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

* git://oss.sgi.com:8090/xfs/linux-2.6:
  [XFS] Don't do I/O beyond eof when unreserving space
  [XFS] Fix use-after-free with buffers
  [XFS] Prevent lockdep false positives when locking two inodes.
  [XFS] Fix barrier status change detection.
  [XFS] Prevent direct I/O from mapping extents beyond eof
  [XFS] Fix regression introduced by remount fixup
  [XFS] Move memory allocations for log tracing out of the critical path

fs/xfs/linux-2.6/xfs_aops.c

@@ -1338,6 +1338,10 @@ __xfs_get_blocks(
 	offset = (xfs_off_t)iblock << inode->i_blkbits;
 	ASSERT(bh_result->b_size >= (1 << inode->i_blkbits));
 	size = bh_result->b_size;
+
+	if (!create && direct && offset >= i_size_read(inode))
+		return 0;
+
 	error = xfs_iomap(XFS_I(inode), offset, size,
 			     create ? flags : BMAPI_READ, &iomap, &niomap);
 	if (error)

fs/xfs/linux-2.6/xfs_super.c

@@ -1302,9 +1302,29 @@ xfs_fs_remount(
 			mp->m_flags &= ~XFS_MOUNT_BARRIER;
 			break;
 		default:
+			/*
+			 * Logically we would return an error here to prevent
+			 * users from believing they might have changed
+			 * mount options using remount which can't be changed.
+			 *
+			 * But unfortunately mount(8) adds all options from
+			 * mtab and fstab to the mount arguments in some cases
+			 * so we can't blindly reject options, but have to
+			 * check for each specified option if it actually
+			 * differs from the currently set option and only
+			 * reject it if that's the case.
+			 *
+			 * Until that is implemented we return success for
+			 * every remount request, and silently ignore all
+			 * options that we can't actually change.
+			 */
+#if 0
 			printk(KERN_INFO
 	"XFS: mount option \"%s\" not supported for remount\n", p);
 			return -EINVAL;
+#else
+			return 0;
+#endif
 		}
 	}
 

fs/xfs/xfs_buf_item.c

@@ -732,6 +732,7 @@ xfs_buf_item_init(
 	bip->bli_item.li_ops = &xfs_buf_item_ops;
 	bip->bli_item.li_mountp = mp;
 	bip->bli_buf = bp;
+	xfs_buf_hold(bp);
 	bip->bli_format.blf_type = XFS_LI_BUF;
 	bip->bli_format.blf_blkno = (__int64_t)XFS_BUF_ADDR(bp);
 	bip->bli_format.blf_len = (ushort)BTOBB(XFS_BUF_COUNT(bp));
@@ -867,6 +868,21 @@ xfs_buf_item_dirty(
 	return (bip->bli_flags & XFS_BLI_DIRTY);
 }
 
+STATIC void
+xfs_buf_item_free(
+	xfs_buf_log_item_t	*bip)
+{
+#ifdef XFS_TRANS_DEBUG
+	kmem_free(bip->bli_orig);
+	kmem_free(bip->bli_logged);
+#endif /* XFS_TRANS_DEBUG */
+
+#ifdef XFS_BLI_TRACE
+	ktrace_free(bip->bli_trace);
+#endif
+	kmem_zone_free(xfs_buf_item_zone, bip);
+}
+
 /*
  * This is called when the buf log item is no longer needed.  It should
  * free the buf log item associated with the given buffer and clear
@@ -887,18 +903,8 @@ xfs_buf_item_relse(
 	    (XFS_BUF_IODONE_FUNC(bp) != NULL)) {
 		XFS_BUF_CLR_IODONE_FUNC(bp);
 	}
-
-#ifdef XFS_TRANS_DEBUG
-	kmem_free(bip->bli_orig);
-	bip->bli_orig = NULL;
-	kmem_free(bip->bli_logged);
-	bip->bli_logged = NULL;
-#endif /* XFS_TRANS_DEBUG */
-
-#ifdef XFS_BLI_TRACE
-	ktrace_free(bip->bli_trace);
-#endif
-	kmem_zone_free(xfs_buf_item_zone, bip);
+	xfs_buf_rele(bp);
+	xfs_buf_item_free(bip);
 }
 
 
@@ -1120,6 +1126,7 @@ xfs_buf_iodone(
 
 	ASSERT(bip->bli_buf == bp);
 
+	xfs_buf_rele(bp);
 	mp = bip->bli_item.li_mountp;
 
 	/*
@@ -1136,18 +1143,7 @@ xfs_buf_iodone(
 	 * xfs_trans_delete_ail() drops the AIL lock.
 	 */
 	xfs_trans_delete_ail(mp, (xfs_log_item_t *)bip);
-
-#ifdef XFS_TRANS_DEBUG
-	kmem_free(bip->bli_orig);
-	bip->bli_orig = NULL;
-	kmem_free(bip->bli_logged);
-	bip->bli_logged = NULL;
-#endif /* XFS_TRANS_DEBUG */
-
-#ifdef XFS_BLI_TRACE
-	ktrace_free(bip->bli_trace);
-#endif
-	kmem_zone_free(xfs_buf_item_zone, bip);
+	xfs_buf_item_free(bip);
 }
 
 #if defined(XFS_BLI_TRACE)

fs/xfs/xfs_dfrag.c

@@ -149,7 +149,14 @@ xfs_swap_extents(
 
 	sbp = &sxp->sx_stat;
 
-	xfs_lock_two_inodes(ip, tip, lock_flags);
+	/*
+	 * we have to do two separate lock calls here to keep lockdep
+	 * happy. If we try to get all the locks in one call, lock will
+	 * report false positives when we drop the ILOCK and regain them
+	 * below.
+	 */
+	xfs_lock_two_inodes(ip, tip, XFS_IOLOCK_EXCL);
+	xfs_lock_two_inodes(ip, tip, XFS_ILOCK_EXCL);
 	locked = 1;
 
 	/* Verify that both files have the same format */

fs/xfs/xfs_log.c

@@ -124,16 +124,27 @@ STATIC void	xlog_verify_tail_lsn(xlog_t *log, xlog_in_core_t *iclog,
 STATIC int	xlog_iclogs_empty(xlog_t *log);
 
 #if defined(XFS_LOG_TRACE)
+
+#define XLOG_TRACE_LOGGRANT_SIZE	2048
+#define XLOG_TRACE_ICLOG_SIZE		256
+
+void
+xlog_trace_loggrant_alloc(xlog_t *log)
+{
+	log->l_grant_trace = ktrace_alloc(XLOG_TRACE_LOGGRANT_SIZE, KM_NOFS);
+}
+
+void
+xlog_trace_loggrant_dealloc(xlog_t *log)
+{
+	ktrace_free(log->l_grant_trace);
+}
+
 void
 xlog_trace_loggrant(xlog_t *log, xlog_ticket_t *tic, xfs_caddr_t string)
 {
 	unsigned long cnts;
 
-	if (!log->l_grant_trace) {
-		log->l_grant_trace = ktrace_alloc(2048, KM_NOSLEEP);
-		if (!log->l_grant_trace)
-			return;
-	}
 	/* ticket counts are 1 byte each */
 	cnts = ((unsigned long)tic->t_ocnt) | ((unsigned long)tic->t_cnt) << 8;
 
@@ -156,11 +167,21 @@ xlog_trace_loggrant(xlog_t *log, xlog_ticket_t *tic, xfs_caddr_t string)
 		     (void *)((unsigned long)tic->t_unit_res));
 }
 
+void
+xlog_trace_iclog_alloc(xlog_in_core_t *iclog)
+{
+	iclog->ic_trace = ktrace_alloc(XLOG_TRACE_ICLOG_SIZE, KM_NOFS);
+}
+
+void
+xlog_trace_iclog_dealloc(xlog_in_core_t *iclog)
+{
+	ktrace_free(iclog->ic_trace);
+}
+
 void
 xlog_trace_iclog(xlog_in_core_t *iclog, uint state)
 {
-	if (!iclog->ic_trace)
-		iclog->ic_trace = ktrace_alloc(256, KM_NOFS);
 	ktrace_enter(iclog->ic_trace,
 		     (void *)((unsigned long)state),
 		     (void *)((unsigned long)current_pid()),
@@ -170,8 +191,15 @@ xlog_trace_iclog(xlog_in_core_t *iclog, uint state)
 		     (void *)NULL, (void *)NULL);
 }
 #else
+
+#define	xlog_trace_loggrant_alloc(log)
+#define	xlog_trace_loggrant_dealloc(log)
 #define	xlog_trace_loggrant(log,tic,string)
+
+#define	xlog_trace_iclog_alloc(iclog)
+#define	xlog_trace_iclog_dealloc(iclog)
 #define	xlog_trace_iclog(iclog,state)
+
 #endif /* XFS_LOG_TRACE */
 
 
@@ -1009,7 +1037,7 @@ xlog_iodone(xfs_buf_t *bp)
 	 * layer, it means the underlyin device no longer supports
 	 * barrier I/O. Warn loudly and turn off barriers.
 	 */
-	if ((l->l_mp->m_flags & XFS_MOUNT_BARRIER) && !XFS_BUF_ORDERED(bp)) {
+	if ((l->l_mp->m_flags & XFS_MOUNT_BARRIER) && !XFS_BUF_ISORDERED(bp)) {
 		l->l_mp->m_flags &= ~XFS_MOUNT_BARRIER;
 		xfs_fs_cmn_err(CE_WARN, l->l_mp,
 				"xlog_iodone: Barriers are no longer supported"
@@ -1231,6 +1259,7 @@ xlog_alloc_log(xfs_mount_t	*mp,
 	spin_lock_init(&log->l_grant_lock);
 	sv_init(&log->l_flush_wait, 0, "flush_wait");
 
+	xlog_trace_loggrant_alloc(log);
 	/* log record size must be multiple of BBSIZE; see xlog_rec_header_t */
 	ASSERT((XFS_BUF_SIZE(bp) & BBMASK) == 0);
 
@@ -1285,6 +1314,8 @@ xlog_alloc_log(xfs_mount_t	*mp,
 		sv_init(&iclog->ic_force_wait, SV_DEFAULT, "iclog-force");
 		sv_init(&iclog->ic_write_wait, SV_DEFAULT, "iclog-write");
 
+		xlog_trace_iclog_alloc(iclog);
+
 		iclogp = &iclog->ic_next;
 	}
 	*iclogp = log->l_iclog;			/* complete ring */
@@ -1565,11 +1596,7 @@ xlog_dealloc_log(xlog_t *log)
 		sv_destroy(&iclog->ic_force_wait);
 		sv_destroy(&iclog->ic_write_wait);
 		xfs_buf_free(iclog->ic_bp);
-#ifdef XFS_LOG_TRACE
-		if (iclog->ic_trace != NULL) {
-			ktrace_free(iclog->ic_trace);
-		}
-#endif
+		xlog_trace_iclog_dealloc(iclog);
 		next_iclog = iclog->ic_next;
 		kmem_free(iclog);
 		iclog = next_iclog;
@@ -1578,14 +1605,7 @@ xlog_dealloc_log(xlog_t *log)
 	spinlock_destroy(&log->l_grant_lock);
 
 	xfs_buf_free(log->l_xbuf);
-#ifdef XFS_LOG_TRACE
-	if (log->l_trace != NULL) {
-		ktrace_free(log->l_trace);
-	}
-	if (log->l_grant_trace != NULL) {
-		ktrace_free(log->l_grant_trace);
-	}
-#endif
+	xlog_trace_loggrant_dealloc(log);
 	log->l_mp->m_log = NULL;
 	kmem_free(log);
 }	/* xlog_dealloc_log */

fs/xfs/xfs_log_priv.h

@@ -448,7 +448,6 @@ typedef struct log {
 	int			l_grant_write_bytes;
 
 #ifdef XFS_LOG_TRACE
-	struct ktrace		*l_trace;
 	struct ktrace		*l_grant_trace;
 #endif
 

fs/xfs/xfs_vnodeops.c

@@ -1838,6 +1838,12 @@ again:
 #endif
 }
 
+/*
+ * xfs_lock_two_inodes() can only be used to lock one type of lock
+ * at a time - the iolock or the ilock, but not both at once. If
+ * we lock both at once, lockdep will report false positives saying
+ * we have violated locking orders.
+ */
 void
 xfs_lock_two_inodes(
 	xfs_inode_t		*ip0,
@@ -1848,6 +1854,8 @@ xfs_lock_two_inodes(
 	int			attempts = 0;
 	xfs_log_item_t		*lp;
 
+	if (lock_mode & (XFS_IOLOCK_SHARED|XFS_IOLOCK_EXCL))
+		ASSERT((lock_mode & (XFS_ILOCK_SHARED|XFS_ILOCK_EXCL)) == 0);
 	ASSERT(ip0->i_ino != ip1->i_ino);
 
 	if (ip0->i_ino > ip1->i_ino) {
@@ -3152,6 +3160,13 @@ error1:	/* Just cancel transaction */
 /*
  * Zero file bytes between startoff and endoff inclusive.
  * The iolock is held exclusive and no blocks are buffered.
+ *
+ * This function is used by xfs_free_file_space() to zero
+ * partial blocks when the range to free is not block aligned.
+ * When unreserving space with boundaries that are not block
+ * aligned we round up the start and round down the end
+ * boundaries and then use this function to zero the parts of
+ * the blocks that got dropped during the rounding.
  */
 STATIC int
 xfs_zero_remaining_bytes(
@@ -3168,6 +3183,17 @@ xfs_zero_remaining_bytes(
 	int			nimap;
 	int			error = 0;
 
+	/*
+	 * Avoid doing I/O beyond eof - it's not necessary
+	 * since nothing can read beyond eof.  The space will
+	 * be zeroed when the file is extended anyway.
+	 */
+	if (startoff >= ip->i_size)
+		return 0;
+
+	if (endoff > ip->i_size)
+		endoff = ip->i_size;
+
 	bp = xfs_buf_get_noaddr(mp->m_sb.sb_blocksize,
 				XFS_IS_REALTIME_INODE(ip) ?
 				mp->m_rtdev_targp : mp->m_ddev_targp);