[s390x, abi_impl] Add i128 support (#4598)

This adds full i128 support to the s390x target, including new filetests
and enabling the existing i128 runtest on s390x.

The ABI requires that i128 is passed and returned via implicit pointer,
but the front end still generates direct i128 types in call.  This means
we have to implement ABI support to implicitly convert i128 types to
pointers when passing arguments.

To do so, we add a new variant ABIArg::ImplicitArg.  This acts like
StructArg, except that the value type is the actual target type,
not a pointer type.  The required conversions have to be inserted
in the prologue and at function call sites.

Note that when dereferencing the implicit pointer in the prologue,
we may require a temp register: the pointer may be passed on the
stack so it needs to be loaded first, but the value register may
be in the wrong class for pointer values.  In this case, we use
the "stack limit" register, which should be available at this
point in the prologue.

For return values, we use a mechanism similar to the one used for
supporting multiple return values in the Wasmtime ABI.  The only
difference is that the hidden pointer to the return buffer must
be the *first*, not last, argument in this case.

(This implements the second half of issue #4565.)

cranelift/filetests/filetests/runtests/i128-arithmetic.clif

@@ -2,6 +2,7 @@ test interpret
 test run
 set enable_llvm_abi_extensions=true
 target aarch64
+target s390x
 target x86_64
 
 function %add_i128(i128, i128) -> i128 {

cranelift/filetests/filetests/runtests/i128-bandnot.clif

@@ -1,5 +1,6 @@
 test run
 target aarch64
+target s390x
 
 function %band_not_i128(i128, i128) -> i128 {
 block0(v0: i128, v1: i128):

cranelift/filetests/filetests/runtests/i128-bextend.clif

@@ -1,6 +1,7 @@
 test interpret
 test run
 target aarch64
+target s390x
 
 function %bextend_b1_b128(b1) -> b128 {
 block0(v0: b1):

cranelift/filetests/filetests/runtests/i128-bint.clif

@@ -2,6 +2,7 @@ test interpret
 test run
 set enable_llvm_abi_extensions=true
 target aarch64
+target s390x
 target x86_64
 
 function %bint_b1_i128_true() -> i128 {

cranelift/filetests/filetests/runtests/i128-bitops-count.clif

@@ -1,6 +1,7 @@
 test run
 set enable_llvm_abi_extensions=true
 target aarch64
+target s390x
 target x86_64
 
 function %ctz_i128(i128) -> i128 {

cranelift/filetests/filetests/runtests/i128-bitops.clif

@@ -1,6 +1,7 @@
 test run
 set enable_llvm_abi_extensions=true
 target aarch64
+target s390x
 target x86_64
 
 function %bnot_i128(i128) -> i128 {

cranelift/filetests/filetests/runtests/i128-bmask.clif

@@ -1,6 +1,7 @@
 test interpret
 test run
 target aarch64
+target s390x
 
 function %bmask_b128_i128(b128) -> i128 {
 block0(v0: b128):

cranelift/filetests/filetests/runtests/i128-bornot.clif

@@ -1,5 +1,6 @@
 test run
 target aarch64
+target s390x
 
 function %bor_not_i128(i128, i128) -> i128 {
 block0(v0: i128, v1: i128):

cranelift/filetests/filetests/runtests/i128-br.clif

@@ -1,6 +1,7 @@
 test run
 set enable_llvm_abi_extensions=true
 target aarch64
+target s390x
 target x86_64
 
 

cranelift/filetests/filetests/runtests/i128-bricmp-overflow.clif

@@ -0,0 +1,61 @@
+test run
+target aarch64
+
+; TODO: Merge this with the main i128-bricmp file when s390x supports overflows.
+; See: https://github.com/bytecodealliance/wasmtime/issues/3060
+
+function %i128_bricmp_of(i128, i128) -> b1 {
+block0(v0: i128,v1: i128):
+    br_icmp.i128 of v0, v1, block2
+    jump block1
+
+block1:
+    v2 = bconst.b1 false
+    return v2
+
+block2:
+    v3 = bconst.b1 true
+    return v3
+}
+; run: %i128_bricmp_of(0, 0) == false
+; run: %i128_bricmp_of(0, 1) == false
+; run: %i128_bricmp_of(0, -1) == false
+; run: %i128_bricmp_of(-1, -1) == false
+; run: %i128_bricmp_of(0x80000000_00000000_00000000_00000000, 0) == false
+; run: %i128_bricmp_of(0x7FFFFFFF_FFFFFFFF_FFFFFFFF_FFFFFFFF, 0) == false
+; run: %i128_bricmp_of(1, 0x7FFFFFFF_FFFFFFFF_FFFFFFFF_FFFFFFFF) == false
+; run: %i128_bricmp_of(0x7FFFFFFF_FFFFFFFF_FFFFFFFF_FFFFFFFF, 1) == false
+; run: %i128_bricmp_of(0x80000000_00000000_00000000_00000000, 1) == true
+; run: %i128_bricmp_of(1, 0x80000000_00000000_00000000_00000000) == true
+; run: %i128_bricmp_of(0x7FFFFFFF_FFFFFFFF_FFFFFFFF_FFFFFFFF, 0x80000000_00000000_00000000_00000000) == true
+; run: %i128_bricmp_of(0x80000000_00000000_00000000_00000000, 0x7FFFFFFF_FFFFFFFF_FFFFFFFF_FFFFFFFF) == true
+; run: %i128_bricmp_of(0x4FFFFFFF_FFFFFFFF_FFFFFFFF_FFFFFFFF, 0x30000000_00000000_00000000_00000000) == false
+; run: %i128_bricmp_of(0x4FFFFFFF_FFFFFFFF_FFFFFFFF_FFFFFFFF, 0x30000000_00000000_00000000_00000001) == false
+
+function %i128_bricmp_nof(i128, i128) -> b1 {
+block0(v0: i128,v1: i128):
+    br_icmp.i128 nof v0, v1, block2
+    jump block1
+
+block1:
+    v2 = bconst.b1 false
+    return v2
+
+block2:
+    v3 = bconst.b1 true
+    return v3
+}
+; run: %i128_bricmp_nof(0, 0) == true
+; run: %i128_bricmp_nof(0, 1) == true
+; run: %i128_bricmp_nof(0, -1) == true
+; run: %i128_bricmp_nof(-1, -1) == true
+; run: %i128_bricmp_nof(0x80000000_00000000_00000000_00000000, 0) == true
+; run: %i128_bricmp_nof(0x7FFFFFFF_FFFFFFFF_FFFFFFFF_FFFFFFFF, 0) == true
+; run: %i128_bricmp_nof(1, 0x7FFFFFFF_FFFFFFFF_FFFFFFFF_FFFFFFFF) == true
+; run: %i128_bricmp_nof(0x7FFFFFFF_FFFFFFFF_FFFFFFFF_FFFFFFFF, 1) == true
+; run: %i128_bricmp_nof(0x80000000_00000000_00000000_00000000, 1) == false
+; run: %i128_bricmp_nof(1, 0x80000000_00000000_00000000_00000000) == false
+; run: %i128_bricmp_nof(0x7FFFFFFF_FFFFFFFF_FFFFFFFF_FFFFFFFF, 0x80000000_00000000_00000000_00000000) == false
+; run: %i128_bricmp_nof(0x80000000_00000000_00000000_00000000, 0x7FFFFFFF_FFFFFFFF_FFFFFFFF_FFFFFFFF) == false
+; run: %i128_bricmp_nof(0x4FFFFFFF_FFFFFFFF_FFFFFFFF_FFFFFFFF, 0x30000000_00000000_00000000_00000000) == true
+; run: %i128_bricmp_nof(0x4FFFFFFF_FFFFFFFF_FFFFFFFF_FFFFFFFF, 0x30000000_00000000_00000000_00000001) == true

cranelift/filetests/filetests/runtests/i128-bricmp.clif

@@ -1,5 +1,6 @@
 test run
 target aarch64
+target s390x
 
 function %i128_bricmp_eq(i128, i128) -> b1 {
 block0(v0: i128, v1: i128):
@@ -245,45 +246,3 @@ block2:
     v3 = bconst.b1 true
     return v3
 }
-; run: %i128_bricmp_of(0, 0) == false
-; run: %i128_bricmp_of(0, 1) == false
-; run: %i128_bricmp_of(0, -1) == false
-; run: %i128_bricmp_of(-1, -1) == false
-; run: %i128_bricmp_of(0x80000000_00000000_00000000_00000000, 0) == false
-; run: %i128_bricmp_of(0x7FFFFFFF_FFFFFFFF_FFFFFFFF_FFFFFFFF, 0) == false
-; run: %i128_bricmp_of(1, 0x7FFFFFFF_FFFFFFFF_FFFFFFFF_FFFFFFFF) == false
-; run: %i128_bricmp_of(0x7FFFFFFF_FFFFFFFF_FFFFFFFF_FFFFFFFF, 1) == false
-; run: %i128_bricmp_of(0x80000000_00000000_00000000_00000000, 1) == true
-; run: %i128_bricmp_of(1, 0x80000000_00000000_00000000_00000000) == true
-; run: %i128_bricmp_of(0x7FFFFFFF_FFFFFFFF_FFFFFFFF_FFFFFFFF, 0x80000000_00000000_00000000_00000000) == true
-; run: %i128_bricmp_of(0x80000000_00000000_00000000_00000000, 0x7FFFFFFF_FFFFFFFF_FFFFFFFF_FFFFFFFF) == true
-; run: %i128_bricmp_of(0x4FFFFFFF_FFFFFFFF_FFFFFFFF_FFFFFFFF, 0x30000000_00000000_00000000_00000000) == false
-; run: %i128_bricmp_of(0x4FFFFFFF_FFFFFFFF_FFFFFFFF_FFFFFFFF, 0x30000000_00000000_00000000_00000001) == false
-
-function %i128_bricmp_nof(i128, i128) -> b1 {
-block0(v0: i128,v1: i128):
-    br_icmp.i128 nof v0, v1, block2
-    jump block1
-
-block1:
-    v2 = bconst.b1 false
-    return v2
-
-block2:
-    v3 = bconst.b1 true
-    return v3
-}
-; run: %i128_bricmp_nof(0, 0) == true
-; run: %i128_bricmp_nof(0, 1) == true
-; run: %i128_bricmp_nof(0, -1) == true
-; run: %i128_bricmp_nof(-1, -1) == true
-; run: %i128_bricmp_nof(0x80000000_00000000_00000000_00000000, 0) == true
-; run: %i128_bricmp_nof(0x7FFFFFFF_FFFFFFFF_FFFFFFFF_FFFFFFFF, 0) == true
-; run: %i128_bricmp_nof(1, 0x7FFFFFFF_FFFFFFFF_FFFFFFFF_FFFFFFFF) == true
-; run: %i128_bricmp_nof(0x7FFFFFFF_FFFFFFFF_FFFFFFFF_FFFFFFFF, 1) == true
-; run: %i128_bricmp_nof(0x80000000_00000000_00000000_00000000, 1) == false
-; run: %i128_bricmp_nof(1, 0x80000000_00000000_00000000_00000000) == false
-; run: %i128_bricmp_nof(0x7FFFFFFF_FFFFFFFF_FFFFFFFF_FFFFFFFF, 0x80000000_00000000_00000000_00000000) == false
-; run: %i128_bricmp_nof(0x80000000_00000000_00000000_00000000, 0x7FFFFFFF_FFFFFFFF_FFFFFFFF_FFFFFFFF) == false
-; run: %i128_bricmp_nof(0x4FFFFFFF_FFFFFFFF_FFFFFFFF_FFFFFFFF, 0x30000000_00000000_00000000_00000000) == true
-; run: %i128_bricmp_nof(0x4FFFFFFF_FFFFFFFF_FFFFFFFF_FFFFFFFF, 0x30000000_00000000_00000000_00000001) == true

cranelift/filetests/filetests/runtests/i128-bxornot.clif

@@ -1,5 +1,6 @@
 test run
 target aarch64
+target s390x
 
 function %bxor_not_i128(i128, i128) -> i128 {
 block0(v0: i128, v1: i128):

cranelift/filetests/filetests/runtests/i128-cls.clif

@@ -1,5 +1,6 @@
 test run
 target aarch64
+target s390x
 
 function %cls_i128(i128) -> i128 {
 block0(v0: i128):

cranelift/filetests/filetests/runtests/i128-concat-split.clif

@@ -1,6 +1,7 @@
 test interpret
 test run
 target aarch64
+target s390x
 target x86_64
 
 function %iconcat_isplit(i64, i64) -> i64, i64 {

cranelift/filetests/filetests/runtests/i128-const.clif

@@ -2,7 +2,7 @@ test interpret
 test run
 set enable_llvm_abi_extensions=true
 target aarch64
-; target s390x TODO: Not yet implemented on s390x
+target s390x
 target x86_64
 
 function %i128_const_0() -> i128 {

cranelift/filetests/filetests/runtests/i128-extend.clif

@@ -2,6 +2,7 @@ test interpret
 test run
 set enable_llvm_abi_extensions=true
 target aarch64
+target s390x
 target x86_64
 
 function %i128_uextend_i64(i64) -> i128 {

cranelift/filetests/filetests/runtests/i128-icmp.clif

@@ -2,6 +2,7 @@ test interpret
 test run
 set enable_llvm_abi_extensions=true
 target aarch64
+target s390x
 target x86_64
 
 function %icmp_eq_i128(i128, i128) -> b1 {

cranelift/filetests/filetests/runtests/i128-ireduce.clif

@@ -2,6 +2,7 @@ test interpret
 test run
 set enable_llvm_abi_extensions=true
 target aarch64
+target s390x
 target x86_64
 
 function %ireduce_128_64(i128) -> i64 {

cranelift/filetests/filetests/runtests/i128-load-store.clif

@@ -2,6 +2,7 @@ test run
 set enable_llvm_abi_extensions=true
 target x86_64
 target aarch64
+target s390x
 
 function %i128_stack_store_load(i128) -> b1 {
     ss0 = explicit_slot 16

cranelift/filetests/filetests/runtests/i128-rotate.clif

@@ -1,6 +1,7 @@
 test run
 set enable_llvm_abi_extensions=true
 target aarch64
+target s390x
 target x86_64
 
 function %rotl(i128, i8) -> i128 {

cranelift/filetests/filetests/runtests/i128-select.clif

@@ -1,6 +1,7 @@
 test run
 set enable_llvm_abi_extensions=true
 target aarch64
+target s390x
 target x86_64
 
 function %i128_select(b1, i128, i128) -> i128 {

cranelift/filetests/filetests/runtests/i128-shifts-small-types.clif

@@ -1,5 +1,6 @@
 test run
 target aarch64
+target s390x
 
 ; TODO: Merge this with the main i128-shifts file when x86_64 passes these.
 

cranelift/filetests/filetests/runtests/i128-shifts.clif

@@ -1,6 +1,7 @@
 test run
 set enable_llvm_abi_extensions=true
 target aarch64
+target s390x
 target x86_64