Updates

Author: JonPsson1

compiler-rt/lib/builtins/trunctfhf2.c

@@ -10,7 +10,7 @@
 #define QUAD_PRECISION
 #include "fp_lib.h"
 
-#if defined(CRT_HAS_TF_MODE)
+#if defined(CRT_HAS_TF_MODE) && defined(COMPILER_RT_HAS_FLOAT16)
 #define SRC_QUAD
 #define DST_HALF
 #include "fp_trunc_impl.inc"

llvm/test/CodeGen/SystemZ/atomicrmw-fadd-04.ll

@@ -0,0 +1,76 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
+; Test atomic half addition. Expect a compare-and-swap loop.
+;
+; RUN: llc < %s -mtriple=s390x-linux-gnu | FileCheck %s
+
+define half @f1(ptr %src, half %b) {
+; CHECK-LABEL: f1:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    stmg %r11, %r15, 88(%r15)
+; CHECK-NEXT:    .cfi_offset %r11, -72
+; CHECK-NEXT:    .cfi_offset %r12, -64
+; CHECK-NEXT:    .cfi_offset %r13, -56
+; CHECK-NEXT:    .cfi_offset %r14, -48
+; CHECK-NEXT:    .cfi_offset %r15, -40
+; CHECK-NEXT:    aghi %r15, -184
+; CHECK-NEXT:    .cfi_def_cfa_offset 344
+; CHECK-NEXT:    std %f8, 176(%r15) # 8-byte Folded Spill
+; CHECK-NEXT:    std %f9, 168(%r15) # 8-byte Folded Spill
+; CHECK-NEXT:    std %f10, 160(%r15) # 8-byte Folded Spill
+; CHECK-NEXT:    .cfi_offset %f8, -168
+; CHECK-NEXT:    .cfi_offset %f9, -176
+; CHECK-NEXT:    .cfi_offset %f10, -184
+; CHECK-NEXT:    lgr %r13, %r2
+; CHECK-NEXT:    lgh %r0, 0(%r2)
+; CHECK-NEXT:    ler %f8, %f0
+; CHECK-NEXT:    sllg %r0, %r0, 48
+; CHECK-NEXT:    ldgr %f9, %r0
+; CHECK-NEXT:    risbg %r12, %r2, 0, 189, 0
+; CHECK-NEXT:    sll %r13, 3
+; CHECK-NEXT:    lcr %r11, %r13
+; CHECK-NEXT:    j .LBB0_2
+; CHECK-NEXT:  .LBB0_1: # %atomicrmw.start
+; CHECK-NEXT:    # in Loop: Header=BB0_2 Depth=1
+; CHECK-NEXT:    sllg %r0, %r3, 48
+; CHECK-NEXT:    ldgr %f9, %r0
+; CHECK-NEXT:    je .LBB0_5
+; CHECK-NEXT:  .LBB0_2: # %atomicrmw.start
+; CHECK-NEXT:    # =>This Loop Header: Depth=1
+; CHECK-NEXT:    # Child Loop BB0_3 Depth 2
+; CHECK-NEXT:    ler %f0, %f8
+; CHECK-NEXT:    brasl %r14, __extendhfsf2@PLT
+; CHECK-NEXT:    ler %f10, %f0
+; CHECK-NEXT:    ler %f0, %f9
+; CHECK-NEXT:    brasl %r14, __extendhfsf2@PLT
+; CHECK-NEXT:    aebr %f0, %f10
+; CHECK-NEXT:    brasl %r14, __truncsfhf2@PLT
+; CHECK-NEXT:    # kill: def $f0h killed $f0h def $f0d
+; CHECK-NEXT:    lgdr %r1, %f0
+; CHECK-NEXT:    l %r0, 0(%r12)
+; CHECK-NEXT:    srlg %r1, %r1, 48
+; CHECK-NEXT:    lgdr %r2, %f9
+; CHECK-NEXT:    srlg %r2, %r2, 48
+; CHECK-NEXT:  .LBB0_3: # %atomicrmw.start
+; CHECK-NEXT:    # Parent Loop BB0_2 Depth=1
+; CHECK-NEXT:    # => This Inner Loop Header: Depth=2
+; CHECK-NEXT:    rll %r3, %r0, 16(%r13)
+; CHECK-NEXT:    risbg %r1, %r3, 32, 47, 0
+; CHECK-NEXT:    llhr %r3, %r3
+; CHECK-NEXT:    cr %r3, %r2
+; CHECK-NEXT:    jlh .LBB0_1
+; CHECK-NEXT:  # %bb.4: # %atomicrmw.start
+; CHECK-NEXT:    # in Loop: Header=BB0_3 Depth=2
+; CHECK-NEXT:    rll %r4, %r1, -16(%r11)
+; CHECK-NEXT:    cs %r0, %r4, 0(%r12)
+; CHECK-NEXT:    jl .LBB0_3
+; CHECK-NEXT:    j .LBB0_1
+; CHECK-NEXT:  .LBB0_5: # %atomicrmw.end
+; CHECK-NEXT:    ler %f0, %f9
+; CHECK-NEXT:    ld %f8, 176(%r15) # 8-byte Folded Reload
+; CHECK-NEXT:    ld %f9, 168(%r15) # 8-byte Folded Reload
+; CHECK-NEXT:    ld %f10, 160(%r15) # 8-byte Folded Reload
+; CHECK-NEXT:    lmg %r11, %r15, 272(%r15)
+; CHECK-NEXT:    br %r14
+  %res = atomicrmw fadd ptr %src, half %b seq_cst
+  ret half %res
+}

llvm/test/CodeGen/SystemZ/fp-half-cmp.ll

@@ -2,7 +2,7 @@
 ; RUN: llc < %s -mtriple=s390x-linux-gnu -mcpu=z16 -verify-machineinstrs \
 ; RUN:   | FileCheck %s
 ;
-; Various tests comparaisons and uses involving 16-bit floating point (half).
+; Some tests with comparisons and their uses involving 16-bit floating point.
 
 ; fcmp half; select half
 define half @fun0(half %Arg0, half %Arg1) {

llvm/test/CodeGen/SystemZ/fp-half-mem.ll

@@ -0,0 +1,65 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
+; RUN: llc < %s -mtriple=s390x-linux-gnu -mcpu=zEC12 -verify-machineinstrs \
+; RUN:   | FileCheck %s --check-prefix=NOVEC
+; RUN: llc < %s -mtriple=s390x-linux-gnu -mcpu=z16 -verify-machineinstrs \
+; RUN:   | FileCheck %s  --check-prefix=VECTOR
+
+declare void @foo(ptr)
+
+; Test an alloca.
+define half @f1() {
+; NOVEC-LABEL: f1:
+; NOVEC:       # %bb.0:
+; NOVEC-NEXT:    stmg %r14, %r15, 112(%r15)
+; NOVEC-NEXT:    .cfi_offset %r14, -48
+; NOVEC-NEXT:    .cfi_offset %r15, -40
+; NOVEC-NEXT:    aghi %r15, -168
+; NOVEC-NEXT:    .cfi_def_cfa_offset 328
+; NOVEC-NEXT:    la %r2, 166(%r15)
+; NOVEC-NEXT:    brasl %r14, foo@PLT
+; NOVEC-NEXT:    lh %r0, 166(%r15)
+; NOVEC-NEXT:    sll %r0, 16
+; NOVEC-NEXT:    risbhg %r0, %r0, 0, 159, 32
+; NOVEC-NEXT:    ldgr %f0, %r0
+; NOVEC-NEXT:    # kill: def $f0h killed $f0h killed $f0d
+; NOVEC-NEXT:    lmg %r14, %r15, 280(%r15)
+; NOVEC-NEXT:    br %r14
+;
+; VECTOR-LABEL: f1:
+; VECTOR:       # %bb.0:
+; VECTOR-NEXT:    stmg %r14, %r15, 112(%r15)
+; VECTOR-NEXT:    .cfi_offset %r14, -48
+; VECTOR-NEXT:    .cfi_offset %r15, -40
+; VECTOR-NEXT:    aghi %r15, -168
+; VECTOR-NEXT:    .cfi_def_cfa_offset 328
+; VECTOR-NEXT:    la %r2, 166(%r15)
+; VECTOR-NEXT:    brasl %r14, foo@PLT
+; VECTOR-NEXT:    vlreph %v0, 166(%r15)
+; VECTOR-NEXT:    lmg %r14, %r15, 280(%r15)
+; VECTOR-NEXT:    br %r14
+  %ptr = alloca half
+  call void @foo(ptr %ptr)
+  %orig = load half, ptr %ptr
+  ret half %orig
+}
+
+; Test accessing a half element of an aggregate type.
+%s.half = type { half, half, half, half, half }
+define half @f2(ptr %P) {
+; NOVEC-LABEL: f2:
+; NOVEC:       # %bb.0:
+; NOVEC-NEXT:    lh %r0, 6(%r2)
+; NOVEC-NEXT:    sll %r0, 16
+; NOVEC-NEXT:    risbhg %r0, %r0, 0, 159, 32
+; NOVEC-NEXT:    ldgr %f0, %r0
+; NOVEC-NEXT:    # kill: def $f0h killed $f0h killed $f0d
+; NOVEC-NEXT:    br %r14
+;
+; VECTOR-LABEL: f2:
+; VECTOR:       # %bb.0:
+; VECTOR-NEXT:    vlreph %v0, 6(%r2)
+; VECTOR-NEXT:    br %r14
+  %gep = getelementptr inbounds %s.half, ptr %P, i64 0, i32 3
+  %res = load half, ptr %gep
+  ret half %res
+}