Re: [patch 1/1] Switch ioctl functions of drivers/scsi/sg.c to unlocked_ioctl

[Andi Kleen <andi@firstfloor.org>]

On Thu, Jan 10, 2008 at 01:03:48PM -0600, James Bottomley wrote:
> 
> On Thu, 2008-01-10 at 19:59 +0100, Andi Kleen wrote:
> > > Really, all this is doing is open coding what the ioctl handler is doing
> > > anyway, isn't it? in which case, why bother to change it at all?
> > 
> > Because once it's open coded it is visible and can then be eliminated.
> > Does SCSI need the BKL at all?
> 
> No, of course not ... it hasn't for ages, otherwise linux performance
> would be in the tank.  If we require the BKL at all in the ioctl path it
> will be to protect the non-SCSI structures we have to use.  Is there any
> guide to which structures in the kernel still require the BKL?

Not many really in the core kernel. Hardly any. Grep for 
lock_kernel to be sure, but there is not much.

It's mostly drivers that still need it.

How about the low level SCSI drivers that might called from the high
level SCSI code? 

Anyways starting these kinds of discussions was the goal of the proposal.
Even if Andre's patch ends up not being merged it would have reached
its goal.

-Andi