forked from OSchip/llvm-project
New testcase to see if -indvars can promote pointer arithmetic to array accesses
llvm-svn: 10577
This commit is contained in:
parent
2d69eae755
commit
a5e3418d98
|
@ -0,0 +1,14 @@
|
|||
; RUN: llvm-as < %s | opt -indvars | llvm-dis | grep indvar
|
||||
|
||||
%G = global int* null
|
||||
|
||||
%Array = external global [40 x int]
|
||||
|
||||
void %test() {
|
||||
br label %Loop
|
||||
Loop:
|
||||
%X = phi int* [getelementptr ([40 x int]* %Array, long 0, long 0), %0], [%X.next, %Loop]
|
||||
%X.next = getelementptr int* %X, long 1
|
||||
store int* %X, int** %G
|
||||
br label %Loop
|
||||
}
|
Loading…
Reference in New Issue