[PATCH] D34200: [PM/unswitch] Teach SimpleLoopUnswitch to do non-trivial unswitching, making it no longer even remotely simple.

[Chandler Carruth via Phabricator via llvm-commits]

chandlerc created this revision.
Herald added subscribers: mzolotukhin, mcrosier, mehdi_amini.

The pass will now be more of a "full loop unswitching" pass rather than
anything substantively simpler than any other approach. I plan to rename
it accordingly once the dust settles.

The key ideas of the new loop unswitcher are carried over for
non-trivial unswitching:

1. Fully unswitch a branch or switch instruction from inside of a loop to outside of it.
2. Update the CFG and IR. This avoids needing to "remember" the unswitched branches as well as avoiding excessively cloning and reliance on complex parts of simplify-cfg to cleanup the cfg.
3. Update the analyses rather than just blowing them away or relying on something else updating them.

Sadly, #3 is somewhat compromised here as the dominator tree updates
were too complex for me to want to reason about. I may take another stab
at it, but it may be best to wait for something like the newly proposed
dynamic dominators. However, we do adhere to #3 w.r.t. LoopInfo.

This approach also an some important principles specific to non-trivial
unswitching: not *all* of the loop will be duplicated when unswitching.
This fact allows us to compute the cost in terms of how much *duplicate*
code is inserted rather than just on raw size. Unswitching conditions
which essentialy partition loops will work regardless of how large the
loop is in reality.

Unfortunately, there is a *lot* of code to implement all of this. And
I'm still not really happy with all of it. I think there is more
factoring and cleanup to be done here, but I wanted to at least get it
out where others can see and comment on it.

Some high level outstanding things that I'd like to defer to subsequent
patches:

- We could be much more clever about not cloning things that will be deleted. In fact, we should be able to delete *nothing* and do a minimal number of clones.
- There are many more interesting selection criteria for which branch to unswitch that we might want to look at. One that I'm interested in particularly are a set of conditions which all exit the loop and which can be merged into a single unswitched test of them.

Anyways, even with somewhat rough code, hopefuly folks can start chewing
on this and giving feedback.

Depends on https://reviews.llvm.org/D34049.


https://reviews.llvm.org/D34200

Files:
  include/llvm/Analysis/LoopInfo.h
  include/llvm/Transforms/Scalar/LoopPassManager.h
  include/llvm/Transforms/Scalar/SimpleLoopUnswitch.h
  lib/Transforms/Scalar/SimpleLoopUnswitch.cpp
  test/Transforms/SimpleLoopUnswitch/2006-06-27-DeadSwitchCase.ll
  test/Transforms/SimpleLoopUnswitch/nontrivial-unswitch-cost.ll
  test/Transforms/SimpleLoopUnswitch/nontrivial-unswitch.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D34200.102518.patch
Type: text/x-patch
Size: 167790 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20170614/a0b52d86/attachment-0001.bin>