Merge branch 'for-linus' of git://oss.sgi.com/xfs/xfs

* 'for-linus' of git://oss.sgi.com/xfs/xfs:
  xfs: xfs_bmap_add_extent_delay_real should init br_startblock
  xfs: fix dquot shaker deadlock
  xfs: handle CIl transaction commit failures correctly
  xfs: limit extsize to size of AGs and/or MAXEXTLEN
  xfs: prevent extsize alignment from exceeding maximum extent size
  xfs: limit extent length for allocation to AG size
  xfs: speculative delayed allocation uses rounddown_power_of_2 badly
  xfs: fix efi item leak on forced shutdown
  xfs: fix log ticket leak on forced shutdown.

10 files changed, 152 insertions, 71 deletions

diff --git a/fs/xfs/linux-2.6/xfs_ioctl.c b/fs/xfs/linux-2.6/xfs_ioctl.c
index b06ede1d0be..f5e2a19e0f8 100644
--- a/fs/xfs/linux-2.6/xfs_ioctl.c
+++ b/fs/xfs/linux-2.6/xfs_ioctl.c
@@ -985,10 +985,22 @@ xfs_ioctl_setattr(
 
 		/*
 		 * Extent size must be a multiple of the appropriate block
-		 * size, if set at all.
+		 * size, if set at all. It must also be smaller than the
+		 * maximum extent size supported by the filesystem.
+		 *
+		 * Also, for non-realtime files, limit the extent size hint to
+		 * half the size of the AGs in the filesystem so alignment
+		 * doesn't result in extents larger than an AG.
 		 */
 		if (fa->fsx_extsize != 0) {
-			xfs_extlen_t	size;
+			xfs_extlen_t    size;
+			xfs_fsblock_t   extsize_fsb;
+
+			extsize_fsb = XFS_B_TO_FSB(mp, fa->fsx_extsize);
+			if (extsize_fsb > MAXEXTLEN) {
+				code = XFS_ERROR(EINVAL);
+				goto error_return;
+			}
 
 			if (XFS_IS_REALTIME_INODE(ip) ||
 			    ((mask & FSX_XFLAGS) &&
@@ -997,6 +1009,10 @@ xfs_ioctl_setattr(
 				       mp->m_sb.sb_blocklog;
 			} else {
 				size = mp->m_sb.sb_blocksize;
+				if (extsize_fsb > mp->m_sb.sb_agblocks / 2) {
+					code = XFS_ERROR(EINVAL);
+					goto error_return;
+				}
 			}
 
 			if (fa->fsx_extsize % size) {
diff --git a/fs/xfs/quota/xfs_qm.c b/fs/xfs/quota/xfs_qm.c
index f8e854b4fde..206a2815ced 100644
--- a/fs/xfs/quota/xfs_qm.c
+++ b/fs/xfs/quota/xfs_qm.c
@@ -1863,12 +1863,14 @@ xfs_qm_dqreclaim_one(void)
 	xfs_dquot_t	*dqpout;
 	xfs_dquot_t	*dqp;
 	int		restarts;
+	int		startagain;
 
 	restarts = 0;
 	dqpout = NULL;
 
 	/* lockorder: hashchainlock, freelistlock, mplistlock, dqlock, dqflock */
-startagain:
+again:
+	startagain = 0;
 	mutex_lock(&xfs_Gqm->qm_dqfrlist_lock);
 
 	list_for_each_entry(dqp, &xfs_Gqm->qm_dqfrlist, q_freelist) {
@@ -1885,13 +1887,10 @@ startagain:
 			ASSERT(! (dqp->dq_flags & XFS_DQ_INACTIVE));
 
 			trace_xfs_dqreclaim_want(dqp);
-
-			xfs_dqunlock(dqp);
-			mutex_unlock(&xfs_Gqm->qm_dqfrlist_lock);
-			if (++restarts >= XFS_QM_RECLAIM_MAX_RESTARTS)
-				return NULL;
 			XQM_STATS_INC(xqmstats.xs_qm_dqwants);
-			goto startagain;
+			restarts++;
+			startagain = 1;
+			goto dqunlock;
 		}
 
 		/*
@@ -1906,23 +1905,20 @@ startagain:
 			ASSERT(list_empty(&dqp->q_mplist));
 			list_del_init(&dqp->q_freelist);
 			xfs_Gqm->qm_dqfrlist_cnt--;
-			xfs_dqunlock(dqp);
 			dqpout = dqp;
 			XQM_STATS_INC(xqmstats.xs_qm_dqinact_reclaims);
-			break;
+			goto dqunlock;
 		}
 
 		ASSERT(dqp->q_hash);
 		ASSERT(!list_empty(&dqp->q_mplist));
 
 		/*
-		 * Try to grab the flush lock. If this dquot is in the process of
-		 * getting flushed to disk, we don't want to reclaim it.
+		 * Try to grab the flush lock. If this dquot is in the process
+		 * of getting flushed to disk, we don't want to reclaim it.
 		 */
-		if (!xfs_dqflock_nowait(dqp)) {
-			xfs_dqunlock(dqp);
-			continue;
-		}
+		if (!xfs_dqflock_nowait(dqp))
+			goto dqunlock;
 
 		/*
 		 * We have the flush lock so we know that this is not in the
@@ -1944,8 +1940,7 @@ startagain:
 				xfs_fs_cmn_err(CE_WARN, mp,
 			"xfs_qm_dqreclaim: dquot %p flush failed", dqp);
 			}
-			xfs_dqunlock(dqp); /* dqflush unlocks dqflock */
-			continue;
+			goto dqunlock;
 		}
 
 		/*
@@ -1967,13 +1962,8 @@ startagain:
 		 */
 		if (!mutex_trylock(&mp->m_quotainfo->qi_dqlist_lock)) {
 			restarts++;
-			mutex_unlock(&dqp->q_hash->qh_lock);
-			xfs_dqfunlock(dqp);
-			xfs_dqunlock(dqp);
-			mutex_unlock(&xfs_Gqm->qm_dqfrlist_lock);
-			if (restarts++ >= XFS_QM_RECLAIM_MAX_RESTARTS)
-				return NULL;
-			goto startagain;
+			startagain = 1;
+			goto qhunlock;
 		}
 
 		ASSERT(dqp->q_nrefs == 0);
@@ -1986,14 +1976,20 @@ startagain:
 		xfs_Gqm->qm_dqfrlist_cnt--;
 		dqpout = dqp;
 		mutex_unlock(&mp->m_quotainfo->qi_dqlist_lock);
+qhunlock:
 		mutex_unlock(&dqp->q_hash->qh_lock);
 dqfunlock:
 		xfs_dqfunlock(dqp);
+dqunlock:
 		xfs_dqunlock(dqp);
 		if (dqpout)
 			break;
 		if (restarts >= XFS_QM_RECLAIM_MAX_RESTARTS)
-			return NULL;
+			break;
+		if (startagain) {
+			mutex_unlock(&xfs_Gqm->qm_dqfrlist_lock);
+			goto again;
+		}
 	}
 	mutex_unlock(&xfs_Gqm->qm_dqfrlist_lock);
 	return dqpout;
diff --git a/fs/xfs/xfs_alloc.h b/fs/xfs/xfs_alloc.h
index 0ab56b32c7e..d0b3bc72005 100644
--- a/fs/xfs/xfs_alloc.h
+++ b/fs/xfs/xfs_alloc.h
@@ -75,6 +75,22 @@ typedef unsigned int xfs_alloctype_t;
 #define XFS_ALLOC_SET_ASIDE(mp)  (4 + ((mp)->m_sb.sb_agcount * 4))
 
 /*
+ * When deciding how much space to allocate out of an AG, we limit the
+ * allocation maximum size to the size the AG. However, we cannot use all the
+ * blocks in the AG - some are permanently used by metadata. These
+ * blocks are generally:
+ *	- the AG superblock, AGF, AGI and AGFL
+ *	- the AGF (bno and cnt) and AGI btree root blocks
+ *	- 4 blocks on the AGFL according to XFS_ALLOC_SET_ASIDE() limits
+ *
+ * The AG headers are sector sized, so the amount of space they take up is
+ * dependent on filesystem geometry. The others are all single blocks.
+ */
+#define XFS_ALLOC_AG_MAX_USABLE(mp)	\
+	((mp)->m_sb.sb_agblocks - XFS_BB_TO_FSB(mp, XFS_FSS_TO_BB(mp, 4)) - 7)
+
+
+/*
  * Argument structure for xfs_alloc routines.
  * This is turned into a structure to avoid having 20 arguments passed
  * down several levels of the stack.
diff --git a/fs/xfs/xfs_bmap.c b/fs/xfs/xfs_bmap.c
index 4111cd3966c..dc3afd7739f 100644
--- a/fs/xfs/xfs_bmap.c
+++ b/fs/xfs/xfs_bmap.c
@@ -1038,17 +1038,34 @@ xfs_bmap_add_extent_delay_real(
 		 * Filling in the middle part of a previous delayed allocation.
 		 * Contiguity is impossible here.
 		 * This case is avoided almost all the time.
+		 *
+		 * We start with a delayed allocation:
+		 *
+		 * +ddddddddddddddddddddddddddddddddddddddddddddddddddddddd+
+		 *  PREV @ idx
+		 *
+	         * and we are allocating:
+		 *                     +rrrrrrrrrrrrrrrrr+
+		 *			      new
+		 *
+		 * and we set it up for insertion as:
+		 * +ddddddddddddddddddd+rrrrrrrrrrrrrrrrr+ddddddddddddddddd+
+		 *                            new
+		 *  PREV @ idx          LEFT              RIGHT
+		 *                      inserted at idx + 1
 		 */
 		temp = new->br_startoff - PREV.br_startoff;
-		trace_xfs_bmap_pre_update(ip, idx, 0, _THIS_IP_);
-		xfs_bmbt_set_blockcount(ep, temp);
-		r[0] = *new;
-		r[1].br_state = PREV.br_state;
-		r[1].br_startblock = 0;
-		r[1].br_startoff = new_endoff;
 		temp2 = PREV.br_startoff + PREV.br_blockcount - new_endoff;
-		r[1].br_blockcount = temp2;
-		xfs_iext_insert(ip, idx + 1, 2, &r[0], state);
+		trace_xfs_bmap_pre_update(ip, idx, 0, _THIS_IP_);
+		xfs_bmbt_set_blockcount(ep, temp);	/* truncate PREV */
+		LEFT = *new;
+		RIGHT.br_state = PREV.br_state;
+		RIGHT.br_startblock = nullstartblock(
+				(int)xfs_bmap_worst_indlen(ip, temp2));
+		RIGHT.br_startoff = new_endoff;
+		RIGHT.br_blockcount = temp2;
+		/* insert LEFT (r[0]) and RIGHT (r[1]) at the same time */
+		xfs_iext_insert(ip, idx + 1, 2, &LEFT, state);
 		ip->i_df.if_lastex = idx + 1;
 		ip->i_d.di_nextents++;
 		if (cur == NULL)
@@ -2430,7 +2447,7 @@ xfs_bmap_btalloc_nullfb(
 		startag = ag = 0;
 
 	pag = xfs_perag_get(mp, ag);
-	while (*blen < ap->alen) {
+	while (*blen < args->maxlen) {
 		if (!pag->pagf_init) {
 			error = xfs_alloc_pagf_init(mp, args->tp, ag,
 						    XFS_ALLOC_FLAG_TRYLOCK);
@@ -2452,7 +2469,7 @@ xfs_bmap_btalloc_nullfb(
 			notinit = 1;
 
 		if (xfs_inode_is_filestream(ap->ip)) {
-			if (*blen >= ap->alen)
+			if (*blen >= args->maxlen)
 				break;
 
 			if (ap->userdata) {
@@ -2498,14 +2515,14 @@ xfs_bmap_btalloc_nullfb(
 	 * If the best seen length is less than the request
 	 * length, use the best as the minimum.
 	 */
-	else if (*blen < ap->alen)
+	else if (*blen < args->maxlen)
 		args->minlen = *blen;
 	/*
-	 * Otherwise we've seen an extent as big as alen,
+	 * Otherwise we've seen an extent as big as maxlen,
 	 * use that as the minimum.
 	 */
 	else
-		args->minlen = ap->alen;
+		args->minlen = args->maxlen;
 
 	/*
 	 * set the failure fallback case to look in the selected
@@ -2573,7 +2590,9 @@ xfs_bmap_btalloc(
 	args.tp = ap->tp;
 	args.mp = mp;
 	args.fsbno = ap->rval;
-	args.maxlen = MIN(ap->alen, mp->m_sb.sb_agblocks);
+
+	/* Trim the allocation back to the maximum an AG can fit. */
+	args.maxlen = MIN(ap->alen, XFS_ALLOC_AG_MAX_USABLE(mp));
 	args.firstblock = ap->firstblock;
 	blen = 0;
 	if (nullfb) {
@@ -2621,7 +2640,7 @@ xfs_bmap_btalloc(
 			/*
 			 * Adjust for alignment
 			 */
-			if (blen > args.alignment && blen <= ap->alen)
+			if (blen > args.alignment && blen <= args.maxlen)
 				args.minlen = blen - args.alignment;
 			args.minalignslop = 0;
 		} else {
@@ -2640,7 +2659,7 @@ xfs_bmap_btalloc(
 			 * of minlen+alignment+slop doesn't go up
 			 * between the calls.
 			 */
-			if (blen > mp->m_dalign && blen <= ap->alen)
+			if (blen > mp->m_dalign && blen <= args.maxlen)
 				nextminlen = blen - mp->m_dalign;
 			else
 				nextminlen = args.minlen;
@@ -4485,6 +4504,16 @@ xfs_bmapi(
 				/* Figure out the extent size, adjust alen */
 				extsz = xfs_get_extsz_hint(ip);
 				if (extsz) {
+					/*
+					 * make sure we don't exceed a single
+					 * extent length when we align the
+					 * extent by reducing length we are
+					 * going to allocate by the maximum
+					 * amount extent size aligment may
+					 * require.
+					 */
+					alen = XFS_FILBLKS_MIN(len,
+						   MAXEXTLEN - (2 * extsz - 1));
 					error = xfs_bmap_extsize_align(mp,
 							&got, &prev, extsz,
 							rt, eof,
diff --git a/fs/xfs/xfs_buf_item.c b/fs/xfs/xfs_buf_item.c
index 98c6f73b675..6f8c21ce0d6 100644
--- a/fs/xfs/xfs_buf_item.c
+++ b/fs/xfs/xfs_buf_item.c
@@ -427,13 +427,15 @@ xfs_buf_item_unpin(
 
 		if (remove) {
 			/*
-			 * We have to remove the log item from the transaction
-			 * as we are about to release our reference to the
-			 * buffer.  If we don't, the unlock that occurs later
-			 * in xfs_trans_uncommit() will ry to reference the
+			 * If we are in a transaction context, we have to
+			 * remove the log item from the transaction as we are
+			 * about to release our reference to the buffer.  If we
+			 * don't, the unlock that occurs later in
+			 * xfs_trans_uncommit() will try to reference the
 			 * buffer which we no longer have a hold on.
 			 */
-			xfs_trans_del_item(lip);
+			if (lip->li_desc)
+				xfs_trans_del_item(lip);
 
 			/*
 			 * Since the transaction no longer refers to the buffer,
diff --git a/fs/xfs/xfs_extfree_item.c b/fs/xfs/xfs_extfree_item.c
index 75f2ef60e57..d22e6262343 100644
--- a/fs/xfs/xfs_extfree_item.c
+++ b/fs/xfs/xfs_extfree_item.c
@@ -138,7 +138,8 @@ xfs_efi_item_unpin(
 
 	if (remove) {
 		ASSERT(!(lip->li_flags & XFS_LI_IN_AIL));
-		xfs_trans_del_item(lip);
+		if (lip->li_desc)
+			xfs_trans_del_item(lip);
 		xfs_efi_item_free(efip);
 		return;
 	}
diff --git a/fs/xfs/xfs_iomap.c b/fs/xfs/xfs_iomap.c
index 55582bd6665..8a0f044750c 100644
--- a/fs/xfs/xfs_iomap.c
+++ b/fs/xfs/xfs_iomap.c
@@ -337,7 +337,12 @@ xfs_iomap_prealloc_size(
 		int shift = 0;
 		int64_t freesp;
 
-		alloc_blocks = XFS_B_TO_FSB(mp, ip->i_size);
+		/*
+		 * rounddown_pow_of_two() returns an undefined result
+		 * if we pass in alloc_blocks = 0. Hence the "+ 1" to
+		 * ensure we always pass in a non-zero value.
+		 */
+		alloc_blocks = XFS_B_TO_FSB(mp, ip->i_size) + 1;
 		alloc_blocks = XFS_FILEOFF_MIN(MAXEXTLEN,
 					rounddown_pow_of_two(alloc_blocks));
 
diff --git a/fs/xfs/xfs_log.h b/fs/xfs/xfs_log.h
index 916eb7db14d..3bd3291ef8d 100644
--- a/fs/xfs/xfs_log.h
+++ b/fs/xfs/xfs_log.h
@@ -191,7 +191,7 @@ void	  xfs_log_ticket_put(struct xlog_ticket *ticket);
 
 xlog_tid_t xfs_log_get_trans_ident(struct xfs_trans *tp);
 
-int	xfs_log_commit_cil(struct xfs_mount *mp, struct xfs_trans *tp,
+void	xfs_log_commit_cil(struct xfs_mount *mp, struct xfs_trans *tp,
 				struct xfs_log_vec *log_vector,
 				xfs_lsn_t *commit_lsn, int flags);
 bool	xfs_log_item_in_current_chkpt(struct xfs_log_item *lip);
diff --git a/fs/xfs/xfs_log_cil.c b/fs/xfs/xfs_log_cil.c
index 9dc8125d04e..9ca59be0897 100644
--- a/fs/xfs/xfs_log_cil.c
+++ b/fs/xfs/xfs_log_cil.c
@@ -543,7 +543,7 @@ xlog_cil_push(
 
 	error = xlog_write(log, &lvhdr, tic, &ctx->start_lsn, NULL, 0);
 	if (error)
-		goto out_abort;
+		goto out_abort_free_ticket;
 
 	/*
 	 * now that we've written the checkpoint into the log, strictly
@@ -569,8 +569,9 @@ restart:
 	}
 	spin_unlock(&cil->xc_cil_lock);
 
+	/* xfs_log_done always frees the ticket on error. */
 	commit_lsn = xfs_log_done(log->l_mp, tic, &commit_iclog, 0);
-	if (error || commit_lsn == -1)
+	if (commit_lsn == -1)
 		goto out_abort;
 
 	/* attach all the transactions w/ busy extents to iclog */
@@ -600,6 +601,8 @@ out_free_ticket:
 	kmem_free(new_ctx);
 	return 0;
 
+out_abort_free_ticket:
+	xfs_log_ticket_put(tic);
 out_abort:
 	xlog_cil_committed(ctx, XFS_LI_ABORTED);
 	return XFS_ERROR(EIO);
@@ -622,7 +625,7 @@ out_abort:
  * background commit, returns without it held once background commits are
  * allowed again.
  */
-int
+void
 xfs_log_commit_cil(
 	struct xfs_mount	*mp,
 	struct xfs_trans	*tp,
@@ -637,11 +640,6 @@ xfs_log_commit_cil(
 	if (flags & XFS_TRANS_RELEASE_LOG_RES)
 		log_flags = XFS_LOG_REL_PERM_RESERV;
 
-	if (XLOG_FORCED_SHUTDOWN(log)) {
-		xlog_cil_free_logvec(log_vector);
-		return XFS_ERROR(EIO);
-	}
-
 	/*
 	 * do all the hard work of formatting items (including memory
 	 * allocation) outside the CIL context lock. This prevents stalling CIL
@@ -701,7 +699,6 @@ xfs_log_commit_cil(
 	 */
 	if (push)
 		xlog_cil_push(log, 0);
-	return 0;
 }
 
 /*
diff --git a/fs/xfs/xfs_trans.c b/fs/xfs/xfs_trans.c
index 33dbc4e0ad6..76922793f64 100644
--- a/fs/xfs/xfs_trans.c
+++ b/fs/xfs/xfs_trans.c
@@ -1446,6 +1446,14 @@ xfs_log_item_batch_insert(
  * Bulk operation version of xfs_trans_committed that takes a log vector of
  * items to insert into the AIL. This uses bulk AIL insertion techniques to
  * minimise lock traffic.
+ *
+ * If we are called with the aborted flag set, it is because a log write during
+ * a CIL checkpoint commit has failed. In this case, all the items in the
+ * checkpoint have already gone through IOP_COMMITED and IOP_UNLOCK, which
+ * means that checkpoint commit abort handling is treated exactly the same
+ * as an iclog write error even though we haven't started any IO yet. Hence in
+ * this case all we need to do is IOP_COMMITTED processing, followed by an
+ * IOP_UNPIN(aborted) call.
  */
 void
 xfs_trans_committed_bulk(
@@ -1472,6 +1480,16 @@ xfs_trans_committed_bulk(
 		if (XFS_LSN_CMP(item_lsn, (xfs_lsn_t)-1) == 0)
 			continue;
 
+		/*
+		 * if we are aborting the operation, no point in inserting the
+		 * object into the AIL as we are in a shutdown situation.
+		 */
+		if (aborted) {
+			ASSERT(XFS_FORCED_SHUTDOWN(ailp->xa_mount));
+			IOP_UNPIN(lip, 1);
+			continue;
+		}
+
 		if (item_lsn != commit_lsn) {
 
 			/*
@@ -1503,20 +1521,24 @@ xfs_trans_committed_bulk(
 }
 
 /*
- * Called from the trans_commit code when we notice that
- * the filesystem is in the middle of a forced shutdown.
+ * Called from the trans_commit code when we notice that the filesystem is in
+ * the middle of a forced shutdown.
+ *
+ * When we are called here, we have already pinned all the items in the
+ * transaction. However, neither IOP_COMMITTING or IOP_UNLOCK has been called
+ * so we can simply walk the items in the transaction, unpin them with an abort
+ * flag and then free the items. Note that unpinning the items can result in
+ * them being freed immediately, so we need to use a safe list traversal method
+ * here.
  */
 STATIC void
 xfs_trans_uncommit(
 	struct xfs_trans	*tp,
 	uint			flags)
 {
-	struct xfs_log_item_desc *lidp;
+	struct xfs_log_item_desc *lidp, *n;
 
-	list_for_each_entry(lidp, &tp->t_items, lid_trans) {
-		/*
-		 * Unpin all but those that aren't dirty.
-		 */
+	list_for_each_entry_safe(lidp, n, &tp->t_items, lid_trans) {
 		if (lidp->lid_flags & XFS_LID_DIRTY)
 			IOP_UNPIN(lidp->lid_item, 1);
 	}
@@ -1733,7 +1755,6 @@ xfs_trans_commit_cil(
 	int			flags)
 {
 	struct xfs_log_vec	*log_vector;
-	int			error;
 
 	/*
 	 * Get each log item to allocate a vector structure for
@@ -1744,9 +1765,7 @@ xfs_trans_commit_cil(
 	if (!log_vector)
 		return ENOMEM;
 
-	error = xfs_log_commit_cil(mp, tp, log_vector, commit_lsn, flags);
-	if (error)
-		return error;
+	xfs_log_commit_cil(mp, tp, log_vector, commit_lsn, flags);
 
 	current_restore_flags_nested(&tp->t_pflags, PF_FSTRANS);
 	xfs_trans_free(tp);