On 2023/6/4 11:04, Theodore Ts'o wrote:
I tried testing to see if this fixed [1], and it appears to be
triggering a lockdep warning[2] at this line in the patch:
[1] https://syzkaller.appspot.com/bug?extid=f4582777a19ec422b517
[2] https://syzkaller.appspot.com/x/report.txt?x=17260843280000
diff --git a/fs/ext4/file.c b/fs/ext4/file.c
index d101b3b0c7da..9df82d72eb90 100644
--- a/fs/ext4/file.c
+++ b/fs/ext4/file.c
@@ -808,6 +809,27 @@ static int ext4_file_mmap(struct file *file, struct vm_area_struct *vma)
if (!daxdev_mapping_supported(vma, dax_dev))
return -EOPNOTSUPP;
+ /*
+ * Writing via mmap has no logic to handle inline data, so we
+ * need to call ext4_convert_inline_data() to convert the inode
+ * to normal format before doing so, otherwise a BUG_ON will be
+ * triggered in ext4_writepages() due to the
+ * EXT4_STATE_MAY_INLINE_DATA flag. Moreover, we need to grab
+ * i_rwsem during conversion, since clearing and setting the
+ * inline data flag may race with ext4_buffered_write_iter()
+ * to trigger a BUG_ON.
+ */
+ if (ext4_has_feature_inline_data(sb) &&
+ vma->vm_flags & VM_SHARED && vma->vm_flags & VM_MAYWRITE) {
+ int err;
+
+ inode_lock(inode); <=================== LOCKDEP warning
+ err = ext4_convert_inline_data(inode);
+ inode_unlock(inode);
+ if (err)
+ return err;
+ }
The details of the lockdep warning from [2], which appears to be a
mmap(2) racing with a buffered write(2) are below. Could you take a
look?
Thanks!
- Ted
Sorry for the late reply!
Had a look at this question which is similar to the one Honza mentioned
earlier.
Concurrency between write and mmap as follows can lead to ABBA deadlocks:
CPU0 CPU1
write(2) mmap(2)
ext4_file_write_iter
ext4_buffered_write_iter
inode_lock(inode) ---> LOCK A
generic_perform_write
ksys_mmap_pgoff
vm_mmap_pgoff
mmap_write_lock_killable(mm) ---> LOCK B
do_mmap
mmap_region
call_mmap
ext4_file_mmap
inode_lock(inode) ---> try LOCK A again
fault_in_iov_iter_readable |
fault_in_readable |
asm_exc_page_fault ABBA deadlock
handle_page_fault |
do_user_addr_fault |
mmap_read_lock(mm) ---> try LOCK B again
Thanks!
--
With Best Regards,
Baokun Li
.
