[PATCH 0/5] pending items for 2.6.37

[PATCH 0/5] pending items for 2.6.37

[Sage Weil]

This is the current set of patches I have queued for the next -rc.  The 
direct-io page dirty ones could really use some review.  Also the buffer
alignment.

Thanks!
sage

---

Henry C Chang (3):
  ceph: fix direct-io on non-page-aligned buffers
  ceph: mark user pages dirty on direct-io reads
  ceph: handle partial result from get_user_pages

Sage Weil (2):
  ceph: fix msgr_init error path
  ceph: fix null pointer dereference in ceph_init_dentry for nfs
    reexport

 fs/ceph/dir.c                |    3 ++-
 fs/ceph/file.c               |   39 +++++++++++++++++++++++----------------
 include/linux/ceph/libceph.h |    6 ++++--
 net/ceph/messenger.c         |    8 +++-----
 net/ceph/pagevec.c           |   15 +++++++++------
 5 files changed, 41 insertions(+), 30 deletions(-)

[PATCH 1/5] ceph: fix msgr_init error path

[Sage Weil]

create_workqueue() returns NULL on failure.

Signed-off-by: Sage Weil <sage@newdream.net>
---
 net/ceph/messenger.c |    8 +++-----
 1 files changed, 3 insertions(+), 5 deletions(-)

diff --git a/net/ceph/messenger.c b/net/ceph/messenger.c
index 1c7a2ec..b6ff4a1 100644
--- a/net/ceph/messenger.c
+++ b/net/ceph/messenger.c
@@ -97,11 +97,9 @@ struct workqueue_struct *ceph_msgr_wq;
 int ceph_msgr_init(void)
 {
 	ceph_msgr_wq = create_workqueue("ceph-msgr");
-	if (IS_ERR(ceph_msgr_wq)) {
-		int ret = PTR_ERR(ceph_msgr_wq);
-		pr_err("msgr_init failed to create workqueue: %d\n", ret);
-		ceph_msgr_wq = NULL;
-		return ret;
+	if (!ceph_msgr_wq) {
+		pr_err("msgr_init failed to create workqueue\n");
+		return -ENOMEM;
 	}
 	return 0;
 }
-- 
1.6.6.1

[PATCH 2/5] ceph: fix direct-io on non-page-aligned buffers

[Sage Weil]

From: Henry C Chang <henry_c_chang@tcloudcomputing.com>

The user buffer may be 512-byte aligned, not page-aligned.  We were
assuming the buffer was page-aligned and only accounting for
non-page-aligned io offsets.

Signed-off-by: Henry C Chang <henry_c_chang@tcloudcomputing.com>
Signed-off-by: Sage Weil <sage@newdream.net>
---
 fs/ceph/file.c |   31 +++++++++++++++++++------------
 1 files changed, 19 insertions(+), 12 deletions(-)

diff --git a/fs/ceph/file.c b/fs/ceph/file.c
index 8d79b89..e860d8f 100644
--- a/fs/ceph/file.c
+++ b/fs/ceph/file.c
@@ -282,7 +282,8 @@ int ceph_release(struct inode *inode, struct file *file)
 static int striped_read(struct inode *inode,
 			u64 off, u64 len,
 			struct page **pages, int num_pages,
-			int *checkeof, bool align_to_pages)
+			int *checkeof, bool align_to_pages,
+			unsigned long buf_align)
 {
 	struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
 	struct ceph_inode_info *ci = ceph_inode(inode);
@@ -307,7 +308,7 @@ static int striped_read(struct inode *inode,
 
 more:
 	if (align_to_pages)
-		page_align = (pos - io_align) & ~PAGE_MASK;
+		page_align = (pos - io_align + buf_align) & ~PAGE_MASK;
 	else
 		page_align = pos & ~PAGE_MASK;
 	this_len = left;
@@ -376,16 +377,18 @@ static ssize_t ceph_sync_read(struct file *file, char __user *data,
 	struct inode *inode = file->f_dentry->d_inode;
 	struct page **pages;
 	u64 off = *poff;
-	int num_pages = calc_pages_for(off, len);
-	int ret;
+	int num_pages, ret;
 
 	dout("sync_read on file %p %llu~%u %s\n", file, off, len,
 	     (file->f_flags & O_DIRECT) ? "O_DIRECT" : "");
 
-	if (file->f_flags & O_DIRECT)
+	if (file->f_flags & O_DIRECT) {
+		num_pages = calc_pages_for((unsigned long)data, len);
 		pages = ceph_get_direct_page_vector(data, num_pages);
-	else
+	} else {
+		num_pages = calc_pages_for(off, len);
 		pages = ceph_alloc_page_vector(num_pages, GFP_NOFS);
+	}
 	if (IS_ERR(pages))
 		return PTR_ERR(pages);
 
@@ -400,7 +403,8 @@ static ssize_t ceph_sync_read(struct file *file, char __user *data,
 		goto done;
 
 	ret = striped_read(inode, off, len, pages, num_pages, checkeof,
-			   file->f_flags & O_DIRECT);
+			   file->f_flags & O_DIRECT,
+			   (unsigned long)data & ~PAGE_MASK);
 
 	if (ret >= 0 && (file->f_flags & O_DIRECT) == 0)
 		ret = ceph_copy_page_vector_to_user(pages, data, off, ret);
@@ -456,6 +460,7 @@ static ssize_t ceph_sync_write(struct file *file, const char __user *data,
 	int do_sync = 0;
 	int check_caps = 0;
 	int page_align, io_align;
+	unsigned long buf_align;
 	int ret;
 	struct timespec mtime = CURRENT_TIME;
 
@@ -471,6 +476,7 @@ static ssize_t ceph_sync_write(struct file *file, const char __user *data,
 		pos = *offset;
 
 	io_align = pos & ~PAGE_MASK;
+	buf_align = (unsigned long)data & ~PAGE_MASK;
 
 	ret = filemap_write_and_wait_range(inode->i_mapping, pos, pos + left);
 	if (ret < 0)
@@ -496,12 +502,15 @@ static ssize_t ceph_sync_write(struct file *file, const char __user *data,
 	 */
 more:
 	len = left;
-	if (file->f_flags & O_DIRECT)
+	if (file->f_flags & O_DIRECT) {
 		/* write from beginning of first page, regardless of
 		   io alignment */
-		page_align = (pos - io_align) & ~PAGE_MASK;
-	else
+		page_align = (pos - io_align + buf_align) & ~PAGE_MASK;
+		num_pages = calc_pages_for((unsigned long)data, len);
+	} else {
 		page_align = pos & ~PAGE_MASK;
+		num_pages = calc_pages_for(pos, len);
+	}
 	req = ceph_osdc_new_request(&fsc->client->osdc, &ci->i_layout,
 				    ceph_vino(inode), pos, &len,
 				    CEPH_OSD_OP_WRITE, flags,
@@ -512,8 +521,6 @@ more:
 	if (!req)
 		return -ENOMEM;
 
-	num_pages = calc_pages_for(pos, len);
-
 	if (file->f_flags & O_DIRECT) {
 		pages = ceph_get_direct_page_vector(data, num_pages);
 		if (IS_ERR(pages)) {
-- 
1.6.6.1

[PATCH 3/5] ceph: fix null pointer dereference in ceph_init_dentry for nfs reexport

[Sage Weil]

The fh_to_dentry etc. methods use ceph_init_dentry(), which assumes that
d_parent is defined.  It isn't for those callers, so check!

Signed-off-by: Sage Weil <sage@newdream.net>
---
 fs/ceph/dir.c |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)

diff --git a/fs/ceph/dir.c b/fs/ceph/dir.c
index 158c700..d902948 100644
--- a/fs/ceph/dir.c
+++ b/fs/ceph/dir.c
@@ -40,7 +40,8 @@ int ceph_init_dentry(struct dentry *dentry)
 	if (dentry->d_fsdata)
 		return 0;
 
-	if (ceph_snap(dentry->d_parent->d_inode) == CEPH_NOSNAP)
+	if (dentry->d_parent == NULL ||   /* nfs fh_to_dentry */
+	    ceph_snap(dentry->d_parent->d_inode) == CEPH_NOSNAP)
 		dentry->d_op = &ceph_dentry_ops;
 	else if (ceph_snap(dentry->d_parent->d_inode) == CEPH_SNAPDIR)
 		dentry->d_op = &ceph_snapdir_dentry_ops;
-- 
1.6.6.1

[PATCH 4/5] ceph: mark user pages dirty on direct-io reads

[Sage Weil]

From: Henry C Chang <henry_c_chang@tcloudcomputing.com>

For read operation, we have to set the argument _write_ of get_user_pages
to 1 since we will write data to pages. Also, we need to SetPageDirty before
releasing these pages.

Signed-off-by: Henry C Chang <henry_c_chang@tcloudcomputing.com>
Signed-off-by: Sage Weil <sage@newdream.net>
---
 fs/ceph/file.c               |    8 ++++----
 include/linux/ceph/libceph.h |    6 ++++--
 net/ceph/pagevec.c           |   11 +++++++----
 3 files changed, 15 insertions(+), 10 deletions(-)

diff --git a/fs/ceph/file.c b/fs/ceph/file.c
index e860d8f..7d0e4a8 100644
--- a/fs/ceph/file.c
+++ b/fs/ceph/file.c
@@ -384,7 +384,7 @@ static ssize_t ceph_sync_read(struct file *file, char __user *data,
 
 	if (file->f_flags & O_DIRECT) {
 		num_pages = calc_pages_for((unsigned long)data, len);
-		pages = ceph_get_direct_page_vector(data, num_pages);
+		pages = ceph_get_direct_page_vector(data, num_pages, true);
 	} else {
 		num_pages = calc_pages_for(off, len);
 		pages = ceph_alloc_page_vector(num_pages, GFP_NOFS);
@@ -413,7 +413,7 @@ static ssize_t ceph_sync_read(struct file *file, char __user *data,
 
 done:
 	if (file->f_flags & O_DIRECT)
-		ceph_put_page_vector(pages, num_pages);
+		ceph_put_page_vector(pages, num_pages, true);
 	else
 		ceph_release_page_vector(pages, num_pages);
 	dout("sync_read result %d\n", ret);
@@ -522,7 +522,7 @@ more:
 		return -ENOMEM;
 
 	if (file->f_flags & O_DIRECT) {
-		pages = ceph_get_direct_page_vector(data, num_pages);
+		pages = ceph_get_direct_page_vector(data, num_pages, false);
 		if (IS_ERR(pages)) {
 			ret = PTR_ERR(pages);
 			goto out;
@@ -572,7 +572,7 @@ more:
 	}
 
 	if (file->f_flags & O_DIRECT)
-		ceph_put_page_vector(pages, num_pages);
+		ceph_put_page_vector(pages, num_pages, false);
 	else if (file->f_flags & O_SYNC)
 		ceph_release_page_vector(pages, num_pages);
 
diff --git a/include/linux/ceph/libceph.h b/include/linux/ceph/libceph.h
index 9e76d35..72c72bf 100644
--- a/include/linux/ceph/libceph.h
+++ b/include/linux/ceph/libceph.h
@@ -227,8 +227,10 @@ extern int ceph_open_session(struct ceph_client *client);
 extern void ceph_release_page_vector(struct page **pages, int num_pages);
 
 extern struct page **ceph_get_direct_page_vector(const char __user *data,
-						 int num_pages);
-extern void ceph_put_page_vector(struct page **pages, int num_pages);
+						 int num_pages,
+						 bool write_page);
+extern void ceph_put_page_vector(struct page **pages, int num_pages,
+				 bool dirty);
 extern void ceph_release_page_vector(struct page **pages, int num_pages);
 extern struct page **ceph_alloc_page_vector(int num_pages, gfp_t flags);
 extern int ceph_copy_user_to_page_vector(struct page **pages,
diff --git a/net/ceph/pagevec.c b/net/ceph/pagevec.c
index ac34fee..01947a5 100644
--- a/net/ceph/pagevec.c
+++ b/net/ceph/pagevec.c
@@ -13,7 +13,7 @@
  * build a vector of user pages
  */
 struct page **ceph_get_direct_page_vector(const char __user *data,
-					  int num_pages)
+					  int num_pages, bool write_page)
 {
 	struct page **pages;
 	int rc;
@@ -24,7 +24,7 @@ struct page **ceph_get_direct_page_vector(const char __user *data,
 
 	down_read(&current->mm->mmap_sem);
 	rc = get_user_pages(current, current->mm, (unsigned long)data,
-			    num_pages, 0, 0, pages, NULL);
+			    num_pages, write_page, 0, pages, NULL);
 	up_read(&current->mm->mmap_sem);
 	if (rc < 0)
 		goto fail;
@@ -36,12 +36,15 @@ fail:
 }
 EXPORT_SYMBOL(ceph_get_direct_page_vector);
 
-void ceph_put_page_vector(struct page **pages, int num_pages)
+void ceph_put_page_vector(struct page **pages, int num_pages, bool dirty)
 {
 	int i;
 
-	for (i = 0; i < num_pages; i++)
+	for (i = 0; i < num_pages; i++) {
+		if (dirty)
+			set_page_dirty_lock(pages[i]);
 		put_page(pages[i]);
+	}
 	kfree(pages);
 }
 EXPORT_SYMBOL(ceph_put_page_vector);
-- 
1.6.6.1

[PATCH 5/5] ceph: handle partial result from get_user_pages

[Sage Weil]

From: Henry C Chang <henry_c_chang@tcloudcomputing.com>

The get_user_pages() helper can return fewer than the requested pages.
Error out in that case, and clean up the partial result.

Signed-off-by: Henry C Chang <henry_c_chang@tcloudcomputing.com>
Signed-off-by: Sage Weil <sage@newdream.net>
---
 net/ceph/pagevec.c |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/net/ceph/pagevec.c b/net/ceph/pagevec.c
index 01947a5..1a040e6 100644
--- a/net/ceph/pagevec.c
+++ b/net/ceph/pagevec.c
@@ -26,12 +26,12 @@ struct page **ceph_get_direct_page_vector(const char __user *data,
 	rc = get_user_pages(current, current->mm, (unsigned long)data,
 			    num_pages, write_page, 0, pages, NULL);
 	up_read(&current->mm->mmap_sem);
-	if (rc < 0)
+	if (rc < num_pages)
 		goto fail;
 	return pages;
 
 fail:
-	kfree(pages);
+	ceph_put_page_vector(pages, rc > 0 ? rc : 0, false);
 	return ERR_PTR(rc);
 }
 EXPORT_SYMBOL(ceph_get_direct_page_vector);
-- 
1.6.6.1