)(xfs_log_item_t*, xfs_lsn_t))
 					xfs_inode_item_committed,
 	.iop_push	= (void(*)(xfs_log_item_t*))xfs_inode_item_push,
-	.iop_abort	= (void(*)(xfs_log_item_t*))xfs_inode_item_abort,
 	.iop_pushbuf	= (void(*)(xfs_log_item_t*))xfs_inode_item_pushbuf,
 	.iop_committing = (void(*)(xfs_log_item_t*, xfs_lsn_t))
 					xfs_inode_item_committing

fs/xfs/xfs_inode_item.h

@@ -25,52 +25,54 @@
  * must be added on to the end.
  */
 typedef struct xfs_inode_log_format {
-	unsigned short		ilf_type;	/* inode log item type */
-	unsigned short		ilf_size;	/* size of this item */
-	uint			ilf_fields;	/* flags for fields logged */
-	ushort			ilf_asize;	/* size of attr d/ext/root */
-	ushort			ilf_dsize;	/* size of data/ext/root */
-	xfs_ino_t		ilf_ino;	/* inode number */
+	__uint16_t		ilf_type;	/* inode log item type */
+	__uint16_t		ilf_size;	/* size of this item */
+	__uint32_t		ilf_fields;	/* flags for fields logged */
+	__uint16_t		ilf_asize;	/* size of attr d/ext/root */
+	__uint16_t		ilf_dsize;	/* size of data/ext/root */
+	__uint64_t		ilf_ino;	/* inode number */
 	union {
-		xfs_dev_t	ilfu_rdev;	/* rdev value for dev inode*/
+		__uint32_t	ilfu_rdev;	/* rdev value for dev inode*/
 		uuid_t		ilfu_uuid;	/* mount point value */
 	} ilf_u;
 	__int64_t		ilf_blkno;	/* blkno of inode buffer */
-	int			ilf_len;	/* len of inode buffer */
-	int			ilf_boffset;	/* off of inode in buffer */
+	__int32_t		ilf_len;	/* len of inode buffer */
+	__int32_t		ilf_boffset;	/* off of inode in buffer */
 } xfs_inode_log_format_t;
 
+#ifndef HAVE_FORMAT32
 typedef struct xfs_inode_log_format_32 {
-	unsigned short		ilf_type;	/* 16: inode log item type */
-	unsigned short		ilf_size;	/* 16: size of this item */
-	uint			ilf_fields;	/* 32: flags for fields logged */
-	ushort			ilf_asize;	/* 32: size of attr d/ext/root */
-	ushort			ilf_dsize;	/* 32: size of data/ext/root */
-	xfs_ino_t		ilf_ino;	/* 64: inode number */
+	__uint16_t		ilf_type;	/* inode log item type */
+	__uint16_t		ilf_size;	/* size of this item */
+	__uint32_t		ilf_fields;	/* flags for fields logged */
+	__uint16_t		ilf_asize;	/* size of attr d/ext/root */
+	__uint16_t		ilf_dsize;	/* size of data/ext/root */
+	__uint64_t		ilf_ino;	/* inode number */
 	union {
-		xfs_dev_t	ilfu_rdev;	/* 32: rdev value for dev inode*/
-		uuid_t		ilfu_uuid;	/* 128: mount point value */
+		__uint32_t	ilfu_rdev;	/* rdev value for dev inode*/
+		uuid_t		ilfu_uuid;	/* mount point value */
 	} ilf_u;
-	__int64_t		ilf_blkno;	/* 64: blkno of inode buffer */
-	int			ilf_len;	/* 32: len of inode buffer */
-	int			ilf_boffset;	/* 32: off of inode in buffer */
+	__int64_t		ilf_blkno;	/* blkno of inode buffer */
+	__int32_t		ilf_len;	/* len of inode buffer */
+	__int32_t		ilf_boffset;	/* off of inode in buffer */
 } __attribute__((packed)) xfs_inode_log_format_32_t;
+#endif
 
 typedef struct xfs_inode_log_format_64 {
-	unsigned short		ilf_type;	/* 16: inode log item type */
-	unsigned short		ilf_size;	/* 16: size of this item */
-	uint			ilf_fields;	/* 32: flags for fields logged */
-	ushort			ilf_asize;	/* 32: size of attr d/ext/root */
-	ushort			ilf_dsize;	/* 32: size of data/ext/root */
-	__uint32_t		ilf_pad;	/* 32: pad for 64 bit boundary */
-	xfs_ino_t		ilf_ino;	/* 64: inode number */
+	__uint16_t		ilf_type;	/* inode log item type */
+	__uint16_t		ilf_size;	/* size of this item */
+	__uint32_t		ilf_fields;	/* flags for fields logged */
+	__uint16_t		ilf_asize;	/* size of attr d/ext/root */
+	__uint16_t		ilf_dsize;	/* size of data/ext/root */
+	__uint32_t		ilf_pad;	/* pad for 64 bit boundary */
+	__uint64_t		ilf_ino;	/* inode number */
 	union {
-		xfs_dev_t	ilfu_rdev;	/* 32: rdev value for dev inode*/
-		uuid_t		ilfu_uuid;	/* 128: mount point value */
+		__uint32_t	ilfu_rdev;	/* rdev value for dev inode*/
+		uuid_t		ilfu_uuid;	/* mount point value */
 	} ilf_u;
-	__int64_t		ilf_blkno;	/* 64: blkno of inode buffer */
-	int			ilf_len;	/* 32: len of inode buffer */
-	int			ilf_boffset;	/* 32: off of inode in buffer */
+	__int64_t		ilf_blkno;	/* blkno of inode buffer */
+	__int32_t		ilf_len;	/* len of inode buffer */
+	__int32_t		ilf_boffset;	/* off of inode in buffer */
 } xfs_inode_log_format_64_t;
 
 /*

fs/xfs/xfs_iomap.c

@@ -398,6 +398,23 @@ xfs_flush_space(
 	return 1;
 }
 
+STATIC int
+xfs_cmn_err_fsblock_zero(
+	xfs_inode_t	*ip,
+	xfs_bmbt_irec_t	*imap)
+{
+	xfs_cmn_err(XFS_PTAG_FSBLOCK_ZERO, CE_ALERT, ip->i_mount,
+			"Access to block zero in inode %llu "
+			"start_block: %llx start_off: %llx "
+			"blkcnt: %llx extent-state: %x\n",
+		(unsigned long long)ip->i_ino,
+		(unsigned long long)imap->br_startblock,
+		(unsigned long long)imap->br_startoff,
+		(unsigned long long)imap->br_blockcount,
+		imap->br_state);
+	return EFSCORRUPTED;
+}
+
 int
 xfs_iomap_write_direct(
 	xfs_inode_t	*ip,
@@ -536,23 +553,17 @@ xfs_iomap_write_direct(
 	 * Copy any maps to caller's array and return any error.
 	 */
 	if (nimaps == 0) {
-		error = (ENOSPC);
+		error = ENOSPC;
+		goto error_out;
+	}
+
+	if (unlikely(!imap.br_startblock && !(io->io_flags & XFS_IOCORE_RT))) {
+		error = xfs_cmn_err_fsblock_zero(ip, &imap);
 		goto error_out;
 	}
 
 	*ret_imap = imap;
 	*nmaps = 1;
-	if ( !(io->io_flags & XFS_IOCORE_RT)  && !ret_imap->br_startblock) {
-                cmn_err(CE_PANIC,"Access to block zero:  fs <%s> inode: %lld "
-                        "start_block : %llx start_off : %llx blkcnt : %llx "
-                        "extent-state : %x \n",
-                        (ip->i_mount)->m_fsname,
-                        (long long)ip->i_ino,
-                        (unsigned long long)ret_imap->br_startblock,
-			(unsigned long long)ret_imap->br_startoff,
-                        (unsigned long long)ret_imap->br_blockcount,
-			ret_imap->br_state);
-        }
 	return 0;
 
 error0:	/* Cancel bmap, unlock inode, unreserve quota blocks, cancel trans */
@@ -715,17 +726,8 @@ retry:
 		goto retry;
 	}
 
-	if (!(io->io_flags & XFS_IOCORE_RT)  && !ret_imap->br_startblock) {
-		cmn_err(CE_PANIC,"Access to block zero:  fs <%s> inode: %lld "
-                        "start_block : %llx start_off : %llx blkcnt : %llx "
-                        "extent-state : %x \n",
-                        (ip->i_mount)->m_fsname,
-                        (long long)ip->i_ino,
-                        (unsigned long long)ret_imap->br_startblock,
-			(unsigned long long)ret_imap->br_startoff,
-                        (unsigned long long)ret_imap->br_blockcount,
-			ret_imap->br_state);
-	}
+	if (unlikely(!imap[0].br_startblock && !(io->io_flags & XFS_IOCORE_RT)))
+		return xfs_cmn_err_fsblock_zero(ip, &imap[0]);
 
 	*ret_imap = imap[0];
 	*nmaps = 1;
@@ -853,24 +855,10 @@ xfs_iomap_write_allocate(
 		 * See if we were able to allocate an extent that
 		 * covers at least part of the callers request
 		 */
-
 		for (i = 0; i < nimaps; i++) {
-			if (!(io->io_flags & XFS_IOCORE_RT)  &&
-			    !imap[i].br_startblock) {
-				cmn_err(CE_PANIC,"Access to block zero:  "
-					"fs <%s> inode: %lld "
-					"start_block : %llx start_off : %llx "
-					"blkcnt : %llx extent-state : %x \n",
-					(ip->i_mount)->m_fsname,
-					(long long)ip->i_ino,
-					(unsigned long long)
-						imap[i].br_startblock,
-					(unsigned long long)
-						imap[i].br_startoff,
-					(unsigned long long)
-				        	imap[i].br_blockcount,
-					imap[i].br_state);
-                        }
+			if (unlikely(!imap[i].br_startblock &&
+				     !(io->io_flags & XFS_IOCORE_RT)))
+				return xfs_cmn_err_fsblock_zero(ip, &imap[i]);
 			if ((offset_fsb >= imap[i].br_startoff) &&
 			    (offset_fsb < (imap[i].br_startoff +
 					   imap[i].br_blockcount))) {
@@ -941,7 +929,7 @@ xfs_iomap_write_unwritten(
 				XFS_WRITE_LOG_COUNT);
 		if (error) {
 			xfs_trans_cancel(tp, 0);
-			goto error0;
+			return XFS_ERROR(error);
 		}
 
 		xfs_ilock(ip, XFS_ILOCK_EXCL);
@@ -967,19 +955,11 @@ xfs_iomap_write_unwritten(
 		error = xfs_trans_commit(tp, XFS_TRANS_RELEASE_LOG_RES, NULL);
 		xfs_iunlock(ip, XFS_ILOCK_EXCL);
 		if (error)
-			goto error0;
-
-		if ( !(io->io_flags & XFS_IOCORE_RT)  && !imap.br_startblock) {
-			cmn_err(CE_PANIC,"Access to block zero:  fs <%s> "
-				"inode: %lld start_block : %llx start_off : "
-				"%llx blkcnt : %llx extent-state : %x \n",
-				(ip->i_mount)->m_fsname,
-				(long long)ip->i_ino,
-				(unsigned long long)imap.br_startblock,
-				(unsigned long long)imap.br_startoff,
-				(unsigned long long)imap.br_blockcount,
-				imap.br_state);
-        	}
+			return XFS_ERROR(error);
+
+		if (unlikely(!imap.br_startblock &&
+			     !(io->io_flags & XFS_IOCORE_RT)))
+			return xfs_cmn_err_fsblock_zero(ip, &imap);
 
 		if ((numblks_fsb = imap.br_blockcount) == 0) {
 			/*
@@ -999,6 +979,5 @@ error_on_bmapi_transaction:
 	xfs_bmap_cancel(&free_list);
 	xfs_trans_cancel(tp, (XFS_TRANS_RELEASE_LOG_RES | XFS_TRANS_ABORT));
 	xfs_iunlock(ip, XFS_ILOCK_EXCL);
-error0:
 	return XFS_ERROR(error);
 }

fs/xfs/xfs_itable.c

@@ -39,6 +39,16 @@
 #include "xfs_error.h"
 #include "xfs_btree.h"
 
+int
+xfs_internal_inum(
+	xfs_mount_t	*mp,
+	xfs_ino_t	ino)
+{
+	return (ino == mp->m_sb.sb_rbmino || ino == mp->m_sb.sb_rsumino ||
+		(XFS_SB_VERSION_HASQUOTA(&mp->m_sb) &&
+		 (ino == mp->m_sb.sb_uquotino || ino == mp->m_sb.sb_gquotino)));
+}
+
 STATIC int
 xfs_bulkstat_one_iget(
 	xfs_mount_t	*mp,		/* mount point for filesystem */
@@ -52,7 +62,8 @@ xfs_bulkstat_one_iget(
 	bhv_vnode_t	*vp;
 	int		error;
 
-	error = xfs_iget(mp, NULL, ino, 0, XFS_ILOCK_SHARED, &ip, bno);
+	error = xfs_iget(mp, NULL, ino,
+			 XFS_IGET_BULKSTAT, XFS_ILOCK_SHARED, &ip, bno);
 	if (error) {
 		*stat = BULKSTAT_RV_NOTHING;
 		return error;
@@ -212,17 +223,12 @@ xfs_bulkstat_one(
 	xfs_dinode_t	*dip;		/* dinode inode pointer */
 
 	dip = (xfs_dinode_t *)dibuff;
+	*stat = BULKSTAT_RV_NOTHING;
 
-	if (!buffer || ino == mp->m_sb.sb_rbmino || ino == mp->m_sb.sb_rsumino ||
-	    (XFS_SB_VERSION_HASQUOTA(&mp->m_sb) &&
-	     (ino == mp->m_sb.sb_uquotino || ino == mp->m_sb.sb_gquotino))) {
-		*stat = BULKSTAT_RV_NOTHING;
+	if (!buffer || xfs_internal_inum(mp, ino))
 		return XFS_ERROR(EINVAL);
-	}
-	if (ubsize < sizeof(*buf)) {
-		*stat = BULKSTAT_RV_NOTHING;
+	if (ubsize < sizeof(*buf))
 		return XFS_ERROR(ENOMEM);
-	}
 
 	buf = kmem_alloc(sizeof(*buf), KM_SLEEP);
 
@@ -238,8 +244,7 @@ xfs_bulkstat_one(
 	}
 
 	if (copy_to_user(buffer, buf, sizeof(*buf)))  {
-		*stat = BULKSTAT_RV_NOTHING;
-		error =  EFAULT;
+		error = EFAULT;
 		goto out_free;
 	}
 
@@ -252,6 +257,46 @@ xfs_bulkstat_one(
 	return error;
 }
 
+/*
+ * Test to see whether we can use the ondisk inode directly, based
+ * on the given bulkstat flags, filling in dipp accordingly.
+ * Returns zero if the inode is dodgey.
+ */
+STATIC int
+xfs_bulkstat_use_dinode(
+	xfs_mount_t	*mp,
+	int		flags,
+	xfs_buf_t	*bp,
+	int		clustidx,
+	xfs_dinode_t	**dipp)
+{
+	xfs_dinode_t	*dip;
+	unsigned int	aformat;
+
+	*dipp = NULL;
+	if (!bp || (flags & BULKSTAT_FG_IGET))
+		return 1;
+	dip = (xfs_dinode_t *)
+			xfs_buf_offset(bp, clustidx << mp->m_sb.sb_inodelog);
+	if (INT_GET(dip->di_core.di_magic, ARCH_CONVERT) != XFS_DINODE_MAGIC ||
+	    !XFS_DINODE_GOOD_VERSION(
+			INT_GET(dip->di_core.di_version, ARCH_CONVERT)))
+		return 0;
+	if (flags & BULKSTAT_FG_QUICK) {
+		*dipp = dip;
+		return 1;
+	}
+	/* BULKSTAT_FG_INLINE: if attr fork is local, or not there, use it */
+	aformat = INT_GET(dip->di_core.di_aformat, ARCH_CONVERT);
+	if ((XFS_CFORK_Q(&dip->di_core) == 0) ||
+	    (aformat == XFS_DINODE_FMT_LOCAL) ||
+	    (aformat == XFS_DINODE_FMT_EXTENTS && !dip->di_core.di_anextents)) {
+		*dipp = dip;
+		return 1;
+	}
+	return 1;
+}
+
 /*
  * Return stat information in bulk (by-inode) for the filesystem.
  */
@@ -284,10 +329,11 @@ xfs_bulkstat(
 	xfs_agino_t		gino;	/* current btree rec's start inode */
 	int			i;	/* loop index */
 	int			icount;	/* count of inodes good in irbuf */
+	size_t			irbsize; /* size of irec buffer in bytes */
 	xfs_ino_t		ino;	/* inode number (filesystem) */
-	xfs_inobt_rec_t		*irbp;	/* current irec buffer pointer */
-	xfs_inobt_rec_t		*irbuf;	/* start of irec buffer */
-	xfs_inobt_rec_t		*irbufend; /* end of good irec buffer entries */
+	xfs_inobt_rec_incore_t	*irbp;	/* current irec buffer pointer */
+	xfs_inobt_rec_incore_t	*irbuf;	/* start of irec buffer */
+	xfs_inobt_rec_incore_t	*irbufend; /* end of good irec buffer entries */
 	xfs_ino_t		lastino=0; /* last inode number returned */
 	int			nbcluster; /* # of blocks in a cluster */
 	int			nicluster; /* # of inodes in a cluster */
@@ -328,13 +374,10 @@ xfs_bulkstat(
 		(XFS_INODE_CLUSTER_SIZE(mp) >> mp->m_sb.sb_inodelog);
 	nimask = ~(nicluster - 1);
 	nbcluster = nicluster >> mp->m_sb.sb_inopblog;
-	/*
-	 * Allocate a page-sized buffer for inode btree records.
-	 * We could try allocating something smaller, but for normal
-	 * calls we'll always (potentially) need the whole page.
-	 */
-	irbuf = kmem_alloc(NBPC, KM_SLEEP);
-	nirbuf = NBPC / sizeof(*irbuf);
+	irbuf = kmem_zalloc_greedy(&irbsize, NBPC, NBPC * 4,
+				   KM_SLEEP | KM_MAYFAIL | KM_LARGE);
+	nirbuf = irbsize / sizeof(*irbuf);
+
 	/*
 	 * Loop over the allocation groups, starting from the last
 	 * inode returned; 0 means start of the allocation group.
@@ -358,7 +401,7 @@ xfs_bulkstat(
 		 * Allocate and initialize a btree cursor for ialloc btree.
 		 */
 		cur = xfs_btree_init_cursor(mp, NULL, agbp, agno, XFS_BTNUM_INO,
-			(xfs_inode_t *)0, 0);
+						(xfs_inode_t *)0, 0);
 		irbp = irbuf;
 		irbufend = irbuf + nirbuf;
 		end_of_ag = 0;
@@ -395,9 +438,9 @@ xfs_bulkstat(
 						gcnt++;
 				}
 				gfree |= XFS_INOBT_MASKN(0, chunkidx);
-				INT_SET(irbp->ir_startino, ARCH_CONVERT, gino);
-				INT_SET(irbp->ir_freecount, ARCH_CONVERT, gcnt);
-				INT_SET(irbp->ir_free, ARCH_CONVERT, gfree);
+				irbp->ir_startino = gino;
+				irbp->ir_freecount = gcnt;
+				irbp->ir_free = gfree;
 				irbp++;
 				agino = gino + XFS_INODES_PER_CHUNK;
 				icount = XFS_INODES_PER_CHUNK - gcnt;
@@ -451,11 +494,27 @@ xfs_bulkstat(
 			}
 			/*
 			 * If this chunk has any allocated inodes, save it.
+			 * Also start read-ahead now for this chunk.
 			 */
 			if (gcnt < XFS_INODES_PER_CHUNK) {
-				INT_SET(irbp->ir_startino, ARCH_CONVERT, gino);
-				INT_SET(irbp->ir_freecount, ARCH_CONVERT, gcnt);
-				INT_SET(irbp->ir_free, ARCH_CONVERT, gfree);
+				/*
+				 * Loop over all clusters in the next chunk.
+				 * Do a readahead if there are any allocated
+				 * inodes in that cluster.
+				 */
+				for (agbno = XFS_AGINO_TO_AGBNO(mp, gino),
+				     chunkidx = 0;
+				     chunkidx < XFS_INODES_PER_CHUNK;
+				     chunkidx += nicluster,
+				     agbno += nbcluster) {
+					if (XFS_INOBT_MASKN(chunkidx,
+							    nicluster) & ~gfree)
+						xfs_btree_reada_bufs(mp, agno,
+							agbno, nbcluster);
+				}
+				irbp->ir_startino = gino;
+				irbp->ir_freecount = gcnt;
+				irbp->ir_free = gfree;
 				irbp++;
 				icount += XFS_INODES_PER_CHUNK - gcnt;
 			}
@@ -478,34 +537,12 @@ xfs_bulkstat(
 		irbufend = irbp;
 		for (irbp = irbuf;
 		     irbp < irbufend && ubleft >= statstruct_size; irbp++) {
-			/*
-			 * Read-ahead the next chunk's worth of inodes.
-			 */
-			if (&irbp[1] < irbufend) {
-				/*
-				 * Loop over all clusters in the next chunk.
-				 * Do a readahead if there are any allocated
-				 * inodes in that cluster.
-				 */
-				for (agbno = XFS_AGINO_TO_AGBNO(mp,
-							INT_GET(irbp[1].ir_startino, ARCH_CONVERT)),
-				     chunkidx = 0;
-				     chunkidx < XFS_INODES_PER_CHUNK;
-				     chunkidx += nicluster,
-				     agbno += nbcluster) {
-					if (XFS_INOBT_MASKN(chunkidx,
-							    nicluster) &
-					    ~(INT_GET(irbp[1].ir_free, ARCH_CONVERT)))
-						xfs_btree_reada_bufs(mp, agno,
-							agbno, nbcluster);
-				}
-			}
 			/*
 			 * Now process this chunk of inodes.
 			 */
-			for (agino = INT_GET(irbp->ir_startino, ARCH_CONVERT), chunkidx = 0, clustidx = 0;
+			for (agino = irbp->ir_startino, chunkidx = clustidx = 0;
 			     ubleft > 0 &&
-				INT_GET(irbp->ir_freecount, ARCH_CONVERT) < XFS_INODES_PER_CHUNK;
+				irbp->ir_freecount < XFS_INODES_PER_CHUNK;
 			     chunkidx++, clustidx++, agino++) {
 				ASSERT(chunkidx < XFS_INODES_PER_CHUNK);
 				/*
@@ -525,11 +562,12 @@ xfs_bulkstat(
 				 */
 				if ((chunkidx & (nicluster - 1)) == 0) {
 					agbno = XFS_AGINO_TO_AGBNO(mp,
-							INT_GET(irbp->ir_startino, ARCH_CONVERT)) +
+							irbp->ir_startino) +
 						((chunkidx & nimask) >>
 						 mp->m_sb.sb_inopblog);
 
-					if (flags & BULKSTAT_FG_QUICK) {
+					if (flags & (BULKSTAT_FG_QUICK |
+						     BULKSTAT_FG_INLINE)) {
 						ino = XFS_AGINO_TO_INO(mp, agno,
 								       agino);
 						bno = XFS_AGB_TO_DADDR(mp, agno,
@@ -543,6 +581,7 @@ xfs_bulkstat(
 								      KM_SLEEP);
 						ip->i_ino = ino;
 						ip->i_mount = mp;
+						spin_lock_init(&ip->i_flags_lock);
 						if (bp)
 							xfs_buf_relse(bp);
 						error = xfs_itobp(mp, NULL, ip,
@@ -564,30 +603,34 @@ xfs_bulkstat(
 				/*
 				 * Skip if this inode is free.
 				 */
-				if (XFS_INOBT_MASK(chunkidx) & INT_GET(irbp->ir_free, ARCH_CONVERT))
+				if (XFS_INOBT_MASK(chunkidx) & irbp->ir_free)
 					continue;
 				/*
 				 * Count used inodes as free so we can tell
 				 * when the chunk is used up.
 				 */
-				INT_MOD(irbp->ir_freecount, ARCH_CONVERT, +1);
+				irbp->ir_freecount++;
 				ino = XFS_AGINO_TO_INO(mp, agno, agino);
 				bno = XFS_AGB_TO_DADDR(mp, agno, agbno);
-				if (flags & BULKSTAT_FG_QUICK) {
-					dip = (xfs_dinode_t *)xfs_buf_offset(bp,
-					      (clustidx << mp->m_sb.sb_inodelog));
-
-					if (INT_GET(dip->di_core.di_magic, ARCH_CONVERT)
-						    != XFS_DINODE_MAGIC
-					    || !XFS_DINODE_GOOD_VERSION(
-						    INT_GET(dip->di_core.di_version, ARCH_CONVERT)))
-						continue;
+				if (!xfs_bulkstat_use_dinode(mp, flags, bp,
+							     clustidx, &dip))
+					continue;
+				/*
+				 * If we need to do an iget, cannot hold bp.
+				 * Drop it, until starting the next cluster.
+				 */
+				if ((flags & BULKSTAT_FG_INLINE) && !dip) {
+					if (bp)
+						xfs_buf_relse(bp);
+					bp = NULL;
 				}
 
 				/*
 				 * Get the inode and fill in a single buffer.
 				 * BULKSTAT_FG_QUICK uses dip to fill it in.
 				 * BULKSTAT_FG_IGET uses igets.
+				 * BULKSTAT_FG_INLINE uses dip if we have an
+				 * inline attr fork, else igets.
 				 * See: xfs_bulkstat_one & xfs_dm_bulkstat_one.
 				 * This is also used to count inodes/blks, etc
 				 * in xfs_qm_quotacheck.
@@ -597,8 +640,15 @@ xfs_bulkstat(
 						ubleft, private_data,
 						bno, &ubused, dip, &fmterror);
 				if (fmterror == BULKSTAT_RV_NOTHING) {
-					if (error == ENOMEM)
+                                        if (error == EFAULT) {
+                                                ubleft = 0;
+                                                rval = error;
+                                                break;
+                                        }
+					else if (error == ENOMEM)
 						ubleft = 0;
+					else
+						lastino = ino;
 					continue;
 				}
 				if (fmterror == BULKSTAT_RV_GIVEUP) {
@@ -633,7 +683,7 @@ xfs_bulkstat(
 	/*
 	 * Done, we're either out of filesystem or space to put the data.
 	 */
-	kmem_free(irbuf, NBPC);
+	kmem_free(irbuf, irbsize);
 	*ubcountp = ubelem;
 	if (agno >= mp->m_sb.sb_agcount) {
 		/*

fs/xfs/xfs_itable.h

@@ -36,15 +36,16 @@ typedef int (*bulkstat_one_pf)(struct xfs_mount	*mp,
 /*
  * Values for stat return value.
  */
-#define	BULKSTAT_RV_NOTHING	0
-#define	BULKSTAT_RV_DIDONE	1
-#define	BULKSTAT_RV_GIVEUP	2
+#define BULKSTAT_RV_NOTHING	0
+#define BULKSTAT_RV_DIDONE	1
+#define BULKSTAT_RV_GIVEUP	2
 
 /*
  * Values for bulkstat flag argument.
  */
-#define	BULKSTAT_FG_IGET	0x1	/* Go through the buffer cache */
-#define	BULKSTAT_FG_QUICK	0x2	/* No iget, walk the dinode cluster */
+#define BULKSTAT_FG_IGET	0x1	/* Go through the buffer cache */
+#define BULKSTAT_FG_QUICK	0x2	/* No iget, walk the dinode cluster */
+#define BULKSTAT_FG_INLINE	0x4	/* No iget if inline attrs */
 
 /*
  * Return stat information in bulk (by-inode) for the filesystem.
@@ -80,6 +81,11 @@ xfs_bulkstat_one(
 	void			*dibuff,
 	int			*stat);
 
+int
+xfs_internal_inum(
+	xfs_mount_t		*mp,
+	xfs_ino_t		ino);
+
 int					/* error status */
 xfs_inumbers(
 	xfs_mount_t		*mp,	/* mount point for filesystem */

fs/xfs/xfs_log.c

@@ -617,7 +617,8 @@ xfs_log_unmount_write(xfs_mount_t *mp)
 		reg[0].i_len  = sizeof(magic);
 		XLOG_VEC_SET_TYPE(&reg[0], XLOG_REG_TYPE_UNMOUNT);
 
-		error = xfs_log_reserve(mp, 600, 1, &tic, XFS_LOG, 0, 0);
+		error = xfs_log_reserve(mp, 600, 1, &tic,
+					XFS_LOG, 0, XLOG_UNMOUNT_REC_TYPE);
 		if (!error) {
 			/* remove inited flag */
 			((xlog_ticket_t *)tic)->t_flags = 0;
@@ -655,8 +656,11 @@ xfs_log_unmount_write(xfs_mount_t *mp)
 		} else {
 			LOG_UNLOCK(log, s);
 		}
-		if (tic)
+		if (tic) {
+			xlog_trace_loggrant(log, tic, "unmount rec");
+			xlog_ungrant_log_space(log, tic);
 			xlog_state_put_ticket(log, tic);
+		}
 	} else {
 		/*
 		 * We're already in forced_shutdown mode, couldn't
@@ -1196,7 +1200,7 @@ xlog_alloc_log(xfs_mount_t	*mp,
 			  kmem_zalloc(sizeof(xlog_in_core_t), KM_SLEEP);
 		iclog = *iclogp;
 		iclog->hic_data = (xlog_in_core_2_t *)
-			  kmem_zalloc(iclogsize, KM_SLEEP);
+			  kmem_zalloc(iclogsize, KM_SLEEP | KM_LARGE);
 
 		iclog->ic_prev = prev_iclog;
 		prev_iclog = iclog;
@@ -2212,9 +2216,13 @@ xlog_state_do_callback(
 
 			iclog = iclog->ic_next;
 		} while (first_iclog != iclog);
-		if (repeats && (repeats % 10) == 0) {
+
+		if (repeats > 5000) {
+			flushcnt += repeats;
+			repeats = 0;
 			xfs_fs_cmn_err(CE_WARN, log->l_mp,
-				"xlog_state_do_callback: looping %d", repeats);
+				"%s: possible infinite loop (%d iterations)",
+				__FUNCTION__, flushcnt);
 		}
 	} while (!ioerrors && loopdidcallbacks);
 
@@ -2246,6 +2254,7 @@ xlog_state_do_callback(
 	}
 #endif
 
+	flushcnt = 0;
 	if (log->l_iclog->ic_state & (XLOG_STATE_ACTIVE|XLOG_STATE_IOERROR)) {
 		flushcnt = log->l_flushcnt;
 		log->l_flushcnt = 0;

fs/xfs/xfs_log.h

@@ -47,17 +47,11 @@ static inline xfs_lsn_t	_lsn_cmp(xfs_lsn_t lsn1, xfs_lsn_t lsn2)
  * Macros, structures, prototypes for interface to the log manager.
  */
 
-/*
- * Flags to xfs_log_mount
- */
-#define XFS_LOG_RECOVER		0x1
-
 /*
  * Flags to xfs_log_done()
  */
 #define XFS_LOG_REL_PERM_RESERV	0x1
 
-
 /*
  * Flags to xfs_log_reserve()
  *
@@ -70,8 +64,6 @@ static inline xfs_lsn_t	_lsn_cmp(xfs_lsn_t lsn1, xfs_lsn_t lsn2)
 #define XFS_LOG_SLEEP		0x0
 #define XFS_LOG_NOSLEEP		0x1
 #define XFS_LOG_PERM_RESERV	0x2
-#define XFS_LOG_RESV_ALL	(XFS_LOG_NOSLEEP|XFS_LOG_PERM_RESERV)
-
 
 /*
  * Flags to xfs_log_force()

fs/xfs/xfs_log_priv.h

@@ -32,7 +32,6 @@ struct xfs_mount;
 #define XLOG_MIN_ICLOGS		2
 #define XLOG_MED_ICLOGS		4
 #define XLOG_MAX_ICLOGS		8
-#define XLOG_CALLBACK_SIZE	10
 #define XLOG_HEADER_MAGIC_NUM	0xFEEDbabe	/* Invalid cycle number */
 #define XLOG_VERSION_1		1
 #define XLOG_VERSION_2		2		/* Large IClogs, Log sunit */
@@ -149,9 +148,6 @@ struct xfs_mount;
 #define XLOG_WAS_CONT_TRANS	0x08	/* Cont this trans into new region */
 #define XLOG_END_TRANS		0x10	/* End a continued transaction */
 #define XLOG_UNMOUNT_TRANS	0x20	/* Unmount a filesystem transaction */
-#define XLOG_SKIP_TRANS		(XLOG_COMMIT_TRANS | XLOG_CONTINUE_TRANS | \
-				 XLOG_WAS_CONT_TRANS | XLOG_END_TRANS | \
-				 XLOG_UNMOUNT_TRANS)
 
 #ifdef __KERNEL__
 /*
@@ -506,6 +502,12 @@ extern int	 xlog_bread(xlog_t *, xfs_daddr_t, int, struct xfs_buf *);
 #define XLOG_TRACE_SLEEP_FLUSH 3
 #define XLOG_TRACE_WAKE_FLUSH  4
 
+/*
+ * Unmount record type is used as a pseudo transaction type for the ticket.
+ * It's value must be outside the range of XFS_TRANS_* values.
+ */
+#define XLOG_UNMOUNT_REC_TYPE	(-1U)
+
 #endif	/* __KERNEL__ */
 
 #endif	/* __XFS_LOG_PRIV_H__ */

fs/xfs/xfs_mount.h

@@ -331,7 +331,7 @@ typedef struct xfs_mount {
 	xfs_agnumber_t		m_agirotor;	/* last ag dir inode alloced */
 	lock_t			m_agirotor_lock;/* .. and lock protecting it */
 	xfs_agnumber_t		m_maxagi;	/* highest inode alloc group */
-	uint			m_ihsize;	/* size of next field */
+	size_t			m_ihsize;	/* size of next field */
 	struct xfs_ihash	*m_ihash;	/* fs private inode hash table*/
 	struct xfs_inode	*m_inodes;	/* active inode list */
 	struct list_head	m_del_inodes;	/* inodes to reclaim */
@@ -541,7 +541,8 @@ static inline xfs_mount_t *xfs_bhvtom(bhv_desc_t *bdp)
 #define XFS_VFSTOM(vfs) xfs_vfstom(vfs)
 static inline xfs_mount_t *xfs_vfstom(bhv_vfs_t *vfs)
 {
-	return XFS_BHVTOM(bhv_lookup(VFS_BHVHEAD(vfs), &xfs_vfsops));
+	return XFS_BHVTOM(bhv_lookup_range(VFS_BHVHEAD(vfs),
+				VFS_POSITION_XFS, VFS_POSITION_XFS));
 }
 
 #define XFS_DADDR_TO_AGNO(mp,d)         xfs_daddr_to_agno(mp,d)

fs/xfs/xfs_quota.h

@@ -281,8 +281,6 @@ typedef struct xfs_qoff_logformat {
 				 XFS_UQUOTA_CHKD|XFS_PQUOTA_ACCT|\
 				 XFS_OQUOTA_ENFD|XFS_OQUOTA_CHKD|\
 				 XFS_GQUOTA_ACCT)
-#define XFS_MOUNT_QUOTA_MASK	(XFS_MOUNT_QUOTA_ALL | XFS_UQUOTA_ACTIVE | \
-				 XFS_GQUOTA_ACTIVE | XFS_PQUOTA_ACTIVE)
 
 
 /*

fs/xfs/xfs_rtalloc.c

@@ -1948,7 +1948,7 @@ xfs_growfs_rt(
 	 */
 	nrextents = nrblocks;
 	do_div(nrextents, in->extsize);
-	nrbmblocks = roundup_64(nrextents, NBBY * sbp->sb_blocksize);
+	nrbmblocks = howmany_64(nrextents, NBBY * sbp->sb_blocksize);
 	nrextslog = xfs_highbit32(nrextents);
 	nrsumlevels = nrextslog + 1;
 	nrsumsize = (uint)sizeof(xfs_suminfo_t) * nrsumlevels * nrbmblocks;
@@ -1976,7 +1976,10 @@ xfs_growfs_rt(
 	if ((error = xfs_growfs_rt_alloc(mp, rsumblocks, nrsumblocks,
 			mp->m_sb.sb_rsumino)))
 		return error;
-	nmp = NULL;
+	/*
+	 * Allocate a new (fake) mount/sb.
+	 */
+	nmp = kmem_alloc(sizeof(*nmp), KM_SLEEP);
 	/*
 	 * Loop over the bitmap blocks.
 	 * We will do everything one bitmap block at a time.
@@ -1987,10 +1990,6 @@ xfs_growfs_rt(
 		     ((sbp->sb_rextents & ((1 << mp->m_blkbit_log) - 1)) != 0);
 	     bmbno < nrbmblocks;
 	     bmbno++) {
-		/*
-		 * Allocate a new (fake) mount/sb.
-		 */
-		nmp = kmem_alloc(sizeof(*nmp), KM_SLEEP);
 		*nmp = *mp;
 		nsbp = &nmp->m_sb;
 		/*
@@ -2018,13 +2017,13 @@ xfs_growfs_rt(
 		cancelflags = 0;
 		if ((error = xfs_trans_reserve(tp, 0,
 				XFS_GROWRTFREE_LOG_RES(nmp), 0, 0, 0)))
-			goto error_exit;
+			break;
 		/*
 		 * Lock out other callers by grabbing the bitmap inode lock.
 		 */
 		if ((error = xfs_trans_iget(mp, tp, mp->m_sb.sb_rbmino, 0,
 						XFS_ILOCK_EXCL, &ip)))
-			goto error_exit;
+			break;
 		ASSERT(ip == mp->m_rbmip);
 		/*
 		 * Update the bitmap inode's size.
@@ -2038,7 +2037,7 @@ xfs_growfs_rt(
 		 */
 		if ((error = xfs_trans_iget(mp, tp, mp->m_sb.sb_rsumino, 0,
 						XFS_ILOCK_EXCL, &ip)))
-			goto error_exit;
+			break;
 		ASSERT(ip == mp->m_rsumip);
 		/*
 		 * Update the summary inode's size.
@@ -2053,7 +2052,7 @@ xfs_growfs_rt(
 		    mp->m_rsumlevels != nmp->m_rsumlevels) {
 			error = xfs_rtcopy_summary(mp, nmp, tp);
 			if (error)
-				goto error_exit;
+				break;
 		}
 		/*
 		 * Update superblock fields.
@@ -2080,17 +2079,12 @@ xfs_growfs_rt(
 		error = xfs_rtfree_range(nmp, tp, sbp->sb_rextents,
 			nsbp->sb_rextents - sbp->sb_rextents, &bp, &sumbno);
 		if (error)
-			goto error_exit;
+			break;
 		/*
 		 * Mark more blocks free in the superblock.
 		 */
 		xfs_trans_mod_sb(tp, XFS_TRANS_SB_FREXTENTS,
 			nsbp->sb_rextents - sbp->sb_rextents);
-		/*
-		 * Free the fake mp structure.
-		 */
-		kmem_free(nmp, sizeof(*nmp));
-		nmp = NULL;
 		/*
 		 * Update mp values into the real mp structure.
 		 */
@@ -2101,15 +2095,15 @@ xfs_growfs_rt(
 		 */
 		xfs_trans_commit(tp, 0, NULL);
 	}
-	return 0;
+
+	if (error)
+		xfs_trans_cancel(tp, cancelflags);
 
 	/*
-	 * Error paths come here.
+	 * Free the fake mp structure.
 	 */
-error_exit:
-	if (nmp)
-		kmem_free(nmp, sizeof(*nmp));
-	xfs_trans_cancel(tp, cancelflags);
+	kmem_free(nmp, sizeof(*nmp));
+
 	return error;
 }
 

fs/xfs/xfs_sb.h

@@ -60,10 +60,6 @@ struct xfs_mount;
 	 XFS_SB_VERSION_LOGV2BIT | \
 	 XFS_SB_VERSION_SECTORBIT | \
 	 XFS_SB_VERSION_MOREBITSBIT)
-#define	XFS_SB_VERSION_OKSASHBITS	\
-	(XFS_SB_VERSION_NUMBITS | \
-	 XFS_SB_VERSION_REALFBITS | \
-	 XFS_SB_VERSION_OKSASHFBITS)
 #define	XFS_SB_VERSION_OKREALBITS	\
 	(XFS_SB_VERSION_NUMBITS | \
 	 XFS_SB_VERSION_OKREALFBITS | \
@@ -81,9 +77,6 @@ struct xfs_mount;
 #define XFS_SB_VERSION2_RESERVED2BIT	0x00000002
 #define XFS_SB_VERSION2_RESERVED4BIT	0x00000004
 #define XFS_SB_VERSION2_ATTR2BIT	0x00000008	/* Inline attr rework */
-#define XFS_SB_VERSION2_SASHFBITS	0xff000000	/* Mask: features that
-							   require changing
-							   PROM and SASH */
 
 #define	XFS_SB_VERSION2_OKREALFBITS	\
 	(XFS_SB_VERSION2_ATTR2BIT)
@@ -238,12 +231,6 @@ static inline int xfs_sb_good_version(xfs_sb_t *sbp)
 }
 #endif /* __KERNEL__ */
 
-#define	XFS_SB_GOOD_SASH_VERSION(sbp)	\
-	((((sbp)->sb_versionnum >= XFS_SB_VERSION_1) && \
-	  ((sbp)->sb_versionnum <= XFS_SB_VERSION_3)) || \
-	 ((XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4) && \
-	  !((sbp)->sb_versionnum & ~XFS_SB_VERSION_OKSASHBITS)))
-
 #define	XFS_SB_VERSION_TONEW(v)	xfs_sb_version_tonew(v)
 static inline unsigned xfs_sb_version_tonew(unsigned v)
 {
@@ -461,15 +448,6 @@ static inline void xfs_sb_version_addattr2(xfs_sb_t *sbp)
  * File system sector to basic block conversions.
  */
 #define XFS_FSS_TO_BB(mp,sec)	((sec) << (mp)->m_sectbb_log)
-#define XFS_BB_TO_FSS(mp,bb)	\
-	(((bb) + (XFS_FSS_TO_BB(mp,1) - 1)) >> (mp)->m_sectbb_log)
-#define XFS_BB_TO_FSST(mp,bb)	((bb) >> (mp)->m_sectbb_log)
-
-/*
- * File system sector to byte conversions.
- */
-#define XFS_FSS_TO_B(mp,sectno)	((xfs_fsize_t)(sectno) << (mp)->m_sb.sb_sectlog)
-#define XFS_B_TO_FSST(mp,b)	(((__uint64_t)(b)) >> (mp)->m_sb.sb_sectlog)
 
 /*
  * File system block to basic block conversions.

fs/xfs/xfs_trans.h

@@ -149,7 +149,6 @@ typedef struct xfs_item_ops {
 	void (*iop_unlock)(xfs_log_item_t *);
 	xfs_lsn_t (*iop_committed)(xfs_log_item_t *, xfs_lsn_t);
 	void (*iop_push)(xfs_log_item_t *);
-	void (*iop_abort)(xfs_log_item_t *);
 	void (*iop_pushbuf)(xfs_log_item_t *);
 	void (*iop_committing)(xfs_log_item_t *, xfs_lsn_t);
 } xfs_item_ops_t;
@@ -163,7 +162,6 @@ typedef struct xfs_item_ops {
 #define IOP_UNLOCK(ip)		(*(ip)->li_ops->iop_unlock)(ip)
 #define IOP_COMMITTED(ip, lsn)	(*(ip)->li_ops->iop_committed)(ip, lsn)
 #define IOP_PUSH(ip)		(*(ip)->li_ops->iop_push)(ip)
-#define IOP_ABORT(ip)		(*(ip)->li_ops->iop_abort)(ip)
 #define IOP_PUSHBUF(ip)		(*(ip)->li_ops->iop_pushbuf)(ip)
 #define IOP_COMMITTING(ip, lsn) (*(ip)->li_ops->iop_committing)(ip, lsn)
 

fs/xfs/xfs_trans_ail.c

@@ -276,7 +276,7 @@ xfs_trans_update_ail(
 	xfs_mount_t	*mp,
 	xfs_log_item_t	*lip,
 	xfs_lsn_t	lsn,
-	unsigned long	s)
+	unsigned long	s) __releases(mp->m_ail_lock)
 {
 	xfs_ail_entry_t		*ailp;
 	xfs_log_item_t		*dlip=NULL;
@@ -328,7 +328,7 @@ void
 xfs_trans_delete_ail(
 	xfs_mount_t	*mp,
 	xfs_log_item_t	*lip,
-	unsigned long	s)
+	unsigned long	s) __releases(mp->m_ail_lock)
 {
 	xfs_ail_entry_t		*ailp;
 	xfs_log_item_t		*dlip;

fs/xfs/xfs_trans_priv.h

@@ -46,11 +46,13 @@ xfs_log_busy_slot_t		*xfs_trans_add_busy(xfs_trans_t *tp,
 /*
  * From xfs_trans_ail.c
  */
-void			xfs_trans_update_ail(struct xfs_mount *,
-				     struct xfs_log_item *, xfs_lsn_t,
-				     unsigned long);
-void			xfs_trans_delete_ail(struct xfs_mount *,
-				     struct xfs_log_item *, unsigned long);
+void			xfs_trans_update_ail(struct xfs_mount *mp,
+				     struct xfs_log_item *lip, xfs_lsn_t lsn,
+				     unsigned long s)
+				     __releases(mp->m_ail_lock);
+void			xfs_trans_delete_ail(struct xfs_mount *mp,
+				     struct xfs_log_item *lip, unsigned long s)
+				     __releases(mp->m_ail_lock);
 struct xfs_log_item	*xfs_trans_first_ail(struct xfs_mount *, int *);
 struct xfs_log_item	*xfs_trans_next_ail(struct xfs_mount *,
 				     struct xfs_log_item *, int *, int *);

fs/xfs/xfs_vfsops.c

@@ -1922,7 +1922,7 @@ xfs_showargs(
 	}
 
 	if (mp->m_flags & XFS_MOUNT_IHASHSIZE)
-		seq_printf(m, "," MNTOPT_IHASHSIZE "=%d", mp->m_ihsize);
+		seq_printf(m, "," MNTOPT_IHASHSIZE "=%d", (int)mp->m_ihsize);
 
 	if (mp->m_flags & XFS_MOUNT_DFLT_IOSIZE)
 		seq_printf(m, "," MNTOPT_ALLOCSIZE "=%dk",

fs/xfs/xfs_vnodeops.c

@@ -2366,10 +2366,15 @@ xfs_remove(
 
 	namelen = VNAMELEN(dentry);
 
+	if (!xfs_get_dir_entry(dentry, &ip)) {
+	        dm_di_mode = ip->i_d.di_mode;
+		IRELE(ip);
+	}
+
 	if (DM_EVENT_ENABLED(dir_vp->v_vfsp, dp, DM_EVENT_REMOVE)) {
 		error = XFS_SEND_NAMESP(mp, DM_EVENT_REMOVE, dir_vp,
 					DM_RIGHT_NULL, NULL, DM_RIGHT_NULL,
-					name, NULL, 0, 0, 0);
+					name, NULL, dm_di_mode, 0, 0);
 		if (error)
 			return error;
 	}
@@ -2995,7 +3000,7 @@ xfs_rmdir(
 	int			cancel_flags;
 	int			committed;
 	bhv_vnode_t		*dir_vp;
-	int			dm_di_mode = 0;
+	int			dm_di_mode = S_IFDIR;
 	int			last_cdp_link;
 	int			namelen;
 	uint			resblks;
@@ -3010,11 +3015,16 @@ xfs_rmdir(
 		return XFS_ERROR(EIO);
 	namelen = VNAMELEN(dentry);
 
+	if (!xfs_get_dir_entry(dentry, &cdp)) {
+	        dm_di_mode = cdp->i_d.di_mode;
+		IRELE(cdp);
+	}
+
 	if (DM_EVENT_ENABLED(dir_vp->v_vfsp, dp, DM_EVENT_REMOVE)) {
 		error = XFS_SEND_NAMESP(mp, DM_EVENT_REMOVE,
 					dir_vp, DM_RIGHT_NULL,
 					NULL, DM_RIGHT_NULL,
-					name, NULL, 0, 0, 0);
+					name, NULL, dm_di_mode, 0, 0);
 		if (error)
 			return XFS_ERROR(error);
 	}
@@ -3834,7 +3844,9 @@ xfs_reclaim(
 		XFS_MOUNT_ILOCK(mp);
 		vn_bhv_remove(VN_BHV_HEAD(vp), XFS_ITOBHV(ip));
 		list_add_tail(&ip->i_reclaim, &mp->m_del_inodes);
+		spin_lock(&ip->i_flags_lock);
 		ip->i_flags |= XFS_IRECLAIMABLE;
+		spin_unlock(&ip->i_flags_lock);
 		XFS_MOUNT_IUNLOCK(mp);
 	}
 	return 0;
@@ -3859,8 +3871,10 @@ xfs_finish_reclaim(
 	 * us.
 	 */
 	write_lock(&ih->ih_lock);
+	spin_lock(&ip->i_flags_lock);
 	if ((ip->i_flags & XFS_IRECLAIM) ||
 	    (!(ip->i_flags & XFS_IRECLAIMABLE) && vp == NULL)) {
+		spin_unlock(&ip->i_flags_lock);
 		write_unlock(&ih->ih_lock);
 		if (locked) {
 			xfs_ifunlock(ip);
@@ -3869,6 +3883,7 @@ xfs_finish_reclaim(
 		return 1;
 	}
 	ip->i_flags |= XFS_IRECLAIM;
+	spin_unlock(&ip->i_flags_lock);
 	write_unlock(&ih->ih_lock);
 
 	/*
@@ -4272,7 +4287,7 @@ xfs_free_file_space(
 	xfs_mount_t		*mp;
 	int			nimap;
 	uint			resblks;
-	int			rounding;
+	uint			rounding;
 	int			rt;
 	xfs_fileoff_t		startoffset_fsb;
 	xfs_trans_t		*tp;
@@ -4313,8 +4328,7 @@ xfs_free_file_space(
 		vn_iowait(vp);	/* wait for the completion of any pending DIOs */
 	}
 
-	rounding = MAX((__uint8_t)(1 << mp->m_sb.sb_blocklog),
-			(__uint8_t)NBPP);
+	rounding = max_t(uint, 1 << mp->m_sb.sb_blocklog, NBPP);
 	ilen = len + (offset & (rounding - 1));
 	ioffset = offset & ~(rounding - 1);
 	if (ilen & (rounding - 1))