Re: [PATCH v5 1/2] tmpfs: Add per-superblock i_ino support

[Chris Down]

zhengbin (A) writes:
> Use spin_lock will affect performance

"Performance" isn't a binary. In discussions, you should avoid invoking the 
performance boogeyman without presenting any real-world data. :-)

We already have to take this spin lock before when setting the free inode 
count. The two sites can be merged, but it seems unnecessary to conflate their 
purpose at this time.