From: Jan Kara <[email protected]>
Under heavy memory and filesystem load, users observe the assertion
mapping->nrpages == 0 in end_writeback() trigger. This can be caused by
page reclaim reclaiming the last page from a mapping in the following
race:
CPU0 CPU1
...
shrink_page_list()
__remove_mapping()
__delete_from_page_cache()
radix_tree_delete()
evict_inode()
truncate_inode_pages()
truncate_inode_pages_range()
pagevec_lookup() - finds nothing
end_writeback()
mapping->nrpages != 0 -> BUG
page->mapping = NULL
mapping->nrpages--
Fix the problem by doing a reliable check of mapping->nrpages under
mapping->tree_lock in end_writeback().
Analyzed by Jay <[email protected]>, lost in LKML, and dug
out by Miklos Szeredi <[email protected]>.
Cc: Jay <[email protected]>
Cc: Miklos Szeredi <[email protected]>
Signed-off-by: Jan Kara <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
---
fs/inode.c | 7 +++++++
include/linux/fs.h | 1 +
mm/truncate.c | 5 +++++
3 files changed, 13 insertions(+)
diff -puN fs/inode.c~mm-fix-assertion-mapping-nrpages-==-0-in-end_writeback
fs/inode.c
--- a/fs/inode.c~mm-fix-assertion-mapping-nrpages-==-0-in-end_writeback
+++ a/fs/inode.c
@@ -423,7 +423,14 @@ EXPORT_SYMBOL(remove_inode_hash);
void end_writeback(struct inode *inode)
{
might_sleep();
+ /*
+ * We have to cycle tree_lock here because reclaim can be still in the
+ * process of removing the last page (in __delete_from_page_cache())
+ * and we must not free mapping under it.
+ */
+ spin_lock_irq(&inode->i_data.tree_lock);
BUG_ON(inode->i_data.nrpages);
+ spin_unlock_irq(&inode->i_data.tree_lock);
BUG_ON(!list_empty(&inode->i_data.private_list));
BUG_ON(!(inode->i_state & I_FREEING));
BUG_ON(inode->i_state & I_CLEAR);
diff -puN
include/linux/fs.h~mm-fix-assertion-mapping-nrpages-==-0-in-end_writeback
include/linux/fs.h
--- a/include/linux/fs.h~mm-fix-assertion-mapping-nrpages-==-0-in-end_writeback
+++ a/include/linux/fs.h
@@ -639,6 +639,7 @@ struct address_space {
struct prio_tree_root i_mmap; /* tree of private and shared
mappings */
struct list_head i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
struct mutex i_mmap_mutex; /* protect tree, count, list */
+ /* Protected by tree_lock together with the radix tree */
unsigned long nrpages; /* number of total pages */
pgoff_t writeback_index;/* writeback starts here */
const struct address_space_operations *a_ops; /* methods */
diff -puN mm/truncate.c~mm-fix-assertion-mapping-nrpages-==-0-in-end_writeback
mm/truncate.c
--- a/mm/truncate.c~mm-fix-assertion-mapping-nrpages-==-0-in-end_writeback
+++ a/mm/truncate.c
@@ -304,6 +304,11 @@ EXPORT_SYMBOL(truncate_inode_pages_range
* @lstart: offset from which to truncate
*
* Called under (and serialised by) inode->i_mutex.
+ *
+ * Note: When this function returns, there can be a page in the process of
+ * deletion (inside __delete_from_page_cache()) in the specified range. Thus
+ * mapping->nrpages can be non-zero when this function returns even after
+ * truncation of the whole mapping.
*/
void truncate_inode_pages(struct address_space *mapping, loff_t lstart)
{
_
_______________________________________________
stable mailing list
[email protected]
http://linux.kernel.org/mailman/listinfo/stable
