Commit Graph

11 Commits

Author SHA1 Message Date
Piotr Padlewski 9530883e8c [Devirtualization] MemDep returns non-local !invariant.group dependencies
Summary:
Memory Dependence Analysis was limited to return only local dependencies
for invariant.group handling. Now it returns NonLocal when it finds it
and then by asking getNonLocalPointerDependency we get found dep.

Thanks to this we are able to devirtualize loops!

    void indirect(A &a, int n) {
      for (int i = 0 ; i < n; i++)
        a.foo();

    }
    void test(int n) {
      A a;
      indirect(a);
    }

After inlining a.foo() will be changed to direct call, even if foo and A::A()
is external (but only if vtable definition is be available).

Reviewers: nlewycky, dberlin, chandlerc, rsmith

Subscribers: mehdi_amini, davide, llvm-commits

Differential Revision: https://reviews.llvm.org/D28137

llvm-svn: 291762
2017-01-12 11:33:58 +00:00
David L Kreitzer 4d7257dfa1 Fix for two constant propagation problems in GVN with the assume intrinsic
instruction.

Patch by Yuanrui Zhang.

Differential Revision: http://reviews.llvm.org/D16100

llvm-svn: 258435
2016-01-21 21:32:35 +00:00
Piotr Padlewski dc9b2cfc50 inariant.group handling in GVN
The most important part required to make clang
devirtualization works ( ͡°͜ʖ ͡°).
The code is able to find non local dependencies, but unfortunatelly
because the caller can only handle local dependencies, I had to add
some restrictions to look for dependencies only in the same BB.

http://reviews.llvm.org/D12992

llvm-svn: 249196
2015-10-02 22:12:22 +00:00
Piotr Padlewski 0c7d8fc1f6 assuem(X) handling in GVN bugfix
There was infinite loop because it was trying to change assume(true) into
assume(true)
Also added handling when assume(false) appear

http://reviews.llvm.org/D12516

llvm-svn: 246697
2015-09-02 20:00:03 +00:00
Piotr Padlewski 28ffcbe1cc Constant propagation after hitting assume(cmp) bugfix
Last time code run into assertion `BBE.isSingleEdge()` in
lib/IR/Dominators.cpp:200.

http://reviews.llvm.org/D12170

llvm-svn: 246696
2015-09-02 19:59:59 +00:00
Piotr Padlewski 14e815c22b Constant propagation after hiting llvm.assume
After hitting @llvm.assume(X) we can:
- propagate equality that X == true
- if X is icmp/fcmp (with eq operation), and one of operand
  is constant we can change all variables with constants in the same BasicBlock

http://reviews.llvm.org/D11918

llvm-svn: 246695
2015-09-02 19:59:53 +00:00
Steven Wu 61db34d12e Revert r246244 and r246243
These two commits cause clang/llvm bootstrap to hang.

llvm-svn: 246279
2015-08-28 06:52:00 +00:00
Piotr Padlewski 3f81ec1e38 Constant propagation after hitting assume(cmp) bugfix
Last time code run into assertion `BBE.isSingleEdge()` in
lib/IR/Dominators.cpp:200.

http://reviews.llvm.org/D12170

llvm-svn: 246244
2015-08-28 01:02:00 +00:00
Piotr Padlewski 63cc5d4627 Constant propagation after hiting llvm.assume
After hitting @llvm.assume(X) we can:
- propagate equality that X == true
- if X is icmp/fcmp (with eq operation), and one of operand
  is constant we can change all variables with constants in the same BasicBlock

http://reviews.llvm.org/D11918

llvm-svn: 246243
2015-08-28 01:01:57 +00:00
Justin Bogner 9f00ebaeda Revert "Constant propagation after hiting llvm.assume"
This was also failing bootstrap:

http://lab.llvm.org:8080/green/job/clang-stage2-configure-Rlto_build

This reverts r245265.

llvm-svn: 245269
2015-08-18 07:00:34 +00:00
Piotr Padlewski 94ca3783b8 Constant propagation after hiting llvm.assume
After hitting @llvm.assume(X) we can:
- propagate equality that X == true
- if X is icmp/fcmp (with eq operation), and one of operand
  is constant we can change all variables with constants in the same BasicBlock

http://reviews.llvm.org/D11918

llvm-svn: 245265
2015-08-18 03:55:30 +00:00