diff --git a/cranelift/codegen/src/isa/aarch64/abi.rs b/cranelift/codegen/src/isa/aarch64/abi.rs
index acd8cc88e5..287b570efd 100644
--- a/cranelift/codegen/src/isa/aarch64/abi.rs
+++ b/cranelift/codegen/src/isa/aarch64/abi.rs
@@ -152,6 +152,11 @@ impl ABIMachineSpec for AArch64MachineDeps {
         64
     }
 
+    /// Return required stack alignment in bytes.
+    fn stack_align(_call_conv: isa::CallConv) -> u32 {
+        16
+    }
+
     fn compute_arg_locs(
         call_conv: isa::CallConv,
         params: &[ir::AbiParam],
diff --git a/cranelift/codegen/src/isa/arm32/abi.rs b/cranelift/codegen/src/isa/arm32/abi.rs
index 4611899af6..8de86ca861 100644
--- a/cranelift/codegen/src/isa/arm32/abi.rs
+++ b/cranelift/codegen/src/isa/arm32/abi.rs
@@ -45,6 +45,11 @@ impl ABIMachineSpec for Arm32MachineDeps {
         32
     }
 
+    /// Return required stack alignment in bytes.
+    fn stack_align(_call_conv: isa::CallConv) -> u32 {
+        8
+    }
+
     fn compute_arg_locs(
         _call_conv: isa::CallConv,
         params: &[ir::AbiParam],
diff --git a/cranelift/codegen/src/isa/x64/abi.rs b/cranelift/codegen/src/isa/x64/abi.rs
index 25d1f58762..48f5d92a77 100644
--- a/cranelift/codegen/src/isa/x64/abi.rs
+++ b/cranelift/codegen/src/isa/x64/abi.rs
@@ -90,6 +90,11 @@ impl ABIMachineSpec for X64ABIMachineSpec {
         64
     }
 
+    /// Return required stack alignment in bytes.
+    fn stack_align(_call_conv: isa::CallConv) -> u32 {
+        16
+    }
+
     fn compute_arg_locs(
         call_conv: isa::CallConv,
         params: &[ir::AbiParam],
diff --git a/cranelift/codegen/src/machinst/abi_impl.rs b/cranelift/codegen/src/machinst/abi_impl.rs
index 2fe61d78d1..70ffaa753f 100644
--- a/cranelift/codegen/src/machinst/abi_impl.rs
+++ b/cranelift/codegen/src/machinst/abi_impl.rs
@@ -216,6 +216,9 @@ pub trait ABIMachineSpec {
         }
     }
 
+    /// Returns required stack alignment in bytes.
+    fn stack_align(call_conv: isa::CallConv) -> u32;
+
     /// Process a list of parameters or return values and allocate them to registers
     /// and stack slots.
     ///
@@ -942,7 +945,7 @@ impl<M: ABIMachineSpec> ABICallee for ABICalleeImpl<M> {
             );
             total_stacksize += self.flags.baldrdash_prologue_words() as u32 * bytes;
         }
-        let mask = 2 * bytes - 1;
+        let mask = M::stack_align(self.call_conv) - 1;
         let total_stacksize = (total_stacksize + mask) & !mask; // 16-align the stack.
 
         if !self.call_conv.extends_baldrdash() {