Expand GEPs in ScalarEvolution expressions. SCEV expressions can now

have pointer types, though in contrast to C pointer types, SCEV
addition is never implicitly scaled. This not only eliminates the
need for special code like IndVars' EliminatePointerRecurrence
and LSR's own GEP expansion code, it also does a better job because
it lets the normal optimizations handle pointer expressions just
like integer expressions.

Also, since LLVM IR GEPs can't directly index into multi-dimensional
VLAs, moving the GEP analysis out of client code and into the SCEV
framework makes it easier for clients to handle multi-dimensional
VLAs the same way as other arrays.

Some existing regression tests show improved optimization.
test/CodeGen/ARM/2007-03-13-InstrSched.ll in particular improved to
the point where if-conversion started kicking in; I turned it off
for this test to preserve the intent of the test.

llvm-svn: 69258

1 files changed, 1 insertions, 1 deletions

diff --git a/llvm/test/CodeGen/ARM/2007-03-13-InstrSched.ll b/llvm/test/CodeGen/ARM/2007-03-13-InstrSched.ll
index 8fdff52f015..1b917f0ac2b 100644
--- a/llvm/test/CodeGen/ARM/2007-03-13-InstrSched.ll
+++ b/llvm/test/CodeGen/ARM/2007-03-13-InstrSched.ll
@@ -1,5 +1,5 @@
 ; RUN: llvm-as < %s | llc -mtriple=arm-apple-darwin -relocation-model=pic \
-; RUN:   -mattr=+v6 -stats |& grep asm-printer | grep 41
+; RUN:   -mattr=+v6 -ifcvt-limit=0 -stats |& grep asm-printer | grep 35
 
 define void @test(i32 %tmp56222, i32 %tmp36224, i32 %tmp46223, i32 %i.0196.0.ph, i32 %tmp8, i32* %tmp1011, i32** %tmp1, i32* %d2.1.out, i32* %d3.1.out, i32* %d0.1.out, i32* %d1.1.out) {
 newFuncRoot: