[llvm-branch-commits] [llvm] [SDAG] Introduce inbounds flag for pointer arithmetic (PR #131862)

[Fabian Ritter via llvm-branch-commits]

ritter-x2a wrote:

> Maybe we could consider adding "ISD::PTRADD"? Lowers to ISD::ADD by default, but targets that want to do weird things with pointer arithmetic could do them.

 That would be helpful. We'd still need an inbounds flag for ISD::PTRADD, but it would certainly be easier to make use of. I'll look into that.

> One other concern, which applies to basically any formulation of this: Since SelectionDAG doesn't have a distinct pointer type, you can't tell whether the pointer operand was produced by an inttoptr. So in some cases, you have an operation marked "inbounds", but it's ambiguous which object it's actually inbounds to. This isn't really a problem at the moment because we do IR-level transforms that remove inttoptr anyway, but if we ever do resolve the IR-level issues, we should have some idea for how we propagate the fix to SelectionDAG.

I can see that's a problem if you'd want to infer that an operation is inbounds, or if you'd want to prove the absence (or presence) of poison/UB. But how is that a problem for generating code? If there is an inbounds flag on a (hypothetical) ISD::PTRADD, we can assume that the operation is inbounds with respect to whatever the address operand is pointing to, no matter if it's the result of integer operations, right?



https://github.com/llvm/llvm-project/pull/131862