[llvm-dev] Field sensitive alias analysis?

[Dmitry Polukhin via llvm-dev]

As far as I can see it is specifics of arrays inside structs. Current TBAA
does distinguish non-array members with path sensitive TBAA (see !1 and !6
in my example below, TBAA has reference to struct !2 and offset). As for
arrays information that it was member of some struct get lost completely
(!7 has nothing about struct !2).

struct S {
  int a;
  int b;
  int c[3];
};

void foo(struct S* p) {
  p->a = 1;
  p->b = 2;
  p->c[0] = 3;
  p->c[1] = 4;
}

define void @foo(%struct.S* nocapture %p) #0 {
entry:
  %a = getelementptr inbounds %struct.S, %struct.S* %p, i64 0, i32 0
  store i32 1, i32* %a, align 4, !tbaa !1
  %b = getelementptr inbounds %struct.S, %struct.S* %p, i64 0, i32 1
  store i32 2, i32* %b, align 4, !tbaa !6
  %arrayidx = getelementptr inbounds %struct.S, %struct.S* %p, i64 0, i32
2, i64 0
  store i32 3, i32* %arrayidx, align 4, !tbaa !7
  %arrayidx2 = getelementptr inbounds %struct.S, %struct.S* %p, i64 0, i32
2, i64 1
  store i32 4, i32* %arrayidx2, align 4, !tbaa !7
  ret void
}

!0 = !{!"clang version 3.8.0 "}
!1 = !{!2, !3, i64 0}
!2 = !{!"S", !3, i64 0, !3, i64 4, !4, i64 8}
!3 = !{!"int", !4, i64 0}
!4 = !{!"omnipotent char", !5, i64 0}
!5 = !{!"Simple C/C++ TBAA"}
!6 = !{!2, !3, i64 4}
!7 = !{!3, !3, i64 0}

I'm just start learning how TBAA in clang works so I don't know why it was
implemented this way.

On Fri, Dec 4, 2015 at 11:06 AM, Vaivaswatha Nagaraj via llvm-dev <
llvm-dev at lists.llvm.org> wrote:

> Hi,
>
> I'm trying to optimize a simple C code and came across a situation where
> invariant code is not being moved out:
>
> On an -O3 compilation, I noticed that the "load" for the loop bounds
> (which remain invariant throughout) happens on each iteration of both the
> loops, even though it is not modified anywhere in the function "bigLoop".
> It seems that alias analysis is not able to say that the writes to one
> field in the structure does not impact the other field, leading to LICM
> being ineffective.
>
> Do any of the alias analyses currently have some kind of field sensitivity
> that can help in this case?
>
> ------------------------- test case ------------------------------------
> #include <stdlib.h>
> #include <stdio.h>
>
> #define SIZE 100
>
> struct AS {
>   int a[SIZE+4];
>   int size;
> } A;
>
> void bigLoop(void)
> {
>   unsigned i, j;
>
>   for (i = 0; i < A.size; i++) {
>     A.a[i+2] +=  A.a[i];
>   }
>   for (i = 0; i < A.size; i++) {
>     A.a[i+2] *=  A.a[i];
>   }
> }
>
> int main()
> {
>   A.size = random()%SIZE;
>   for (unsigned i = 0; i < A.size; i++) {
>     A.a[i] = random()%23;
>   }
>   bigLoop();
>   return 0;
> }
>
> Thanks,
>
>   - Vaivaswatha
>
> _______________________________________________
> LLVM Developers mailing list
> llvm-dev at lists.llvm.org
> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20151204/2bfc6c0a/attachment.html>