[PATCH] D21921: [proof of concept] Port old PM CGSCC visitation logic to new PM

[Sean Silva via llvm-commits]

silvas added a comment.

With this experience of doing this proof of concept, I can now say with confidence that porting the old CGSCC visitation order interferes minimally with what Chandler is doing in http://reviews.llvm.org/D21464
(I know this because every change that he will have to undo will be a change that I had to do in preparing that patch; I can say that in fact they were a very small part of preparing this patch.
In fact, there is no reason why Chandler's new logic can't live in parallel with this kind of port of the old logic; the hardest part of that is likely to just be deciding on the name for distinguishing the two.
)
There is thus little downside to moving forward with porting the old PM visitation logic to the new PM.

I would like to establish consensus for moving forward with porting the old PM visitation logic to the new PM (or discovering a good reason why that would be a bad idea!). The huge advantage of doing this is that it is as NFC as possible.


Repository:
  rL LLVM

http://reviews.llvm.org/D21921