fs: Protect write paths by sb_start_write - sb_end_write
There are several entry points which dirty pages in a filesystem. mmap
(handled by block_page_mkwrite()), buffered write (handled by
__generic_file_aio_write()), splice write (generic_file_splice_write),
truncate, and fallocate (these can dirty last partial page - handled inside
each filesystem separately). Protect these places with sb_start_write() and
sb_end_write().
->page_mkwrite() calls are particularly complex since they are called with
mmap_sem held and thus we cannot use standard sb_start_write() due to lock
ordering constraints. We solve the problem by using a special freeze protection
sb_start_pagefault() which ranks below mmap_sem.
BugLink: https://bugs.launchpad.net/bugs/897421
Tested-by: Kamal Mostafa <kamal@canonical.com>
Tested-by: Peter M. Petrakis <peter.petrakis@canonical.com>
Tested-by: Dann Frazier <dann.frazier@canonical.com>
Tested-by: Massimo Morana <massimo.morana@canonical.com>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
diff --git a/mm/filemap_xip.c b/mm/filemap_xip.c
index 80b34ef..13e013b 100644
--- a/mm/filemap_xip.c
+++ b/mm/filemap_xip.c
@@ -402,6 +402,8 @@
loff_t pos;
ssize_t ret;
+ sb_start_write(inode->i_sb);
+
mutex_lock(&inode->i_mutex);
if (!access_ok(VERIFY_READ, buf, len)) {
@@ -412,8 +414,6 @@
pos = *ppos;
count = len;
- vfs_check_frozen(inode->i_sb, SB_FREEZE_WRITE);
-
/* We can write back this queue in page reclaim */
current->backing_dev_info = mapping->backing_dev_info;
@@ -437,6 +437,7 @@
current->backing_dev_info = NULL;
out_up:
mutex_unlock(&inode->i_mutex);
+ sb_end_write(inode->i_sb);
return ret;
}
EXPORT_SYMBOL_GPL(xip_file_write);