Re: Exiting with locks still held (was Re: [PATCH] kmemleak: Fix scheduling-while-atomic bug)

[Linus Torvalds <torvalds@linux-foundation.org>]

On Thu, 2 Jul 2009, Catalin Marinas wrote:
> 
> Initially, the scan_mutex was acquired in kmemleak_open() and released
> in kmemleak_release() (corresponding to /sys/kernel/debug/kmemleak
> operations). This was causing some lockdep reports when the file was
> closed from a different task than the one opening it. This patch moves
> the scan_mutex acquiring in kmemleak_write() or kmemleak_seq_show().

This is better, but not really how you are supposed to do it.

The whole seq-file thing is very much _designed_ for taking a lock at the 
beginning of the operation, and releasing it at the end. It's a very 
common pattern.

But you should _not_ do it in the "show" routine. If you do, you're always 
going to be racy wrt lseek() and friends.

What you _should_ do is to take the lock in the "seq_start" routine, and 
release it in "seq_stop". The "seq_show" routine may be called multiple 
times in between.

For a trivial example, see the drivers/char/misc.c file. Note how it needs 
to hold the lock over the whole list traversal, and how seqfiles allows it 
to do that quite naturally.

		Linus