From 156bda8bc708cca60e9de18743d833c8d97dd7ff Mon Sep 17 00:00:00 2001 From: Afonso Bordado Date: Tue, 23 Aug 2022 11:42:24 +0100 Subject: [PATCH] Avoid masking shift amounts (#1268) Cranelift 0.87 now follows its own documentation regarding shift amounts, and implicitly masks them if the arch requires it. [0] [0]: https://github.com/bytecodealliance/wasmtime/commit/05089321740a07757dff0a285176b2651a49aae2 --- src/num.rs | 25 ++++++------------------- 1 file changed, 6 insertions(+), 19 deletions(-) diff --git a/src/num.rs b/src/num.rs index 4ce8adb182e..679974710df 100644 --- a/src/num.rs +++ b/src/num.rs @@ -150,18 +150,12 @@ pub(crate) fn codegen_int_binop<'tcx>( BinOp::BitXor => b.bxor(lhs, rhs), BinOp::BitAnd => b.band(lhs, rhs), BinOp::BitOr => b.bor(lhs, rhs), - BinOp::Shl => { - let lhs_ty = fx.bcx.func.dfg.value_type(lhs); - let actual_shift = fx.bcx.ins().band_imm(rhs, i64::from(lhs_ty.bits() - 1)); - fx.bcx.ins().ishl(lhs, actual_shift) - } + BinOp::Shl => b.ishl(lhs, rhs), BinOp::Shr => { - let lhs_ty = fx.bcx.func.dfg.value_type(lhs); - let actual_shift = fx.bcx.ins().band_imm(rhs, i64::from(lhs_ty.bits() - 1)); if signed { - fx.bcx.ins().sshr(lhs, actual_shift) + b.sshr(lhs, rhs) } else { - fx.bcx.ins().ushr(lhs, actual_shift) + b.ushr(lhs, rhs) } } // Compare binops handles by `codegen_binop`. @@ -279,22 +273,15 @@ pub(crate) fn codegen_checked_int_binop<'tcx>( } } BinOp::Shl => { - let lhs_ty = fx.bcx.func.dfg.value_type(lhs); - let masked_shift = fx.bcx.ins().band_imm(rhs, i64::from(lhs_ty.bits() - 1)); - let val = fx.bcx.ins().ishl(lhs, masked_shift); + let val = fx.bcx.ins().ishl(lhs, rhs); let ty = fx.bcx.func.dfg.value_type(val); let max_shift = i64::from(ty.bits()) - 1; let has_overflow = fx.bcx.ins().icmp_imm(IntCC::UnsignedGreaterThan, rhs, max_shift); (val, has_overflow) } BinOp::Shr => { - let lhs_ty = fx.bcx.func.dfg.value_type(lhs); - let masked_shift = fx.bcx.ins().band_imm(rhs, i64::from(lhs_ty.bits() - 1)); - let val = if !signed { - fx.bcx.ins().ushr(lhs, masked_shift) - } else { - fx.bcx.ins().sshr(lhs, masked_shift) - }; + let val = + if !signed { fx.bcx.ins().ushr(lhs, rhs) } else { fx.bcx.ins().sshr(lhs, rhs) }; let ty = fx.bcx.func.dfg.value_type(val); let max_shift = i64::from(ty.bits()) - 1; let has_overflow = fx.bcx.ins().icmp_imm(IntCC::UnsignedGreaterThan, rhs, max_shift);