Skip to contentSkip to conversation

f2fs: fix deadlock during inline_data conversion

PATCH New Jaegeuk Kim 1 participant 3 messages Nov 25, 2014 – Nov 26, 2014

Subsystems (2)
VFS F2FS file system
lore
Share & tools
lore

Jaegeuk Kim

author maintainer
#0 lore raw

PATCH 1/2 f2fs: fix deadlock during inline_data conversion

A deadlock can be occurred:
Thread 1] Thread 2]
- f2fs_write_data_pages - f2fs_write_begin
- lock_page(page #0)
- grab_cache_page(page #X)
- get_node_page(inode_page)
- grab_cache_page(page #0) : to convert inline_data
- f2fs_write_data_page
- f2fs_write_inline_data
- get_node_page(inode_page)

In this case, trying to lock inode_page and page #0 causes deadlock. In order to avoid this, this patch adds a rule for this locking policy, which is that page #0 should be locked followed by inode_page lock.

Signed-off-by Jaegeuk Kim <[email protected]>
---
 fs/f2fs/data.c | 28 ++++++++++++++--------------
 1 file changed, 14 insertions(+), 14 deletions(-)
fs/f2fs/data.c +14 -14
index 12dd58a..c7bc626 100644
@@ -936,6 +936,17 @@ static int f2fs_write_begin(struct file *file, struct address_space *mapping,
 	trace_f2fs_write_begin(inode, pos, len, flags);
 
 	f2fs_balance_fs(sbi);
+
+	/*
+	 * We should check this at this moment to avoid deadlock on inode page
+	 * and #0 page. The locking rule for inline_data conversion should be:
+	 * lock_page(page #0) -> lock_page(inode_page)
+	 */
+	if (index != 0) {
+		err = f2fs_convert_inline_inode(inode);
+		if (err)
+			goto fail;
+	}
 repeat:
 	page = grab_cache_page_write_begin(mapping, index, flags);
 	if (!page) {

@@ -960,21 +971,10 @@ repeat:
 			set_inode_flag(F2FS_I(inode), FI_DATA_EXIST);
 			sync_inode_page(&dn);
 			goto put_next;
-		} else if (page->index == 0) {
-			err = f2fs_convert_inline_page(&dn, page);
-			if (err)
-				goto put_fail;
-		} else {
-			struct page *p = grab_cache_page(inode->i_mapping, 0);
-			if (!p) {
-				err = -ENOMEM;
-				goto put_fail;
-			}
-			err = f2fs_convert_inline_page(&dn, p);
-			f2fs_put_page(p, 1);
-			if (err)
-				goto put_fail;
 		}
+		err = f2fs_convert_inline_page(&dn, page);
+		if (err)
+			goto put_fail;
 	}
 	err = f2fs_reserve_block(&dn, index);
 	if (err)
Signature · 9 lines
2.1.1


------------------------------------------------------------------------------
Download BIRT iHub F-Type - The Free Enterprise-Grade BIRT Server
from Actuate! Instantly Supercharge Your Business Reports and Dashboards
with Interactivity, Sharing, Native Excel Exports, App Integration & more
Get technology previously reserved for billion-dollar corporations, FREE
http://pubads.g.doubleclick.net/gampad/clk?id=157005751&iu=/4140/ostg.clktrk
author maintainer
#1 lore raw

PATCH 2/2 f2fs: make clean the page before writing

If a page is set to be written to the disk, we can make clean the page.

Signed-off-by Jaegeuk Kim <[email protected]>
---
 fs/f2fs/inline.c | 3 +++
 1 file changed, 3 insertions(+)
fs/f2fs/inline.c +3 -0
index 914b6d3..7c96c27 100644
@@ -110,6 +110,9 @@ int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page)
 	kunmap_atomic(dst_addr);
 	SetPageUptodate(page);
 no_update:
+	/* clear dirty state */
+	clear_page_dirty_for_io(page);
+
 	/* write data page to try to make data consistent */
 	set_page_writeback(page);
Signature · 1 line
2.1.1
author maintainer
#2 lore raw

PATCH v2 2/2 f2fs: make clean the page before writing

Change log from v1:
o fix to handle # of dirty pages

If a page is set to be written to the disk, we can make clean the page.

Signed-off-by Jaegeuk Kim <[email protected]>
---
 fs/f2fs/inline.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)
fs/f2fs/inline.c +6 -1
index 914b6d3..e27f290 100644
@@ -84,7 +84,7 @@ int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page)
 		.type = DATA,
 		.rw = WRITE_SYNC | REQ_PRIO,
 	};
-	int err;
+	int dirty, err;
 
 	f2fs_bug_on(F2FS_I_SB(dn->inode), page->index);
 

@@ -110,12 +110,17 @@ int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page)
 	kunmap_atomic(dst_addr);
 	SetPageUptodate(page);
 no_update:
+	/* clear dirty state */
+	dirty = clear_page_dirty_for_io(page);
+
 	/* write data page to try to make data consistent */
 	set_page_writeback(page);
 
 	write_data_page(page, dn, &new_blk_addr, &fio);
 	update_extent_cache(new_blk_addr, dn);
 	f2fs_wait_on_page_writeback(page, DATA);
+	if (dirty)
+		inode_dec_dirty_pages(dn->inode);
 
 	/* clear inline data and flag after data writeback */
 	truncate_inline_data(dn->inode_page, 0);
Signature · 1 line
2.1.1