X-Git-Url: http://git.onelab.eu/?a=blobdiff_plain;f=fs%2Focfs2%2Ffile.c;fp=fs%2Focfs2%2Ffile.c;h=5a3d1fd4d28a04cefa4bc8d6b0158e129341460d;hb=97bf2856c6014879bd04983a3e9dfcdac1e7fe85;hp=8a4048b55fdc41d5dda4dc5e8b9908e4a765cc64;hpb=76828883507a47dae78837ab5dec5a5b4513c667;p=linux-2.6.git

diff --git a/fs/ocfs2/file.c b/fs/ocfs2/file.c
index 8a4048b55..5a3d1fd4d 100644
--- a/fs/ocfs2/file.c
+++ b/fs/ocfs2/file.c
@@ -30,6 +30,9 @@
 #include <linux/highmem.h>
 #include <linux/pagemap.h>
 #include <linux/uio.h>
+#include <linux/sched.h>
+#include <linux/pipe_fs_i.h>
+#include <linux/mount.h>
 
 #define MLOG_MASK_PREFIX ML_INODE
 #include <cluster/masklog.h>
@@ -44,6 +47,7 @@
 #include "file.h"
 #include "sysfile.h"
 #include "inode.h"
+#include "ioctl.h"
 #include "journal.h"
 #include "mmap.h"
 #include "suballoc.h"
@@ -64,7 +68,7 @@ static int ocfs2_file_open(struct inode *inode, struct file *file)
 	struct ocfs2_inode_info *oi = OCFS2_I(inode);
 
 	mlog_entry("(0x%p, 0x%p, '%.*s')\n", inode, file,
-		   file->f_dentry->d_name.len, file->f_dentry->d_name.name);
+		   file->f_path.dentry->d_name.len, file->f_path.dentry->d_name.name);
 
 	spin_lock(&oi->ip_lock);
 
@@ -94,8 +98,8 @@ static int ocfs2_file_release(struct inode *inode, struct file *file)
 	struct ocfs2_inode_info *oi = OCFS2_I(inode);
 
 	mlog_entry("(0x%p, 0x%p, '%.*s')\n", inode, file,
-		       file->f_dentry->d_name.len,
-		       file->f_dentry->d_name.name);
+		       file->f_path.dentry->d_name.len,
+		       file->f_path.dentry->d_name.name);
 
 	spin_lock(&oi->ip_lock);
 	if (!--oi->ip_open_count)
@@ -132,7 +136,77 @@ bail:
 	return (err < 0) ? -EIO : 0;
 }
 
-int ocfs2_set_inode_size(struct ocfs2_journal_handle *handle,
+int ocfs2_should_update_atime(struct inode *inode,
+			      struct vfsmount *vfsmnt)
+{
+	struct timespec now;
+	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
+
+	if (ocfs2_is_hard_readonly(osb) || ocfs2_is_soft_readonly(osb))
+		return 0;
+
+	if ((inode->i_flags & S_NOATIME) ||
+	    ((inode->i_sb->s_flags & MS_NODIRATIME) && S_ISDIR(inode->i_mode)))
+		return 0;
+
+	/*
+	 * We can be called with no vfsmnt structure - NFSD will
+	 * sometimes do this.
+	 *
+	 * Note that our action here is different than touch_atime() -
+	 * if we can't tell whether this is a noatime mount, then we
+	 * don't know whether to trust the value of s_atime_quantum.
+	 */
+	if (vfsmnt == NULL)
+		return 0;
+
+	if ((vfsmnt->mnt_flags & MNT_NOATIME) ||
+	    ((vfsmnt->mnt_flags & MNT_NODIRATIME) && S_ISDIR(inode->i_mode)))
+		return 0;
+
+	if (vfsmnt->mnt_flags & MNT_RELATIME) {
+		if ((timespec_compare(&inode->i_atime, &inode->i_mtime) <= 0) ||
+		    (timespec_compare(&inode->i_atime, &inode->i_ctime) <= 0))
+			return 1;
+
+		return 0;
+	}
+
+	now = CURRENT_TIME;
+	if ((now.tv_sec - inode->i_atime.tv_sec <= osb->s_atime_quantum))
+		return 0;
+	else
+		return 1;
+}
+
+int ocfs2_update_inode_atime(struct inode *inode,
+			     struct buffer_head *bh)
+{
+	int ret;
+	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
+	handle_t *handle;
+
+	mlog_entry_void();
+
+	handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS);
+	if (handle == NULL) {
+		ret = -ENOMEM;
+		mlog_errno(ret);
+		goto out;
+	}
+
+	inode->i_atime = CURRENT_TIME;
+	ret = ocfs2_mark_inode_dirty(handle, inode, bh);
+	if (ret < 0)
+		mlog_errno(ret);
+
+	ocfs2_commit_trans(OCFS2_SB(inode->i_sb), handle);
+out:
+	mlog_exit(ret);
+	return ret;
+}
+
+int ocfs2_set_inode_size(handle_t *handle,
 			 struct inode *inode,
 			 struct buffer_head *fe_bh,
 			 u64 new_i_size)
@@ -161,10 +235,9 @@ static int ocfs2_simple_size_update(struct inode *inode,
 {
 	int ret;
 	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
-	struct ocfs2_journal_handle *handle = NULL;
+	handle_t *handle = NULL;
 
-	handle = ocfs2_start_trans(osb, NULL,
-				   OCFS2_INODE_UPDATE_CREDITS);
+	handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS);
 	if (handle == NULL) {
 		ret = -ENOMEM;
 		mlog_errno(ret);
@@ -176,7 +249,7 @@ static int ocfs2_simple_size_update(struct inode *inode,
 	if (ret < 0)
 		mlog_errno(ret);
 
-	ocfs2_commit_trans(handle);
+	ocfs2_commit_trans(osb, handle);
 out:
 	return ret;
 }
@@ -187,14 +260,14 @@ static int ocfs2_orphan_for_truncate(struct ocfs2_super *osb,
 				     u64 new_i_size)
 {
 	int status;
-	struct ocfs2_journal_handle *handle;
+	handle_t *handle;
 
 	mlog_entry_void();
 
 	/* TODO: This needs to actually orphan the inode in this
 	 * transaction. */
 
-	handle = ocfs2_start_trans(osb, NULL, OCFS2_INODE_UPDATE_CREDITS);
+	handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS);
 	if (IS_ERR(handle)) {
 		status = PTR_ERR(handle);
 		mlog_errno(status);
@@ -205,7 +278,7 @@ static int ocfs2_orphan_for_truncate(struct ocfs2_super *osb,
 	if (status < 0)
 		mlog_errno(status);
 
-	ocfs2_commit_trans(handle);
+	ocfs2_commit_trans(osb, handle);
 out:
 	mlog_exit(status);
 	return status;
@@ -220,8 +293,9 @@ static int ocfs2_truncate_file(struct inode *inode,
 	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
 	struct ocfs2_truncate_context *tc = NULL;
 
-	mlog_entry("(inode = %"MLFu64", new_i_size = %"MLFu64"\n",
-		   OCFS2_I(inode)->ip_blkno, new_i_size);
+	mlog_entry("(inode = %llu, new_i_size = %llu\n",
+		   (unsigned long long)OCFS2_I(inode)->ip_blkno,
+		   (unsigned long long)new_i_size);
 
 	truncate_inode_pages(inode->i_mapping, new_i_size);
 
@@ -233,29 +307,43 @@ static int ocfs2_truncate_file(struct inode *inode,
 	}
 
 	mlog_bug_on_msg(le64_to_cpu(fe->i_size) != i_size_read(inode),
-			"Inode %"MLFu64", inode i_size = %lld != di "
-			"i_size = %"MLFu64", i_flags = 0x%x\n",
-			OCFS2_I(inode)->ip_blkno,
+			"Inode %llu, inode i_size = %lld != di "
+			"i_size = %llu, i_flags = 0x%x\n",
+			(unsigned long long)OCFS2_I(inode)->ip_blkno,
 			i_size_read(inode),
-			le64_to_cpu(fe->i_size), le32_to_cpu(fe->i_flags));
+			(unsigned long long)le64_to_cpu(fe->i_size),
+			le32_to_cpu(fe->i_flags));
 
 	if (new_i_size > le64_to_cpu(fe->i_size)) {
-		mlog(0, "asked to truncate file with size (%"MLFu64") "
-		     "to size (%"MLFu64")!\n",
-		     le64_to_cpu(fe->i_size), new_i_size);
+		mlog(0, "asked to truncate file with size (%llu) to size (%llu)!\n",
+		     (unsigned long long)le64_to_cpu(fe->i_size),
+		     (unsigned long long)new_i_size);
 		status = -EINVAL;
 		mlog_errno(status);
 		goto bail;
 	}
 
-	mlog(0, "inode %"MLFu64", i_size = %"MLFu64", new_i_size = %"MLFu64"\n",
-	     le64_to_cpu(fe->i_blkno), le64_to_cpu(fe->i_size), new_i_size);
+	mlog(0, "inode %llu, i_size = %llu, new_i_size = %llu\n",
+	     (unsigned long long)le64_to_cpu(fe->i_blkno),
+	     (unsigned long long)le64_to_cpu(fe->i_size),
+	     (unsigned long long)new_i_size);
 
 	/* lets handle the simple truncate cases before doing any more
 	 * cluster locking. */
 	if (new_i_size == le64_to_cpu(fe->i_size))
 		goto bail;
 
+	/* This forces other nodes to sync and drop their pages. Do
+	 * this even if we have a truncate without allocation change -
+	 * ocfs2 cluster sizes can be much greater than page size, so
+	 * we have to truncate them anyway.  */
+	status = ocfs2_data_lock(inode, 1);
+	if (status < 0) {
+		mlog_errno(status);
+		goto bail;
+	}
+	ocfs2_data_unlock(inode, 1);
+
 	if (le32_to_cpu(fe->i_clusters) ==
 	    ocfs2_clusters_for_bytes(osb->sb, new_i_size)) {
 		mlog(0, "fe->i_clusters = %u, so we do a simple truncate\n",
@@ -268,14 +356,6 @@ static int ocfs2_truncate_file(struct inode *inode,
 		goto bail;
 	}
 
-	/* This forces other nodes to sync and drop their pages */
-	status = ocfs2_data_lock(inode, 1);
-	if (status < 0) {
-		mlog_errno(status);
-		goto bail;
-	}
-	ocfs2_data_unlock(inode, 1);
-
 	/* alright, we're going to need to do a full blown alloc size
 	 * change. Orphan the inode so that recovery can complete the
 	 * truncate if necessary. This does the task of marking
@@ -319,7 +399,7 @@ int ocfs2_do_extend_allocation(struct ocfs2_super *osb,
 			       struct inode *inode,
 			       u32 clusters_to_add,
 			       struct buffer_head *fe_bh,
-			       struct ocfs2_journal_handle *handle,
+			       handle_t *handle,
 			       struct ocfs2_alloc_context *data_ac,
 			       struct ocfs2_alloc_context *meta_ac,
 			       enum ocfs2_alloc_restarted *reason_ret)
@@ -378,8 +458,8 @@ int ocfs2_do_extend_allocation(struct ocfs2_super *osb,
 	}
 
 	block = ocfs2_clusters_to_blocks(osb->sb, bit_off);
-	mlog(0, "Allocating %u clusters at block %u for inode %"MLFu64"\n",
-	     num_bits, bit_off, OCFS2_I(inode)->ip_blkno);
+	mlog(0, "Allocating %u clusters at block %u for inode %llu\n",
+	     num_bits, bit_off, (unsigned long long)OCFS2_I(inode)->ip_blkno);
 	status = ocfs2_insert_extent(osb, handle, inode, fe_bh, block,
 				     num_bits, meta_ac);
 	if (status < 0) {
@@ -424,7 +504,7 @@ static int ocfs2_extend_allocation(struct inode *inode,
 	u32 prev_clusters;
 	struct buffer_head *bh = NULL;
 	struct ocfs2_dinode *fe = NULL;
-	struct ocfs2_journal_handle *handle = NULL;
+	handle_t *handle = NULL;
 	struct ocfs2_alloc_context *data_ac = NULL;
 	struct ocfs2_alloc_context *meta_ac = NULL;
 	enum ocfs2_alloc_restarted why;
@@ -449,18 +529,11 @@ static int ocfs2_extend_allocation(struct inode *inode,
 restart_all:
 	BUG_ON(le32_to_cpu(fe->i_clusters) != OCFS2_I(inode)->ip_clusters);
 
-	mlog(0, "extend inode %"MLFu64", i_size = %lld, fe->i_clusters = %u, "
+	mlog(0, "extend inode %llu, i_size = %lld, fe->i_clusters = %u, "
 	     "clusters_to_add = %u\n",
-	     OCFS2_I(inode)->ip_blkno, i_size_read(inode),
+	     (unsigned long long)OCFS2_I(inode)->ip_blkno, i_size_read(inode),
 	     fe->i_clusters, clusters_to_add);
 
-	handle = ocfs2_alloc_handle(osb);
-	if (handle == NULL) {
-		status = -ENOMEM;
-		mlog_errno(status);
-		goto leave;
-	}
-
 	num_free_extents = ocfs2_num_free_extents(osb,
 						  inode,
 						  fe);
@@ -471,10 +544,7 @@ restart_all:
 	}
 
 	if (!num_free_extents) {
-		status = ocfs2_reserve_new_metadata(osb,
-						    handle,
-						    fe,
-						    &meta_ac);
+		status = ocfs2_reserve_new_metadata(osb, fe, &meta_ac);
 		if (status < 0) {
 			if (status != -ENOSPC)
 				mlog_errno(status);
@@ -482,10 +552,7 @@ restart_all:
 		}
 	}
 
-	status = ocfs2_reserve_clusters(osb,
-					handle,
-					clusters_to_add,
-					&data_ac);
+	status = ocfs2_reserve_clusters(osb, clusters_to_add, &data_ac);
 	if (status < 0) {
 		if (status != -ENOSPC)
 			mlog_errno(status);
@@ -500,7 +567,7 @@ restart_all:
 	drop_alloc_sem = 1;
 
 	credits = ocfs2_calc_extend_credits(osb->sb, fe, clusters_to_add);
-	handle = ocfs2_start_trans(osb, handle, credits);
+	handle = ocfs2_start_trans(osb, credits);
 	if (IS_ERR(handle)) {
 		status = PTR_ERR(handle);
 		handle = NULL;
@@ -569,8 +636,8 @@ restarted_transaction:
 		}
 	}
 
-	mlog(0, "fe: i_clusters = %u, i_size=%"MLFu64"\n",
-	     fe->i_clusters, fe->i_size);
+	mlog(0, "fe: i_clusters = %u, i_size=%llu\n",
+	     fe->i_clusters, (unsigned long long)fe->i_size);
 	mlog(0, "inode: ip_clusters=%u, i_size=%lld\n",
 	     OCFS2_I(inode)->ip_clusters, i_size_read(inode));
 
@@ -580,7 +647,7 @@ leave:
 		drop_alloc_sem = 0;
 	}
 	if (handle) {
-		ocfs2_commit_trans(handle);
+		ocfs2_commit_trans(osb, handle);
 		handle = NULL;
 	}
 	if (data_ac) {
@@ -606,7 +673,8 @@ leave:
 
 /* Some parts of this taken from generic_cont_expand, which turned out
  * to be too fragile to do exactly what we need without us having to
- * worry about recursive locking in ->commit_write(). */
+ * worry about recursive locking in ->prepare_write() and
+ * ->commit_write(). */
 static int ocfs2_write_zero_page(struct inode *inode,
 				 u64 size)
 {
@@ -614,7 +682,7 @@ static int ocfs2_write_zero_page(struct inode *inode,
 	struct page *page;
 	unsigned long index;
 	unsigned int offset;
-	struct ocfs2_journal_handle *handle = NULL;
+	handle_t *handle = NULL;
 	int ret;
 
 	offset = (size & (PAGE_CACHE_SIZE-1)); /* Within page */
@@ -634,7 +702,7 @@ static int ocfs2_write_zero_page(struct inode *inode,
 		goto out;
 	}
 
-	ret = ocfs2_prepare_write(NULL, page, offset, offset);
+	ret = ocfs2_prepare_write_nolock(inode, page, offset, offset);
 	if (ret < 0) {
 		mlog_errno(ret);
 		goto out_unlock;
@@ -658,7 +726,7 @@ static int ocfs2_write_zero_page(struct inode *inode,
 		ret = 0;
 
 	if (handle)
-		ocfs2_commit_trans(handle);
+		ocfs2_commit_trans(OCFS2_SB(inode->i_sb), handle);
 out_unlock:
 	unlock_page(page);
 	page_cache_release(page);
@@ -682,19 +750,38 @@ static int ocfs2_zero_extend(struct inode *inode,
 		}
 
 		start_off += sb->s_blocksize;
+
+		/*
+		 * Very large extends have the potential to lock up
+		 * the cpu for extended periods of time.
+		 */
+		cond_resched();
 	}
 
 out:
 	return ret;
 }
 
+/* 
+ * A tail_to_skip value > 0 indicates that we're being called from
+ * ocfs2_file_aio_write(). This has the following implications:
+ *
+ * - we don't want to update i_size
+ * - di_bh will be NULL, which is fine because it's only used in the
+ *   case where we want to update i_size.
+ * - ocfs2_zero_extend() will then only be filling the hole created
+ *   between i_size and the start of the write.
+ */
 static int ocfs2_extend_file(struct inode *inode,
 			     struct buffer_head *di_bh,
-			     u64 new_i_size)
+			     u64 new_i_size,
+			     size_t tail_to_skip)
 {
 	int ret = 0;
 	u32 clusters_to_add;
 
+	BUG_ON(!tail_to_skip && !di_bh);
+
 	/* setattr sometimes calls us like this. */
 	if (new_i_size == 0)
 		goto out;
@@ -706,28 +793,49 @@ static int ocfs2_extend_file(struct inode *inode,
 	clusters_to_add = ocfs2_clusters_for_bytes(inode->i_sb, new_i_size) - 
 		OCFS2_I(inode)->ip_clusters;
 
+	/* 
+	 * protect the pages that ocfs2_zero_extend is going to be
+	 * pulling into the page cache.. we do this before the
+	 * metadata extend so that we don't get into the situation
+	 * where we've extended the metadata but can't get the data
+	 * lock to zero.
+	 */
+	ret = ocfs2_data_lock(inode, 1);
+	if (ret < 0) {
+		mlog_errno(ret);
+		goto out;
+	}
+
 	if (clusters_to_add) {
 		ret = ocfs2_extend_allocation(inode, clusters_to_add);
 		if (ret < 0) {
 			mlog_errno(ret);
-			goto out;
-		}
-
-		ret = ocfs2_zero_extend(inode, new_i_size);
-		if (ret < 0) {
-			mlog_errno(ret);
-			goto out;
+			goto out_unlock;
 		}
-	} 
+	}
 
-	/* No allocation required, we just use this helper to
-	 * do a trivial update of i_size. */
-	ret = ocfs2_simple_size_update(inode, di_bh, new_i_size);
+	/*
+	 * Call this even if we don't add any clusters to the tree. We
+	 * still need to zero the area between the old i_size and the
+	 * new i_size.
+	 */
+	ret = ocfs2_zero_extend(inode, (u64)new_i_size - tail_to_skip);
 	if (ret < 0) {
 		mlog_errno(ret);
-		goto out;
+		goto out_unlock;
 	}
 
+	if (!tail_to_skip) {
+		/* We're being called from ocfs2_setattr() which wants
+		 * us to update i_size */
+		ret = ocfs2_simple_size_update(inode, di_bh, new_i_size);
+		if (ret < 0)
+			mlog_errno(ret);
+	}
+
+out_unlock:
+	ocfs2_data_unlock(inode, 1);
+
 out:
 	return ret;
 }
@@ -739,7 +847,7 @@ int ocfs2_setattr(struct dentry *dentry, struct iattr *attr)
 	struct super_block *sb = inode->i_sb;
 	struct ocfs2_super *osb = OCFS2_SB(sb);
 	struct buffer_head *bh = NULL;
-	struct ocfs2_journal_handle *handle = NULL;
+	handle_t *handle = NULL;
 
 	mlog_entry("(0x%p, '%.*s')\n", dentry,
 	           dentry->d_name.len, dentry->d_name.name);
@@ -750,13 +858,15 @@ int ocfs2_setattr(struct dentry *dentry, struct iattr *attr)
 		mlog(0, "uid change: %d\n", attr->ia_uid);
 	if (attr->ia_valid & ATTR_GID)
 		mlog(0, "gid change: %d\n", attr->ia_gid);
+	if (attr->ia_valid & ATTR_TAG)
+		mlog(0, "tag change: %d\n", attr->ia_tag);
 	if (attr->ia_valid & ATTR_SIZE)
 		mlog(0, "size change...\n");
 	if (attr->ia_valid & (ATTR_ATIME | ATTR_MTIME | ATTR_CTIME))
 		mlog(0, "time change...\n");
 
 #define OCFS2_VALID_ATTRS (ATTR_ATIME | ATTR_MTIME | ATTR_CTIME | ATTR_SIZE \
-			   | ATTR_GID | ATTR_UID | ATTR_MODE)
+			   | ATTR_GID | ATTR_UID | ATTR_TAG | ATTR_MODE)
 	if (!(attr->ia_valid & OCFS2_VALID_ATTRS)) {
 		mlog(0, "can't handle attrs: 0x%x\n", attr->ia_valid);
 		return 0;
@@ -775,7 +885,7 @@ int ocfs2_setattr(struct dentry *dentry, struct iattr *attr)
 		}
 	}
 
-	status = ocfs2_meta_lock(inode, NULL, &bh, 1);
+	status = ocfs2_meta_lock(inode, &bh, 1);
 	if (status < 0) {
 		if (status != -ENOENT)
 			mlog_errno(status);
@@ -786,7 +896,7 @@ int ocfs2_setattr(struct dentry *dentry, struct iattr *attr)
 		if (i_size_read(inode) > attr->ia_size)
 			status = ocfs2_truncate_file(inode, bh, attr->ia_size);
 		else
-			status = ocfs2_extend_file(inode, bh, attr->ia_size);
+			status = ocfs2_extend_file(inode, bh, attr->ia_size, 0);
 		if (status < 0) {
 			if (status != -ENOSPC)
 				mlog_errno(status);
@@ -795,7 +905,7 @@ int ocfs2_setattr(struct dentry *dentry, struct iattr *attr)
 		}
 	}
 
-	handle = ocfs2_start_trans(osb, NULL, OCFS2_INODE_UPDATE_CREDITS);
+	handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS);
 	if (IS_ERR(handle)) {
 		status = PTR_ERR(handle);
 		mlog_errno(status);
@@ -813,7 +923,7 @@ int ocfs2_setattr(struct dentry *dentry, struct iattr *attr)
 		mlog_errno(status);
 
 bail_commit:
-	ocfs2_commit_trans(handle);
+	ocfs2_commit_trans(osb, handle);
 bail_unlock:
 	ocfs2_meta_unlock(inode, 1);
 bail_unlock_rw:
@@ -856,19 +966,39 @@ bail:
 	return err;
 }
 
+int ocfs2_permission(struct inode *inode, int mask, struct nameidata *nd)
+{
+	int ret;
+
+	mlog_entry_void();
+
+	ret = ocfs2_meta_lock(inode, NULL, 0);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ret = generic_permission(inode, mask, NULL);
+
+	ocfs2_meta_unlock(inode, 0);
+out:
+	mlog_exit(ret);
+	return ret;
+}
+
 static int ocfs2_write_remove_suid(struct inode *inode)
 {
 	int ret;
 	struct buffer_head *bh = NULL;
 	struct ocfs2_inode_info *oi = OCFS2_I(inode);
-	struct ocfs2_journal_handle *handle;
+	handle_t *handle;
 	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
 	struct ocfs2_dinode *di;
 
-	mlog_entry("(Inode %"MLFu64", mode 0%o)\n", oi->ip_blkno,
-		   inode->i_mode);
+	mlog_entry("(Inode %llu, mode 0%o)\n",
+		   (unsigned long long)oi->ip_blkno, inode->i_mode);
 
-	handle = ocfs2_start_trans(osb, NULL, OCFS2_INODE_UPDATE_CREDITS);
+	handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS);
 	if (handle == NULL) {
 		ret = -ENOMEM;
 		mlog_errno(ret);
@@ -901,77 +1031,29 @@ static int ocfs2_write_remove_suid(struct inode *inode)
 out_bh:
 	brelse(bh);
 out_trans:
-	ocfs2_commit_trans(handle);
+	ocfs2_commit_trans(osb, handle);
 out:
 	mlog_exit(ret);
 	return ret;
 }
 
-static inline int ocfs2_write_should_remove_suid(struct inode *inode)
-{
-	mode_t mode = inode->i_mode;
-
-	if (!capable(CAP_FSETID)) {
-		if (unlikely(mode & S_ISUID))
-			return 1;
-
-		if (unlikely((mode & S_ISGID) && (mode & S_IXGRP)))
-			return 1;
-	}
-	return 0;
-}
-
-static ssize_t ocfs2_file_aio_write(struct kiocb *iocb,
-				    const char __user *buf,
-				    size_t count,
-				    loff_t pos)
+static int ocfs2_prepare_inode_for_write(struct dentry *dentry,
+					 loff_t *ppos,
+					 size_t count,
+					 int appending)
 {
-	struct iovec local_iov = { .iov_base = (void __user *)buf,
-				   .iov_len = count };
-	int ret, rw_level = -1, meta_level = -1, have_alloc_sem = 0;
+	int ret = 0, meta_level = appending;
+	struct inode *inode = dentry->d_inode;
 	u32 clusters;
-	struct file *filp = iocb->ki_filp;
-	struct inode *inode = filp->f_dentry->d_inode;
 	loff_t newsize, saved_pos;
 
-	mlog_entry("(0x%p, 0x%p, %u, '%.*s')\n", filp, buf,
-		   (unsigned int)count,
-		   filp->f_dentry->d_name.len,
-		   filp->f_dentry->d_name.name);
-
-	/* happy write of zero bytes */
-	if (count == 0)
-		return 0;
-
-	if (!inode) {
-		mlog(0, "bad inode\n");
-		return -EIO;
-	}
-
-	mutex_lock(&inode->i_mutex);
-	/* to match setattr's i_mutex -> i_alloc_sem -> rw_lock ordering */
-	if (filp->f_flags & O_DIRECT) {
-		have_alloc_sem = 1;
-		down_read(&inode->i_alloc_sem);
-	}
-
-	/* concurrent O_DIRECT writes are allowed */
-	rw_level = (filp->f_flags & O_DIRECT) ? 0 : 1;
-	ret = ocfs2_rw_lock(inode, rw_level);
-	if (ret < 0) {
-		rw_level = -1;
-		mlog_errno(ret);
-		goto out;
-	}
-
 	/* 
 	 * We sample i_size under a read level meta lock to see if our write
 	 * is extending the file, if it is we back off and get a write level
 	 * meta lock.
 	 */
-	meta_level = (filp->f_flags & O_APPEND) ? 1 : 0;
 	for(;;) {
-		ret = ocfs2_meta_lock(inode, NULL, NULL, meta_level);
+		ret = ocfs2_meta_lock(inode, NULL, meta_level);
 		if (ret < 0) {
 			meta_level = -1;
 			mlog_errno(ret);
@@ -987,7 +1069,7 @@ static ssize_t ocfs2_file_aio_write(struct kiocb *iocb,
 		 * inode. There's also the dinode i_size state which
 		 * can be lost via setattr during extending writes (we
 		 * set inode->i_size at the end of a write. */
-		if (ocfs2_write_should_remove_suid(inode)) {
+		if (should_remove_suid(dentry)) {
 			if (meta_level == 0) {
 				ocfs2_meta_unlock(inode, meta_level);
 				meta_level = 1;
@@ -997,17 +1079,17 @@ static ssize_t ocfs2_file_aio_write(struct kiocb *iocb,
 			ret = ocfs2_write_remove_suid(inode);
 			if (ret < 0) {
 				mlog_errno(ret);
-				goto out;
+				goto out_unlock;
 			}
 		}
 
 		/* work on a copy of ppos until we're sure that we won't have
 		 * to recalculate it due to relocking. */
-		if (filp->f_flags & O_APPEND) {
+		if (appending) {
 			saved_pos = i_size_read(inode);
 			mlog(0, "O_APPEND: inode->i_size=%llu\n", saved_pos);
 		} else {
-			saved_pos = iocb->ki_pos;
+			saved_pos = *ppos;
 		}
 		newsize = count + saved_pos;
 
@@ -1042,33 +1124,71 @@ static ssize_t ocfs2_file_aio_write(struct kiocb *iocb,
 		if (!clusters)
 			break;
 
-		ret = ocfs2_extend_allocation(inode, clusters);
+		ret = ocfs2_extend_file(inode, NULL, newsize, count);
 		if (ret < 0) {
 			if (ret != -ENOSPC)
 				mlog_errno(ret);
-			goto out;
-		}
-
-		/* Fill any holes which would've been created by this
-		 * write. If we're O_APPEND, this will wind up
-		 * (correctly) being a noop. */
-		ret = ocfs2_zero_extend(inode, (u64) newsize - count);
-		if (ret < 0) {
-			mlog_errno(ret);
-			goto out;
+			goto out_unlock;
 		}
 		break;
 	}
 
-	/* ok, we're done with i_size and alloc work */
-	iocb->ki_pos = saved_pos;
+	if (appending)
+		*ppos = saved_pos;
+
+out_unlock:
 	ocfs2_meta_unlock(inode, meta_level);
-	meta_level = -1;
+
+out:
+	return ret;
+}
+
+static ssize_t ocfs2_file_aio_write(struct kiocb *iocb,
+				    const struct iovec *iov,
+				    unsigned long nr_segs,
+				    loff_t pos)
+{
+	int ret, rw_level, have_alloc_sem = 0;
+	struct file *filp = iocb->ki_filp;
+	struct inode *inode = filp->f_path.dentry->d_inode;
+	int appending = filp->f_flags & O_APPEND ? 1 : 0;
+
+	mlog_entry("(0x%p, %u, '%.*s')\n", filp,
+		   (unsigned int)nr_segs,
+		   filp->f_path.dentry->d_name.len,
+		   filp->f_path.dentry->d_name.name);
+
+	/* happy write of zero bytes */
+	if (iocb->ki_left == 0)
+		return 0;
+
+	mutex_lock(&inode->i_mutex);
+	/* to match setattr's i_mutex -> i_alloc_sem -> rw_lock ordering */
+	if (filp->f_flags & O_DIRECT) {
+		have_alloc_sem = 1;
+		down_read(&inode->i_alloc_sem);
+	}
+
+	/* concurrent O_DIRECT writes are allowed */
+	rw_level = (filp->f_flags & O_DIRECT) ? 0 : 1;
+	ret = ocfs2_rw_lock(inode, rw_level);
+	if (ret < 0) {
+		rw_level = -1;
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ret = ocfs2_prepare_inode_for_write(filp->f_path.dentry, &iocb->ki_pos,
+					    iocb->ki_left, appending);
+	if (ret < 0) {
+		mlog_errno(ret);
+		goto out;
+	}
 
 	/* communicate with ocfs2_dio_end_io */
 	ocfs2_iocb_set_rw_locked(iocb);
 
-	ret = generic_file_aio_write_nolock(iocb, &local_iov, 1, &iocb->ki_pos);
+	ret = generic_file_aio_write_nolock(iocb, iov, nr_segs, iocb->ki_pos);
 
 	/* buffered aio wouldn't have proper lock coverage today */
 	BUG_ON(ret == -EIOCBQUEUED && !(filp->f_flags & O_DIRECT));
@@ -1089,8 +1209,6 @@ static ssize_t ocfs2_file_aio_write(struct kiocb *iocb,
 	}
 
 out:
-	if (meta_level != -1)
-		ocfs2_meta_unlock(inode, meta_level);
 	if (have_alloc_sem)
 		up_read(&inode->i_alloc_sem);
 	if (rw_level != -1) 
@@ -1101,19 +1219,90 @@ out:
 	return ret;
 }
 
+static ssize_t ocfs2_file_splice_write(struct pipe_inode_info *pipe,
+				       struct file *out,
+				       loff_t *ppos,
+				       size_t len,
+				       unsigned int flags)
+{
+	int ret;
+	struct inode *inode = out->f_path.dentry->d_inode;
+
+	mlog_entry("(0x%p, 0x%p, %u, '%.*s')\n", out, pipe,
+		   (unsigned int)len,
+		   out->f_path.dentry->d_name.len,
+		   out->f_path.dentry->d_name.name);
+
+	inode_double_lock(inode, pipe->inode);
+
+	ret = ocfs2_rw_lock(inode, 1);
+	if (ret < 0) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ret = ocfs2_prepare_inode_for_write(out->f_path.dentry, ppos, len, 0);
+	if (ret < 0) {
+		mlog_errno(ret);
+		goto out_unlock;
+	}
+
+	/* ok, we're done with i_size and alloc work */
+	ret = generic_file_splice_write_nolock(pipe, out, ppos, len, flags);
+
+out_unlock:
+	ocfs2_rw_unlock(inode, 1);
+out:
+	inode_double_unlock(inode, pipe->inode);
+
+	mlog_exit(ret);
+	return ret;
+}
+
+static ssize_t ocfs2_file_splice_read(struct file *in,
+				      loff_t *ppos,
+				      struct pipe_inode_info *pipe,
+				      size_t len,
+				      unsigned int flags)
+{
+	int ret = 0;
+	struct inode *inode = in->f_path.dentry->d_inode;
+
+	mlog_entry("(0x%p, 0x%p, %u, '%.*s')\n", in, pipe,
+		   (unsigned int)len,
+		   in->f_path.dentry->d_name.len,
+		   in->f_path.dentry->d_name.name);
+
+	/*
+	 * See the comment in ocfs2_file_aio_read()
+	 */
+	ret = ocfs2_meta_lock(inode, NULL, 0);
+	if (ret < 0) {
+		mlog_errno(ret);
+		goto bail;
+	}
+	ocfs2_meta_unlock(inode, 0);
+
+	ret = generic_file_splice_read(in, ppos, pipe, len, flags);
+
+bail:
+	mlog_exit(ret);
+	return ret;
+}
+
 static ssize_t ocfs2_file_aio_read(struct kiocb *iocb,
-				   char __user *buf,
-				   size_t count,
+				   const struct iovec *iov,
+				   unsigned long nr_segs,
 				   loff_t pos)
 {
-	int ret = 0, rw_level = -1, have_alloc_sem = 0;
+	int ret = 0, rw_level = -1, have_alloc_sem = 0, lock_level = 0;
 	struct file *filp = iocb->ki_filp;
-	struct inode *inode = filp->f_dentry->d_inode;
+	struct inode *inode = filp->f_path.dentry->d_inode;
 
-	mlog_entry("(0x%p, 0x%p, %u, '%.*s')\n", filp, buf,
-		   (unsigned int)count,
-		   filp->f_dentry->d_name.len,
-		   filp->f_dentry->d_name.name);
+	mlog_entry("(0x%p, %u, '%.*s')\n", filp,
+		   (unsigned int)nr_segs,
+		   filp->f_path.dentry->d_name.len,
+		   filp->f_path.dentry->d_name.name);
 
 	if (!inode) {
 		ret = -EINVAL;
@@ -1139,7 +1328,23 @@ static ssize_t ocfs2_file_aio_read(struct kiocb *iocb,
 		ocfs2_iocb_set_rw_locked(iocb);
 	}
 
-	ret = generic_file_aio_read(iocb, buf, count, iocb->ki_pos);
+	/*
+	 * We're fine letting folks race truncates and extending
+	 * writes with read across the cluster, just like they can
+	 * locally. Hence no rw_lock during read.
+	 * 
+	 * Take and drop the meta data lock to update inode fields
+	 * like i_size. This allows the checks down below
+	 * generic_file_aio_read() a chance of actually working. 
+	 */
+	ret = ocfs2_meta_lock_atime(inode, filp->f_vfsmnt, &lock_level);
+	if (ret < 0) {
+		mlog_errno(ret);
+		goto bail;
+	}
+	ocfs2_meta_unlock(inode, lock_level);
+
+	ret = generic_file_aio_read(iocb, iov, nr_segs, iocb->ki_pos);
 	if (ret == -EINVAL)
 		mlog(ML_ERROR, "generic_file_aio_read returned -EINVAL\n");
 
@@ -1165,14 +1370,17 @@ bail:
 struct inode_operations ocfs2_file_iops = {
 	.setattr	= ocfs2_setattr,
 	.getattr	= ocfs2_getattr,
+	.sync_flags     = ocfs2_sync_flags,
+	.permission	= ocfs2_permission,
 };
 
 struct inode_operations ocfs2_special_file_iops = {
 	.setattr	= ocfs2_setattr,
 	.getattr	= ocfs2_getattr,
+	.permission	= ocfs2_permission,
 };
 
-struct file_operations ocfs2_fops = {
+const struct file_operations ocfs2_fops = {
 	.read		= do_sync_read,
 	.write		= do_sync_write,
 	.sendfile	= generic_file_sendfile,
@@ -1182,10 +1390,14 @@ struct file_operations ocfs2_fops = {
 	.open		= ocfs2_file_open,
 	.aio_read	= ocfs2_file_aio_read,
 	.aio_write	= ocfs2_file_aio_write,
+	.ioctl		= ocfs2_ioctl,
+	.splice_read	= ocfs2_file_splice_read,
+	.splice_write	= ocfs2_file_splice_write,
 };
 
-struct file_operations ocfs2_dops = {
+const struct file_operations ocfs2_dops = {
 	.read		= generic_read_dir,
 	.readdir	= ocfs2_readdir,
 	.fsync		= ocfs2_sync_file,
+	.ioctl		= ocfs2_ioctl,
 };