`nuw nsw` not deduced for `add 1` inbounds of range-restricted length

[scottmcm]

(Context: I tried adding assumes to slice lengths in rust -- rust-lang/rust#122926 -- and was surprised that certain things didn't optimize away that I would have expected to. The example here is basically if i < slice.len() { do_something_with(i + 1); }).)

I tried the following: https://llvm.godbolt.org/z/qM5bnMxxj

define noundef i64 @src(i64 noundef %x.1, i64 noundef %i) unnamed_addr #0 {
start:
  %0 = icmp sge i64 %x.1, 0
  tail call void @llvm.assume(i1 %0)
  %inbounds = icmp ult i64 %i, %x.1
  br i1 %inbounds, label %do_something, label %done

do_something:
  %next = add i64 %i, 1  ;  <-- look here
  br label %done

done:
  %r = phi i64 [ %next, %do_something ], [-1, %start ]
  ret i64 %r
}

expecting to see the add become add nuw nsw because of the range restriction on %x.1, but it doesn't.

Alive proof that it would be ok to do so here: https://alive2.llvm.org/ce/z/3jHXNu

[dtcxzyw]

It has been supported by CVP: https://godbolt.org/z/ezdrzn4Yv.

Did you mean: https://alive2.llvm.org/ce/z/xNoymj

define i32 @src(i32 %x.1, i32 noundef %i) {
  %cond = icmp sgt i32 %x.1, -1
  tail call void @llvm.assume(i1 %cond)
  %inbounds = icmp ult i32 %i, %x.1
  %next = add i32 %i, 1
  %spec.select = select i1 %inbounds, i32 %next, i32 -1
  ret i32 %spec.select
}

define i32 @tgt(i32 %x.1, i32 noundef %i) {
  %cond = icmp sgt i32 %x.1, -1
  tail call void @llvm.assume(i1 %cond)
  %inbounds = icmp ult i32 %i, %x.1
  %next = add nuw nsw i32 %i, 1
  %spec.select = select i1 %inbounds, i32 %next, i32 -1
  ret i32 %spec.select
}

[scottmcm]

Did you mean

I guess so? I don't know enough about phase ordering to say confidently whether CVP can actually handle the scenario I was hitting, just that running the normal O3 passes on the provided source IR doesn't give me an output with add nuw nsw.

[dtcxzyw]

Closed as there is little benefit to use block value in CVP.

diff --git a/llvm/lib/Analysis/LazyValueInfo.cpp b/llvm/lib/Analysis/LazyValueInfo.cpp
index 4c023ed5ed8f..a35d250608ea 100644
--- a/llvm/lib/Analysis/LazyValueInfo.cpp
+++ b/llvm/lib/Analysis/LazyValueInfo.cpp
@@ -1531,7 +1531,7 @@ ValueLatticeElement LazyValueInfoImpl::getValueInBlock(Value *V, BasicBlock *BB,
   LLVM_DEBUG(dbgs() << "LVI Getting block end value " << *V << " at '"
                     << BB->getName() << "'\n");
 
-  assert(BlockValueStack.empty() && BlockValueSet.empty());
+  // assert(BlockValueStack.empty() && BlockValueSet.empty());
   std::optional<ValueLatticeElement> OptResult = getBlockValue(V, BB, CxtI);
   if (!OptResult) {
     solve();
@@ -1601,12 +1601,12 @@ ValueLatticeElement LazyValueInfoImpl::getValueAtUse(const Use &U) {
         break;
       if (CurrU->getOperandNo() == 1)
         CondVal =
-            *getValueFromCondition(V, SI->getCondition(), /*IsTrueDest*/ true,
-                                   /*UseBlockValue*/ false);
+            getValueFromCondition(V, SI->getCondition(), /*IsTrueDest=*/true,
+                                   /*UseBlockValue=*/true);
       else if (CurrU->getOperandNo() == 2)
         CondVal =
-            *getValueFromCondition(V, SI->getCondition(), /*IsTrueDest*/ false,
-                                   /*UseBlockValue*/ false);
+            getValueFromCondition(V, SI->getCondition(), /*IsTrueDest=*/false,
+                                   /*UseBlockValue=*/true);
     } else if (auto *PHI = dyn_cast<PHINode>(CurrI)) {
       // TODO: Use non-local query?
       CondVal = *getEdgeValueLocal(V, PHI->getIncomingBlock(*CurrU),