[v4,06/22] RISC-V FPU Support

Commit Message

Michael Clark Feb. 5, 2018, 6:22 a.m. UTC

Helper routines for FPU instructions and NaN definitions.

Signed-off-by: Michael Clark <[e-mail]>
---
 fpu/softfloat-specialize.h |   7 +-
 target/riscv/fpu_helper.c  | 375 +++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 379 insertions(+), 3 deletions(-)
 create mode 100644 target/riscv/fpu_helper.c

Comments

Richard Henderson Feb. 5, 2018, 2:01 p.m. UTC | #1

On 02/04/2018 10:22 PM, Michael Clark wrote:
> +uint64_t helper_fmsub_s(CPURISCVState *env, uint64_t frs1, uint64_t frs2,
> +                        uint64_t frs3)
> +{
> +    return float32_muladd(frs1, frs2, frs3 ^ (uint32_t)INT32_MIN, 0,
> +                          &env->fp_status);

Missing the requested change to use float32_chs or float_muladd_negate_product
et al throughout the fma family of helpers.

Probably float_muladd_* since you've re-enabled default_nan_mode.


r~

Michael Clark Feb. 5, 2018, 10:01 p.m. UTC | #2

On Tue, Feb 6, 2018 at 3:01 AM, Richard Henderson <
[e-mail]> wrote:

> On 02/04/2018 10:22 PM, Michael Clark wrote:
> > +uint64_t helper_fmsub_s(CPURISCVState *env, uint64_t frs1, uint64_t
> frs2,
> > +                        uint64_t frs3)
> > +{
> > +    return float32_muladd(frs1, frs2, frs3 ^ (uint32_t)INT32_MIN, 0,
> > +                          &env->fp_status);
>
> Missing the requested change to use float32_chs or
> float_muladd_negate_product
> et al throughout the fma family of helpers.
>
> Probably float_muladd_* since you've re-enabled default_nan_mode.


I'll go through your comments again. I wasn't sure whether you had some of
these changes in your tree

I was previously a little hesitant about change floating point stuff but
now i've got riscv-tests running i'm less worried.

We are currently failing several floating point tests.

rv64ua-v-amoadd_d
rv64ua-v-amoadd_w
rv64ua-v-amoand_d
rv64ua-v-amoand_w
rv64ua-v-amomax_d
rv64ua-v-amomax_w
rv64ua-v-amomaxu_d
rv64ua-v-amomaxu_w
rv64ua-v-amomin_d
rv64ua-v-amomin_w
rv64ua-v-amominu_d
rv64ua-v-amominu_w
rv64ua-v-amoor_d
rv64ua-v-amoor_w
rv64ua-v-amoswap_d
rv64ua-v-amoswap_w
rv64ua-v-amoxor_d
rv64ua-v-amoxor_w
rv64ua-v-lrsc
rv64uc-v-rvc
rv64ud-v-fadd
rv64ud-v-fclass
rv64ud-v-fcmp
rv64ud-v-fcvt
rv64ud-v-fcvt_w
rv64ud-v-fdiv
rv64ud-v-fmadd
rv64ud-v-fmin
*** FAILED *** (tohost = 20
)rv64ud-v-ldst
*** FAILED *** (tohost = 6)
rv64ud-v-move
*** FAILED *** (tohost = 40)
rv64ud-v-recoding
rv64ud-v-structural
rv64uf-v-fadd
rv64uf-v-fclass
rv64uf-v-fcmp
rv64uf-v-fcvt
rv64uf-v-fcvt_w
rv64uf-v-fdiv
rv64uf-v-fmadd
rv64uf-v-fmin
*** FAILED *** (tohost = 20)
rv64uf-v-ldst
rv64uf-v-move
rv64uf-v-recoding
rv64ui-v-add
rv64ui-v-addi
rv64ui-v-addiw
rv64ui-v-addw
rv64ui-v-and
rv64ui-v-andi
rv64ui-v-auipc
rv64ui-v-beq
rv64ui-v-bge
rv64ui-v-bgeu
rv64ui-v-blt
rv64ui-v-bltu
rv64ui-v-bne
rv64ui-v-fence_i
rv64ui-v-jal
rv64ui-v-jalr
rv64ui-v-lb
rv64ui-v-lbu
rv64ui-v-ld
rv64ui-v-lh
rv64ui-v-lhu
rv64ui-v-lui
rv64ui-v-lw
rv64ui-v-lwu
rv64ui-v-or
rv64ui-v-ori
rv64ui-v-sb
rv64ui-v-sd
rv64ui-v-sh
rv64ui-v-simple
rv64ui-v-sll
rv64ui-v-slli
rv64ui-v-slliw
rv64ui-v-sllw
rv64ui-v-slt
rv64ui-v-slti
rv64ui-v-sltiu
rv64ui-v-sltu
rv64ui-v-sra
rv64ui-v-srai
rv64ui-v-sraiw
rv64ui-v-sraw
rv64ui-v-srl
rv64ui-v-srli
rv64ui-v-srliw
rv64ui-v-srlw
rv64ui-v-sub
rv64ui-v-subw
rv64ui-v-sw
rv64ui-v-xor
rv64ui-v-xori
rv64um-v-div
rv64um-v-divu
rv64um-v-divuw
rv64um-v-divw
rv64um-v-mul
rv64um-v-mulh
rv64um-v-mulhsu
rv64um-v-mulhu
rv64um-v-mulw
rv64um-v-rem
rv64um-v-remu
rv64um-v-remuw
rv64um-v-remw

Patch

diff --git a/fpu/softfloat-specialize.h b/fpu/softfloat-specialize.h
index de2c5d5..49ee578 100644
--- a/fpu/softfloat-specialize.h
+++ b/fpu/softfloat-specialize.h
@@ -114,7 +114,8 @@  float32 float32_default_nan(float_status *status)
 #if defined(TARGET_SPARC) || defined(TARGET_M68K)
     return const_float32(0x7FFFFFFF);
 #elif defined(TARGET_PPC) || defined(TARGET_ARM) || defined(TARGET_ALPHA) || \
-      defined(TARGET_XTENSA) || defined(TARGET_S390X) || defined(TARGET_TRICORE)
+      defined(TARGET_XTENSA) || defined(TARGET_S390X) || \
+      defined(TARGET_TRICORE) || defined(TARGET_RISCV)
     return const_float32(0x7FC00000);
 #elif defined(TARGET_HPPA)
     return const_float32(0x7FA00000);
@@ -139,7 +140,7 @@  float64 float64_default_nan(float_status *status)
 #if defined(TARGET_SPARC) || defined(TARGET_M68K)
     return const_float64(LIT64(0x7FFFFFFFFFFFFFFF));
 #elif defined(TARGET_PPC) || defined(TARGET_ARM) || defined(TARGET_ALPHA) || \
-      defined(TARGET_S390X)
+      defined(TARGET_S390X) || defined(TARGET_RISCV)
     return const_float64(LIT64(0x7FF8000000000000));
 #elif defined(TARGET_HPPA)
     return const_float64(LIT64(0x7FF4000000000000));
@@ -189,7 +190,7 @@  float128 float128_default_nan(float_status *status)
         r.high = LIT64(0x7FFF7FFFFFFFFFFF);
     } else {
         r.low = LIT64(0x0000000000000000);
-#if defined(TARGET_S390X) || defined(TARGET_PPC)
+#if defined(TARGET_S390X) || defined(TARGET_PPC) || defined(TARGET_RISCV)
         r.high = LIT64(0x7FFF800000000000);
 #else
         r.high = LIT64(0xFFFF800000000000);
diff --git a/target/riscv/fpu_helper.c b/target/riscv/fpu_helper.c
new file mode 100644
index 0000000..42d3633
--- /dev/null
+++ b/target/riscv/fpu_helper.c
@@ -0,0 +1,375 @@ 
+/*
+ * RISC-V FPU Emulation Helpers for QEMU.
+ *
+ * Author: Sagar Karandikar, [e-mail]
+ *
+ *
+ * This library is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2 of the License, or (at your option) any later version.
+ *
+ * This library is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with this library; if not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include "qemu/osdep.h"
+#include <stdlib.h>
+#include "cpu.h"
+#include "qemu/host-utils.h"
+#include "exec/exec-all.h"
+#include "exec/helper-proto.h"
+
+target_ulong cpu_riscv_get_fflags(CPURISCVState *env)
+{
+    int soft = get_float_exception_flags(&env->fp_status);
+    target_ulong hard = 0;
+
+    hard |= (soft & float_flag_inexact) ? FPEXC_NX : 0;
+    hard |= (soft & float_flag_underflow) ? FPEXC_UF : 0;
+    hard |= (soft & float_flag_overflow) ? FPEXC_OF : 0;
+    hard |= (soft & float_flag_divbyzero) ? FPEXC_DZ : 0;
+    hard |= (soft & float_flag_invalid) ? FPEXC_NV : 0;
+
+    return hard;
+}
+
+void cpu_riscv_set_fflags(CPURISCVState *env, target_ulong hard)
+{
+    int soft = 0;
+
+    soft |= (hard & FPEXC_NX) ? float_flag_inexact : 0;
+    soft |= (hard & FPEXC_UF) ? float_flag_underflow : 0;
+    soft |= (hard & FPEXC_OF) ? float_flag_overflow : 0;
+    soft |= (hard & FPEXC_DZ) ? float_flag_divbyzero : 0;
+    soft |= (hard & FPEXC_NV) ? float_flag_invalid : 0;
+
+    set_float_exception_flags(soft, &env->fp_status);
+}
+
+void helper_set_rounding_mode(CPURISCVState *env, uint32_t rm)
+{
+    int softrm;
+
+    if (rm == 7) {
+        rm = env->frm;
+    }
+    switch (rm) {
+    case 0:
+        softrm = float_round_nearest_even;
+        break;
+    case 1:
+        softrm = float_round_to_zero;
+        break;
+    case 2:
+        softrm = float_round_down;
+        break;
+    case 3:
+        softrm = float_round_up;
+        break;
+    case 4:
+        softrm = float_round_ties_away;
+        break;
+    default:
+        do_raise_exception_err(env, RISCV_EXCP_ILLEGAL_INST, GETPC());
+    }
+
+    set_float_rounding_mode(softrm, &env->fp_status);
+}
+
+uint64_t helper_fmadd_s(CPURISCVState *env, uint64_t frs1, uint64_t frs2,
+                        uint64_t frs3)
+{
+    return float32_muladd(frs1, frs2, frs3, 0, &env->fp_status);
+}
+
+uint64_t helper_fmadd_d(CPURISCVState *env, uint64_t frs1, uint64_t frs2,
+                        uint64_t frs3)
+{
+    return float64_muladd(frs1, frs2, frs3, 0, &env->fp_status);
+}
+
+uint64_t helper_fmsub_s(CPURISCVState *env, uint64_t frs1, uint64_t frs2,
+                        uint64_t frs3)
+{
+    return float32_muladd(frs1, frs2, frs3 ^ (uint32_t)INT32_MIN, 0,
+                          &env->fp_status);
+}
+
+uint64_t helper_fmsub_d(CPURISCVState *env, uint64_t frs1, uint64_t frs2,
+                        uint64_t frs3)
+{
+    return float64_muladd(frs1, frs2, frs3 ^ (uint64_t)INT64_MIN, 0,
+                          &env->fp_status);
+}
+
+uint64_t helper_fnmsub_s(CPURISCVState *env, uint64_t frs1, uint64_t frs2,
+                         uint64_t frs3)
+{
+    return float32_muladd(frs1 ^ (uint32_t)INT32_MIN, frs2, frs3, 0,
+                          &env->fp_status);
+}
+
+uint64_t helper_fnmsub_d(CPURISCVState *env, uint64_t frs1, uint64_t frs2,
+                         uint64_t frs3)
+{
+    return float64_muladd(frs1 ^ (uint64_t)INT64_MIN, frs2, frs3, 0,
+                          &env->fp_status);
+}
+
+uint64_t helper_fnmadd_s(CPURISCVState *env, uint64_t frs1, uint64_t frs2,
+                         uint64_t frs3)
+{
+    return float32_muladd(frs1 ^ (uint32_t)INT32_MIN, frs2,
+                          frs3 ^ (uint32_t)INT32_MIN, 0, &env->fp_status);
+}
+
+uint64_t helper_fnmadd_d(CPURISCVState *env, uint64_t frs1, uint64_t frs2,
+                         uint64_t frs3)
+{
+    return float64_muladd(frs1 ^ (uint64_t)INT64_MIN, frs2,
+                          frs3 ^ (uint64_t)INT64_MIN, 0, &env->fp_status);
+}
+
+uint64_t helper_fadd_s(CPURISCVState *env, uint64_t frs1, uint64_t frs2)
+{
+    return float32_add(frs1, frs2, &env->fp_status);
+}
+
+uint64_t helper_fsub_s(CPURISCVState *env, uint64_t frs1, uint64_t frs2)
+{
+    return float32_sub(frs1, frs2, &env->fp_status);
+}
+
+uint64_t helper_fmul_s(CPURISCVState *env, uint64_t frs1, uint64_t frs2)
+{
+    return float32_mul(frs1, frs2, &env->fp_status);
+}
+
+uint64_t helper_fdiv_s(CPURISCVState *env, uint64_t frs1, uint64_t frs2)
+{
+    return float32_div(frs1, frs2, &env->fp_status);
+}
+
+uint64_t helper_fmin_s(CPURISCVState *env, uint64_t frs1, uint64_t frs2)
+{
+    return float32_minnum(frs1, frs2, &env->fp_status);
+}
+
+uint64_t helper_fmax_s(CPURISCVState *env, uint64_t frs1, uint64_t frs2)
+{
+    return float32_maxnum(frs1, frs2, &env->fp_status);
+}
+
+uint64_t helper_fsqrt_s(CPURISCVState *env, uint64_t frs1)
+{
+    return float32_sqrt(frs1, &env->fp_status);
+}
+
+target_ulong helper_fle_s(CPURISCVState *env, uint64_t frs1, uint64_t frs2)
+{
+    return float32_le(frs1, frs2, &env->fp_status);
+}
+
+target_ulong helper_flt_s(CPURISCVState *env, uint64_t frs1, uint64_t frs2)
+{
+    return float32_lt(frs1, frs2, &env->fp_status);
+}
+
+target_ulong helper_feq_s(CPURISCVState *env, uint64_t frs1, uint64_t frs2)
+{
+    return float32_eq_quiet(frs1, frs2, &env->fp_status);
+}
+
+target_ulong helper_fcvt_w_s(CPURISCVState *env, uint64_t frs1)
+{
+    return float32_to_int32(frs1, &env->fp_status);
+}
+
+target_ulong helper_fcvt_wu_s(CPURISCVState *env, uint64_t frs1)
+{
+    return (int32_t)float32_to_uint32(frs1, &env->fp_status);
+}
+
+#if defined(TARGET_RISCV64)
+uint64_t helper_fcvt_l_s(CPURISCVState *env, uint64_t frs1)
+{
+    return float32_to_int64(frs1, &env->fp_status);
+}
+
+uint64_t helper_fcvt_lu_s(CPURISCVState *env, uint64_t frs1)
+{
+    return float32_to_uint64(frs1, &env->fp_status);
+}
+#endif
+
+uint64_t helper_fcvt_s_w(CPURISCVState *env, target_ulong rs1)
+{
+    return int32_to_float32((int32_t)rs1, &env->fp_status);
+}
+
+uint64_t helper_fcvt_s_wu(CPURISCVState *env, target_ulong rs1)
+{
+    return uint32_to_float32((uint32_t)rs1, &env->fp_status);
+}
+
+#if defined(TARGET_RISCV64)
+uint64_t helper_fcvt_s_l(CPURISCVState *env, uint64_t rs1)
+{
+    return int64_to_float32(rs1, &env->fp_status);
+}
+
+uint64_t helper_fcvt_s_lu(CPURISCVState *env, uint64_t rs1)
+{
+    return uint64_to_float32(rs1, &env->fp_status);
+}
+#endif
+
+target_ulong helper_fclass_s(uint64_t frs1)
+{
+    float32 f = frs1;
+    bool sign = float32_is_neg(f);
+
+    if (float32_is_infinity(f)) {
+        return sign ? 1 << 0 : 1 << 7;
+    } else if (float32_is_zero(f)) {
+        return sign ? 1 << 3 : 1 << 4;
+    } else if (float32_is_zero_or_denormal(f)) {
+        return sign ? 1 << 2 : 1 << 5;
+    } else if (float32_is_any_nan(f)) {
+        float_status s = { }; /* for snan_bit_is_one */
+        return float32_is_quiet_nan(f, &s) ? 1 << 9 : 1 << 8;
+    } else {
+        return sign ? 1 << 1 : 1 << 6;
+    }
+}
+
+uint64_t helper_fadd_d(CPURISCVState *env, uint64_t frs1, uint64_t frs2)
+{
+    return float64_add(frs1, frs2, &env->fp_status);
+}
+
+uint64_t helper_fsub_d(CPURISCVState *env, uint64_t frs1, uint64_t frs2)
+{
+    return float64_sub(frs1, frs2, &env->fp_status);
+}
+
+uint64_t helper_fmul_d(CPURISCVState *env, uint64_t frs1, uint64_t frs2)
+{
+    return float64_mul(frs1, frs2, &env->fp_status);
+}
+
+uint64_t helper_fdiv_d(CPURISCVState *env, uint64_t frs1, uint64_t frs2)
+{
+    return float64_div(frs1, frs2, &env->fp_status);
+}
+
+uint64_t helper_fmin_d(CPURISCVState *env, uint64_t frs1, uint64_t frs2)
+{
+    return float64_minnum(frs1, frs2, &env->fp_status);
+}
+
+uint64_t helper_fmax_d(CPURISCVState *env, uint64_t frs1, uint64_t frs2)
+{
+    return float64_maxnum(frs1, frs2, &env->fp_status);
+}
+
+uint64_t helper_fcvt_s_d(CPURISCVState *env, uint64_t rs1)
+{
+    rs1 = float64_to_float32(rs1, &env->fp_status);
+    return float32_maybe_silence_nan(rs1, &env->fp_status);
+}
+
+uint64_t helper_fcvt_d_s(CPURISCVState *env, uint64_t rs1)
+{
+    rs1 = float32_to_float64(rs1, &env->fp_status);
+    return float64_maybe_silence_nan(rs1, &env->fp_status);
+}
+
+uint64_t helper_fsqrt_d(CPURISCVState *env, uint64_t frs1)
+{
+    return float64_sqrt(frs1, &env->fp_status);
+}
+
+target_ulong helper_fle_d(CPURISCVState *env, uint64_t frs1, uint64_t frs2)
+{
+    return float64_le(frs1, frs2, &env->fp_status);
+}
+
+target_ulong helper_flt_d(CPURISCVState *env, uint64_t frs1, uint64_t frs2)
+{
+    return float64_lt(frs1, frs2, &env->fp_status);
+}
+
+target_ulong helper_feq_d(CPURISCVState *env, uint64_t frs1, uint64_t frs2)
+{
+    return float64_eq_quiet(frs1, frs2, &env->fp_status);
+}
+
+target_ulong helper_fcvt_w_d(CPURISCVState *env, uint64_t frs1)
+{
+    return float64_to_int32(frs1, &env->fp_status);
+}
+
+target_ulong helper_fcvt_wu_d(CPURISCVState *env, uint64_t frs1)
+{
+    return (int32_t)float64_to_uint32(frs1, &env->fp_status);
+}
+
+#if defined(TARGET_RISCV64)
+uint64_t helper_fcvt_l_d(CPURISCVState *env, uint64_t frs1)
+{
+    return float64_to_int64(frs1, &env->fp_status);
+}
+
+uint64_t helper_fcvt_lu_d(CPURISCVState *env, uint64_t frs1)
+{
+    return float64_to_uint64(frs1, &env->fp_status);
+}
+#endif
+
+uint64_t helper_fcvt_d_w(CPURISCVState *env, target_ulong rs1)
+{
+    return int32_to_float64((int32_t)rs1, &env->fp_status);
+}
+
+uint64_t helper_fcvt_d_wu(CPURISCVState *env, target_ulong rs1)
+{
+    return uint32_to_float64((uint32_t)rs1, &env->fp_status);
+}
+
+#if defined(TARGET_RISCV64)
+uint64_t helper_fcvt_d_l(CPURISCVState *env, uint64_t rs1)
+{
+    return int64_to_float64(rs1, &env->fp_status);
+}
+
+uint64_t helper_fcvt_d_lu(CPURISCVState *env, uint64_t rs1)
+{
+    return uint64_to_float64(rs1, &env->fp_status);
+}
+#endif
+
+target_ulong helper_fclass_d(uint64_t frs1)
+{
+    float64 f = frs1;
+    bool sign = float64_is_neg(f);
+
+    if (float64_is_infinity(f)) {
+        return sign ? 1 << 0 : 1 << 7;
+    } else if (float64_is_zero(f)) {
+        return sign ? 1 << 3 : 1 << 4;
+    } else if (float64_is_zero_or_denormal(f)) {
+        return sign ? 1 << 2 : 1 << 5;
+    } else if (float64_is_any_nan(f)) {
+        float_status s = { }; /* for snan_bit_is_one */
+        return float64_is_quiet_nan(f, &s) ? 1 << 9 : 1 << 8;
+    } else {
+        return sign ? 1 << 1 : 1 << 6;
+    }
+}