* [PATCH 0/6] support subsets of Float-Point in Integer Registers extensions
@ 2021-12-24 3:49 ` liweiwei
0 siblings, 0 replies; 30+ messages in thread
From: liweiwei @ 2021-12-24 3:49 UTC (permalink / raw)
To: palmer, alistair.francis, bin.meng, qemu-riscv, qemu-devel
Cc: wangjunqiang, lazyparser, ardxwe, liweiwei
This patchset implements RISC-V Float-Point in Integer Registers extensions(Version 1.0.0-rc), which includes Zfinx, Zdinx, Zhinx and Zhinxmin extension.
Specification:
https://github.com/riscv/riscv-zfinx/blob/main/zfinx-1.0.0-rc.pdf
The port is available here:
https://github.com/plctlab/plct-qemu/tree/plct-zfinx-upstream
To test this implementation, specify cpu argument with 'Zfinx =true,Zdinx=true,Zhinx=true,Zhinxmin=true' with 'g=false,f=false,d=false,Zfh=false,Zfhmin-false'
This implementation can pass gcc tests, ci result can be found in https://ci.rvperf.org/job/plct-qemu-zfinx-upstream/.
liweiwei (6):
target/riscv: add cfg properties for zfinx, zdinx and zhinx{min}
target/riscv: add support for unique fpr read/write with support for
zfinx
target/riscv: add support for zfinx
target/riscv: add support for zdinx
target/riscv: add support for zhinx/zhinxmin
target/riscv: expose zfinx, zdinx, zhinx{min} properties
roms/SLOF | 2 +-
target/riscv/cpu.c | 16 +
target/riscv/cpu.h | 4 +
target/riscv/fpu_helper.c | 120 ++++----
target/riscv/helper.h | 4 +-
target/riscv/insn_trans/trans_rvd.c.inc | 252 +++++++++++-----
target/riscv/insn_trans/trans_rvf.c.inc | 330 ++++++++++++++-------
target/riscv/insn_trans/trans_rvzfh.c.inc | 342 +++++++++++++++-------
target/riscv/internals.h | 12 +-
target/riscv/translate.c | 177 +++++++++++
10 files changed, 914 insertions(+), 345 deletions(-)
--
2.17.1
^ permalink raw reply [flat|nested] 30+ messages in thread
* [PATCH 1/6] target/riscv: add cfg properties for zfinx, zdinx and zhinx{min}
2021-12-24 3:49 ` liweiwei
@ 2021-12-24 3:49 ` liweiwei
-1 siblings, 0 replies; 30+ messages in thread
From: liweiwei @ 2021-12-24 3:49 UTC (permalink / raw)
To: palmer, alistair.francis, bin.meng, qemu-riscv, qemu-devel
Cc: wangjunqiang, liweiwei, lazyparser, ardxwe
Co-authored-by: ardxwe <ardxwe@gmail.com>
Signed-off-by: liweiwei <liweiwei@iscas.ac.cn>
Signed-off-by: wangjunqiang <wangjunqiang@iscas.ac.cn>
---
roms/SLOF | 2 +-
target/riscv/cpu.c | 12 ++++++++++++
target/riscv/cpu.h | 4 ++++
target/riscv/translate.c | 8 ++++++++
4 files changed, 25 insertions(+), 1 deletion(-)
diff --git a/roms/SLOF b/roms/SLOF
index a6906b024c..dd0dcaa1c1 160000
--- a/roms/SLOF
+++ b/roms/SLOF
@@ -1 +1 @@
-Subproject commit a6906b024c6cca5a86496f51eb4bfee3a0c36148
+Subproject commit dd0dcaa1c1085c159ddab709c7f274b3917be8bd
diff --git a/target/riscv/cpu.c b/target/riscv/cpu.c
index 6ef3314bce..a5fa14f2ac 100644
--- a/target/riscv/cpu.c
+++ b/target/riscv/cpu.c
@@ -491,6 +491,11 @@ static void riscv_cpu_realize(DeviceState *dev, Error **errp)
cpu->cfg.ext_d = true;
}
+ if (cpu->cfg.ext_zdinx || cpu->cfg.ext_zhinx ||
+ cpu->cfg.ext_zhinxmin ) {
+ cpu->cfg.ext_zfinx = true;
+ }
+
/* Set the ISA extensions, checks should have happened above */
if (cpu->cfg.ext_i) {
ext |= RVI;
@@ -565,6 +570,13 @@ static void riscv_cpu_realize(DeviceState *dev, Error **errp)
if (cpu->cfg.ext_j) {
ext |= RVJ;
}
+ if (cpu->cfg.ext_zfinx && ((ext & (RVF | RVD)) || cpu->cfg.ext_zfh ||
+ cpu->cfg.ext_zfhmin)) {
+ error_setg(errp,
+ "'Zfinx' cannot be supported together with 'F', 'D', 'Zfh',"
+ " 'Zfhmin'");
+ return;
+ }
set_misa(env, env->misa_mxl, ext);
}
diff --git a/target/riscv/cpu.h b/target/riscv/cpu.h
index dc10f27093..6fba31c5cd 100644
--- a/target/riscv/cpu.h
+++ b/target/riscv/cpu.h
@@ -315,8 +315,12 @@ struct RISCVCPU {
bool ext_counters;
bool ext_ifencei;
bool ext_icsr;
+ bool ext_zdinx;
bool ext_zfh;
bool ext_zfhmin;
+ bool ext_zfinx;
+ bool ext_zhinx;
+ bool ext_zhinxmin;
char *priv_spec;
char *user_spec;
diff --git a/target/riscv/translate.c b/target/riscv/translate.c
index 5df6c0d800..8b1cdacf50 100644
--- a/target/riscv/translate.c
+++ b/target/riscv/translate.c
@@ -76,8 +76,12 @@ typedef struct DisasContext {
RISCVMXL ol;
bool virt_enabled;
bool ext_ifencei;
+ bool ext_zdinx;
bool ext_zfh;
bool ext_zfhmin;
+ bool ext_zfinx;
+ bool ext_zhinx;
+ bool ext_zhinxmin;
bool hlsx;
/* vector extension */
bool vill;
@@ -703,8 +707,12 @@ static void riscv_tr_init_disas_context(DisasContextBase *dcbase, CPUState *cs)
ctx->misa_ext = env->misa_ext;
ctx->frm = -1; /* unknown rounding mode */
ctx->ext_ifencei = cpu->cfg.ext_ifencei;
+ ctx->ext_zdinx = cpu->cfg.ext_zdinx;
ctx->ext_zfh = cpu->cfg.ext_zfh;
ctx->ext_zfhmin = cpu->cfg.ext_zfhmin;
+ ctx->ext_zfinx = cpu->cfg.ext_zfinx;
+ ctx->ext_zhinx = cpu->cfg.ext_zhinx;
+ ctx->ext_zhinxmin = cpu->cfg.ext_zhinxmin;
ctx->vlen = cpu->cfg.vlen;
ctx->elen = cpu->cfg.elen;
ctx->mstatus_hs_fs = FIELD_EX32(tb_flags, TB_FLAGS, MSTATUS_HS_FS);
--
2.17.1
^ permalink raw reply related [flat|nested] 30+ messages in thread
* [PATCH 1/6] target/riscv: add cfg properties for zfinx, zdinx and zhinx{min}
@ 2021-12-24 3:49 ` liweiwei
0 siblings, 0 replies; 30+ messages in thread
From: liweiwei @ 2021-12-24 3:49 UTC (permalink / raw)
To: palmer, alistair.francis, bin.meng, qemu-riscv, qemu-devel
Cc: wangjunqiang, lazyparser, ardxwe, liweiwei
Co-authored-by: ardxwe <ardxwe@gmail.com>
Signed-off-by: liweiwei <liweiwei@iscas.ac.cn>
Signed-off-by: wangjunqiang <wangjunqiang@iscas.ac.cn>
---
roms/SLOF | 2 +-
target/riscv/cpu.c | 12 ++++++++++++
target/riscv/cpu.h | 4 ++++
target/riscv/translate.c | 8 ++++++++
4 files changed, 25 insertions(+), 1 deletion(-)
diff --git a/roms/SLOF b/roms/SLOF
index a6906b024c..dd0dcaa1c1 160000
--- a/roms/SLOF
+++ b/roms/SLOF
@@ -1 +1 @@
-Subproject commit a6906b024c6cca5a86496f51eb4bfee3a0c36148
+Subproject commit dd0dcaa1c1085c159ddab709c7f274b3917be8bd
diff --git a/target/riscv/cpu.c b/target/riscv/cpu.c
index 6ef3314bce..a5fa14f2ac 100644
--- a/target/riscv/cpu.c
+++ b/target/riscv/cpu.c
@@ -491,6 +491,11 @@ static void riscv_cpu_realize(DeviceState *dev, Error **errp)
cpu->cfg.ext_d = true;
}
+ if (cpu->cfg.ext_zdinx || cpu->cfg.ext_zhinx ||
+ cpu->cfg.ext_zhinxmin ) {
+ cpu->cfg.ext_zfinx = true;
+ }
+
/* Set the ISA extensions, checks should have happened above */
if (cpu->cfg.ext_i) {
ext |= RVI;
@@ -565,6 +570,13 @@ static void riscv_cpu_realize(DeviceState *dev, Error **errp)
if (cpu->cfg.ext_j) {
ext |= RVJ;
}
+ if (cpu->cfg.ext_zfinx && ((ext & (RVF | RVD)) || cpu->cfg.ext_zfh ||
+ cpu->cfg.ext_zfhmin)) {
+ error_setg(errp,
+ "'Zfinx' cannot be supported together with 'F', 'D', 'Zfh',"
+ " 'Zfhmin'");
+ return;
+ }
set_misa(env, env->misa_mxl, ext);
}
diff --git a/target/riscv/cpu.h b/target/riscv/cpu.h
index dc10f27093..6fba31c5cd 100644
--- a/target/riscv/cpu.h
+++ b/target/riscv/cpu.h
@@ -315,8 +315,12 @@ struct RISCVCPU {
bool ext_counters;
bool ext_ifencei;
bool ext_icsr;
+ bool ext_zdinx;
bool ext_zfh;
bool ext_zfhmin;
+ bool ext_zfinx;
+ bool ext_zhinx;
+ bool ext_zhinxmin;
char *priv_spec;
char *user_spec;
diff --git a/target/riscv/translate.c b/target/riscv/translate.c
index 5df6c0d800..8b1cdacf50 100644
--- a/target/riscv/translate.c
+++ b/target/riscv/translate.c
@@ -76,8 +76,12 @@ typedef struct DisasContext {
RISCVMXL ol;
bool virt_enabled;
bool ext_ifencei;
+ bool ext_zdinx;
bool ext_zfh;
bool ext_zfhmin;
+ bool ext_zfinx;
+ bool ext_zhinx;
+ bool ext_zhinxmin;
bool hlsx;
/* vector extension */
bool vill;
@@ -703,8 +707,12 @@ static void riscv_tr_init_disas_context(DisasContextBase *dcbase, CPUState *cs)
ctx->misa_ext = env->misa_ext;
ctx->frm = -1; /* unknown rounding mode */
ctx->ext_ifencei = cpu->cfg.ext_ifencei;
+ ctx->ext_zdinx = cpu->cfg.ext_zdinx;
ctx->ext_zfh = cpu->cfg.ext_zfh;
ctx->ext_zfhmin = cpu->cfg.ext_zfhmin;
+ ctx->ext_zfinx = cpu->cfg.ext_zfinx;
+ ctx->ext_zhinx = cpu->cfg.ext_zhinx;
+ ctx->ext_zhinxmin = cpu->cfg.ext_zhinxmin;
ctx->vlen = cpu->cfg.vlen;
ctx->elen = cpu->cfg.elen;
ctx->mstatus_hs_fs = FIELD_EX32(tb_flags, TB_FLAGS, MSTATUS_HS_FS);
--
2.17.1
^ permalink raw reply related [flat|nested] 30+ messages in thread
* Re: [PATCH 1/6] target/riscv: add cfg properties for zfinx, zdinx and zhinx{min}
2021-12-24 3:49 ` liweiwei
(?)
@ 2021-12-24 21:40 ` Richard Henderson
-1 siblings, 0 replies; 30+ messages in thread
From: Richard Henderson @ 2021-12-24 21:40 UTC (permalink / raw)
To: liweiwei, palmer, alistair.francis, bin.meng, qemu-riscv, qemu-devel
Cc: wangjunqiang, lazyparser, ardxwe
On 12/23/21 7:49 PM, liweiwei wrote:
> Co-authored-by: ardxwe<ardxwe@gmail.com>
> Signed-off-by: liweiwei<liweiwei@iscas.ac.cn>
> Signed-off-by: wangjunqiang<wangjunqiang@iscas.ac.cn>
> ---
> roms/SLOF | 2 +-
> target/riscv/cpu.c | 12 ++++++++++++
> target/riscv/cpu.h | 4 ++++
> target/riscv/translate.c | 8 ++++++++
> 4 files changed, 25 insertions(+), 1 deletion(-)
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
r~
^ permalink raw reply [flat|nested] 30+ messages in thread
* Re: [PATCH 1/6] target/riscv: add cfg properties for zfinx, zdinx and zhinx{min}
2021-12-24 3:49 ` liweiwei
@ 2022-01-03 22:47 ` Alistair Francis
-1 siblings, 0 replies; 30+ messages in thread
From: Alistair Francis @ 2022-01-03 22:47 UTC (permalink / raw)
To: liweiwei
Cc: lazyparser, open list:RISC-V, wangjunqiang, Bin Meng,
qemu-devel@nongnu.org Developers, ardxwe, Alistair Francis,
Palmer Dabbelt
On Fri, Dec 24, 2021 at 1:51 PM liweiwei <liweiwei@iscas.ac.cn> wrote:
>
> Co-authored-by: ardxwe <ardxwe@gmail.com>
> Signed-off-by: liweiwei <liweiwei@iscas.ac.cn>
> Signed-off-by: wangjunqiang <wangjunqiang@iscas.ac.cn>
> ---
> roms/SLOF | 2 +-
> target/riscv/cpu.c | 12 ++++++++++++
> target/riscv/cpu.h | 4 ++++
> target/riscv/translate.c | 8 ++++++++
> 4 files changed, 25 insertions(+), 1 deletion(-)
>
> diff --git a/roms/SLOF b/roms/SLOF
> index a6906b024c..dd0dcaa1c1 160000
> --- a/roms/SLOF
> +++ b/roms/SLOF
> @@ -1 +1 @@
> -Subproject commit a6906b024c6cca5a86496f51eb4bfee3a0c36148
> +Subproject commit dd0dcaa1c1085c159ddab709c7f274b3917be8bd
It looks like you accidentally changed a submodule.
Alistair
^ permalink raw reply [flat|nested] 30+ messages in thread
* Re: [PATCH 1/6] target/riscv: add cfg properties for zfinx, zdinx and zhinx{min}
@ 2022-01-03 22:47 ` Alistair Francis
0 siblings, 0 replies; 30+ messages in thread
From: Alistair Francis @ 2022-01-03 22:47 UTC (permalink / raw)
To: liweiwei
Cc: Palmer Dabbelt, Alistair Francis, Bin Meng, open list:RISC-V,
qemu-devel@nongnu.org Developers, wangjunqiang, lazyparser,
ardxwe
On Fri, Dec 24, 2021 at 1:51 PM liweiwei <liweiwei@iscas.ac.cn> wrote:
>
> Co-authored-by: ardxwe <ardxwe@gmail.com>
> Signed-off-by: liweiwei <liweiwei@iscas.ac.cn>
> Signed-off-by: wangjunqiang <wangjunqiang@iscas.ac.cn>
> ---
> roms/SLOF | 2 +-
> target/riscv/cpu.c | 12 ++++++++++++
> target/riscv/cpu.h | 4 ++++
> target/riscv/translate.c | 8 ++++++++
> 4 files changed, 25 insertions(+), 1 deletion(-)
>
> diff --git a/roms/SLOF b/roms/SLOF
> index a6906b024c..dd0dcaa1c1 160000
> --- a/roms/SLOF
> +++ b/roms/SLOF
> @@ -1 +1 @@
> -Subproject commit a6906b024c6cca5a86496f51eb4bfee3a0c36148
> +Subproject commit dd0dcaa1c1085c159ddab709c7f274b3917be8bd
It looks like you accidentally changed a submodule.
Alistair
^ permalink raw reply [flat|nested] 30+ messages in thread
* [PATCH 2/6] target/riscv: add support for unique fpr read/write with support for zfinx
2021-12-24 3:49 ` liweiwei
@ 2021-12-24 3:49 ` liweiwei
-1 siblings, 0 replies; 30+ messages in thread
From: liweiwei @ 2021-12-24 3:49 UTC (permalink / raw)
To: palmer, alistair.francis, bin.meng, qemu-riscv, qemu-devel
Cc: wangjunqiang, liweiwei, lazyparser, ardxwe
Co-authored-by: ardxwe <ardxwe@gmail.com>
Signed-off-by: liweiwei <liweiwei@iscas.ac.cn>
Signed-off-by: wangjunqiang <wangjunqiang@iscas.ac.cn>
---
target/riscv/translate.c | 169 +++++++++++++++++++++++++++++++++++++++
1 file changed, 169 insertions(+)
diff --git a/target/riscv/translate.c b/target/riscv/translate.c
index 8b1cdacf50..bac42e60bd 100644
--- a/target/riscv/translate.c
+++ b/target/riscv/translate.c
@@ -104,10 +104,13 @@ typedef struct DisasContext {
target_ulong vstart;
bool vl_eq_vlmax;
uint8_t ntemp;
+ uint8_t nftemp;
CPUState *cs;
TCGv zero;
/* Space for 3 operands plus 1 extra for address computation. */
TCGv temp[4];
+ /* Space for 4 float point operands */
+ TCGv_i64 ftemp[4];
/* PointerMasking extension */
bool pm_enabled;
TCGv pm_mask;
@@ -295,6 +298,165 @@ static void gen_set_gpr(DisasContext *ctx, int reg_num, TCGv t)
}
}
+static TCGv_i64 ftemp_new(DisasContext *ctx)
+{
+ assert(ctx->nftemp < ARRAY_SIZE(ctx->ftemp));
+ return ctx->ftemp[ctx->nftemp++] = tcg_temp_new_i64();
+}
+
+static TCGv_i64 get_fpr_hs(DisasContext *ctx, int reg_num)
+{
+ if (ctx->ext_zfinx) {
+ switch (get_ol(ctx)) {
+ case MXL_RV32:
+#ifdef TARGET_RISCV32
+ {
+ TCGv_i64 t = ftemp_new(ctx);
+ if (reg_num == 0) {
+ tcg_gen_concat_i32_i64(t, ctx->zero, ctx->zero);
+ } else {
+ tcg_gen_concat_i32_i64(t, cpu_gpr[reg_num], ctx->zero);
+ }
+ return t;
+ }
+#else
+ /* fall through */
+ case MXL_RV64:
+ if (reg_num == 0) {
+ return ctx->zero;
+ } else {
+ return cpu_gpr[reg_num];
+ }
+#endif
+ default:
+ g_assert_not_reached();
+ }
+ } else {
+ return cpu_fpr[reg_num];
+ }
+}
+
+static TCGv_i64 get_fpr_d(DisasContext *ctx, int reg_num)
+{
+ if (ctx->ext_zfinx) {
+ switch (get_ol(ctx)) {
+ case MXL_RV32:
+ if (reg_num & 1) {
+ gen_exception_illegal(ctx);
+ return NULL;
+ } else {
+#ifdef TARGET_RISCV32
+ TCGv_i64 t = ftemp_new(ctx);
+ if (reg_num == 0) {
+ tcg_gen_concat_i32_i64(t, ctx->zero, ctx->zero);
+ } else {
+ tcg_gen_concat_i32_i64(t, cpu_gpr[reg_num], cpu_gpr[reg_num + 1]);
+ }
+ return t;
+ }
+#else
+ if (reg_num == 0) {
+ return ctx->zero;
+ } else {
+ TCGv_i64 t = ftemp_new(ctx);
+ tcg_gen_deposit_i64(t, cpu_gpr[reg_num], cpu_gpr[reg_num + 1], 32, 32);
+ return t;
+ }
+ }
+ case MXL_RV64:
+ if (reg_num == 0) {
+ return ctx->zero;
+ } else {
+ return cpu_gpr[reg_num];
+ }
+#endif
+ default:
+ g_assert_not_reached();
+ }
+ } else {
+ return cpu_fpr[reg_num];
+ }
+}
+
+static TCGv_i64 dest_fpr(DisasContext *ctx, int reg_num)
+{
+ if (ctx->ext_zfinx) {
+ switch (get_ol(ctx)) {
+ case MXL_RV32:
+ return ftemp_new(ctx);
+#ifdef TARGET_RISCV64
+ case MXL_RV64:
+ if (reg_num == 0) {
+ return ftemp_new(ctx);
+ } else {
+ return cpu_gpr[reg_num];
+ }
+#endif
+ default:
+ g_assert_not_reached();
+ }
+ } else {
+ return cpu_fpr[reg_num];
+ }
+}
+
+static void gen_set_fpr_hs(DisasContext *ctx, int reg_num, TCGv_i64 t)
+{
+ if (ctx->ext_zfinx) {
+ if (reg_num != 0) {
+ switch (get_ol(ctx)) {
+ case MXL_RV32:
+#ifdef TARGET_RISCV32
+ tcg_gen_extrl_i64_i32(cpu_gpr[reg_num], t);
+ break;
+#else
+ tcg_gen_ext32s_i64(cpu_gpr[reg_num], t);
+ break;
+ case MXL_RV64:
+ tcg_gen_mov_i64(cpu_gpr[reg_num], t);
+ break;
+#endif
+ default:
+ g_assert_not_reached();
+ }
+ }
+ } else {
+ tcg_gen_mov_i64(cpu_fpr[reg_num], t);
+ }
+}
+
+static void gen_set_fpr_d(DisasContext *ctx, int reg_num, TCGv_i64 t)
+{
+ if (ctx->ext_zfinx) {
+ if (reg_num != 0) {
+ switch (get_ol(ctx)) {
+ case MXL_RV32:
+ if (reg_num & 1) {
+ gen_exception_illegal(ctx);
+ } else {
+#ifdef TARGET_RISCV32
+ tcg_gen_extrl_i64_i32(cpu_gpr[reg_num], t);
+ tcg_gen_extrh_i64_i32(cpu_gpr[reg_num + 1], t);
+ }
+ break;
+#else
+ tcg_gen_ext32s_i64(cpu_gpr[reg_num], t);
+ tcg_gen_sari_i64(cpu_gpr[reg_num + 1], t, 32);
+ }
+ break;
+ case MXL_RV64:
+ tcg_gen_mov_i64(cpu_gpr[reg_num], t);
+ break;
+#endif
+ default:
+ g_assert_not_reached();
+ }
+ }
+ } else {
+ tcg_gen_mov_i64(cpu_fpr[reg_num], t);
+ }
+}
+
static void gen_jal(DisasContext *ctx, int rd, target_ulong imm)
{
target_ulong next_pc;
@@ -727,6 +889,8 @@ static void riscv_tr_init_disas_context(DisasContextBase *dcbase, CPUState *cs)
ctx->cs = cs;
ctx->ntemp = 0;
memset(ctx->temp, 0, sizeof(ctx->temp));
+ ctx->nftemp = 0;
+ memset(ctx->ftemp, 0, sizeof(ctx->ftemp));
ctx->pm_enabled = FIELD_EX32(tb_flags, TB_FLAGS, PM_ENABLED);
int priv = tb_flags & TB_FLAGS_PRIV_MMU_MASK;
ctx->pm_mask = pm_mask[priv];
@@ -761,6 +925,11 @@ static void riscv_tr_translate_insn(DisasContextBase *dcbase, CPUState *cpu)
ctx->temp[i] = NULL;
}
ctx->ntemp = 0;
+ for (int i = ctx->nftemp - 1; i >= 0; --i) {
+ tcg_temp_free_i64(ctx->ftemp[i]);
+ ctx->ftemp[i] = NULL;
+ }
+ ctx->nftemp = 0;
if (ctx->base.is_jmp == DISAS_NEXT) {
target_ulong page_start;
--
2.17.1
^ permalink raw reply related [flat|nested] 30+ messages in thread
* [PATCH 2/6] target/riscv: add support for unique fpr read/write with support for zfinx
@ 2021-12-24 3:49 ` liweiwei
0 siblings, 0 replies; 30+ messages in thread
From: liweiwei @ 2021-12-24 3:49 UTC (permalink / raw)
To: palmer, alistair.francis, bin.meng, qemu-riscv, qemu-devel
Cc: wangjunqiang, lazyparser, ardxwe, liweiwei
Co-authored-by: ardxwe <ardxwe@gmail.com>
Signed-off-by: liweiwei <liweiwei@iscas.ac.cn>
Signed-off-by: wangjunqiang <wangjunqiang@iscas.ac.cn>
---
target/riscv/translate.c | 169 +++++++++++++++++++++++++++++++++++++++
1 file changed, 169 insertions(+)
diff --git a/target/riscv/translate.c b/target/riscv/translate.c
index 8b1cdacf50..bac42e60bd 100644
--- a/target/riscv/translate.c
+++ b/target/riscv/translate.c
@@ -104,10 +104,13 @@ typedef struct DisasContext {
target_ulong vstart;
bool vl_eq_vlmax;
uint8_t ntemp;
+ uint8_t nftemp;
CPUState *cs;
TCGv zero;
/* Space for 3 operands plus 1 extra for address computation. */
TCGv temp[4];
+ /* Space for 4 float point operands */
+ TCGv_i64 ftemp[4];
/* PointerMasking extension */
bool pm_enabled;
TCGv pm_mask;
@@ -295,6 +298,165 @@ static void gen_set_gpr(DisasContext *ctx, int reg_num, TCGv t)
}
}
+static TCGv_i64 ftemp_new(DisasContext *ctx)
+{
+ assert(ctx->nftemp < ARRAY_SIZE(ctx->ftemp));
+ return ctx->ftemp[ctx->nftemp++] = tcg_temp_new_i64();
+}
+
+static TCGv_i64 get_fpr_hs(DisasContext *ctx, int reg_num)
+{
+ if (ctx->ext_zfinx) {
+ switch (get_ol(ctx)) {
+ case MXL_RV32:
+#ifdef TARGET_RISCV32
+ {
+ TCGv_i64 t = ftemp_new(ctx);
+ if (reg_num == 0) {
+ tcg_gen_concat_i32_i64(t, ctx->zero, ctx->zero);
+ } else {
+ tcg_gen_concat_i32_i64(t, cpu_gpr[reg_num], ctx->zero);
+ }
+ return t;
+ }
+#else
+ /* fall through */
+ case MXL_RV64:
+ if (reg_num == 0) {
+ return ctx->zero;
+ } else {
+ return cpu_gpr[reg_num];
+ }
+#endif
+ default:
+ g_assert_not_reached();
+ }
+ } else {
+ return cpu_fpr[reg_num];
+ }
+}
+
+static TCGv_i64 get_fpr_d(DisasContext *ctx, int reg_num)
+{
+ if (ctx->ext_zfinx) {
+ switch (get_ol(ctx)) {
+ case MXL_RV32:
+ if (reg_num & 1) {
+ gen_exception_illegal(ctx);
+ return NULL;
+ } else {
+#ifdef TARGET_RISCV32
+ TCGv_i64 t = ftemp_new(ctx);
+ if (reg_num == 0) {
+ tcg_gen_concat_i32_i64(t, ctx->zero, ctx->zero);
+ } else {
+ tcg_gen_concat_i32_i64(t, cpu_gpr[reg_num], cpu_gpr[reg_num + 1]);
+ }
+ return t;
+ }
+#else
+ if (reg_num == 0) {
+ return ctx->zero;
+ } else {
+ TCGv_i64 t = ftemp_new(ctx);
+ tcg_gen_deposit_i64(t, cpu_gpr[reg_num], cpu_gpr[reg_num + 1], 32, 32);
+ return t;
+ }
+ }
+ case MXL_RV64:
+ if (reg_num == 0) {
+ return ctx->zero;
+ } else {
+ return cpu_gpr[reg_num];
+ }
+#endif
+ default:
+ g_assert_not_reached();
+ }
+ } else {
+ return cpu_fpr[reg_num];
+ }
+}
+
+static TCGv_i64 dest_fpr(DisasContext *ctx, int reg_num)
+{
+ if (ctx->ext_zfinx) {
+ switch (get_ol(ctx)) {
+ case MXL_RV32:
+ return ftemp_new(ctx);
+#ifdef TARGET_RISCV64
+ case MXL_RV64:
+ if (reg_num == 0) {
+ return ftemp_new(ctx);
+ } else {
+ return cpu_gpr[reg_num];
+ }
+#endif
+ default:
+ g_assert_not_reached();
+ }
+ } else {
+ return cpu_fpr[reg_num];
+ }
+}
+
+static void gen_set_fpr_hs(DisasContext *ctx, int reg_num, TCGv_i64 t)
+{
+ if (ctx->ext_zfinx) {
+ if (reg_num != 0) {
+ switch (get_ol(ctx)) {
+ case MXL_RV32:
+#ifdef TARGET_RISCV32
+ tcg_gen_extrl_i64_i32(cpu_gpr[reg_num], t);
+ break;
+#else
+ tcg_gen_ext32s_i64(cpu_gpr[reg_num], t);
+ break;
+ case MXL_RV64:
+ tcg_gen_mov_i64(cpu_gpr[reg_num], t);
+ break;
+#endif
+ default:
+ g_assert_not_reached();
+ }
+ }
+ } else {
+ tcg_gen_mov_i64(cpu_fpr[reg_num], t);
+ }
+}
+
+static void gen_set_fpr_d(DisasContext *ctx, int reg_num, TCGv_i64 t)
+{
+ if (ctx->ext_zfinx) {
+ if (reg_num != 0) {
+ switch (get_ol(ctx)) {
+ case MXL_RV32:
+ if (reg_num & 1) {
+ gen_exception_illegal(ctx);
+ } else {
+#ifdef TARGET_RISCV32
+ tcg_gen_extrl_i64_i32(cpu_gpr[reg_num], t);
+ tcg_gen_extrh_i64_i32(cpu_gpr[reg_num + 1], t);
+ }
+ break;
+#else
+ tcg_gen_ext32s_i64(cpu_gpr[reg_num], t);
+ tcg_gen_sari_i64(cpu_gpr[reg_num + 1], t, 32);
+ }
+ break;
+ case MXL_RV64:
+ tcg_gen_mov_i64(cpu_gpr[reg_num], t);
+ break;
+#endif
+ default:
+ g_assert_not_reached();
+ }
+ }
+ } else {
+ tcg_gen_mov_i64(cpu_fpr[reg_num], t);
+ }
+}
+
static void gen_jal(DisasContext *ctx, int rd, target_ulong imm)
{
target_ulong next_pc;
@@ -727,6 +889,8 @@ static void riscv_tr_init_disas_context(DisasContextBase *dcbase, CPUState *cs)
ctx->cs = cs;
ctx->ntemp = 0;
memset(ctx->temp, 0, sizeof(ctx->temp));
+ ctx->nftemp = 0;
+ memset(ctx->ftemp, 0, sizeof(ctx->ftemp));
ctx->pm_enabled = FIELD_EX32(tb_flags, TB_FLAGS, PM_ENABLED);
int priv = tb_flags & TB_FLAGS_PRIV_MMU_MASK;
ctx->pm_mask = pm_mask[priv];
@@ -761,6 +925,11 @@ static void riscv_tr_translate_insn(DisasContextBase *dcbase, CPUState *cpu)
ctx->temp[i] = NULL;
}
ctx->ntemp = 0;
+ for (int i = ctx->nftemp - 1; i >= 0; --i) {
+ tcg_temp_free_i64(ctx->ftemp[i]);
+ ctx->ftemp[i] = NULL;
+ }
+ ctx->nftemp = 0;
if (ctx->base.is_jmp == DISAS_NEXT) {
target_ulong page_start;
--
2.17.1
^ permalink raw reply related [flat|nested] 30+ messages in thread
* Re: [PATCH 2/6] target/riscv: add support for unique fpr read/write with support for zfinx
2021-12-24 3:49 ` liweiwei
(?)
@ 2021-12-24 22:00 ` Richard Henderson
2021-12-25 3:13 ` liweiwei
-1 siblings, 1 reply; 30+ messages in thread
From: Richard Henderson @ 2021-12-24 22:00 UTC (permalink / raw)
To: liweiwei, palmer, alistair.francis, bin.meng, qemu-riscv, qemu-devel
Cc: wangjunqiang, lazyparser, ardxwe
On 12/23/21 7:49 PM, liweiwei wrote:
> +static TCGv_i64 get_fpr_hs(DisasContext *ctx, int reg_num)
> +{
> + if (ctx->ext_zfinx) {
> + switch (get_ol(ctx)) {
> + case MXL_RV32:
> +#ifdef TARGET_RISCV32
> + if (reg_num == 0) {
> + tcg_gen_concat_i32_i64(t, ctx->zero, ctx->zero);
return tcg_constant_i64(0);
You could hoist this above the switch.
> + tcg_gen_concat_i32_i64(t, cpu_gpr[reg_num], ctx->zero);
tcg_gen_extu_i32_i64 -- though I would think a signed extraction would be more natural, as
compared with the signed value you'll get from the RV64 case.
> + case MXL_RV64:
> + if (reg_num == 0) {
> + return ctx->zero;
> + } else {
> + return cpu_gpr[reg_num];
> + }
> +#endif
> + default:
> + g_assert_not_reached();
> + }
> + } else {
> + return cpu_fpr[reg_num];
> + }
It is tempting to reverse the sense of the zfinx if, to eliminate this case first, and
keep the indentation level down.
> +static TCGv_i64 get_fpr_d(DisasContext *ctx, int reg_num)
> +{
> + if (ctx->ext_zfinx) {
> + switch (get_ol(ctx)) {
> + case MXL_RV32:
> + if (reg_num & 1) {
> + gen_exception_illegal(ctx);
> + return NULL;
Not keen on checking this here. It should be done earlier.
> + } else {
> +#ifdef TARGET_RISCV32
> + TCGv_i64 t = ftemp_new(ctx);
> + if (reg_num == 0) {
> + tcg_gen_concat_i32_i64(t, ctx->zero, ctx->zero);
> + } else {
> + tcg_gen_concat_i32_i64(t, cpu_gpr[reg_num], cpu_gpr[reg_num + 1]);
> + }
> + return t;
> + }
> +#else
> + if (reg_num == 0) {
> + return ctx->zero;
> + } else {
> + TCGv_i64 t = ftemp_new(ctx);
> + tcg_gen_deposit_i64(t, cpu_gpr[reg_num], cpu_gpr[reg_num + 1], 32, 32);
> + return t;
> + }
> + }
> + case MXL_RV64:
> + if (reg_num == 0) {
> + return ctx->zero;
> + } else {
> + return cpu_gpr[reg_num];
> + }
> +#endif
> + default:
> + g_assert_not_reached();
> + }
> + } else {
> + return cpu_fpr[reg_num];
> + }
Very similar comments to above.
> +static void gen_set_fpr_d(DisasContext *ctx, int reg_num, TCGv_i64 t)
> +{
> + if (ctx->ext_zfinx) {
> + if (reg_num != 0) {
> + switch (get_ol(ctx)) {
> + case MXL_RV32:
> + if (reg_num & 1) {
> + gen_exception_illegal(ctx);
This is *far* too late to diagnose illegal insn. You've already modified global state in
the FPSCR, or have taken an fpu exception in fpu_helper.c.
> + } else {
> +#ifdef TARGET_RISCV32
> + tcg_gen_extrl_i64_i32(cpu_gpr[reg_num], t);
> + tcg_gen_extrh_i64_i32(cpu_gpr[reg_num + 1], t);
> + }
tcg_gen_extr_i64_i32 does both at once.
Never split braces around ifdefs like this.
r~
^ permalink raw reply [flat|nested] 30+ messages in thread
* Re: [PATCH 2/6] target/riscv: add support for unique fpr read/write with support for zfinx
2021-12-24 22:00 ` Richard Henderson
@ 2021-12-25 3:13 ` liweiwei
2021-12-25 22:00 ` Richard Henderson
0 siblings, 1 reply; 30+ messages in thread
From: liweiwei @ 2021-12-25 3:13 UTC (permalink / raw)
To: Richard Henderson, palmer, alistair.francis, bin.meng,
qemu-riscv, qemu-devel
Cc: wangjunqiang, lazyparser, ardxwe
Thanks for your comments.
在 2021/12/25 上午6:00, Richard Henderson 写道:
> On 12/23/21 7:49 PM, liweiwei wrote:
>> +static TCGv_i64 get_fpr_hs(DisasContext *ctx, int reg_num)
>> +{
>> + if (ctx->ext_zfinx) {
>> + switch (get_ol(ctx)) {
>> + case MXL_RV32:
>> +#ifdef TARGET_RISCV32
>> + if (reg_num == 0) {
>> + tcg_gen_concat_i32_i64(t, ctx->zero, ctx->zero);
>
> return tcg_constant_i64(0);
> You could hoist this above the switch.
>
OK.
>> + tcg_gen_concat_i32_i64(t, cpu_gpr[reg_num], ctx->zero);
> tcg_gen_extu_i32_i64 -- though I would think a signed extraction would
> be more natural, as compared with the signed value you'll get from the
> RV64 case.
>
In RV64 case, this should be nan-boxing value( upper bits are all
ones). However, zfinx will not check nan-boxing of source, the upper 32
bits have no effect on the final result. So I think both zero-extended
or sign-extended are OK.
>> + case MXL_RV64:
>> + if (reg_num == 0) {
>> + return ctx->zero;
>> + } else {
>> + return cpu_gpr[reg_num];
>> + }
>> +#endif
>> + default:
>> + g_assert_not_reached();
>> + }
>
>> + } else {
>> + return cpu_fpr[reg_num];
>> + }
>
> It is tempting to reverse the sense of the zfinx if, to eliminate this
> case first, and keep the indentation level down.
>
OK I'll update this.
>
>> +static TCGv_i64 get_fpr_d(DisasContext *ctx, int reg_num)
>> +{
>> + if (ctx->ext_zfinx) {
>> + switch (get_ol(ctx)) {
>> + case MXL_RV32:
>> + if (reg_num & 1) {
>> + gen_exception_illegal(ctx);
>> + return NULL;
>
> Not keen on checking this here. It should be done earlier.
>
OK. I'll put this check into the trans_* function
>> + } else {
>> +#ifdef TARGET_RISCV32
>> + TCGv_i64 t = ftemp_new(ctx);
>> + if (reg_num == 0) {
>> + tcg_gen_concat_i32_i64(t, ctx->zero, ctx->zero);
>> + } else {
>> + tcg_gen_concat_i32_i64(t, cpu_gpr[reg_num],
>> cpu_gpr[reg_num + 1]);
>> + }
>> + return t;
>> + }
>> +#else
>> + if (reg_num == 0) {
>> + return ctx->zero;
>> + } else {
>> + TCGv_i64 t = ftemp_new(ctx);
>> + tcg_gen_deposit_i64(t, cpu_gpr[reg_num],
>> cpu_gpr[reg_num + 1], 32, 32);
>> + return t;
>> + }
>> + }
>> + case MXL_RV64:
>> + if (reg_num == 0) {
>> + return ctx->zero;
>> + } else {
>> + return cpu_gpr[reg_num];
>> + }
>> +#endif
>> + default:
>> + g_assert_not_reached();
>> + }
>> + } else {
>> + return cpu_fpr[reg_num];
>> + }
>
> Very similar comments to above.
>
>> +static void gen_set_fpr_d(DisasContext *ctx, int reg_num, TCGv_i64 t)
>> +{
>> + if (ctx->ext_zfinx) {
>> + if (reg_num != 0) {
>> + switch (get_ol(ctx)) {
>> + case MXL_RV32:
>> + if (reg_num & 1) {
>> + gen_exception_illegal(ctx);
>
> This is *far* too late to diagnose illegal insn. You've already
> modified global state in the FPSCR, or have taken an fpu exception in
> fpu_helper.c.
OK. I'll put this check into the trans_* function too.
>
>> + } else {
>> +#ifdef TARGET_RISCV32
>> + tcg_gen_extrl_i64_i32(cpu_gpr[reg_num], t);
>> + tcg_gen_extrh_i64_i32(cpu_gpr[reg_num + 1], t);
>> + }
>
> tcg_gen_extr_i64_i32 does both at once.
> Never split braces around ifdefs like this.
OK. I'll update this.
>
>
> r~
^ permalink raw reply [flat|nested] 30+ messages in thread
* Re: [PATCH 2/6] target/riscv: add support for unique fpr read/write with support for zfinx
2021-12-25 3:13 ` liweiwei
@ 2021-12-25 22:00 ` Richard Henderson
2021-12-26 1:42 ` liweiwei
0 siblings, 1 reply; 30+ messages in thread
From: Richard Henderson @ 2021-12-25 22:00 UTC (permalink / raw)
To: liweiwei, palmer, alistair.francis, bin.meng, qemu-riscv, qemu-devel
Cc: wangjunqiang, lazyparser, ardxwe
On 12/24/21 7:13 PM, liweiwei wrote:
> In RV64 case, this should be nan-boxing value( upper bits are all ones). However, zfinx
> will not check nan-boxing of source, the upper 32 bits have no effect on the final result.
> So I think both zero-extended or sign-extended are OK.
There is no nanboxing in zfinx at all -- values are sign-extended.
r~
^ permalink raw reply [flat|nested] 30+ messages in thread
* Re: [PATCH 2/6] target/riscv: add support for unique fpr read/write with support for zfinx
2021-12-25 22:00 ` Richard Henderson
@ 2021-12-26 1:42 ` liweiwei
2021-12-26 1:54 ` liweiwei
2021-12-26 3:48 ` Richard Henderson
0 siblings, 2 replies; 30+ messages in thread
From: liweiwei @ 2021-12-26 1:42 UTC (permalink / raw)
To: Richard Henderson, palmer, alistair.francis, bin.meng,
qemu-riscv, qemu-devel
Cc: wangjunqiang, lazyparser, ardxwe
[-- Attachment #1: Type: text/plain, Size: 722 bytes --]
Sorry. In the old spec(version 0.41), nanboxing is not totally disabled,
but "NaN-boxing is limited to |XLEN| bits, not |FLEN| bits". Taking
misa.mxl into acount, if misa.mxl is RV32, and maximum is RV64, this
should be sign-extended. Is there any other new update for nanboxing to
the spec?
在 2021/12/26 上午6:00, Richard Henderson 写道:
> On 12/24/21 7:13 PM, liweiwei wrote:
>> In RV64 case, this should be nan-boxing value( upper bits are all
>> ones). However, zfinx will not check nan-boxing of source, the upper
>> 32 bits have no effect on the final result. So I think both
>> zero-extended or sign-extended are OK.
>
> There is no nanboxing in zfinx at all -- values are sign-extended.
>
>
> r~
[-- Attachment #2: Type: text/html, Size: 1231 bytes --]
^ permalink raw reply [flat|nested] 30+ messages in thread
* Re: [PATCH 2/6] target/riscv: add support for unique fpr read/write with support for zfinx
2021-12-26 1:42 ` liweiwei
@ 2021-12-26 1:54 ` liweiwei
2021-12-26 3:48 ` Richard Henderson
1 sibling, 0 replies; 30+ messages in thread
From: liweiwei @ 2021-12-26 1:54 UTC (permalink / raw)
To: Richard Henderson, palmer, alistair.francis, bin.meng,
qemu-riscv, qemu-devel
Cc: wangjunqiang, lazyparser, ardxwe
[-- Attachment #1: Type: text/plain, Size: 999 bytes --]
OK. It have been changed to sign-extended in the "processing of Narrower
Values" section of the new spec(version 1.0.0.rc) . I'll fix this and
update other nan-boxing processing in current implementation.
在 2021/12/26 上午9:42, liweiwei 写道:
>
> Sorry. In the old spec(version 0.41), nanboxing is not totally
> disabled, but "NaN-boxing is limited to |XLEN| bits, not |FLEN| bits".
> Taking misa.mxl into acount, if misa.mxl is RV32, and maximum is RV64,
> this should be sign-extended. Is there any other new update for
> nanboxing to the spec?
>
> 在 2021/12/26 上午6:00, Richard Henderson 写道:
>> On 12/24/21 7:13 PM, liweiwei wrote:
>>> In RV64 case, this should be nan-boxing value( upper bits are all
>>> ones). However, zfinx will not check nan-boxing of source, the
>>> upper 32 bits have no effect on the final result. So I think both
>>> zero-extended or sign-extended are OK.
>>
>> There is no nanboxing in zfinx at all -- values are sign-extended.
>>
>>
>> r~
[-- Attachment #2: Type: text/html, Size: 1786 bytes --]
^ permalink raw reply [flat|nested] 30+ messages in thread
* Re: [PATCH 2/6] target/riscv: add support for unique fpr read/write with support for zfinx
2021-12-26 1:42 ` liweiwei
2021-12-26 1:54 ` liweiwei
@ 2021-12-26 3:48 ` Richard Henderson
1 sibling, 0 replies; 30+ messages in thread
From: Richard Henderson @ 2021-12-26 3:48 UTC (permalink / raw)
To: liweiwei, palmer, alistair.francis, bin.meng, qemu-riscv, qemu-devel
Cc: wangjunqiang, lazyparser, ardxwe
On 12/25/21 5:42 PM, liweiwei wrote:
> Sorry. In the old spec(version 0.41), nanboxing is not totally disabled, but "NaN-boxing
> is limited to |XLEN| bits, not |FLEN| bits". Taking misa.mxl into acount, if misa.mxl is
> RV32, and maximum is RV64, this should be sign-extended. Is there any other new update for
> nanboxing to the spec?
Yes, in 1.0.0-rc, it's all gone.
r~
^ permalink raw reply [flat|nested] 30+ messages in thread
* [PATCH 3/6] target/riscv: add support for zfinx
2021-12-24 3:49 ` liweiwei
@ 2021-12-24 3:49 ` liweiwei
-1 siblings, 0 replies; 30+ messages in thread
From: liweiwei @ 2021-12-24 3:49 UTC (permalink / raw)
To: palmer, alistair.francis, bin.meng, qemu-riscv, qemu-devel
Cc: wangjunqiang, liweiwei, lazyparser, ardxwe
- update extension check REQUIRE_ZFINX_OR_F
- update single float point register read/write
- disable nanbox_s check
Co-authored-by: ardxwe <ardxwe@gmail.com>
Signed-off-by: liweiwei <liweiwei@iscas.ac.cn>
Signed-off-by: wangjunqiang <wangjunqiang@iscas.ac.cn>
---
target/riscv/fpu_helper.c | 60 ++---
target/riscv/helper.h | 2 +-
target/riscv/insn_trans/trans_rvf.c.inc | 330 ++++++++++++++++--------
target/riscv/internals.h | 6 +-
4 files changed, 263 insertions(+), 135 deletions(-)
diff --git a/target/riscv/fpu_helper.c b/target/riscv/fpu_helper.c
index 4a5982d594..0895fa3c7e 100644
--- a/target/riscv/fpu_helper.c
+++ b/target/riscv/fpu_helper.c
@@ -98,9 +98,9 @@ static uint64_t do_fmadd_h(CPURISCVState *env, uint64_t rs1, uint64_t rs2,
static uint64_t do_fmadd_s(CPURISCVState *env, uint64_t rs1, uint64_t rs2,
uint64_t rs3, int flags)
{
- float32 frs1 = check_nanbox_s(rs1);
- float32 frs2 = check_nanbox_s(rs2);
- float32 frs3 = check_nanbox_s(rs3);
+ float32 frs1 = check_nanbox_s(env, rs1);
+ float32 frs2 = check_nanbox_s(env, rs2);
+ float32 frs3 = check_nanbox_s(env, rs3);
return nanbox_s(float32_muladd(frs1, frs2, frs3, flags, &env->fp_status));
}
@@ -183,36 +183,36 @@ uint64_t helper_fnmadd_h(CPURISCVState *env, uint64_t frs1, uint64_t frs2,
uint64_t helper_fadd_s(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
{
- float32 frs1 = check_nanbox_s(rs1);
- float32 frs2 = check_nanbox_s(rs2);
+ float32 frs1 = check_nanbox_s(env, rs1);
+ float32 frs2 = check_nanbox_s(env, rs2);
return nanbox_s(float32_add(frs1, frs2, &env->fp_status));
}
uint64_t helper_fsub_s(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
{
- float32 frs1 = check_nanbox_s(rs1);
- float32 frs2 = check_nanbox_s(rs2);
+ float32 frs1 = check_nanbox_s(env, rs1);
+ float32 frs2 = check_nanbox_s(env, rs2);
return nanbox_s(float32_sub(frs1, frs2, &env->fp_status));
}
uint64_t helper_fmul_s(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
{
- float32 frs1 = check_nanbox_s(rs1);
- float32 frs2 = check_nanbox_s(rs2);
+ float32 frs1 = check_nanbox_s(env, rs1);
+ float32 frs2 = check_nanbox_s(env, rs2);
return nanbox_s(float32_mul(frs1, frs2, &env->fp_status));
}
uint64_t helper_fdiv_s(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
{
- float32 frs1 = check_nanbox_s(rs1);
- float32 frs2 = check_nanbox_s(rs2);
+ float32 frs1 = check_nanbox_s(env, rs1);
+ float32 frs2 = check_nanbox_s(env, rs2);
return nanbox_s(float32_div(frs1, frs2, &env->fp_status));
}
uint64_t helper_fmin_s(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
{
- float32 frs1 = check_nanbox_s(rs1);
- float32 frs2 = check_nanbox_s(rs2);
+ float32 frs1 = check_nanbox_s(env, rs1);
+ float32 frs2 = check_nanbox_s(env, rs2);
return nanbox_s(env->priv_ver < PRIV_VERSION_1_11_0 ?
float32_minnum(frs1, frs2, &env->fp_status) :
float32_minimum_number(frs1, frs2, &env->fp_status));
@@ -220,8 +220,8 @@ uint64_t helper_fmin_s(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
uint64_t helper_fmax_s(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
{
- float32 frs1 = check_nanbox_s(rs1);
- float32 frs2 = check_nanbox_s(rs2);
+ float32 frs1 = check_nanbox_s(env, rs1);
+ float32 frs2 = check_nanbox_s(env, rs2);
return nanbox_s(env->priv_ver < PRIV_VERSION_1_11_0 ?
float32_maxnum(frs1, frs2, &env->fp_status) :
float32_maximum_number(frs1, frs2, &env->fp_status));
@@ -229,52 +229,52 @@ uint64_t helper_fmax_s(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
uint64_t helper_fsqrt_s(CPURISCVState *env, uint64_t rs1)
{
- float32 frs1 = check_nanbox_s(rs1);
+ float32 frs1 = check_nanbox_s(env, rs1);
return nanbox_s(float32_sqrt(frs1, &env->fp_status));
}
target_ulong helper_fle_s(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
{
- float32 frs1 = check_nanbox_s(rs1);
- float32 frs2 = check_nanbox_s(rs2);
+ float32 frs1 = check_nanbox_s(env, rs1);
+ float32 frs2 = check_nanbox_s(env, rs2);
return float32_le(frs1, frs2, &env->fp_status);
}
target_ulong helper_flt_s(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
{
- float32 frs1 = check_nanbox_s(rs1);
- float32 frs2 = check_nanbox_s(rs2);
+ float32 frs1 = check_nanbox_s(env, rs1);
+ float32 frs2 = check_nanbox_s(env, rs2);
return float32_lt(frs1, frs2, &env->fp_status);
}
target_ulong helper_feq_s(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
{
- float32 frs1 = check_nanbox_s(rs1);
- float32 frs2 = check_nanbox_s(rs2);
+ float32 frs1 = check_nanbox_s(env, rs1);
+ float32 frs2 = check_nanbox_s(env, rs2);
return float32_eq_quiet(frs1, frs2, &env->fp_status);
}
target_ulong helper_fcvt_w_s(CPURISCVState *env, uint64_t rs1)
{
- float32 frs1 = check_nanbox_s(rs1);
+ float32 frs1 = check_nanbox_s(env, rs1);
return float32_to_int32(frs1, &env->fp_status);
}
target_ulong helper_fcvt_wu_s(CPURISCVState *env, uint64_t rs1)
{
- float32 frs1 = check_nanbox_s(rs1);
+ float32 frs1 = check_nanbox_s(env, rs1);
return (int32_t)float32_to_uint32(frs1, &env->fp_status);
}
target_ulong helper_fcvt_l_s(CPURISCVState *env, uint64_t rs1)
{
- float32 frs1 = check_nanbox_s(rs1);
+ float32 frs1 = check_nanbox_s(env, rs1);
return float32_to_int64(frs1, &env->fp_status);
}
target_ulong helper_fcvt_lu_s(CPURISCVState *env, uint64_t rs1)
{
- float32 frs1 = check_nanbox_s(rs1);
+ float32 frs1 = check_nanbox_s(env, rs1);
return float32_to_uint64(frs1, &env->fp_status);
}
@@ -298,9 +298,9 @@ uint64_t helper_fcvt_s_lu(CPURISCVState *env, target_ulong rs1)
return nanbox_s(uint64_to_float32(rs1, &env->fp_status));
}
-target_ulong helper_fclass_s(uint64_t rs1)
+target_ulong helper_fclass_s(CPURISCVState *env, uint64_t rs1)
{
- float32 frs1 = check_nanbox_s(rs1);
+ float32 frs1 = check_nanbox_s(env, rs1);
return fclass_s(frs1);
}
@@ -345,7 +345,7 @@ uint64_t helper_fcvt_s_d(CPURISCVState *env, uint64_t rs1)
uint64_t helper_fcvt_d_s(CPURISCVState *env, uint64_t rs1)
{
- float32 frs1 = check_nanbox_s(rs1);
+ float32 frs1 = check_nanbox_s(env, rs1);
return float32_to_float64(frs1, &env->fp_status);
}
@@ -539,7 +539,7 @@ uint64_t helper_fcvt_h_lu(CPURISCVState *env, target_ulong rs1)
uint64_t helper_fcvt_h_s(CPURISCVState *env, uint64_t rs1)
{
- float32 frs1 = check_nanbox_s(rs1);
+ float32 frs1 = check_nanbox_s(env, rs1);
return nanbox_h(float32_to_float16(frs1, true, &env->fp_status));
}
diff --git a/target/riscv/helper.h b/target/riscv/helper.h
index c15497e4a1..816acca993 100644
--- a/target/riscv/helper.h
+++ b/target/riscv/helper.h
@@ -38,7 +38,7 @@ DEF_HELPER_FLAGS_2(fcvt_s_w, TCG_CALL_NO_RWG, i64, env, tl)
DEF_HELPER_FLAGS_2(fcvt_s_wu, TCG_CALL_NO_RWG, i64, env, tl)
DEF_HELPER_FLAGS_2(fcvt_s_l, TCG_CALL_NO_RWG, i64, env, tl)
DEF_HELPER_FLAGS_2(fcvt_s_lu, TCG_CALL_NO_RWG, i64, env, tl)
-DEF_HELPER_FLAGS_1(fclass_s, TCG_CALL_NO_RWG_SE, tl, i64)
+DEF_HELPER_FLAGS_2(fclass_s, TCG_CALL_NO_RWG_SE, tl, env, i64)
/* Floating Point - Double Precision */
DEF_HELPER_FLAGS_3(fadd_d, TCG_CALL_NO_RWG, i64, env, i64, i64)
diff --git a/target/riscv/insn_trans/trans_rvf.c.inc b/target/riscv/insn_trans/trans_rvf.c.inc
index b5459249c4..881e265222 100644
--- a/target/riscv/insn_trans/trans_rvf.c.inc
+++ b/target/riscv/insn_trans/trans_rvf.c.inc
@@ -20,7 +20,14 @@
#define REQUIRE_FPU do {\
if (ctx->mstatus_fs == 0) \
- return false; \
+ if (!ctx->ext_zfinx) \
+ return false; \
+} while (0)
+
+#define REQUIRE_ZFINX_OR_F(ctx) do {\
+ if (!ctx->ext_zfinx) {\
+ REQUIRE_EXT(ctx, RVF); \
+ } \
} while (0)
static bool trans_flw(DisasContext *ctx, arg_flw *a)
@@ -70,10 +77,16 @@ static bool trans_fsw(DisasContext *ctx, arg_fsw *a)
static bool trans_fmadd_s(DisasContext *ctx, arg_fmadd_s *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
+ TCGv_i64 src3 = get_fpr_hs(ctx, a->rs3);
+
gen_set_rm(ctx, a->rm);
- gen_helper_fmadd_s(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1],
- cpu_fpr[a->rs2], cpu_fpr[a->rs3]);
+ gen_helper_fmadd_s(dest, cpu_env, src1, src2, src3);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -81,10 +94,16 @@ static bool trans_fmadd_s(DisasContext *ctx, arg_fmadd_s *a)
static bool trans_fmsub_s(DisasContext *ctx, arg_fmsub_s *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
+ TCGv_i64 src3 = get_fpr_hs(ctx, a->rs3);
+
gen_set_rm(ctx, a->rm);
- gen_helper_fmsub_s(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1],
- cpu_fpr[a->rs2], cpu_fpr[a->rs3]);
+ gen_helper_fmsub_s(dest, cpu_env, src1, src2, src3);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -92,10 +111,16 @@ static bool trans_fmsub_s(DisasContext *ctx, arg_fmsub_s *a)
static bool trans_fnmsub_s(DisasContext *ctx, arg_fnmsub_s *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
+ TCGv_i64 src3 = get_fpr_hs(ctx, a->rs3);
+
gen_set_rm(ctx, a->rm);
- gen_helper_fnmsub_s(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1],
- cpu_fpr[a->rs2], cpu_fpr[a->rs3]);
+ gen_helper_fnmsub_s(dest, cpu_env, src1, src2, src3);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -103,10 +128,16 @@ static bool trans_fnmsub_s(DisasContext *ctx, arg_fnmsub_s *a)
static bool trans_fnmadd_s(DisasContext *ctx, arg_fnmadd_s *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
+ TCGv_i64 src3 = get_fpr_hs(ctx, a->rs3);
+
gen_set_rm(ctx, a->rm);
- gen_helper_fnmadd_s(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1],
- cpu_fpr[a->rs2], cpu_fpr[a->rs3]);
+ gen_helper_fnmadd_s(dest, cpu_env, src1, src2, src3);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -114,11 +145,15 @@ static bool trans_fnmadd_s(DisasContext *ctx, arg_fnmadd_s *a)
static bool trans_fadd_s(DisasContext *ctx, arg_fadd_s *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
gen_set_rm(ctx, a->rm);
- gen_helper_fadd_s(cpu_fpr[a->rd], cpu_env,
- cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ gen_helper_fadd_s(dest, cpu_env, src1, src2);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -126,11 +161,15 @@ static bool trans_fadd_s(DisasContext *ctx, arg_fadd_s *a)
static bool trans_fsub_s(DisasContext *ctx, arg_fsub_s *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
gen_set_rm(ctx, a->rm);
- gen_helper_fsub_s(cpu_fpr[a->rd], cpu_env,
- cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ gen_helper_fsub_s(dest, cpu_env, src1, src2);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -138,11 +177,15 @@ static bool trans_fsub_s(DisasContext *ctx, arg_fsub_s *a)
static bool trans_fmul_s(DisasContext *ctx, arg_fmul_s *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
gen_set_rm(ctx, a->rm);
- gen_helper_fmul_s(cpu_fpr[a->rd], cpu_env,
- cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ gen_helper_fmul_s(dest, cpu_env, src1, src2);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -150,11 +193,15 @@ static bool trans_fmul_s(DisasContext *ctx, arg_fmul_s *a)
static bool trans_fdiv_s(DisasContext *ctx, arg_fdiv_s *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
gen_set_rm(ctx, a->rm);
- gen_helper_fdiv_s(cpu_fpr[a->rd], cpu_env,
- cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ gen_helper_fdiv_s(dest, cpu_env, src1, src2);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -162,10 +209,14 @@ static bool trans_fdiv_s(DisasContext *ctx, arg_fdiv_s *a)
static bool trans_fsqrt_s(DisasContext *ctx, arg_fsqrt_s *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
gen_set_rm(ctx, a->rm);
- gen_helper_fsqrt_s(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1]);
+ gen_helper_fsqrt_s(dest, cpu_env, src1);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -173,22 +224,40 @@ static bool trans_fsqrt_s(DisasContext *ctx, arg_fsqrt_s *a)
static bool trans_fsgnj_s(DisasContext *ctx, arg_fsgnj_s *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
if (a->rs1 == a->rs2) { /* FMOV */
- gen_check_nanbox_s(cpu_fpr[a->rd], cpu_fpr[a->rs1]);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ if (ctx->ext_zfinx) {
+ gen_nanbox_s(dest, src1);
+ } else {
+ gen_check_nanbox_s(dest, src1);
+ }
} else { /* FSGNJ */
- TCGv_i64 rs1 = tcg_temp_new_i64();
- TCGv_i64 rs2 = tcg_temp_new_i64();
-
- gen_check_nanbox_s(rs1, cpu_fpr[a->rs1]);
- gen_check_nanbox_s(rs2, cpu_fpr[a->rs2]);
-
- /* This formulation retains the nanboxing of rs2. */
- tcg_gen_deposit_i64(cpu_fpr[a->rd], rs2, rs1, 0, 31);
- tcg_temp_free_i64(rs1);
- tcg_temp_free_i64(rs2);
+ TCGv_i64 rs1, rs2;
+ if (!ctx->ext_zfinx) {
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
+ rs1 = tcg_temp_new_i64();
+ rs2 = tcg_temp_new_i64();
+ gen_check_nanbox_s(rs1, src1);
+ gen_check_nanbox_s(rs2, src2);
+ } else {
+ rs1 = get_fpr_hs(ctx, a->rs1);
+ rs2 = get_fpr_hs(ctx, a->rs2);
+ }
+
+ /* This formulation retains the nanboxing of rs2 in normal 'F'. */
+ tcg_gen_deposit_i64(dest, rs2, rs1, 0, 31);
+ if (!ctx->ext_zfinx) {
+ tcg_temp_free_i64(rs1);
+ tcg_temp_free_i64(rs2);
+ } else {
+ gen_nanbox_s(dest, dest);
+ }
}
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -198,16 +267,27 @@ static bool trans_fsgnjn_s(DisasContext *ctx, arg_fsgnjn_s *a)
TCGv_i64 rs1, rs2, mask;
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
- rs1 = tcg_temp_new_i64();
- gen_check_nanbox_s(rs1, cpu_fpr[a->rs1]);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ if (!ctx->ext_zfinx) {
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ rs1 = tcg_temp_new_i64();
+ gen_check_nanbox_s(rs1, src1);
+ } else {
+ rs1 = get_fpr_hs(ctx, a->rs1);
+ }
if (a->rs1 == a->rs2) { /* FNEG */
- tcg_gen_xori_i64(cpu_fpr[a->rd], rs1, MAKE_64BIT_MASK(31, 1));
+ tcg_gen_xori_i64(dest, rs1, MAKE_64BIT_MASK(31, 1));
} else {
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
rs2 = tcg_temp_new_i64();
- gen_check_nanbox_s(rs2, cpu_fpr[a->rs2]);
+ if (!ctx->ext_zfinx) {
+ gen_check_nanbox_s(rs2, src2);
+ } else {
+ tcg_gen_mov_i64(rs2, src2);
+ }
/*
* Replace bit 31 in rs1 with inverse in rs2.
@@ -215,13 +295,17 @@ static bool trans_fsgnjn_s(DisasContext *ctx, arg_fsgnjn_s *a)
*/
mask = tcg_constant_i64(~MAKE_64BIT_MASK(31, 1));
tcg_gen_nor_i64(rs2, rs2, mask);
- tcg_gen_and_i64(rs1, mask, rs1);
- tcg_gen_or_i64(cpu_fpr[a->rd], rs1, rs2);
-
+ tcg_gen_and_i64(dest, mask, rs1);
+ tcg_gen_or_i64(dest, dest, rs2);
+
tcg_temp_free_i64(rs2);
}
- tcg_temp_free_i64(rs1);
-
+ if (ctx->ext_zfinx) {
+ gen_nanbox_s(dest, dest);
+ } else {
+ tcg_temp_free_i64(rs1);
+ }
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -231,28 +315,44 @@ static bool trans_fsgnjx_s(DisasContext *ctx, arg_fsgnjx_s *a)
TCGv_i64 rs1, rs2;
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
- rs1 = tcg_temp_new_i64();
- gen_check_nanbox_s(rs1, cpu_fpr[a->rs1]);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ if (!ctx->ext_zfinx) {
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ rs1 = tcg_temp_new_i64();
+ gen_check_nanbox_s(rs1, src1);
+ } else {
+ rs1 = get_fpr_hs(ctx, a->rs1);
+ }
if (a->rs1 == a->rs2) { /* FABS */
- tcg_gen_andi_i64(cpu_fpr[a->rd], rs1, ~MAKE_64BIT_MASK(31, 1));
+ tcg_gen_andi_i64(dest, rs1, ~MAKE_64BIT_MASK(31, 1));
} else {
- rs2 = tcg_temp_new_i64();
- gen_check_nanbox_s(rs2, cpu_fpr[a->rs2]);
+ if (!ctx->ext_zfinx) {
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
+ rs2 = tcg_temp_new_i64();
+ gen_check_nanbox_s(rs2, src2);
+ } else {
+ rs2 = get_fpr_hs(ctx, a->rs2);
+ }
/*
* Xor bit 31 in rs1 with that in rs2.
* This formulation retains the nanboxing of rs1.
*/
- tcg_gen_andi_i64(rs2, rs2, MAKE_64BIT_MASK(31, 1));
- tcg_gen_xor_i64(cpu_fpr[a->rd], rs1, rs2);
-
- tcg_temp_free_i64(rs2);
+ tcg_gen_andi_i64(dest, rs2, MAKE_64BIT_MASK(31, 1));
+ tcg_gen_xor_i64(dest, rs1, dest);
+ if (!ctx->ext_zfinx) {
+ tcg_temp_free_i64(rs2);
+ }
}
- tcg_temp_free_i64(rs1);
-
+ if (ctx->ext_zfinx) {
+ gen_nanbox_s(dest, dest);
+ } else {
+ tcg_temp_free_i64(rs1);
+ }
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -260,10 +360,14 @@ static bool trans_fsgnjx_s(DisasContext *ctx, arg_fsgnjx_s *a)
static bool trans_fmin_s(DisasContext *ctx, arg_fmin_s *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
- gen_helper_fmin_s(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1],
- cpu_fpr[a->rs2]);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
+
+ gen_helper_fmin_s(dest, cpu_env, src1, src2);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -271,10 +375,14 @@ static bool trans_fmin_s(DisasContext *ctx, arg_fmin_s *a)
static bool trans_fmax_s(DisasContext *ctx, arg_fmax_s *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
- gen_helper_fmax_s(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1],
- cpu_fpr[a->rs2]);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
+
+ gen_helper_fmax_s(dest, cpu_env, src1, src2);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -282,12 +390,13 @@ static bool trans_fmax_s(DisasContext *ctx, arg_fmax_s *a)
static bool trans_fcvt_w_s(DisasContext *ctx, arg_fcvt_w_s *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_w_s(dest, cpu_env, cpu_fpr[a->rs1]);
+ gen_helper_fcvt_w_s(dest, cpu_env, src1);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -295,12 +404,13 @@ static bool trans_fcvt_w_s(DisasContext *ctx, arg_fcvt_w_s *a)
static bool trans_fcvt_wu_s(DisasContext *ctx, arg_fcvt_wu_s *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_wu_s(dest, cpu_env, cpu_fpr[a->rs1]);
+ gen_helper_fcvt_wu_s(dest, cpu_env, src1);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -309,14 +419,14 @@ static bool trans_fmv_x_w(DisasContext *ctx, arg_fmv_x_w *a)
{
/* NOTE: This was FMV.X.S in an earlier version of the ISA spec! */
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
-
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
#if defined(TARGET_RISCV64)
- tcg_gen_ext32s_tl(dest, cpu_fpr[a->rs1]);
+ tcg_gen_ext32s_tl(dest, src1);
#else
- tcg_gen_extrl_i64_i32(dest, cpu_fpr[a->rs1]);
+ tcg_gen_extrl_i64_i32(dest, src1);
#endif
gen_set_gpr(ctx, a->rd, dest);
@@ -326,11 +436,13 @@ static bool trans_fmv_x_w(DisasContext *ctx, arg_fmv_x_w *a)
static bool trans_feq_s(DisasContext *ctx, arg_feq_s *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
- gen_helper_feq_s(dest, cpu_env, cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ gen_helper_feq_s(dest, cpu_env, src1, src2);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -338,11 +450,13 @@ static bool trans_feq_s(DisasContext *ctx, arg_feq_s *a)
static bool trans_flt_s(DisasContext *ctx, arg_flt_s *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
- gen_helper_flt_s(dest, cpu_env, cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ gen_helper_flt_s(dest, cpu_env, src1, src2);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -350,11 +464,13 @@ static bool trans_flt_s(DisasContext *ctx, arg_flt_s *a)
static bool trans_fle_s(DisasContext *ctx, arg_fle_s *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
- gen_helper_fle_s(dest, cpu_env, cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ gen_helper_fle_s(dest, cpu_env, src1, src2);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -362,11 +478,12 @@ static bool trans_fle_s(DisasContext *ctx, arg_fle_s *a)
static bool trans_fclass_s(DisasContext *ctx, arg_fclass_s *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
- gen_helper_fclass_s(dest, cpu_fpr[a->rs1]);
+ gen_helper_fclass_s(dest, cpu_env, src1);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -374,13 +491,14 @@ static bool trans_fclass_s(DisasContext *ctx, arg_fclass_s *a)
static bool trans_fcvt_s_w(DisasContext *ctx, arg_fcvt_s_w *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
TCGv src = get_gpr(ctx, a->rs1, EXT_SIGN);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_s_w(cpu_fpr[a->rd], cpu_env, src);
-
+ gen_helper_fcvt_s_w(dest, cpu_env, src);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -388,13 +506,14 @@ static bool trans_fcvt_s_w(DisasContext *ctx, arg_fcvt_s_w *a)
static bool trans_fcvt_s_wu(DisasContext *ctx, arg_fcvt_s_wu *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
TCGv src = get_gpr(ctx, a->rs1, EXT_ZERO);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_s_wu(cpu_fpr[a->rd], cpu_env, src);
-
+ gen_helper_fcvt_s_wu(dest, cpu_env, src);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -403,13 +522,14 @@ static bool trans_fmv_w_x(DisasContext *ctx, arg_fmv_w_x *a)
{
/* NOTE: This was FMV.S.X in an earlier version of the ISA spec! */
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
TCGv src = get_gpr(ctx, a->rs1, EXT_ZERO);
- tcg_gen_extu_tl_i64(cpu_fpr[a->rd], src);
- gen_nanbox_s(cpu_fpr[a->rd], cpu_fpr[a->rd]);
-
+ tcg_gen_extu_tl_i64(dest, src);
+ gen_nanbox_s(dest, dest);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -418,12 +538,13 @@ static bool trans_fcvt_l_s(DisasContext *ctx, arg_fcvt_l_s *a)
{
REQUIRE_64BIT(ctx);
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_l_s(dest, cpu_env, cpu_fpr[a->rs1]);
+ gen_helper_fcvt_l_s(dest, cpu_env, src1);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -432,12 +553,13 @@ static bool trans_fcvt_lu_s(DisasContext *ctx, arg_fcvt_lu_s *a)
{
REQUIRE_64BIT(ctx);
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_lu_s(dest, cpu_env, cpu_fpr[a->rs1]);
+ gen_helper_fcvt_lu_s(dest, cpu_env, src1);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -446,13 +568,14 @@ static bool trans_fcvt_s_l(DisasContext *ctx, arg_fcvt_s_l *a)
{
REQUIRE_64BIT(ctx);
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
TCGv src = get_gpr(ctx, a->rs1, EXT_SIGN);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_s_l(cpu_fpr[a->rd], cpu_env, src);
-
+ gen_helper_fcvt_s_l(dest, cpu_env, src);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -461,13 +584,14 @@ static bool trans_fcvt_s_lu(DisasContext *ctx, arg_fcvt_s_lu *a)
{
REQUIRE_64BIT(ctx);
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
TCGv src = get_gpr(ctx, a->rs1, EXT_ZERO);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_s_lu(cpu_fpr[a->rd], cpu_env, src);
-
+ gen_helper_fcvt_s_lu(dest, cpu_env, src);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
diff --git a/target/riscv/internals.h b/target/riscv/internals.h
index 065e8162a2..9f3f3319f2 100644
--- a/target/riscv/internals.h
+++ b/target/riscv/internals.h
@@ -51,8 +51,12 @@ static inline uint64_t nanbox_s(float32 f)
return f | MAKE_64BIT_MASK(32, 32);
}
-static inline float32 check_nanbox_s(uint64_t f)
+static inline float32 check_nanbox_s(CPURISCVState *env, uint64_t f)
{
+ /* Disable nanbox check when enable zfinx */
+ if (RISCV_CPU(env_cpu(env))->cfg.ext_zfinx)
+ return (uint32_t)f;
+
uint64_t mask = MAKE_64BIT_MASK(32, 32);
if (likely((f & mask) == mask)) {
--
2.17.1
^ permalink raw reply related [flat|nested] 30+ messages in thread
* [PATCH 3/6] target/riscv: add support for zfinx
@ 2021-12-24 3:49 ` liweiwei
0 siblings, 0 replies; 30+ messages in thread
From: liweiwei @ 2021-12-24 3:49 UTC (permalink / raw)
To: palmer, alistair.francis, bin.meng, qemu-riscv, qemu-devel
Cc: wangjunqiang, lazyparser, ardxwe, liweiwei
- update extension check REQUIRE_ZFINX_OR_F
- update single float point register read/write
- disable nanbox_s check
Co-authored-by: ardxwe <ardxwe@gmail.com>
Signed-off-by: liweiwei <liweiwei@iscas.ac.cn>
Signed-off-by: wangjunqiang <wangjunqiang@iscas.ac.cn>
---
target/riscv/fpu_helper.c | 60 ++---
target/riscv/helper.h | 2 +-
target/riscv/insn_trans/trans_rvf.c.inc | 330 ++++++++++++++++--------
target/riscv/internals.h | 6 +-
4 files changed, 263 insertions(+), 135 deletions(-)
diff --git a/target/riscv/fpu_helper.c b/target/riscv/fpu_helper.c
index 4a5982d594..0895fa3c7e 100644
--- a/target/riscv/fpu_helper.c
+++ b/target/riscv/fpu_helper.c
@@ -98,9 +98,9 @@ static uint64_t do_fmadd_h(CPURISCVState *env, uint64_t rs1, uint64_t rs2,
static uint64_t do_fmadd_s(CPURISCVState *env, uint64_t rs1, uint64_t rs2,
uint64_t rs3, int flags)
{
- float32 frs1 = check_nanbox_s(rs1);
- float32 frs2 = check_nanbox_s(rs2);
- float32 frs3 = check_nanbox_s(rs3);
+ float32 frs1 = check_nanbox_s(env, rs1);
+ float32 frs2 = check_nanbox_s(env, rs2);
+ float32 frs3 = check_nanbox_s(env, rs3);
return nanbox_s(float32_muladd(frs1, frs2, frs3, flags, &env->fp_status));
}
@@ -183,36 +183,36 @@ uint64_t helper_fnmadd_h(CPURISCVState *env, uint64_t frs1, uint64_t frs2,
uint64_t helper_fadd_s(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
{
- float32 frs1 = check_nanbox_s(rs1);
- float32 frs2 = check_nanbox_s(rs2);
+ float32 frs1 = check_nanbox_s(env, rs1);
+ float32 frs2 = check_nanbox_s(env, rs2);
return nanbox_s(float32_add(frs1, frs2, &env->fp_status));
}
uint64_t helper_fsub_s(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
{
- float32 frs1 = check_nanbox_s(rs1);
- float32 frs2 = check_nanbox_s(rs2);
+ float32 frs1 = check_nanbox_s(env, rs1);
+ float32 frs2 = check_nanbox_s(env, rs2);
return nanbox_s(float32_sub(frs1, frs2, &env->fp_status));
}
uint64_t helper_fmul_s(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
{
- float32 frs1 = check_nanbox_s(rs1);
- float32 frs2 = check_nanbox_s(rs2);
+ float32 frs1 = check_nanbox_s(env, rs1);
+ float32 frs2 = check_nanbox_s(env, rs2);
return nanbox_s(float32_mul(frs1, frs2, &env->fp_status));
}
uint64_t helper_fdiv_s(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
{
- float32 frs1 = check_nanbox_s(rs1);
- float32 frs2 = check_nanbox_s(rs2);
+ float32 frs1 = check_nanbox_s(env, rs1);
+ float32 frs2 = check_nanbox_s(env, rs2);
return nanbox_s(float32_div(frs1, frs2, &env->fp_status));
}
uint64_t helper_fmin_s(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
{
- float32 frs1 = check_nanbox_s(rs1);
- float32 frs2 = check_nanbox_s(rs2);
+ float32 frs1 = check_nanbox_s(env, rs1);
+ float32 frs2 = check_nanbox_s(env, rs2);
return nanbox_s(env->priv_ver < PRIV_VERSION_1_11_0 ?
float32_minnum(frs1, frs2, &env->fp_status) :
float32_minimum_number(frs1, frs2, &env->fp_status));
@@ -220,8 +220,8 @@ uint64_t helper_fmin_s(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
uint64_t helper_fmax_s(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
{
- float32 frs1 = check_nanbox_s(rs1);
- float32 frs2 = check_nanbox_s(rs2);
+ float32 frs1 = check_nanbox_s(env, rs1);
+ float32 frs2 = check_nanbox_s(env, rs2);
return nanbox_s(env->priv_ver < PRIV_VERSION_1_11_0 ?
float32_maxnum(frs1, frs2, &env->fp_status) :
float32_maximum_number(frs1, frs2, &env->fp_status));
@@ -229,52 +229,52 @@ uint64_t helper_fmax_s(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
uint64_t helper_fsqrt_s(CPURISCVState *env, uint64_t rs1)
{
- float32 frs1 = check_nanbox_s(rs1);
+ float32 frs1 = check_nanbox_s(env, rs1);
return nanbox_s(float32_sqrt(frs1, &env->fp_status));
}
target_ulong helper_fle_s(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
{
- float32 frs1 = check_nanbox_s(rs1);
- float32 frs2 = check_nanbox_s(rs2);
+ float32 frs1 = check_nanbox_s(env, rs1);
+ float32 frs2 = check_nanbox_s(env, rs2);
return float32_le(frs1, frs2, &env->fp_status);
}
target_ulong helper_flt_s(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
{
- float32 frs1 = check_nanbox_s(rs1);
- float32 frs2 = check_nanbox_s(rs2);
+ float32 frs1 = check_nanbox_s(env, rs1);
+ float32 frs2 = check_nanbox_s(env, rs2);
return float32_lt(frs1, frs2, &env->fp_status);
}
target_ulong helper_feq_s(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
{
- float32 frs1 = check_nanbox_s(rs1);
- float32 frs2 = check_nanbox_s(rs2);
+ float32 frs1 = check_nanbox_s(env, rs1);
+ float32 frs2 = check_nanbox_s(env, rs2);
return float32_eq_quiet(frs1, frs2, &env->fp_status);
}
target_ulong helper_fcvt_w_s(CPURISCVState *env, uint64_t rs1)
{
- float32 frs1 = check_nanbox_s(rs1);
+ float32 frs1 = check_nanbox_s(env, rs1);
return float32_to_int32(frs1, &env->fp_status);
}
target_ulong helper_fcvt_wu_s(CPURISCVState *env, uint64_t rs1)
{
- float32 frs1 = check_nanbox_s(rs1);
+ float32 frs1 = check_nanbox_s(env, rs1);
return (int32_t)float32_to_uint32(frs1, &env->fp_status);
}
target_ulong helper_fcvt_l_s(CPURISCVState *env, uint64_t rs1)
{
- float32 frs1 = check_nanbox_s(rs1);
+ float32 frs1 = check_nanbox_s(env, rs1);
return float32_to_int64(frs1, &env->fp_status);
}
target_ulong helper_fcvt_lu_s(CPURISCVState *env, uint64_t rs1)
{
- float32 frs1 = check_nanbox_s(rs1);
+ float32 frs1 = check_nanbox_s(env, rs1);
return float32_to_uint64(frs1, &env->fp_status);
}
@@ -298,9 +298,9 @@ uint64_t helper_fcvt_s_lu(CPURISCVState *env, target_ulong rs1)
return nanbox_s(uint64_to_float32(rs1, &env->fp_status));
}
-target_ulong helper_fclass_s(uint64_t rs1)
+target_ulong helper_fclass_s(CPURISCVState *env, uint64_t rs1)
{
- float32 frs1 = check_nanbox_s(rs1);
+ float32 frs1 = check_nanbox_s(env, rs1);
return fclass_s(frs1);
}
@@ -345,7 +345,7 @@ uint64_t helper_fcvt_s_d(CPURISCVState *env, uint64_t rs1)
uint64_t helper_fcvt_d_s(CPURISCVState *env, uint64_t rs1)
{
- float32 frs1 = check_nanbox_s(rs1);
+ float32 frs1 = check_nanbox_s(env, rs1);
return float32_to_float64(frs1, &env->fp_status);
}
@@ -539,7 +539,7 @@ uint64_t helper_fcvt_h_lu(CPURISCVState *env, target_ulong rs1)
uint64_t helper_fcvt_h_s(CPURISCVState *env, uint64_t rs1)
{
- float32 frs1 = check_nanbox_s(rs1);
+ float32 frs1 = check_nanbox_s(env, rs1);
return nanbox_h(float32_to_float16(frs1, true, &env->fp_status));
}
diff --git a/target/riscv/helper.h b/target/riscv/helper.h
index c15497e4a1..816acca993 100644
--- a/target/riscv/helper.h
+++ b/target/riscv/helper.h
@@ -38,7 +38,7 @@ DEF_HELPER_FLAGS_2(fcvt_s_w, TCG_CALL_NO_RWG, i64, env, tl)
DEF_HELPER_FLAGS_2(fcvt_s_wu, TCG_CALL_NO_RWG, i64, env, tl)
DEF_HELPER_FLAGS_2(fcvt_s_l, TCG_CALL_NO_RWG, i64, env, tl)
DEF_HELPER_FLAGS_2(fcvt_s_lu, TCG_CALL_NO_RWG, i64, env, tl)
-DEF_HELPER_FLAGS_1(fclass_s, TCG_CALL_NO_RWG_SE, tl, i64)
+DEF_HELPER_FLAGS_2(fclass_s, TCG_CALL_NO_RWG_SE, tl, env, i64)
/* Floating Point - Double Precision */
DEF_HELPER_FLAGS_3(fadd_d, TCG_CALL_NO_RWG, i64, env, i64, i64)
diff --git a/target/riscv/insn_trans/trans_rvf.c.inc b/target/riscv/insn_trans/trans_rvf.c.inc
index b5459249c4..881e265222 100644
--- a/target/riscv/insn_trans/trans_rvf.c.inc
+++ b/target/riscv/insn_trans/trans_rvf.c.inc
@@ -20,7 +20,14 @@
#define REQUIRE_FPU do {\
if (ctx->mstatus_fs == 0) \
- return false; \
+ if (!ctx->ext_zfinx) \
+ return false; \
+} while (0)
+
+#define REQUIRE_ZFINX_OR_F(ctx) do {\
+ if (!ctx->ext_zfinx) {\
+ REQUIRE_EXT(ctx, RVF); \
+ } \
} while (0)
static bool trans_flw(DisasContext *ctx, arg_flw *a)
@@ -70,10 +77,16 @@ static bool trans_fsw(DisasContext *ctx, arg_fsw *a)
static bool trans_fmadd_s(DisasContext *ctx, arg_fmadd_s *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
+ TCGv_i64 src3 = get_fpr_hs(ctx, a->rs3);
+
gen_set_rm(ctx, a->rm);
- gen_helper_fmadd_s(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1],
- cpu_fpr[a->rs2], cpu_fpr[a->rs3]);
+ gen_helper_fmadd_s(dest, cpu_env, src1, src2, src3);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -81,10 +94,16 @@ static bool trans_fmadd_s(DisasContext *ctx, arg_fmadd_s *a)
static bool trans_fmsub_s(DisasContext *ctx, arg_fmsub_s *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
+ TCGv_i64 src3 = get_fpr_hs(ctx, a->rs3);
+
gen_set_rm(ctx, a->rm);
- gen_helper_fmsub_s(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1],
- cpu_fpr[a->rs2], cpu_fpr[a->rs3]);
+ gen_helper_fmsub_s(dest, cpu_env, src1, src2, src3);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -92,10 +111,16 @@ static bool trans_fmsub_s(DisasContext *ctx, arg_fmsub_s *a)
static bool trans_fnmsub_s(DisasContext *ctx, arg_fnmsub_s *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
+ TCGv_i64 src3 = get_fpr_hs(ctx, a->rs3);
+
gen_set_rm(ctx, a->rm);
- gen_helper_fnmsub_s(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1],
- cpu_fpr[a->rs2], cpu_fpr[a->rs3]);
+ gen_helper_fnmsub_s(dest, cpu_env, src1, src2, src3);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -103,10 +128,16 @@ static bool trans_fnmsub_s(DisasContext *ctx, arg_fnmsub_s *a)
static bool trans_fnmadd_s(DisasContext *ctx, arg_fnmadd_s *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
+ TCGv_i64 src3 = get_fpr_hs(ctx, a->rs3);
+
gen_set_rm(ctx, a->rm);
- gen_helper_fnmadd_s(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1],
- cpu_fpr[a->rs2], cpu_fpr[a->rs3]);
+ gen_helper_fnmadd_s(dest, cpu_env, src1, src2, src3);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -114,11 +145,15 @@ static bool trans_fnmadd_s(DisasContext *ctx, arg_fnmadd_s *a)
static bool trans_fadd_s(DisasContext *ctx, arg_fadd_s *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
gen_set_rm(ctx, a->rm);
- gen_helper_fadd_s(cpu_fpr[a->rd], cpu_env,
- cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ gen_helper_fadd_s(dest, cpu_env, src1, src2);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -126,11 +161,15 @@ static bool trans_fadd_s(DisasContext *ctx, arg_fadd_s *a)
static bool trans_fsub_s(DisasContext *ctx, arg_fsub_s *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
gen_set_rm(ctx, a->rm);
- gen_helper_fsub_s(cpu_fpr[a->rd], cpu_env,
- cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ gen_helper_fsub_s(dest, cpu_env, src1, src2);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -138,11 +177,15 @@ static bool trans_fsub_s(DisasContext *ctx, arg_fsub_s *a)
static bool trans_fmul_s(DisasContext *ctx, arg_fmul_s *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
gen_set_rm(ctx, a->rm);
- gen_helper_fmul_s(cpu_fpr[a->rd], cpu_env,
- cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ gen_helper_fmul_s(dest, cpu_env, src1, src2);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -150,11 +193,15 @@ static bool trans_fmul_s(DisasContext *ctx, arg_fmul_s *a)
static bool trans_fdiv_s(DisasContext *ctx, arg_fdiv_s *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
gen_set_rm(ctx, a->rm);
- gen_helper_fdiv_s(cpu_fpr[a->rd], cpu_env,
- cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ gen_helper_fdiv_s(dest, cpu_env, src1, src2);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -162,10 +209,14 @@ static bool trans_fdiv_s(DisasContext *ctx, arg_fdiv_s *a)
static bool trans_fsqrt_s(DisasContext *ctx, arg_fsqrt_s *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
gen_set_rm(ctx, a->rm);
- gen_helper_fsqrt_s(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1]);
+ gen_helper_fsqrt_s(dest, cpu_env, src1);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -173,22 +224,40 @@ static bool trans_fsqrt_s(DisasContext *ctx, arg_fsqrt_s *a)
static bool trans_fsgnj_s(DisasContext *ctx, arg_fsgnj_s *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
if (a->rs1 == a->rs2) { /* FMOV */
- gen_check_nanbox_s(cpu_fpr[a->rd], cpu_fpr[a->rs1]);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ if (ctx->ext_zfinx) {
+ gen_nanbox_s(dest, src1);
+ } else {
+ gen_check_nanbox_s(dest, src1);
+ }
} else { /* FSGNJ */
- TCGv_i64 rs1 = tcg_temp_new_i64();
- TCGv_i64 rs2 = tcg_temp_new_i64();
-
- gen_check_nanbox_s(rs1, cpu_fpr[a->rs1]);
- gen_check_nanbox_s(rs2, cpu_fpr[a->rs2]);
-
- /* This formulation retains the nanboxing of rs2. */
- tcg_gen_deposit_i64(cpu_fpr[a->rd], rs2, rs1, 0, 31);
- tcg_temp_free_i64(rs1);
- tcg_temp_free_i64(rs2);
+ TCGv_i64 rs1, rs2;
+ if (!ctx->ext_zfinx) {
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
+ rs1 = tcg_temp_new_i64();
+ rs2 = tcg_temp_new_i64();
+ gen_check_nanbox_s(rs1, src1);
+ gen_check_nanbox_s(rs2, src2);
+ } else {
+ rs1 = get_fpr_hs(ctx, a->rs1);
+ rs2 = get_fpr_hs(ctx, a->rs2);
+ }
+
+ /* This formulation retains the nanboxing of rs2 in normal 'F'. */
+ tcg_gen_deposit_i64(dest, rs2, rs1, 0, 31);
+ if (!ctx->ext_zfinx) {
+ tcg_temp_free_i64(rs1);
+ tcg_temp_free_i64(rs2);
+ } else {
+ gen_nanbox_s(dest, dest);
+ }
}
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -198,16 +267,27 @@ static bool trans_fsgnjn_s(DisasContext *ctx, arg_fsgnjn_s *a)
TCGv_i64 rs1, rs2, mask;
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
- rs1 = tcg_temp_new_i64();
- gen_check_nanbox_s(rs1, cpu_fpr[a->rs1]);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ if (!ctx->ext_zfinx) {
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ rs1 = tcg_temp_new_i64();
+ gen_check_nanbox_s(rs1, src1);
+ } else {
+ rs1 = get_fpr_hs(ctx, a->rs1);
+ }
if (a->rs1 == a->rs2) { /* FNEG */
- tcg_gen_xori_i64(cpu_fpr[a->rd], rs1, MAKE_64BIT_MASK(31, 1));
+ tcg_gen_xori_i64(dest, rs1, MAKE_64BIT_MASK(31, 1));
} else {
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
rs2 = tcg_temp_new_i64();
- gen_check_nanbox_s(rs2, cpu_fpr[a->rs2]);
+ if (!ctx->ext_zfinx) {
+ gen_check_nanbox_s(rs2, src2);
+ } else {
+ tcg_gen_mov_i64(rs2, src2);
+ }
/*
* Replace bit 31 in rs1 with inverse in rs2.
@@ -215,13 +295,17 @@ static bool trans_fsgnjn_s(DisasContext *ctx, arg_fsgnjn_s *a)
*/
mask = tcg_constant_i64(~MAKE_64BIT_MASK(31, 1));
tcg_gen_nor_i64(rs2, rs2, mask);
- tcg_gen_and_i64(rs1, mask, rs1);
- tcg_gen_or_i64(cpu_fpr[a->rd], rs1, rs2);
-
+ tcg_gen_and_i64(dest, mask, rs1);
+ tcg_gen_or_i64(dest, dest, rs2);
+
tcg_temp_free_i64(rs2);
}
- tcg_temp_free_i64(rs1);
-
+ if (ctx->ext_zfinx) {
+ gen_nanbox_s(dest, dest);
+ } else {
+ tcg_temp_free_i64(rs1);
+ }
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -231,28 +315,44 @@ static bool trans_fsgnjx_s(DisasContext *ctx, arg_fsgnjx_s *a)
TCGv_i64 rs1, rs2;
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
- rs1 = tcg_temp_new_i64();
- gen_check_nanbox_s(rs1, cpu_fpr[a->rs1]);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ if (!ctx->ext_zfinx) {
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ rs1 = tcg_temp_new_i64();
+ gen_check_nanbox_s(rs1, src1);
+ } else {
+ rs1 = get_fpr_hs(ctx, a->rs1);
+ }
if (a->rs1 == a->rs2) { /* FABS */
- tcg_gen_andi_i64(cpu_fpr[a->rd], rs1, ~MAKE_64BIT_MASK(31, 1));
+ tcg_gen_andi_i64(dest, rs1, ~MAKE_64BIT_MASK(31, 1));
} else {
- rs2 = tcg_temp_new_i64();
- gen_check_nanbox_s(rs2, cpu_fpr[a->rs2]);
+ if (!ctx->ext_zfinx) {
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
+ rs2 = tcg_temp_new_i64();
+ gen_check_nanbox_s(rs2, src2);
+ } else {
+ rs2 = get_fpr_hs(ctx, a->rs2);
+ }
/*
* Xor bit 31 in rs1 with that in rs2.
* This formulation retains the nanboxing of rs1.
*/
- tcg_gen_andi_i64(rs2, rs2, MAKE_64BIT_MASK(31, 1));
- tcg_gen_xor_i64(cpu_fpr[a->rd], rs1, rs2);
-
- tcg_temp_free_i64(rs2);
+ tcg_gen_andi_i64(dest, rs2, MAKE_64BIT_MASK(31, 1));
+ tcg_gen_xor_i64(dest, rs1, dest);
+ if (!ctx->ext_zfinx) {
+ tcg_temp_free_i64(rs2);
+ }
}
- tcg_temp_free_i64(rs1);
-
+ if (ctx->ext_zfinx) {
+ gen_nanbox_s(dest, dest);
+ } else {
+ tcg_temp_free_i64(rs1);
+ }
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -260,10 +360,14 @@ static bool trans_fsgnjx_s(DisasContext *ctx, arg_fsgnjx_s *a)
static bool trans_fmin_s(DisasContext *ctx, arg_fmin_s *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
- gen_helper_fmin_s(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1],
- cpu_fpr[a->rs2]);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
+
+ gen_helper_fmin_s(dest, cpu_env, src1, src2);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -271,10 +375,14 @@ static bool trans_fmin_s(DisasContext *ctx, arg_fmin_s *a)
static bool trans_fmax_s(DisasContext *ctx, arg_fmax_s *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
- gen_helper_fmax_s(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1],
- cpu_fpr[a->rs2]);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
+
+ gen_helper_fmax_s(dest, cpu_env, src1, src2);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -282,12 +390,13 @@ static bool trans_fmax_s(DisasContext *ctx, arg_fmax_s *a)
static bool trans_fcvt_w_s(DisasContext *ctx, arg_fcvt_w_s *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_w_s(dest, cpu_env, cpu_fpr[a->rs1]);
+ gen_helper_fcvt_w_s(dest, cpu_env, src1);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -295,12 +404,13 @@ static bool trans_fcvt_w_s(DisasContext *ctx, arg_fcvt_w_s *a)
static bool trans_fcvt_wu_s(DisasContext *ctx, arg_fcvt_wu_s *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_wu_s(dest, cpu_env, cpu_fpr[a->rs1]);
+ gen_helper_fcvt_wu_s(dest, cpu_env, src1);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -309,14 +419,14 @@ static bool trans_fmv_x_w(DisasContext *ctx, arg_fmv_x_w *a)
{
/* NOTE: This was FMV.X.S in an earlier version of the ISA spec! */
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
-
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
#if defined(TARGET_RISCV64)
- tcg_gen_ext32s_tl(dest, cpu_fpr[a->rs1]);
+ tcg_gen_ext32s_tl(dest, src1);
#else
- tcg_gen_extrl_i64_i32(dest, cpu_fpr[a->rs1]);
+ tcg_gen_extrl_i64_i32(dest, src1);
#endif
gen_set_gpr(ctx, a->rd, dest);
@@ -326,11 +436,13 @@ static bool trans_fmv_x_w(DisasContext *ctx, arg_fmv_x_w *a)
static bool trans_feq_s(DisasContext *ctx, arg_feq_s *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
- gen_helper_feq_s(dest, cpu_env, cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ gen_helper_feq_s(dest, cpu_env, src1, src2);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -338,11 +450,13 @@ static bool trans_feq_s(DisasContext *ctx, arg_feq_s *a)
static bool trans_flt_s(DisasContext *ctx, arg_flt_s *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
- gen_helper_flt_s(dest, cpu_env, cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ gen_helper_flt_s(dest, cpu_env, src1, src2);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -350,11 +464,13 @@ static bool trans_flt_s(DisasContext *ctx, arg_flt_s *a)
static bool trans_fle_s(DisasContext *ctx, arg_fle_s *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
- gen_helper_fle_s(dest, cpu_env, cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ gen_helper_fle_s(dest, cpu_env, src1, src2);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -362,11 +478,12 @@ static bool trans_fle_s(DisasContext *ctx, arg_fle_s *a)
static bool trans_fclass_s(DisasContext *ctx, arg_fclass_s *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
- gen_helper_fclass_s(dest, cpu_fpr[a->rs1]);
+ gen_helper_fclass_s(dest, cpu_env, src1);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -374,13 +491,14 @@ static bool trans_fclass_s(DisasContext *ctx, arg_fclass_s *a)
static bool trans_fcvt_s_w(DisasContext *ctx, arg_fcvt_s_w *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
TCGv src = get_gpr(ctx, a->rs1, EXT_SIGN);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_s_w(cpu_fpr[a->rd], cpu_env, src);
-
+ gen_helper_fcvt_s_w(dest, cpu_env, src);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -388,13 +506,14 @@ static bool trans_fcvt_s_w(DisasContext *ctx, arg_fcvt_s_w *a)
static bool trans_fcvt_s_wu(DisasContext *ctx, arg_fcvt_s_wu *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
TCGv src = get_gpr(ctx, a->rs1, EXT_ZERO);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_s_wu(cpu_fpr[a->rd], cpu_env, src);
-
+ gen_helper_fcvt_s_wu(dest, cpu_env, src);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -403,13 +522,14 @@ static bool trans_fmv_w_x(DisasContext *ctx, arg_fmv_w_x *a)
{
/* NOTE: This was FMV.S.X in an earlier version of the ISA spec! */
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
TCGv src = get_gpr(ctx, a->rs1, EXT_ZERO);
- tcg_gen_extu_tl_i64(cpu_fpr[a->rd], src);
- gen_nanbox_s(cpu_fpr[a->rd], cpu_fpr[a->rd]);
-
+ tcg_gen_extu_tl_i64(dest, src);
+ gen_nanbox_s(dest, dest);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -418,12 +538,13 @@ static bool trans_fcvt_l_s(DisasContext *ctx, arg_fcvt_l_s *a)
{
REQUIRE_64BIT(ctx);
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_l_s(dest, cpu_env, cpu_fpr[a->rs1]);
+ gen_helper_fcvt_l_s(dest, cpu_env, src1);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -432,12 +553,13 @@ static bool trans_fcvt_lu_s(DisasContext *ctx, arg_fcvt_lu_s *a)
{
REQUIRE_64BIT(ctx);
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_lu_s(dest, cpu_env, cpu_fpr[a->rs1]);
+ gen_helper_fcvt_lu_s(dest, cpu_env, src1);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -446,13 +568,14 @@ static bool trans_fcvt_s_l(DisasContext *ctx, arg_fcvt_s_l *a)
{
REQUIRE_64BIT(ctx);
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
TCGv src = get_gpr(ctx, a->rs1, EXT_SIGN);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_s_l(cpu_fpr[a->rd], cpu_env, src);
-
+ gen_helper_fcvt_s_l(dest, cpu_env, src);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -461,13 +584,14 @@ static bool trans_fcvt_s_lu(DisasContext *ctx, arg_fcvt_s_lu *a)
{
REQUIRE_64BIT(ctx);
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVF);
+ REQUIRE_ZFINX_OR_F(ctx);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
TCGv src = get_gpr(ctx, a->rs1, EXT_ZERO);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_s_lu(cpu_fpr[a->rd], cpu_env, src);
-
+ gen_helper_fcvt_s_lu(dest, cpu_env, src);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
diff --git a/target/riscv/internals.h b/target/riscv/internals.h
index 065e8162a2..9f3f3319f2 100644
--- a/target/riscv/internals.h
+++ b/target/riscv/internals.h
@@ -51,8 +51,12 @@ static inline uint64_t nanbox_s(float32 f)
return f | MAKE_64BIT_MASK(32, 32);
}
-static inline float32 check_nanbox_s(uint64_t f)
+static inline float32 check_nanbox_s(CPURISCVState *env, uint64_t f)
{
+ /* Disable nanbox check when enable zfinx */
+ if (RISCV_CPU(env_cpu(env))->cfg.ext_zfinx)
+ return (uint32_t)f;
+
uint64_t mask = MAKE_64BIT_MASK(32, 32);
if (likely((f & mask) == mask)) {
--
2.17.1
^ permalink raw reply related [flat|nested] 30+ messages in thread
* Re: [PATCH 3/6] target/riscv: add support for zfinx
2021-12-24 3:49 ` liweiwei
(?)
@ 2021-12-24 22:26 ` Richard Henderson
2021-12-25 3:24 ` liweiwei
-1 siblings, 1 reply; 30+ messages in thread
From: Richard Henderson @ 2021-12-24 22:26 UTC (permalink / raw)
To: liweiwei, palmer, alistair.francis, bin.meng, qemu-riscv, qemu-devel
Cc: wangjunqiang, lazyparser, ardxwe
On 12/23/21 7:49 PM, liweiwei wrote:
> static bool trans_fsgnj_s(DisasContext *ctx, arg_fsgnj_s *a)
> {
> REQUIRE_FPU;
> - REQUIRE_EXT(ctx, RVF);
> + REQUIRE_ZFINX_OR_F(ctx);
>
> + TCGv_i64 dest = dest_fpr(ctx, a->rd);
> if (a->rs1 == a->rs2) { /* FMOV */
> - gen_check_nanbox_s(cpu_fpr[a->rd], cpu_fpr[a->rs1]);
> + TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
> + if (ctx->ext_zfinx) {
> + gen_nanbox_s(dest, src1);
Sign-extend, not nanbox. Or, since you handle sign-extension in gen_set_gpr_hs, nothing
at all -- just tcg_gen_mov_i64.
> } else { /* FSGNJ */
> - TCGv_i64 rs1 = tcg_temp_new_i64();
> - TCGv_i64 rs2 = tcg_temp_new_i64();
> -
> - gen_check_nanbox_s(rs1, cpu_fpr[a->rs1]);
> - gen_check_nanbox_s(rs2, cpu_fpr[a->rs2]);
> -
> - /* This formulation retains the nanboxing of rs2. */
> - tcg_gen_deposit_i64(cpu_fpr[a->rd], rs2, rs1, 0, 31);
> - tcg_temp_free_i64(rs1);
> - tcg_temp_free_i64(rs2);
> + TCGv_i64 rs1, rs2;
> + if (!ctx->ext_zfinx) {
> + TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
> + TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
> + rs1 = tcg_temp_new_i64();
> + rs2 = tcg_temp_new_i64();
> + gen_check_nanbox_s(rs1, src1);
> + gen_check_nanbox_s(rs2, src2);
> + } else {
> + rs1 = get_fpr_hs(ctx, a->rs1);
> + rs2 = get_fpr_hs(ctx, a->rs2);
> + }
> +
> + /* This formulation retains the nanboxing of rs2 in normal 'F'. */
> + tcg_gen_deposit_i64(dest, rs2, rs1, 0, 31);
> + if (!ctx->ext_zfinx) {
> + tcg_temp_free_i64(rs1);
> + tcg_temp_free_i64(rs2);
> + } else {
> + gen_nanbox_s(dest, dest);
> + }
This is tangled enough that I think you should check zfinx at one higher indent level, and
not do conditional allocate followed by conditional free.
> diff --git a/target/riscv/internals.h b/target/riscv/internals.h
> index 065e8162a2..9f3f3319f2 100644
> --- a/target/riscv/internals.h
> +++ b/target/riscv/internals.h
> @@ -51,8 +51,12 @@ static inline uint64_t nanbox_s(float32 f)
> return f | MAKE_64BIT_MASK(32, 32);
> }
>
> -static inline float32 check_nanbox_s(uint64_t f)
> +static inline float32 check_nanbox_s(CPURISCVState *env, uint64_t f)
> {
> + /* Disable nanbox check when enable zfinx */
> + if (RISCV_CPU(env_cpu(env))->cfg.ext_zfinx)
> + return (uint32_t)f;
> +
Braces.
r~
^ permalink raw reply [flat|nested] 30+ messages in thread
* Re: [PATCH 3/6] target/riscv: add support for zfinx
2021-12-24 22:26 ` Richard Henderson
@ 2021-12-25 3:24 ` liweiwei
0 siblings, 0 replies; 30+ messages in thread
From: liweiwei @ 2021-12-25 3:24 UTC (permalink / raw)
To: Richard Henderson, palmer, alistair.francis, bin.meng,
qemu-riscv, qemu-devel
Cc: wangjunqiang, lazyparser, ardxwe
在 2021/12/25 上午6:26, Richard Henderson 写道:
> On 12/23/21 7:49 PM, liweiwei wrote:
>> static bool trans_fsgnj_s(DisasContext *ctx, arg_fsgnj_s *a)
>> {
>> REQUIRE_FPU;
>> - REQUIRE_EXT(ctx, RVF);
>> + REQUIRE_ZFINX_OR_F(ctx);
>> + TCGv_i64 dest = dest_fpr(ctx, a->rd);
>> if (a->rs1 == a->rs2) { /* FMOV */
>> - gen_check_nanbox_s(cpu_fpr[a->rd], cpu_fpr[a->rs1]);
>> + TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
>> + if (ctx->ext_zfinx) {
>> + gen_nanbox_s(dest, src1);
>
> Sign-extend, not nanbox. Or, since you handle sign-extension in
> gen_set_gpr_hs, nothing at all -- just tcg_gen_mov_i64.
Yeah. I did gen_nanbox_s here is to make the dest nan-boxing just as
other instructions. It's truely unnecessary here. I'll try maximize the
reuse of code here.
>
>> } else { /* FSGNJ */
>> - TCGv_i64 rs1 = tcg_temp_new_i64();
>> - TCGv_i64 rs2 = tcg_temp_new_i64();
>> -
>> - gen_check_nanbox_s(rs1, cpu_fpr[a->rs1]);
>> - gen_check_nanbox_s(rs2, cpu_fpr[a->rs2]);
>> -
>> - /* This formulation retains the nanboxing of rs2. */
>> - tcg_gen_deposit_i64(cpu_fpr[a->rd], rs2, rs1, 0, 31);
>> - tcg_temp_free_i64(rs1);
>> - tcg_temp_free_i64(rs2);
>> + TCGv_i64 rs1, rs2;
>> + if (!ctx->ext_zfinx) {
>> + TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
>> + TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
>> + rs1 = tcg_temp_new_i64();
>> + rs2 = tcg_temp_new_i64();
>> + gen_check_nanbox_s(rs1, src1);
>> + gen_check_nanbox_s(rs2, src2);
>> + } else {
>> + rs1 = get_fpr_hs(ctx, a->rs1);
>> + rs2 = get_fpr_hs(ctx, a->rs2);
>> + }
>> +
>> + /* This formulation retains the nanboxing of rs2 in normal
>> 'F'. */
>> + tcg_gen_deposit_i64(dest, rs2, rs1, 0, 31);
>> + if (!ctx->ext_zfinx) {
>> + tcg_temp_free_i64(rs1);
>> + tcg_temp_free_i64(rs2);
>> + } else {
>> + gen_nanbox_s(dest, dest);
>> + }
>
> This is tangled enough that I think you should check zfinx at one
> higher indent level, and not do conditional allocate followed by
> conditional free.
OK, I'll divide them. By the way I did in this way to maximize the reuse
of code.
>
>> diff --git a/target/riscv/internals.h b/target/riscv/internals.h
>> index 065e8162a2..9f3f3319f2 100644
>> --- a/target/riscv/internals.h
>> +++ b/target/riscv/internals.h
>> @@ -51,8 +51,12 @@ static inline uint64_t nanbox_s(float32 f)
>> return f | MAKE_64BIT_MASK(32, 32);
>> }
>> -static inline float32 check_nanbox_s(uint64_t f)
>> +static inline float32 check_nanbox_s(CPURISCVState *env, uint64_t f)
>> {
>> + /* Disable nanbox check when enable zfinx */
>> + if (RISCV_CPU(env_cpu(env))->cfg.ext_zfinx)
>> + return (uint32_t)f;
>> +
>
> Braces.
>
>
> r~
^ permalink raw reply [flat|nested] 30+ messages in thread
* [PATCH 4/6] target/riscv: add support for zdinx
2021-12-24 3:49 ` liweiwei
@ 2021-12-24 3:49 ` liweiwei
-1 siblings, 0 replies; 30+ messages in thread
From: liweiwei @ 2021-12-24 3:49 UTC (permalink / raw)
To: palmer, alistair.francis, bin.meng, qemu-riscv, qemu-devel
Cc: wangjunqiang, liweiwei, lazyparser, ardxwe
-- update extension check REQUIRE_ZDINX_OR_D
-- update double float point register read/write
Co-authored-by: ardxwe <ardxwe@gmail.com>
Signed-off-by: liweiwei <liweiwei@iscas.ac.cn>
Signed-off-by: wangjunqiang <wangjunqiang@iscas.ac.cn>
---
target/riscv/insn_trans/trans_rvd.c.inc | 252 ++++++++++++++++--------
1 file changed, 174 insertions(+), 78 deletions(-)
diff --git a/target/riscv/insn_trans/trans_rvd.c.inc b/target/riscv/insn_trans/trans_rvd.c.inc
index 64fb0046f7..084a5c5fbe 100644
--- a/target/riscv/insn_trans/trans_rvd.c.inc
+++ b/target/riscv/insn_trans/trans_rvd.c.inc
@@ -18,6 +18,12 @@
* this program. If not, see <http://www.gnu.org/licenses/>.
*/
+#define REQUIRE_ZDINX_OR_D(ctx) do {\
+ if (!ctx->ext_zdinx) {\
+ REQUIRE_EXT(ctx, RVD); \
+ } \
+} while (0)
+
static bool trans_fld(DisasContext *ctx, arg_fld *a)
{
TCGv addr;
@@ -62,10 +68,16 @@ static bool trans_fsd(DisasContext *ctx, arg_fsd *a)
static bool trans_fmadd_d(DisasContext *ctx, arg_fmadd_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_d(ctx, a->rs2);
+ TCGv_i64 src3 = get_fpr_d(ctx, a->rs3);
+
gen_set_rm(ctx, a->rm);
- gen_helper_fmadd_d(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1],
- cpu_fpr[a->rs2], cpu_fpr[a->rs3]);
+ gen_helper_fmadd_d(dest, cpu_env, src1, src2, src3);
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -73,10 +85,16 @@ static bool trans_fmadd_d(DisasContext *ctx, arg_fmadd_d *a)
static bool trans_fmsub_d(DisasContext *ctx, arg_fmsub_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_d(ctx, a->rs2);
+ TCGv_i64 src3 = get_fpr_d(ctx, a->rs3);
+
gen_set_rm(ctx, a->rm);
- gen_helper_fmsub_d(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1],
- cpu_fpr[a->rs2], cpu_fpr[a->rs3]);
+ gen_helper_fmsub_d(dest, cpu_env, src1, src2, src3);
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -84,10 +102,16 @@ static bool trans_fmsub_d(DisasContext *ctx, arg_fmsub_d *a)
static bool trans_fnmsub_d(DisasContext *ctx, arg_fnmsub_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_d(ctx, a->rs2);
+ TCGv_i64 src3 = get_fpr_d(ctx, a->rs3);
+
gen_set_rm(ctx, a->rm);
- gen_helper_fnmsub_d(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1],
- cpu_fpr[a->rs2], cpu_fpr[a->rs3]);
+ gen_helper_fnmsub_d(dest, cpu_env, src1, src2, src3);
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -95,10 +119,16 @@ static bool trans_fnmsub_d(DisasContext *ctx, arg_fnmsub_d *a)
static bool trans_fnmadd_d(DisasContext *ctx, arg_fnmadd_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_d(ctx, a->rs2);
+ TCGv_i64 src3 = get_fpr_d(ctx, a->rs3);
+
gen_set_rm(ctx, a->rm);
- gen_helper_fnmadd_d(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1],
- cpu_fpr[a->rs2], cpu_fpr[a->rs3]);
+ gen_helper_fnmadd_d(dest, cpu_env, src1, src2, src3);
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -106,12 +136,15 @@ static bool trans_fnmadd_d(DisasContext *ctx, arg_fnmadd_d *a)
static bool trans_fadd_d(DisasContext *ctx, arg_fadd_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
- gen_set_rm(ctx, a->rm);
- gen_helper_fadd_d(cpu_fpr[a->rd], cpu_env,
- cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_d(ctx, a->rs2);
+ gen_set_rm(ctx, a->rm);
+ gen_helper_fadd_d(dest, cpu_env, src1, src2);
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -119,12 +152,15 @@ static bool trans_fadd_d(DisasContext *ctx, arg_fadd_d *a)
static bool trans_fsub_d(DisasContext *ctx, arg_fsub_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
- gen_set_rm(ctx, a->rm);
- gen_helper_fsub_d(cpu_fpr[a->rd], cpu_env,
- cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_d(ctx, a->rs2);
+ gen_set_rm(ctx, a->rm);
+ gen_helper_fsub_d(dest, cpu_env, src1, src2);
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -132,12 +168,15 @@ static bool trans_fsub_d(DisasContext *ctx, arg_fsub_d *a)
static bool trans_fmul_d(DisasContext *ctx, arg_fmul_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
- gen_set_rm(ctx, a->rm);
- gen_helper_fmul_d(cpu_fpr[a->rd], cpu_env,
- cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_d(ctx, a->rs2);
+ gen_set_rm(ctx, a->rm);
+ gen_helper_fmul_d(dest, cpu_env, src1, src2);
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -145,12 +184,15 @@ static bool trans_fmul_d(DisasContext *ctx, arg_fmul_d *a)
static bool trans_fdiv_d(DisasContext *ctx, arg_fdiv_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
- gen_set_rm(ctx, a->rm);
- gen_helper_fdiv_d(cpu_fpr[a->rd], cpu_env,
- cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_d(ctx, a->rs2);
+ gen_set_rm(ctx, a->rm);
+ gen_helper_fdiv_d(dest, cpu_env, src1, src2);
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -158,23 +200,34 @@ static bool trans_fdiv_d(DisasContext *ctx, arg_fdiv_d *a)
static bool trans_fsqrt_d(DisasContext *ctx, arg_fsqrt_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
- gen_set_rm(ctx, a->rm);
- gen_helper_fsqrt_d(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1]);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ gen_set_rm(ctx, a->rm);
+ gen_helper_fsqrt_d(dest, cpu_env, src1);
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
static bool trans_fsgnj_d(DisasContext *ctx, arg_fsgnj_d *a)
{
+ REQUIRE_FPU;
+ REQUIRE_ZDINX_OR_D(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_d(ctx, a->rs2);
+
if (a->rs1 == a->rs2) { /* FMOV */
- tcg_gen_mov_i64(cpu_fpr[a->rd], cpu_fpr[a->rs1]);
+ tcg_gen_mov_i64(dest, src1);
} else {
- tcg_gen_deposit_i64(cpu_fpr[a->rd], cpu_fpr[a->rs2],
- cpu_fpr[a->rs1], 0, 63);
+ tcg_gen_deposit_i64(dest, src2,
+ src1, 0, 63);
}
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -182,15 +235,21 @@ static bool trans_fsgnj_d(DisasContext *ctx, arg_fsgnj_d *a)
static bool trans_fsgnjn_d(DisasContext *ctx, arg_fsgnjn_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_d(ctx, a->rs2);
+
if (a->rs1 == a->rs2) { /* FNEG */
- tcg_gen_xori_i64(cpu_fpr[a->rd], cpu_fpr[a->rs1], INT64_MIN);
+ tcg_gen_xori_i64(dest, src1, INT64_MIN);
} else {
TCGv_i64 t0 = tcg_temp_new_i64();
- tcg_gen_not_i64(t0, cpu_fpr[a->rs2]);
- tcg_gen_deposit_i64(cpu_fpr[a->rd], t0, cpu_fpr[a->rs1], 0, 63);
+ tcg_gen_not_i64(t0, src2);
+ tcg_gen_deposit_i64(dest, t0, src1, 0, 63);
tcg_temp_free_i64(t0);
}
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -198,15 +257,21 @@ static bool trans_fsgnjn_d(DisasContext *ctx, arg_fsgnjn_d *a)
static bool trans_fsgnjx_d(DisasContext *ctx, arg_fsgnjx_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_d(ctx, a->rs2);
+
if (a->rs1 == a->rs2) { /* FABS */
- tcg_gen_andi_i64(cpu_fpr[a->rd], cpu_fpr[a->rs1], ~INT64_MIN);
+ tcg_gen_andi_i64(dest, src1, ~INT64_MIN);
} else {
TCGv_i64 t0 = tcg_temp_new_i64();
- tcg_gen_andi_i64(t0, cpu_fpr[a->rs2], INT64_MIN);
- tcg_gen_xor_i64(cpu_fpr[a->rd], cpu_fpr[a->rs1], t0);
+ tcg_gen_andi_i64(t0, src2, INT64_MIN);
+ tcg_gen_xor_i64(dest, src1, t0);
tcg_temp_free_i64(t0);
}
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -214,11 +279,14 @@ static bool trans_fsgnjx_d(DisasContext *ctx, arg_fsgnjx_d *a)
static bool trans_fmin_d(DisasContext *ctx, arg_fmin_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
- gen_helper_fmin_d(cpu_fpr[a->rd], cpu_env,
- cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_d(ctx, a->rs2);
+ gen_helper_fmin_d(dest, cpu_env, src1, src2);
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -226,11 +294,14 @@ static bool trans_fmin_d(DisasContext *ctx, arg_fmin_d *a)
static bool trans_fmax_d(DisasContext *ctx, arg_fmax_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
- gen_helper_fmax_d(cpu_fpr[a->rd], cpu_env,
- cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_d(ctx, a->rs2);
+ gen_helper_fmax_d(dest, cpu_env, src1, src2);
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -238,11 +309,14 @@ static bool trans_fmax_d(DisasContext *ctx, arg_fmax_d *a)
static bool trans_fcvt_s_d(DisasContext *ctx, arg_fcvt_s_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
- gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_s_d(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1]);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ gen_set_rm(ctx, a->rm);
+ gen_helper_fcvt_s_d(dest, cpu_env, src1);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -250,11 +324,14 @@ static bool trans_fcvt_s_d(DisasContext *ctx, arg_fcvt_s_d *a)
static bool trans_fcvt_d_s(DisasContext *ctx, arg_fcvt_d_s *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
- gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_d_s(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1]);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ gen_set_rm(ctx, a->rm);
+ gen_helper_fcvt_d_s(dest, cpu_env, src1);
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -262,11 +339,13 @@ static bool trans_fcvt_d_s(DisasContext *ctx, arg_fcvt_d_s *a)
static bool trans_feq_d(DisasContext *ctx, arg_feq_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_d(ctx, a->rs2);
- gen_helper_feq_d(dest, cpu_env, cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ gen_helper_feq_d(dest, cpu_env, src1, src2);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -274,11 +353,13 @@ static bool trans_feq_d(DisasContext *ctx, arg_feq_d *a)
static bool trans_flt_d(DisasContext *ctx, arg_flt_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_d(ctx, a->rs2);
- gen_helper_flt_d(dest, cpu_env, cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ gen_helper_flt_d(dest, cpu_env, src1, src2);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -286,11 +367,13 @@ static bool trans_flt_d(DisasContext *ctx, arg_flt_d *a)
static bool trans_fle_d(DisasContext *ctx, arg_fle_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_d(ctx, a->rs2);
- gen_helper_fle_d(dest, cpu_env, cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ gen_helper_fle_d(dest, cpu_env, src1, src2);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -298,11 +381,12 @@ static bool trans_fle_d(DisasContext *ctx, arg_fle_d *a)
static bool trans_fclass_d(DisasContext *ctx, arg_fclass_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
- gen_helper_fclass_d(dest, cpu_fpr[a->rs1]);
+ gen_helper_fclass_d(dest, src1);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -310,12 +394,13 @@ static bool trans_fclass_d(DisasContext *ctx, arg_fclass_d *a)
static bool trans_fcvt_w_d(DisasContext *ctx, arg_fcvt_w_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_w_d(dest, cpu_env, cpu_fpr[a->rs1]);
+ gen_helper_fcvt_w_d(dest, cpu_env, src1);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -323,12 +408,13 @@ static bool trans_fcvt_w_d(DisasContext *ctx, arg_fcvt_w_d *a)
static bool trans_fcvt_wu_d(DisasContext *ctx, arg_fcvt_wu_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_wu_d(dest, cpu_env, cpu_fpr[a->rs1]);
+ gen_helper_fcvt_wu_d(dest, cpu_env, src1);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -336,12 +422,14 @@ static bool trans_fcvt_wu_d(DisasContext *ctx, arg_fcvt_wu_d *a)
static bool trans_fcvt_d_w(DisasContext *ctx, arg_fcvt_d_w *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
TCGv src = get_gpr(ctx, a->rs1, EXT_SIGN);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_d_w(cpu_fpr[a->rd], cpu_env, src);
+ gen_helper_fcvt_d_w(dest, cpu_env, src);
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
@@ -350,12 +438,14 @@ static bool trans_fcvt_d_w(DisasContext *ctx, arg_fcvt_d_w *a)
static bool trans_fcvt_d_wu(DisasContext *ctx, arg_fcvt_d_wu *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
TCGv src = get_gpr(ctx, a->rs1, EXT_ZERO);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_d_wu(cpu_fpr[a->rd], cpu_env, src);
+ gen_helper_fcvt_d_wu(dest, cpu_env, src);
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
@@ -365,12 +455,13 @@ static bool trans_fcvt_l_d(DisasContext *ctx, arg_fcvt_l_d *a)
{
REQUIRE_64BIT(ctx);
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_l_d(dest, cpu_env, cpu_fpr[a->rs1]);
+ gen_helper_fcvt_l_d(dest, cpu_env, src1);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -379,12 +470,13 @@ static bool trans_fcvt_lu_d(DisasContext *ctx, arg_fcvt_lu_d *a)
{
REQUIRE_64BIT(ctx);
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_lu_d(dest, cpu_env, cpu_fpr[a->rs1]);
+ gen_helper_fcvt_lu_d(dest, cpu_env, src1);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -407,12 +499,14 @@ static bool trans_fcvt_d_l(DisasContext *ctx, arg_fcvt_d_l *a)
{
REQUIRE_64BIT(ctx);
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
TCGv src = get_gpr(ctx, a->rs1, EXT_SIGN);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_d_l(cpu_fpr[a->rd], cpu_env, src);
+ gen_helper_fcvt_d_l(dest, cpu_env, src);
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
@@ -422,12 +516,14 @@ static bool trans_fcvt_d_lu(DisasContext *ctx, arg_fcvt_d_lu *a)
{
REQUIRE_64BIT(ctx);
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
TCGv src = get_gpr(ctx, a->rs1, EXT_ZERO);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_d_lu(cpu_fpr[a->rd], cpu_env, src);
+ gen_helper_fcvt_d_lu(dest, cpu_env, src);
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
--
2.17.1
^ permalink raw reply related [flat|nested] 30+ messages in thread
* [PATCH 4/6] target/riscv: add support for zdinx
@ 2021-12-24 3:49 ` liweiwei
0 siblings, 0 replies; 30+ messages in thread
From: liweiwei @ 2021-12-24 3:49 UTC (permalink / raw)
To: palmer, alistair.francis, bin.meng, qemu-riscv, qemu-devel
Cc: wangjunqiang, lazyparser, ardxwe, liweiwei
-- update extension check REQUIRE_ZDINX_OR_D
-- update double float point register read/write
Co-authored-by: ardxwe <ardxwe@gmail.com>
Signed-off-by: liweiwei <liweiwei@iscas.ac.cn>
Signed-off-by: wangjunqiang <wangjunqiang@iscas.ac.cn>
---
target/riscv/insn_trans/trans_rvd.c.inc | 252 ++++++++++++++++--------
1 file changed, 174 insertions(+), 78 deletions(-)
diff --git a/target/riscv/insn_trans/trans_rvd.c.inc b/target/riscv/insn_trans/trans_rvd.c.inc
index 64fb0046f7..084a5c5fbe 100644
--- a/target/riscv/insn_trans/trans_rvd.c.inc
+++ b/target/riscv/insn_trans/trans_rvd.c.inc
@@ -18,6 +18,12 @@
* this program. If not, see <http://www.gnu.org/licenses/>.
*/
+#define REQUIRE_ZDINX_OR_D(ctx) do {\
+ if (!ctx->ext_zdinx) {\
+ REQUIRE_EXT(ctx, RVD); \
+ } \
+} while (0)
+
static bool trans_fld(DisasContext *ctx, arg_fld *a)
{
TCGv addr;
@@ -62,10 +68,16 @@ static bool trans_fsd(DisasContext *ctx, arg_fsd *a)
static bool trans_fmadd_d(DisasContext *ctx, arg_fmadd_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_d(ctx, a->rs2);
+ TCGv_i64 src3 = get_fpr_d(ctx, a->rs3);
+
gen_set_rm(ctx, a->rm);
- gen_helper_fmadd_d(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1],
- cpu_fpr[a->rs2], cpu_fpr[a->rs3]);
+ gen_helper_fmadd_d(dest, cpu_env, src1, src2, src3);
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -73,10 +85,16 @@ static bool trans_fmadd_d(DisasContext *ctx, arg_fmadd_d *a)
static bool trans_fmsub_d(DisasContext *ctx, arg_fmsub_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_d(ctx, a->rs2);
+ TCGv_i64 src3 = get_fpr_d(ctx, a->rs3);
+
gen_set_rm(ctx, a->rm);
- gen_helper_fmsub_d(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1],
- cpu_fpr[a->rs2], cpu_fpr[a->rs3]);
+ gen_helper_fmsub_d(dest, cpu_env, src1, src2, src3);
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -84,10 +102,16 @@ static bool trans_fmsub_d(DisasContext *ctx, arg_fmsub_d *a)
static bool trans_fnmsub_d(DisasContext *ctx, arg_fnmsub_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_d(ctx, a->rs2);
+ TCGv_i64 src3 = get_fpr_d(ctx, a->rs3);
+
gen_set_rm(ctx, a->rm);
- gen_helper_fnmsub_d(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1],
- cpu_fpr[a->rs2], cpu_fpr[a->rs3]);
+ gen_helper_fnmsub_d(dest, cpu_env, src1, src2, src3);
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -95,10 +119,16 @@ static bool trans_fnmsub_d(DisasContext *ctx, arg_fnmsub_d *a)
static bool trans_fnmadd_d(DisasContext *ctx, arg_fnmadd_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_d(ctx, a->rs2);
+ TCGv_i64 src3 = get_fpr_d(ctx, a->rs3);
+
gen_set_rm(ctx, a->rm);
- gen_helper_fnmadd_d(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1],
- cpu_fpr[a->rs2], cpu_fpr[a->rs3]);
+ gen_helper_fnmadd_d(dest, cpu_env, src1, src2, src3);
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -106,12 +136,15 @@ static bool trans_fnmadd_d(DisasContext *ctx, arg_fnmadd_d *a)
static bool trans_fadd_d(DisasContext *ctx, arg_fadd_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
- gen_set_rm(ctx, a->rm);
- gen_helper_fadd_d(cpu_fpr[a->rd], cpu_env,
- cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_d(ctx, a->rs2);
+ gen_set_rm(ctx, a->rm);
+ gen_helper_fadd_d(dest, cpu_env, src1, src2);
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -119,12 +152,15 @@ static bool trans_fadd_d(DisasContext *ctx, arg_fadd_d *a)
static bool trans_fsub_d(DisasContext *ctx, arg_fsub_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
- gen_set_rm(ctx, a->rm);
- gen_helper_fsub_d(cpu_fpr[a->rd], cpu_env,
- cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_d(ctx, a->rs2);
+ gen_set_rm(ctx, a->rm);
+ gen_helper_fsub_d(dest, cpu_env, src1, src2);
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -132,12 +168,15 @@ static bool trans_fsub_d(DisasContext *ctx, arg_fsub_d *a)
static bool trans_fmul_d(DisasContext *ctx, arg_fmul_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
- gen_set_rm(ctx, a->rm);
- gen_helper_fmul_d(cpu_fpr[a->rd], cpu_env,
- cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_d(ctx, a->rs2);
+ gen_set_rm(ctx, a->rm);
+ gen_helper_fmul_d(dest, cpu_env, src1, src2);
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -145,12 +184,15 @@ static bool trans_fmul_d(DisasContext *ctx, arg_fmul_d *a)
static bool trans_fdiv_d(DisasContext *ctx, arg_fdiv_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
- gen_set_rm(ctx, a->rm);
- gen_helper_fdiv_d(cpu_fpr[a->rd], cpu_env,
- cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_d(ctx, a->rs2);
+ gen_set_rm(ctx, a->rm);
+ gen_helper_fdiv_d(dest, cpu_env, src1, src2);
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -158,23 +200,34 @@ static bool trans_fdiv_d(DisasContext *ctx, arg_fdiv_d *a)
static bool trans_fsqrt_d(DisasContext *ctx, arg_fsqrt_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
- gen_set_rm(ctx, a->rm);
- gen_helper_fsqrt_d(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1]);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ gen_set_rm(ctx, a->rm);
+ gen_helper_fsqrt_d(dest, cpu_env, src1);
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
static bool trans_fsgnj_d(DisasContext *ctx, arg_fsgnj_d *a)
{
+ REQUIRE_FPU;
+ REQUIRE_ZDINX_OR_D(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_d(ctx, a->rs2);
+
if (a->rs1 == a->rs2) { /* FMOV */
- tcg_gen_mov_i64(cpu_fpr[a->rd], cpu_fpr[a->rs1]);
+ tcg_gen_mov_i64(dest, src1);
} else {
- tcg_gen_deposit_i64(cpu_fpr[a->rd], cpu_fpr[a->rs2],
- cpu_fpr[a->rs1], 0, 63);
+ tcg_gen_deposit_i64(dest, src2,
+ src1, 0, 63);
}
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -182,15 +235,21 @@ static bool trans_fsgnj_d(DisasContext *ctx, arg_fsgnj_d *a)
static bool trans_fsgnjn_d(DisasContext *ctx, arg_fsgnjn_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_d(ctx, a->rs2);
+
if (a->rs1 == a->rs2) { /* FNEG */
- tcg_gen_xori_i64(cpu_fpr[a->rd], cpu_fpr[a->rs1], INT64_MIN);
+ tcg_gen_xori_i64(dest, src1, INT64_MIN);
} else {
TCGv_i64 t0 = tcg_temp_new_i64();
- tcg_gen_not_i64(t0, cpu_fpr[a->rs2]);
- tcg_gen_deposit_i64(cpu_fpr[a->rd], t0, cpu_fpr[a->rs1], 0, 63);
+ tcg_gen_not_i64(t0, src2);
+ tcg_gen_deposit_i64(dest, t0, src1, 0, 63);
tcg_temp_free_i64(t0);
}
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -198,15 +257,21 @@ static bool trans_fsgnjn_d(DisasContext *ctx, arg_fsgnjn_d *a)
static bool trans_fsgnjx_d(DisasContext *ctx, arg_fsgnjx_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_d(ctx, a->rs2);
+
if (a->rs1 == a->rs2) { /* FABS */
- tcg_gen_andi_i64(cpu_fpr[a->rd], cpu_fpr[a->rs1], ~INT64_MIN);
+ tcg_gen_andi_i64(dest, src1, ~INT64_MIN);
} else {
TCGv_i64 t0 = tcg_temp_new_i64();
- tcg_gen_andi_i64(t0, cpu_fpr[a->rs2], INT64_MIN);
- tcg_gen_xor_i64(cpu_fpr[a->rd], cpu_fpr[a->rs1], t0);
+ tcg_gen_andi_i64(t0, src2, INT64_MIN);
+ tcg_gen_xor_i64(dest, src1, t0);
tcg_temp_free_i64(t0);
}
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -214,11 +279,14 @@ static bool trans_fsgnjx_d(DisasContext *ctx, arg_fsgnjx_d *a)
static bool trans_fmin_d(DisasContext *ctx, arg_fmin_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
- gen_helper_fmin_d(cpu_fpr[a->rd], cpu_env,
- cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_d(ctx, a->rs2);
+ gen_helper_fmin_d(dest, cpu_env, src1, src2);
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -226,11 +294,14 @@ static bool trans_fmin_d(DisasContext *ctx, arg_fmin_d *a)
static bool trans_fmax_d(DisasContext *ctx, arg_fmax_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
- gen_helper_fmax_d(cpu_fpr[a->rd], cpu_env,
- cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_d(ctx, a->rs2);
+ gen_helper_fmax_d(dest, cpu_env, src1, src2);
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -238,11 +309,14 @@ static bool trans_fmax_d(DisasContext *ctx, arg_fmax_d *a)
static bool trans_fcvt_s_d(DisasContext *ctx, arg_fcvt_s_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
- gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_s_d(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1]);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ gen_set_rm(ctx, a->rm);
+ gen_helper_fcvt_s_d(dest, cpu_env, src1);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -250,11 +324,14 @@ static bool trans_fcvt_s_d(DisasContext *ctx, arg_fcvt_s_d *a)
static bool trans_fcvt_d_s(DisasContext *ctx, arg_fcvt_d_s *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
- gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_d_s(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1]);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ gen_set_rm(ctx, a->rm);
+ gen_helper_fcvt_d_s(dest, cpu_env, src1);
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -262,11 +339,13 @@ static bool trans_fcvt_d_s(DisasContext *ctx, arg_fcvt_d_s *a)
static bool trans_feq_d(DisasContext *ctx, arg_feq_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_d(ctx, a->rs2);
- gen_helper_feq_d(dest, cpu_env, cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ gen_helper_feq_d(dest, cpu_env, src1, src2);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -274,11 +353,13 @@ static bool trans_feq_d(DisasContext *ctx, arg_feq_d *a)
static bool trans_flt_d(DisasContext *ctx, arg_flt_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_d(ctx, a->rs2);
- gen_helper_flt_d(dest, cpu_env, cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ gen_helper_flt_d(dest, cpu_env, src1, src2);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -286,11 +367,13 @@ static bool trans_flt_d(DisasContext *ctx, arg_flt_d *a)
static bool trans_fle_d(DisasContext *ctx, arg_fle_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_d(ctx, a->rs2);
- gen_helper_fle_d(dest, cpu_env, cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ gen_helper_fle_d(dest, cpu_env, src1, src2);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -298,11 +381,12 @@ static bool trans_fle_d(DisasContext *ctx, arg_fle_d *a)
static bool trans_fclass_d(DisasContext *ctx, arg_fclass_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
- gen_helper_fclass_d(dest, cpu_fpr[a->rs1]);
+ gen_helper_fclass_d(dest, src1);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -310,12 +394,13 @@ static bool trans_fclass_d(DisasContext *ctx, arg_fclass_d *a)
static bool trans_fcvt_w_d(DisasContext *ctx, arg_fcvt_w_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_w_d(dest, cpu_env, cpu_fpr[a->rs1]);
+ gen_helper_fcvt_w_d(dest, cpu_env, src1);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -323,12 +408,13 @@ static bool trans_fcvt_w_d(DisasContext *ctx, arg_fcvt_w_d *a)
static bool trans_fcvt_wu_d(DisasContext *ctx, arg_fcvt_wu_d *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_wu_d(dest, cpu_env, cpu_fpr[a->rs1]);
+ gen_helper_fcvt_wu_d(dest, cpu_env, src1);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -336,12 +422,14 @@ static bool trans_fcvt_wu_d(DisasContext *ctx, arg_fcvt_wu_d *a)
static bool trans_fcvt_d_w(DisasContext *ctx, arg_fcvt_d_w *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
TCGv src = get_gpr(ctx, a->rs1, EXT_SIGN);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_d_w(cpu_fpr[a->rd], cpu_env, src);
+ gen_helper_fcvt_d_w(dest, cpu_env, src);
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
@@ -350,12 +438,14 @@ static bool trans_fcvt_d_w(DisasContext *ctx, arg_fcvt_d_w *a)
static bool trans_fcvt_d_wu(DisasContext *ctx, arg_fcvt_d_wu *a)
{
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
TCGv src = get_gpr(ctx, a->rs1, EXT_ZERO);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_d_wu(cpu_fpr[a->rd], cpu_env, src);
+ gen_helper_fcvt_d_wu(dest, cpu_env, src);
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
@@ -365,12 +455,13 @@ static bool trans_fcvt_l_d(DisasContext *ctx, arg_fcvt_l_d *a)
{
REQUIRE_64BIT(ctx);
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_l_d(dest, cpu_env, cpu_fpr[a->rs1]);
+ gen_helper_fcvt_l_d(dest, cpu_env, src1);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -379,12 +470,13 @@ static bool trans_fcvt_lu_d(DisasContext *ctx, arg_fcvt_lu_d *a)
{
REQUIRE_64BIT(ctx);
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_lu_d(dest, cpu_env, cpu_fpr[a->rs1]);
+ gen_helper_fcvt_lu_d(dest, cpu_env, src1);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -407,12 +499,14 @@ static bool trans_fcvt_d_l(DisasContext *ctx, arg_fcvt_d_l *a)
{
REQUIRE_64BIT(ctx);
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
TCGv src = get_gpr(ctx, a->rs1, EXT_SIGN);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_d_l(cpu_fpr[a->rd], cpu_env, src);
+ gen_helper_fcvt_d_l(dest, cpu_env, src);
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
@@ -422,12 +516,14 @@ static bool trans_fcvt_d_lu(DisasContext *ctx, arg_fcvt_d_lu *a)
{
REQUIRE_64BIT(ctx);
REQUIRE_FPU;
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZDINX_OR_D(ctx);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
TCGv src = get_gpr(ctx, a->rs1, EXT_ZERO);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_d_lu(cpu_fpr[a->rd], cpu_env, src);
+ gen_helper_fcvt_d_lu(dest, cpu_env, src);
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
--
2.17.1
^ permalink raw reply related [flat|nested] 30+ messages in thread
* Re: [PATCH 4/6] target/riscv: add support for zdinx
2021-12-24 3:49 ` liweiwei
(?)
@ 2021-12-24 22:30 ` Richard Henderson
2021-12-25 3:27 ` liweiwei
-1 siblings, 1 reply; 30+ messages in thread
From: Richard Henderson @ 2021-12-24 22:30 UTC (permalink / raw)
To: liweiwei, palmer, alistair.francis, bin.meng, qemu-riscv, qemu-devel
Cc: wangjunqiang, lazyparser, ardxwe
On 12/23/21 7:49 PM, liweiwei wrote:
> static bool trans_fsgnj_d(DisasContext *ctx, arg_fsgnj_d *a)
> {
> + REQUIRE_FPU;
> + REQUIRE_ZDINX_OR_D(ctx);
> +
> + TCGv_i64 dest = dest_fpr(ctx, a->rd);
> + TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
> + TCGv_i64 src2 = get_fpr_d(ctx, a->rs2);
> +
> if (a->rs1 == a->rs2) { /* FMOV */
Applies to the F version as well, but we should not assemble src2 when we don't need it.
> - tcg_gen_mov_i64(cpu_fpr[a->rd], cpu_fpr[a->rs1]);
> + tcg_gen_mov_i64(dest, src1);
I think this can just be dest = get_fpr_d(ctx, a->src1), leaving the final "move" to
gen_set_fpr_d.
r~
^ permalink raw reply [flat|nested] 30+ messages in thread
* Re: [PATCH 4/6] target/riscv: add support for zdinx
2021-12-24 22:30 ` Richard Henderson
@ 2021-12-25 3:27 ` liweiwei
0 siblings, 0 replies; 30+ messages in thread
From: liweiwei @ 2021-12-25 3:27 UTC (permalink / raw)
To: Richard Henderson, palmer, alistair.francis, bin.meng,
qemu-riscv, qemu-devel
Cc: wangjunqiang, lazyparser, ardxwe
在 2021/12/25 上午6:30, Richard Henderson 写道:
> On 12/23/21 7:49 PM, liweiwei wrote:
>> static bool trans_fsgnj_d(DisasContext *ctx, arg_fsgnj_d *a)
>> {
>> + REQUIRE_FPU;
>> + REQUIRE_ZDINX_OR_D(ctx);
>> +
>> + TCGv_i64 dest = dest_fpr(ctx, a->rd);
>> + TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
>> + TCGv_i64 src2 = get_fpr_d(ctx, a->rs2);
>> +
>> if (a->rs1 == a->rs2) { /* FMOV */
>
> Applies to the F version as well, but we should not assemble src2 when
> we don't need it.
OK, I'll fix this.
>
>> - tcg_gen_mov_i64(cpu_fpr[a->rd], cpu_fpr[a->rs1]);
>> + tcg_gen_mov_i64(dest, src1);
>
>
> I think this can just be dest = get_fpr_d(ctx, a->src1), leaving the
> final "move" to gen_set_fpr_d.
OK. I'll update this.
>
>
> r~
^ permalink raw reply [flat|nested] 30+ messages in thread
* [PATCH 5/6] target/riscv: add support for zhinx/zhinxmin
2021-12-24 3:49 ` liweiwei
@ 2021-12-24 3:49 ` liweiwei
-1 siblings, 0 replies; 30+ messages in thread
From: liweiwei @ 2021-12-24 3:49 UTC (permalink / raw)
To: palmer, alistair.francis, bin.meng, qemu-riscv, qemu-devel
Cc: wangjunqiang, liweiwei, lazyparser, ardxwe
- update extension check REQUIRE_ZHINX_OR_ZFH and REQUIRE_ZFH_OR_ZFHMIN_OR_ZHINX_OR_ZHINXMIN
- update half float point register read/write
- disable nanbox_h check
Signed-off-by: liweiwei <liweiwei@iscas.ac.cn>
Signed-off-by: wangjunqiang <wangjunqiang@iscas.ac.cn>
---
target/riscv/fpu_helper.c | 60 ++--
target/riscv/helper.h | 2 +-
target/riscv/insn_trans/trans_rvzfh.c.inc | 342 +++++++++++++++-------
target/riscv/internals.h | 6 +-
4 files changed, 279 insertions(+), 131 deletions(-)
diff --git a/target/riscv/fpu_helper.c b/target/riscv/fpu_helper.c
index 0895fa3c7e..2032800dfe 100644
--- a/target/riscv/fpu_helper.c
+++ b/target/riscv/fpu_helper.c
@@ -89,9 +89,9 @@ void helper_set_rod_rounding_mode(CPURISCVState *env)
static uint64_t do_fmadd_h(CPURISCVState *env, uint64_t rs1, uint64_t rs2,
uint64_t rs3, int flags)
{
- float16 frs1 = check_nanbox_h(rs1);
- float16 frs2 = check_nanbox_h(rs2);
- float16 frs3 = check_nanbox_h(rs3);
+ float16 frs1 = check_nanbox_h(env, rs1);
+ float16 frs2 = check_nanbox_h(env, rs2);
+ float16 frs3 = check_nanbox_h(env, rs3);
return nanbox_h(float16_muladd(frs1, frs2, frs3, flags, &env->fp_status));
}
@@ -416,36 +416,36 @@ target_ulong helper_fclass_d(uint64_t frs1)
uint64_t helper_fadd_h(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
{
- float16 frs1 = check_nanbox_h(rs1);
- float16 frs2 = check_nanbox_h(rs2);
+ float16 frs1 = check_nanbox_h(env, rs1);
+ float16 frs2 = check_nanbox_h(env, rs2);
return nanbox_h(float16_add(frs1, frs2, &env->fp_status));
}
uint64_t helper_fsub_h(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
{
- float16 frs1 = check_nanbox_h(rs1);
- float16 frs2 = check_nanbox_h(rs2);
+ float16 frs1 = check_nanbox_h(env, rs1);
+ float16 frs2 = check_nanbox_h(env, rs2);
return nanbox_h(float16_sub(frs1, frs2, &env->fp_status));
}
uint64_t helper_fmul_h(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
{
- float16 frs1 = check_nanbox_h(rs1);
- float16 frs2 = check_nanbox_h(rs2);
+ float16 frs1 = check_nanbox_h(env, rs1);
+ float16 frs2 = check_nanbox_h(env, rs2);
return nanbox_h(float16_mul(frs1, frs2, &env->fp_status));
}
uint64_t helper_fdiv_h(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
{
- float16 frs1 = check_nanbox_h(rs1);
- float16 frs2 = check_nanbox_h(rs2);
+ float16 frs1 = check_nanbox_h(env, rs1);
+ float16 frs2 = check_nanbox_h(env, rs2);
return nanbox_h(float16_div(frs1, frs2, &env->fp_status));
}
uint64_t helper_fmin_h(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
{
- float16 frs1 = check_nanbox_h(rs1);
- float16 frs2 = check_nanbox_h(rs2);
+ float16 frs1 = check_nanbox_h(env, rs1);
+ float16 frs2 = check_nanbox_h(env, rs2);
return nanbox_h(env->priv_ver < PRIV_VERSION_1_11_0 ?
float16_minnum(frs1, frs2, &env->fp_status) :
float16_minimum_number(frs1, frs2, &env->fp_status));
@@ -453,8 +453,8 @@ uint64_t helper_fmin_h(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
uint64_t helper_fmax_h(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
{
- float16 frs1 = check_nanbox_h(rs1);
- float16 frs2 = check_nanbox_h(rs2);
+ float16 frs1 = check_nanbox_h(env, rs1);
+ float16 frs2 = check_nanbox_h(env, rs2);
return nanbox_h(env->priv_ver < PRIV_VERSION_1_11_0 ?
float16_maxnum(frs1, frs2, &env->fp_status) :
float16_maximum_number(frs1, frs2, &env->fp_status));
@@ -462,58 +462,58 @@ uint64_t helper_fmax_h(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
uint64_t helper_fsqrt_h(CPURISCVState *env, uint64_t rs1)
{
- float16 frs1 = check_nanbox_h(rs1);
+ float16 frs1 = check_nanbox_h(env, rs1);
return nanbox_h(float16_sqrt(frs1, &env->fp_status));
}
target_ulong helper_fle_h(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
{
- float16 frs1 = check_nanbox_h(rs1);
- float16 frs2 = check_nanbox_h(rs2);
+ float16 frs1 = check_nanbox_h(env, rs1);
+ float16 frs2 = check_nanbox_h(env, rs2);
return float16_le(frs1, frs2, &env->fp_status);
}
target_ulong helper_flt_h(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
{
- float16 frs1 = check_nanbox_h(rs1);
- float16 frs2 = check_nanbox_h(rs2);
+ float16 frs1 = check_nanbox_h(env, rs1);
+ float16 frs2 = check_nanbox_h(env, rs2);
return float16_lt(frs1, frs2, &env->fp_status);
}
target_ulong helper_feq_h(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
{
- float16 frs1 = check_nanbox_h(rs1);
- float16 frs2 = check_nanbox_h(rs2);
+ float16 frs1 = check_nanbox_h(env, rs1);
+ float16 frs2 = check_nanbox_h(env, rs2);
return float16_eq_quiet(frs1, frs2, &env->fp_status);
}
-target_ulong helper_fclass_h(uint64_t rs1)
+target_ulong helper_fclass_h(CPURISCVState *env, uint64_t rs1)
{
- float16 frs1 = check_nanbox_h(rs1);
+ float16 frs1 = check_nanbox_h(env, rs1);
return fclass_h(frs1);
}
target_ulong helper_fcvt_w_h(CPURISCVState *env, uint64_t rs1)
{
- float16 frs1 = check_nanbox_h(rs1);
+ float16 frs1 = check_nanbox_h(env, rs1);
return float16_to_int32(frs1, &env->fp_status);
}
target_ulong helper_fcvt_wu_h(CPURISCVState *env, uint64_t rs1)
{
- float16 frs1 = check_nanbox_h(rs1);
+ float16 frs1 = check_nanbox_h(env, rs1);
return (int32_t)float16_to_uint32(frs1, &env->fp_status);
}
target_ulong helper_fcvt_l_h(CPURISCVState *env, uint64_t rs1)
{
- float16 frs1 = check_nanbox_h(rs1);
+ float16 frs1 = check_nanbox_h(env, rs1);
return float16_to_int64(frs1, &env->fp_status);
}
target_ulong helper_fcvt_lu_h(CPURISCVState *env, uint64_t rs1)
{
- float16 frs1 = check_nanbox_h(rs1);
+ float16 frs1 = check_nanbox_h(env, rs1);
return float16_to_uint64(frs1, &env->fp_status);
}
@@ -545,7 +545,7 @@ uint64_t helper_fcvt_h_s(CPURISCVState *env, uint64_t rs1)
uint64_t helper_fcvt_s_h(CPURISCVState *env, uint64_t rs1)
{
- float16 frs1 = check_nanbox_h(rs1);
+ float16 frs1 = check_nanbox_h(env, rs1);
return nanbox_s(float16_to_float32(frs1, true, &env->fp_status));
}
@@ -556,6 +556,6 @@ uint64_t helper_fcvt_h_d(CPURISCVState *env, uint64_t rs1)
uint64_t helper_fcvt_d_h(CPURISCVState *env, uint64_t rs1)
{
- float16 frs1 = check_nanbox_h(rs1);
+ float16 frs1 = check_nanbox_h(env, rs1);
return float16_to_float64(frs1, true, &env->fp_status);
}
diff --git a/target/riscv/helper.h b/target/riscv/helper.h
index 816acca993..9765238ab9 100644
--- a/target/riscv/helper.h
+++ b/target/riscv/helper.h
@@ -90,7 +90,7 @@ DEF_HELPER_FLAGS_2(fcvt_h_w, TCG_CALL_NO_RWG, i64, env, tl)
DEF_HELPER_FLAGS_2(fcvt_h_wu, TCG_CALL_NO_RWG, i64, env, tl)
DEF_HELPER_FLAGS_2(fcvt_h_l, TCG_CALL_NO_RWG, i64, env, tl)
DEF_HELPER_FLAGS_2(fcvt_h_lu, TCG_CALL_NO_RWG, i64, env, tl)
-DEF_HELPER_FLAGS_1(fclass_h, TCG_CALL_NO_RWG_SE, tl, i64)
+DEF_HELPER_FLAGS_2(fclass_h, TCG_CALL_NO_RWG_SE, tl, env, i64)
/* Special functions */
DEF_HELPER_2(csrr, tl, env, int)
diff --git a/target/riscv/insn_trans/trans_rvzfh.c.inc b/target/riscv/insn_trans/trans_rvzfh.c.inc
index 5a7cac8958..5917eb251b 100644
--- a/target/riscv/insn_trans/trans_rvzfh.c.inc
+++ b/target/riscv/insn_trans/trans_rvzfh.c.inc
@@ -22,12 +22,25 @@
} \
} while (0)
+#define REQUIRE_ZHINX_OR_ZFH(ctx) do { \
+ if (!ctx->ext_zhinx && !ctx->ext_zfh) { \
+ return false; \
+ } \
+} while (0)
+
#define REQUIRE_ZFH_OR_ZFHMIN(ctx) do { \
if (!(ctx->ext_zfh || ctx->ext_zfhmin)) { \
return false; \
} \
} while (0)
+#define REQUIRE_ZFH_OR_ZFHMIN_OR_ZHINX_OR_ZHINXMIN(ctx) do { \
+ if (!(ctx->ext_zfh || ctx->ext_zfhmin || \
+ ctx->ext_zhinx || ctx->ext_zhinxmin)) { \
+ return false; \
+ } \
+} while (0)
+
static bool trans_flh(DisasContext *ctx, arg_flh *a)
{
TCGv_i64 dest;
@@ -73,11 +86,16 @@ static bool trans_fsh(DisasContext *ctx, arg_fsh *a)
static bool trans_fmadd_h(DisasContext *ctx, arg_fmadd_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
+ TCGv_i64 src3 = get_fpr_hs(ctx, a->rs3);
gen_set_rm(ctx, a->rm);
- gen_helper_fmadd_h(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1],
- cpu_fpr[a->rs2], cpu_fpr[a->rs3]);
+ gen_helper_fmadd_h(dest, cpu_env, src1, src2, src3);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -85,11 +103,16 @@ static bool trans_fmadd_h(DisasContext *ctx, arg_fmadd_h *a)
static bool trans_fmsub_h(DisasContext *ctx, arg_fmsub_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
+ TCGv_i64 src3 = get_fpr_hs(ctx, a->rs3);
gen_set_rm(ctx, a->rm);
- gen_helper_fmsub_h(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1],
- cpu_fpr[a->rs2], cpu_fpr[a->rs3]);
+ gen_helper_fmsub_h(dest, cpu_env, src1, src2, src3);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -97,11 +120,16 @@ static bool trans_fmsub_h(DisasContext *ctx, arg_fmsub_h *a)
static bool trans_fnmsub_h(DisasContext *ctx, arg_fnmsub_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
+ TCGv_i64 src3 = get_fpr_hs(ctx, a->rs3);
gen_set_rm(ctx, a->rm);
- gen_helper_fnmsub_h(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1],
- cpu_fpr[a->rs2], cpu_fpr[a->rs3]);
+ gen_helper_fnmsub_h(dest, cpu_env, src1, src2, src3);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -109,11 +137,16 @@ static bool trans_fnmsub_h(DisasContext *ctx, arg_fnmsub_h *a)
static bool trans_fnmadd_h(DisasContext *ctx, arg_fnmadd_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
+ TCGv_i64 src3 = get_fpr_hs(ctx, a->rs3);
gen_set_rm(ctx, a->rm);
- gen_helper_fnmadd_h(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1],
- cpu_fpr[a->rs2], cpu_fpr[a->rs3]);
+ gen_helper_fnmadd_h(dest, cpu_env, src1, src2, src3);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -121,11 +154,15 @@ static bool trans_fnmadd_h(DisasContext *ctx, arg_fnmadd_h *a)
static bool trans_fadd_h(DisasContext *ctx, arg_fadd_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
gen_set_rm(ctx, a->rm);
- gen_helper_fadd_h(cpu_fpr[a->rd], cpu_env,
- cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ gen_helper_fadd_h(dest, cpu_env, src1, src2);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -133,11 +170,15 @@ static bool trans_fadd_h(DisasContext *ctx, arg_fadd_h *a)
static bool trans_fsub_h(DisasContext *ctx, arg_fsub_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
gen_set_rm(ctx, a->rm);
- gen_helper_fsub_h(cpu_fpr[a->rd], cpu_env,
- cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ gen_helper_fsub_h(dest, cpu_env, src1, src2);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -145,11 +186,15 @@ static bool trans_fsub_h(DisasContext *ctx, arg_fsub_h *a)
static bool trans_fmul_h(DisasContext *ctx, arg_fmul_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
gen_set_rm(ctx, a->rm);
- gen_helper_fmul_h(cpu_fpr[a->rd], cpu_env,
- cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ gen_helper_fmul_h(dest, cpu_env, src1, src2);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -157,11 +202,15 @@ static bool trans_fmul_h(DisasContext *ctx, arg_fmul_h *a)
static bool trans_fdiv_h(DisasContext *ctx, arg_fdiv_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
gen_set_rm(ctx, a->rm);
- gen_helper_fdiv_h(cpu_fpr[a->rd], cpu_env,
- cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ gen_helper_fdiv_h(dest, cpu_env, src1, src2);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -169,10 +218,14 @@ static bool trans_fdiv_h(DisasContext *ctx, arg_fdiv_h *a)
static bool trans_fsqrt_h(DisasContext *ctx, arg_fsqrt_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
gen_set_rm(ctx, a->rm);
- gen_helper_fsqrt_h(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1]);
+ gen_helper_fsqrt_h(dest, cpu_env, src1);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -180,23 +233,41 @@ static bool trans_fsqrt_h(DisasContext *ctx, arg_fsqrt_h *a)
static bool trans_fsgnj_h(DisasContext *ctx, arg_fsgnj_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
if (a->rs1 == a->rs2) { /* FMOV */
- gen_check_nanbox_h(cpu_fpr[a->rd], cpu_fpr[a->rs1]);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ if (ctx->ext_zfinx) {
+ gen_nanbox_h(dest, src1);
+ } else {
+ gen_check_nanbox_h(dest, src1);
+ }
} else {
- TCGv_i64 rs1 = tcg_temp_new_i64();
- TCGv_i64 rs2 = tcg_temp_new_i64();
-
- gen_check_nanbox_h(rs1, cpu_fpr[a->rs1]);
- gen_check_nanbox_h(rs2, cpu_fpr[a->rs2]);
-
- /* This formulation retains the nanboxing of rs2. */
- tcg_gen_deposit_i64(cpu_fpr[a->rd], rs2, rs1, 0, 15);
- tcg_temp_free_i64(rs1);
- tcg_temp_free_i64(rs2);
+ TCGv_i64 rs1, rs2;
+ if (!ctx->ext_zfinx) {
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
+ rs1 = tcg_temp_new_i64();
+ rs2 = tcg_temp_new_i64();
+ gen_check_nanbox_h(rs1, src1);
+ gen_check_nanbox_h(rs2, src2);
+ } else {
+ rs1 = get_fpr_hs(ctx, a->rs1);
+ rs2 = get_fpr_hs(ctx, a->rs2);
+ }
+
+ /* This formulation retains the nanboxing of rs2 in normal 'Zfh'. */
+ tcg_gen_deposit_i64(dest, rs2, rs1, 0, 15);
+ if (!ctx->ext_zfinx) {
+ tcg_temp_free_i64(rs1);
+ tcg_temp_free_i64(rs2);
+ } else {
+ gen_nanbox_h(dest, dest);
+ }
}
-
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -206,16 +277,27 @@ static bool trans_fsgnjn_h(DisasContext *ctx, arg_fsgnjn_h *a)
TCGv_i64 rs1, rs2, mask;
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
- rs1 = tcg_temp_new_i64();
- gen_check_nanbox_h(rs1, cpu_fpr[a->rs1]);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ if (!ctx->ext_zfinx) {
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ rs1 = tcg_temp_new_i64();
+ gen_check_nanbox_h(rs1, src1);
+ } else {
+ rs1 = get_fpr_hs(ctx, a->rs1);
+ }
if (a->rs1 == a->rs2) { /* FNEG */
- tcg_gen_xori_i64(cpu_fpr[a->rd], rs1, MAKE_64BIT_MASK(15, 1));
+ tcg_gen_xori_i64(dest, rs1, MAKE_64BIT_MASK(15, 1));
} else {
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
rs2 = tcg_temp_new_i64();
- gen_check_nanbox_h(rs2, cpu_fpr[a->rs2]);
+ if (!ctx->ext_zfinx) {
+ gen_check_nanbox_h(rs2, src2);
+ } else {
+ tcg_gen_mov_i64(rs2, src2);
+ }
/*
* Replace bit 15 in rs1 with inverse in rs2.
@@ -224,12 +306,17 @@ static bool trans_fsgnjn_h(DisasContext *ctx, arg_fsgnjn_h *a)
mask = tcg_const_i64(~MAKE_64BIT_MASK(15, 1));
tcg_gen_not_i64(rs2, rs2);
tcg_gen_andc_i64(rs2, rs2, mask);
- tcg_gen_and_i64(rs1, mask, rs1);
- tcg_gen_or_i64(cpu_fpr[a->rd], rs1, rs2);
+ tcg_gen_and_i64(dest, mask, rs1);
+ tcg_gen_or_i64(dest, dest, rs2);
tcg_temp_free_i64(mask);
tcg_temp_free_i64(rs2);
}
+ if (ctx->ext_zfinx) {
+ gen_nanbox_h(dest, dest);
+ } else {
+ tcg_temp_free_i64(rs1);
+ }
mark_fs_dirty(ctx);
return true;
}
@@ -239,27 +326,44 @@ static bool trans_fsgnjx_h(DisasContext *ctx, arg_fsgnjx_h *a)
TCGv_i64 rs1, rs2;
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
- rs1 = tcg_temp_new_i64();
- gen_check_nanbox_s(rs1, cpu_fpr[a->rs1]);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ if (!ctx->ext_zfinx) {
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ rs1 = tcg_temp_new_i64();
+ gen_check_nanbox_h(rs1, src1);
+ } else {
+ rs1 = get_fpr_hs(ctx, a->rs1);
+ }
if (a->rs1 == a->rs2) { /* FABS */
- tcg_gen_andi_i64(cpu_fpr[a->rd], rs1, ~MAKE_64BIT_MASK(15, 1));
+ tcg_gen_andi_i64(dest, rs1, ~MAKE_64BIT_MASK(15, 1));
} else {
- rs2 = tcg_temp_new_i64();
- gen_check_nanbox_s(rs2, cpu_fpr[a->rs2]);
+ if (!ctx->ext_zfinx) {
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
+ rs2 = tcg_temp_new_i64();
+ gen_check_nanbox_h(rs2, src2);
+ } else {
+ rs2 = get_fpr_hs(ctx, a->rs2);
+ }
/*
* Xor bit 15 in rs1 with that in rs2.
* This formulation retains the nanboxing of rs1.
*/
- tcg_gen_andi_i64(rs2, rs2, MAKE_64BIT_MASK(15, 1));
- tcg_gen_xor_i64(cpu_fpr[a->rd], rs1, rs2);
+ tcg_gen_andi_i64(dest, rs2, MAKE_64BIT_MASK(15, 1));
+ tcg_gen_xor_i64(dest, rs1, dest);
- tcg_temp_free_i64(rs2);
+ if (!ctx->ext_zfinx) {
+ tcg_temp_free_i64(rs2);
+ }
+ }
+ if (ctx->ext_zfinx) {
+ gen_nanbox_h(dest, dest);
+ } else {
+ tcg_temp_free_i64(rs1);
}
-
mark_fs_dirty(ctx);
return true;
}
@@ -267,10 +371,14 @@ static bool trans_fsgnjx_h(DisasContext *ctx, arg_fsgnjx_h *a)
static bool trans_fmin_h(DisasContext *ctx, arg_fmin_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
- gen_helper_fmin_h(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1],
- cpu_fpr[a->rs2]);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
+
+ gen_helper_fmin_h(dest, cpu_env, src1, src2);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -278,10 +386,14 @@ static bool trans_fmin_h(DisasContext *ctx, arg_fmin_h *a)
static bool trans_fmax_h(DisasContext *ctx, arg_fmax_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
- gen_helper_fmax_h(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1],
- cpu_fpr[a->rs2]);
+ gen_helper_fmax_h(dest, cpu_env, src1, src2);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -289,10 +401,14 @@ static bool trans_fmax_h(DisasContext *ctx, arg_fmax_h *a)
static bool trans_fcvt_s_h(DisasContext *ctx, arg_fcvt_s_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH_OR_ZFHMIN(ctx);
+ REQUIRE_ZFH_OR_ZFHMIN_OR_ZHINX_OR_ZHINXMIN(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_s_h(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1]);
+ gen_helper_fcvt_s_h(dest, cpu_env, src1);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
@@ -302,26 +418,32 @@ static bool trans_fcvt_s_h(DisasContext *ctx, arg_fcvt_s_h *a)
static bool trans_fcvt_d_h(DisasContext *ctx, arg_fcvt_d_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH_OR_ZFHMIN(ctx);
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZFH_OR_ZFHMIN_OR_ZHINX_OR_ZHINXMIN(ctx);
+ REQUIRE_ZDINX_OR_D(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_d_h(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1]);
+ gen_helper_fcvt_d_h(dest, cpu_env, src1);
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
-
return true;
}
static bool trans_fcvt_h_s(DisasContext *ctx, arg_fcvt_h_s *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH_OR_ZFHMIN(ctx);
+ REQUIRE_ZFH_OR_ZFHMIN_OR_ZHINX_OR_ZHINXMIN(ctx);
- gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_h_s(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1]);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ gen_set_rm(ctx, a->rm);
+ gen_helper_fcvt_h_s(dest, cpu_env, src1);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
@@ -330,12 +452,15 @@ static bool trans_fcvt_h_s(DisasContext *ctx, arg_fcvt_h_s *a)
static bool trans_fcvt_h_d(DisasContext *ctx, arg_fcvt_h_d *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH_OR_ZFHMIN(ctx);
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZFH_OR_ZFHMIN_OR_ZHINX_OR_ZHINXMIN(ctx);
+ REQUIRE_ZDINX_OR_D(ctx);
- gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_h_d(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1]);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ gen_set_rm(ctx, a->rm);
+ gen_helper_fcvt_h_d(dest, cpu_env, src1);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
@@ -344,11 +469,13 @@ static bool trans_fcvt_h_d(DisasContext *ctx, arg_fcvt_h_d *a)
static bool trans_feq_h(DisasContext *ctx, arg_feq_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
- gen_helper_feq_h(dest, cpu_env, cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ gen_helper_feq_h(dest, cpu_env, src1, src2);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -356,11 +483,13 @@ static bool trans_feq_h(DisasContext *ctx, arg_feq_h *a)
static bool trans_flt_h(DisasContext *ctx, arg_flt_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
- gen_helper_flt_h(dest, cpu_env, cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ gen_helper_flt_h(dest, cpu_env, src1, src2);
gen_set_gpr(ctx, a->rd, dest);
return true;
@@ -369,11 +498,13 @@ static bool trans_flt_h(DisasContext *ctx, arg_flt_h *a)
static bool trans_fle_h(DisasContext *ctx, arg_fle_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
- gen_helper_fle_h(dest, cpu_env, cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ gen_helper_fle_h(dest, cpu_env, src1, src2);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -381,11 +512,12 @@ static bool trans_fle_h(DisasContext *ctx, arg_fle_h *a)
static bool trans_fclass_h(DisasContext *ctx, arg_fclass_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
- gen_helper_fclass_h(dest, cpu_fpr[a->rs1]);
+ gen_helper_fclass_h(dest, cpu_env, src1);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -393,12 +525,13 @@ static bool trans_fclass_h(DisasContext *ctx, arg_fclass_h *a)
static bool trans_fcvt_w_h(DisasContext *ctx, arg_fcvt_w_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_w_h(dest, cpu_env, cpu_fpr[a->rs1]);
+ gen_helper_fcvt_w_h(dest, cpu_env, src1);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -406,12 +539,13 @@ static bool trans_fcvt_w_h(DisasContext *ctx, arg_fcvt_w_h *a)
static bool trans_fcvt_wu_h(DisasContext *ctx, arg_fcvt_wu_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_wu_h(dest, cpu_env, cpu_fpr[a->rs1]);
+ gen_helper_fcvt_wu_h(dest, cpu_env, src1);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -419,12 +553,14 @@ static bool trans_fcvt_wu_h(DisasContext *ctx, arg_fcvt_wu_h *a)
static bool trans_fcvt_h_w(DisasContext *ctx, arg_fcvt_h_w *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
TCGv t0 = get_gpr(ctx, a->rs1, EXT_SIGN);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_h_w(cpu_fpr[a->rd], cpu_env, t0);
+ gen_helper_fcvt_h_w(dest, cpu_env, t0);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
@@ -433,12 +569,14 @@ static bool trans_fcvt_h_w(DisasContext *ctx, arg_fcvt_h_w *a)
static bool trans_fcvt_h_wu(DisasContext *ctx, arg_fcvt_h_wu *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
TCGv t0 = get_gpr(ctx, a->rs1, EXT_SIGN);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_h_wu(cpu_fpr[a->rd], cpu_env, t0);
+ gen_helper_fcvt_h_wu(dest, cpu_env, t0);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
@@ -482,12 +620,13 @@ static bool trans_fcvt_l_h(DisasContext *ctx, arg_fcvt_l_h *a)
{
REQUIRE_64BIT(ctx);
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_l_h(dest, cpu_env, cpu_fpr[a->rs1]);
+ gen_helper_fcvt_l_h(dest, cpu_env, src1);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -496,12 +635,13 @@ static bool trans_fcvt_lu_h(DisasContext *ctx, arg_fcvt_lu_h *a)
{
REQUIRE_64BIT(ctx);
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_lu_h(dest, cpu_env, cpu_fpr[a->rs1]);
+ gen_helper_fcvt_lu_h(dest, cpu_env, src1);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -510,12 +650,14 @@ static bool trans_fcvt_h_l(DisasContext *ctx, arg_fcvt_h_l *a)
{
REQUIRE_64BIT(ctx);
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
TCGv t0 = get_gpr(ctx, a->rs1, EXT_SIGN);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_h_l(cpu_fpr[a->rd], cpu_env, t0);
+ gen_helper_fcvt_h_l(dest, cpu_env, t0);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
@@ -525,12 +667,14 @@ static bool trans_fcvt_h_lu(DisasContext *ctx, arg_fcvt_h_lu *a)
{
REQUIRE_64BIT(ctx);
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
TCGv t0 = get_gpr(ctx, a->rs1, EXT_SIGN);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_h_lu(cpu_fpr[a->rd], cpu_env, t0);
+ gen_helper_fcvt_h_lu(dest, cpu_env, t0);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
diff --git a/target/riscv/internals.h b/target/riscv/internals.h
index 9f3f3319f2..9f046e26bb 100644
--- a/target/riscv/internals.h
+++ b/target/riscv/internals.h
@@ -71,8 +71,12 @@ static inline uint64_t nanbox_h(float16 f)
return f | MAKE_64BIT_MASK(16, 48);
}
-static inline float16 check_nanbox_h(uint64_t f)
+static inline float16 check_nanbox_h(CPURISCVState *env, uint64_t f)
{
+ /* Disable nanbox check when enable zfinx */
+ if (RISCV_CPU(env_cpu(env))->cfg.ext_zfinx)
+ return (uint16_t)f;
+
uint64_t mask = MAKE_64BIT_MASK(16, 48);
if (likely((f & mask) == mask)) {
--
2.17.1
^ permalink raw reply related [flat|nested] 30+ messages in thread
* [PATCH 5/6] target/riscv: add support for zhinx/zhinxmin
@ 2021-12-24 3:49 ` liweiwei
0 siblings, 0 replies; 30+ messages in thread
From: liweiwei @ 2021-12-24 3:49 UTC (permalink / raw)
To: palmer, alistair.francis, bin.meng, qemu-riscv, qemu-devel
Cc: wangjunqiang, lazyparser, ardxwe, liweiwei
- update extension check REQUIRE_ZHINX_OR_ZFH and REQUIRE_ZFH_OR_ZFHMIN_OR_ZHINX_OR_ZHINXMIN
- update half float point register read/write
- disable nanbox_h check
Signed-off-by: liweiwei <liweiwei@iscas.ac.cn>
Signed-off-by: wangjunqiang <wangjunqiang@iscas.ac.cn>
---
target/riscv/fpu_helper.c | 60 ++--
target/riscv/helper.h | 2 +-
target/riscv/insn_trans/trans_rvzfh.c.inc | 342 +++++++++++++++-------
target/riscv/internals.h | 6 +-
4 files changed, 279 insertions(+), 131 deletions(-)
diff --git a/target/riscv/fpu_helper.c b/target/riscv/fpu_helper.c
index 0895fa3c7e..2032800dfe 100644
--- a/target/riscv/fpu_helper.c
+++ b/target/riscv/fpu_helper.c
@@ -89,9 +89,9 @@ void helper_set_rod_rounding_mode(CPURISCVState *env)
static uint64_t do_fmadd_h(CPURISCVState *env, uint64_t rs1, uint64_t rs2,
uint64_t rs3, int flags)
{
- float16 frs1 = check_nanbox_h(rs1);
- float16 frs2 = check_nanbox_h(rs2);
- float16 frs3 = check_nanbox_h(rs3);
+ float16 frs1 = check_nanbox_h(env, rs1);
+ float16 frs2 = check_nanbox_h(env, rs2);
+ float16 frs3 = check_nanbox_h(env, rs3);
return nanbox_h(float16_muladd(frs1, frs2, frs3, flags, &env->fp_status));
}
@@ -416,36 +416,36 @@ target_ulong helper_fclass_d(uint64_t frs1)
uint64_t helper_fadd_h(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
{
- float16 frs1 = check_nanbox_h(rs1);
- float16 frs2 = check_nanbox_h(rs2);
+ float16 frs1 = check_nanbox_h(env, rs1);
+ float16 frs2 = check_nanbox_h(env, rs2);
return nanbox_h(float16_add(frs1, frs2, &env->fp_status));
}
uint64_t helper_fsub_h(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
{
- float16 frs1 = check_nanbox_h(rs1);
- float16 frs2 = check_nanbox_h(rs2);
+ float16 frs1 = check_nanbox_h(env, rs1);
+ float16 frs2 = check_nanbox_h(env, rs2);
return nanbox_h(float16_sub(frs1, frs2, &env->fp_status));
}
uint64_t helper_fmul_h(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
{
- float16 frs1 = check_nanbox_h(rs1);
- float16 frs2 = check_nanbox_h(rs2);
+ float16 frs1 = check_nanbox_h(env, rs1);
+ float16 frs2 = check_nanbox_h(env, rs2);
return nanbox_h(float16_mul(frs1, frs2, &env->fp_status));
}
uint64_t helper_fdiv_h(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
{
- float16 frs1 = check_nanbox_h(rs1);
- float16 frs2 = check_nanbox_h(rs2);
+ float16 frs1 = check_nanbox_h(env, rs1);
+ float16 frs2 = check_nanbox_h(env, rs2);
return nanbox_h(float16_div(frs1, frs2, &env->fp_status));
}
uint64_t helper_fmin_h(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
{
- float16 frs1 = check_nanbox_h(rs1);
- float16 frs2 = check_nanbox_h(rs2);
+ float16 frs1 = check_nanbox_h(env, rs1);
+ float16 frs2 = check_nanbox_h(env, rs2);
return nanbox_h(env->priv_ver < PRIV_VERSION_1_11_0 ?
float16_minnum(frs1, frs2, &env->fp_status) :
float16_minimum_number(frs1, frs2, &env->fp_status));
@@ -453,8 +453,8 @@ uint64_t helper_fmin_h(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
uint64_t helper_fmax_h(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
{
- float16 frs1 = check_nanbox_h(rs1);
- float16 frs2 = check_nanbox_h(rs2);
+ float16 frs1 = check_nanbox_h(env, rs1);
+ float16 frs2 = check_nanbox_h(env, rs2);
return nanbox_h(env->priv_ver < PRIV_VERSION_1_11_0 ?
float16_maxnum(frs1, frs2, &env->fp_status) :
float16_maximum_number(frs1, frs2, &env->fp_status));
@@ -462,58 +462,58 @@ uint64_t helper_fmax_h(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
uint64_t helper_fsqrt_h(CPURISCVState *env, uint64_t rs1)
{
- float16 frs1 = check_nanbox_h(rs1);
+ float16 frs1 = check_nanbox_h(env, rs1);
return nanbox_h(float16_sqrt(frs1, &env->fp_status));
}
target_ulong helper_fle_h(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
{
- float16 frs1 = check_nanbox_h(rs1);
- float16 frs2 = check_nanbox_h(rs2);
+ float16 frs1 = check_nanbox_h(env, rs1);
+ float16 frs2 = check_nanbox_h(env, rs2);
return float16_le(frs1, frs2, &env->fp_status);
}
target_ulong helper_flt_h(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
{
- float16 frs1 = check_nanbox_h(rs1);
- float16 frs2 = check_nanbox_h(rs2);
+ float16 frs1 = check_nanbox_h(env, rs1);
+ float16 frs2 = check_nanbox_h(env, rs2);
return float16_lt(frs1, frs2, &env->fp_status);
}
target_ulong helper_feq_h(CPURISCVState *env, uint64_t rs1, uint64_t rs2)
{
- float16 frs1 = check_nanbox_h(rs1);
- float16 frs2 = check_nanbox_h(rs2);
+ float16 frs1 = check_nanbox_h(env, rs1);
+ float16 frs2 = check_nanbox_h(env, rs2);
return float16_eq_quiet(frs1, frs2, &env->fp_status);
}
-target_ulong helper_fclass_h(uint64_t rs1)
+target_ulong helper_fclass_h(CPURISCVState *env, uint64_t rs1)
{
- float16 frs1 = check_nanbox_h(rs1);
+ float16 frs1 = check_nanbox_h(env, rs1);
return fclass_h(frs1);
}
target_ulong helper_fcvt_w_h(CPURISCVState *env, uint64_t rs1)
{
- float16 frs1 = check_nanbox_h(rs1);
+ float16 frs1 = check_nanbox_h(env, rs1);
return float16_to_int32(frs1, &env->fp_status);
}
target_ulong helper_fcvt_wu_h(CPURISCVState *env, uint64_t rs1)
{
- float16 frs1 = check_nanbox_h(rs1);
+ float16 frs1 = check_nanbox_h(env, rs1);
return (int32_t)float16_to_uint32(frs1, &env->fp_status);
}
target_ulong helper_fcvt_l_h(CPURISCVState *env, uint64_t rs1)
{
- float16 frs1 = check_nanbox_h(rs1);
+ float16 frs1 = check_nanbox_h(env, rs1);
return float16_to_int64(frs1, &env->fp_status);
}
target_ulong helper_fcvt_lu_h(CPURISCVState *env, uint64_t rs1)
{
- float16 frs1 = check_nanbox_h(rs1);
+ float16 frs1 = check_nanbox_h(env, rs1);
return float16_to_uint64(frs1, &env->fp_status);
}
@@ -545,7 +545,7 @@ uint64_t helper_fcvt_h_s(CPURISCVState *env, uint64_t rs1)
uint64_t helper_fcvt_s_h(CPURISCVState *env, uint64_t rs1)
{
- float16 frs1 = check_nanbox_h(rs1);
+ float16 frs1 = check_nanbox_h(env, rs1);
return nanbox_s(float16_to_float32(frs1, true, &env->fp_status));
}
@@ -556,6 +556,6 @@ uint64_t helper_fcvt_h_d(CPURISCVState *env, uint64_t rs1)
uint64_t helper_fcvt_d_h(CPURISCVState *env, uint64_t rs1)
{
- float16 frs1 = check_nanbox_h(rs1);
+ float16 frs1 = check_nanbox_h(env, rs1);
return float16_to_float64(frs1, true, &env->fp_status);
}
diff --git a/target/riscv/helper.h b/target/riscv/helper.h
index 816acca993..9765238ab9 100644
--- a/target/riscv/helper.h
+++ b/target/riscv/helper.h
@@ -90,7 +90,7 @@ DEF_HELPER_FLAGS_2(fcvt_h_w, TCG_CALL_NO_RWG, i64, env, tl)
DEF_HELPER_FLAGS_2(fcvt_h_wu, TCG_CALL_NO_RWG, i64, env, tl)
DEF_HELPER_FLAGS_2(fcvt_h_l, TCG_CALL_NO_RWG, i64, env, tl)
DEF_HELPER_FLAGS_2(fcvt_h_lu, TCG_CALL_NO_RWG, i64, env, tl)
-DEF_HELPER_FLAGS_1(fclass_h, TCG_CALL_NO_RWG_SE, tl, i64)
+DEF_HELPER_FLAGS_2(fclass_h, TCG_CALL_NO_RWG_SE, tl, env, i64)
/* Special functions */
DEF_HELPER_2(csrr, tl, env, int)
diff --git a/target/riscv/insn_trans/trans_rvzfh.c.inc b/target/riscv/insn_trans/trans_rvzfh.c.inc
index 5a7cac8958..5917eb251b 100644
--- a/target/riscv/insn_trans/trans_rvzfh.c.inc
+++ b/target/riscv/insn_trans/trans_rvzfh.c.inc
@@ -22,12 +22,25 @@
} \
} while (0)
+#define REQUIRE_ZHINX_OR_ZFH(ctx) do { \
+ if (!ctx->ext_zhinx && !ctx->ext_zfh) { \
+ return false; \
+ } \
+} while (0)
+
#define REQUIRE_ZFH_OR_ZFHMIN(ctx) do { \
if (!(ctx->ext_zfh || ctx->ext_zfhmin)) { \
return false; \
} \
} while (0)
+#define REQUIRE_ZFH_OR_ZFHMIN_OR_ZHINX_OR_ZHINXMIN(ctx) do { \
+ if (!(ctx->ext_zfh || ctx->ext_zfhmin || \
+ ctx->ext_zhinx || ctx->ext_zhinxmin)) { \
+ return false; \
+ } \
+} while (0)
+
static bool trans_flh(DisasContext *ctx, arg_flh *a)
{
TCGv_i64 dest;
@@ -73,11 +86,16 @@ static bool trans_fsh(DisasContext *ctx, arg_fsh *a)
static bool trans_fmadd_h(DisasContext *ctx, arg_fmadd_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
+ TCGv_i64 src3 = get_fpr_hs(ctx, a->rs3);
gen_set_rm(ctx, a->rm);
- gen_helper_fmadd_h(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1],
- cpu_fpr[a->rs2], cpu_fpr[a->rs3]);
+ gen_helper_fmadd_h(dest, cpu_env, src1, src2, src3);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -85,11 +103,16 @@ static bool trans_fmadd_h(DisasContext *ctx, arg_fmadd_h *a)
static bool trans_fmsub_h(DisasContext *ctx, arg_fmsub_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
+ TCGv_i64 src3 = get_fpr_hs(ctx, a->rs3);
gen_set_rm(ctx, a->rm);
- gen_helper_fmsub_h(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1],
- cpu_fpr[a->rs2], cpu_fpr[a->rs3]);
+ gen_helper_fmsub_h(dest, cpu_env, src1, src2, src3);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -97,11 +120,16 @@ static bool trans_fmsub_h(DisasContext *ctx, arg_fmsub_h *a)
static bool trans_fnmsub_h(DisasContext *ctx, arg_fnmsub_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
+ TCGv_i64 src3 = get_fpr_hs(ctx, a->rs3);
gen_set_rm(ctx, a->rm);
- gen_helper_fnmsub_h(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1],
- cpu_fpr[a->rs2], cpu_fpr[a->rs3]);
+ gen_helper_fnmsub_h(dest, cpu_env, src1, src2, src3);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -109,11 +137,16 @@ static bool trans_fnmsub_h(DisasContext *ctx, arg_fnmsub_h *a)
static bool trans_fnmadd_h(DisasContext *ctx, arg_fnmadd_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
+ TCGv_i64 src3 = get_fpr_hs(ctx, a->rs3);
gen_set_rm(ctx, a->rm);
- gen_helper_fnmadd_h(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1],
- cpu_fpr[a->rs2], cpu_fpr[a->rs3]);
+ gen_helper_fnmadd_h(dest, cpu_env, src1, src2, src3);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -121,11 +154,15 @@ static bool trans_fnmadd_h(DisasContext *ctx, arg_fnmadd_h *a)
static bool trans_fadd_h(DisasContext *ctx, arg_fadd_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
gen_set_rm(ctx, a->rm);
- gen_helper_fadd_h(cpu_fpr[a->rd], cpu_env,
- cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ gen_helper_fadd_h(dest, cpu_env, src1, src2);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -133,11 +170,15 @@ static bool trans_fadd_h(DisasContext *ctx, arg_fadd_h *a)
static bool trans_fsub_h(DisasContext *ctx, arg_fsub_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
gen_set_rm(ctx, a->rm);
- gen_helper_fsub_h(cpu_fpr[a->rd], cpu_env,
- cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ gen_helper_fsub_h(dest, cpu_env, src1, src2);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -145,11 +186,15 @@ static bool trans_fsub_h(DisasContext *ctx, arg_fsub_h *a)
static bool trans_fmul_h(DisasContext *ctx, arg_fmul_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
gen_set_rm(ctx, a->rm);
- gen_helper_fmul_h(cpu_fpr[a->rd], cpu_env,
- cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ gen_helper_fmul_h(dest, cpu_env, src1, src2);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -157,11 +202,15 @@ static bool trans_fmul_h(DisasContext *ctx, arg_fmul_h *a)
static bool trans_fdiv_h(DisasContext *ctx, arg_fdiv_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
gen_set_rm(ctx, a->rm);
- gen_helper_fdiv_h(cpu_fpr[a->rd], cpu_env,
- cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ gen_helper_fdiv_h(dest, cpu_env, src1, src2);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -169,10 +218,14 @@ static bool trans_fdiv_h(DisasContext *ctx, arg_fdiv_h *a)
static bool trans_fsqrt_h(DisasContext *ctx, arg_fsqrt_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
gen_set_rm(ctx, a->rm);
- gen_helper_fsqrt_h(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1]);
+ gen_helper_fsqrt_h(dest, cpu_env, src1);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -180,23 +233,41 @@ static bool trans_fsqrt_h(DisasContext *ctx, arg_fsqrt_h *a)
static bool trans_fsgnj_h(DisasContext *ctx, arg_fsgnj_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
if (a->rs1 == a->rs2) { /* FMOV */
- gen_check_nanbox_h(cpu_fpr[a->rd], cpu_fpr[a->rs1]);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ if (ctx->ext_zfinx) {
+ gen_nanbox_h(dest, src1);
+ } else {
+ gen_check_nanbox_h(dest, src1);
+ }
} else {
- TCGv_i64 rs1 = tcg_temp_new_i64();
- TCGv_i64 rs2 = tcg_temp_new_i64();
-
- gen_check_nanbox_h(rs1, cpu_fpr[a->rs1]);
- gen_check_nanbox_h(rs2, cpu_fpr[a->rs2]);
-
- /* This formulation retains the nanboxing of rs2. */
- tcg_gen_deposit_i64(cpu_fpr[a->rd], rs2, rs1, 0, 15);
- tcg_temp_free_i64(rs1);
- tcg_temp_free_i64(rs2);
+ TCGv_i64 rs1, rs2;
+ if (!ctx->ext_zfinx) {
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
+ rs1 = tcg_temp_new_i64();
+ rs2 = tcg_temp_new_i64();
+ gen_check_nanbox_h(rs1, src1);
+ gen_check_nanbox_h(rs2, src2);
+ } else {
+ rs1 = get_fpr_hs(ctx, a->rs1);
+ rs2 = get_fpr_hs(ctx, a->rs2);
+ }
+
+ /* This formulation retains the nanboxing of rs2 in normal 'Zfh'. */
+ tcg_gen_deposit_i64(dest, rs2, rs1, 0, 15);
+ if (!ctx->ext_zfinx) {
+ tcg_temp_free_i64(rs1);
+ tcg_temp_free_i64(rs2);
+ } else {
+ gen_nanbox_h(dest, dest);
+ }
}
-
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -206,16 +277,27 @@ static bool trans_fsgnjn_h(DisasContext *ctx, arg_fsgnjn_h *a)
TCGv_i64 rs1, rs2, mask;
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
- rs1 = tcg_temp_new_i64();
- gen_check_nanbox_h(rs1, cpu_fpr[a->rs1]);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ if (!ctx->ext_zfinx) {
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ rs1 = tcg_temp_new_i64();
+ gen_check_nanbox_h(rs1, src1);
+ } else {
+ rs1 = get_fpr_hs(ctx, a->rs1);
+ }
if (a->rs1 == a->rs2) { /* FNEG */
- tcg_gen_xori_i64(cpu_fpr[a->rd], rs1, MAKE_64BIT_MASK(15, 1));
+ tcg_gen_xori_i64(dest, rs1, MAKE_64BIT_MASK(15, 1));
} else {
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
rs2 = tcg_temp_new_i64();
- gen_check_nanbox_h(rs2, cpu_fpr[a->rs2]);
+ if (!ctx->ext_zfinx) {
+ gen_check_nanbox_h(rs2, src2);
+ } else {
+ tcg_gen_mov_i64(rs2, src2);
+ }
/*
* Replace bit 15 in rs1 with inverse in rs2.
@@ -224,12 +306,17 @@ static bool trans_fsgnjn_h(DisasContext *ctx, arg_fsgnjn_h *a)
mask = tcg_const_i64(~MAKE_64BIT_MASK(15, 1));
tcg_gen_not_i64(rs2, rs2);
tcg_gen_andc_i64(rs2, rs2, mask);
- tcg_gen_and_i64(rs1, mask, rs1);
- tcg_gen_or_i64(cpu_fpr[a->rd], rs1, rs2);
+ tcg_gen_and_i64(dest, mask, rs1);
+ tcg_gen_or_i64(dest, dest, rs2);
tcg_temp_free_i64(mask);
tcg_temp_free_i64(rs2);
}
+ if (ctx->ext_zfinx) {
+ gen_nanbox_h(dest, dest);
+ } else {
+ tcg_temp_free_i64(rs1);
+ }
mark_fs_dirty(ctx);
return true;
}
@@ -239,27 +326,44 @@ static bool trans_fsgnjx_h(DisasContext *ctx, arg_fsgnjx_h *a)
TCGv_i64 rs1, rs2;
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
- rs1 = tcg_temp_new_i64();
- gen_check_nanbox_s(rs1, cpu_fpr[a->rs1]);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ if (!ctx->ext_zfinx) {
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ rs1 = tcg_temp_new_i64();
+ gen_check_nanbox_h(rs1, src1);
+ } else {
+ rs1 = get_fpr_hs(ctx, a->rs1);
+ }
if (a->rs1 == a->rs2) { /* FABS */
- tcg_gen_andi_i64(cpu_fpr[a->rd], rs1, ~MAKE_64BIT_MASK(15, 1));
+ tcg_gen_andi_i64(dest, rs1, ~MAKE_64BIT_MASK(15, 1));
} else {
- rs2 = tcg_temp_new_i64();
- gen_check_nanbox_s(rs2, cpu_fpr[a->rs2]);
+ if (!ctx->ext_zfinx) {
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
+ rs2 = tcg_temp_new_i64();
+ gen_check_nanbox_h(rs2, src2);
+ } else {
+ rs2 = get_fpr_hs(ctx, a->rs2);
+ }
/*
* Xor bit 15 in rs1 with that in rs2.
* This formulation retains the nanboxing of rs1.
*/
- tcg_gen_andi_i64(rs2, rs2, MAKE_64BIT_MASK(15, 1));
- tcg_gen_xor_i64(cpu_fpr[a->rd], rs1, rs2);
+ tcg_gen_andi_i64(dest, rs2, MAKE_64BIT_MASK(15, 1));
+ tcg_gen_xor_i64(dest, rs1, dest);
- tcg_temp_free_i64(rs2);
+ if (!ctx->ext_zfinx) {
+ tcg_temp_free_i64(rs2);
+ }
+ }
+ if (ctx->ext_zfinx) {
+ gen_nanbox_h(dest, dest);
+ } else {
+ tcg_temp_free_i64(rs1);
}
-
mark_fs_dirty(ctx);
return true;
}
@@ -267,10 +371,14 @@ static bool trans_fsgnjx_h(DisasContext *ctx, arg_fsgnjx_h *a)
static bool trans_fmin_h(DisasContext *ctx, arg_fmin_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
- gen_helper_fmin_h(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1],
- cpu_fpr[a->rs2]);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
+
+ gen_helper_fmin_h(dest, cpu_env, src1, src2);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -278,10 +386,14 @@ static bool trans_fmin_h(DisasContext *ctx, arg_fmin_h *a)
static bool trans_fmax_h(DisasContext *ctx, arg_fmax_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
- gen_helper_fmax_h(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1],
- cpu_fpr[a->rs2]);
+ gen_helper_fmax_h(dest, cpu_env, src1, src2);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -289,10 +401,14 @@ static bool trans_fmax_h(DisasContext *ctx, arg_fmax_h *a)
static bool trans_fcvt_s_h(DisasContext *ctx, arg_fcvt_s_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH_OR_ZFHMIN(ctx);
+ REQUIRE_ZFH_OR_ZFHMIN_OR_ZHINX_OR_ZHINXMIN(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_s_h(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1]);
+ gen_helper_fcvt_s_h(dest, cpu_env, src1);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
@@ -302,26 +418,32 @@ static bool trans_fcvt_s_h(DisasContext *ctx, arg_fcvt_s_h *a)
static bool trans_fcvt_d_h(DisasContext *ctx, arg_fcvt_d_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH_OR_ZFHMIN(ctx);
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZFH_OR_ZFHMIN_OR_ZHINX_OR_ZHINXMIN(ctx);
+ REQUIRE_ZDINX_OR_D(ctx);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_d_h(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1]);
+ gen_helper_fcvt_d_h(dest, cpu_env, src1);
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
-
return true;
}
static bool trans_fcvt_h_s(DisasContext *ctx, arg_fcvt_h_s *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH_OR_ZFHMIN(ctx);
+ REQUIRE_ZFH_OR_ZFHMIN_OR_ZHINX_OR_ZHINXMIN(ctx);
- gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_h_s(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1]);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ gen_set_rm(ctx, a->rm);
+ gen_helper_fcvt_h_s(dest, cpu_env, src1);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
@@ -330,12 +452,15 @@ static bool trans_fcvt_h_s(DisasContext *ctx, arg_fcvt_h_s *a)
static bool trans_fcvt_h_d(DisasContext *ctx, arg_fcvt_h_d *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH_OR_ZFHMIN(ctx);
- REQUIRE_EXT(ctx, RVD);
+ REQUIRE_ZFH_OR_ZFHMIN_OR_ZHINX_OR_ZHINXMIN(ctx);
+ REQUIRE_ZDINX_OR_D(ctx);
- gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_h_d(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1]);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ gen_set_rm(ctx, a->rm);
+ gen_helper_fcvt_h_d(dest, cpu_env, src1);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
@@ -344,11 +469,13 @@ static bool trans_fcvt_h_d(DisasContext *ctx, arg_fcvt_h_d *a)
static bool trans_feq_h(DisasContext *ctx, arg_feq_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
- gen_helper_feq_h(dest, cpu_env, cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ gen_helper_feq_h(dest, cpu_env, src1, src2);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -356,11 +483,13 @@ static bool trans_feq_h(DisasContext *ctx, arg_feq_h *a)
static bool trans_flt_h(DisasContext *ctx, arg_flt_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
- gen_helper_flt_h(dest, cpu_env, cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ gen_helper_flt_h(dest, cpu_env, src1, src2);
gen_set_gpr(ctx, a->rd, dest);
return true;
@@ -369,11 +498,13 @@ static bool trans_flt_h(DisasContext *ctx, arg_flt_h *a)
static bool trans_fle_h(DisasContext *ctx, arg_fle_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_hs(ctx, a->rs2);
- gen_helper_fle_h(dest, cpu_env, cpu_fpr[a->rs1], cpu_fpr[a->rs2]);
+ gen_helper_fle_h(dest, cpu_env, src1, src2);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -381,11 +512,12 @@ static bool trans_fle_h(DisasContext *ctx, arg_fle_h *a)
static bool trans_fclass_h(DisasContext *ctx, arg_fclass_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
- gen_helper_fclass_h(dest, cpu_fpr[a->rs1]);
+ gen_helper_fclass_h(dest, cpu_env, src1);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -393,12 +525,13 @@ static bool trans_fclass_h(DisasContext *ctx, arg_fclass_h *a)
static bool trans_fcvt_w_h(DisasContext *ctx, arg_fcvt_w_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_w_h(dest, cpu_env, cpu_fpr[a->rs1]);
+ gen_helper_fcvt_w_h(dest, cpu_env, src1);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -406,12 +539,13 @@ static bool trans_fcvt_w_h(DisasContext *ctx, arg_fcvt_w_h *a)
static bool trans_fcvt_wu_h(DisasContext *ctx, arg_fcvt_wu_h *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_wu_h(dest, cpu_env, cpu_fpr[a->rs1]);
+ gen_helper_fcvt_wu_h(dest, cpu_env, src1);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -419,12 +553,14 @@ static bool trans_fcvt_wu_h(DisasContext *ctx, arg_fcvt_wu_h *a)
static bool trans_fcvt_h_w(DisasContext *ctx, arg_fcvt_h_w *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
TCGv t0 = get_gpr(ctx, a->rs1, EXT_SIGN);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_h_w(cpu_fpr[a->rd], cpu_env, t0);
+ gen_helper_fcvt_h_w(dest, cpu_env, t0);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
@@ -433,12 +569,14 @@ static bool trans_fcvt_h_w(DisasContext *ctx, arg_fcvt_h_w *a)
static bool trans_fcvt_h_wu(DisasContext *ctx, arg_fcvt_h_wu *a)
{
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
TCGv t0 = get_gpr(ctx, a->rs1, EXT_SIGN);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_h_wu(cpu_fpr[a->rd], cpu_env, t0);
+ gen_helper_fcvt_h_wu(dest, cpu_env, t0);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
@@ -482,12 +620,13 @@ static bool trans_fcvt_l_h(DisasContext *ctx, arg_fcvt_l_h *a)
{
REQUIRE_64BIT(ctx);
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_l_h(dest, cpu_env, cpu_fpr[a->rs1]);
+ gen_helper_fcvt_l_h(dest, cpu_env, src1);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -496,12 +635,13 @@ static bool trans_fcvt_lu_h(DisasContext *ctx, arg_fcvt_lu_h *a)
{
REQUIRE_64BIT(ctx);
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
TCGv dest = dest_gpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_lu_h(dest, cpu_env, cpu_fpr[a->rs1]);
+ gen_helper_fcvt_lu_h(dest, cpu_env, src1);
gen_set_gpr(ctx, a->rd, dest);
return true;
}
@@ -510,12 +650,14 @@ static bool trans_fcvt_h_l(DisasContext *ctx, arg_fcvt_h_l *a)
{
REQUIRE_64BIT(ctx);
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
TCGv t0 = get_gpr(ctx, a->rs1, EXT_SIGN);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_h_l(cpu_fpr[a->rd], cpu_env, t0);
+ gen_helper_fcvt_h_l(dest, cpu_env, t0);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
@@ -525,12 +667,14 @@ static bool trans_fcvt_h_lu(DisasContext *ctx, arg_fcvt_h_lu *a)
{
REQUIRE_64BIT(ctx);
REQUIRE_FPU;
- REQUIRE_ZFH(ctx);
+ REQUIRE_ZHINX_OR_ZFH(ctx);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
TCGv t0 = get_gpr(ctx, a->rs1, EXT_SIGN);
gen_set_rm(ctx, a->rm);
- gen_helper_fcvt_h_lu(cpu_fpr[a->rd], cpu_env, t0);
+ gen_helper_fcvt_h_lu(dest, cpu_env, t0);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
diff --git a/target/riscv/internals.h b/target/riscv/internals.h
index 9f3f3319f2..9f046e26bb 100644
--- a/target/riscv/internals.h
+++ b/target/riscv/internals.h
@@ -71,8 +71,12 @@ static inline uint64_t nanbox_h(float16 f)
return f | MAKE_64BIT_MASK(16, 48);
}
-static inline float16 check_nanbox_h(uint64_t f)
+static inline float16 check_nanbox_h(CPURISCVState *env, uint64_t f)
{
+ /* Disable nanbox check when enable zfinx */
+ if (RISCV_CPU(env_cpu(env))->cfg.ext_zfinx)
+ return (uint16_t)f;
+
uint64_t mask = MAKE_64BIT_MASK(16, 48);
if (likely((f & mask) == mask)) {
--
2.17.1
^ permalink raw reply related [flat|nested] 30+ messages in thread
* Re: [PATCH 5/6] target/riscv: add support for zhinx/zhinxmin
2021-12-24 3:49 ` liweiwei
(?)
@ 2021-12-24 22:32 ` Richard Henderson
2021-12-25 3:35 ` liweiwei
-1 siblings, 1 reply; 30+ messages in thread
From: Richard Henderson @ 2021-12-24 22:32 UTC (permalink / raw)
To: liweiwei, palmer, alistair.francis, bin.meng, qemu-riscv, qemu-devel
Cc: wangjunqiang, lazyparser, ardxwe
On 12/23/21 7:49 PM, liweiwei wrote:
> +static inline float16 check_nanbox_h(CPURISCVState *env, uint64_t f)
> {
> + /* Disable nanbox check when enable zfinx */
> + if (RISCV_CPU(env_cpu(env))->cfg.ext_zfinx)
> + return (uint16_t)f;
Braces.
r~
^ permalink raw reply [flat|nested] 30+ messages in thread
* Re: [PATCH 5/6] target/riscv: add support for zhinx/zhinxmin
2021-12-24 22:32 ` Richard Henderson
@ 2021-12-25 3:35 ` liweiwei
0 siblings, 0 replies; 30+ messages in thread
From: liweiwei @ 2021-12-25 3:35 UTC (permalink / raw)
To: Richard Henderson, palmer, alistair.francis, bin.meng,
qemu-riscv, qemu-devel
Cc: wangjunqiang, lazyparser, ardxwe
在 2021/12/25 上午6:32, Richard Henderson 写道:
> On 12/23/21 7:49 PM, liweiwei wrote:
>> +static inline float16 check_nanbox_h(CPURISCVState *env, uint64_t f)
>> {
>> + /* Disable nanbox check when enable zfinx */
>> + if (RISCV_CPU(env_cpu(env))->cfg.ext_zfinx)
>> + return (uint16_t)f;
>
> Braces.
>
OK. I'll add this later.
> r~
^ permalink raw reply [flat|nested] 30+ messages in thread
* [PATCH 6/6] target/riscv: expose zfinx, zdinx, zhinx{min} properties
2021-12-24 3:49 ` liweiwei
@ 2021-12-24 3:49 ` liweiwei
-1 siblings, 0 replies; 30+ messages in thread
From: liweiwei @ 2021-12-24 3:49 UTC (permalink / raw)
To: palmer, alistair.francis, bin.meng, qemu-riscv, qemu-devel
Cc: wangjunqiang, liweiwei, lazyparser, ardxwe
Co-authored-by: ardxwe <ardxwe@gmail.com>
Signed-off-by: liweiwei <liweiwei@iscas.ac.cn>
Signed-off-by: wangjunqiang <wangjunqiang@iscas.ac.cn>
---
target/riscv/cpu.c | 4 ++++
1 file changed, 4 insertions(+)
diff --git a/target/riscv/cpu.c b/target/riscv/cpu.c
index a5fa14f2ac..dbd15693be 100644
--- a/target/riscv/cpu.c
+++ b/target/riscv/cpu.c
@@ -657,6 +657,10 @@ static Property riscv_cpu_properties[] = {
DEFINE_PROP_BOOL("zbb", RISCVCPU, cfg.ext_zbb, true),
DEFINE_PROP_BOOL("zbc", RISCVCPU, cfg.ext_zbc, true),
DEFINE_PROP_BOOL("zbs", RISCVCPU, cfg.ext_zbs, true),
+ DEFINE_PROP_BOOL("Zdinx", RISCVCPU, cfg.ext_zdinx, false),
+ DEFINE_PROP_BOOL("Zfinx", RISCVCPU, cfg.ext_zfinx, false),
+ DEFINE_PROP_BOOL("Zhinx", RISCVCPU, cfg.ext_zhinx, false),
+ DEFINE_PROP_BOOL("Zhinxmin", RISCVCPU, cfg.ext_zhinxmin, false),
DEFINE_PROP_BOOL("x-h", RISCVCPU, cfg.ext_h, false),
DEFINE_PROP_BOOL("x-j", RISCVCPU, cfg.ext_j, false),
/* ePMP 0.9.3 */
--
2.17.1
^ permalink raw reply related [flat|nested] 30+ messages in thread
* [PATCH 6/6] target/riscv: expose zfinx, zdinx, zhinx{min} properties
@ 2021-12-24 3:49 ` liweiwei
0 siblings, 0 replies; 30+ messages in thread
From: liweiwei @ 2021-12-24 3:49 UTC (permalink / raw)
To: palmer, alistair.francis, bin.meng, qemu-riscv, qemu-devel
Cc: wangjunqiang, lazyparser, ardxwe, liweiwei
Co-authored-by: ardxwe <ardxwe@gmail.com>
Signed-off-by: liweiwei <liweiwei@iscas.ac.cn>
Signed-off-by: wangjunqiang <wangjunqiang@iscas.ac.cn>
---
target/riscv/cpu.c | 4 ++++
1 file changed, 4 insertions(+)
diff --git a/target/riscv/cpu.c b/target/riscv/cpu.c
index a5fa14f2ac..dbd15693be 100644
--- a/target/riscv/cpu.c
+++ b/target/riscv/cpu.c
@@ -657,6 +657,10 @@ static Property riscv_cpu_properties[] = {
DEFINE_PROP_BOOL("zbb", RISCVCPU, cfg.ext_zbb, true),
DEFINE_PROP_BOOL("zbc", RISCVCPU, cfg.ext_zbc, true),
DEFINE_PROP_BOOL("zbs", RISCVCPU, cfg.ext_zbs, true),
+ DEFINE_PROP_BOOL("Zdinx", RISCVCPU, cfg.ext_zdinx, false),
+ DEFINE_PROP_BOOL("Zfinx", RISCVCPU, cfg.ext_zfinx, false),
+ DEFINE_PROP_BOOL("Zhinx", RISCVCPU, cfg.ext_zhinx, false),
+ DEFINE_PROP_BOOL("Zhinxmin", RISCVCPU, cfg.ext_zhinxmin, false),
DEFINE_PROP_BOOL("x-h", RISCVCPU, cfg.ext_h, false),
DEFINE_PROP_BOOL("x-j", RISCVCPU, cfg.ext_j, false),
/* ePMP 0.9.3 */
--
2.17.1
^ permalink raw reply related [flat|nested] 30+ messages in thread
* Re: [PATCH 6/6] target/riscv: expose zfinx, zdinx, zhinx{min} properties
2021-12-24 3:49 ` liweiwei
(?)
@ 2021-12-24 22:32 ` Richard Henderson
-1 siblings, 0 replies; 30+ messages in thread
From: Richard Henderson @ 2021-12-24 22:32 UTC (permalink / raw)
To: liweiwei, palmer, alistair.francis, bin.meng, qemu-riscv, qemu-devel
Cc: wangjunqiang, lazyparser, ardxwe
On 12/23/21 7:49 PM, liweiwei wrote:
> Co-authored-by: ardxwe <ardxwe@gmail.com>
> Signed-off-by: liweiwei <liweiwei@iscas.ac.cn>
> Signed-off-by: wangjunqiang <wangjunqiang@iscas.ac.cn>
> ---
> target/riscv/cpu.c | 4 ++++
> 1 file changed, 4 insertions(+)
>
> diff --git a/target/riscv/cpu.c b/target/riscv/cpu.c
> index a5fa14f2ac..dbd15693be 100644
> --- a/target/riscv/cpu.c
> +++ b/target/riscv/cpu.c
> @@ -657,6 +657,10 @@ static Property riscv_cpu_properties[] = {
> DEFINE_PROP_BOOL("zbb", RISCVCPU, cfg.ext_zbb, true),
> DEFINE_PROP_BOOL("zbc", RISCVCPU, cfg.ext_zbc, true),
> DEFINE_PROP_BOOL("zbs", RISCVCPU, cfg.ext_zbs, true),
> + DEFINE_PROP_BOOL("Zdinx", RISCVCPU, cfg.ext_zdinx, false),
> + DEFINE_PROP_BOOL("Zfinx", RISCVCPU, cfg.ext_zfinx, false),
> + DEFINE_PROP_BOOL("Zhinx", RISCVCPU, cfg.ext_zhinx, false),
> + DEFINE_PROP_BOOL("Zhinxmin", RISCVCPU, cfg.ext_zhinxmin, false),
> DEFINE_PROP_BOOL("x-h", RISCVCPU, cfg.ext_h, false),
> DEFINE_PROP_BOOL("x-j", RISCVCPU, cfg.ext_j, false),
> /* ePMP 0.9.3 */
>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
r~
^ permalink raw reply [flat|nested] 30+ messages in thread