target/arm: Implement fp16 for VCEQ, VCGE, VCGT comparisons

Convert the Neon floating-point vector comparison ops VCEQ,
VCGE and VCGT over to using a gvec helper and use this to
implement the fp16 case.

(We put the float16_ceq() etc functions above the DO_2OP()
macro definition because later when we convert the
compare-against-zero instructions we'll want their
definitions to be visible at that point in the source file.)

Backports ad505db233b89b7fd4b5a98b6f0e8ac8d05b11db

qemu/aarch64.h

@@ -1272,6 +1272,12 @@
 #define helper_gvec_fcaddh helper_gvec_fcaddh_aarch64
 #define helper_gvec_fcadds helper_gvec_fcadds_aarch64
 #define helper_gvec_fcaddd helper_gvec_fcaddd_aarch64
+#define helper_gvec_fceq_h helper_gvec_fceq_h_aarch64
+#define helper_gvec_fceq_s helper_gvec_fceq_s_aarch64
+#define helper_gvec_fcge_h helper_gvec_fcge_h_aarch64
+#define helper_gvec_fcge_s helper_gvec_fcge_s_aarch64
+#define helper_gvec_fcgt_h helper_gvec_fcgt_h_aarch64
+#define helper_gvec_fcgt_s helper_gvec_fcgt_s_aarch64
 #define helper_gvec_fcmlad helper_gvec_fcmlad_aarch64
 #define helper_gvec_fcmlah helper_gvec_fcmlah_aarch64
 #define helper_gvec_fcmlah_idx helper_gvec_fcmlah_idx_aarch64

qemu/aarch64eb.h

@@ -1272,6 +1272,12 @@
 #define helper_gvec_fcaddh helper_gvec_fcaddh_aarch64eb
 #define helper_gvec_fcadds helper_gvec_fcadds_aarch64eb
 #define helper_gvec_fcaddd helper_gvec_fcaddd_aarch64eb
+#define helper_gvec_fceq_h helper_gvec_fceq_h_aarch64eb
+#define helper_gvec_fceq_s helper_gvec_fceq_s_aarch64eb
+#define helper_gvec_fcge_h helper_gvec_fcge_h_aarch64eb
+#define helper_gvec_fcge_s helper_gvec_fcge_s_aarch64eb
+#define helper_gvec_fcgt_h helper_gvec_fcgt_h_aarch64eb
+#define helper_gvec_fcgt_s helper_gvec_fcgt_s_aarch64eb
 #define helper_gvec_fcmlad helper_gvec_fcmlad_aarch64eb
 #define helper_gvec_fcmlah helper_gvec_fcmlah_aarch64eb
 #define helper_gvec_fcmlah_idx helper_gvec_fcmlah_idx_aarch64eb

qemu/arm.h

@@ -1272,6 +1272,12 @@
 #define helper_gvec_fcaddh helper_gvec_fcaddh_arm
 #define helper_gvec_fcadds helper_gvec_fcadds_arm
 #define helper_gvec_fcaddd helper_gvec_fcaddd_arm
+#define helper_gvec_fceq_h helper_gvec_fceq_h_arm
+#define helper_gvec_fceq_s helper_gvec_fceq_s_arm
+#define helper_gvec_fcge_h helper_gvec_fcge_h_arm
+#define helper_gvec_fcge_s helper_gvec_fcge_s_arm
+#define helper_gvec_fcgt_h helper_gvec_fcgt_h_arm
+#define helper_gvec_fcgt_s helper_gvec_fcgt_s_arm
 #define helper_gvec_fcmlad helper_gvec_fcmlad_arm
 #define helper_gvec_fcmlah helper_gvec_fcmlah_arm
 #define helper_gvec_fcmlah_idx helper_gvec_fcmlah_idx_arm

qemu/armeb.h

@@ -1272,6 +1272,12 @@
 #define helper_gvec_fcaddh helper_gvec_fcaddh_armeb
 #define helper_gvec_fcadds helper_gvec_fcadds_armeb
 #define helper_gvec_fcaddd helper_gvec_fcaddd_armeb
+#define helper_gvec_fceq_h helper_gvec_fceq_h_armeb
+#define helper_gvec_fceq_s helper_gvec_fceq_s_armeb
+#define helper_gvec_fcge_h helper_gvec_fcge_h_armeb
+#define helper_gvec_fcge_s helper_gvec_fcge_s_armeb
+#define helper_gvec_fcgt_h helper_gvec_fcgt_h_armeb
+#define helper_gvec_fcgt_s helper_gvec_fcgt_s_armeb
 #define helper_gvec_fcmlad helper_gvec_fcmlad_armeb
 #define helper_gvec_fcmlah helper_gvec_fcmlah_armeb
 #define helper_gvec_fcmlah_idx helper_gvec_fcmlah_idx_armeb

qemu/header_gen.py

@@ -1278,6 +1278,12 @@ symbols = (
     'helper_gvec_fcaddh',
     'helper_gvec_fcadds',
     'helper_gvec_fcaddd',
+    'helper_gvec_fceq_h',
+    'helper_gvec_fceq_s',
+    'helper_gvec_fcge_h',
+    'helper_gvec_fcge_s',
+    'helper_gvec_fcgt_h',
+    'helper_gvec_fcgt_s',
     'helper_gvec_fcmlad',
     'helper_gvec_fcmlah',
     'helper_gvec_fcmlah_idx',

qemu/m68k.h

@@ -1272,6 +1272,12 @@
 #define helper_gvec_fcaddh helper_gvec_fcaddh_m68k
 #define helper_gvec_fcadds helper_gvec_fcadds_m68k
 #define helper_gvec_fcaddd helper_gvec_fcaddd_m68k
+#define helper_gvec_fceq_h helper_gvec_fceq_h_m68k
+#define helper_gvec_fceq_s helper_gvec_fceq_s_m68k
+#define helper_gvec_fcge_h helper_gvec_fcge_h_m68k
+#define helper_gvec_fcge_s helper_gvec_fcge_s_m68k
+#define helper_gvec_fcgt_h helper_gvec_fcgt_h_m68k
+#define helper_gvec_fcgt_s helper_gvec_fcgt_s_m68k
 #define helper_gvec_fcmlad helper_gvec_fcmlad_m68k
 #define helper_gvec_fcmlah helper_gvec_fcmlah_m68k
 #define helper_gvec_fcmlah_idx helper_gvec_fcmlah_idx_m68k

qemu/mips.h

@@ -1272,6 +1272,12 @@
 #define helper_gvec_fcaddh helper_gvec_fcaddh_mips
 #define helper_gvec_fcadds helper_gvec_fcadds_mips
 #define helper_gvec_fcaddd helper_gvec_fcaddd_mips
+#define helper_gvec_fceq_h helper_gvec_fceq_h_mips
+#define helper_gvec_fceq_s helper_gvec_fceq_s_mips
+#define helper_gvec_fcge_h helper_gvec_fcge_h_mips
+#define helper_gvec_fcge_s helper_gvec_fcge_s_mips
+#define helper_gvec_fcgt_h helper_gvec_fcgt_h_mips
+#define helper_gvec_fcgt_s helper_gvec_fcgt_s_mips
 #define helper_gvec_fcmlad helper_gvec_fcmlad_mips
 #define helper_gvec_fcmlah helper_gvec_fcmlah_mips
 #define helper_gvec_fcmlah_idx helper_gvec_fcmlah_idx_mips

qemu/mips64.h

@@ -1272,6 +1272,12 @@
 #define helper_gvec_fcaddh helper_gvec_fcaddh_mips64
 #define helper_gvec_fcadds helper_gvec_fcadds_mips64
 #define helper_gvec_fcaddd helper_gvec_fcaddd_mips64
+#define helper_gvec_fceq_h helper_gvec_fceq_h_mips64
+#define helper_gvec_fceq_s helper_gvec_fceq_s_mips64
+#define helper_gvec_fcge_h helper_gvec_fcge_h_mips64
+#define helper_gvec_fcge_s helper_gvec_fcge_s_mips64
+#define helper_gvec_fcgt_h helper_gvec_fcgt_h_mips64
+#define helper_gvec_fcgt_s helper_gvec_fcgt_s_mips64
 #define helper_gvec_fcmlad helper_gvec_fcmlad_mips64
 #define helper_gvec_fcmlah helper_gvec_fcmlah_mips64
 #define helper_gvec_fcmlah_idx helper_gvec_fcmlah_idx_mips64

qemu/mips64el.h

@@ -1272,6 +1272,12 @@
 #define helper_gvec_fcaddh helper_gvec_fcaddh_mips64el
 #define helper_gvec_fcadds helper_gvec_fcadds_mips64el
 #define helper_gvec_fcaddd helper_gvec_fcaddd_mips64el
+#define helper_gvec_fceq_h helper_gvec_fceq_h_mips64el
+#define helper_gvec_fceq_s helper_gvec_fceq_s_mips64el
+#define helper_gvec_fcge_h helper_gvec_fcge_h_mips64el
+#define helper_gvec_fcge_s helper_gvec_fcge_s_mips64el
+#define helper_gvec_fcgt_h helper_gvec_fcgt_h_mips64el
+#define helper_gvec_fcgt_s helper_gvec_fcgt_s_mips64el
 #define helper_gvec_fcmlad helper_gvec_fcmlad_mips64el
 #define helper_gvec_fcmlah helper_gvec_fcmlah_mips64el
 #define helper_gvec_fcmlah_idx helper_gvec_fcmlah_idx_mips64el

qemu/mipsel.h

@@ -1272,6 +1272,12 @@
 #define helper_gvec_fcaddh helper_gvec_fcaddh_mipsel
 #define helper_gvec_fcadds helper_gvec_fcadds_mipsel
 #define helper_gvec_fcaddd helper_gvec_fcaddd_mipsel
+#define helper_gvec_fceq_h helper_gvec_fceq_h_mipsel
+#define helper_gvec_fceq_s helper_gvec_fceq_s_mipsel
+#define helper_gvec_fcge_h helper_gvec_fcge_h_mipsel
+#define helper_gvec_fcge_s helper_gvec_fcge_s_mipsel
+#define helper_gvec_fcgt_h helper_gvec_fcgt_h_mipsel
+#define helper_gvec_fcgt_s helper_gvec_fcgt_s_mipsel
 #define helper_gvec_fcmlad helper_gvec_fcmlad_mipsel
 #define helper_gvec_fcmlah helper_gvec_fcmlah_mipsel
 #define helper_gvec_fcmlah_idx helper_gvec_fcmlah_idx_mipsel

qemu/powerpc.h

@@ -1272,6 +1272,12 @@
 #define helper_gvec_fcaddh helper_gvec_fcaddh_powerpc
 #define helper_gvec_fcadds helper_gvec_fcadds_powerpc
 #define helper_gvec_fcaddd helper_gvec_fcaddd_powerpc
+#define helper_gvec_fceq_h helper_gvec_fceq_h_powerpc
+#define helper_gvec_fceq_s helper_gvec_fceq_s_powerpc
+#define helper_gvec_fcge_h helper_gvec_fcge_h_powerpc
+#define helper_gvec_fcge_s helper_gvec_fcge_s_powerpc
+#define helper_gvec_fcgt_h helper_gvec_fcgt_h_powerpc
+#define helper_gvec_fcgt_s helper_gvec_fcgt_s_powerpc
 #define helper_gvec_fcmlad helper_gvec_fcmlad_powerpc
 #define helper_gvec_fcmlah helper_gvec_fcmlah_powerpc
 #define helper_gvec_fcmlah_idx helper_gvec_fcmlah_idx_powerpc

qemu/riscv32.h

@@ -1272,6 +1272,12 @@
 #define helper_gvec_fcaddh helper_gvec_fcaddh_riscv32
 #define helper_gvec_fcadds helper_gvec_fcadds_riscv32
 #define helper_gvec_fcaddd helper_gvec_fcaddd_riscv32
+#define helper_gvec_fceq_h helper_gvec_fceq_h_riscv32
+#define helper_gvec_fceq_s helper_gvec_fceq_s_riscv32
+#define helper_gvec_fcge_h helper_gvec_fcge_h_riscv32
+#define helper_gvec_fcge_s helper_gvec_fcge_s_riscv32
+#define helper_gvec_fcgt_h helper_gvec_fcgt_h_riscv32
+#define helper_gvec_fcgt_s helper_gvec_fcgt_s_riscv32
 #define helper_gvec_fcmlad helper_gvec_fcmlad_riscv32
 #define helper_gvec_fcmlah helper_gvec_fcmlah_riscv32
 #define helper_gvec_fcmlah_idx helper_gvec_fcmlah_idx_riscv32

qemu/riscv64.h

@@ -1272,6 +1272,12 @@
 #define helper_gvec_fcaddh helper_gvec_fcaddh_riscv64
 #define helper_gvec_fcadds helper_gvec_fcadds_riscv64
 #define helper_gvec_fcaddd helper_gvec_fcaddd_riscv64
+#define helper_gvec_fceq_h helper_gvec_fceq_h_riscv64
+#define helper_gvec_fceq_s helper_gvec_fceq_s_riscv64
+#define helper_gvec_fcge_h helper_gvec_fcge_h_riscv64
+#define helper_gvec_fcge_s helper_gvec_fcge_s_riscv64
+#define helper_gvec_fcgt_h helper_gvec_fcgt_h_riscv64
+#define helper_gvec_fcgt_s helper_gvec_fcgt_s_riscv64
 #define helper_gvec_fcmlad helper_gvec_fcmlad_riscv64
 #define helper_gvec_fcmlah helper_gvec_fcmlah_riscv64
 #define helper_gvec_fcmlah_idx helper_gvec_fcmlah_idx_riscv64

qemu/sparc.h

@@ -1272,6 +1272,12 @@
 #define helper_gvec_fcaddh helper_gvec_fcaddh_sparc
 #define helper_gvec_fcadds helper_gvec_fcadds_sparc
 #define helper_gvec_fcaddd helper_gvec_fcaddd_sparc
+#define helper_gvec_fceq_h helper_gvec_fceq_h_sparc
+#define helper_gvec_fceq_s helper_gvec_fceq_s_sparc
+#define helper_gvec_fcge_h helper_gvec_fcge_h_sparc
+#define helper_gvec_fcge_s helper_gvec_fcge_s_sparc
+#define helper_gvec_fcgt_h helper_gvec_fcgt_h_sparc
+#define helper_gvec_fcgt_s helper_gvec_fcgt_s_sparc
 #define helper_gvec_fcmlad helper_gvec_fcmlad_sparc
 #define helper_gvec_fcmlah helper_gvec_fcmlah_sparc
 #define helper_gvec_fcmlah_idx helper_gvec_fcmlah_idx_sparc

qemu/sparc64.h

@@ -1272,6 +1272,12 @@
 #define helper_gvec_fcaddh helper_gvec_fcaddh_sparc64
 #define helper_gvec_fcadds helper_gvec_fcadds_sparc64
 #define helper_gvec_fcaddd helper_gvec_fcaddd_sparc64
+#define helper_gvec_fceq_h helper_gvec_fceq_h_sparc64
+#define helper_gvec_fceq_s helper_gvec_fceq_s_sparc64
+#define helper_gvec_fcge_h helper_gvec_fcge_h_sparc64
+#define helper_gvec_fcge_s helper_gvec_fcge_s_sparc64
+#define helper_gvec_fcgt_h helper_gvec_fcgt_h_sparc64
+#define helper_gvec_fcgt_s helper_gvec_fcgt_s_sparc64
 #define helper_gvec_fcmlad helper_gvec_fcmlad_sparc64
 #define helper_gvec_fcmlah helper_gvec_fcmlah_sparc64
 #define helper_gvec_fcmlah_idx helper_gvec_fcmlah_idx_sparc64

qemu/target/arm/helper.h

@@ -630,6 +630,15 @@ DEF_HELPER_FLAGS_5(gvec_fmul_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32)
 DEF_HELPER_FLAGS_5(gvec_fabd_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32)
 DEF_HELPER_FLAGS_5(gvec_fabd_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32)
 
+DEF_HELPER_FLAGS_5(gvec_fceq_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32)
+DEF_HELPER_FLAGS_5(gvec_fceq_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32)
+
+DEF_HELPER_FLAGS_5(gvec_fcge_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32)
+DEF_HELPER_FLAGS_5(gvec_fcge_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32)
+
+DEF_HELPER_FLAGS_5(gvec_fcgt_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32)
+DEF_HELPER_FLAGS_5(gvec_fcgt_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32)
+
 DEF_HELPER_FLAGS_5(gvec_ftsmul_h, TCG_CALL_NO_RWG,
                    void, ptr, ptr, ptr, ptr, i32)
 DEF_HELPER_FLAGS_5(gvec_ftsmul_s, TCG_CALL_NO_RWG,

qemu/target/arm/translate-neon.inc.c

@@ -1131,6 +1131,9 @@ DO_3S_FP_GVEC(VADD, gen_helper_gvec_fadd_s, gen_helper_gvec_fadd_h)
 DO_3S_FP_GVEC(VSUB, gen_helper_gvec_fsub_s, gen_helper_gvec_fsub_h)
 DO_3S_FP_GVEC(VABD, gen_helper_gvec_fabd_s, gen_helper_gvec_fabd_h)
 DO_3S_FP_GVEC(VMUL, gen_helper_gvec_fmul_s, gen_helper_gvec_fmul_h)
+DO_3S_FP_GVEC(VCEQ, gen_helper_gvec_fceq_s, gen_helper_gvec_fceq_h)
+DO_3S_FP_GVEC(VCGE, gen_helper_gvec_fcge_s, gen_helper_gvec_fcge_h)
+DO_3S_FP_GVEC(VCGT, gen_helper_gvec_fcgt_s, gen_helper_gvec_fcgt_h)
 
 /*
  * For all the functions using this macro, size == 1 means fp16,
@@ -1146,9 +1149,6 @@ DO_3S_FP_GVEC(VMUL, gen_helper_gvec_fmul_s, gen_helper_gvec_fmul_h)
         return do_3same_fp(s, a, FUNC, READS_VD);                   \
     }
 
-DO_3S_FP(VCEQ, gen_helper_neon_ceq_f32, false)
-DO_3S_FP(VCGE, gen_helper_neon_cge_f32, false)
-DO_3S_FP(VCGT, gen_helper_neon_cgt_f32, false)
 DO_3S_FP(VACGE, gen_helper_neon_acge_f32, false)
 DO_3S_FP(VACGT, gen_helper_neon_acgt_f32, false)
 DO_3S_FP(VMAX, gen_helper_vfp_maxs, false)

qemu/target/arm/vec_helper.c

@@ -657,6 +657,41 @@ void HELPER(gvec_fcmlad)(void *vd, void *vn, void *vm,
     clear_tail(d, opr_sz, simd_maxsz(desc));
 }
 
+/*
+ * Floating point comparisons producing an integer result (all 1s or all 0s).
+ * Note that EQ doesn't signal InvalidOp for QNaNs but GE and GT do.
+ * Softfloat routines return 0/1, which we convert to the 0/-1 Neon requires.
+ */
+static uint16_t float16_ceq(float16 op1, float16 op2, float_status *stat)
+{
+    return -float16_eq_quiet(op1, op2, stat);
+}
+
+static uint32_t float32_ceq(float32 op1, float32 op2, float_status *stat)
+{
+    return -float32_eq_quiet(op1, op2, stat);
+}
+
+static uint16_t float16_cge(float16 op1, float16 op2, float_status *stat)
+{
+    return -float16_le(op2, op1, stat);
+}
+
+static uint32_t float32_cge(float32 op1, float32 op2, float_status *stat)
+{
+    return -float32_le(op2, op1, stat);
+}
+
+static uint16_t float16_cgt(float16 op1, float16 op2, float_status *stat)
+{
+    return -float16_lt(op2, op1, stat);
+}
+
+static uint32_t float32_cgt(float32 op1, float32 op2, float_status *stat)
+{
+    return -float32_lt(op2, op1, stat);
+}
+
 #define DO_2OP(NAME, FUNC, TYPE) \
 void HELPER(NAME)(void *vd, void *vn, void *stat, uint32_t desc)  \
 {                                                                 \
@@ -748,6 +783,15 @@ DO_3OP(gvec_ftsmul_d, float64_ftsmul, float64)
 DO_3OP(gvec_fabd_h, float16_abd, float16)
 DO_3OP(gvec_fabd_s, float32_abd, float32)
 
+DO_3OP(gvec_fceq_h, float16_ceq, float16)
+DO_3OP(gvec_fceq_s, float32_ceq, float32)
+
+DO_3OP(gvec_fcge_h, float16_cge, float16)
+DO_3OP(gvec_fcge_s, float32_cge, float32)
+
+DO_3OP(gvec_fcgt_h, float16_cgt, float16)
+DO_3OP(gvec_fcgt_s, float32_cgt, float32)
+
 #ifdef TARGET_AARCH64
 
 DO_3OP(gvec_recps_h, helper_recpsf_f16, float16)

qemu/x86_64.h

@@ -1272,6 +1272,12 @@
 #define helper_gvec_fcaddh helper_gvec_fcaddh_x86_64
 #define helper_gvec_fcadds helper_gvec_fcadds_x86_64
 #define helper_gvec_fcaddd helper_gvec_fcaddd_x86_64
+#define helper_gvec_fceq_h helper_gvec_fceq_h_x86_64
+#define helper_gvec_fceq_s helper_gvec_fceq_s_x86_64
+#define helper_gvec_fcge_h helper_gvec_fcge_h_x86_64
+#define helper_gvec_fcge_s helper_gvec_fcge_s_x86_64
+#define helper_gvec_fcgt_h helper_gvec_fcgt_h_x86_64
+#define helper_gvec_fcgt_s helper_gvec_fcgt_s_x86_64
 #define helper_gvec_fcmlad helper_gvec_fcmlad_x86_64
 #define helper_gvec_fcmlah helper_gvec_fcmlah_x86_64
 #define helper_gvec_fcmlah_idx helper_gvec_fcmlah_idx_x86_64