bct_p *is* used in loop.c:strength_reduction

[David Edelsohn]

>>>>> "Kaveh R Ghazi" writes:

Kaveh> My original patches included conditionalizations for these
Kaveh> attributes but Jeff suggested that they were making the code too
Kaveh> unreadable and asked that I just use ATTRIBUTE_UNUSED unconditionally.
Kaveh> (See the discussion starting with:
Kaveh> http://www.cygnus.com/ml/egcs-patches/1998-Oct/0481.html )

	As I said in later messages on this subject, this is a change in
policy to which I was not privy.  Based on our conversations from June
1998:

Thu Jun 25 15:47:55 1998  Kaveh R. Ghazi  <ghazi@caip.rutgers.edu>

        * rs6000.c: (small_data_operand): Likewise for parameters `op' and
	`mode' but only when !TARGET_ELF.

and the reasons for making the choice at that time, your recent patch
would have wrong and potentially dangerous.

	Given the new information and policy, your patch was correct and
you obviously did more cross-checking.

Kaveh> I would appreciate it, if you would please undo your Oct 15th
Kaveh> reversion of my original patch.  There was no cause for you to have
Kaveh> done this.

	Done.

David