Bill Wendling f7da4e939f Fix for PR929. The PHI nodes were being gone through for each instruction
in a successor block for every block...resulting in some O(N^k) algorithm
which wasn't very good for performance. Calculating this information up
front and keeping it in a map made it much faster.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30697 91177308-0d34-0410-b5e6-96231b3b80d8
2006-10-03 07:20:20 +00:00
..
2006-09-01 06:08:16 +00:00
2006-08-22 16:06:27 +00:00
2006-06-05 16:29:06 +00:00
2006-06-05 16:29:06 +00:00
2005-04-21 20:19:05 +00:00
2006-09-28 00:37:43 +00:00
2006-07-26 16:18:00 +00:00
2006-07-26 16:18:00 +00:00
2006-09-08 06:43:00 +00:00
2006-07-26 16:55:39 +00:00
2006-08-28 00:09:00 +00:00
2006-06-05 16:29:06 +00:00