Re: Fix knfsd readahead cache in 2.4.15

[David S. Miller]

   From: Neil Brown <[e-mail]>
   Date: Tue, 27 Nov 2001 10:35:26 +1100 (EST)
   
   This is definately a bug, but I don't think it is quite as dramatic as
   you suggest.
   
   The "struct raparms" that ra points to will almost always be the last
   one on the list, so ra->p_next will almost always be NULL anyway.
   Nevertheless, it is a bug and your fix looks good.

There are other problems remaining, this function is a logical
mess.

1) depth is computed in the loop, but thrown away.
   It is basically reassigned to a constant before
   being used to index the statistics it is for.

2) raparm_cache is reassigned, and since the ra param list is
   NULL terminated this can make the list shorter and shorter
   and shorter until it is one entry deep.

Yikes :)