[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[PULL 08/27] tcg/aarch64: Simplify constraints on qemu_ld/st
From: |
Richard Henderson |
Subject: |
[PULL 08/27] tcg/aarch64: Simplify constraints on qemu_ld/st |
Date: |
Tue, 30 May 2023 11:59:30 -0700 |
Adjust the softmmu tlb to use TMP[0-2], not any of the normally available
registers. Since we handle overlap betwen inputs and helper arguments,
we can allow any allocatable reg.
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
tcg/aarch64/tcg-target-con-set.h | 2 --
tcg/aarch64/tcg-target-con-str.h | 1 -
tcg/aarch64/tcg-target.c.inc | 45 ++++++++++++++------------------
3 files changed, 19 insertions(+), 29 deletions(-)
diff --git a/tcg/aarch64/tcg-target-con-set.h b/tcg/aarch64/tcg-target-con-set.h
index d6c6866878..109f4ab97c 100644
--- a/tcg/aarch64/tcg-target-con-set.h
+++ b/tcg/aarch64/tcg-target-con-set.h
@@ -10,11 +10,9 @@
* tcg-target-con-str.h; the constraint combination is inclusive or.
*/
C_O0_I1(r)
-C_O0_I2(lZ, l)
C_O0_I2(r, rA)
C_O0_I2(rZ, r)
C_O0_I2(w, r)
-C_O1_I1(r, l)
C_O1_I1(r, r)
C_O1_I1(w, r)
C_O1_I1(w, w)
diff --git a/tcg/aarch64/tcg-target-con-str.h b/tcg/aarch64/tcg-target-con-str.h
index 00adb64594..fb1a845b4f 100644
--- a/tcg/aarch64/tcg-target-con-str.h
+++ b/tcg/aarch64/tcg-target-con-str.h
@@ -9,7 +9,6 @@
* REGS(letter, register_mask)
*/
REGS('r', ALL_GENERAL_REGS)
-REGS('l', ALL_QLDST_REGS)
REGS('w', ALL_VECTOR_REGS)
/*
diff --git a/tcg/aarch64/tcg-target.c.inc b/tcg/aarch64/tcg-target.c.inc
index 5e7ac6fb76..00d9033e2f 100644
--- a/tcg/aarch64/tcg-target.c.inc
+++ b/tcg/aarch64/tcg-target.c.inc
@@ -132,14 +132,6 @@ static bool patch_reloc(tcg_insn_unit *code_ptr, int type,
#define ALL_GENERAL_REGS 0xffffffffu
#define ALL_VECTOR_REGS 0xffffffff00000000ull
-#ifdef CONFIG_SOFTMMU
-#define ALL_QLDST_REGS \
- (ALL_GENERAL_REGS & ~((1 << TCG_REG_X0) | (1 << TCG_REG_X1) | \
- (1 << TCG_REG_X2) | (1 << TCG_REG_X3)))
-#else
-#define ALL_QLDST_REGS ALL_GENERAL_REGS
-#endif
-
/* Match a constant valid for addition (12-bit, optionally shifted). */
static inline bool is_aimm(uint64_t val)
{
@@ -1648,7 +1640,7 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s,
HostAddress *h,
unsigned s_bits = opc & MO_SIZE;
unsigned s_mask = (1u << s_bits) - 1;
unsigned mem_index = get_mmuidx(oi);
- TCGReg x3;
+ TCGReg addr_adj;
TCGType mask_type;
uint64_t compare_mask;
@@ -1660,27 +1652,27 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext
*s, HostAddress *h,
mask_type = (s->page_bits + s->tlb_dyn_max_bits > 32
? TCG_TYPE_I64 : TCG_TYPE_I32);
- /* Load env_tlb(env)->f[mmu_idx].{mask,table} into {x0,x1}. */
+ /* Load env_tlb(env)->f[mmu_idx].{mask,table} into {tmp0,tmp1}. */
QEMU_BUILD_BUG_ON(TLB_MASK_TABLE_OFS(0) > 0);
QEMU_BUILD_BUG_ON(TLB_MASK_TABLE_OFS(0) < -512);
QEMU_BUILD_BUG_ON(offsetof(CPUTLBDescFast, mask) != 0);
QEMU_BUILD_BUG_ON(offsetof(CPUTLBDescFast, table) != 8);
- tcg_out_insn(s, 3314, LDP, TCG_REG_X0, TCG_REG_X1, TCG_AREG0,
+ tcg_out_insn(s, 3314, LDP, TCG_REG_TMP0, TCG_REG_TMP1, TCG_AREG0,
TLB_MASK_TABLE_OFS(mem_index), 1, 0);
/* Extract the TLB index from the address into X0. */
tcg_out_insn(s, 3502S, AND_LSR, mask_type == TCG_TYPE_I64,
- TCG_REG_X0, TCG_REG_X0, addr_reg,
+ TCG_REG_TMP0, TCG_REG_TMP0, addr_reg,
s->page_bits - CPU_TLB_ENTRY_BITS);
- /* Add the tlb_table pointer, creating the CPUTLBEntry address into X1. */
- tcg_out_insn(s, 3502, ADD, 1, TCG_REG_X1, TCG_REG_X1, TCG_REG_X0);
+ /* Add the tlb_table pointer, forming the CPUTLBEntry address in TMP1. */
+ tcg_out_insn(s, 3502, ADD, 1, TCG_REG_TMP1, TCG_REG_TMP1, TCG_REG_TMP0);
- /* Load the tlb comparator into X0, and the fast path addend into X1. */
- tcg_out_ld(s, addr_type, TCG_REG_X0, TCG_REG_X1,
+ /* Load the tlb comparator into TMP0, and the fast path addend into TMP1.
*/
+ tcg_out_ld(s, addr_type, TCG_REG_TMP0, TCG_REG_TMP1,
is_ld ? offsetof(CPUTLBEntry, addr_read)
: offsetof(CPUTLBEntry, addr_write));
- tcg_out_ld(s, TCG_TYPE_PTR, TCG_REG_X1, TCG_REG_X1,
+ tcg_out_ld(s, TCG_TYPE_PTR, TCG_REG_TMP1, TCG_REG_TMP1,
offsetof(CPUTLBEntry, addend));
/*
@@ -1689,25 +1681,26 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext
*s, HostAddress *h,
* cross pages using the address of the last byte of the access.
*/
if (a_mask >= s_mask) {
- x3 = addr_reg;
+ addr_adj = addr_reg;
} else {
+ addr_adj = TCG_REG_TMP2;
tcg_out_insn(s, 3401, ADDI, addr_type,
- TCG_REG_X3, addr_reg, s_mask - a_mask);
- x3 = TCG_REG_X3;
+ addr_adj, addr_reg, s_mask - a_mask);
}
compare_mask = (uint64_t)s->page_mask | a_mask;
- /* Store the page mask part of the address into X3. */
- tcg_out_logicali(s, I3404_ANDI, addr_type, TCG_REG_X3, x3, compare_mask);
+ /* Store the page mask part of the address into TMP2. */
+ tcg_out_logicali(s, I3404_ANDI, addr_type, TCG_REG_TMP2,
+ addr_adj, compare_mask);
/* Perform the address comparison. */
- tcg_out_cmp(s, addr_type, TCG_REG_X0, TCG_REG_X3, 0);
+ tcg_out_cmp(s, addr_type, TCG_REG_TMP0, TCG_REG_TMP2, 0);
/* If not equal, we jump to the slow path. */
ldst->label_ptr[0] = s->code_ptr;
tcg_out_insn(s, 3202, B_C, TCG_COND_NE, 0);
- h->base = TCG_REG_X1,
+ h->base = TCG_REG_TMP1;
h->index = addr_reg;
h->index_ext = addr_type;
#else
@@ -2802,12 +2795,12 @@ static TCGConstraintSetIndex
tcg_target_op_def(TCGOpcode op)
case INDEX_op_qemu_ld_a64_i32:
case INDEX_op_qemu_ld_a32_i64:
case INDEX_op_qemu_ld_a64_i64:
- return C_O1_I1(r, l);
+ return C_O1_I1(r, r);
case INDEX_op_qemu_st_a32_i32:
case INDEX_op_qemu_st_a64_i32:
case INDEX_op_qemu_st_a32_i64:
case INDEX_op_qemu_st_a64_i64:
- return C_O0_I2(lZ, l);
+ return C_O0_I2(rZ, r);
case INDEX_op_deposit_i32:
case INDEX_op_deposit_i64:
--
2.34.1
- [PULL 13/27] accel/tcg: Extract store_atom_insert_al16 to host header, (continued)
- [PULL 13/27] accel/tcg: Extract store_atom_insert_al16 to host header, Richard Henderson, 2023/05/30
- [PULL 18/27] decodetree: Add --test-for-error, Richard Henderson, 2023/05/30
- [PULL 15/27] accel/tcg: Add aarch64 lse2 load_atom_extract_al16_or_al8, Richard Henderson, 2023/05/30
- [PULL 19/27] decodetree: Fix recursion in prop_format and build_tree, Richard Henderson, 2023/05/30
- [PULL 27/27] tests/decode: Add tests for various named-field cases, Richard Henderson, 2023/05/30
- [PULL 26/27] scripts/decodetree: Implement named field support, Richard Henderson, 2023/05/30
- [PULL 05/27] tcg/i386: Support 128-bit load/store, Richard Henderson, 2023/05/30
- [PULL 09/27] tcg/aarch64: Support 128-bit load/store, Richard Henderson, 2023/05/30
- [PULL 11/27] tcg/s390x: Support 128-bit load/store, Richard Henderson, 2023/05/30
- [PULL 10/27] tcg/ppc: Support 128-bit load/store, Richard Henderson, 2023/05/30
- [PULL 08/27] tcg/aarch64: Simplify constraints on qemu_ld/st,
Richard Henderson <=
- [PULL 07/27] tcg/aarch64: Reserve TCG_REG_TMP1, TCG_REG_TMP2, Richard Henderson, 2023/05/30
- [PULL 16/27] accel/tcg: Add aarch64 store_atom_insert_al16, Richard Henderson, 2023/05/30
- [PULL 12/27] accel/tcg: Extract load_atom_extract_al16_or_al8 to host header, Richard Henderson, 2023/05/30
- [PULL 20/27] decodetree: Diagnose empty pattern group, Richard Henderson, 2023/05/30
- [PULL 23/27] docs: Document decodetree named field syntax, Richard Henderson, 2023/05/30
- [PULL 25/27] scripts/decodetree: Implement a topological sort, Richard Henderson, 2023/05/30
- Re: [PULL 00/27] tcg patch queue, Richard Henderson, 2023/05/30