diff --git a/cranelift/codegen/src/isa/aarch64/lower.isle b/cranelift/codegen/src/isa/aarch64/lower.isle
index 8c5d8bea81..5b9651947d 100644
--- a/cranelift/codegen/src/isa/aarch64/lower.isle
+++ b/cranelift/codegen/src/isa/aarch64/lower.isle
@@ -1528,6 +1528,11 @@
 (rule (lower (has_type $I64 (bswap x)))
       (a64_rev64 $I64 x))
 
+(rule (lower (has_type $I128 (bswap x)))
+      (value_regs
+       (a64_rev64 $I64 (value_regs_get x 1))
+       (a64_rev64 $I64 (value_regs_get x 0))))
+
 ;;;; Rules for `bmask` ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 
 ;; Bmask tests the value against zero, and uses `csetm` to assert the result.
diff --git a/cranelift/codegen/src/isa/s390x/lower.isle b/cranelift/codegen/src/isa/s390x/lower.isle
index 9e0e43bb2d..7970286059 100644
--- a/cranelift/codegen/src/isa/s390x/lower.isle
+++ b/cranelift/codegen/src/isa/s390x/lower.isle
@@ -1190,15 +1190,8 @@
 
 ;;;; Rules for `bswap` ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 
-(rule (lower (has_type $I16 (bswap x)))
-      (lshr_imm $I32 (bswap_reg $I32 x) 16))
-
-(rule (lower (has_type $I32 (bswap x)))
-      (bswap_reg $I32 x))
-
-(rule (lower (has_type $I64 (bswap x)))
-      (bswap_reg $I64 x))
-
+(rule (lower (has_type ty (bswap x)))
+      (bitrev_bytes ty x))
 
 ;;;; Rules for `clz` ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 
diff --git a/cranelift/codegen/src/isa/x64/lower.isle b/cranelift/codegen/src/isa/x64/lower.isle
index 18d3f01890..2060ea7337 100644
--- a/cranelift/codegen/src/isa/x64/lower.isle
+++ b/cranelift/codegen/src/isa/x64/lower.isle
@@ -2060,6 +2060,11 @@
 (rule (lower (has_type $I64 (bswap src)))
       (x64_bswap $I64 src))
 
+(rule (lower (has_type $I128 (bswap src)))
+      (value_regs
+       (x64_bswap $I64 (value_regs_get_gpr src 1))
+       (x64_bswap $I64 (value_regs_get_gpr src 0))))
+
 ;; Rules for `is_null` ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 
 ;; Null references are represented by the constant value `0`.
diff --git a/cranelift/filetests/filetests/runtests/i128-bswap.clif b/cranelift/filetests/filetests/runtests/i128-bswap.clif
index 657c32d11a..62651992c3 100644
--- a/cranelift/filetests/filetests/runtests/i128-bswap.clif
+++ b/cranelift/filetests/filetests/runtests/i128-bswap.clif
@@ -1,4 +1,9 @@
 test interpret
+test run
+set enable_llvm_abi_extensions
+target x86_64
+target aarch64
+target s390x
 
 function %bswap_i128(i128) -> i128 {
 block0(v0: i128):
@@ -9,4 +14,3 @@ block0(v0: i128):
 ; run: %bswap_i128(1) == 0x01000000_00000000_00000000_00000000
 ; run: %bswap_i128(0x12345678_9ABCDEF0_CAFEF00D_F00DCAFE) == 0xFECA0DF0_0DF0FECA_F0DEBC9A_78563412
 ; run: %bswap_i128(-2) == 0xFEFFFFFF_FFFFFFFF_FFFFFFFF_FFFFFFFF
-
diff --git a/cranelift/fuzzgen/src/function_generator.rs b/cranelift/fuzzgen/src/function_generator.rs
index 6e129501eb..df19663d6b 100644
--- a/cranelift/fuzzgen/src/function_generator.rs
+++ b/cranelift/fuzzgen/src/function_generator.rs
@@ -672,12 +672,10 @@ const OPCODE_SIGNATURES: &'static [(
     (Opcode::Bmask, &[I64], &[I128], insert_opcode),
     (Opcode::Bmask, &[I128], &[I128], insert_opcode),
     // Bswap
-    (Opcode::Bswap, &[I16, I16], &[I16], insert_opcode),
-    (Opcode::Bswap, &[I32, I32], &[I32], insert_opcode),
-    (Opcode::Bswap, &[I64, I64], &[I64], insert_opcode),
-    // I128 version not yet implemented.
-    #[cfg(not(any(target_arch = "x86_64", target_arch = "aarch64")))]
-    (Opcode::Bswap, &[I128, I128], &[I128], insert_opcode),
+    (Opcode::Bswap, &[I16], &[I16], insert_opcode),
+    (Opcode::Bswap, &[I32], &[I32], insert_opcode),
+    (Opcode::Bswap, &[I64], &[I64], insert_opcode),
+    (Opcode::Bswap, &[I128], &[I128], insert_opcode),
     // Fadd
     (Opcode::Fadd, &[F32, F32], &[F32], insert_opcode),
     (Opcode::Fadd, &[F64, F64], &[F64], insert_opcode),