From: Richard Henderson <richard.henderson@linaro.org>
To: qemu-devel@nongnu.org
Cc: peter.maydell@linaro.org
Subject: [PATCH v3 13/16] target/arm: Pass pointer to qc to qrdmla/qrdmls
Date: Fri, 8 May 2020 08:21:57 -0700 [thread overview]
Message-ID: <20200508152200.6547-14-richard.henderson@linaro.org> (raw)
In-Reply-To: <20200508152200.6547-1-richard.henderson@linaro.org>
Pass a pointer directly to env->vfp.qc[0], rather than env.
This will allow SVE2, which does not modify QC, to pass a
pointer to dummy storage.
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
target/arm/translate.c | 18 ++++++++---
target/arm/vec_helper.c | 70 +++++++++++++++++++++++------------------
2 files changed, 54 insertions(+), 34 deletions(-)
diff --git a/target/arm/translate.c b/target/arm/translate.c
index 8321644f25..0e2cf6028a 100644
--- a/target/arm/translate.c
+++ b/target/arm/translate.c
@@ -3629,6 +3629,18 @@ static const uint8_t neon_2rm_sizes[] = {
[NEON_2RM_VCVT_UF] = 0x4,
};
+static void gen_gvec_fn3_qc(uint32_t rd_ofs, uint32_t rn_ofs, uint32_t rm_ofs,
+ uint32_t opr_sz, uint32_t max_sz,
+ gen_helper_gvec_3_ptr *fn)
+{
+ TCGv_ptr qc_ptr = tcg_temp_new_ptr();
+
+ tcg_gen_addi_ptr(qc_ptr, cpu_env, offsetof(CPUARMState, vfp.qc));
+ tcg_gen_gvec_3_ptr(rd_ofs, rn_ofs, rm_ofs, qc_ptr,
+ opr_sz, max_sz, 0, fn);
+ tcg_temp_free_ptr(qc_ptr);
+}
+
void gen_gvec_sqrdmlah_qc(unsigned vece, uint32_t rd_ofs, uint32_t rn_ofs,
uint32_t rm_ofs, uint32_t opr_sz, uint32_t max_sz)
{
@@ -3636,8 +3648,7 @@ void gen_gvec_sqrdmlah_qc(unsigned vece, uint32_t rd_ofs, uint32_t rn_ofs,
gen_helper_gvec_qrdmlah_s16, gen_helper_gvec_qrdmlah_s32
};
tcg_debug_assert(vece >= 1 && vece <= 2);
- tcg_gen_gvec_3_ptr(rd_ofs, rn_ofs, rm_ofs, cpu_env,
- opr_sz, max_sz, 0, fns[vece - 1]);
+ gen_gvec_fn3_qc(rd_ofs, rn_ofs, rm_ofs, opr_sz, max_sz, fns[vece - 1]);
}
void gen_gvec_sqrdmlsh_qc(unsigned vece, uint32_t rd_ofs, uint32_t rn_ofs,
@@ -3647,8 +3658,7 @@ void gen_gvec_sqrdmlsh_qc(unsigned vece, uint32_t rd_ofs, uint32_t rn_ofs,
gen_helper_gvec_qrdmlsh_s16, gen_helper_gvec_qrdmlsh_s32
};
tcg_debug_assert(vece >= 1 && vece <= 2);
- tcg_gen_gvec_3_ptr(rd_ofs, rn_ofs, rm_ofs, cpu_env,
- opr_sz, max_sz, 0, fns[vece - 1]);
+ gen_gvec_fn3_qc(rd_ofs, rn_ofs, rm_ofs, opr_sz, max_sz, fns[vece - 1]);
}
#define GEN_CMP0(NAME, COND) \
diff --git a/target/arm/vec_helper.c b/target/arm/vec_helper.c
index 096fea67ef..6aa2ca0827 100644
--- a/target/arm/vec_helper.c
+++ b/target/arm/vec_helper.c
@@ -36,8 +36,6 @@
#define H4(x) (x)
#endif
-#define SET_QC() env->vfp.qc[0] = 1
-
static void clear_tail(void *vd, uintptr_t opr_sz, uintptr_t max_sz)
{
uint64_t *d = vd + opr_sz;
@@ -49,8 +47,8 @@ static void clear_tail(void *vd, uintptr_t opr_sz, uintptr_t max_sz)
}
/* Signed saturating rounding doubling multiply-accumulate high half, 16-bit */
-static uint16_t inl_qrdmlah_s16(CPUARMState *env, int16_t src1,
- int16_t src2, int16_t src3)
+static int16_t inl_qrdmlah_s16(int16_t src1, int16_t src2,
+ int16_t src3, uint32_t *sat)
{
/* Simplify:
* = ((a3 << 16) + ((e1 * e2) << 1) + (1 << 15)) >> 16
@@ -60,7 +58,7 @@ static uint16_t inl_qrdmlah_s16(CPUARMState *env, int16_t src1,
ret = ((int32_t)src3 << 15) + ret + (1 << 14);
ret >>= 15;
if (ret != (int16_t)ret) {
- SET_QC();
+ *sat = 1;
ret = (ret < 0 ? -0x8000 : 0x7fff);
}
return ret;
@@ -69,30 +67,30 @@ static uint16_t inl_qrdmlah_s16(CPUARMState *env, int16_t src1,
uint32_t HELPER(neon_qrdmlah_s16)(CPUARMState *env, uint32_t src1,
uint32_t src2, uint32_t src3)
{
- uint16_t e1 = inl_qrdmlah_s16(env, src1, src2, src3);
- uint16_t e2 = inl_qrdmlah_s16(env, src1 >> 16, src2 >> 16, src3 >> 16);
+ uint32_t *sat = &env->vfp.qc[0];
+ uint16_t e1 = inl_qrdmlah_s16(src1, src2, src3, sat);
+ uint16_t e2 = inl_qrdmlah_s16(src1 >> 16, src2 >> 16, src3 >> 16, sat);
return deposit32(e1, 16, 16, e2);
}
void HELPER(gvec_qrdmlah_s16)(void *vd, void *vn, void *vm,
- void *ve, uint32_t desc)
+ void *vq, uint32_t desc)
{
uintptr_t opr_sz = simd_oprsz(desc);
int16_t *d = vd;
int16_t *n = vn;
int16_t *m = vm;
- CPUARMState *env = ve;
uintptr_t i;
for (i = 0; i < opr_sz / 2; ++i) {
- d[i] = inl_qrdmlah_s16(env, n[i], m[i], d[i]);
+ d[i] = inl_qrdmlah_s16(n[i], m[i], d[i], vq);
}
clear_tail(d, opr_sz, simd_maxsz(desc));
}
/* Signed saturating rounding doubling multiply-subtract high half, 16-bit */
-static uint16_t inl_qrdmlsh_s16(CPUARMState *env, int16_t src1,
- int16_t src2, int16_t src3)
+static int16_t inl_qrdmlsh_s16(int16_t src1, int16_t src2,
+ int16_t src3, uint32_t *sat)
{
/* Similarly, using subtraction:
* = ((a3 << 16) - ((e1 * e2) << 1) + (1 << 15)) >> 16
@@ -102,7 +100,7 @@ static uint16_t inl_qrdmlsh_s16(CPUARMState *env, int16_t src1,
ret = ((int32_t)src3 << 15) - ret + (1 << 14);
ret >>= 15;
if (ret != (int16_t)ret) {
- SET_QC();
+ *sat = 1;
ret = (ret < 0 ? -0x8000 : 0x7fff);
}
return ret;
@@ -111,85 +109,97 @@ static uint16_t inl_qrdmlsh_s16(CPUARMState *env, int16_t src1,
uint32_t HELPER(neon_qrdmlsh_s16)(CPUARMState *env, uint32_t src1,
uint32_t src2, uint32_t src3)
{
- uint16_t e1 = inl_qrdmlsh_s16(env, src1, src2, src3);
- uint16_t e2 = inl_qrdmlsh_s16(env, src1 >> 16, src2 >> 16, src3 >> 16);
+ uint32_t *sat = &env->vfp.qc[0];
+ uint16_t e1 = inl_qrdmlsh_s16(src1, src2, src3, sat);
+ uint16_t e2 = inl_qrdmlsh_s16(src1 >> 16, src2 >> 16, src3 >> 16, sat);
return deposit32(e1, 16, 16, e2);
}
void HELPER(gvec_qrdmlsh_s16)(void *vd, void *vn, void *vm,
- void *ve, uint32_t desc)
+ void *vq, uint32_t desc)
{
uintptr_t opr_sz = simd_oprsz(desc);
int16_t *d = vd;
int16_t *n = vn;
int16_t *m = vm;
- CPUARMState *env = ve;
uintptr_t i;
for (i = 0; i < opr_sz / 2; ++i) {
- d[i] = inl_qrdmlsh_s16(env, n[i], m[i], d[i]);
+ d[i] = inl_qrdmlsh_s16(n[i], m[i], d[i], vq);
}
clear_tail(d, opr_sz, simd_maxsz(desc));
}
/* Signed saturating rounding doubling multiply-accumulate high half, 32-bit */
-uint32_t HELPER(neon_qrdmlah_s32)(CPUARMState *env, int32_t src1,
- int32_t src2, int32_t src3)
+static int32_t inl_qrdmlah_s32(int32_t src1, int32_t src2,
+ int32_t src3, uint32_t *sat)
{
/* Simplify similarly to int_qrdmlah_s16 above. */
int64_t ret = (int64_t)src1 * src2;
ret = ((int64_t)src3 << 31) + ret + (1 << 30);
ret >>= 31;
if (ret != (int32_t)ret) {
- SET_QC();
+ *sat = 1;
ret = (ret < 0 ? INT32_MIN : INT32_MAX);
}
return ret;
}
+uint32_t HELPER(neon_qrdmlah_s32)(CPUARMState *env, int32_t src1,
+ int32_t src2, int32_t src3)
+{
+ uint32_t *sat = &env->vfp.qc[0];
+ return inl_qrdmlah_s32(src1, src2, src3, sat);
+}
+
void HELPER(gvec_qrdmlah_s32)(void *vd, void *vn, void *vm,
- void *ve, uint32_t desc)
+ void *vq, uint32_t desc)
{
uintptr_t opr_sz = simd_oprsz(desc);
int32_t *d = vd;
int32_t *n = vn;
int32_t *m = vm;
- CPUARMState *env = ve;
uintptr_t i;
for (i = 0; i < opr_sz / 4; ++i) {
- d[i] = helper_neon_qrdmlah_s32(env, n[i], m[i], d[i]);
+ d[i] = inl_qrdmlah_s32(n[i], m[i], d[i], vq);
}
clear_tail(d, opr_sz, simd_maxsz(desc));
}
/* Signed saturating rounding doubling multiply-subtract high half, 32-bit */
-uint32_t HELPER(neon_qrdmlsh_s32)(CPUARMState *env, int32_t src1,
- int32_t src2, int32_t src3)
+static int32_t inl_qrdmlsh_s32(int32_t src1, int32_t src2,
+ int32_t src3, uint32_t *sat)
{
/* Simplify similarly to int_qrdmlsh_s16 above. */
int64_t ret = (int64_t)src1 * src2;
ret = ((int64_t)src3 << 31) - ret + (1 << 30);
ret >>= 31;
if (ret != (int32_t)ret) {
- SET_QC();
+ *sat = 1;
ret = (ret < 0 ? INT32_MIN : INT32_MAX);
}
return ret;
}
+uint32_t HELPER(neon_qrdmlsh_s32)(CPUARMState *env, int32_t src1,
+ int32_t src2, int32_t src3)
+{
+ uint32_t *sat = &env->vfp.qc[0];
+ return inl_qrdmlsh_s32(src1, src2, src3, sat);
+}
+
void HELPER(gvec_qrdmlsh_s32)(void *vd, void *vn, void *vm,
- void *ve, uint32_t desc)
+ void *vq, uint32_t desc)
{
uintptr_t opr_sz = simd_oprsz(desc);
int32_t *d = vd;
int32_t *n = vn;
int32_t *m = vm;
- CPUARMState *env = ve;
uintptr_t i;
for (i = 0; i < opr_sz / 4; ++i) {
- d[i] = helper_neon_qrdmlsh_s32(env, n[i], m[i], d[i]);
+ d[i] = inl_qrdmlsh_s32(n[i], m[i], d[i], vq);
}
clear_tail(d, opr_sz, simd_maxsz(desc));
}
--
2.20.1
next prev parent reply other threads:[~2020-05-08 15:44 UTC|newest]
Thread overview: 38+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-05-08 15:21 [PATCH v3 00/16] target/arm: partial vector cleanup Richard Henderson
2020-05-08 15:21 ` [PATCH v3 01/16] target/arm: Create gen_gvec_[us]sra Richard Henderson
2020-05-12 13:20 ` Peter Maydell
2020-05-08 15:21 ` [PATCH v3 02/16] target/arm: Create gen_gvec_{u,s}{rshr,rsra} Richard Henderson
2020-05-12 13:09 ` Peter Maydell
2020-05-12 13:46 ` Peter Maydell
2020-05-13 2:04 ` Richard Henderson
2020-05-12 13:51 ` Peter Maydell
2020-05-13 2:06 ` Richard Henderson
2020-05-08 15:21 ` [PATCH v3 03/16] target/arm: Create gen_gvec_{sri,sli} Richard Henderson
2020-05-12 13:52 ` Peter Maydell
2020-05-08 15:21 ` [PATCH v3 04/16] target/arm: Remove unnecessary range check for VSHL Richard Henderson
2020-05-12 13:53 ` Peter Maydell
2020-05-08 15:21 ` [PATCH v3 05/16] target/arm: Tidy handle_vec_simd_shri Richard Henderson
2020-05-12 13:56 ` Peter Maydell
2020-05-08 15:21 ` [PATCH v3 06/16] target/arm: Create gen_gvec_{ceq,clt,cle,cgt,cge}0 Richard Henderson
2020-05-12 14:10 ` [PATCH v3 06/16] target/arm: Create gen_gvec_{ceq, clt, cle, cgt, cge}0 Peter Maydell
2020-05-08 15:21 ` [PATCH v3 07/16] target/arm: Create gen_gvec_{mla,mls} Richard Henderson
2020-05-12 14:11 ` Peter Maydell
2020-05-08 15:21 ` [PATCH v3 08/16] target/arm: Swap argument order for VSHL during decode Richard Henderson
2020-05-12 14:14 ` Peter Maydell
2020-05-08 15:21 ` [PATCH v3 09/16] target/arm: Create gen_gvec_{cmtst,ushl,sshl} Richard Henderson
2020-05-12 14:16 ` Peter Maydell
2020-05-08 15:21 ` [PATCH v3 10/16] target/arm: Create gen_gvec_{uqadd, sqadd, uqsub, sqsub} Richard Henderson
2020-05-12 14:18 ` Peter Maydell
2020-05-08 15:21 ` [PATCH v3 11/16] target/arm: Remove fp_status from helper_{recpe, rsqrte}_u32 Richard Henderson
2020-05-12 14:19 ` Peter Maydell
2020-05-08 15:21 ` [PATCH v3 12/16] target/arm: Create gen_gvec_{qrdmla,qrdmls} Richard Henderson
2020-05-12 14:20 ` Peter Maydell
2020-05-08 15:21 ` Richard Henderson [this message]
2020-05-12 14:28 ` [PATCH v3 13/16] target/arm: Pass pointer to qc to qrdmla/qrdmls Peter Maydell
2020-05-08 15:21 ` [PATCH v3 14/16] target/arm: Clear tail in gvec_fmul_idx_*, gvec_fmla_idx_* Richard Henderson
2020-05-12 14:29 ` Peter Maydell
2020-05-08 15:21 ` [PATCH v3 15/16] target/arm: Vectorize SABD/UABD Richard Henderson
2020-05-12 14:40 ` Peter Maydell
2020-05-08 15:22 ` [PATCH v3 16/16] target/arm: Vectorize SABA/UABA Richard Henderson
2020-05-12 14:41 ` Peter Maydell
2020-05-12 12:55 ` [PATCH v3 00/16] target/arm: partial vector cleanup Peter Maydell
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20200508152200.6547-14-richard.henderson@linaro.org \
--to=richard.henderson@linaro.org \
--cc=peter.maydell@linaro.org \
--cc=qemu-devel@nongnu.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).