On Tue, Jun 11, 2013 at 07:50:31AM +0100, Srinivas KANDAGATLA wrote:

You are right, It does not make sense to use BIT() macro for field which
has more than 1 bit. I think using mix of both BIT() and the old style
will make code look bit confusing to reader, Also no other mach code in
the kernel use BIT while configuring L2 controller. So am going to drop
the idea of using BIT here and leave the code as it is.

I'd suggest putting a comment in the code to that effect.  With the way
"cleanups" get done, I wouldn't be surprised if this attracts a lot of
people wanting to do a trivial "1 << bit" -> "BIT(bit)" conversions.

One of the problems of open source is that you can say "no" to a patch
like that until you're blue in the face, but it will eventually make
its way in via some path.

Just one of the reasons I consider BIT() to be evil and an inappropriate
macro.