[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [PATCH v2 16/54] tcg: Introduce tcg_out_movext
|
From: |
Philippe Mathieu-Daudé |
|
Subject: |
Re: [PATCH v2 16/54] tcg: Introduce tcg_out_movext |
|
Date: |
Sat, 22 Apr 2023 01:02:46 +0200 |
|
User-agent: |
Mozilla/5.0 (Macintosh; Intel Mac OS X 10.15; rv:102.0) Gecko/20100101 Thunderbird/102.10.0 |
On 11/4/23 03:04, Richard Henderson wrote:
This is common code in most qemu_{ld,st} slow paths, extending the
input value for the store helper data argument or extending the
return value from the load helper.
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
tcg/tcg.c | 63 ++++++++++++++++++++++++++++++++
tcg/aarch64/tcg-target.c.inc | 8 +---
tcg/arm/tcg-target.c.inc | 16 ++------
tcg/i386/tcg-target.c.inc | 30 +++------------
tcg/loongarch64/tcg-target.c.inc | 53 ++++-----------------------
tcg/ppc/tcg-target.c.inc | 38 +++++--------------
tcg/riscv/tcg-target.c.inc | 13 +------
tcg/s390x/tcg-target.c.inc | 19 ++--------
tcg/sparc64/tcg-target.c.inc | 32 ++++------------
9 files changed, 104 insertions(+), 168 deletions(-)
diff --git a/tcg/arm/tcg-target.c.inc b/tcg/arm/tcg-target.c.inc
index 1820655ee3..f865294861 100644
--- a/tcg/arm/tcg-target.c.inc
+++ b/tcg/arm/tcg-target.c.inc
@@ -1567,17 +1567,7 @@ static bool tcg_out_qemu_ld_slow_path(TCGContext *s,
TCGLabelQemuLdst *lb)
datalo = lb->datalo_reg;
datahi = lb->datahi_reg;
- switch (opc & MO_SSIZE) {
- case MO_SB:
- tcg_out_ext8s(s, TCG_TYPE_I32, datalo, TCG_REG_R0);
- break;
- case MO_SW:
- tcg_out_ext16s(s, TCG_TYPE_I32, datalo, TCG_REG_R0);
- break;
- default:
- tcg_out_mov_reg(s, COND_AL, datalo, TCG_REG_R0);
- break;
- case MO_UQ:
+ if ((opc & MO_SIZE) == MO_64) {
if (datalo != TCG_REG_R1) {
tcg_out_mov_reg(s, COND_AL, datalo, TCG_REG_R0);
tcg_out_mov_reg(s, COND_AL, datahi, TCG_REG_R1);
@@ -1589,7 +1579,9 @@ static bool tcg_out_qemu_ld_slow_path(TCGContext *s,
TCGLabelQemuLdst *lb)
tcg_out_mov_reg(s, COND_AL, datahi, TCG_REG_R1);
tcg_out_mov_reg(s, COND_AL, datalo, TCG_REG_TMP);
}
- break;
+ } else {
+ tcg_out_movext(s, TCG_TYPE_I32, lb->datalo_reg,
Why not use 'datalo' like in i386?
+ TCG_TYPE_I32, opc & MO_SSIZE, TCG_REG_R0);
}
tcg_out_goto(s, COND_AL, lb->raddr);
diff --git a/tcg/i386/tcg-target.c.inc b/tcg/i386/tcg-target.c.inc
index a166a195c4..4847da7e1a 100644
--- a/tcg/i386/tcg-target.c.inc
+++ b/tcg/i386/tcg-target.c.inc
@@ -1946,28 +1946,8 @@ static bool tcg_out_qemu_ld_slow_path(TCGContext *s,
TCGLabelQemuLdst *l)
tcg_out_branch(s, 1, qemu_ld_helpers[opc & (MO_BSWAP | MO_SIZE)]);
data_reg = l->datalo_reg;
- switch (opc & MO_SSIZE) {
- case MO_SB:
- tcg_out_ext8s(s, l->type, data_reg, TCG_REG_EAX);
- break;
- case MO_SW:
- tcg_out_ext16s(s, l->type, data_reg, TCG_REG_EAX);
- break;
-#if TCG_TARGET_REG_BITS == 64
- case MO_SL:
- tcg_out_ext32s(s, data_reg, TCG_REG_EAX);
- break;
-#endif
- case MO_UB:
- case MO_UW:
- /* Note that the helpers have zero-extended to tcg_target_long. */
- case MO_UL:
- tcg_out_mov(s, TCG_TYPE_I32, data_reg, TCG_REG_EAX);
- break;
- case MO_UQ:
- if (TCG_TARGET_REG_BITS == 64) {
- tcg_out_mov(s, TCG_TYPE_I64, data_reg, TCG_REG_RAX);
- } else if (data_reg == TCG_REG_EDX) {
+ if (TCG_TARGET_REG_BITS == 32 && (opc & MO_SIZE) == MO_64) {
+ if (data_reg == TCG_REG_EDX) {
/* xchg %edx, %eax */
tcg_out_opc(s, OPC_XCHG_ax_r32 + TCG_REG_EDX, 0, 0, 0);
tcg_out_mov(s, TCG_TYPE_I32, l->datahi_reg, TCG_REG_EAX);
@@ -1975,9 +1955,9 @@ static bool tcg_out_qemu_ld_slow_path(TCGContext *s,
TCGLabelQemuLdst *l)
tcg_out_mov(s, TCG_TYPE_I32, data_reg, TCG_REG_EAX);
tcg_out_mov(s, TCG_TYPE_I32, l->datahi_reg, TCG_REG_EDX);
}
- break;
- default:
- g_assert_not_reached();
+ } else {
+ tcg_out_movext(s, l->type, data_reg,
+ TCG_TYPE_REG, opc & MO_SSIZE, TCG_REG_EAX);
}
/* Jump to the code corresponding to next IR of qemu_st */
[I'm skipping the ppc change hopping Daniel can review it]
diff --git a/tcg/ppc/tcg-target.c.inc b/tcg/ppc/tcg-target.c.inc
index 4c4178b700..b1d9c0bbe4 100644
--- a/tcg/ppc/tcg-target.c.inc
+++ b/tcg/ppc/tcg-target.c.inc
@@ -1971,10 +1971,6 @@ static const uint32_t qemu_stx_opc[(MO_SIZE + MO_BSWAP)
+ 1] = {
[MO_BSWAP | MO_UQ] = STDBRX,
};
-static const uint32_t qemu_exts_opc[4] = {
- EXTSB, EXTSH, EXTSW, 0
-};
-
#if defined (CONFIG_SOFTMMU)
/* helper signature: helper_ld_mmu(CPUState *env, target_ulong addr,
* int mmu_idx, uintptr_t ra)
@@ -2168,11 +2164,9 @@ static bool tcg_out_qemu_ld_slow_path(TCGContext *s,
TCGLabelQemuLdst *lb)
if (TCG_TARGET_REG_BITS == 32 && (opc & MO_SIZE) == MO_64) {
tcg_out_mov(s, TCG_TYPE_I32, lo, TCG_REG_R4);
tcg_out_mov(s, TCG_TYPE_I32, hi, TCG_REG_R3);
- } else if (opc & MO_SIGN) {
- uint32_t insn = qemu_exts_opc[opc & MO_SIZE];
- tcg_out32(s, insn | RA(lo) | RS(TCG_REG_R3));
} else {
- tcg_out_mov(s, TCG_TYPE_REG, lo, TCG_REG_R3);
+ tcg_out_movext(s, lb->type, lo,
+ TCG_TYPE_REG, opc & MO_SSIZE, TCG_REG_R3);
}
tcg_out_b(s, 0, lb->raddr);
@@ -2206,25 +2200,13 @@ static bool tcg_out_qemu_st_slow_path(TCGContext *s,
TCGLabelQemuLdst *lb)
lo = lb->datalo_reg;
hi = lb->datahi_reg;
- if (TCG_TARGET_REG_BITS == 32) {
- switch (s_bits) {
- case MO_64:
- arg |= (TCG_TARGET_CALL_ARG_I64 == TCG_CALL_ARG_EVEN);
- tcg_out_mov(s, TCG_TYPE_I32, arg++, hi);
- /* FALLTHRU */
- case MO_32:
- tcg_out_mov(s, TCG_TYPE_I32, arg++, lo);
- break;
- default:
- tcg_out_rlw(s, RLWINM, arg++, lo, 0, 32 - (8 << s_bits), 31);
- break;
- }
+ if (TCG_TARGET_REG_BITS == 32 && s_bits == MO_64) {
+ arg |= (TCG_TARGET_CALL_ARG_I64 == TCG_CALL_ARG_EVEN);
+ tcg_out_mov(s, TCG_TYPE_I32, arg++, hi);
+ tcg_out_mov(s, TCG_TYPE_I32, arg++, lo);
} else {
- if (s_bits == MO_64) {
- tcg_out_mov(s, TCG_TYPE_I64, arg++, lo);
- } else {
- tcg_out_rld(s, RLDICL, arg++, lo, 0, 64 - (8 << s_bits));
- }
+ tcg_out_movext(s, s_bits == MO_64 ? TCG_TYPE_I64 : TCG_TYPE_I32,
+ arg++, lb->type, s_bits, lo);
}
tcg_out_movi(s, TCG_TYPE_I32, arg++, oi);
@@ -2371,8 +2353,8 @@ static void tcg_out_qemu_ld(TCGContext *s, const TCGArg
*args, bool is_64)
} else {
insn = qemu_ldx_opc[opc & (MO_SIZE | MO_BSWAP)];
tcg_out32(s, insn | TAB(datalo, rbase, addrlo));
- insn = qemu_exts_opc[s_bits];
- tcg_out32(s, insn | RA(datalo) | RS(datalo));
+ tcg_out_movext(s, TCG_TYPE_REG, datalo,
+ TCG_TYPE_REG, opc & MO_SSIZE, datalo);
}
}
diff --git a/tcg/sparc64/tcg-target.c.inc b/tcg/sparc64/tcg-target.c.inc
index 18ddd6bb9f..99ba0fdc2b 100644
--- a/tcg/sparc64/tcg-target.c.inc
+++ b/tcg/sparc64/tcg-target.c.inc
@@ -917,26 +917,6 @@ static void tcg_out_mb(TCGContext *s, TCGArg a0)
static const tcg_insn_unit *qemu_ld_trampoline[(MO_SSIZE | MO_BSWAP) + 1];
static const tcg_insn_unit *qemu_st_trampoline[(MO_SIZE | MO_BSWAP) + 1];
-static void emit_extend(TCGContext *s, TCGReg r, int op)
-{
- /* Emit zero extend of 8, 16 or 32 bit data as
- * required by the MO_* value op; do nothing for 64 bit.
- */
- switch (op & MO_SIZE) {
- case MO_8:
- tcg_out_ext8u(s, r, r);
- break;
- case MO_16:
- tcg_out_ext16u(s, r, r);
- break;
- case MO_32:
- tcg_out_ext32u(s, r, r);
- break;
- case MO_64:
- break;
- }
-}
-
static void build_trampolines(TCGContext *s)
{
static void * const qemu_ld_helpers[] = {
@@ -993,8 +973,6 @@ static void build_trampolines(TCGContext *s)
}
qemu_st_trampoline[i] = tcg_splitwx_to_rx(s->code_ptr);
- emit_extend(s, TCG_REG_O2, i);
-
/* Set the retaddr operand. */
tcg_out_mov(s, TCG_TYPE_PTR, TCG_REG_O4, TCG_REG_O7);
@@ -1341,7 +1319,7 @@ static void tcg_out_qemu_ld(TCGContext *s, TCGReg data, TCGReg addr,
}
static void tcg_out_qemu_st(TCGContext *s, TCGReg data, TCGReg addr,
- MemOpIdx oi)
+ MemOpIdx oi, bool is64)
Why not directly pass 'TCGType data_type' instead of is64?
Otherwise (except ppc),
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
{
MemOp memop = get_memop(oi);
tcg_insn_unit *label_ptr;
@@ -1367,7 +1345,9 @@ static void tcg_out_qemu_st(TCGContext *s, TCGReg data,
TCGReg addr,
/* TLB Miss. */
tcg_out_mov(s, TCG_TYPE_REG, TCG_REG_O1, addrz);
- tcg_out_mov(s, TCG_TYPE_REG, TCG_REG_O2, data);
+ tcg_out_movext(s, (memop & MO_SIZE) == MO_64 ? TCG_TYPE_I64 : TCG_TYPE_I32,
+ TCG_REG_O2, is64 ? TCG_TYPE_I64 : TCG_TYPE_I32,
+ memop & MO_SIZE, data);
func = qemu_st_trampoline[memop & (MO_BSWAP | MO_SIZE)];
tcg_debug_assert(func != NULL);
@@ -1658,8 +1638,10 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc,
tcg_out_qemu_ld(s, a0, a1, a2, true);
break;
case INDEX_op_qemu_st_i32:
+ tcg_out_qemu_st(s, a0, a1, a2, false);
+ break;
case INDEX_op_qemu_st_i64:
- tcg_out_qemu_st(s, a0, a1, a2);
+ tcg_out_qemu_st(s, a0, a1, a2, true);
break;
case INDEX_op_ld32s_i64:
- [PATCH v2 09/54] tcg: Split out tcg_out_exts_i32_i64, (continued)
- [PATCH v2 09/54] tcg: Split out tcg_out_exts_i32_i64, Richard Henderson, 2023/04/10
- [PATCH v2 10/54] tcg/loongarch64: Conditionalize tcg_out_exts_i32_i64, Richard Henderson, 2023/04/10
- [PATCH v2 12/54] tcg/riscv: Conditionalize tcg_out_exts_i32_i64, Richard Henderson, 2023/04/10
- [PATCH v2 11/54] tcg/mips: Conditionalize tcg_out_exts_i32_i64, Richard Henderson, 2023/04/10
- [PATCH v2 14/54] tcg/i386: Conditionalize tcg_out_extu_i32_i64, Richard Henderson, 2023/04/10
- [PATCH v2 13/54] tcg: Split out tcg_out_extu_i32_i64, Richard Henderson, 2023/04/10
- [PATCH v2 16/54] tcg: Introduce tcg_out_movext, Richard Henderson, 2023/04/10
- Re: [PATCH v2 16/54] tcg: Introduce tcg_out_movext,
Philippe Mathieu-Daudé <=
- [PATCH v2 18/54] tcg: Introduce tcg_out_movext2, Richard Henderson, 2023/04/10
- [PATCH v2 15/54] tcg: Split out tcg_out_extrl_i64_i32, Richard Henderson, 2023/04/10
- [PATCH v2 22/54] tcg/arm: Rationalize args to tcg_out_qemu_{ld,st}, Richard Henderson, 2023/04/10
- [PATCH v2 24/54] tcg/loongarch64: Rationalize args to tcg_out_qemu_{ld, st}, Richard Henderson, 2023/04/10
- [PATCH v2 25/54] tcg/ppc: Rationalize args to tcg_out_qemu_{ld,st}, Richard Henderson, 2023/04/10
- [PATCH v2 26/54] tcg/s390x: Pass TCGType to tcg_out_qemu_{ld,st}, Richard Henderson, 2023/04/10