* [PATCH v5 0/6] support subsets of Float-Point in Integer Registers extensions
@ 2022-01-28 13:16 ` Weiwei Li
0 siblings, 0 replies; 14+ messages in thread
From: Weiwei Li @ 2022-01-28 13:16 UTC (permalink / raw)
To: richard.henderson, palmer, alistair.francis, bin.meng,
qemu-riscv, qemu-devel
Cc: wangjunqiang, Weiwei Li, lazyparser, ardxwe
This patchset implements RISC-V Float-Point in Integer Registers extensions(Version 1.0), which includes Zfinx, Zdinx, Zhinx and Zhinxmin extension.
Specification:
https://github.com/riscv/riscv-zfinx/blob/main/zfinx-1.0.0.pdf
The port is available here:
https://github.com/plctlab/plct-qemu/tree/plct-zfinx-upstream-v5
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/.
v5:
* put definition of ftemp and nftemp together, add comments for them
* sperate the declare of variable i from loop
v4:
* combine register pair check for rv32 zdinx
* clear mstatus.FS when RVF is disabled by write_misa
v3:
* delete unused reset for mstatus.FS
* use positive test for RVF instead of negative test for ZFINX
* replace get_ol with get_xl
* use tcg_gen_concat_tl_i64 to unify tcg_gen_concat_i32_i64 and tcg_gen_deposit_i64
v2:
* hardwire mstatus.FS to zero when enable zfinx
* do register-pair check at the begin of translation
* optimize partial implemention as suggested
Weiwei Li (6):
target/riscv: add cfg properties for zfinx, zdinx and zhinx{min}
target/riscv: hardwire mstatus.FS to zero when enable 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
target/riscv/cpu.c | 17 ++
target/riscv/cpu.h | 4 +
target/riscv/cpu_helper.c | 6 +-
target/riscv/csr.c | 25 +-
target/riscv/fpu_helper.c | 178 ++++++------
target/riscv/helper.h | 4 +-
target/riscv/insn_trans/trans_rvd.c.inc | 285 ++++++++++++++-----
target/riscv/insn_trans/trans_rvf.c.inc | 314 +++++++++++++-------
target/riscv/insn_trans/trans_rvzfh.c.inc | 332 +++++++++++++++-------
target/riscv/internals.h | 32 ++-
target/riscv/translate.c | 157 +++++++++-
11 files changed, 982 insertions(+), 372 deletions(-)
--
2.17.1
^ permalink raw reply [flat|nested] 14+ messages in thread
* [PATCH v5 0/6] support subsets of Float-Point in Integer Registers extensions
@ 2022-01-28 13:16 ` Weiwei Li
0 siblings, 0 replies; 14+ messages in thread
From: Weiwei Li @ 2022-01-28 13:16 UTC (permalink / raw)
To: richard.henderson, palmer, alistair.francis, bin.meng,
qemu-riscv, qemu-devel
Cc: wangjunqiang, lazyparser, ardxwe, Weiwei Li
This patchset implements RISC-V Float-Point in Integer Registers extensions(Version 1.0), which includes Zfinx, Zdinx, Zhinx and Zhinxmin extension.
Specification:
https://github.com/riscv/riscv-zfinx/blob/main/zfinx-1.0.0.pdf
The port is available here:
https://github.com/plctlab/plct-qemu/tree/plct-zfinx-upstream-v5
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/.
v5:
* put definition of ftemp and nftemp together, add comments for them
* sperate the declare of variable i from loop
v4:
* combine register pair check for rv32 zdinx
* clear mstatus.FS when RVF is disabled by write_misa
v3:
* delete unused reset for mstatus.FS
* use positive test for RVF instead of negative test for ZFINX
* replace get_ol with get_xl
* use tcg_gen_concat_tl_i64 to unify tcg_gen_concat_i32_i64 and tcg_gen_deposit_i64
v2:
* hardwire mstatus.FS to zero when enable zfinx
* do register-pair check at the begin of translation
* optimize partial implemention as suggested
Weiwei Li (6):
target/riscv: add cfg properties for zfinx, zdinx and zhinx{min}
target/riscv: hardwire mstatus.FS to zero when enable 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
target/riscv/cpu.c | 17 ++
target/riscv/cpu.h | 4 +
target/riscv/cpu_helper.c | 6 +-
target/riscv/csr.c | 25 +-
target/riscv/fpu_helper.c | 178 ++++++------
target/riscv/helper.h | 4 +-
target/riscv/insn_trans/trans_rvd.c.inc | 285 ++++++++++++++-----
target/riscv/insn_trans/trans_rvf.c.inc | 314 +++++++++++++-------
target/riscv/insn_trans/trans_rvzfh.c.inc | 332 +++++++++++++++-------
target/riscv/internals.h | 32 ++-
target/riscv/translate.c | 157 +++++++++-
11 files changed, 982 insertions(+), 372 deletions(-)
--
2.17.1
^ permalink raw reply [flat|nested] 14+ messages in thread
* [PATCH v5 1/6] target/riscv: add cfg properties for zfinx, zdinx and zhinx{min}
2022-01-28 13:16 ` Weiwei Li
@ 2022-01-28 13:16 ` Weiwei Li
-1 siblings, 0 replies; 14+ messages in thread
From: Weiwei Li @ 2022-01-28 13:16 UTC (permalink / raw)
To: richard.henderson, palmer, alistair.francis, bin.meng,
qemu-riscv, qemu-devel
Cc: wangjunqiang, Weiwei Li, lazyparser, ardxwe
Co-authored-by: ardxwe <ardxwe@gmail.com>
Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
---
target/riscv/cpu.c | 12 ++++++++++++
target/riscv/cpu.h | 4 ++++
target/riscv/translate.c | 8 ++++++++
3 files changed, 24 insertions(+)
diff --git a/target/riscv/cpu.c b/target/riscv/cpu.c
index 1cb0436187..7006e6647b 100644
--- a/target/riscv/cpu.c
+++ b/target/riscv/cpu.c
@@ -569,6 +569,11 @@ 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;
@@ -647,6 +652,13 @@ 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 55635d68d5..be46840541 100644
--- a/target/riscv/cpu.h
+++ b/target/riscv/cpu.h
@@ -341,8 +341,12 @@ 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;
bool ext_zve32f;
bool ext_zve64f;
diff --git a/target/riscv/translate.c b/target/riscv/translate.c
index f0bbe80875..1484586fe7 100644
--- a/target/riscv/translate.c
+++ b/target/riscv/translate.c
@@ -77,8 +77,12 @@ 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 ext_zve32f;
bool ext_zve64f;
bool hlsx;
@@ -909,8 +913,12 @@ 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->ext_zve32f = cpu->cfg.ext_zve32f;
ctx->ext_zve64f = cpu->cfg.ext_zve64f;
ctx->vlen = cpu->cfg.vlen;
--
2.17.1
^ permalink raw reply related [flat|nested] 14+ messages in thread
* [PATCH v5 1/6] target/riscv: add cfg properties for zfinx, zdinx and zhinx{min}
@ 2022-01-28 13:16 ` Weiwei Li
0 siblings, 0 replies; 14+ messages in thread
From: Weiwei Li @ 2022-01-28 13:16 UTC (permalink / raw)
To: richard.henderson, palmer, alistair.francis, bin.meng,
qemu-riscv, qemu-devel
Cc: wangjunqiang, lazyparser, ardxwe, Weiwei Li
Co-authored-by: ardxwe <ardxwe@gmail.com>
Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
---
target/riscv/cpu.c | 12 ++++++++++++
target/riscv/cpu.h | 4 ++++
target/riscv/translate.c | 8 ++++++++
3 files changed, 24 insertions(+)
diff --git a/target/riscv/cpu.c b/target/riscv/cpu.c
index 1cb0436187..7006e6647b 100644
--- a/target/riscv/cpu.c
+++ b/target/riscv/cpu.c
@@ -569,6 +569,11 @@ 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;
@@ -647,6 +652,13 @@ 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 55635d68d5..be46840541 100644
--- a/target/riscv/cpu.h
+++ b/target/riscv/cpu.h
@@ -341,8 +341,12 @@ 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;
bool ext_zve32f;
bool ext_zve64f;
diff --git a/target/riscv/translate.c b/target/riscv/translate.c
index f0bbe80875..1484586fe7 100644
--- a/target/riscv/translate.c
+++ b/target/riscv/translate.c
@@ -77,8 +77,12 @@ 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 ext_zve32f;
bool ext_zve64f;
bool hlsx;
@@ -909,8 +913,12 @@ 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->ext_zve32f = cpu->cfg.ext_zve32f;
ctx->ext_zve64f = cpu->cfg.ext_zve64f;
ctx->vlen = cpu->cfg.vlen;
--
2.17.1
^ permalink raw reply related [flat|nested] 14+ messages in thread
* [PATCH v5 2/6] target/riscv: hardwire mstatus.FS to zero when enable zfinx
2022-01-28 13:16 ` Weiwei Li
@ 2022-01-28 13:16 ` Weiwei Li
-1 siblings, 0 replies; 14+ messages in thread
From: Weiwei Li @ 2022-01-28 13:16 UTC (permalink / raw)
To: richard.henderson, palmer, alistair.francis, bin.meng,
qemu-riscv, qemu-devel
Cc: wangjunqiang, Weiwei Li, lazyparser, ardxwe
Co-authored-by: ardxwe <ardxwe@gmail.com>
Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
---
target/riscv/cpu_helper.c | 6 +++++-
target/riscv/csr.c | 25 ++++++++++++++++++++-----
target/riscv/translate.c | 4 ++++
3 files changed, 29 insertions(+), 6 deletions(-)
diff --git a/target/riscv/cpu_helper.c b/target/riscv/cpu_helper.c
index 327a2c4f1d..023e011923 100644
--- a/target/riscv/cpu_helper.c
+++ b/target/riscv/cpu_helper.c
@@ -223,9 +223,13 @@ bool riscv_cpu_vector_enabled(CPURISCVState *env)
void riscv_cpu_swap_hypervisor_regs(CPURISCVState *env)
{
- uint64_t mstatus_mask = MSTATUS_MXR | MSTATUS_SUM | MSTATUS_FS |
+ uint64_t mstatus_mask = MSTATUS_MXR | MSTATUS_SUM |
MSTATUS_SPP | MSTATUS_SPIE | MSTATUS_SIE |
MSTATUS64_UXL | MSTATUS_VS;
+
+ if (riscv_has_ext(env, RVF)) {
+ mstatus_mask |= MSTATUS_FS;
+ }
bool current_virt = riscv_cpu_virt_enabled(env);
g_assert(riscv_has_ext(env, RVH));
diff --git a/target/riscv/csr.c b/target/riscv/csr.c
index e5f9d4ef93..238e1d1858 100644
--- a/target/riscv/csr.c
+++ b/target/riscv/csr.c
@@ -38,7 +38,8 @@ void riscv_set_csr_ops(int csrno, riscv_csr_operations *ops)
static RISCVException fs(CPURISCVState *env, int csrno)
{
#if !defined(CONFIG_USER_ONLY)
- if (!env->debugger && !riscv_cpu_fp_enabled(env)) {
+ if (!env->debugger && !riscv_cpu_fp_enabled(env) &&
+ !RISCV_CPU(env_cpu(env))->cfg.ext_zfinx) {
return RISCV_EXCP_ILLEGAL_INST;
}
#endif
@@ -238,7 +239,9 @@ static RISCVException write_fflags(CPURISCVState *env, int csrno,
target_ulong val)
{
#if !defined(CONFIG_USER_ONLY)
- env->mstatus |= MSTATUS_FS;
+ if (riscv_has_ext(env, RVF)) {
+ env->mstatus |= MSTATUS_FS;
+ }
#endif
riscv_cpu_set_fflags(env, val & (FSR_AEXC >> FSR_AEXC_SHIFT));
return RISCV_EXCP_NONE;
@@ -255,7 +258,9 @@ static RISCVException write_frm(CPURISCVState *env, int csrno,
target_ulong val)
{
#if !defined(CONFIG_USER_ONLY)
- env->mstatus |= MSTATUS_FS;
+ if (riscv_has_ext(env, RVF)) {
+ env->mstatus |= MSTATUS_FS;
+ }
#endif
env->frm = val & (FSR_RD >> FSR_RD_SHIFT);
return RISCV_EXCP_NONE;
@@ -273,7 +278,9 @@ static RISCVException write_fcsr(CPURISCVState *env, int csrno,
target_ulong val)
{
#if !defined(CONFIG_USER_ONLY)
- env->mstatus |= MSTATUS_FS;
+ if (riscv_has_ext(env, RVF)) {
+ env->mstatus |= MSTATUS_FS;
+ }
#endif
env->frm = (val & FSR_RD) >> FSR_RD_SHIFT;
riscv_cpu_set_fflags(env, (val & FSR_AEXC) >> FSR_AEXC_SHIFT);
@@ -580,10 +587,14 @@ static RISCVException write_mstatus(CPURISCVState *env, int csrno,
tlb_flush(env_cpu(env));
}
mask = MSTATUS_SIE | MSTATUS_SPIE | MSTATUS_MIE | MSTATUS_MPIE |
- MSTATUS_SPP | MSTATUS_FS | MSTATUS_MPRV | MSTATUS_SUM |
+ MSTATUS_SPP | MSTATUS_MPRV | MSTATUS_SUM |
MSTATUS_MPP | MSTATUS_MXR | MSTATUS_TVM | MSTATUS_TSR |
MSTATUS_TW | MSTATUS_VS;
+ if (riscv_has_ext(env, RVF)) {
+ mask |= MSTATUS_FS;
+ }
+
if (xl != MXL_RV32 || env->debugger) {
/*
* RV32: MPV and GVA are not in mstatus. The current plan is to
@@ -715,6 +726,10 @@ static RISCVException write_misa(CPURISCVState *env, int csrno,
return RISCV_EXCP_NONE;
}
+ if (!(val & RVF)) {
+ env->mstatus &= ~MSTATUS_FS;
+ }
+
/* flush translation cache */
tb_flush(env_cpu(env));
env->misa_ext = val;
diff --git a/target/riscv/translate.c b/target/riscv/translate.c
index 1484586fe7..c9baf2a87a 100644
--- a/target/riscv/translate.c
+++ b/target/riscv/translate.c
@@ -423,6 +423,10 @@ static void mark_fs_dirty(DisasContext *ctx)
{
TCGv tmp;
+ if (!has_ext(ctx, RVF)) {
+ return;
+ }
+
if (ctx->mstatus_fs != MSTATUS_FS) {
/* Remember the state change for the rest of the TB. */
ctx->mstatus_fs = MSTATUS_FS;
--
2.17.1
^ permalink raw reply related [flat|nested] 14+ messages in thread
* [PATCH v5 2/6] target/riscv: hardwire mstatus.FS to zero when enable zfinx
@ 2022-01-28 13:16 ` Weiwei Li
0 siblings, 0 replies; 14+ messages in thread
From: Weiwei Li @ 2022-01-28 13:16 UTC (permalink / raw)
To: richard.henderson, palmer, alistair.francis, bin.meng,
qemu-riscv, qemu-devel
Cc: wangjunqiang, lazyparser, ardxwe, Weiwei Li
Co-authored-by: ardxwe <ardxwe@gmail.com>
Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
---
target/riscv/cpu_helper.c | 6 +++++-
target/riscv/csr.c | 25 ++++++++++++++++++++-----
target/riscv/translate.c | 4 ++++
3 files changed, 29 insertions(+), 6 deletions(-)
diff --git a/target/riscv/cpu_helper.c b/target/riscv/cpu_helper.c
index 327a2c4f1d..023e011923 100644
--- a/target/riscv/cpu_helper.c
+++ b/target/riscv/cpu_helper.c
@@ -223,9 +223,13 @@ bool riscv_cpu_vector_enabled(CPURISCVState *env)
void riscv_cpu_swap_hypervisor_regs(CPURISCVState *env)
{
- uint64_t mstatus_mask = MSTATUS_MXR | MSTATUS_SUM | MSTATUS_FS |
+ uint64_t mstatus_mask = MSTATUS_MXR | MSTATUS_SUM |
MSTATUS_SPP | MSTATUS_SPIE | MSTATUS_SIE |
MSTATUS64_UXL | MSTATUS_VS;
+
+ if (riscv_has_ext(env, RVF)) {
+ mstatus_mask |= MSTATUS_FS;
+ }
bool current_virt = riscv_cpu_virt_enabled(env);
g_assert(riscv_has_ext(env, RVH));
diff --git a/target/riscv/csr.c b/target/riscv/csr.c
index e5f9d4ef93..238e1d1858 100644
--- a/target/riscv/csr.c
+++ b/target/riscv/csr.c
@@ -38,7 +38,8 @@ void riscv_set_csr_ops(int csrno, riscv_csr_operations *ops)
static RISCVException fs(CPURISCVState *env, int csrno)
{
#if !defined(CONFIG_USER_ONLY)
- if (!env->debugger && !riscv_cpu_fp_enabled(env)) {
+ if (!env->debugger && !riscv_cpu_fp_enabled(env) &&
+ !RISCV_CPU(env_cpu(env))->cfg.ext_zfinx) {
return RISCV_EXCP_ILLEGAL_INST;
}
#endif
@@ -238,7 +239,9 @@ static RISCVException write_fflags(CPURISCVState *env, int csrno,
target_ulong val)
{
#if !defined(CONFIG_USER_ONLY)
- env->mstatus |= MSTATUS_FS;
+ if (riscv_has_ext(env, RVF)) {
+ env->mstatus |= MSTATUS_FS;
+ }
#endif
riscv_cpu_set_fflags(env, val & (FSR_AEXC >> FSR_AEXC_SHIFT));
return RISCV_EXCP_NONE;
@@ -255,7 +258,9 @@ static RISCVException write_frm(CPURISCVState *env, int csrno,
target_ulong val)
{
#if !defined(CONFIG_USER_ONLY)
- env->mstatus |= MSTATUS_FS;
+ if (riscv_has_ext(env, RVF)) {
+ env->mstatus |= MSTATUS_FS;
+ }
#endif
env->frm = val & (FSR_RD >> FSR_RD_SHIFT);
return RISCV_EXCP_NONE;
@@ -273,7 +278,9 @@ static RISCVException write_fcsr(CPURISCVState *env, int csrno,
target_ulong val)
{
#if !defined(CONFIG_USER_ONLY)
- env->mstatus |= MSTATUS_FS;
+ if (riscv_has_ext(env, RVF)) {
+ env->mstatus |= MSTATUS_FS;
+ }
#endif
env->frm = (val & FSR_RD) >> FSR_RD_SHIFT;
riscv_cpu_set_fflags(env, (val & FSR_AEXC) >> FSR_AEXC_SHIFT);
@@ -580,10 +587,14 @@ static RISCVException write_mstatus(CPURISCVState *env, int csrno,
tlb_flush(env_cpu(env));
}
mask = MSTATUS_SIE | MSTATUS_SPIE | MSTATUS_MIE | MSTATUS_MPIE |
- MSTATUS_SPP | MSTATUS_FS | MSTATUS_MPRV | MSTATUS_SUM |
+ MSTATUS_SPP | MSTATUS_MPRV | MSTATUS_SUM |
MSTATUS_MPP | MSTATUS_MXR | MSTATUS_TVM | MSTATUS_TSR |
MSTATUS_TW | MSTATUS_VS;
+ if (riscv_has_ext(env, RVF)) {
+ mask |= MSTATUS_FS;
+ }
+
if (xl != MXL_RV32 || env->debugger) {
/*
* RV32: MPV and GVA are not in mstatus. The current plan is to
@@ -715,6 +726,10 @@ static RISCVException write_misa(CPURISCVState *env, int csrno,
return RISCV_EXCP_NONE;
}
+ if (!(val & RVF)) {
+ env->mstatus &= ~MSTATUS_FS;
+ }
+
/* flush translation cache */
tb_flush(env_cpu(env));
env->misa_ext = val;
diff --git a/target/riscv/translate.c b/target/riscv/translate.c
index 1484586fe7..c9baf2a87a 100644
--- a/target/riscv/translate.c
+++ b/target/riscv/translate.c
@@ -423,6 +423,10 @@ static void mark_fs_dirty(DisasContext *ctx)
{
TCGv tmp;
+ if (!has_ext(ctx, RVF)) {
+ return;
+ }
+
if (ctx->mstatus_fs != MSTATUS_FS) {
/* Remember the state change for the rest of the TB. */
ctx->mstatus_fs = MSTATUS_FS;
--
2.17.1
^ permalink raw reply related [flat|nested] 14+ messages in thread
* [PATCH v5 3/6] target/riscv: add support for zfinx
2022-01-28 13:16 ` Weiwei Li
@ 2022-01-28 13:16 ` Weiwei Li
-1 siblings, 0 replies; 14+ messages in thread
From: Weiwei Li @ 2022-01-28 13:16 UTC (permalink / raw)
To: richard.henderson, palmer, alistair.francis, bin.meng,
qemu-riscv, qemu-devel
Cc: wangjunqiang, Weiwei Li, 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: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
---
target/riscv/fpu_helper.c | 89 +++----
target/riscv/helper.h | 2 +-
target/riscv/insn_trans/trans_rvf.c.inc | 314 ++++++++++++++++--------
target/riscv/internals.h | 16 +-
target/riscv/translate.c | 93 ++++++-
5 files changed, 369 insertions(+), 145 deletions(-)
diff --git a/target/riscv/fpu_helper.c b/target/riscv/fpu_helper.c
index 4a5982d594..63ca703459 100644
--- a/target/riscv/fpu_helper.c
+++ b/target/riscv/fpu_helper.c
@@ -98,10 +98,11 @@ 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);
- return nanbox_s(float32_muladd(frs1, frs2, frs3, flags, &env->fp_status));
+ float32 frs1 = check_nanbox_s(env, rs1);
+ float32 frs2 = check_nanbox_s(env, rs2);
+ float32 frs3 = check_nanbox_s(env, rs3);
+ return nanbox_s(env, float32_muladd(frs1, frs2, frs3, flags,
+ &env->fp_status));
}
uint64_t helper_fmadd_s(CPURISCVState *env, uint64_t frs1, uint64_t frs2,
@@ -183,124 +184,124 @@ 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);
- return nanbox_s(float32_add(frs1, frs2, &env->fp_status));
+ float32 frs1 = check_nanbox_s(env, rs1);
+ float32 frs2 = check_nanbox_s(env, rs2);
+ return nanbox_s(env, 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);
- return nanbox_s(float32_sub(frs1, frs2, &env->fp_status));
+ float32 frs1 = check_nanbox_s(env, rs1);
+ float32 frs2 = check_nanbox_s(env, rs2);
+ return nanbox_s(env, 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);
- return nanbox_s(float32_mul(frs1, frs2, &env->fp_status));
+ float32 frs1 = check_nanbox_s(env, rs1);
+ float32 frs2 = check_nanbox_s(env, rs2);
+ return nanbox_s(env, 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);
- return nanbox_s(float32_div(frs1, frs2, &env->fp_status));
+ float32 frs1 = check_nanbox_s(env, rs1);
+ float32 frs2 = check_nanbox_s(env, rs2);
+ return nanbox_s(env, 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);
- return nanbox_s(env->priv_ver < PRIV_VERSION_1_11_0 ?
+ float32 frs1 = check_nanbox_s(env, rs1);
+ float32 frs2 = check_nanbox_s(env, rs2);
+ return nanbox_s(env, env->priv_ver < PRIV_VERSION_1_11_0 ?
float32_minnum(frs1, frs2, &env->fp_status) :
float32_minimum_number(frs1, frs2, &env->fp_status));
}
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);
- return nanbox_s(env->priv_ver < PRIV_VERSION_1_11_0 ?
+ float32 frs1 = check_nanbox_s(env, rs1);
+ float32 frs2 = check_nanbox_s(env, rs2);
+ return nanbox_s(env, env->priv_ver < PRIV_VERSION_1_11_0 ?
float32_maxnum(frs1, frs2, &env->fp_status) :
float32_maximum_number(frs1, frs2, &env->fp_status));
}
uint64_t helper_fsqrt_s(CPURISCVState *env, uint64_t rs1)
{
- float32 frs1 = check_nanbox_s(rs1);
- return nanbox_s(float32_sqrt(frs1, &env->fp_status));
+ float32 frs1 = check_nanbox_s(env, rs1);
+ return nanbox_s(env, 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);
}
uint64_t helper_fcvt_s_w(CPURISCVState *env, target_ulong rs1)
{
- return nanbox_s(int32_to_float32((int32_t)rs1, &env->fp_status));
+ return nanbox_s(env, int32_to_float32((int32_t)rs1, &env->fp_status));
}
uint64_t helper_fcvt_s_wu(CPURISCVState *env, target_ulong rs1)
{
- return nanbox_s(uint32_to_float32((uint32_t)rs1, &env->fp_status));
+ return nanbox_s(env, uint32_to_float32((uint32_t)rs1, &env->fp_status));
}
uint64_t helper_fcvt_s_l(CPURISCVState *env, target_ulong rs1)
{
- return nanbox_s(int64_to_float32(rs1, &env->fp_status));
+ return nanbox_s(env, int64_to_float32(rs1, &env->fp_status));
}
uint64_t helper_fcvt_s_lu(CPURISCVState *env, target_ulong rs1)
{
- return nanbox_s(uint64_to_float32(rs1, &env->fp_status));
+ return nanbox_s(env, 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);
}
@@ -340,12 +341,12 @@ uint64_t helper_fmax_d(CPURISCVState *env, uint64_t frs1, uint64_t frs2)
uint64_t helper_fcvt_s_d(CPURISCVState *env, uint64_t rs1)
{
- return nanbox_s(float64_to_float32(rs1, &env->fp_status));
+ return nanbox_s(env, float64_to_float32(rs1, &env->fp_status));
}
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);
}
@@ -539,14 +540,14 @@ 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));
}
uint64_t helper_fcvt_s_h(CPURISCVState *env, uint64_t rs1)
{
float16 frs1 = check_nanbox_h(rs1);
- return nanbox_s(float16_to_float32(frs1, true, &env->fp_status));
+ return nanbox_s(env, float16_to_float32(frs1, true, &env->fp_status));
}
uint64_t helper_fcvt_h_d(CPURISCVState *env, uint64_t rs1)
diff --git a/target/riscv/helper.h b/target/riscv/helper.h
index 72cc2582f4..89195aad9d 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 0aac87f7db..68a305edd6 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)
@@ -55,10 +62,16 @@ 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;
}
@@ -66,10 +79,16 @@ 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;
}
@@ -77,10 +96,16 @@ 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;
}
@@ -88,10 +113,16 @@ 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;
}
@@ -99,11 +130,15 @@ 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;
}
@@ -111,11 +146,15 @@ 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;
}
@@ -123,11 +162,15 @@ 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;
}
@@ -135,11 +178,15 @@ 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;
}
@@ -147,10 +194,14 @@ 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;
}
@@ -158,22 +209,37 @@ 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);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
if (a->rs1 == a->rs2) { /* FMOV */
- gen_check_nanbox_s(cpu_fpr[a->rd], cpu_fpr[a->rs1]);
+ if (!ctx->ext_zfinx) {
+ gen_check_nanbox_s(dest, src1);
+ } else {
+ tcg_gen_ext32s_i64(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 src2 = get_fpr_hs(ctx, a->rs2);
+
+ if (!ctx->ext_zfinx) {
+ TCGv_i64 rs1 = tcg_temp_new_i64();
+ TCGv_i64 rs2 = tcg_temp_new_i64();
+ gen_check_nanbox_s(rs1, src1);
+ gen_check_nanbox_s(rs2, src2);
+
+ /* This formulation retains the nanboxing of rs2 in normal 'F'. */
+ tcg_gen_deposit_i64(dest, rs2, rs1, 0, 31);
+
+ tcg_temp_free_i64(rs1);
+ tcg_temp_free_i64(rs2);
+ } else {
+ tcg_gen_deposit_i64(dest, src2, src1, 0, 31);
+ tcg_gen_ext32s_i64(dest, dest);
+ }
}
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -183,16 +249,27 @@ 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);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ rs1 = tcg_temp_new_i64();
+ if (!ctx->ext_zfinx) {
+ gen_check_nanbox_s(rs1, src1);
+ } else {
+ tcg_gen_mov_i64(rs1, src1);
+ }
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.
@@ -200,13 +277,17 @@ 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);
}
+ /* signed-extended intead of nanboxing for result if enable zfinx */
+ if (ctx->ext_zfinx) {
+ tcg_gen_ext32s_i64(dest, dest);
+ }
+ gen_set_fpr_hs(ctx, a->rd, dest);
tcg_temp_free_i64(rs1);
-
mark_fs_dirty(ctx);
return true;
}
@@ -216,28 +297,45 @@ 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);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
rs1 = tcg_temp_new_i64();
- gen_check_nanbox_s(rs1, cpu_fpr[a->rs1]);
+
+ if (!ctx->ext_zfinx) {
+ gen_check_nanbox_s(rs1, src1);
+ } else {
+ tcg_gen_mov_i64(rs1, src1);
+ }
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 {
+ 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);
+ }
/*
* 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_gen_andi_i64(dest, rs2, MAKE_64BIT_MASK(31, 1));
+ tcg_gen_xor_i64(dest, rs1, dest);
tcg_temp_free_i64(rs2);
}
+ /* signed-extended intead of nanboxing for result if enable zfinx */
+ if (ctx->ext_zfinx) {
+ tcg_gen_ext32s_i64(dest, dest);
+ }
tcg_temp_free_i64(rs1);
-
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -245,10 +343,14 @@ 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);
+
+ 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(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1],
- cpu_fpr[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;
}
@@ -256,10 +358,14 @@ 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);
+
+ 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(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1],
- cpu_fpr[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;
}
@@ -267,12 +373,13 @@ 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;
}
@@ -280,12 +387,13 @@ 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;
}
@@ -294,14 +402,14 @@ 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);
@@ -311,11 +419,13 @@ 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;
}
@@ -323,11 +433,13 @@ 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;
}
@@ -335,11 +447,13 @@ 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;
}
@@ -347,11 +461,12 @@ 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;
}
@@ -359,13 +474,14 @@ 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;
}
@@ -373,13 +489,14 @@ 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;
}
@@ -388,13 +505,14 @@ 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;
}
@@ -403,12 +521,13 @@ 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;
}
@@ -417,12 +536,13 @@ 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;
}
@@ -431,13 +551,14 @@ 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;
}
@@ -446,13 +567,14 @@ 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..6237bb3115 100644
--- a/target/riscv/internals.h
+++ b/target/riscv/internals.h
@@ -46,13 +46,23 @@ enum {
RISCV_FRM_ROD = 8, /* Round to Odd */
};
-static inline uint64_t nanbox_s(float32 f)
+static inline uint64_t nanbox_s(CPURISCVState *env, float32 f)
{
- return f | MAKE_64BIT_MASK(32, 32);
+ /* the value is sign-extended instead of NaN-boxing for zfinx */
+ if (RISCV_CPU(env_cpu(env))->cfg.ext_zfinx) {
+ return (int32_t)f;
+ } else {
+ 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 NaN-boxing 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)) {
diff --git a/target/riscv/translate.c b/target/riscv/translate.c
index c9baf2a87a..6376c35fd7 100644
--- a/target/riscv/translate.c
+++ b/target/riscv/translate.c
@@ -111,6 +111,9 @@ typedef struct DisasContext {
TCGv zero;
/* Space for 3 operands plus 1 extra for address computation. */
TCGv temp[4];
+ /* Space for 4 operands(1 dest and <=3 src) for float point computation */
+ TCGv_i64 ftemp[4];
+ uint8_t nftemp;
/* PointerMasking extension */
bool pm_mask_enabled;
bool pm_base_enabled;
@@ -377,6 +380,86 @@ static void gen_set_gpr128(DisasContext *ctx, int reg_num, TCGv rl, TCGv rh)
}
}
+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) {
+ return cpu_fpr[reg_num];
+ }
+
+ if (reg_num == 0) {
+ return tcg_constant_i64(0);
+ }
+ switch (get_xl(ctx)) {
+ case MXL_RV32:
+#ifdef TARGET_RISCV32
+ {
+ TCGv_i64 t = ftemp_new(ctx);
+ tcg_gen_ext_i32_i64(t, cpu_gpr[reg_num]);
+ return t;
+ }
+#else
+ /* fall through */
+ case MXL_RV64:
+ return cpu_gpr[reg_num];
+#endif
+ default:
+ g_assert_not_reached();
+ }
+}
+
+static TCGv_i64 dest_fpr(DisasContext *ctx, int reg_num)
+{
+ if (!ctx->ext_zfinx) {
+ return cpu_fpr[reg_num];
+ }
+
+ if (reg_num == 0) {
+ return ftemp_new(ctx);
+ }
+
+ switch (get_xl(ctx)) {
+ case MXL_RV32:
+ return ftemp_new(ctx);
+#ifdef TARGET_RISCV64
+ case MXL_RV64:
+ return cpu_gpr[reg_num];
+#endif
+ default:
+ g_assert_not_reached();
+ }
+}
+
+/* assume t is nanboxing (for normal) or sign-extended (for zfinx) */
+static void gen_set_fpr_hs(DisasContext *ctx, int reg_num, TCGv_i64 t)
+{
+ if (!ctx->ext_zfinx) {
+ tcg_gen_mov_i64(cpu_fpr[reg_num], t);
+ return;
+ }
+ if (reg_num != 0) {
+ switch (get_xl(ctx)) {
+ case MXL_RV32:
+#ifdef TARGET_RISCV32
+ tcg_gen_extrl_i64_i32(cpu_gpr[reg_num], t);
+ break;
+#else
+ /* fall through */
+ case MXL_RV64:
+ tcg_gen_mov_i64(cpu_gpr[reg_num], t);
+ break;
+#endif
+ default:
+ g_assert_not_reached();
+ }
+ }
+}
+
static void gen_jal(DisasContext *ctx, int rd, target_ulong imm)
{
target_ulong next_pc;
@@ -940,6 +1023,8 @@ 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_mask_enabled = FIELD_EX32(tb_flags, TB_FLAGS, PM_MASK_ENABLED);
ctx->pm_base_enabled = FIELD_EX32(tb_flags, TB_FLAGS, PM_BASE_ENABLED);
ctx->zero = tcg_constant_tl(0);
@@ -961,16 +1046,22 @@ static void riscv_tr_translate_insn(DisasContextBase *dcbase, CPUState *cpu)
DisasContext *ctx = container_of(dcbase, DisasContext, base);
CPURISCVState *env = cpu->env_ptr;
uint16_t opcode16 = translator_lduw(env, &ctx->base, ctx->base.pc_next);
+ int i;
ctx->ol = ctx->xl;
decode_opc(env, ctx, opcode16);
ctx->base.pc_next = ctx->pc_succ_insn;
- for (int i = ctx->ntemp - 1; i >= 0; --i) {
+ for (i = ctx->ntemp - 1; i >= 0; --i) {
tcg_temp_free(ctx->temp[i]);
ctx->temp[i] = NULL;
}
ctx->ntemp = 0;
+ for (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] 14+ messages in thread
* [PATCH v5 3/6] target/riscv: add support for zfinx
@ 2022-01-28 13:16 ` Weiwei Li
0 siblings, 0 replies; 14+ messages in thread
From: Weiwei Li @ 2022-01-28 13:16 UTC (permalink / raw)
To: richard.henderson, palmer, alistair.francis, bin.meng,
qemu-riscv, qemu-devel
Cc: wangjunqiang, lazyparser, ardxwe, Weiwei Li
- 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: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
---
target/riscv/fpu_helper.c | 89 +++----
target/riscv/helper.h | 2 +-
target/riscv/insn_trans/trans_rvf.c.inc | 314 ++++++++++++++++--------
target/riscv/internals.h | 16 +-
target/riscv/translate.c | 93 ++++++-
5 files changed, 369 insertions(+), 145 deletions(-)
diff --git a/target/riscv/fpu_helper.c b/target/riscv/fpu_helper.c
index 4a5982d594..63ca703459 100644
--- a/target/riscv/fpu_helper.c
+++ b/target/riscv/fpu_helper.c
@@ -98,10 +98,11 @@ 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);
- return nanbox_s(float32_muladd(frs1, frs2, frs3, flags, &env->fp_status));
+ float32 frs1 = check_nanbox_s(env, rs1);
+ float32 frs2 = check_nanbox_s(env, rs2);
+ float32 frs3 = check_nanbox_s(env, rs3);
+ return nanbox_s(env, float32_muladd(frs1, frs2, frs3, flags,
+ &env->fp_status));
}
uint64_t helper_fmadd_s(CPURISCVState *env, uint64_t frs1, uint64_t frs2,
@@ -183,124 +184,124 @@ 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);
- return nanbox_s(float32_add(frs1, frs2, &env->fp_status));
+ float32 frs1 = check_nanbox_s(env, rs1);
+ float32 frs2 = check_nanbox_s(env, rs2);
+ return nanbox_s(env, 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);
- return nanbox_s(float32_sub(frs1, frs2, &env->fp_status));
+ float32 frs1 = check_nanbox_s(env, rs1);
+ float32 frs2 = check_nanbox_s(env, rs2);
+ return nanbox_s(env, 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);
- return nanbox_s(float32_mul(frs1, frs2, &env->fp_status));
+ float32 frs1 = check_nanbox_s(env, rs1);
+ float32 frs2 = check_nanbox_s(env, rs2);
+ return nanbox_s(env, 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);
- return nanbox_s(float32_div(frs1, frs2, &env->fp_status));
+ float32 frs1 = check_nanbox_s(env, rs1);
+ float32 frs2 = check_nanbox_s(env, rs2);
+ return nanbox_s(env, 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);
- return nanbox_s(env->priv_ver < PRIV_VERSION_1_11_0 ?
+ float32 frs1 = check_nanbox_s(env, rs1);
+ float32 frs2 = check_nanbox_s(env, rs2);
+ return nanbox_s(env, env->priv_ver < PRIV_VERSION_1_11_0 ?
float32_minnum(frs1, frs2, &env->fp_status) :
float32_minimum_number(frs1, frs2, &env->fp_status));
}
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);
- return nanbox_s(env->priv_ver < PRIV_VERSION_1_11_0 ?
+ float32 frs1 = check_nanbox_s(env, rs1);
+ float32 frs2 = check_nanbox_s(env, rs2);
+ return nanbox_s(env, env->priv_ver < PRIV_VERSION_1_11_0 ?
float32_maxnum(frs1, frs2, &env->fp_status) :
float32_maximum_number(frs1, frs2, &env->fp_status));
}
uint64_t helper_fsqrt_s(CPURISCVState *env, uint64_t rs1)
{
- float32 frs1 = check_nanbox_s(rs1);
- return nanbox_s(float32_sqrt(frs1, &env->fp_status));
+ float32 frs1 = check_nanbox_s(env, rs1);
+ return nanbox_s(env, 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);
}
uint64_t helper_fcvt_s_w(CPURISCVState *env, target_ulong rs1)
{
- return nanbox_s(int32_to_float32((int32_t)rs1, &env->fp_status));
+ return nanbox_s(env, int32_to_float32((int32_t)rs1, &env->fp_status));
}
uint64_t helper_fcvt_s_wu(CPURISCVState *env, target_ulong rs1)
{
- return nanbox_s(uint32_to_float32((uint32_t)rs1, &env->fp_status));
+ return nanbox_s(env, uint32_to_float32((uint32_t)rs1, &env->fp_status));
}
uint64_t helper_fcvt_s_l(CPURISCVState *env, target_ulong rs1)
{
- return nanbox_s(int64_to_float32(rs1, &env->fp_status));
+ return nanbox_s(env, int64_to_float32(rs1, &env->fp_status));
}
uint64_t helper_fcvt_s_lu(CPURISCVState *env, target_ulong rs1)
{
- return nanbox_s(uint64_to_float32(rs1, &env->fp_status));
+ return nanbox_s(env, 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);
}
@@ -340,12 +341,12 @@ uint64_t helper_fmax_d(CPURISCVState *env, uint64_t frs1, uint64_t frs2)
uint64_t helper_fcvt_s_d(CPURISCVState *env, uint64_t rs1)
{
- return nanbox_s(float64_to_float32(rs1, &env->fp_status));
+ return nanbox_s(env, float64_to_float32(rs1, &env->fp_status));
}
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);
}
@@ -539,14 +540,14 @@ 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));
}
uint64_t helper_fcvt_s_h(CPURISCVState *env, uint64_t rs1)
{
float16 frs1 = check_nanbox_h(rs1);
- return nanbox_s(float16_to_float32(frs1, true, &env->fp_status));
+ return nanbox_s(env, float16_to_float32(frs1, true, &env->fp_status));
}
uint64_t helper_fcvt_h_d(CPURISCVState *env, uint64_t rs1)
diff --git a/target/riscv/helper.h b/target/riscv/helper.h
index 72cc2582f4..89195aad9d 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 0aac87f7db..68a305edd6 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)
@@ -55,10 +62,16 @@ 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;
}
@@ -66,10 +79,16 @@ 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;
}
@@ -77,10 +96,16 @@ 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;
}
@@ -88,10 +113,16 @@ 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;
}
@@ -99,11 +130,15 @@ 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;
}
@@ -111,11 +146,15 @@ 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;
}
@@ -123,11 +162,15 @@ 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;
}
@@ -135,11 +178,15 @@ 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;
}
@@ -147,10 +194,14 @@ 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;
}
@@ -158,22 +209,37 @@ 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);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
if (a->rs1 == a->rs2) { /* FMOV */
- gen_check_nanbox_s(cpu_fpr[a->rd], cpu_fpr[a->rs1]);
+ if (!ctx->ext_zfinx) {
+ gen_check_nanbox_s(dest, src1);
+ } else {
+ tcg_gen_ext32s_i64(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 src2 = get_fpr_hs(ctx, a->rs2);
+
+ if (!ctx->ext_zfinx) {
+ TCGv_i64 rs1 = tcg_temp_new_i64();
+ TCGv_i64 rs2 = tcg_temp_new_i64();
+ gen_check_nanbox_s(rs1, src1);
+ gen_check_nanbox_s(rs2, src2);
+
+ /* This formulation retains the nanboxing of rs2 in normal 'F'. */
+ tcg_gen_deposit_i64(dest, rs2, rs1, 0, 31);
+
+ tcg_temp_free_i64(rs1);
+ tcg_temp_free_i64(rs2);
+ } else {
+ tcg_gen_deposit_i64(dest, src2, src1, 0, 31);
+ tcg_gen_ext32s_i64(dest, dest);
+ }
}
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -183,16 +249,27 @@ 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);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
+ rs1 = tcg_temp_new_i64();
+ if (!ctx->ext_zfinx) {
+ gen_check_nanbox_s(rs1, src1);
+ } else {
+ tcg_gen_mov_i64(rs1, src1);
+ }
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.
@@ -200,13 +277,17 @@ 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);
}
+ /* signed-extended intead of nanboxing for result if enable zfinx */
+ if (ctx->ext_zfinx) {
+ tcg_gen_ext32s_i64(dest, dest);
+ }
+ gen_set_fpr_hs(ctx, a->rd, dest);
tcg_temp_free_i64(rs1);
-
mark_fs_dirty(ctx);
return true;
}
@@ -216,28 +297,45 @@ 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);
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
rs1 = tcg_temp_new_i64();
- gen_check_nanbox_s(rs1, cpu_fpr[a->rs1]);
+
+ if (!ctx->ext_zfinx) {
+ gen_check_nanbox_s(rs1, src1);
+ } else {
+ tcg_gen_mov_i64(rs1, src1);
+ }
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 {
+ 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);
+ }
/*
* 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_gen_andi_i64(dest, rs2, MAKE_64BIT_MASK(31, 1));
+ tcg_gen_xor_i64(dest, rs1, dest);
tcg_temp_free_i64(rs2);
}
+ /* signed-extended intead of nanboxing for result if enable zfinx */
+ if (ctx->ext_zfinx) {
+ tcg_gen_ext32s_i64(dest, dest);
+ }
tcg_temp_free_i64(rs1);
-
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -245,10 +343,14 @@ 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);
+
+ 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(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1],
- cpu_fpr[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;
}
@@ -256,10 +358,14 @@ 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);
+
+ 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(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1],
- cpu_fpr[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;
}
@@ -267,12 +373,13 @@ 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;
}
@@ -280,12 +387,13 @@ 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;
}
@@ -294,14 +402,14 @@ 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);
@@ -311,11 +419,13 @@ 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;
}
@@ -323,11 +433,13 @@ 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;
}
@@ -335,11 +447,13 @@ 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;
}
@@ -347,11 +461,12 @@ 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;
}
@@ -359,13 +474,14 @@ 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;
}
@@ -373,13 +489,14 @@ 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;
}
@@ -388,13 +505,14 @@ 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;
}
@@ -403,12 +521,13 @@ 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;
}
@@ -417,12 +536,13 @@ 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;
}
@@ -431,13 +551,14 @@ 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;
}
@@ -446,13 +567,14 @@ 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..6237bb3115 100644
--- a/target/riscv/internals.h
+++ b/target/riscv/internals.h
@@ -46,13 +46,23 @@ enum {
RISCV_FRM_ROD = 8, /* Round to Odd */
};
-static inline uint64_t nanbox_s(float32 f)
+static inline uint64_t nanbox_s(CPURISCVState *env, float32 f)
{
- return f | MAKE_64BIT_MASK(32, 32);
+ /* the value is sign-extended instead of NaN-boxing for zfinx */
+ if (RISCV_CPU(env_cpu(env))->cfg.ext_zfinx) {
+ return (int32_t)f;
+ } else {
+ 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 NaN-boxing 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)) {
diff --git a/target/riscv/translate.c b/target/riscv/translate.c
index c9baf2a87a..6376c35fd7 100644
--- a/target/riscv/translate.c
+++ b/target/riscv/translate.c
@@ -111,6 +111,9 @@ typedef struct DisasContext {
TCGv zero;
/* Space for 3 operands plus 1 extra for address computation. */
TCGv temp[4];
+ /* Space for 4 operands(1 dest and <=3 src) for float point computation */
+ TCGv_i64 ftemp[4];
+ uint8_t nftemp;
/* PointerMasking extension */
bool pm_mask_enabled;
bool pm_base_enabled;
@@ -377,6 +380,86 @@ static void gen_set_gpr128(DisasContext *ctx, int reg_num, TCGv rl, TCGv rh)
}
}
+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) {
+ return cpu_fpr[reg_num];
+ }
+
+ if (reg_num == 0) {
+ return tcg_constant_i64(0);
+ }
+ switch (get_xl(ctx)) {
+ case MXL_RV32:
+#ifdef TARGET_RISCV32
+ {
+ TCGv_i64 t = ftemp_new(ctx);
+ tcg_gen_ext_i32_i64(t, cpu_gpr[reg_num]);
+ return t;
+ }
+#else
+ /* fall through */
+ case MXL_RV64:
+ return cpu_gpr[reg_num];
+#endif
+ default:
+ g_assert_not_reached();
+ }
+}
+
+static TCGv_i64 dest_fpr(DisasContext *ctx, int reg_num)
+{
+ if (!ctx->ext_zfinx) {
+ return cpu_fpr[reg_num];
+ }
+
+ if (reg_num == 0) {
+ return ftemp_new(ctx);
+ }
+
+ switch (get_xl(ctx)) {
+ case MXL_RV32:
+ return ftemp_new(ctx);
+#ifdef TARGET_RISCV64
+ case MXL_RV64:
+ return cpu_gpr[reg_num];
+#endif
+ default:
+ g_assert_not_reached();
+ }
+}
+
+/* assume t is nanboxing (for normal) or sign-extended (for zfinx) */
+static void gen_set_fpr_hs(DisasContext *ctx, int reg_num, TCGv_i64 t)
+{
+ if (!ctx->ext_zfinx) {
+ tcg_gen_mov_i64(cpu_fpr[reg_num], t);
+ return;
+ }
+ if (reg_num != 0) {
+ switch (get_xl(ctx)) {
+ case MXL_RV32:
+#ifdef TARGET_RISCV32
+ tcg_gen_extrl_i64_i32(cpu_gpr[reg_num], t);
+ break;
+#else
+ /* fall through */
+ case MXL_RV64:
+ tcg_gen_mov_i64(cpu_gpr[reg_num], t);
+ break;
+#endif
+ default:
+ g_assert_not_reached();
+ }
+ }
+}
+
static void gen_jal(DisasContext *ctx, int rd, target_ulong imm)
{
target_ulong next_pc;
@@ -940,6 +1023,8 @@ 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_mask_enabled = FIELD_EX32(tb_flags, TB_FLAGS, PM_MASK_ENABLED);
ctx->pm_base_enabled = FIELD_EX32(tb_flags, TB_FLAGS, PM_BASE_ENABLED);
ctx->zero = tcg_constant_tl(0);
@@ -961,16 +1046,22 @@ static void riscv_tr_translate_insn(DisasContextBase *dcbase, CPUState *cpu)
DisasContext *ctx = container_of(dcbase, DisasContext, base);
CPURISCVState *env = cpu->env_ptr;
uint16_t opcode16 = translator_lduw(env, &ctx->base, ctx->base.pc_next);
+ int i;
ctx->ol = ctx->xl;
decode_opc(env, ctx, opcode16);
ctx->base.pc_next = ctx->pc_succ_insn;
- for (int i = ctx->ntemp - 1; i >= 0; --i) {
+ for (i = ctx->ntemp - 1; i >= 0; --i) {
tcg_temp_free(ctx->temp[i]);
ctx->temp[i] = NULL;
}
ctx->ntemp = 0;
+ for (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] 14+ messages in thread
* [PATCH v5 4/6] target/riscv: add support for zdinx
2022-01-28 13:16 ` Weiwei Li
@ 2022-01-28 13:16 ` Weiwei Li
-1 siblings, 0 replies; 14+ messages in thread
From: Weiwei Li @ 2022-01-28 13:16 UTC (permalink / raw)
To: richard.henderson, palmer, alistair.francis, bin.meng,
qemu-riscv, qemu-devel
Cc: wangjunqiang, Weiwei Li, 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: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
---
target/riscv/insn_trans/trans_rvd.c.inc | 285 +++++++++++++++++-------
target/riscv/translate.c | 52 +++++
2 files changed, 259 insertions(+), 78 deletions(-)
diff --git a/target/riscv/insn_trans/trans_rvd.c.inc b/target/riscv/insn_trans/trans_rvd.c.inc
index 091ed3a8ad..748343f261 100644
--- a/target/riscv/insn_trans/trans_rvd.c.inc
+++ b/target/riscv/insn_trans/trans_rvd.c.inc
@@ -18,6 +18,19 @@
* 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)
+
+#define REQUIRE_EVEN(ctx, reg) do { \
+ if (ctx->ext_zdinx && (get_xl(ctx) == MXL_RV32) && \
+ ((reg) & 0x1)) { \
+ return false; \
+ } \
+} while (0)
+
static bool trans_fld(DisasContext *ctx, arg_fld *a)
{
TCGv addr;
@@ -47,10 +60,17 @@ 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);
+ REQUIRE_EVEN(ctx, a->rd | a->rs1 | a->rs2 | a->rs3);
+
+ 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;
}
@@ -58,10 +78,17 @@ 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);
+ REQUIRE_EVEN(ctx, a->rd | a->rs1 | a->rs2 | a->rs3);
+
+ 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;
}
@@ -69,10 +96,17 @@ 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);
+ REQUIRE_EVEN(ctx, a->rd | a->rs1 | a->rs2 | a->rs3);
+
+ 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;
}
@@ -80,10 +114,17 @@ 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);
+ REQUIRE_EVEN(ctx, a->rd | a->rs1 | a->rs2 | a->rs3);
+
+ 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;
}
@@ -91,12 +132,16 @@ 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);
+ REQUIRE_EVEN(ctx, a->rd | a->rs1 | a->rs2);
- 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;
}
@@ -104,12 +149,16 @@ 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);
+ REQUIRE_EVEN(ctx, a->rd | a->rs1 | a->rs2);
- 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;
}
@@ -117,12 +166,16 @@ 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);
+ REQUIRE_EVEN(ctx, a->rd | a->rs1 | a->rs2);
- 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;
}
@@ -130,12 +183,16 @@ 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);
+ REQUIRE_EVEN(ctx, a->rd | a->rs1 | a->rs2);
- 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;
}
@@ -143,23 +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);
+ REQUIRE_EVEN(ctx, a->rd | a->rs1);
- 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);
+ REQUIRE_EVEN(ctx, a->rd | a->rs1 | a->rs2);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
if (a->rs1 == a->rs2) { /* FMOV */
- tcg_gen_mov_i64(cpu_fpr[a->rd], cpu_fpr[a->rs1]);
+ dest = get_fpr_d(ctx, a->rs1);
} else {
- tcg_gen_deposit_i64(cpu_fpr[a->rd], cpu_fpr[a->rs2],
- cpu_fpr[a->rs1], 0, 63);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_d(ctx, a->rs2);
+ tcg_gen_deposit_i64(dest, src2, src1, 0, 63);
}
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -167,15 +235,22 @@ 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);
+ REQUIRE_EVEN(ctx, a->rd | a->rs1 | a->rs2);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+
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 src2 = get_fpr_d(ctx, a->rs2);
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;
}
@@ -183,15 +258,22 @@ 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);
+ REQUIRE_EVEN(ctx, a->rd | a->rs1 | a->rs2);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+
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 src2 = get_fpr_d(ctx, a->rs2);
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;
}
@@ -199,11 +281,15 @@ 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);
+ REQUIRE_EVEN(ctx, a->rd | a->rs1 | a->rs2);
- 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;
}
@@ -211,11 +297,15 @@ 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);
+ REQUIRE_EVEN(ctx, a->rd | a->rs1 | a->rs2);
- 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;
}
@@ -223,11 +313,15 @@ 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);
+ REQUIRE_EVEN(ctx, a->rs1);
- 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;
}
@@ -235,11 +329,15 @@ 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);
+ REQUIRE_EVEN(ctx, a->rd);
- 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;
}
@@ -247,11 +345,14 @@ 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);
+ REQUIRE_EVEN(ctx, a->rs1 | a->rs2);
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;
}
@@ -259,11 +360,14 @@ 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);
+ REQUIRE_EVEN(ctx, a->rs1 | a->rs2);
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;
}
@@ -271,11 +375,14 @@ 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);
+ REQUIRE_EVEN(ctx, a->rs1 | a->rs2);
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;
}
@@ -283,11 +390,13 @@ 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);
+ REQUIRE_EVEN(ctx, a->rs1);
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;
}
@@ -295,12 +404,14 @@ 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);
+ REQUIRE_EVEN(ctx, a->rs1);
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;
}
@@ -308,12 +419,14 @@ 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);
+ REQUIRE_EVEN(ctx, a->rs1);
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;
}
@@ -321,12 +434,15 @@ 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);
+ REQUIRE_EVEN(ctx, a->rd);
+ 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;
@@ -335,12 +451,15 @@ 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);
+ REQUIRE_EVEN(ctx, a->rd);
+ 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;
@@ -350,12 +469,14 @@ 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);
+ REQUIRE_EVEN(ctx, a->rs1);
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;
}
@@ -364,12 +485,14 @@ 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);
+ REQUIRE_EVEN(ctx, a->rs1);
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;
}
@@ -392,12 +515,15 @@ 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);
+ REQUIRE_EVEN(ctx, a->rd);
+ 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;
@@ -407,12 +533,15 @@ 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);
+ REQUIRE_EVEN(ctx, a->rd);
+ 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;
diff --git a/target/riscv/translate.c b/target/riscv/translate.c
index 6376c35fd7..091b5b25cc 100644
--- a/target/riscv/translate.c
+++ b/target/riscv/translate.c
@@ -413,6 +413,31 @@ static TCGv_i64 get_fpr_hs(DisasContext *ctx, int reg_num)
}
}
+static TCGv_i64 get_fpr_d(DisasContext *ctx, int reg_num)
+{
+ if (!ctx->ext_zfinx) {
+ return cpu_fpr[reg_num];
+ }
+
+ if (reg_num == 0) {
+ return tcg_constant_i64(0);
+ }
+ switch (get_xl(ctx)) {
+ case MXL_RV32:
+ {
+ TCGv_i64 t = ftemp_new(ctx);
+ tcg_gen_concat_tl_i64(t, cpu_gpr[reg_num], cpu_gpr[reg_num + 1]);
+ return t;
+ }
+#ifdef TARGET_RISCV64
+ case MXL_RV64:
+ return cpu_gpr[reg_num];
+#endif
+ default:
+ g_assert_not_reached();
+ }
+}
+
static TCGv_i64 dest_fpr(DisasContext *ctx, int reg_num)
{
if (!ctx->ext_zfinx) {
@@ -460,6 +485,33 @@ static void gen_set_fpr_hs(DisasContext *ctx, int reg_num, TCGv_i64 t)
}
}
+static void gen_set_fpr_d(DisasContext *ctx, int reg_num, TCGv_i64 t)
+{
+ if (!ctx->ext_zfinx) {
+ tcg_gen_mov_i64(cpu_fpr[reg_num], t);
+ return;
+ }
+
+ if (reg_num != 0) {
+ switch (get_xl(ctx)) {
+ case MXL_RV32:
+#ifdef TARGET_RISCV32
+ tcg_gen_extr_i64_i32(cpu_gpr[reg_num], 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();
+ }
+ }
+}
+
static void gen_jal(DisasContext *ctx, int rd, target_ulong imm)
{
target_ulong next_pc;
--
2.17.1
^ permalink raw reply related [flat|nested] 14+ messages in thread
* [PATCH v5 4/6] target/riscv: add support for zdinx
@ 2022-01-28 13:16 ` Weiwei Li
0 siblings, 0 replies; 14+ messages in thread
From: Weiwei Li @ 2022-01-28 13:16 UTC (permalink / raw)
To: richard.henderson, palmer, alistair.francis, bin.meng,
qemu-riscv, qemu-devel
Cc: wangjunqiang, lazyparser, ardxwe, Weiwei Li
-- update extension check REQUIRE_ZDINX_OR_D
-- update double float point register read/write
Co-authored-by: ardxwe <ardxwe@gmail.com>
Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
---
target/riscv/insn_trans/trans_rvd.c.inc | 285 +++++++++++++++++-------
target/riscv/translate.c | 52 +++++
2 files changed, 259 insertions(+), 78 deletions(-)
diff --git a/target/riscv/insn_trans/trans_rvd.c.inc b/target/riscv/insn_trans/trans_rvd.c.inc
index 091ed3a8ad..748343f261 100644
--- a/target/riscv/insn_trans/trans_rvd.c.inc
+++ b/target/riscv/insn_trans/trans_rvd.c.inc
@@ -18,6 +18,19 @@
* 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)
+
+#define REQUIRE_EVEN(ctx, reg) do { \
+ if (ctx->ext_zdinx && (get_xl(ctx) == MXL_RV32) && \
+ ((reg) & 0x1)) { \
+ return false; \
+ } \
+} while (0)
+
static bool trans_fld(DisasContext *ctx, arg_fld *a)
{
TCGv addr;
@@ -47,10 +60,17 @@ 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);
+ REQUIRE_EVEN(ctx, a->rd | a->rs1 | a->rs2 | a->rs3);
+
+ 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;
}
@@ -58,10 +78,17 @@ 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);
+ REQUIRE_EVEN(ctx, a->rd | a->rs1 | a->rs2 | a->rs3);
+
+ 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;
}
@@ -69,10 +96,17 @@ 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);
+ REQUIRE_EVEN(ctx, a->rd | a->rs1 | a->rs2 | a->rs3);
+
+ 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;
}
@@ -80,10 +114,17 @@ 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);
+ REQUIRE_EVEN(ctx, a->rd | a->rs1 | a->rs2 | a->rs3);
+
+ 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;
}
@@ -91,12 +132,16 @@ 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);
+ REQUIRE_EVEN(ctx, a->rd | a->rs1 | a->rs2);
- 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;
}
@@ -104,12 +149,16 @@ 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);
+ REQUIRE_EVEN(ctx, a->rd | a->rs1 | a->rs2);
- 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;
}
@@ -117,12 +166,16 @@ 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);
+ REQUIRE_EVEN(ctx, a->rd | a->rs1 | a->rs2);
- 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;
}
@@ -130,12 +183,16 @@ 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);
+ REQUIRE_EVEN(ctx, a->rd | a->rs1 | a->rs2);
- 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;
}
@@ -143,23 +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);
+ REQUIRE_EVEN(ctx, a->rd | a->rs1);
- 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);
+ REQUIRE_EVEN(ctx, a->rd | a->rs1 | a->rs2);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
if (a->rs1 == a->rs2) { /* FMOV */
- tcg_gen_mov_i64(cpu_fpr[a->rd], cpu_fpr[a->rs1]);
+ dest = get_fpr_d(ctx, a->rs1);
} else {
- tcg_gen_deposit_i64(cpu_fpr[a->rd], cpu_fpr[a->rs2],
- cpu_fpr[a->rs1], 0, 63);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+ TCGv_i64 src2 = get_fpr_d(ctx, a->rs2);
+ tcg_gen_deposit_i64(dest, src2, src1, 0, 63);
}
+ gen_set_fpr_d(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -167,15 +235,22 @@ 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);
+ REQUIRE_EVEN(ctx, a->rd | a->rs1 | a->rs2);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+
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 src2 = get_fpr_d(ctx, a->rs2);
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;
}
@@ -183,15 +258,22 @@ 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);
+ REQUIRE_EVEN(ctx, a->rd | a->rs1 | a->rs2);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_d(ctx, a->rs1);
+
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 src2 = get_fpr_d(ctx, a->rs2);
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;
}
@@ -199,11 +281,15 @@ 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);
+ REQUIRE_EVEN(ctx, a->rd | a->rs1 | a->rs2);
- 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;
}
@@ -211,11 +297,15 @@ 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);
+ REQUIRE_EVEN(ctx, a->rd | a->rs1 | a->rs2);
- 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;
}
@@ -223,11 +313,15 @@ 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);
+ REQUIRE_EVEN(ctx, a->rs1);
- 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;
}
@@ -235,11 +329,15 @@ 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);
+ REQUIRE_EVEN(ctx, a->rd);
- 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;
}
@@ -247,11 +345,14 @@ 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);
+ REQUIRE_EVEN(ctx, a->rs1 | a->rs2);
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;
}
@@ -259,11 +360,14 @@ 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);
+ REQUIRE_EVEN(ctx, a->rs1 | a->rs2);
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;
}
@@ -271,11 +375,14 @@ 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);
+ REQUIRE_EVEN(ctx, a->rs1 | a->rs2);
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;
}
@@ -283,11 +390,13 @@ 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);
+ REQUIRE_EVEN(ctx, a->rs1);
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;
}
@@ -295,12 +404,14 @@ 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);
+ REQUIRE_EVEN(ctx, a->rs1);
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;
}
@@ -308,12 +419,14 @@ 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);
+ REQUIRE_EVEN(ctx, a->rs1);
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;
}
@@ -321,12 +434,15 @@ 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);
+ REQUIRE_EVEN(ctx, a->rd);
+ 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;
@@ -335,12 +451,15 @@ 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);
+ REQUIRE_EVEN(ctx, a->rd);
+ 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;
@@ -350,12 +469,14 @@ 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);
+ REQUIRE_EVEN(ctx, a->rs1);
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;
}
@@ -364,12 +485,14 @@ 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);
+ REQUIRE_EVEN(ctx, a->rs1);
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;
}
@@ -392,12 +515,15 @@ 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);
+ REQUIRE_EVEN(ctx, a->rd);
+ 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;
@@ -407,12 +533,15 @@ 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);
+ REQUIRE_EVEN(ctx, a->rd);
+ 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;
diff --git a/target/riscv/translate.c b/target/riscv/translate.c
index 6376c35fd7..091b5b25cc 100644
--- a/target/riscv/translate.c
+++ b/target/riscv/translate.c
@@ -413,6 +413,31 @@ static TCGv_i64 get_fpr_hs(DisasContext *ctx, int reg_num)
}
}
+static TCGv_i64 get_fpr_d(DisasContext *ctx, int reg_num)
+{
+ if (!ctx->ext_zfinx) {
+ return cpu_fpr[reg_num];
+ }
+
+ if (reg_num == 0) {
+ return tcg_constant_i64(0);
+ }
+ switch (get_xl(ctx)) {
+ case MXL_RV32:
+ {
+ TCGv_i64 t = ftemp_new(ctx);
+ tcg_gen_concat_tl_i64(t, cpu_gpr[reg_num], cpu_gpr[reg_num + 1]);
+ return t;
+ }
+#ifdef TARGET_RISCV64
+ case MXL_RV64:
+ return cpu_gpr[reg_num];
+#endif
+ default:
+ g_assert_not_reached();
+ }
+}
+
static TCGv_i64 dest_fpr(DisasContext *ctx, int reg_num)
{
if (!ctx->ext_zfinx) {
@@ -460,6 +485,33 @@ static void gen_set_fpr_hs(DisasContext *ctx, int reg_num, TCGv_i64 t)
}
}
+static void gen_set_fpr_d(DisasContext *ctx, int reg_num, TCGv_i64 t)
+{
+ if (!ctx->ext_zfinx) {
+ tcg_gen_mov_i64(cpu_fpr[reg_num], t);
+ return;
+ }
+
+ if (reg_num != 0) {
+ switch (get_xl(ctx)) {
+ case MXL_RV32:
+#ifdef TARGET_RISCV32
+ tcg_gen_extr_i64_i32(cpu_gpr[reg_num], 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();
+ }
+ }
+}
+
static void gen_jal(DisasContext *ctx, int rd, target_ulong imm)
{
target_ulong next_pc;
--
2.17.1
^ permalink raw reply related [flat|nested] 14+ messages in thread
* [PATCH v5 5/6] target/riscv: add support for zhinx/zhinxmin
2022-01-28 13:16 ` Weiwei Li
@ 2022-01-28 13:16 ` Weiwei Li
-1 siblings, 0 replies; 14+ messages in thread
From: Weiwei Li @ 2022-01-28 13:16 UTC (permalink / raw)
To: richard.henderson, palmer, alistair.francis, bin.meng,
qemu-riscv, qemu-devel
Cc: wangjunqiang, Weiwei Li, 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: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
---
target/riscv/fpu_helper.c | 89 +++---
target/riscv/helper.h | 2 +-
target/riscv/insn_trans/trans_rvzfh.c.inc | 332 +++++++++++++++-------
target/riscv/internals.h | 16 +-
4 files changed, 296 insertions(+), 143 deletions(-)
diff --git a/target/riscv/fpu_helper.c b/target/riscv/fpu_helper.c
index 63ca703459..5699c9517f 100644
--- a/target/riscv/fpu_helper.c
+++ b/target/riscv/fpu_helper.c
@@ -89,10 +89,11 @@ 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);
- return nanbox_h(float16_muladd(frs1, frs2, frs3, flags, &env->fp_status));
+ float16 frs1 = check_nanbox_h(env, rs1);
+ float16 frs2 = check_nanbox_h(env, rs2);
+ float16 frs3 = check_nanbox_h(env, rs3);
+ return nanbox_h(env, float16_muladd(frs1, frs2, frs3, flags,
+ &env->fp_status));
}
static uint64_t do_fmadd_s(CPURISCVState *env, uint64_t rs1, uint64_t rs2,
@@ -417,146 +418,146 @@ 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);
- return nanbox_h(float16_add(frs1, frs2, &env->fp_status));
+ float16 frs1 = check_nanbox_h(env, rs1);
+ float16 frs2 = check_nanbox_h(env, rs2);
+ return nanbox_h(env, 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);
- return nanbox_h(float16_sub(frs1, frs2, &env->fp_status));
+ float16 frs1 = check_nanbox_h(env, rs1);
+ float16 frs2 = check_nanbox_h(env, rs2);
+ return nanbox_h(env, 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);
- return nanbox_h(float16_mul(frs1, frs2, &env->fp_status));
+ float16 frs1 = check_nanbox_h(env, rs1);
+ float16 frs2 = check_nanbox_h(env, rs2);
+ return nanbox_h(env, 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);
- return nanbox_h(float16_div(frs1, frs2, &env->fp_status));
+ float16 frs1 = check_nanbox_h(env, rs1);
+ float16 frs2 = check_nanbox_h(env, rs2);
+ return nanbox_h(env, 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);
- return nanbox_h(env->priv_ver < PRIV_VERSION_1_11_0 ?
+ float16 frs1 = check_nanbox_h(env, rs1);
+ float16 frs2 = check_nanbox_h(env, rs2);
+ return nanbox_h(env, env->priv_ver < PRIV_VERSION_1_11_0 ?
float16_minnum(frs1, frs2, &env->fp_status) :
float16_minimum_number(frs1, frs2, &env->fp_status));
}
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);
- return nanbox_h(env->priv_ver < PRIV_VERSION_1_11_0 ?
+ float16 frs1 = check_nanbox_h(env, rs1);
+ float16 frs2 = check_nanbox_h(env, rs2);
+ return nanbox_h(env, env->priv_ver < PRIV_VERSION_1_11_0 ?
float16_maxnum(frs1, frs2, &env->fp_status) :
float16_maximum_number(frs1, frs2, &env->fp_status));
}
uint64_t helper_fsqrt_h(CPURISCVState *env, uint64_t rs1)
{
- float16 frs1 = check_nanbox_h(rs1);
- return nanbox_h(float16_sqrt(frs1, &env->fp_status));
+ float16 frs1 = check_nanbox_h(env, rs1);
+ return nanbox_h(env, 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);
}
uint64_t helper_fcvt_h_w(CPURISCVState *env, target_ulong rs1)
{
- return nanbox_h(int32_to_float16((int32_t)rs1, &env->fp_status));
+ return nanbox_h(env, int32_to_float16((int32_t)rs1, &env->fp_status));
}
uint64_t helper_fcvt_h_wu(CPURISCVState *env, target_ulong rs1)
{
- return nanbox_h(uint32_to_float16((uint32_t)rs1, &env->fp_status));
+ return nanbox_h(env, uint32_to_float16((uint32_t)rs1, &env->fp_status));
}
uint64_t helper_fcvt_h_l(CPURISCVState *env, target_ulong rs1)
{
- return nanbox_h(int64_to_float16(rs1, &env->fp_status));
+ return nanbox_h(env, int64_to_float16(rs1, &env->fp_status));
}
uint64_t helper_fcvt_h_lu(CPURISCVState *env, target_ulong rs1)
{
- return nanbox_h(uint64_to_float16(rs1, &env->fp_status));
+ return nanbox_h(env, uint64_to_float16(rs1, &env->fp_status));
}
uint64_t helper_fcvt_h_s(CPURISCVState *env, uint64_t rs1)
{
float32 frs1 = check_nanbox_s(env, rs1);
- return nanbox_h(float32_to_float16(frs1, true, &env->fp_status));
+ return nanbox_h(env, float32_to_float16(frs1, true, &env->fp_status));
}
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(env, float16_to_float32(frs1, true, &env->fp_status));
}
uint64_t helper_fcvt_h_d(CPURISCVState *env, uint64_t rs1)
{
- return nanbox_h(float64_to_float16(rs1, true, &env->fp_status));
+ return nanbox_h(env, float64_to_float16(rs1, true, &env->fp_status));
}
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 89195aad9d..26bbab2fab 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..dd7484bc90 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 +233,37 @@ 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);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
if (a->rs1 == a->rs2) { /* FMOV */
- gen_check_nanbox_h(cpu_fpr[a->rd], cpu_fpr[a->rs1]);
+ if (!ctx->ext_zfinx) {
+ gen_check_nanbox_h(dest, src1);
+ } else {
+ tcg_gen_ext16s_i64(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 src2 = get_fpr_hs(ctx, a->rs2);
+
+ if (!ctx->ext_zfinx) {
+ TCGv_i64 rs1 = tcg_temp_new_i64();
+ TCGv_i64 rs2 = tcg_temp_new_i64();
+ gen_check_nanbox_h(rs1, src1);
+ gen_check_nanbox_h(rs2, src2);
+
+ /* This formulation retains the nanboxing of rs2 in normal 'Zfh'. */
+ tcg_gen_deposit_i64(dest, rs2, rs1, 0, 15);
+
+ tcg_temp_free_i64(rs1);
+ tcg_temp_free_i64(rs2);
+ } else {
+ tcg_gen_deposit_i64(dest, src2, src1, 0, 15);
+ tcg_gen_ext16s_i64(dest, dest);
+ }
}
-
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -206,16 +273,29 @@ 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);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
rs1 = tcg_temp_new_i64();
- gen_check_nanbox_h(rs1, cpu_fpr[a->rs1]);
+ if (!ctx->ext_zfinx) {
+ gen_check_nanbox_h(rs1, src1);
+ } else {
+ tcg_gen_mov_i64(rs1, src1);
+ }
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 +304,17 @@ 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);
}
+ /* signed-extended intead of nanboxing for result if enable zfinx */
+ if (ctx->ext_zfinx) {
+ tcg_gen_ext16s_i64(dest, dest);
+ }
+ tcg_temp_free_i64(rs1);
mark_fs_dirty(ctx);
return true;
}
@@ -239,27 +324,44 @@ 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);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
rs1 = tcg_temp_new_i64();
- gen_check_nanbox_s(rs1, cpu_fpr[a->rs1]);
+ if (!ctx->ext_zfinx) {
+ gen_check_nanbox_h(rs1, src1);
+ } else {
+ tcg_gen_mov_i64(rs1, src1);
+ }
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 {
+ 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_h(rs2, src2);
+ } else {
+ tcg_gen_mov_i64(rs2, src2);
+ }
/*
* 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);
}
-
+ /* signed-extended intead of nanboxing for result if enable zfinx */
+ if (ctx->ext_zfinx) {
+ tcg_gen_ext16s_i64(dest, dest);
+ }
+ tcg_temp_free_i64(rs1);
mark_fs_dirty(ctx);
return true;
}
@@ -267,10 +369,14 @@ 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);
+
+ 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(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1],
- cpu_fpr[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 +384,14 @@ 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);
- gen_helper_fmax_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_fmax_h(dest, cpu_env, src1, src2);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -289,10 +399,14 @@ 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 +416,32 @@ 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 +450,15 @@ 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 +467,13 @@ 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 +481,13 @@ 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 +496,13 @@ 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 +510,12 @@ 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 +523,13 @@ 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 +537,13 @@ 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 +551,14 @@ 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 +567,14 @@ 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 +618,13 @@ 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 +633,13 @@ 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 +648,14 @@ 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 +665,14 @@ 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 6237bb3115..dbb322bfa7 100644
--- a/target/riscv/internals.h
+++ b/target/riscv/internals.h
@@ -72,13 +72,23 @@ static inline float32 check_nanbox_s(CPURISCVState *env, uint64_t f)
}
}
-static inline uint64_t nanbox_h(float16 f)
+static inline uint64_t nanbox_h(CPURISCVState *env, float16 f)
{
- return f | MAKE_64BIT_MASK(16, 48);
+ /* the value is sign-extended instead of NaN-boxing for zfinx */
+ if (RISCV_CPU(env_cpu(env))->cfg.ext_zfinx) {
+ return (int16_t)f;
+ } else {
+ 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] 14+ messages in thread
* [PATCH v5 5/6] target/riscv: add support for zhinx/zhinxmin
@ 2022-01-28 13:16 ` Weiwei Li
0 siblings, 0 replies; 14+ messages in thread
From: Weiwei Li @ 2022-01-28 13:16 UTC (permalink / raw)
To: richard.henderson, palmer, alistair.francis, bin.meng,
qemu-riscv, qemu-devel
Cc: wangjunqiang, lazyparser, ardxwe, Weiwei Li
- 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: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
---
target/riscv/fpu_helper.c | 89 +++---
target/riscv/helper.h | 2 +-
target/riscv/insn_trans/trans_rvzfh.c.inc | 332 +++++++++++++++-------
target/riscv/internals.h | 16 +-
4 files changed, 296 insertions(+), 143 deletions(-)
diff --git a/target/riscv/fpu_helper.c b/target/riscv/fpu_helper.c
index 63ca703459..5699c9517f 100644
--- a/target/riscv/fpu_helper.c
+++ b/target/riscv/fpu_helper.c
@@ -89,10 +89,11 @@ 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);
- return nanbox_h(float16_muladd(frs1, frs2, frs3, flags, &env->fp_status));
+ float16 frs1 = check_nanbox_h(env, rs1);
+ float16 frs2 = check_nanbox_h(env, rs2);
+ float16 frs3 = check_nanbox_h(env, rs3);
+ return nanbox_h(env, float16_muladd(frs1, frs2, frs3, flags,
+ &env->fp_status));
}
static uint64_t do_fmadd_s(CPURISCVState *env, uint64_t rs1, uint64_t rs2,
@@ -417,146 +418,146 @@ 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);
- return nanbox_h(float16_add(frs1, frs2, &env->fp_status));
+ float16 frs1 = check_nanbox_h(env, rs1);
+ float16 frs2 = check_nanbox_h(env, rs2);
+ return nanbox_h(env, 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);
- return nanbox_h(float16_sub(frs1, frs2, &env->fp_status));
+ float16 frs1 = check_nanbox_h(env, rs1);
+ float16 frs2 = check_nanbox_h(env, rs2);
+ return nanbox_h(env, 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);
- return nanbox_h(float16_mul(frs1, frs2, &env->fp_status));
+ float16 frs1 = check_nanbox_h(env, rs1);
+ float16 frs2 = check_nanbox_h(env, rs2);
+ return nanbox_h(env, 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);
- return nanbox_h(float16_div(frs1, frs2, &env->fp_status));
+ float16 frs1 = check_nanbox_h(env, rs1);
+ float16 frs2 = check_nanbox_h(env, rs2);
+ return nanbox_h(env, 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);
- return nanbox_h(env->priv_ver < PRIV_VERSION_1_11_0 ?
+ float16 frs1 = check_nanbox_h(env, rs1);
+ float16 frs2 = check_nanbox_h(env, rs2);
+ return nanbox_h(env, env->priv_ver < PRIV_VERSION_1_11_0 ?
float16_minnum(frs1, frs2, &env->fp_status) :
float16_minimum_number(frs1, frs2, &env->fp_status));
}
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);
- return nanbox_h(env->priv_ver < PRIV_VERSION_1_11_0 ?
+ float16 frs1 = check_nanbox_h(env, rs1);
+ float16 frs2 = check_nanbox_h(env, rs2);
+ return nanbox_h(env, env->priv_ver < PRIV_VERSION_1_11_0 ?
float16_maxnum(frs1, frs2, &env->fp_status) :
float16_maximum_number(frs1, frs2, &env->fp_status));
}
uint64_t helper_fsqrt_h(CPURISCVState *env, uint64_t rs1)
{
- float16 frs1 = check_nanbox_h(rs1);
- return nanbox_h(float16_sqrt(frs1, &env->fp_status));
+ float16 frs1 = check_nanbox_h(env, rs1);
+ return nanbox_h(env, 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);
}
uint64_t helper_fcvt_h_w(CPURISCVState *env, target_ulong rs1)
{
- return nanbox_h(int32_to_float16((int32_t)rs1, &env->fp_status));
+ return nanbox_h(env, int32_to_float16((int32_t)rs1, &env->fp_status));
}
uint64_t helper_fcvt_h_wu(CPURISCVState *env, target_ulong rs1)
{
- return nanbox_h(uint32_to_float16((uint32_t)rs1, &env->fp_status));
+ return nanbox_h(env, uint32_to_float16((uint32_t)rs1, &env->fp_status));
}
uint64_t helper_fcvt_h_l(CPURISCVState *env, target_ulong rs1)
{
- return nanbox_h(int64_to_float16(rs1, &env->fp_status));
+ return nanbox_h(env, int64_to_float16(rs1, &env->fp_status));
}
uint64_t helper_fcvt_h_lu(CPURISCVState *env, target_ulong rs1)
{
- return nanbox_h(uint64_to_float16(rs1, &env->fp_status));
+ return nanbox_h(env, uint64_to_float16(rs1, &env->fp_status));
}
uint64_t helper_fcvt_h_s(CPURISCVState *env, uint64_t rs1)
{
float32 frs1 = check_nanbox_s(env, rs1);
- return nanbox_h(float32_to_float16(frs1, true, &env->fp_status));
+ return nanbox_h(env, float32_to_float16(frs1, true, &env->fp_status));
}
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(env, float16_to_float32(frs1, true, &env->fp_status));
}
uint64_t helper_fcvt_h_d(CPURISCVState *env, uint64_t rs1)
{
- return nanbox_h(float64_to_float16(rs1, true, &env->fp_status));
+ return nanbox_h(env, float64_to_float16(rs1, true, &env->fp_status));
}
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 89195aad9d..26bbab2fab 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..dd7484bc90 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 +233,37 @@ 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);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
if (a->rs1 == a->rs2) { /* FMOV */
- gen_check_nanbox_h(cpu_fpr[a->rd], cpu_fpr[a->rs1]);
+ if (!ctx->ext_zfinx) {
+ gen_check_nanbox_h(dest, src1);
+ } else {
+ tcg_gen_ext16s_i64(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 src2 = get_fpr_hs(ctx, a->rs2);
+
+ if (!ctx->ext_zfinx) {
+ TCGv_i64 rs1 = tcg_temp_new_i64();
+ TCGv_i64 rs2 = tcg_temp_new_i64();
+ gen_check_nanbox_h(rs1, src1);
+ gen_check_nanbox_h(rs2, src2);
+
+ /* This formulation retains the nanboxing of rs2 in normal 'Zfh'. */
+ tcg_gen_deposit_i64(dest, rs2, rs1, 0, 15);
+
+ tcg_temp_free_i64(rs1);
+ tcg_temp_free_i64(rs2);
+ } else {
+ tcg_gen_deposit_i64(dest, src2, src1, 0, 15);
+ tcg_gen_ext16s_i64(dest, dest);
+ }
}
-
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -206,16 +273,29 @@ 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);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
rs1 = tcg_temp_new_i64();
- gen_check_nanbox_h(rs1, cpu_fpr[a->rs1]);
+ if (!ctx->ext_zfinx) {
+ gen_check_nanbox_h(rs1, src1);
+ } else {
+ tcg_gen_mov_i64(rs1, src1);
+ }
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 +304,17 @@ 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);
}
+ /* signed-extended intead of nanboxing for result if enable zfinx */
+ if (ctx->ext_zfinx) {
+ tcg_gen_ext16s_i64(dest, dest);
+ }
+ tcg_temp_free_i64(rs1);
mark_fs_dirty(ctx);
return true;
}
@@ -239,27 +324,44 @@ 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);
+
+ TCGv_i64 dest = dest_fpr(ctx, a->rd);
+ TCGv_i64 src1 = get_fpr_hs(ctx, a->rs1);
rs1 = tcg_temp_new_i64();
- gen_check_nanbox_s(rs1, cpu_fpr[a->rs1]);
+ if (!ctx->ext_zfinx) {
+ gen_check_nanbox_h(rs1, src1);
+ } else {
+ tcg_gen_mov_i64(rs1, src1);
+ }
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 {
+ 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_h(rs2, src2);
+ } else {
+ tcg_gen_mov_i64(rs2, src2);
+ }
/*
* 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);
}
-
+ /* signed-extended intead of nanboxing for result if enable zfinx */
+ if (ctx->ext_zfinx) {
+ tcg_gen_ext16s_i64(dest, dest);
+ }
+ tcg_temp_free_i64(rs1);
mark_fs_dirty(ctx);
return true;
}
@@ -267,10 +369,14 @@ 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);
+
+ 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(cpu_fpr[a->rd], cpu_env, cpu_fpr[a->rs1],
- cpu_fpr[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 +384,14 @@ 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);
- gen_helper_fmax_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_fmax_h(dest, cpu_env, src1, src2);
+ gen_set_fpr_hs(ctx, a->rd, dest);
mark_fs_dirty(ctx);
return true;
}
@@ -289,10 +399,14 @@ 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 +416,32 @@ 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 +450,15 @@ 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 +467,13 @@ 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 +481,13 @@ 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 +496,13 @@ 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 +510,12 @@ 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 +523,13 @@ 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 +537,13 @@ 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 +551,14 @@ 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 +567,14 @@ 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 +618,13 @@ 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 +633,13 @@ 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 +648,14 @@ 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 +665,14 @@ 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 6237bb3115..dbb322bfa7 100644
--- a/target/riscv/internals.h
+++ b/target/riscv/internals.h
@@ -72,13 +72,23 @@ static inline float32 check_nanbox_s(CPURISCVState *env, uint64_t f)
}
}
-static inline uint64_t nanbox_h(float16 f)
+static inline uint64_t nanbox_h(CPURISCVState *env, float16 f)
{
- return f | MAKE_64BIT_MASK(16, 48);
+ /* the value is sign-extended instead of NaN-boxing for zfinx */
+ if (RISCV_CPU(env_cpu(env))->cfg.ext_zfinx) {
+ return (int16_t)f;
+ } else {
+ 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] 14+ messages in thread
* [PATCH v5 6/6] target/riscv: expose zfinx, zdinx, zhinx{min} properties
2022-01-28 13:16 ` Weiwei Li
@ 2022-01-28 13:16 ` Weiwei Li
-1 siblings, 0 replies; 14+ messages in thread
From: Weiwei Li @ 2022-01-28 13:16 UTC (permalink / raw)
To: richard.henderson, palmer, alistair.francis, bin.meng,
qemu-riscv, qemu-devel
Cc: wangjunqiang, Weiwei Li, lazyparser, ardxwe
Co-authored-by: ardxwe <ardxwe@gmail.com>
Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
---
target/riscv/cpu.c | 5 +++++
1 file changed, 5 insertions(+)
diff --git a/target/riscv/cpu.c b/target/riscv/cpu.c
index 7006e6647b..e96d0a73f5 100644
--- a/target/riscv/cpu.c
+++ b/target/riscv/cpu.c
@@ -746,6 +746,11 @@ static Property riscv_cpu_properties[] = {
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),
+
/* These are experimental so mark with 'x-' */
DEFINE_PROP_BOOL("x-j", RISCVCPU, cfg.ext_j, false),
/* ePMP 0.9.3 */
--
2.17.1
^ permalink raw reply related [flat|nested] 14+ messages in thread
* [PATCH v5 6/6] target/riscv: expose zfinx, zdinx, zhinx{min} properties
@ 2022-01-28 13:16 ` Weiwei Li
0 siblings, 0 replies; 14+ messages in thread
From: Weiwei Li @ 2022-01-28 13:16 UTC (permalink / raw)
To: richard.henderson, palmer, alistair.francis, bin.meng,
qemu-riscv, qemu-devel
Cc: wangjunqiang, lazyparser, ardxwe, Weiwei Li
Co-authored-by: ardxwe <ardxwe@gmail.com>
Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
---
target/riscv/cpu.c | 5 +++++
1 file changed, 5 insertions(+)
diff --git a/target/riscv/cpu.c b/target/riscv/cpu.c
index 7006e6647b..e96d0a73f5 100644
--- a/target/riscv/cpu.c
+++ b/target/riscv/cpu.c
@@ -746,6 +746,11 @@ static Property riscv_cpu_properties[] = {
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),
+
/* These are experimental so mark with 'x-' */
DEFINE_PROP_BOOL("x-j", RISCVCPU, cfg.ext_j, false),
/* ePMP 0.9.3 */
--
2.17.1
^ permalink raw reply related [flat|nested] 14+ messages in thread
end of thread, other threads:[~2022-01-28 14:22 UTC | newest]
Thread overview: 14+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-01-28 13:16 [PATCH v5 0/6] support subsets of Float-Point in Integer Registers extensions Weiwei Li
2022-01-28 13:16 ` Weiwei Li
2022-01-28 13:16 ` [PATCH v5 1/6] target/riscv: add cfg properties for zfinx, zdinx and zhinx{min} Weiwei Li
2022-01-28 13:16 ` Weiwei Li
2022-01-28 13:16 ` [PATCH v5 2/6] target/riscv: hardwire mstatus.FS to zero when enable zfinx Weiwei Li
2022-01-28 13:16 ` Weiwei Li
2022-01-28 13:16 ` [PATCH v5 3/6] target/riscv: add support for zfinx Weiwei Li
2022-01-28 13:16 ` Weiwei Li
2022-01-28 13:16 ` [PATCH v5 4/6] target/riscv: add support for zdinx Weiwei Li
2022-01-28 13:16 ` Weiwei Li
2022-01-28 13:16 ` [PATCH v5 5/6] target/riscv: add support for zhinx/zhinxmin Weiwei Li
2022-01-28 13:16 ` Weiwei Li
2022-01-28 13:16 ` [PATCH v5 6/6] target/riscv: expose zfinx, zdinx, zhinx{min} properties Weiwei Li
2022-01-28 13:16 ` Weiwei Li
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.