s390x: Enable more runtests, and fix a few bugs (#4516)

This enables more runtests to be executed on s390x.  Doing so
uncovered a two back-end bugs, which are fixed as well:

- The result of cls was always off by one.
- The result of popcnt.i16 has uninitialized high bits.

In addition, I found a bug in the load-op-store.clif test case:
     v3 = heap_addr.i64 heap0, v1, 4
     v4 = iconst.i64 42
     store.i32 v4, v3
This was clearly intended to perform a 32-bit store, but
actually performs a 64-bit store (it seems the type annotation
of the store opcode is ignored, and the type of the operand
is used instead).  That bug did not show any noticable symptoms
on little-endian architectures, but broke on big-endian.

cranelift/filetests/filetests/isa/s390x/bitops.clif

@@ -93,7 +93,7 @@ block0(v0: i64):
 ;   srag %r5, %r2, 63
 ;   xgrk %r3, %r2, %r5
 ;   flogr %r0, %r3
-;   lgr %r2, %r0
+;   aghik %r2, %r0, -1
 ;   br %r14
 
 function %cls_i32(i32) -> i32 {
@@ -107,7 +107,7 @@ block0(v0: i32):
 ;   srag %r3, %r5, 63
 ;   xgr %r5, %r3
 ;   flogr %r0, %r5
-;   ahik %r2, %r0, -32
+;   ahik %r2, %r0, -33
 ;   br %r14
 
 function %cls_i16(i16) -> i16 {
@@ -121,7 +121,7 @@ block0(v0: i16):
 ;   srag %r3, %r5, 63
 ;   xgr %r5, %r3
 ;   flogr %r0, %r5
-;   ahik %r2, %r0, -48
+;   ahik %r2, %r0, -49
 ;   br %r14
 
 function %cls_i8(i8) -> i8 {
@@ -135,7 +135,7 @@ block0(v0: i8):
 ;   srag %r3, %r5, 63
 ;   xgr %r5, %r3
 ;   flogr %r0, %r5
-;   ahik %r2, %r0, -56
+;   ahik %r2, %r0, -57
 ;   br %r14
 
 function %ctz_i64(i64) -> i64 {
@@ -238,9 +238,9 @@ block0(v0: i16):
 
 ; block0:
 ;   popcnt %r5, %r2
-;   sllk %r3, %r5, 8
-;   ar %r5, %r3
-;   srlk %r2, %r5, 8
+;   srlk %r3, %r5, 8
+;   ark %r2, %r5, %r3
+;   nill %r2, 255
 ;   br %r14
 
 function %popcnt_i8(i8) -> i8 {

cranelift/filetests/filetests/runtests/bint.clif

@@ -1,6 +1,7 @@
 test interpret
 test run
 target aarch64
+target s390x
 target x86_64
 
 function %bint_b1_i8_true() -> i8 {

cranelift/filetests/filetests/runtests/cls.clif

@@ -1,6 +1,7 @@
 test interpret
 test run
 target aarch64
+target s390x
 ; not implemented on `x86_64`
 
 function %cls_i8(i8) -> i8 {

cranelift/filetests/filetests/runtests/clz.clif

@@ -1,6 +1,7 @@
 test interpret
 test run
 target aarch64
+target s390x
 target x86_64
 target x86_64 has_lzcnt
 

cranelift/filetests/filetests/runtests/conversions.clif

@@ -1,6 +1,7 @@
 test run
 
 target x86_64
+target s390x
 target aarch64
 
 function %fpromote_f32_f64(i64 vmctx, i64, f32) -> f64 {

cranelift/filetests/filetests/runtests/ctz.clif

@@ -1,6 +1,7 @@
 test interpret
 test run
 target aarch64
+target s390x
 target x86_64
 target x86_64 has_bmi1
 

cranelift/filetests/filetests/runtests/iabs.clif

@@ -1,6 +1,7 @@
 test interpret
 test run
 target aarch64
+target s390x
 ; x86_64 only supports vector iabs
 
 function %iabs_i8(i8) -> i8 {
@@ -41,4 +42,4 @@ block0(v0: i64):
 ; run: %iabs_i64(0) == 0
 ; run: %iabs_i64(9223372036854775807) == 9223372036854775807
 ; run: %iabs_i64(-9223372036854775807) == 9223372036854775807
-; run: %iabs_i64(-9223372036854775808) == -9223372036854775808
+; run: %iabs_i64(-9223372036854775808) == -9223372036854775808

cranelift/filetests/filetests/runtests/icmp-ugt.clif

@@ -1,6 +1,7 @@
 test interpret
 test run
 target aarch64
+target s390x
 target x86_64
 
 function %icmp_ugt_i8(i8, i8) -> b1 {

cranelift/filetests/filetests/runtests/ireduce.clif

@@ -1,6 +1,7 @@
 test interpret
 test run
 target aarch64
+target s390x
 target x86_64
 
 function %ireduce_i16_i8(i16) -> i8 {

cranelift/filetests/filetests/runtests/load-op-store.clif

@@ -1,5 +1,6 @@
 test run
 target x86_64
+target s390x
 target aarch64
 
 function %load_op_store_iadd_i64(i64 vmctx, i64, i64) -> i64 {
@@ -28,7 +29,7 @@ function %load_op_store_iadd_i32(i64 vmctx, i64, i32) -> i32 {
 
 block0(v0: i64, v1: i64, v2: i32):
     v3 = heap_addr.i64 heap0, v1, 4
-    v4 = iconst.i64 42
+    v4 = iconst.i32 42
     store.i32 v4, v3
     v5 = load.i32 v3
     v6 = iadd.i32 v5, v2

cranelift/filetests/filetests/runtests/popcnt.clif

@@ -1,6 +1,7 @@
 test interpret
 test run
 target aarch64
+target s390x
 target x86_64
 target x86_64 has_popcnt
 

cranelift/filetests/filetests/runtests/select.clif

@@ -1,5 +1,6 @@
 test interpret
 test run
+target s390x
 target x86_64
 
 function %select_eq_f32(f32, f32) -> i32 {

cranelift/filetests/filetests/runtests/simd-scalartovector-aarch64.clif

@@ -1,5 +1,6 @@
 test run
 target aarch64
+target s390x
 ; i8 and i16 are invalid source sizes for x86_64
 
 function %scalartovector_i8(i8) -> i8x16 {
@@ -16,4 +17,4 @@ block0(v0: i16):
     return v1
 }
 ; run: %scalartovector_i16(1) == [1 0 0 0 0 0 0 0]
-; run: %scalartovector_i16(65535) == [65535 0 0 0 0 0 0 0]
+; run: %scalartovector_i16(65535) == [65535 0 0 0 0 0 0 0]

cranelift/filetests/filetests/runtests/simd-scalartovector.clif

@@ -1,5 +1,6 @@
 test run
 target aarch64
+target s390x
 set enable_simd
 target x86_64 has_sse3 has_ssse3 has_sse41
 

cranelift/filetests/filetests/runtests/smulhi-aarch64.clif

@@ -1,6 +1,7 @@
 test interpret
 test run
 target aarch64
+target s390x
 ; x86_64 backend only supports `i16`, `i32`, and `i64` types.
 
 function %smulhi_i8(i8, i8) -> i8 {
@@ -10,4 +11,4 @@ block0(v0: i8, v1: i8):
 }
 ; run: %smulhi_i8(-2, -4) == 0
 ; run: %smulhi_i8(2, -4) == -1
-; run: %smulhi_i8(127, 127) == 63
+; run: %smulhi_i8(127, 127) == 63

cranelift/filetests/filetests/runtests/smulhi.clif

@@ -1,6 +1,7 @@
 test interpret
 test run
 target aarch64
+target s390x
 set enable_simd
 target x86_64 has_sse3 has_ssse3 has_sse41