An optimization error about loop unrolling


Bugzilla Link	PR42382
Status	NEW
Importance	P enhancement
Reported by	witstorm (witstorm@163.com)
Reported on	2019-06-25 04:18:49 -0700
Last modified on	2019-06-25 05:33:15 -0700
Version	7.0
Hardware	PC Linux
CC	htmldeveloper@gmail.com, llvm-bugs@lists.llvm.org, t.p.northover@gmail.com
Fixed by commit(s)
Attachments
Blocks
Blocked by
See also

In the experiment, when clang uses O2 and O3 optimization options, the pass of
Loop unrolling will cause an result error.The procedure is as follows：
typedef int A __attribute__((vector _size(64),aligned(64)));
A Aa,Ab,Ai;
static __inline A __attribute__((__always_inline__))
SET_A (int __A,int __B,int __C,int __D,int __E,int __F,int __G,int __H,int
__I,int __J,int __K,int __L,int __M,int __N,int __O,int __P){
  union{
   int __a[16] __attribute__((aligned(64)));
   A __v;
}__u;
  __u.__a[0]=__A;
  __u.__a[1]=__B;
  __u.__a[2]=__C;
  __u.__a[3]=__D;
  __u.__a[4]=__E;
  __u.__a[5]=__F;
  __u.__a[6]=__G;
  __u.__a[7]=__H;
  __u.__a[8]=__I;
  __u.__a[9]=__J;
  __u.__a[10]=__K;
  __u.__a[11]=__L;
  __u.__a[12]=__M;
  __u.__a[13]=__N;
  __u.__a[14]=__O;
  __u.__a[15]=__P;
return __u.__v;
}
int main(){
   int r[16]={2047,2047,2047,2047,2047,2047,2047,2047,2047,2047,2047,2047,2047,2047,2047,2047};
   int i;
   for(i=1024;i>=1;i>>=1){
   Ai=SET_A(i,i,i,i,i,i,i,i,i,i,i,i,i,i,i,i);
   Aa^=Aa>>i;
   Ab+=Aa+Ai;
}
}
You can reproduce the error with the following command：
clang -arch=x86 -static -O2 -emit-llvm -S a.c
Loop unrolling optimization produces incorrect values in IR.

Quuxplusone / LLVMBugzillaTest

An optimization error about loop unrolling #41352