On 2021/8/31 11:02, Theodore Ts'o wrote:

On Thu, Aug 26, 2021 at 09:04:12PM +0800, Zhang Yi wrote:

quoted

So this patch initialize the inode buffer by filling the in-mem inode
contents if we skip read I/O, ensure that the buffer is really uptodate.

Signed-off-by: Zhang Yi <yi.zhang@huawei.com>
---
 fs/ext4/inode.c | 22 ++++++++++++++++------
 1 file changed, 16 insertions(+), 6 deletions(-)

diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
index 3c36e701e30e..8b37f55b04ad 100644
--- a/fs/ext4/inode.c
+++ b/fs/ext4/inode.c

@@ -4446,8 +4446,8 @@ static int ext4_fill_raw_inode(struct inode *inode, struct ext4_inode *raw_inode
  * inode.
  */
 static int __ext4_get_inode_loc(struct super_block *sb, unsigned long ino,
-				struct ext4_iloc *iloc, int in_mem,
-				ext4_fsblk_t *ret_block)
+				struct inode *inode, struct ext4_iloc *iloc,
+				int in_mem, ext4_fsblk_t *ret_block)

In this patch you've added a new argument 'inode'.  However, if in_mem
is true, and inode is NULL, the kernel will crash with a null pointer
dereference.  Furthermore, whenever in_mem is false, the callers pass
in NULL for inode.

Given that, perhaps we should just drop the in_mem argument, and then
instead of

	if (in_mem) {

we do:

	if (inode && !ext4_test_inode_state(inode, EXT4_STATE_XATTR) {

with the comments adjusted accordingly?

I think it will make the code a bit simpler and readable.

What do you think?

Yes，although I have already prevent passing 'in_mem' is true but 'inode' is
NULL in ext4_get_inode_loc(), using two arguments show the inode in-mem case
is not safe. I will remove the 'in_mem' parameter.

Thanks,
Yi.