PCH merge breakage on PA

[Richard Henderson]

On Wed, Jun 05, 2002 at 05:43:12PM -0700, David S. Miller wrote:
>    I know where a lot of the added compile time is coming from, every
>    INSN just got bigger with Jan's recent changes that added the 'B'
>    things to all the INSNs.
> 
> No, in fact TWO new members were added to each and every INSN, not
> just one.

Yes, but we removed two external arrays that mirrored the insn list.

Since not every insn uid is used, there must have been unused holes
in these arrays, so memory usage should have _decreased_ during the
times that these arrays were used.  Which was the bulk of an 
optimizing compile.


r~