[Qemu-devel] [PATCH 01/10] target-ppc: optimize fabs, fnabs, fneg

[Aurelien Jarno <aurelien@aurel32.net>]

fabs, fnabs and fneg are just flipping the bit sign of an FP register,
this can be implemented in TCG instead of using softfloat.

Cc: Alexander Graf <agraf@suse.de>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
---
 target-ppc/fpu_helper.c |   31 -------------------------------
 target-ppc/helper.h     |    3 ---
 target-ppc/translate.c  |   40 ++++++++++++++++++++++++++++++++++------
 3 files changed, 34 insertions(+), 40 deletions(-)

diff --git a/target-ppc/fpu_helper.c b/target-ppc/fpu_helper.c
index 9e779ea..2f0db4e 100644
--- a/target-ppc/fpu_helper.c
+++ b/target-ppc/fpu_helper.c
@@ -595,37 +595,6 @@ uint64_t helper_fdiv(CPUPPCState *env, uint64_t arg1, uint64_t arg2)
     return farg1.ll;
 }
 
-/* fabs */
-uint64_t helper_fabs(CPUPPCState *env, uint64_t arg)
-{
-    CPU_DoubleU farg;
-
-    farg.ll = arg;
-    farg.d = float64_abs(farg.d);
-    return farg.ll;
-}
-
-/* fnabs */
-uint64_t helper_fnabs(CPUPPCState *env, uint64_t arg)
-{
-    CPU_DoubleU farg;
-
-    farg.ll = arg;
-    farg.d = float64_abs(farg.d);
-    farg.d = float64_chs(farg.d);
-    return farg.ll;
-}
-
-/* fneg */
-uint64_t helper_fneg(CPUPPCState *env, uint64_t arg)
-{
-    CPU_DoubleU farg;
-
-    farg.ll = arg;
-    farg.d = float64_chs(farg.d);
-    return farg.ll;
-}
-
 /* fctiw - fctiw. */
 uint64_t helper_fctiw(CPUPPCState *env, uint64_t arg)
 {
diff --git a/target-ppc/helper.h b/target-ppc/helper.h
index d33ee66..07397b2 100644
--- a/target-ppc/helper.h
+++ b/target-ppc/helper.h
@@ -80,9 +80,6 @@ DEF_HELPER_4(fmadd, i64, env, i64, i64, i64)
 DEF_HELPER_4(fmsub, i64, env, i64, i64, i64)
 DEF_HELPER_4(fnmadd, i64, env, i64, i64, i64)
 DEF_HELPER_4(fnmsub, i64, env, i64, i64, i64)
-DEF_HELPER_2(fabs, i64, env, i64)
-DEF_HELPER_2(fnabs, i64, env, i64)
-DEF_HELPER_2(fneg, i64, env, i64)
 DEF_HELPER_2(fsqrt, i64, env, i64)
 DEF_HELPER_2(fre, i64, env, i64)
 DEF_HELPER_2(fres, i64, env, i64)
diff --git a/target-ppc/translate.c b/target-ppc/translate.c
index 5e741d1..68cd1c5 100644
--- a/target-ppc/translate.c
+++ b/target-ppc/translate.c
@@ -2161,7 +2161,16 @@ static void gen_fcmpu(DisasContext *ctx)
 /***                         Floating-point move                           ***/
 /* fabs */
 /* XXX: beware that fabs never checks for NaNs nor update FPSCR */
-GEN_FLOAT_B(abs, 0x08, 0x08, 0, PPC_FLOAT);
+static void gen_fabs(DisasContext *ctx)
+{
+    if (unlikely(!ctx->fpu_enabled)) {
+        gen_exception(ctx, POWERPC_EXCP_FPU);
+        return;
+    }
+    tcg_gen_andi_i64(cpu_fpr[rD(ctx->opcode)], cpu_fpr[rB(ctx->opcode)],
+                     ~(1LL << 63));
+    gen_compute_fprf(cpu_fpr[rD(ctx->opcode)], 0, Rc(ctx->opcode) != 0);
+}
 
 /* fmr  - fmr. */
 /* XXX: beware that fmr never checks for NaNs nor update FPSCR */
@@ -2177,10 +2186,29 @@ static void gen_fmr(DisasContext *ctx)
 
 /* fnabs */
 /* XXX: beware that fnabs never checks for NaNs nor update FPSCR */
-GEN_FLOAT_B(nabs, 0x08, 0x04, 0, PPC_FLOAT);
+static void gen_fnabs(DisasContext *ctx)
+{
+    if (unlikely(!ctx->fpu_enabled)) {
+        gen_exception(ctx, POWERPC_EXCP_FPU);
+        return;
+    }
+    tcg_gen_ori_i64(cpu_fpr[rD(ctx->opcode)], cpu_fpr[rB(ctx->opcode)],
+                    1LL << 63);
+    gen_compute_fprf(cpu_fpr[rD(ctx->opcode)], 0, Rc(ctx->opcode) != 0);
+}
+
 /* fneg */
 /* XXX: beware that fneg never checks for NaNs nor update FPSCR */
-GEN_FLOAT_B(neg, 0x08, 0x01, 0, PPC_FLOAT);
+static void gen_fneg(DisasContext *ctx)
+{
+    if (unlikely(!ctx->fpu_enabled)) {
+        gen_exception(ctx, POWERPC_EXCP_FPU);
+        return;
+    }
+    tcg_gen_xori_i64(cpu_fpr[rD(ctx->opcode)], cpu_fpr[rB(ctx->opcode)],
+                     1LL << 63);
+    gen_compute_fprf(cpu_fpr[rD(ctx->opcode)], 0, Rc(ctx->opcode) != 0);
+}
 
 /***                  Floating-Point status & ctrl register                ***/
 
@@ -8476,7 +8504,10 @@ GEN_HANDLER(fsqrt, 0x3F, 0x16, 0xFF, 0x001F07C0, PPC_FLOAT_FSQRT),
 GEN_HANDLER(fsqrts, 0x3B, 0x16, 0xFF, 0x001F07C0, PPC_FLOAT_FSQRT),
 GEN_HANDLER(fcmpo, 0x3F, 0x00, 0x01, 0x00600001, PPC_FLOAT),
 GEN_HANDLER(fcmpu, 0x3F, 0x00, 0x00, 0x00600001, PPC_FLOAT),
+GEN_HANDLER(fabs, 0x3F, 0x08, 0x08, 0x001F0000, PPC_FLOAT),
 GEN_HANDLER(fmr, 0x3F, 0x08, 0x02, 0x001F0000, PPC_FLOAT),
+GEN_HANDLER(fnabs, 0x3F, 0x08, 0x04, 0x001F0000, PPC_FLOAT),
+GEN_HANDLER(fneg, 0x3F, 0x08, 0x01, 0x001F0000, PPC_FLOAT),
 GEN_HANDLER(mcrfs, 0x3F, 0x00, 0x02, 0x0063F801, PPC_FLOAT),
 GEN_HANDLER(mffs, 0x3F, 0x07, 0x12, 0x001FF800, PPC_FLOAT),
 GEN_HANDLER(mtfsb0, 0x3F, 0x06, 0x02, 0x001FF800, PPC_FLOAT),
@@ -8833,9 +8864,6 @@ GEN_FLOAT_B(rin, 0x08, 0x0C, 1, PPC_FLOAT_EXT),
 GEN_FLOAT_B(riz, 0x08, 0x0D, 1, PPC_FLOAT_EXT),
 GEN_FLOAT_B(rip, 0x08, 0x0E, 1, PPC_FLOAT_EXT),
 GEN_FLOAT_B(rim, 0x08, 0x0F, 1, PPC_FLOAT_EXT),
-GEN_FLOAT_B(abs, 0x08, 0x08, 0, PPC_FLOAT),
-GEN_FLOAT_B(nabs, 0x08, 0x04, 0, PPC_FLOAT),
-GEN_FLOAT_B(neg, 0x08, 0x01, 0, PPC_FLOAT),
 
 #undef GEN_LD
 #undef GEN_LDU
-- 
1.7.10.4