Re: devfs: BKL *not* taken while opening devices

[Arjan van de Ven <arjanv@redhat.com>]

On Tue, Apr 30, 2002 at 09:42:32AM -0700, Dave Hansen wrote:
> Russell King wrote:
> > On Mon, Apr 29, 2002 at 06:21:34PM +0100, Arjan van de Ven wrote:
> > 
> >>I'm not convinced of that. It's not nearly a critical path and it's
> >>better to get even the "dumb" drivers safe than to risk having big
> >>security holes in there for years to come.
> > 
> > Would it be worth dropping a  BUG_ON(!kernel_locked()) in tty_open() to
> > catch this type of error?  The tty code heavily relies on the BKL.
> > 
> > This way, such locking problems would get caught early, since everyone
> > uses the tty code during boot, right?
> 
> I like the idea.  But, while we're at it, does anyone have a good enough 
> grasp of locking the the TTY layer that we can start peeling some of the 
> BKL out of there?  Somebody was doing tests over a serial console here 
> and the lockmeter data showed horrible BKL contention and hold times.

I really really doubt that fixing contention will make serial ports go
faster... it'll just move to another lock since I suspect we're
just waiting for hardware