+cc hch given his response
On Mon, Jul 07, 2025 at 02:47:38PM +0200, Konstantin Komarov wrote:
> This reverts commit 69505fe98f198ee813898cbcaf6770949636430b.
>
> Make lock acquiring conditional to avoid the deadlock.
>
> Fixes: 69505fe98f19 ("fs/ntfs3: Replace inode_trylock with inode_lock")
> Reported-by: syzbot+a5d1c9dfa91705cd2f6d@xxxxxxxxxxxxxxxxxxxxxxxxx
> Suggested-by: Lorenzo Stoakes <lorenzo.stoakes@xxxxxxxxxx>
OK you made me have to go searching to figure out where I suggested this :) it
was in:
https://lore.kernel.org/all/2076fb41-4017-4ae4-99df-acfb6148a408@lucifer.local/
Which is in fact syzbot+a91fcdbd2698f99db8f4@xxxxxxxxxxxxxxxxxxxxxxxxx ...
I'm guessing the reported-by you've mentioned here is the same thing but would
have been helpful to link the original report.
Anyway I'm with Christoph here - your commit message is really not great,
esp. since you don't link the report where I provided you with an analysis.
So I suggest you:
- Summarise what I said there in commit message
- Provide a Reported-by tag for the syzkaller report I gave, since it's more
useful than the other one (which is presumably a dupe anyway)
- Provide a Closes tag linking that report so it's easy for people to see.
Thanks!
> Signed-off-by: Konstantin Komarov <almaz.alexandrovich@xxxxxxxxxxxxxxxxxxxx>
> ---
> fs/ntfs3/file.c | 5 ++++-
> 1 file changed, 4 insertions(+), 1 deletion(-)
>
> diff --git a/fs/ntfs3/file.c b/fs/ntfs3/file.c
> index 65fb27d1e17c..2e321b84a1ed 100644
> --- a/fs/ntfs3/file.c
> +++ b/fs/ntfs3/file.c
> @@ -322,7 +322,10 @@ static int ntfs_file_mmap(struct file *file, struct vm_area_struct *vma)
> }
>
> if (ni->i_valid < to) {
> - inode_lock(inode);
> + if (!inode_trylock(inode)) {
> + err = -EAGAIN;
> + goto out;
> + }
> err = ntfs_extend_initialized_size(file, ni,
> ni->i_valid, to);
> inode_unlock(inode);
> --
> 2.43.0
>
