[ValueTracking][RISCV] Use ConstantRange::getUnsignedMax instead of getUpper to simplify some code.

[topperc]

This avoids the need to subtract 1 and explain why.

[llvmbot]

@llvm/pr-subscribers-llvm-analysis

Author: Craig Topper (topperc)

Changes

This avoids the need to subtract 1 and explain why.

---

Full diff: https://github.com/llvm/llvm-project/pull/96816.diff

1 Files Affected:

• (modified) llvm/lib/Analysis/ValueTracking.cpp (+1-3)

diff --git a/llvm/lib/Analysis/ValueTracking.cpp b/llvm/lib/Analysis/ValueTracking.cpp
index 777b2792a1ed2..1dfc6cfac4551 100644
--- a/llvm/lib/Analysis/ValueTracking.cpp
+++ b/llvm/lib/Analysis/ValueTracking.cpp
@@ -1740,10 +1740,8 @@ static void computeKnownBitsFromOperator(const Operator *I,
             cast<ConstantInt>(II->getArgOperand(HasAVL))->getZExtValue());
         RISCVII::VLMUL VLMUL = static_cast<RISCVII::VLMUL>(
             cast<ConstantInt>(II->getArgOperand(1 + HasAVL))->getZExtValue());
-        // The Range is [Lower, Upper), so we need to subtract 1 here to get the
-        // real upper value.
         uint64_t MaxVLEN =
-            (Range.getUpper().getZExtValue() - 1) * RISCV::RVVBitsPerBlock;
+            Range.getUnsignedMax().getZExtValue() * RISCV::RVVBitsPerBlock;
         uint64_t MaxVL = MaxVLEN / RISCVVType::getSEWLMULRatio(SEW, VLMUL);
 
         // Result of vsetvli must be not larger than AVL.