From 3a3b282879e83efdee1cb752e75351725e07e90a Mon Sep 17 00:00:00 2001 From: Richard Henderson Date: Sat, 15 Feb 2025 11:45:21 -0800 Subject: [PATCH 01/28] tcg: Remove last traces of TCG_TARGET_NEED_POOL_LABELS MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit These should have been removed with the rest. There are a couple of hosts which can emit guest_base into the constant pool: aarch64, mips64, ppc64, riscv64. Fixes: a417ef835058 ("tcg: Remove TCG_TARGET_NEED_LDST_LABELS and TCG_TARGET_NEED_POOL_LABELS") Signed-off-by: Richard Henderson Reviewed-by: Philippe Mathieu-Daudé --- tcg/tcg.c | 4 ---- tcg/tci/tcg-target.h | 1 - 2 files changed, 5 deletions(-) diff --git a/tcg/tcg.c b/tcg/tcg.c index 43b6712286..53de13df71 100644 --- a/tcg/tcg.c +++ b/tcg/tcg.c @@ -1598,21 +1598,17 @@ void tcg_prologue_init(void) tcg_qemu_tb_exec = (tcg_prologue_fn *)tcg_splitwx_to_rx(s->code_ptr); #endif -#ifdef TCG_TARGET_NEED_POOL_LABELS s->pool_labels = NULL; -#endif qemu_thread_jit_write(); /* Generate the prologue. */ tcg_target_qemu_prologue(s); -#ifdef TCG_TARGET_NEED_POOL_LABELS /* Allow the prologue to put e.g. guest_base into a pool entry. */ { int result = tcg_out_pool_finalize(s); tcg_debug_assert(result == 0); } -#endif prologue_size = tcg_current_code_size(s); perf_report_prologue(s->code_gen_ptr, prologue_size); diff --git a/tcg/tci/tcg-target.h b/tcg/tci/tcg-target.h index a9ca493d20..bd03aa1bc4 100644 --- a/tcg/tci/tcg-target.h +++ b/tcg/tci/tcg-target.h @@ -72,6 +72,5 @@ typedef enum { } TCGReg; #define HAVE_TCG_QEMU_TB_EXEC -#define TCG_TARGET_NEED_POOL_LABELS #endif /* TCG_TARGET_H */ From f441b4d19b289f55a378b8d033994f45a333b581 Mon Sep 17 00:00:00 2001 From: Richard Henderson Date: Sun, 2 Feb 2025 18:03:51 -0800 Subject: [PATCH 02/28] tcg: Remove TCG_OVERSIZED_GUEST MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit This is now prohibited in configuration. Reviewed-by: Alex Bennée Signed-off-by: Richard Henderson --- accel/tcg/cputlb.c | 7 ------- accel/tcg/tcg-all.c | 9 ++++----- docs/devel/multi-thread-tcg.rst | 1 - include/qemu/atomic.h | 18 +++-------------- include/tcg/oversized-guest.h | 23 ---------------------- target/arm/ptw.c | 34 --------------------------------- target/riscv/cpu_helper.c | 13 +------------ 7 files changed, 8 insertions(+), 97 deletions(-) delete mode 100644 include/tcg/oversized-guest.h diff --git a/accel/tcg/cputlb.c b/accel/tcg/cputlb.c index b4ccf0cdcb..17e2251695 100644 --- a/accel/tcg/cputlb.c +++ b/accel/tcg/cputlb.c @@ -47,7 +47,6 @@ #include "qemu/plugin-memory.h" #endif #include "tcg/tcg-ldst.h" -#include "tcg/oversized-guest.h" /* DEBUG defines, enable DEBUG_TLB_LOG to log to the CPU_LOG_MMU target */ /* #define DEBUG_TLB */ @@ -118,12 +117,8 @@ static inline uint64_t tlb_read_idx(const CPUTLBEntry *entry, return qatomic_read(ptr); #else const uint64_t *ptr = &entry->addr_idx[access_type]; -# if TCG_OVERSIZED_GUEST - return *ptr; -# else /* ofs might correspond to .addr_write, so use qatomic_read */ return qatomic_read(ptr); -# endif #endif } @@ -908,8 +903,6 @@ static void tlb_reset_dirty_range_locked(CPUTLBEntry *tlb_entry, uint32_t *ptr_write = (uint32_t *)&tlb_entry->addr_write; ptr_write += HOST_BIG_ENDIAN; qatomic_set(ptr_write, *ptr_write | TLB_NOTDIRTY); -#elif TCG_OVERSIZED_GUEST - tlb_entry->addr_write |= TLB_NOTDIRTY; #else qatomic_set(&tlb_entry->addr_write, tlb_entry->addr_write | TLB_NOTDIRTY); diff --git a/accel/tcg/tcg-all.c b/accel/tcg/tcg-all.c index 95adaacee8..c1a30b0121 100644 --- a/accel/tcg/tcg-all.c +++ b/accel/tcg/tcg-all.c @@ -28,7 +28,6 @@ #include "exec/replay-core.h" #include "system/cpu-timers.h" #include "tcg/startup.h" -#include "tcg/oversized-guest.h" #include "qapi/error.h" #include "qemu/error-report.h" #include "qemu/accel.h" @@ -41,6 +40,8 @@ #include "hw/boards.h" #endif #include "internal-common.h" +#include "cpu-param.h" + struct TCGState { AccelState parent_obj; @@ -72,7 +73,7 @@ DECLARE_INSTANCE_CHECKER(TCGState, TCG_STATE, static bool default_mttcg_enabled(void) { - if (icount_enabled() || TCG_OVERSIZED_GUEST) { + if (icount_enabled()) { return false; } #ifdef TARGET_SUPPORTS_MTTCG @@ -145,9 +146,7 @@ static void tcg_set_thread(Object *obj, const char *value, Error **errp) TCGState *s = TCG_STATE(obj); if (strcmp(value, "multi") == 0) { - if (TCG_OVERSIZED_GUEST) { - error_setg(errp, "No MTTCG when guest word size > hosts"); - } else if (icount_enabled()) { + if (icount_enabled()) { error_setg(errp, "No MTTCG when icount is enabled"); } else { #ifndef TARGET_SUPPORTS_MTTCG diff --git a/docs/devel/multi-thread-tcg.rst b/docs/devel/multi-thread-tcg.rst index 7fd0a07633..b0f473961d 100644 --- a/docs/devel/multi-thread-tcg.rst +++ b/docs/devel/multi-thread-tcg.rst @@ -37,7 +37,6 @@ if: * forced by --accel tcg,thread=single * enabling --icount mode -* 64 bit guests on 32 bit hosts (TCG_OVERSIZED_GUEST) In the general case of running translated code there should be no inter-vCPU dependencies and all vCPUs should be able to run at full diff --git a/include/qemu/atomic.h b/include/qemu/atomic.h index 7a3f2e6576..f80cba24cf 100644 --- a/include/qemu/atomic.h +++ b/include/qemu/atomic.h @@ -56,25 +56,13 @@ */ #define signal_barrier() __atomic_signal_fence(__ATOMIC_SEQ_CST) -/* Sanity check that the size of an atomic operation isn't "overly large". +/* + * Sanity check that the size of an atomic operation isn't "overly large". * Despite the fact that e.g. i686 has 64-bit atomic operations, we do not * want to use them because we ought not need them, and this lets us do a * bit of sanity checking that other 32-bit hosts might build. - * - * That said, we have a problem on 64-bit ILP32 hosts in that in order to - * sync with TCG_OVERSIZED_GUEST, this must match TCG_TARGET_REG_BITS. - * We'd prefer not want to pull in everything else TCG related, so handle - * those few cases by hand. - * - * Note that x32 is fully detected with __x86_64__ + _ILP32, and that for - * Sparc we always force the use of sparcv9 in configure. MIPS n32 (ILP32) & - * n64 (LP64) ABIs are both detected using __mips64. */ -#if defined(__x86_64__) || defined(__sparc__) || defined(__mips64) -# define ATOMIC_REG_SIZE 8 -#else -# define ATOMIC_REG_SIZE sizeof(void *) -#endif +#define ATOMIC_REG_SIZE sizeof(void *) /* Weak atomic operations prevent the compiler moving other * loads/stores past the atomic operation load/store. However there is diff --git a/include/tcg/oversized-guest.h b/include/tcg/oversized-guest.h deleted file mode 100644 index 641b9749ff..0000000000 --- a/include/tcg/oversized-guest.h +++ /dev/null @@ -1,23 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -/* - * Define TCG_OVERSIZED_GUEST - * Copyright (c) 2008 Fabrice Bellard - */ - -#ifndef EXEC_TCG_OVERSIZED_GUEST_H -#define EXEC_TCG_OVERSIZED_GUEST_H - -#include "tcg-target-reg-bits.h" -#include "cpu-param.h" - -/* - * Oversized TCG guests make things like MTTCG hard - * as we can't use atomics for cputlb updates. - */ -#if TARGET_LONG_BITS > TCG_TARGET_REG_BITS -#define TCG_OVERSIZED_GUEST 1 -#else -#define TCG_OVERSIZED_GUEST 0 -#endif - -#endif diff --git a/target/arm/ptw.c b/target/arm/ptw.c index 64bb6878a4..4330900348 100644 --- a/target/arm/ptw.c +++ b/target/arm/ptw.c @@ -16,9 +16,6 @@ #include "internals.h" #include "cpu-features.h" #include "idau.h" -#ifdef CONFIG_TCG -# include "tcg/oversized-guest.h" -#endif typedef struct S1Translate { /* @@ -840,7 +837,6 @@ static uint64_t arm_casq_ptw(CPUARMState *env, uint64_t old_val, ptw->out_rw = true; } -#ifdef CONFIG_ATOMIC64 if (ptw->out_be) { old_val = cpu_to_be64(old_val); new_val = cpu_to_be64(new_val); @@ -852,36 +848,6 @@ static uint64_t arm_casq_ptw(CPUARMState *env, uint64_t old_val, cur_val = qatomic_cmpxchg__nocheck((uint64_t *)host, old_val, new_val); cur_val = le64_to_cpu(cur_val); } -#else - /* - * We can't support the full 64-bit atomic cmpxchg on the host. - * Because this is only used for FEAT_HAFDBS, which is only for AA64, - * we know that TCG_OVERSIZED_GUEST is set, which means that we are - * running in round-robin mode and could only race with dma i/o. - */ -#if !TCG_OVERSIZED_GUEST -# error "Unexpected configuration" -#endif - bool locked = bql_locked(); - if (!locked) { - bql_lock(); - } - if (ptw->out_be) { - cur_val = ldq_be_p(host); - if (cur_val == old_val) { - stq_be_p(host, new_val); - } - } else { - cur_val = ldq_le_p(host); - if (cur_val == old_val) { - stq_le_p(host, new_val); - } - } - if (!locked) { - bql_unlock(); - } -#endif - return cur_val; #else /* AArch32 does not have FEAT_HADFS; non-TCG guests only use debug-mode. */ diff --git a/target/riscv/cpu_helper.c b/target/riscv/cpu_helper.c index e1dfc4ecbf..8ff6d900f2 100644 --- a/target/riscv/cpu_helper.c +++ b/target/riscv/cpu_helper.c @@ -32,7 +32,6 @@ #include "system/cpu-timers.h" #include "cpu_bits.h" #include "debug.h" -#include "tcg/oversized-guest.h" #include "pmp.h" int riscv_env_mmu_index(CPURISCVState *env, bool ifetch) @@ -1167,9 +1166,7 @@ static int get_physical_address(CPURISCVState *env, hwaddr *physical, hwaddr pte_addr; int i; -#if !TCG_OVERSIZED_GUEST -restart: -#endif + restart: for (i = 0; i < levels; i++, ptshift -= ptidxbits) { target_ulong idx; if (i == 0) { @@ -1388,13 +1385,6 @@ restart: false, MEMTXATTRS_UNSPECIFIED); if (memory_region_is_ram(mr)) { target_ulong *pte_pa = qemu_map_ram_ptr(mr->ram_block, addr1); -#if TCG_OVERSIZED_GUEST - /* - * MTTCG is not enabled on oversized TCG guests so - * page table updates do not need to be atomic - */ - *pte_pa = pte = updated_pte; -#else target_ulong old_pte; if (riscv_cpu_sxl(env) == MXL_RV32) { old_pte = qatomic_cmpxchg((uint32_t *)pte_pa, pte, updated_pte); @@ -1405,7 +1395,6 @@ restart: goto restart; } pte = updated_pte; -#endif } else { /* * Misconfigured PTE in ROM (AD bits are not preset) or From a0ecb8e49418ab0bb01d47493fafe6a0a357e952 Mon Sep 17 00:00:00 2001 From: Richard Henderson Date: Tue, 4 Feb 2025 11:06:26 -0800 Subject: [PATCH 03/28] tcg: Drop support for two address registers in gen_ldst MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Reviewed-by: Philippe Mathieu-Daudé Signed-off-by: Richard Henderson --- tcg/tcg-op-ldst.c | 21 +++------------------ tcg/tcg.c | 4 +--- 2 files changed, 4 insertions(+), 21 deletions(-) diff --git a/tcg/tcg-op-ldst.c b/tcg/tcg-op-ldst.c index 77271e0193..7ba9a3ef7e 100644 --- a/tcg/tcg-op-ldst.c +++ b/tcg/tcg-op-ldst.c @@ -91,25 +91,10 @@ static MemOp tcg_canonicalize_memop(MemOp op, bool is64, bool st) static void gen_ldst(TCGOpcode opc, TCGType type, TCGTemp *vl, TCGTemp *vh, TCGTemp *addr, MemOpIdx oi) { - if (TCG_TARGET_REG_BITS == 64 || tcg_ctx->addr_type == TCG_TYPE_I32) { - if (vh) { - tcg_gen_op4(opc, type, temp_arg(vl), temp_arg(vh), - temp_arg(addr), oi); - } else { - tcg_gen_op3(opc, type, temp_arg(vl), temp_arg(addr), oi); - } + if (vh) { + tcg_gen_op4(opc, type, temp_arg(vl), temp_arg(vh), temp_arg(addr), oi); } else { - /* See TCGV_LOW/HIGH. */ - TCGTemp *al = addr + HOST_BIG_ENDIAN; - TCGTemp *ah = addr + !HOST_BIG_ENDIAN; - - if (vh) { - tcg_gen_op5(opc, type, temp_arg(vl), temp_arg(vh), - temp_arg(al), temp_arg(ah), oi); - } else { - tcg_gen_op4(opc, type, temp_arg(vl), - temp_arg(al), temp_arg(ah), oi); - } + tcg_gen_op3(opc, type, temp_arg(vl), temp_arg(addr), oi); } } diff --git a/tcg/tcg.c b/tcg/tcg.c index 53de13df71..14c2d38160 100644 --- a/tcg/tcg.c +++ b/tcg/tcg.c @@ -1690,9 +1690,7 @@ void tcg_func_start(TCGContext *s) s->emit_before_op = NULL; QSIMPLEQ_INIT(&s->labels); - tcg_debug_assert(s->addr_type == TCG_TYPE_I32 || - s->addr_type == TCG_TYPE_I64); - + tcg_debug_assert(s->addr_type <= TCG_TYPE_REG); tcg_debug_assert(s->insn_start_words > 0); } From 50b7a197e1d1782f9366d5e43d1f94700f6236c8 Mon Sep 17 00:00:00 2001 From: Richard Henderson Date: Tue, 4 Feb 2025 13:46:09 -0800 Subject: [PATCH 04/28] tcg: Merge INDEX_op_qemu_*_{a32,a64}_* MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Since 64-on-32 is now unsupported, guest addresses always fit in one host register. Drop the replication of opcodes. Reviewed-by: Philippe Mathieu-Daudé Signed-off-by: Richard Henderson --- include/tcg/tcg-opc.h | 28 ++------ tcg/aarch64/tcg-target.c.inc | 36 ++++------ tcg/arm/tcg-target.c.inc | 40 +++-------- tcg/i386/tcg-target.c.inc | 69 ++++-------------- tcg/loongarch64/tcg-target.c.inc | 36 ++++------ tcg/mips/tcg-target.c.inc | 51 +++---------- tcg/optimize.c | 21 ++---- tcg/ppc/tcg-target.c.inc | 68 ++++-------------- tcg/riscv/tcg-target.c.inc | 24 +++---- tcg/s390x/tcg-target.c.inc | 36 ++++------ tcg/sparc64/tcg-target.c.inc | 24 +++---- tcg/tcg-op-ldst.c | 82 +++++---------------- tcg/tcg.c | 42 ++++------- tcg/tci.c | 119 ++++++------------------------- tcg/tci/tcg-target.c.inc | 60 ++++------------ 15 files changed, 177 insertions(+), 559 deletions(-) diff --git a/include/tcg/tcg-opc.h b/include/tcg/tcg-opc.h index 9383e295f4..5bf78b0764 100644 --- a/include/tcg/tcg-opc.h +++ b/include/tcg/tcg-opc.h @@ -188,36 +188,22 @@ DEF(goto_ptr, 0, 1, 0, TCG_OPF_BB_EXIT | TCG_OPF_BB_END) DEF(plugin_cb, 0, 0, 1, TCG_OPF_NOT_PRESENT) DEF(plugin_mem_cb, 0, 1, 1, TCG_OPF_NOT_PRESENT) -/* Replicate ld/st ops for 32 and 64-bit guest addresses. */ -DEF(qemu_ld_a32_i32, 1, 1, 1, +DEF(qemu_ld_i32, 1, 1, 1, TCG_OPF_CALL_CLOBBER | TCG_OPF_SIDE_EFFECTS) -DEF(qemu_st_a32_i32, 0, 1 + 1, 1, +DEF(qemu_st_i32, 0, 1 + 1, 1, TCG_OPF_CALL_CLOBBER | TCG_OPF_SIDE_EFFECTS) -DEF(qemu_ld_a32_i64, DATA64_ARGS, 1, 1, +DEF(qemu_ld_i64, DATA64_ARGS, 1, 1, TCG_OPF_CALL_CLOBBER | TCG_OPF_SIDE_EFFECTS) -DEF(qemu_st_a32_i64, 0, DATA64_ARGS + 1, 1, - TCG_OPF_CALL_CLOBBER | TCG_OPF_SIDE_EFFECTS) - -DEF(qemu_ld_a64_i32, 1, DATA64_ARGS, 1, - TCG_OPF_CALL_CLOBBER | TCG_OPF_SIDE_EFFECTS) -DEF(qemu_st_a64_i32, 0, 1 + DATA64_ARGS, 1, - TCG_OPF_CALL_CLOBBER | TCG_OPF_SIDE_EFFECTS) -DEF(qemu_ld_a64_i64, DATA64_ARGS, DATA64_ARGS, 1, - TCG_OPF_CALL_CLOBBER | TCG_OPF_SIDE_EFFECTS) -DEF(qemu_st_a64_i64, 0, DATA64_ARGS + DATA64_ARGS, 1, +DEF(qemu_st_i64, 0, DATA64_ARGS + 1, 1, TCG_OPF_CALL_CLOBBER | TCG_OPF_SIDE_EFFECTS) /* Only used by i386 to cope with stupid register constraints. */ -DEF(qemu_st8_a32_i32, 0, 1 + 1, 1, - TCG_OPF_CALL_CLOBBER | TCG_OPF_SIDE_EFFECTS) -DEF(qemu_st8_a64_i32, 0, 1 + DATA64_ARGS, 1, +DEF(qemu_st8_i32, 0, 1 + 1, 1, TCG_OPF_CALL_CLOBBER | TCG_OPF_SIDE_EFFECTS) /* Only for 64-bit hosts at the moment. */ -DEF(qemu_ld_a32_i128, 2, 1, 1, TCG_OPF_CALL_CLOBBER | TCG_OPF_SIDE_EFFECTS) -DEF(qemu_ld_a64_i128, 2, 1, 1, TCG_OPF_CALL_CLOBBER | TCG_OPF_SIDE_EFFECTS) -DEF(qemu_st_a32_i128, 0, 3, 1, TCG_OPF_CALL_CLOBBER | TCG_OPF_SIDE_EFFECTS) -DEF(qemu_st_a64_i128, 0, 3, 1, TCG_OPF_CALL_CLOBBER | TCG_OPF_SIDE_EFFECTS) +DEF(qemu_ld_i128, 2, 1, 1, TCG_OPF_CALL_CLOBBER | TCG_OPF_SIDE_EFFECTS) +DEF(qemu_st_i128, 0, 3, 1, TCG_OPF_CALL_CLOBBER | TCG_OPF_SIDE_EFFECTS) /* Host vector support. */ diff --git a/tcg/aarch64/tcg-target.c.inc b/tcg/aarch64/tcg-target.c.inc index 66eb4b73b5..45dc2c649b 100644 --- a/tcg/aarch64/tcg-target.c.inc +++ b/tcg/aarch64/tcg-target.c.inc @@ -2398,24 +2398,18 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, TCGType ext, tcg_out_insn(s, 3506, CSEL, ext, a0, REG0(3), REG0(4), args[5]); break; - case INDEX_op_qemu_ld_a32_i32: - case INDEX_op_qemu_ld_a64_i32: - case INDEX_op_qemu_ld_a32_i64: - case INDEX_op_qemu_ld_a64_i64: + case INDEX_op_qemu_ld_i32: + case INDEX_op_qemu_ld_i64: tcg_out_qemu_ld(s, a0, a1, a2, ext); break; - 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: + case INDEX_op_qemu_st_i32: + case INDEX_op_qemu_st_i64: tcg_out_qemu_st(s, REG0(0), a1, a2, ext); break; - case INDEX_op_qemu_ld_a32_i128: - case INDEX_op_qemu_ld_a64_i128: + case INDEX_op_qemu_ld_i128: tcg_out_qemu_ldst_i128(s, a0, a1, a2, args[3], true); break; - case INDEX_op_qemu_st_a32_i128: - case INDEX_op_qemu_st_a64_i128: + case INDEX_op_qemu_st_i128: tcg_out_qemu_ldst_i128(s, REG0(0), REG0(1), a2, args[3], false); break; @@ -3084,21 +3078,15 @@ tcg_target_op_def(TCGOpcode op, TCGType type, unsigned flags) case INDEX_op_movcond_i64: return C_O1_I4(r, r, rC, rZ, rZ); - case INDEX_op_qemu_ld_a32_i32: - case INDEX_op_qemu_ld_a64_i32: - case INDEX_op_qemu_ld_a32_i64: - case INDEX_op_qemu_ld_a64_i64: + case INDEX_op_qemu_ld_i32: + case INDEX_op_qemu_ld_i64: return C_O1_I1(r, r); - case INDEX_op_qemu_ld_a32_i128: - case INDEX_op_qemu_ld_a64_i128: + case INDEX_op_qemu_ld_i128: return C_O2_I1(r, 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: + case INDEX_op_qemu_st_i32: + case INDEX_op_qemu_st_i64: return C_O0_I2(rZ, r); - case INDEX_op_qemu_st_a32_i128: - case INDEX_op_qemu_st_a64_i128: + case INDEX_op_qemu_st_i128: return C_O0_I3(rZ, rZ, r); case INDEX_op_deposit_i32: diff --git a/tcg/arm/tcg-target.c.inc b/tcg/arm/tcg-target.c.inc index 12dad7307f..05bb367a39 100644 --- a/tcg/arm/tcg-target.c.inc +++ b/tcg/arm/tcg-target.c.inc @@ -2071,37 +2071,21 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, TCGType type, ARITH_MOV, args[0], 0, 0); break; - case INDEX_op_qemu_ld_a32_i32: + case INDEX_op_qemu_ld_i32: tcg_out_qemu_ld(s, args[0], -1, args[1], -1, args[2], TCG_TYPE_I32); break; - case INDEX_op_qemu_ld_a64_i32: - tcg_out_qemu_ld(s, args[0], -1, args[1], args[2], - args[3], TCG_TYPE_I32); - break; - case INDEX_op_qemu_ld_a32_i64: + case INDEX_op_qemu_ld_i64: tcg_out_qemu_ld(s, args[0], args[1], args[2], -1, args[3], TCG_TYPE_I64); break; - case INDEX_op_qemu_ld_a64_i64: - tcg_out_qemu_ld(s, args[0], args[1], args[2], args[3], - args[4], TCG_TYPE_I64); - break; - case INDEX_op_qemu_st_a32_i32: + case INDEX_op_qemu_st_i32: tcg_out_qemu_st(s, args[0], -1, args[1], -1, args[2], TCG_TYPE_I32); break; - case INDEX_op_qemu_st_a64_i32: - tcg_out_qemu_st(s, args[0], -1, args[1], args[2], - args[3], TCG_TYPE_I32); - break; - case INDEX_op_qemu_st_a32_i64: + case INDEX_op_qemu_st_i64: tcg_out_qemu_st(s, args[0], args[1], args[2], -1, args[3], TCG_TYPE_I64); break; - case INDEX_op_qemu_st_a64_i64: - tcg_out_qemu_st(s, args[0], args[1], args[2], args[3], - args[4], TCG_TYPE_I64); - break; case INDEX_op_bswap16_i32: tcg_out_bswap16(s, COND_AL, args[0], args[1], args[2]); @@ -2243,22 +2227,14 @@ tcg_target_op_def(TCGOpcode op, TCGType type, unsigned flags) case INDEX_op_setcond2_i32: return C_O1_I4(r, r, r, rI, rI); - case INDEX_op_qemu_ld_a32_i32: + case INDEX_op_qemu_ld_i32: return C_O1_I1(r, q); - case INDEX_op_qemu_ld_a64_i32: - return C_O1_I2(r, q, q); - case INDEX_op_qemu_ld_a32_i64: + case INDEX_op_qemu_ld_i64: return C_O2_I1(e, p, q); - case INDEX_op_qemu_ld_a64_i64: - return C_O2_I2(e, p, q, q); - case INDEX_op_qemu_st_a32_i32: + case INDEX_op_qemu_st_i32: return C_O0_I2(q, q); - case INDEX_op_qemu_st_a64_i32: - return C_O0_I3(q, q, q); - case INDEX_op_qemu_st_a32_i64: + case INDEX_op_qemu_st_i64: return C_O0_I3(Q, p, q); - case INDEX_op_qemu_st_a64_i64: - return C_O0_I4(Q, p, q, q); case INDEX_op_st_vec: return C_O0_I2(w, r); diff --git a/tcg/i386/tcg-target.c.inc b/tcg/i386/tcg-target.c.inc index 2cac151331..ca6e8abc57 100644 --- a/tcg/i386/tcg-target.c.inc +++ b/tcg/i386/tcg-target.c.inc @@ -2879,62 +2879,33 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, TCGType type, tcg_out_modrm(s, OPC_GRP3_Ev + rexw, EXT3_NOT, a0); break; - case INDEX_op_qemu_ld_a64_i32: - if (TCG_TARGET_REG_BITS == 32) { - tcg_out_qemu_ld(s, a0, -1, a1, a2, args[3], TCG_TYPE_I32); - break; - } - /* fall through */ - case INDEX_op_qemu_ld_a32_i32: + case INDEX_op_qemu_ld_i32: tcg_out_qemu_ld(s, a0, -1, a1, -1, a2, TCG_TYPE_I32); break; - case INDEX_op_qemu_ld_a32_i64: + case INDEX_op_qemu_ld_i64: if (TCG_TARGET_REG_BITS == 64) { tcg_out_qemu_ld(s, a0, -1, a1, -1, a2, TCG_TYPE_I64); } else { tcg_out_qemu_ld(s, a0, a1, a2, -1, args[3], TCG_TYPE_I64); } break; - case INDEX_op_qemu_ld_a64_i64: - if (TCG_TARGET_REG_BITS == 64) { - tcg_out_qemu_ld(s, a0, -1, a1, -1, a2, TCG_TYPE_I64); - } else { - tcg_out_qemu_ld(s, a0, a1, a2, args[3], args[4], TCG_TYPE_I64); - } - break; - case INDEX_op_qemu_ld_a32_i128: - case INDEX_op_qemu_ld_a64_i128: + case INDEX_op_qemu_ld_i128: tcg_debug_assert(TCG_TARGET_REG_BITS == 64); tcg_out_qemu_ld(s, a0, a1, a2, -1, args[3], TCG_TYPE_I128); break; - case INDEX_op_qemu_st_a64_i32: - case INDEX_op_qemu_st8_a64_i32: - if (TCG_TARGET_REG_BITS == 32) { - tcg_out_qemu_st(s, a0, -1, a1, a2, args[3], TCG_TYPE_I32); - break; - } - /* fall through */ - case INDEX_op_qemu_st_a32_i32: - case INDEX_op_qemu_st8_a32_i32: + case INDEX_op_qemu_st_i32: + case INDEX_op_qemu_st8_i32: tcg_out_qemu_st(s, a0, -1, a1, -1, a2, TCG_TYPE_I32); break; - case INDEX_op_qemu_st_a32_i64: + case INDEX_op_qemu_st_i64: if (TCG_TARGET_REG_BITS == 64) { tcg_out_qemu_st(s, a0, -1, a1, -1, a2, TCG_TYPE_I64); } else { tcg_out_qemu_st(s, a0, a1, a2, -1, args[3], TCG_TYPE_I64); } break; - case INDEX_op_qemu_st_a64_i64: - if (TCG_TARGET_REG_BITS == 64) { - tcg_out_qemu_st(s, a0, -1, a1, -1, a2, TCG_TYPE_I64); - } else { - tcg_out_qemu_st(s, a0, a1, a2, args[3], args[4], TCG_TYPE_I64); - } - break; - case INDEX_op_qemu_st_a32_i128: - case INDEX_op_qemu_st_a64_i128: + case INDEX_op_qemu_st_i128: tcg_debug_assert(TCG_TARGET_REG_BITS == 64); tcg_out_qemu_st(s, a0, a1, a2, -1, args[3], TCG_TYPE_I128); break; @@ -3824,36 +3795,24 @@ tcg_target_op_def(TCGOpcode op, TCGType type, unsigned flags) case INDEX_op_clz_i64: return have_lzcnt ? C_N1_I2(r, r, rW) : C_N1_I2(r, r, r); - case INDEX_op_qemu_ld_a32_i32: + case INDEX_op_qemu_ld_i32: return C_O1_I1(r, L); - case INDEX_op_qemu_ld_a64_i32: - return TCG_TARGET_REG_BITS == 64 ? C_O1_I1(r, L) : C_O1_I2(r, L, L); - case INDEX_op_qemu_st_a32_i32: + case INDEX_op_qemu_st_i32: return C_O0_I2(L, L); - case INDEX_op_qemu_st_a64_i32: - return TCG_TARGET_REG_BITS == 64 ? C_O0_I2(L, L) : C_O0_I3(L, L, L); - case INDEX_op_qemu_st8_a32_i32: + case INDEX_op_qemu_st8_i32: return C_O0_I2(s, L); - case INDEX_op_qemu_st8_a64_i32: - return TCG_TARGET_REG_BITS == 64 ? C_O0_I2(s, L) : C_O0_I3(s, L, L); - case INDEX_op_qemu_ld_a32_i64: + case INDEX_op_qemu_ld_i64: return TCG_TARGET_REG_BITS == 64 ? C_O1_I1(r, L) : C_O2_I1(r, r, L); - case INDEX_op_qemu_ld_a64_i64: - return TCG_TARGET_REG_BITS == 64 ? C_O1_I1(r, L) : C_O2_I2(r, r, L, L); - case INDEX_op_qemu_st_a32_i64: + case INDEX_op_qemu_st_i64: return TCG_TARGET_REG_BITS == 64 ? C_O0_I2(L, L) : C_O0_I3(L, L, L); - case INDEX_op_qemu_st_a64_i64: - return TCG_TARGET_REG_BITS == 64 ? C_O0_I2(L, L) : C_O0_I4(L, L, L, L); - case INDEX_op_qemu_ld_a32_i128: - case INDEX_op_qemu_ld_a64_i128: + case INDEX_op_qemu_ld_i128: tcg_debug_assert(TCG_TARGET_REG_BITS == 64); return C_O2_I1(r, r, L); - case INDEX_op_qemu_st_a32_i128: - case INDEX_op_qemu_st_a64_i128: + case INDEX_op_qemu_st_i128: tcg_debug_assert(TCG_TARGET_REG_BITS == 64); return C_O0_I3(L, L, L); diff --git a/tcg/loongarch64/tcg-target.c.inc b/tcg/loongarch64/tcg-target.c.inc index cebe8dd354..4f32bf3e97 100644 --- a/tcg/loongarch64/tcg-target.c.inc +++ b/tcg/loongarch64/tcg-target.c.inc @@ -1675,28 +1675,22 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, TCGType type, tcg_out_ldst(s, OPC_ST_D, a0, a1, a2); break; - case INDEX_op_qemu_ld_a32_i32: - case INDEX_op_qemu_ld_a64_i32: + case INDEX_op_qemu_ld_i32: tcg_out_qemu_ld(s, a0, a1, a2, TCG_TYPE_I32); break; - case INDEX_op_qemu_ld_a32_i64: - case INDEX_op_qemu_ld_a64_i64: + case INDEX_op_qemu_ld_i64: tcg_out_qemu_ld(s, a0, a1, a2, TCG_TYPE_I64); break; - case INDEX_op_qemu_ld_a32_i128: - case INDEX_op_qemu_ld_a64_i128: + case INDEX_op_qemu_ld_i128: tcg_out_qemu_ldst_i128(s, a0, a1, a2, a3, true); break; - case INDEX_op_qemu_st_a32_i32: - case INDEX_op_qemu_st_a64_i32: + case INDEX_op_qemu_st_i32: tcg_out_qemu_st(s, a0, a1, a2, TCG_TYPE_I32); break; - case INDEX_op_qemu_st_a32_i64: - case INDEX_op_qemu_st_a64_i64: + case INDEX_op_qemu_st_i64: tcg_out_qemu_st(s, a0, a1, a2, TCG_TYPE_I64); break; - case INDEX_op_qemu_st_a32_i128: - case INDEX_op_qemu_st_a64_i128: + case INDEX_op_qemu_st_i128: tcg_out_qemu_ldst_i128(s, a0, a1, a2, a3, false); break; @@ -2233,18 +2227,14 @@ tcg_target_op_def(TCGOpcode op, TCGType type, unsigned flags) case INDEX_op_st32_i64: case INDEX_op_st_i32: case INDEX_op_st_i64: - 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: + case INDEX_op_qemu_st_i32: + case INDEX_op_qemu_st_i64: return C_O0_I2(rZ, r); - case INDEX_op_qemu_ld_a32_i128: - case INDEX_op_qemu_ld_a64_i128: + case INDEX_op_qemu_ld_i128: return C_N2_I1(r, r, r); - case INDEX_op_qemu_st_a32_i128: - case INDEX_op_qemu_st_a64_i128: + case INDEX_op_qemu_st_i128: return C_O0_I3(r, r, r); case INDEX_op_brcond_i32: @@ -2290,10 +2280,8 @@ tcg_target_op_def(TCGOpcode op, TCGType type, unsigned flags) case INDEX_op_ld32u_i64: case INDEX_op_ld_i32: case INDEX_op_ld_i64: - case INDEX_op_qemu_ld_a32_i32: - case INDEX_op_qemu_ld_a64_i32: - case INDEX_op_qemu_ld_a32_i64: - case INDEX_op_qemu_ld_a64_i64: + case INDEX_op_qemu_ld_i32: + case INDEX_op_qemu_ld_i64: return C_O1_I1(r, r); case INDEX_op_andc_i32: diff --git a/tcg/mips/tcg-target.c.inc b/tcg/mips/tcg-target.c.inc index 99f6ef6c76..b1d512ca2a 100644 --- a/tcg/mips/tcg-target.c.inc +++ b/tcg/mips/tcg-target.c.inc @@ -2095,53 +2095,27 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, TCGType type, tcg_out_setcond2(s, args[5], a0, a1, a2, args[3], args[4]); break; - case INDEX_op_qemu_ld_a64_i32: - if (TCG_TARGET_REG_BITS == 32) { - tcg_out_qemu_ld(s, a0, 0, a1, a2, args[3], TCG_TYPE_I32); - break; - } - /* fall through */ - case INDEX_op_qemu_ld_a32_i32: + case INDEX_op_qemu_ld_i32: tcg_out_qemu_ld(s, a0, 0, a1, 0, a2, TCG_TYPE_I32); break; - case INDEX_op_qemu_ld_a32_i64: + case INDEX_op_qemu_ld_i64: if (TCG_TARGET_REG_BITS == 64) { tcg_out_qemu_ld(s, a0, 0, a1, 0, a2, TCG_TYPE_I64); } else { tcg_out_qemu_ld(s, a0, a1, a2, 0, args[3], TCG_TYPE_I64); } break; - case INDEX_op_qemu_ld_a64_i64: - if (TCG_TARGET_REG_BITS == 64) { - tcg_out_qemu_ld(s, a0, 0, a1, 0, a2, TCG_TYPE_I64); - } else { - tcg_out_qemu_ld(s, a0, a1, a2, args[3], args[4], TCG_TYPE_I64); - } - break; - case INDEX_op_qemu_st_a64_i32: - if (TCG_TARGET_REG_BITS == 32) { - tcg_out_qemu_st(s, a0, 0, a1, a2, args[3], TCG_TYPE_I32); - break; - } - /* fall through */ - case INDEX_op_qemu_st_a32_i32: + case INDEX_op_qemu_st_i32: tcg_out_qemu_st(s, a0, 0, a1, 0, a2, TCG_TYPE_I32); break; - case INDEX_op_qemu_st_a32_i64: + case INDEX_op_qemu_st_i64: if (TCG_TARGET_REG_BITS == 64) { tcg_out_qemu_st(s, a0, 0, a1, 0, a2, TCG_TYPE_I64); } else { tcg_out_qemu_st(s, a0, a1, a2, 0, args[3], TCG_TYPE_I64); } break; - case INDEX_op_qemu_st_a64_i64: - if (TCG_TARGET_REG_BITS == 64) { - tcg_out_qemu_st(s, a0, 0, a1, 0, a2, TCG_TYPE_I64); - } else { - tcg_out_qemu_st(s, a0, a1, a2, args[3], args[4], TCG_TYPE_I64); - } - break; case INDEX_op_add2_i32: tcg_out_addsub2(s, a0, a1, a2, args[3], args[4], args[5], @@ -2301,23 +2275,14 @@ tcg_target_op_def(TCGOpcode op, TCGType type, unsigned flags) case INDEX_op_brcond2_i32: return C_O0_I4(rZ, rZ, rZ, rZ); - case INDEX_op_qemu_ld_a32_i32: + case INDEX_op_qemu_ld_i32: return C_O1_I1(r, r); - case INDEX_op_qemu_ld_a64_i32: - return TCG_TARGET_REG_BITS == 64 ? C_O1_I1(r, r) : C_O1_I2(r, r, r); - case INDEX_op_qemu_st_a32_i32: + case INDEX_op_qemu_st_i32: return C_O0_I2(rZ, r); - case INDEX_op_qemu_st_a64_i32: - return TCG_TARGET_REG_BITS == 64 ? C_O0_I2(rZ, r) : C_O0_I3(rZ, r, r); - case INDEX_op_qemu_ld_a32_i64: + case INDEX_op_qemu_ld_i64: return TCG_TARGET_REG_BITS == 64 ? C_O1_I1(r, r) : C_O2_I1(r, r, r); - case INDEX_op_qemu_ld_a64_i64: - return TCG_TARGET_REG_BITS == 64 ? C_O1_I1(r, r) : C_O2_I2(r, r, r, r); - case INDEX_op_qemu_st_a32_i64: + case INDEX_op_qemu_st_i64: return TCG_TARGET_REG_BITS == 64 ? C_O0_I2(rZ, r) : C_O0_I3(rZ, rZ, r); - case INDEX_op_qemu_st_a64_i64: - return (TCG_TARGET_REG_BITS == 64 ? C_O0_I2(rZ, r) - : C_O0_I4(rZ, rZ, r, r)); default: return C_NotImplemented; diff --git a/tcg/optimize.c b/tcg/optimize.c index bca11cc427..f922f86a1d 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -3011,29 +3011,22 @@ void tcg_optimize(TCGContext *s) CASE_OP_32_64_VEC(orc): done = fold_orc(&ctx, op); break; - case INDEX_op_qemu_ld_a32_i32: - case INDEX_op_qemu_ld_a64_i32: + case INDEX_op_qemu_ld_i32: done = fold_qemu_ld_1reg(&ctx, op); break; - case INDEX_op_qemu_ld_a32_i64: - case INDEX_op_qemu_ld_a64_i64: + case INDEX_op_qemu_ld_i64: if (TCG_TARGET_REG_BITS == 64) { done = fold_qemu_ld_1reg(&ctx, op); break; } QEMU_FALLTHROUGH; - case INDEX_op_qemu_ld_a32_i128: - case INDEX_op_qemu_ld_a64_i128: + case INDEX_op_qemu_ld_i128: done = fold_qemu_ld_2reg(&ctx, op); break; - case INDEX_op_qemu_st8_a32_i32: - case INDEX_op_qemu_st8_a64_i32: - 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: - case INDEX_op_qemu_st_a32_i128: - case INDEX_op_qemu_st_a64_i128: + case INDEX_op_qemu_st8_i32: + case INDEX_op_qemu_st_i32: + case INDEX_op_qemu_st_i64: + case INDEX_op_qemu_st_i128: done = fold_qemu_st(&ctx, op); break; CASE_OP_32_64(rem): diff --git a/tcg/ppc/tcg-target.c.inc b/tcg/ppc/tcg-target.c.inc index 6e711cd53f..801cb6f3cb 100644 --- a/tcg/ppc/tcg-target.c.inc +++ b/tcg/ppc/tcg-target.c.inc @@ -3308,17 +3308,10 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, TCGType type, tcg_out32(s, MODUD | TAB(args[0], args[1], args[2])); break; - case INDEX_op_qemu_ld_a64_i32: - if (TCG_TARGET_REG_BITS == 32) { - tcg_out_qemu_ld(s, args[0], -1, args[1], args[2], - args[3], TCG_TYPE_I32); - break; - } - /* fall through */ - case INDEX_op_qemu_ld_a32_i32: + case INDEX_op_qemu_ld_i32: tcg_out_qemu_ld(s, args[0], -1, args[1], -1, args[2], TCG_TYPE_I32); break; - case INDEX_op_qemu_ld_a32_i64: + case INDEX_op_qemu_ld_i64: if (TCG_TARGET_REG_BITS == 64) { tcg_out_qemu_ld(s, args[0], -1, args[1], -1, args[2], TCG_TYPE_I64); @@ -3327,32 +3320,15 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, TCGType type, args[3], TCG_TYPE_I64); } break; - case INDEX_op_qemu_ld_a64_i64: - if (TCG_TARGET_REG_BITS == 64) { - tcg_out_qemu_ld(s, args[0], -1, args[1], -1, - args[2], TCG_TYPE_I64); - } else { - tcg_out_qemu_ld(s, args[0], args[1], args[2], args[3], - args[4], TCG_TYPE_I64); - } - break; - case INDEX_op_qemu_ld_a32_i128: - case INDEX_op_qemu_ld_a64_i128: + case INDEX_op_qemu_ld_i128: tcg_debug_assert(TCG_TARGET_REG_BITS == 64); tcg_out_qemu_ldst_i128(s, args[0], args[1], args[2], args[3], true); break; - case INDEX_op_qemu_st_a64_i32: - if (TCG_TARGET_REG_BITS == 32) { - tcg_out_qemu_st(s, args[0], -1, args[1], args[2], - args[3], TCG_TYPE_I32); - break; - } - /* fall through */ - case INDEX_op_qemu_st_a32_i32: + case INDEX_op_qemu_st_i32: tcg_out_qemu_st(s, args[0], -1, args[1], -1, args[2], TCG_TYPE_I32); break; - case INDEX_op_qemu_st_a32_i64: + case INDEX_op_qemu_st_i64: if (TCG_TARGET_REG_BITS == 64) { tcg_out_qemu_st(s, args[0], -1, args[1], -1, args[2], TCG_TYPE_I64); @@ -3361,17 +3337,7 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, TCGType type, args[3], TCG_TYPE_I64); } break; - case INDEX_op_qemu_st_a64_i64: - if (TCG_TARGET_REG_BITS == 64) { - tcg_out_qemu_st(s, args[0], -1, args[1], -1, - args[2], TCG_TYPE_I64); - } else { - tcg_out_qemu_st(s, args[0], args[1], args[2], args[3], - args[4], TCG_TYPE_I64); - } - break; - case INDEX_op_qemu_st_a32_i128: - case INDEX_op_qemu_st_a64_i128: + case INDEX_op_qemu_st_i128: tcg_debug_assert(TCG_TARGET_REG_BITS == 64); tcg_out_qemu_ldst_i128(s, args[0], args[1], args[2], args[3], false); break; @@ -4306,29 +4272,19 @@ tcg_target_op_def(TCGOpcode op, TCGType type, unsigned flags) case INDEX_op_sub2_i32: return C_O2_I4(r, r, rI, rZM, r, r); - case INDEX_op_qemu_ld_a32_i32: + case INDEX_op_qemu_ld_i32: return C_O1_I1(r, r); - case INDEX_op_qemu_ld_a64_i32: - return TCG_TARGET_REG_BITS == 64 ? C_O1_I1(r, r) : C_O1_I2(r, r, r); - case INDEX_op_qemu_ld_a32_i64: + case INDEX_op_qemu_ld_i64: return TCG_TARGET_REG_BITS == 64 ? C_O1_I1(r, r) : C_O2_I1(r, r, r); - case INDEX_op_qemu_ld_a64_i64: - return TCG_TARGET_REG_BITS == 64 ? C_O1_I1(r, r) : C_O2_I2(r, r, r, r); - case INDEX_op_qemu_st_a32_i32: + case INDEX_op_qemu_st_i32: return C_O0_I2(r, r); - case INDEX_op_qemu_st_a64_i32: + case INDEX_op_qemu_st_i64: return TCG_TARGET_REG_BITS == 64 ? C_O0_I2(r, r) : C_O0_I3(r, r, r); - case INDEX_op_qemu_st_a32_i64: - return TCG_TARGET_REG_BITS == 64 ? C_O0_I2(r, r) : C_O0_I3(r, r, r); - case INDEX_op_qemu_st_a64_i64: - return TCG_TARGET_REG_BITS == 64 ? C_O0_I2(r, r) : C_O0_I4(r, r, r, r); - case INDEX_op_qemu_ld_a32_i128: - case INDEX_op_qemu_ld_a64_i128: + case INDEX_op_qemu_ld_i128: return C_N1O1_I1(o, m, r); - case INDEX_op_qemu_st_a32_i128: - case INDEX_op_qemu_st_a64_i128: + case INDEX_op_qemu_st_i128: return C_O0_I3(o, m, r); case INDEX_op_add_vec: diff --git a/tcg/riscv/tcg-target.c.inc b/tcg/riscv/tcg-target.c.inc index 61dc310c1a..55a3398712 100644 --- a/tcg/riscv/tcg-target.c.inc +++ b/tcg/riscv/tcg-target.c.inc @@ -2309,20 +2309,16 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, TCGType type, args[3], const_args[3], args[4], const_args[4]); break; - case INDEX_op_qemu_ld_a32_i32: - case INDEX_op_qemu_ld_a64_i32: + case INDEX_op_qemu_ld_i32: tcg_out_qemu_ld(s, a0, a1, a2, TCG_TYPE_I32); break; - case INDEX_op_qemu_ld_a32_i64: - case INDEX_op_qemu_ld_a64_i64: + case INDEX_op_qemu_ld_i64: tcg_out_qemu_ld(s, a0, a1, a2, TCG_TYPE_I64); break; - case INDEX_op_qemu_st_a32_i32: - case INDEX_op_qemu_st_a64_i32: + case INDEX_op_qemu_st_i32: tcg_out_qemu_st(s, a0, a1, a2, TCG_TYPE_I32); break; - case INDEX_op_qemu_st_a32_i64: - case INDEX_op_qemu_st_a64_i64: + case INDEX_op_qemu_st_i64: tcg_out_qemu_st(s, a0, a1, a2, TCG_TYPE_I64); break; @@ -2761,15 +2757,11 @@ tcg_target_op_def(TCGOpcode op, TCGType type, unsigned flags) case INDEX_op_sub2_i64: return C_O2_I4(r, r, rZ, rZ, rM, rM); - case INDEX_op_qemu_ld_a32_i32: - case INDEX_op_qemu_ld_a64_i32: - case INDEX_op_qemu_ld_a32_i64: - case INDEX_op_qemu_ld_a64_i64: + case INDEX_op_qemu_ld_i32: + case INDEX_op_qemu_ld_i64: 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: + case INDEX_op_qemu_st_i32: + case INDEX_op_qemu_st_i64: return C_O0_I2(rZ, r); case INDEX_op_st_vec: diff --git a/tcg/s390x/tcg-target.c.inc b/tcg/s390x/tcg-target.c.inc index dc7722dc31..6786e7b316 100644 --- a/tcg/s390x/tcg-target.c.inc +++ b/tcg/s390x/tcg-target.c.inc @@ -2455,28 +2455,22 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, TCGType type, args[2], const_args[2], args[3], const_args[3], args[4]); break; - case INDEX_op_qemu_ld_a32_i32: - case INDEX_op_qemu_ld_a64_i32: + case INDEX_op_qemu_ld_i32: tcg_out_qemu_ld(s, args[0], args[1], args[2], TCG_TYPE_I32); break; - case INDEX_op_qemu_ld_a32_i64: - case INDEX_op_qemu_ld_a64_i64: + case INDEX_op_qemu_ld_i64: tcg_out_qemu_ld(s, args[0], args[1], args[2], TCG_TYPE_I64); break; - case INDEX_op_qemu_st_a32_i32: - case INDEX_op_qemu_st_a64_i32: + case INDEX_op_qemu_st_i32: tcg_out_qemu_st(s, args[0], args[1], args[2], TCG_TYPE_I32); break; - case INDEX_op_qemu_st_a32_i64: - case INDEX_op_qemu_st_a64_i64: + case INDEX_op_qemu_st_i64: tcg_out_qemu_st(s, args[0], args[1], args[2], TCG_TYPE_I64); break; - case INDEX_op_qemu_ld_a32_i128: - case INDEX_op_qemu_ld_a64_i128: + case INDEX_op_qemu_ld_i128: tcg_out_qemu_ldst_i128(s, args[0], args[1], args[2], args[3], true); break; - case INDEX_op_qemu_st_a32_i128: - case INDEX_op_qemu_st_a64_i128: + case INDEX_op_qemu_st_i128: tcg_out_qemu_ldst_i128(s, args[0], args[1], args[2], args[3], false); break; @@ -3366,21 +3360,15 @@ tcg_target_op_def(TCGOpcode op, TCGType type, unsigned flags) case INDEX_op_ctpop_i64: return C_O1_I1(r, r); - case INDEX_op_qemu_ld_a32_i32: - case INDEX_op_qemu_ld_a64_i32: - case INDEX_op_qemu_ld_a32_i64: - case INDEX_op_qemu_ld_a64_i64: + case INDEX_op_qemu_ld_i32: + case INDEX_op_qemu_ld_i64: return C_O1_I1(r, r); - case INDEX_op_qemu_st_a32_i64: - case INDEX_op_qemu_st_a64_i64: - case INDEX_op_qemu_st_a32_i32: - case INDEX_op_qemu_st_a64_i32: + case INDEX_op_qemu_st_i64: + case INDEX_op_qemu_st_i32: return C_O0_I2(r, r); - case INDEX_op_qemu_ld_a32_i128: - case INDEX_op_qemu_ld_a64_i128: + case INDEX_op_qemu_ld_i128: return C_O2_I1(o, m, r); - case INDEX_op_qemu_st_a32_i128: - case INDEX_op_qemu_st_a64_i128: + case INDEX_op_qemu_st_i128: return C_O0_I3(o, m, r); case INDEX_op_deposit_i32: diff --git a/tcg/sparc64/tcg-target.c.inc b/tcg/sparc64/tcg-target.c.inc index 733cb51651..ea0a3b8692 100644 --- a/tcg/sparc64/tcg-target.c.inc +++ b/tcg/sparc64/tcg-target.c.inc @@ -1426,20 +1426,16 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, TCGType type, tcg_out_arithi(s, a1, a0, 32, SHIFT_SRLX); break; - case INDEX_op_qemu_ld_a32_i32: - case INDEX_op_qemu_ld_a64_i32: + case INDEX_op_qemu_ld_i32: tcg_out_qemu_ld(s, a0, a1, a2, TCG_TYPE_I32); break; - case INDEX_op_qemu_ld_a32_i64: - case INDEX_op_qemu_ld_a64_i64: + case INDEX_op_qemu_ld_i64: tcg_out_qemu_ld(s, a0, a1, a2, TCG_TYPE_I64); break; - case INDEX_op_qemu_st_a32_i32: - case INDEX_op_qemu_st_a64_i32: + case INDEX_op_qemu_st_i32: tcg_out_qemu_st(s, a0, a1, a2, TCG_TYPE_I32); break; - case INDEX_op_qemu_st_a32_i64: - case INDEX_op_qemu_st_a64_i64: + case INDEX_op_qemu_st_i64: tcg_out_qemu_st(s, a0, a1, a2, TCG_TYPE_I64); break; @@ -1570,10 +1566,8 @@ tcg_target_op_def(TCGOpcode op, TCGType type, unsigned flags) case INDEX_op_extu_i32_i64: case INDEX_op_extract_i64: case INDEX_op_sextract_i64: - case INDEX_op_qemu_ld_a32_i32: - case INDEX_op_qemu_ld_a64_i32: - case INDEX_op_qemu_ld_a32_i64: - case INDEX_op_qemu_ld_a64_i64: + case INDEX_op_qemu_ld_i32: + case INDEX_op_qemu_ld_i64: return C_O1_I1(r, r); case INDEX_op_st8_i32: @@ -1583,10 +1577,8 @@ tcg_target_op_def(TCGOpcode op, TCGType type, unsigned flags) case INDEX_op_st_i32: case INDEX_op_st32_i64: case INDEX_op_st_i64: - 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: + case INDEX_op_qemu_st_i32: + case INDEX_op_qemu_st_i64: return C_O0_I2(rZ, r); case INDEX_op_add_i32: diff --git a/tcg/tcg-op-ldst.c b/tcg/tcg-op-ldst.c index 7ba9a3ef7e..73838e2701 100644 --- a/tcg/tcg-op-ldst.c +++ b/tcg/tcg-op-ldst.c @@ -217,7 +217,6 @@ static void tcg_gen_qemu_ld_i32_int(TCGv_i32 val, TCGTemp *addr, MemOp orig_memop; MemOpIdx orig_oi, oi; TCGv_i64 copy_addr; - TCGOpcode opc; tcg_gen_req_mo(TCG_MO_LD_LD | TCG_MO_ST_LD); orig_memop = memop = tcg_canonicalize_memop(memop, 0, 0); @@ -233,12 +232,8 @@ static void tcg_gen_qemu_ld_i32_int(TCGv_i32 val, TCGTemp *addr, } copy_addr = plugin_maybe_preserve_addr(addr); - if (tcg_ctx->addr_type == TCG_TYPE_I32) { - opc = INDEX_op_qemu_ld_a32_i32; - } else { - opc = INDEX_op_qemu_ld_a64_i32; - } - gen_ldst(opc, TCG_TYPE_I32, tcgv_i32_temp(val), NULL, addr, oi); + gen_ldst(INDEX_op_qemu_ld_i32, TCG_TYPE_I32, + tcgv_i32_temp(val), NULL, addr, oi); plugin_gen_mem_callbacks_i32(val, copy_addr, addr, orig_oi, QEMU_PLUGIN_MEM_R); @@ -295,17 +290,9 @@ static void tcg_gen_qemu_st_i32_int(TCGv_i32 val, TCGTemp *addr, } if (TCG_TARGET_HAS_qemu_st8_i32 && (memop & MO_SIZE) == MO_8) { - if (tcg_ctx->addr_type == TCG_TYPE_I32) { - opc = INDEX_op_qemu_st8_a32_i32; - } else { - opc = INDEX_op_qemu_st8_a64_i32; - } + opc = INDEX_op_qemu_st8_i32; } else { - if (tcg_ctx->addr_type == TCG_TYPE_I32) { - opc = INDEX_op_qemu_st_a32_i32; - } else { - opc = INDEX_op_qemu_st_a64_i32; - } + opc = INDEX_op_qemu_st_i32; } gen_ldst(opc, TCG_TYPE_I32, tcgv_i32_temp(val), NULL, addr, oi); plugin_gen_mem_callbacks_i32(val, NULL, addr, orig_oi, QEMU_PLUGIN_MEM_W); @@ -329,7 +316,6 @@ static void tcg_gen_qemu_ld_i64_int(TCGv_i64 val, TCGTemp *addr, MemOp orig_memop; MemOpIdx orig_oi, oi; TCGv_i64 copy_addr; - TCGOpcode opc; if (TCG_TARGET_REG_BITS == 32 && (memop & MO_SIZE) < MO_64) { tcg_gen_qemu_ld_i32_int(TCGV_LOW(val), addr, idx, memop); @@ -355,12 +341,7 @@ static void tcg_gen_qemu_ld_i64_int(TCGv_i64 val, TCGTemp *addr, } copy_addr = plugin_maybe_preserve_addr(addr); - if (tcg_ctx->addr_type == TCG_TYPE_I32) { - opc = INDEX_op_qemu_ld_a32_i64; - } else { - opc = INDEX_op_qemu_ld_a64_i64; - } - gen_ldst_i64(opc, val, addr, oi); + gen_ldst_i64(INDEX_op_qemu_ld_i64, val, addr, oi); plugin_gen_mem_callbacks_i64(val, copy_addr, addr, orig_oi, QEMU_PLUGIN_MEM_R); @@ -397,7 +378,6 @@ static void tcg_gen_qemu_st_i64_int(TCGv_i64 val, TCGTemp *addr, { TCGv_i64 swap = NULL; MemOpIdx orig_oi, oi; - TCGOpcode opc; if (TCG_TARGET_REG_BITS == 32 && (memop & MO_SIZE) < MO_64) { tcg_gen_qemu_st_i32_int(TCGV_LOW(val), addr, idx, memop); @@ -428,12 +408,7 @@ static void tcg_gen_qemu_st_i64_int(TCGv_i64 val, TCGTemp *addr, oi = make_memop_idx(memop, idx); } - if (tcg_ctx->addr_type == TCG_TYPE_I32) { - opc = INDEX_op_qemu_st_a32_i64; - } else { - opc = INDEX_op_qemu_st_a64_i64; - } - gen_ldst_i64(opc, val, addr, oi); + gen_ldst_i64(INDEX_op_qemu_st_i64, val, addr, oi); plugin_gen_mem_callbacks_i64(val, NULL, addr, orig_oi, QEMU_PLUGIN_MEM_W); if (swap) { @@ -545,7 +520,6 @@ static void tcg_gen_qemu_ld_i128_int(TCGv_i128 val, TCGTemp *addr, { MemOpIdx orig_oi; TCGv_i64 ext_addr = NULL; - TCGOpcode opc; check_max_alignment(memop_alignment_bits(memop)); tcg_gen_req_mo(TCG_MO_LD_LD | TCG_MO_ST_LD); @@ -573,12 +547,7 @@ static void tcg_gen_qemu_ld_i128_int(TCGv_i128 val, TCGTemp *addr, hi = TCGV128_HIGH(val); } - if (tcg_ctx->addr_type == TCG_TYPE_I32) { - opc = INDEX_op_qemu_ld_a32_i128; - } else { - opc = INDEX_op_qemu_ld_a64_i128; - } - gen_ldst(opc, TCG_TYPE_I128, tcgv_i64_temp(lo), + gen_ldst(INDEX_op_qemu_ld_i128, TCG_TYPE_I128, tcgv_i64_temp(lo), tcgv_i64_temp(hi), addr, oi); if (need_bswap) { @@ -594,12 +563,6 @@ static void tcg_gen_qemu_ld_i128_int(TCGv_i128 val, TCGTemp *addr, canonicalize_memop_i128_as_i64(mop, memop); need_bswap = (mop[0] ^ memop) & MO_BSWAP; - if (tcg_ctx->addr_type == TCG_TYPE_I32) { - opc = INDEX_op_qemu_ld_a32_i64; - } else { - opc = INDEX_op_qemu_ld_a64_i64; - } - /* * Since there are no global TCGv_i128, there is no visible state * changed if the second load faults. Load directly into the two @@ -613,7 +576,8 @@ static void tcg_gen_qemu_ld_i128_int(TCGv_i128 val, TCGTemp *addr, y = TCGV128_LOW(val); } - gen_ldst_i64(opc, x, addr, make_memop_idx(mop[0], idx)); + gen_ldst_i64(INDEX_op_qemu_ld_i64, x, addr, + make_memop_idx(mop[0], idx)); if (need_bswap) { tcg_gen_bswap64_i64(x, x); @@ -629,7 +593,8 @@ static void tcg_gen_qemu_ld_i128_int(TCGv_i128 val, TCGTemp *addr, addr_p8 = tcgv_i64_temp(t); } - gen_ldst_i64(opc, y, addr_p8, make_memop_idx(mop[1], idx)); + gen_ldst_i64(INDEX_op_qemu_ld_i64, y, addr_p8, + make_memop_idx(mop[1], idx)); tcg_temp_free_internal(addr_p8); if (need_bswap) { @@ -663,7 +628,6 @@ static void tcg_gen_qemu_st_i128_int(TCGv_i128 val, TCGTemp *addr, { MemOpIdx orig_oi; TCGv_i64 ext_addr = NULL; - TCGOpcode opc; check_max_alignment(memop_alignment_bits(memop)); tcg_gen_req_mo(TCG_MO_ST_LD | TCG_MO_ST_ST); @@ -694,13 +658,8 @@ static void tcg_gen_qemu_st_i128_int(TCGv_i128 val, TCGTemp *addr, hi = TCGV128_HIGH(val); } - if (tcg_ctx->addr_type == TCG_TYPE_I32) { - opc = INDEX_op_qemu_st_a32_i128; - } else { - opc = INDEX_op_qemu_st_a64_i128; - } - gen_ldst(opc, TCG_TYPE_I128, tcgv_i64_temp(lo), - tcgv_i64_temp(hi), addr, oi); + gen_ldst(INDEX_op_qemu_st_i128, TCG_TYPE_I128, + tcgv_i64_temp(lo), tcgv_i64_temp(hi), addr, oi); if (need_bswap) { tcg_temp_free_i64(lo); @@ -713,12 +672,6 @@ static void tcg_gen_qemu_st_i128_int(TCGv_i128 val, TCGTemp *addr, canonicalize_memop_i128_as_i64(mop, memop); - if (tcg_ctx->addr_type == TCG_TYPE_I32) { - opc = INDEX_op_qemu_st_a32_i64; - } else { - opc = INDEX_op_qemu_st_a64_i64; - } - if ((memop & MO_BSWAP) == MO_LE) { x = TCGV128_LOW(val); y = TCGV128_HIGH(val); @@ -733,7 +686,8 @@ static void tcg_gen_qemu_st_i128_int(TCGv_i128 val, TCGTemp *addr, x = b; } - gen_ldst_i64(opc, x, addr, make_memop_idx(mop[0], idx)); + gen_ldst_i64(INDEX_op_qemu_st_i64, x, addr, + make_memop_idx(mop[0], idx)); if (tcg_ctx->addr_type == TCG_TYPE_I32) { TCGv_i32 t = tcg_temp_ebb_new_i32(); @@ -747,10 +701,12 @@ static void tcg_gen_qemu_st_i128_int(TCGv_i128 val, TCGTemp *addr, if (b) { tcg_gen_bswap64_i64(b, y); - gen_ldst_i64(opc, b, addr_p8, make_memop_idx(mop[1], idx)); + gen_ldst_i64(INDEX_op_qemu_st_i64, b, addr_p8, + make_memop_idx(mop[1], idx)); tcg_temp_free_i64(b); } else { - gen_ldst_i64(opc, y, addr_p8, make_memop_idx(mop[1], idx)); + gen_ldst_i64(INDEX_op_qemu_st_i64, y, addr_p8, + make_memop_idx(mop[1], idx)); } tcg_temp_free_internal(addr_p8); } else { diff --git a/tcg/tcg.c b/tcg/tcg.c index 14c2d38160..fef93b25ff 100644 --- a/tcg/tcg.c +++ b/tcg/tcg.c @@ -2147,24 +2147,17 @@ bool tcg_op_supported(TCGOpcode op, TCGType type, unsigned flags) case INDEX_op_exit_tb: case INDEX_op_goto_tb: case INDEX_op_goto_ptr: - case INDEX_op_qemu_ld_a32_i32: - case INDEX_op_qemu_ld_a64_i32: - case INDEX_op_qemu_st_a32_i32: - case INDEX_op_qemu_st_a64_i32: - case INDEX_op_qemu_ld_a32_i64: - case INDEX_op_qemu_ld_a64_i64: - case INDEX_op_qemu_st_a32_i64: - case INDEX_op_qemu_st_a64_i64: + case INDEX_op_qemu_ld_i32: + case INDEX_op_qemu_st_i32: + case INDEX_op_qemu_ld_i64: + case INDEX_op_qemu_st_i64: return true; - case INDEX_op_qemu_st8_a32_i32: - case INDEX_op_qemu_st8_a64_i32: + case INDEX_op_qemu_st8_i32: return TCG_TARGET_HAS_qemu_st8_i32; - case INDEX_op_qemu_ld_a32_i128: - case INDEX_op_qemu_ld_a64_i128: - case INDEX_op_qemu_st_a32_i128: - case INDEX_op_qemu_st_a64_i128: + case INDEX_op_qemu_ld_i128: + case INDEX_op_qemu_st_i128: return TCG_TARGET_HAS_qemu_ldst_i128; case INDEX_op_mov_i32: @@ -2862,20 +2855,13 @@ void tcg_dump_ops(TCGContext *s, FILE *f, bool have_prefs) } i = 1; break; - case INDEX_op_qemu_ld_a32_i32: - case INDEX_op_qemu_ld_a64_i32: - case INDEX_op_qemu_st_a32_i32: - case INDEX_op_qemu_st_a64_i32: - case INDEX_op_qemu_st8_a32_i32: - case INDEX_op_qemu_st8_a64_i32: - case INDEX_op_qemu_ld_a32_i64: - case INDEX_op_qemu_ld_a64_i64: - case INDEX_op_qemu_st_a32_i64: - case INDEX_op_qemu_st_a64_i64: - case INDEX_op_qemu_ld_a32_i128: - case INDEX_op_qemu_ld_a64_i128: - case INDEX_op_qemu_st_a32_i128: - case INDEX_op_qemu_st_a64_i128: + case INDEX_op_qemu_ld_i32: + case INDEX_op_qemu_st_i32: + case INDEX_op_qemu_st8_i32: + case INDEX_op_qemu_ld_i64: + case INDEX_op_qemu_st_i64: + case INDEX_op_qemu_ld_i128: + case INDEX_op_qemu_st_i128: { const char *s_al, *s_op, *s_at; MemOpIdx oi = op->args[k++]; diff --git a/tcg/tci.c b/tcg/tci.c index 8c1c53424d..d223258efe 100644 --- a/tcg/tci.c +++ b/tcg/tci.c @@ -154,16 +154,6 @@ static void tci_args_rrrbb(uint32_t insn, TCGReg *r0, TCGReg *r1, *i4 = extract32(insn, 26, 6); } -static void tci_args_rrrrr(uint32_t insn, TCGReg *r0, TCGReg *r1, - TCGReg *r2, TCGReg *r3, TCGReg *r4) -{ - *r0 = extract32(insn, 8, 4); - *r1 = extract32(insn, 12, 4); - *r2 = extract32(insn, 16, 4); - *r3 = extract32(insn, 20, 4); - *r4 = extract32(insn, 24, 4); -} - static void tci_args_rrrr(uint32_t insn, TCGReg *r0, TCGReg *r1, TCGReg *r2, TCGReg *r3) { @@ -912,43 +902,21 @@ uintptr_t QEMU_DISABLE_CFI tcg_qemu_tb_exec(CPUArchState *env, tb_ptr = ptr; break; - case INDEX_op_qemu_ld_a32_i32: + case INDEX_op_qemu_ld_i32: tci_args_rrm(insn, &r0, &r1, &oi); - taddr = (uint32_t)regs[r1]; - goto do_ld_i32; - case INDEX_op_qemu_ld_a64_i32: - if (TCG_TARGET_REG_BITS == 64) { - tci_args_rrm(insn, &r0, &r1, &oi); - taddr = regs[r1]; - } else { - tci_args_rrrr(insn, &r0, &r1, &r2, &r3); - taddr = tci_uint64(regs[r2], regs[r1]); - oi = regs[r3]; - } - do_ld_i32: + taddr = regs[r1]; regs[r0] = tci_qemu_ld(env, taddr, oi, tb_ptr); break; - case INDEX_op_qemu_ld_a32_i64: - if (TCG_TARGET_REG_BITS == 64) { - tci_args_rrm(insn, &r0, &r1, &oi); - taddr = (uint32_t)regs[r1]; - } else { - tci_args_rrrr(insn, &r0, &r1, &r2, &r3); - taddr = (uint32_t)regs[r2]; - oi = regs[r3]; - } - goto do_ld_i64; - case INDEX_op_qemu_ld_a64_i64: + case INDEX_op_qemu_ld_i64: if (TCG_TARGET_REG_BITS == 64) { tci_args_rrm(insn, &r0, &r1, &oi); taddr = regs[r1]; } else { - tci_args_rrrrr(insn, &r0, &r1, &r2, &r3, &r4); - taddr = tci_uint64(regs[r3], regs[r2]); - oi = regs[r4]; + tci_args_rrrr(insn, &r0, &r1, &r2, &r3); + taddr = regs[r2]; + oi = regs[r3]; } - do_ld_i64: tmp64 = tci_qemu_ld(env, taddr, oi, tb_ptr); if (TCG_TARGET_REG_BITS == 32) { tci_write_reg64(regs, r1, r0, tmp64); @@ -957,47 +925,23 @@ uintptr_t QEMU_DISABLE_CFI tcg_qemu_tb_exec(CPUArchState *env, } break; - case INDEX_op_qemu_st_a32_i32: + case INDEX_op_qemu_st_i32: tci_args_rrm(insn, &r0, &r1, &oi); - taddr = (uint32_t)regs[r1]; - goto do_st_i32; - case INDEX_op_qemu_st_a64_i32: - if (TCG_TARGET_REG_BITS == 64) { - tci_args_rrm(insn, &r0, &r1, &oi); - taddr = regs[r1]; - } else { - tci_args_rrrr(insn, &r0, &r1, &r2, &r3); - taddr = tci_uint64(regs[r2], regs[r1]); - oi = regs[r3]; - } - do_st_i32: + taddr = regs[r1]; tci_qemu_st(env, taddr, regs[r0], oi, tb_ptr); break; - case INDEX_op_qemu_st_a32_i64: - if (TCG_TARGET_REG_BITS == 64) { - tci_args_rrm(insn, &r0, &r1, &oi); - tmp64 = regs[r0]; - taddr = (uint32_t)regs[r1]; - } else { - tci_args_rrrr(insn, &r0, &r1, &r2, &r3); - tmp64 = tci_uint64(regs[r1], regs[r0]); - taddr = (uint32_t)regs[r2]; - oi = regs[r3]; - } - goto do_st_i64; - case INDEX_op_qemu_st_a64_i64: + case INDEX_op_qemu_st_i64: if (TCG_TARGET_REG_BITS == 64) { tci_args_rrm(insn, &r0, &r1, &oi); tmp64 = regs[r0]; taddr = regs[r1]; } else { - tci_args_rrrrr(insn, &r0, &r1, &r2, &r3, &r4); + tci_args_rrrr(insn, &r0, &r1, &r2, &r3); tmp64 = tci_uint64(regs[r1], regs[r0]); - taddr = tci_uint64(regs[r3], regs[r2]); - oi = regs[r4]; + taddr = regs[r2]; + oi = regs[r3]; } - do_st_i64: tci_qemu_st(env, taddr, tmp64, oi, tb_ptr); break; @@ -1269,42 +1213,21 @@ int print_insn_tci(bfd_vma addr, disassemble_info *info) str_r(r3), str_r(r4), str_r(r5)); break; - case INDEX_op_qemu_ld_a32_i32: - case INDEX_op_qemu_st_a32_i32: - len = 1 + 1; - goto do_qemu_ldst; - case INDEX_op_qemu_ld_a32_i64: - case INDEX_op_qemu_st_a32_i64: - case INDEX_op_qemu_ld_a64_i32: - case INDEX_op_qemu_st_a64_i32: - len = 1 + DIV_ROUND_UP(64, TCG_TARGET_REG_BITS); - goto do_qemu_ldst; - case INDEX_op_qemu_ld_a64_i64: - case INDEX_op_qemu_st_a64_i64: - len = 2 * DIV_ROUND_UP(64, TCG_TARGET_REG_BITS); - goto do_qemu_ldst; - do_qemu_ldst: - switch (len) { - case 2: - tci_args_rrm(insn, &r0, &r1, &oi); - info->fprintf_func(info->stream, "%-12s %s, %s, %x", - op_name, str_r(r0), str_r(r1), oi); - break; - case 3: + case INDEX_op_qemu_ld_i64: + case INDEX_op_qemu_st_i64: + if (TCG_TARGET_REG_BITS == 32) { tci_args_rrrr(insn, &r0, &r1, &r2, &r3); info->fprintf_func(info->stream, "%-12s %s, %s, %s, %s", op_name, str_r(r0), str_r(r1), str_r(r2), str_r(r3)); break; - case 4: - tci_args_rrrrr(insn, &r0, &r1, &r2, &r3, &r4); - info->fprintf_func(info->stream, "%-12s %s, %s, %s, %s, %s", - op_name, str_r(r0), str_r(r1), - str_r(r2), str_r(r3), str_r(r4)); - break; - default: - g_assert_not_reached(); } + /* fall through */ + case INDEX_op_qemu_ld_i32: + case INDEX_op_qemu_st_i32: + tci_args_rrm(insn, &r0, &r1, &oi); + info->fprintf_func(info->stream, "%-12s %s, %s, %x", + op_name, str_r(r0), str_r(r1), oi); break; case 0: diff --git a/tcg/tci/tcg-target.c.inc b/tcg/tci/tcg-target.c.inc index d6c77325a3..36e018dd19 100644 --- a/tcg/tci/tcg-target.c.inc +++ b/tcg/tci/tcg-target.c.inc @@ -169,22 +169,14 @@ tcg_target_op_def(TCGOpcode op, TCGType type, unsigned flags) case INDEX_op_setcond2_i32: return C_O1_I4(r, r, r, r, r); - case INDEX_op_qemu_ld_a32_i32: + case INDEX_op_qemu_ld_i32: return C_O1_I1(r, r); - case INDEX_op_qemu_ld_a64_i32: - return TCG_TARGET_REG_BITS == 64 ? C_O1_I1(r, r) : C_O1_I2(r, r, r); - case INDEX_op_qemu_ld_a32_i64: + case INDEX_op_qemu_ld_i64: return TCG_TARGET_REG_BITS == 64 ? C_O1_I1(r, r) : C_O2_I1(r, r, r); - case INDEX_op_qemu_ld_a64_i64: - return TCG_TARGET_REG_BITS == 64 ? C_O1_I1(r, r) : C_O2_I2(r, r, r, r); - case INDEX_op_qemu_st_a32_i32: + case INDEX_op_qemu_st_i32: return C_O0_I2(r, r); - case INDEX_op_qemu_st_a64_i32: + case INDEX_op_qemu_st_i64: return TCG_TARGET_REG_BITS == 64 ? C_O0_I2(r, r) : C_O0_I3(r, r, r); - case INDEX_op_qemu_st_a32_i64: - return TCG_TARGET_REG_BITS == 64 ? C_O0_I2(r, r) : C_O0_I3(r, r, r); - case INDEX_op_qemu_st_a64_i64: - return TCG_TARGET_REG_BITS == 64 ? C_O0_I2(r, r) : C_O0_I4(r, r, r, r); default: return C_NotImplemented; @@ -422,20 +414,6 @@ static void tcg_out_op_rrrbb(TCGContext *s, TCGOpcode op, TCGReg r0, tcg_out32(s, insn); } -static void tcg_out_op_rrrrr(TCGContext *s, TCGOpcode op, TCGReg r0, - TCGReg r1, TCGReg r2, TCGReg r3, TCGReg r4) -{ - tcg_insn_unit insn = 0; - - insn = deposit32(insn, 0, 8, op); - insn = deposit32(insn, 8, 4, r0); - insn = deposit32(insn, 12, 4, r1); - insn = deposit32(insn, 16, 4, r2); - insn = deposit32(insn, 20, 4, r3); - insn = deposit32(insn, 24, 4, r4); - tcg_out32(s, insn); -} - static void tcg_out_op_rrrr(TCGContext *s, TCGOpcode op, TCGReg r0, TCGReg r1, TCGReg r2, TCGReg r3) { @@ -833,29 +811,21 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, TCGType type, tcg_out_op_rrrr(s, opc, args[0], args[1], args[2], args[3]); break; - case INDEX_op_qemu_ld_a32_i32: - case INDEX_op_qemu_st_a32_i32: - tcg_out_op_rrm(s, opc, args[0], args[1], args[2]); - break; - case INDEX_op_qemu_ld_a64_i32: - case INDEX_op_qemu_st_a64_i32: - case INDEX_op_qemu_ld_a32_i64: - case INDEX_op_qemu_st_a32_i64: - if (TCG_TARGET_REG_BITS == 64) { - tcg_out_op_rrm(s, opc, args[0], args[1], args[2]); - } else { + case INDEX_op_qemu_ld_i64: + case INDEX_op_qemu_st_i64: + if (TCG_TARGET_REG_BITS == 32) { tcg_out_movi(s, TCG_TYPE_I32, TCG_REG_TMP, args[3]); tcg_out_op_rrrr(s, opc, args[0], args[1], args[2], TCG_REG_TMP); + break; } - break; - case INDEX_op_qemu_ld_a64_i64: - case INDEX_op_qemu_st_a64_i64: - if (TCG_TARGET_REG_BITS == 64) { - tcg_out_op_rrm(s, opc, args[0], args[1], args[2]); + /* fall through */ + case INDEX_op_qemu_ld_i32: + case INDEX_op_qemu_st_i32: + if (TCG_TARGET_REG_BITS == 64 && s->addr_type == TCG_TYPE_I32) { + tcg_out_ext32u(s, TCG_REG_TMP, args[1]); + tcg_out_op_rrm(s, opc, args[0], TCG_REG_TMP, args[2]); } else { - tcg_out_movi(s, TCG_TYPE_I32, TCG_REG_TMP, args[4]); - tcg_out_op_rrrrr(s, opc, args[0], args[1], - args[2], args[3], TCG_REG_TMP); + tcg_out_op_rrm(s, opc, args[0], args[1], args[2]); } break; From d9a8889f6d10f586e28f4bfb8611cf0bc36fcaa9 Mon Sep 17 00:00:00 2001 From: Richard Henderson Date: Tue, 4 Feb 2025 17:20:55 -0800 Subject: [PATCH 05/28] tcg/arm: Drop addrhi from prepare_host_addr MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The guest address will now always be TCG_TYPE_I32. Reviewed-by: Philippe Mathieu-Daudé Signed-off-by: Richard Henderson --- tcg/arm/tcg-target.c.inc | 73 +++++++++++++--------------------------- 1 file changed, 23 insertions(+), 50 deletions(-) diff --git a/tcg/arm/tcg-target.c.inc b/tcg/arm/tcg-target.c.inc index 05bb367a39..93a3ccaf66 100644 --- a/tcg/arm/tcg-target.c.inc +++ b/tcg/arm/tcg-target.c.inc @@ -676,14 +676,8 @@ static void tcg_out_ldrd_r(TCGContext *s, ARMCond cond, TCGReg rt, tcg_out_memop_r(s, cond, INSN_LDRD_REG, rt, rn, rm, 1, 1, 0); } -static void __attribute__((unused)) -tcg_out_ldrd_rwb(TCGContext *s, ARMCond cond, TCGReg rt, TCGReg rn, TCGReg rm) -{ - tcg_out_memop_r(s, cond, INSN_LDRD_REG, rt, rn, rm, 1, 1, 1); -} - -static void __attribute__((unused)) -tcg_out_strd_8(TCGContext *s, ARMCond cond, TCGReg rt, TCGReg rn, int imm8) +static void tcg_out_strd_8(TCGContext *s, ARMCond cond, TCGReg rt, + TCGReg rn, int imm8) { tcg_out_memop_8(s, cond, INSN_STRD_IMM, rt, rn, imm8, 1, 0); } @@ -1455,8 +1449,7 @@ static bool tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *lb) #define MIN_TLB_MASK_TABLE_OFS -256 static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, - TCGReg addrlo, TCGReg addrhi, - MemOpIdx oi, bool is_ld) + TCGReg addr, MemOpIdx oi, bool is_ld) { TCGLabelQemuLdst *ldst = NULL; MemOp opc = get_memop(oi); @@ -1465,14 +1458,14 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, if (tcg_use_softmmu) { *h = (HostAddress){ .cond = COND_AL, - .base = addrlo, + .base = addr, .index = TCG_REG_R1, .index_scratch = true, }; } else { *h = (HostAddress){ .cond = COND_AL, - .base = addrlo, + .base = addr, .index = guest_base ? TCG_REG_GUEST_BASE : -1, .index_scratch = false, }; @@ -1492,8 +1485,7 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, ldst = new_ldst_label(s); ldst->is_ld = is_ld; ldst->oi = oi; - ldst->addrlo_reg = addrlo; - ldst->addrhi_reg = addrhi; + ldst->addrlo_reg = addr; /* Load cpu->neg.tlb.f[mmu_idx].{mask,table} into {r0,r1}. */ QEMU_BUILD_BUG_ON(offsetof(CPUTLBDescFast, mask) != 0); @@ -1501,30 +1493,20 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, tcg_out_ldrd_8(s, COND_AL, TCG_REG_R0, TCG_AREG0, fast_off); /* Extract the tlb index from the address into R0. */ - tcg_out_dat_reg(s, COND_AL, ARITH_AND, TCG_REG_R0, TCG_REG_R0, addrlo, + tcg_out_dat_reg(s, COND_AL, ARITH_AND, TCG_REG_R0, TCG_REG_R0, addr, SHIFT_IMM_LSR(s->page_bits - CPU_TLB_ENTRY_BITS)); /* * Add the tlb_table pointer, creating the CPUTLBEntry address in R1. - * Load the tlb comparator into R2/R3 and the fast path addend into R1. + * Load the tlb comparator into R2 and the fast path addend into R1. */ QEMU_BUILD_BUG_ON(HOST_BIG_ENDIAN); if (cmp_off == 0) { - if (s->addr_type == TCG_TYPE_I32) { - tcg_out_ld32_rwb(s, COND_AL, TCG_REG_R2, - TCG_REG_R1, TCG_REG_R0); - } else { - tcg_out_ldrd_rwb(s, COND_AL, TCG_REG_R2, - TCG_REG_R1, TCG_REG_R0); - } + tcg_out_ld32_rwb(s, COND_AL, TCG_REG_R2, TCG_REG_R1, TCG_REG_R0); } else { tcg_out_dat_reg(s, COND_AL, ARITH_ADD, TCG_REG_R1, TCG_REG_R1, TCG_REG_R0, 0); - if (s->addr_type == TCG_TYPE_I32) { - tcg_out_ld32_12(s, COND_AL, TCG_REG_R2, TCG_REG_R1, cmp_off); - } else { - tcg_out_ldrd_8(s, COND_AL, TCG_REG_R2, TCG_REG_R1, cmp_off); - } + tcg_out_ld32_12(s, COND_AL, TCG_REG_R2, TCG_REG_R1, cmp_off); } /* Load the tlb addend. */ @@ -1543,11 +1525,11 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, * This leaves the least significant alignment bits unchanged, and of * course must be zero. */ - t_addr = addrlo; + t_addr = addr; if (a_mask < s_mask) { t_addr = TCG_REG_R0; tcg_out_dat_imm(s, COND_AL, ARITH_ADD, t_addr, - addrlo, s_mask - a_mask); + addr, s_mask - a_mask); } if (use_armv7_instructions && s->page_bits <= 16) { tcg_out_movi32(s, COND_AL, TCG_REG_TMP, ~(s->page_mask | a_mask)); @@ -1558,7 +1540,7 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, } else { if (a_mask) { tcg_debug_assert(a_mask <= 0xff); - tcg_out_dat_imm(s, COND_AL, ARITH_TST, 0, addrlo, a_mask); + tcg_out_dat_imm(s, COND_AL, ARITH_TST, 0, addr, a_mask); } tcg_out_dat_reg(s, COND_AL, ARITH_MOV, TCG_REG_TMP, 0, t_addr, SHIFT_IMM_LSR(s->page_bits)); @@ -1566,21 +1548,16 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, 0, TCG_REG_R2, TCG_REG_TMP, SHIFT_IMM_LSL(s->page_bits)); } - - if (s->addr_type != TCG_TYPE_I32) { - tcg_out_dat_reg(s, COND_EQ, ARITH_CMP, 0, TCG_REG_R3, addrhi, 0); - } } else if (a_mask) { ldst = new_ldst_label(s); ldst->is_ld = is_ld; ldst->oi = oi; - ldst->addrlo_reg = addrlo; - ldst->addrhi_reg = addrhi; + ldst->addrlo_reg = addr; /* We are expecting alignment to max out at 7 */ tcg_debug_assert(a_mask <= 0xff); /* tst addr, #mask */ - tcg_out_dat_imm(s, COND_AL, ARITH_TST, 0, addrlo, a_mask); + tcg_out_dat_imm(s, COND_AL, ARITH_TST, 0, addr, a_mask); } return ldst; @@ -1678,14 +1655,13 @@ static void tcg_out_qemu_ld_direct(TCGContext *s, MemOp opc, TCGReg datalo, } static void tcg_out_qemu_ld(TCGContext *s, TCGReg datalo, TCGReg datahi, - TCGReg addrlo, TCGReg addrhi, - MemOpIdx oi, TCGType data_type) + TCGReg addr, MemOpIdx oi, TCGType data_type) { MemOp opc = get_memop(oi); TCGLabelQemuLdst *ldst; HostAddress h; - ldst = prepare_host_addr(s, &h, addrlo, addrhi, oi, true); + ldst = prepare_host_addr(s, &h, addr, oi, true); if (ldst) { ldst->type = data_type; ldst->datalo_reg = datalo; @@ -1764,14 +1740,13 @@ static void tcg_out_qemu_st_direct(TCGContext *s, MemOp opc, TCGReg datalo, } static void tcg_out_qemu_st(TCGContext *s, TCGReg datalo, TCGReg datahi, - TCGReg addrlo, TCGReg addrhi, - MemOpIdx oi, TCGType data_type) + TCGReg addr, MemOpIdx oi, TCGType data_type) { MemOp opc = get_memop(oi); TCGLabelQemuLdst *ldst; HostAddress h; - ldst = prepare_host_addr(s, &h, addrlo, addrhi, oi, false); + ldst = prepare_host_addr(s, &h, addr, oi, false); if (ldst) { ldst->type = data_type; ldst->datalo_reg = datalo; @@ -2072,19 +2047,17 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, TCGType type, break; case INDEX_op_qemu_ld_i32: - tcg_out_qemu_ld(s, args[0], -1, args[1], -1, args[2], TCG_TYPE_I32); + tcg_out_qemu_ld(s, args[0], -1, args[1], args[2], TCG_TYPE_I32); break; case INDEX_op_qemu_ld_i64: - tcg_out_qemu_ld(s, args[0], args[1], args[2], -1, - args[3], TCG_TYPE_I64); + tcg_out_qemu_ld(s, args[0], args[1], args[2], args[3], TCG_TYPE_I64); break; case INDEX_op_qemu_st_i32: - tcg_out_qemu_st(s, args[0], -1, args[1], -1, args[2], TCG_TYPE_I32); + tcg_out_qemu_st(s, args[0], -1, args[1], args[2], TCG_TYPE_I32); break; case INDEX_op_qemu_st_i64: - tcg_out_qemu_st(s, args[0], args[1], args[2], -1, - args[3], TCG_TYPE_I64); + tcg_out_qemu_st(s, args[0], args[1], args[2], args[3], TCG_TYPE_I64); break; case INDEX_op_bswap16_i32: From dc8e2f8f7840835e51f23b891b75a79101efc91c Mon Sep 17 00:00:00 2001 From: Richard Henderson Date: Tue, 4 Feb 2025 17:36:35 -0800 Subject: [PATCH 06/28] tcg/i386: Drop addrhi from prepare_host_addr MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The guest address will now always fit in one register. Reviewed-by: Philippe Mathieu-Daudé Signed-off-by: Richard Henderson --- tcg/i386/tcg-target.c.inc | 56 ++++++++++++++------------------------- 1 file changed, 20 insertions(+), 36 deletions(-) diff --git a/tcg/i386/tcg-target.c.inc b/tcg/i386/tcg-target.c.inc index ca6e8abc57..b33fe7fe23 100644 --- a/tcg/i386/tcg-target.c.inc +++ b/tcg/i386/tcg-target.c.inc @@ -2169,8 +2169,7 @@ static inline int setup_guest_base_seg(void) * is required and fill in @h with the host address for the fast path. */ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, - TCGReg addrlo, TCGReg addrhi, - MemOpIdx oi, bool is_ld) + TCGReg addr, MemOpIdx oi, bool is_ld) { TCGLabelQemuLdst *ldst = NULL; MemOp opc = get_memop(oi); @@ -2184,7 +2183,7 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, } else { *h = x86_guest_base; } - h->base = addrlo; + h->base = addr; h->aa = atom_and_align_for_opc(s, opc, MO_ATOM_IFALIGN, s_bits == MO_128); a_mask = (1 << h->aa.align) - 1; @@ -2202,8 +2201,7 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, ldst = new_ldst_label(s); ldst->is_ld = is_ld; ldst->oi = oi; - ldst->addrlo_reg = addrlo; - ldst->addrhi_reg = addrhi; + ldst->addrlo_reg = addr; if (TCG_TARGET_REG_BITS == 64) { ttype = s->addr_type; @@ -2217,7 +2215,7 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, } } - tcg_out_mov(s, tlbtype, TCG_REG_L0, addrlo); + tcg_out_mov(s, tlbtype, TCG_REG_L0, addr); tcg_out_shifti(s, SHIFT_SHR + tlbrexw, TCG_REG_L0, s->page_bits - CPU_TLB_ENTRY_BITS); @@ -2233,10 +2231,10 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, * check that we don't cross pages for the complete access. */ if (a_mask >= s_mask) { - tcg_out_mov(s, ttype, TCG_REG_L1, addrlo); + tcg_out_mov(s, ttype, TCG_REG_L1, addr); } else { tcg_out_modrm_offset(s, OPC_LEA + trexw, TCG_REG_L1, - addrlo, s_mask - a_mask); + addr, s_mask - a_mask); } tlb_mask = s->page_mask | a_mask; tgen_arithi(s, ARITH_AND + trexw, TCG_REG_L1, tlb_mask, 0); @@ -2250,17 +2248,6 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, ldst->label_ptr[0] = s->code_ptr; s->code_ptr += 4; - if (TCG_TARGET_REG_BITS == 32 && s->addr_type == TCG_TYPE_I64) { - /* cmp 4(TCG_REG_L0), addrhi */ - tcg_out_modrm_offset(s, OPC_CMP_GvEv, addrhi, - TCG_REG_L0, cmp_ofs + 4); - - /* jne slow_path */ - tcg_out_opc(s, OPC_JCC_long + JCC_JNE, 0, 0, 0); - ldst->label_ptr[1] = s->code_ptr; - s->code_ptr += 4; - } - /* TLB Hit. */ tcg_out_ld(s, TCG_TYPE_PTR, TCG_REG_L0, TCG_REG_L0, offsetof(CPUTLBEntry, addend)); @@ -2270,11 +2257,10 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, ldst = new_ldst_label(s); ldst->is_ld = is_ld; ldst->oi = oi; - ldst->addrlo_reg = addrlo; - ldst->addrhi_reg = addrhi; + ldst->addrlo_reg = addr; /* jne slow_path */ - jcc = tcg_out_cmp(s, TCG_COND_TSTNE, addrlo, a_mask, true, false); + jcc = tcg_out_cmp(s, TCG_COND_TSTNE, addr, a_mask, true, false); tcg_out_opc(s, OPC_JCC_long + jcc, 0, 0, 0); ldst->label_ptr[0] = s->code_ptr; s->code_ptr += 4; @@ -2446,13 +2432,12 @@ static void tcg_out_qemu_ld_direct(TCGContext *s, TCGReg datalo, TCGReg datahi, } static void tcg_out_qemu_ld(TCGContext *s, TCGReg datalo, TCGReg datahi, - TCGReg addrlo, TCGReg addrhi, - MemOpIdx oi, TCGType data_type) + TCGReg addr, MemOpIdx oi, TCGType data_type) { TCGLabelQemuLdst *ldst; HostAddress h; - ldst = prepare_host_addr(s, &h, addrlo, addrhi, oi, true); + ldst = prepare_host_addr(s, &h, addr, oi, true); tcg_out_qemu_ld_direct(s, datalo, datahi, h, data_type, get_memop(oi)); if (ldst) { @@ -2574,13 +2559,12 @@ static void tcg_out_qemu_st_direct(TCGContext *s, TCGReg datalo, TCGReg datahi, } static void tcg_out_qemu_st(TCGContext *s, TCGReg datalo, TCGReg datahi, - TCGReg addrlo, TCGReg addrhi, - MemOpIdx oi, TCGType data_type) + TCGReg addr, MemOpIdx oi, TCGType data_type) { TCGLabelQemuLdst *ldst; HostAddress h; - ldst = prepare_host_addr(s, &h, addrlo, addrhi, oi, false); + ldst = prepare_host_addr(s, &h, addr, oi, false); tcg_out_qemu_st_direct(s, datalo, datahi, h, get_memop(oi)); if (ldst) { @@ -2880,34 +2864,34 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, TCGType type, break; case INDEX_op_qemu_ld_i32: - tcg_out_qemu_ld(s, a0, -1, a1, -1, a2, TCG_TYPE_I32); + tcg_out_qemu_ld(s, a0, -1, a1, a2, TCG_TYPE_I32); break; case INDEX_op_qemu_ld_i64: if (TCG_TARGET_REG_BITS == 64) { - tcg_out_qemu_ld(s, a0, -1, a1, -1, a2, TCG_TYPE_I64); + tcg_out_qemu_ld(s, a0, -1, a1, a2, TCG_TYPE_I64); } else { - tcg_out_qemu_ld(s, a0, a1, a2, -1, args[3], TCG_TYPE_I64); + tcg_out_qemu_ld(s, a0, a1, a2, args[3], TCG_TYPE_I64); } break; case INDEX_op_qemu_ld_i128: tcg_debug_assert(TCG_TARGET_REG_BITS == 64); - tcg_out_qemu_ld(s, a0, a1, a2, -1, args[3], TCG_TYPE_I128); + tcg_out_qemu_ld(s, a0, a1, a2, args[3], TCG_TYPE_I128); break; case INDEX_op_qemu_st_i32: case INDEX_op_qemu_st8_i32: - tcg_out_qemu_st(s, a0, -1, a1, -1, a2, TCG_TYPE_I32); + tcg_out_qemu_st(s, a0, -1, a1, a2, TCG_TYPE_I32); break; case INDEX_op_qemu_st_i64: if (TCG_TARGET_REG_BITS == 64) { - tcg_out_qemu_st(s, a0, -1, a1, -1, a2, TCG_TYPE_I64); + tcg_out_qemu_st(s, a0, -1, a1, a2, TCG_TYPE_I64); } else { - tcg_out_qemu_st(s, a0, a1, a2, -1, args[3], TCG_TYPE_I64); + tcg_out_qemu_st(s, a0, a1, a2, args[3], TCG_TYPE_I64); } break; case INDEX_op_qemu_st_i128: tcg_debug_assert(TCG_TARGET_REG_BITS == 64); - tcg_out_qemu_st(s, a0, a1, a2, -1, args[3], TCG_TYPE_I128); + tcg_out_qemu_st(s, a0, a1, a2, args[3], TCG_TYPE_I128); break; OP_32_64(mulu2): From 0d000618d92344e0575018dcaa8ebc01e2d589b3 Mon Sep 17 00:00:00 2001 From: Richard Henderson Date: Tue, 4 Feb 2025 17:43:23 -0800 Subject: [PATCH 07/28] tcg/mips: Drop addrhi from prepare_host_addr MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The guest address will now always fit in one register. Reviewed-by: Philippe Mathieu-Daudé Signed-off-by: Richard Henderson --- tcg/mips/tcg-target.c.inc | 62 ++++++++++++++------------------------- 1 file changed, 22 insertions(+), 40 deletions(-) diff --git a/tcg/mips/tcg-target.c.inc b/tcg/mips/tcg-target.c.inc index b1d512ca2a..153ce1f3c3 100644 --- a/tcg/mips/tcg-target.c.inc +++ b/tcg/mips/tcg-target.c.inc @@ -1217,8 +1217,7 @@ bool tcg_target_has_memory_bswap(MemOp memop) * is required and fill in @h with the host address for the fast path. */ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, - TCGReg addrlo, TCGReg addrhi, - MemOpIdx oi, bool is_ld) + TCGReg addr, MemOpIdx oi, bool is_ld) { TCGType addr_type = s->addr_type; TCGLabelQemuLdst *ldst = NULL; @@ -1245,8 +1244,7 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, ldst = new_ldst_label(s); ldst->is_ld = is_ld; ldst->oi = oi; - ldst->addrlo_reg = addrlo; - ldst->addrhi_reg = addrhi; + ldst->addrlo_reg = addr; /* Load tlb_mask[mmu_idx] and tlb_table[mmu_idx]. */ tcg_out_ld(s, TCG_TYPE_PTR, TCG_TMP0, TCG_AREG0, mask_off); @@ -1254,11 +1252,10 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, /* Extract the TLB index from the address into TMP3. */ if (TCG_TARGET_REG_BITS == 32 || addr_type == TCG_TYPE_I32) { - tcg_out_opc_sa(s, OPC_SRL, TCG_TMP3, addrlo, + tcg_out_opc_sa(s, OPC_SRL, TCG_TMP3, addr, s->page_bits - CPU_TLB_ENTRY_BITS); } else { - tcg_out_dsrl(s, TCG_TMP3, addrlo, - s->page_bits - CPU_TLB_ENTRY_BITS); + tcg_out_dsrl(s, TCG_TMP3, addr, s->page_bits - CPU_TLB_ENTRY_BITS); } tcg_out_opc_reg(s, OPC_AND, TCG_TMP3, TCG_TMP3, TCG_TMP0); @@ -1288,48 +1285,35 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, tcg_out_opc_imm(s, (TCG_TARGET_REG_BITS == 32 || addr_type == TCG_TYPE_I32 ? OPC_ADDIU : OPC_DADDIU), - TCG_TMP2, addrlo, s_mask - a_mask); + TCG_TMP2, addr, s_mask - a_mask); tcg_out_opc_reg(s, OPC_AND, TCG_TMP1, TCG_TMP1, TCG_TMP2); } else { - tcg_out_opc_reg(s, OPC_AND, TCG_TMP1, TCG_TMP1, addrlo); + tcg_out_opc_reg(s, OPC_AND, TCG_TMP1, TCG_TMP1, addr); } /* Zero extend a 32-bit guest address for a 64-bit host. */ if (TCG_TARGET_REG_BITS == 64 && addr_type == TCG_TYPE_I32) { - tcg_out_ext32u(s, TCG_TMP2, addrlo); - addrlo = TCG_TMP2; + tcg_out_ext32u(s, TCG_TMP2, addr); + addr = TCG_TMP2; } ldst->label_ptr[0] = s->code_ptr; tcg_out_opc_br(s, OPC_BNE, TCG_TMP1, TCG_TMP0); - /* Load and test the high half tlb comparator. */ - if (TCG_TARGET_REG_BITS == 32 && addr_type != TCG_TYPE_I32) { - /* delay slot */ - tcg_out_ldst(s, OPC_LW, TCG_TMP0, TCG_TMP3, cmp_off + HI_OFF); - - /* Load the tlb addend for the fast path. */ - tcg_out_ld(s, TCG_TYPE_PTR, TCG_TMP3, TCG_TMP3, add_off); - - ldst->label_ptr[1] = s->code_ptr; - tcg_out_opc_br(s, OPC_BNE, addrhi, TCG_TMP0); - } - /* delay slot */ base = TCG_TMP3; - tcg_out_opc_reg(s, ALIAS_PADD, base, TCG_TMP3, addrlo); + tcg_out_opc_reg(s, ALIAS_PADD, base, TCG_TMP3, addr); } else { if (a_mask && (use_mips32r6_instructions || a_bits != s_bits)) { ldst = new_ldst_label(s); ldst->is_ld = is_ld; ldst->oi = oi; - ldst->addrlo_reg = addrlo; - ldst->addrhi_reg = addrhi; + ldst->addrlo_reg = addr; /* We are expecting a_bits to max out at 7, much lower than ANDI. */ tcg_debug_assert(a_bits < 16); - tcg_out_opc_imm(s, OPC_ANDI, TCG_TMP0, addrlo, a_mask); + tcg_out_opc_imm(s, OPC_ANDI, TCG_TMP0, addr, a_mask); ldst->label_ptr[0] = s->code_ptr; if (use_mips32r6_instructions) { @@ -1340,7 +1324,7 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, } } - base = addrlo; + base = addr; if (TCG_TARGET_REG_BITS == 64 && addr_type == TCG_TYPE_I32) { tcg_out_ext32u(s, TCG_REG_A0, base); base = TCG_REG_A0; @@ -1460,14 +1444,13 @@ static void tcg_out_qemu_ld_unalign(TCGContext *s, TCGReg lo, TCGReg hi, } static void tcg_out_qemu_ld(TCGContext *s, TCGReg datalo, TCGReg datahi, - TCGReg addrlo, TCGReg addrhi, - MemOpIdx oi, TCGType data_type) + TCGReg addr, MemOpIdx oi, TCGType data_type) { MemOp opc = get_memop(oi); TCGLabelQemuLdst *ldst; HostAddress h; - ldst = prepare_host_addr(s, &h, addrlo, addrhi, oi, true); + ldst = prepare_host_addr(s, &h, addr, oi, true); if (use_mips32r6_instructions || h.aa.align >= (opc & MO_SIZE)) { tcg_out_qemu_ld_direct(s, datalo, datahi, h.base, opc, data_type); @@ -1547,14 +1530,13 @@ static void tcg_out_qemu_st_unalign(TCGContext *s, TCGReg lo, TCGReg hi, } static void tcg_out_qemu_st(TCGContext *s, TCGReg datalo, TCGReg datahi, - TCGReg addrlo, TCGReg addrhi, - MemOpIdx oi, TCGType data_type) + TCGReg addr, MemOpIdx oi, TCGType data_type) { MemOp opc = get_memop(oi); TCGLabelQemuLdst *ldst; HostAddress h; - ldst = prepare_host_addr(s, &h, addrlo, addrhi, oi, false); + ldst = prepare_host_addr(s, &h, addr, oi, false); if (use_mips32r6_instructions || h.aa.align >= (opc & MO_SIZE)) { tcg_out_qemu_st_direct(s, datalo, datahi, h.base, opc); @@ -2096,24 +2078,24 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, TCGType type, break; case INDEX_op_qemu_ld_i32: - tcg_out_qemu_ld(s, a0, 0, a1, 0, a2, TCG_TYPE_I32); + tcg_out_qemu_ld(s, a0, 0, a1, a2, TCG_TYPE_I32); break; case INDEX_op_qemu_ld_i64: if (TCG_TARGET_REG_BITS == 64) { - tcg_out_qemu_ld(s, a0, 0, a1, 0, a2, TCG_TYPE_I64); + tcg_out_qemu_ld(s, a0, 0, a1, a2, TCG_TYPE_I64); } else { - tcg_out_qemu_ld(s, a0, a1, a2, 0, args[3], TCG_TYPE_I64); + tcg_out_qemu_ld(s, a0, a1, a2, args[3], TCG_TYPE_I64); } break; case INDEX_op_qemu_st_i32: - tcg_out_qemu_st(s, a0, 0, a1, 0, a2, TCG_TYPE_I32); + tcg_out_qemu_st(s, a0, 0, a1, a2, TCG_TYPE_I32); break; case INDEX_op_qemu_st_i64: if (TCG_TARGET_REG_BITS == 64) { - tcg_out_qemu_st(s, a0, 0, a1, 0, a2, TCG_TYPE_I64); + tcg_out_qemu_st(s, a0, 0, a1, a2, TCG_TYPE_I64); } else { - tcg_out_qemu_st(s, a0, a1, a2, 0, args[3], TCG_TYPE_I64); + tcg_out_qemu_st(s, a0, a1, a2, args[3], TCG_TYPE_I64); } break; From 7a967f34660c06f8d304fecd118f046fe21cb261 Mon Sep 17 00:00:00 2001 From: Richard Henderson Date: Tue, 4 Feb 2025 17:49:19 -0800 Subject: [PATCH 08/28] tcg/ppc: Drop addrhi from prepare_host_addr MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The guest address will now always fit in one register. Reviewed-by: Philippe Mathieu-Daudé Signed-off-by: Richard Henderson --- tcg/ppc/tcg-target.c.inc | 75 ++++++++++++---------------------------- 1 file changed, 23 insertions(+), 52 deletions(-) diff --git a/tcg/ppc/tcg-target.c.inc b/tcg/ppc/tcg-target.c.inc index 801cb6f3cb..74b93f4b57 100644 --- a/tcg/ppc/tcg-target.c.inc +++ b/tcg/ppc/tcg-target.c.inc @@ -2438,8 +2438,7 @@ bool tcg_target_has_memory_bswap(MemOp memop) * is required and fill in @h with the host address for the fast path. */ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, - TCGReg addrlo, TCGReg addrhi, - MemOpIdx oi, bool is_ld) + TCGReg addr, MemOpIdx oi, bool is_ld) { TCGType addr_type = s->addr_type; TCGLabelQemuLdst *ldst = NULL; @@ -2474,8 +2473,7 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, ldst = new_ldst_label(s); ldst->is_ld = is_ld; ldst->oi = oi; - ldst->addrlo_reg = addrlo; - ldst->addrhi_reg = addrhi; + ldst->addrlo_reg = addr; /* Load tlb_mask[mmu_idx] and tlb_table[mmu_idx]. */ tcg_out_ld(s, TCG_TYPE_PTR, TCG_REG_TMP1, TCG_AREG0, mask_off); @@ -2483,10 +2481,10 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, /* Extract the page index, shifted into place for tlb index. */ if (TCG_TARGET_REG_BITS == 32) { - tcg_out_shri32(s, TCG_REG_R0, addrlo, + tcg_out_shri32(s, TCG_REG_R0, addr, s->page_bits - CPU_TLB_ENTRY_BITS); } else { - tcg_out_shri64(s, TCG_REG_R0, addrlo, + tcg_out_shri64(s, TCG_REG_R0, addr, s->page_bits - CPU_TLB_ENTRY_BITS); } tcg_out32(s, AND | SAB(TCG_REG_TMP1, TCG_REG_TMP1, TCG_REG_R0)); @@ -2534,10 +2532,10 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, if (a_bits < s_bits) { a_bits = s_bits; } - tcg_out_rlw(s, RLWINM, TCG_REG_R0, addrlo, 0, + tcg_out_rlw(s, RLWINM, TCG_REG_R0, addr, 0, (32 - a_bits) & 31, 31 - s->page_bits); } else { - TCGReg t = addrlo; + TCGReg t = addr; /* * If the access is unaligned, we need to make sure we fail if we @@ -2566,30 +2564,8 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, } } - if (TCG_TARGET_REG_BITS == 32 && addr_type != TCG_TYPE_I32) { - /* Low part comparison into cr7. */ - tcg_out_cmp(s, TCG_COND_EQ, TCG_REG_R0, TCG_REG_TMP2, - 0, 7, TCG_TYPE_I32); - - /* Load the high part TLB comparator into TMP2. */ - tcg_out_ld(s, TCG_TYPE_I32, TCG_REG_TMP2, TCG_REG_TMP1, - cmp_off + 4 * !HOST_BIG_ENDIAN); - - /* Load addend, deferred for this case. */ - tcg_out_ld(s, TCG_TYPE_PTR, TCG_REG_TMP1, TCG_REG_TMP1, - offsetof(CPUTLBEntry, addend)); - - /* High part comparison into cr6. */ - tcg_out_cmp(s, TCG_COND_EQ, addrhi, TCG_REG_TMP2, - 0, 6, TCG_TYPE_I32); - - /* Combine comparisons into cr0. */ - tcg_out32(s, CRAND | BT(0, CR_EQ) | BA(6, CR_EQ) | BB(7, CR_EQ)); - } else { - /* Full comparison into cr0. */ - tcg_out_cmp(s, TCG_COND_EQ, TCG_REG_R0, TCG_REG_TMP2, - 0, 0, addr_type); - } + /* Full comparison into cr0. */ + tcg_out_cmp(s, TCG_COND_EQ, TCG_REG_R0, TCG_REG_TMP2, 0, 0, addr_type); /* Load a pointer into the current opcode w/conditional branch-link. */ ldst->label_ptr[0] = s->code_ptr; @@ -2601,12 +2577,11 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, ldst = new_ldst_label(s); ldst->is_ld = is_ld; ldst->oi = oi; - ldst->addrlo_reg = addrlo; - ldst->addrhi_reg = addrhi; + ldst->addrlo_reg = addr; /* We are expecting a_bits to max out at 7, much lower than ANDI. */ tcg_debug_assert(a_bits < 16); - tcg_out32(s, ANDI | SAI(addrlo, TCG_REG_R0, (1 << a_bits) - 1)); + tcg_out32(s, ANDI | SAI(addr, TCG_REG_R0, (1 << a_bits) - 1)); ldst->label_ptr[0] = s->code_ptr; tcg_out32(s, BC | BI(0, CR_EQ) | BO_COND_FALSE | LK); @@ -2617,24 +2592,23 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, if (TCG_TARGET_REG_BITS == 64 && addr_type == TCG_TYPE_I32) { /* Zero-extend the guest address for use in the host address. */ - tcg_out_ext32u(s, TCG_REG_TMP2, addrlo); + tcg_out_ext32u(s, TCG_REG_TMP2, addr); h->index = TCG_REG_TMP2; } else { - h->index = addrlo; + h->index = addr; } return ldst; } static void tcg_out_qemu_ld(TCGContext *s, TCGReg datalo, TCGReg datahi, - TCGReg addrlo, TCGReg addrhi, - MemOpIdx oi, TCGType data_type) + TCGReg addr, MemOpIdx oi, TCGType data_type) { MemOp opc = get_memop(oi); TCGLabelQemuLdst *ldst; HostAddress h; - ldst = prepare_host_addr(s, &h, addrlo, addrhi, oi, true); + ldst = prepare_host_addr(s, &h, addr, oi, true); if (TCG_TARGET_REG_BITS == 32 && (opc & MO_SIZE) == MO_64) { if (opc & MO_BSWAP) { @@ -2678,14 +2652,13 @@ static void tcg_out_qemu_ld(TCGContext *s, TCGReg datalo, TCGReg datahi, } static void tcg_out_qemu_st(TCGContext *s, TCGReg datalo, TCGReg datahi, - TCGReg addrlo, TCGReg addrhi, - MemOpIdx oi, TCGType data_type) + TCGReg addr, MemOpIdx oi, TCGType data_type) { MemOp opc = get_memop(oi); TCGLabelQemuLdst *ldst; HostAddress h; - ldst = prepare_host_addr(s, &h, addrlo, addrhi, oi, false); + ldst = prepare_host_addr(s, &h, addr, oi, false); if (TCG_TARGET_REG_BITS == 32 && (opc & MO_SIZE) == MO_64) { if (opc & MO_BSWAP) { @@ -2729,7 +2702,7 @@ static void tcg_out_qemu_ldst_i128(TCGContext *s, TCGReg datalo, TCGReg datahi, uint32_t insn; TCGReg index; - ldst = prepare_host_addr(s, &h, addr_reg, -1, oi, is_ld); + ldst = prepare_host_addr(s, &h, addr_reg, oi, is_ld); /* Compose the final address, as LQ/STQ have no indexing. */ index = h.index; @@ -3309,14 +3282,13 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, TCGType type, break; case INDEX_op_qemu_ld_i32: - tcg_out_qemu_ld(s, args[0], -1, args[1], -1, args[2], TCG_TYPE_I32); + tcg_out_qemu_ld(s, args[0], -1, args[1], args[2], TCG_TYPE_I32); break; case INDEX_op_qemu_ld_i64: if (TCG_TARGET_REG_BITS == 64) { - tcg_out_qemu_ld(s, args[0], -1, args[1], -1, - args[2], TCG_TYPE_I64); + tcg_out_qemu_ld(s, args[0], -1, args[1], args[2], TCG_TYPE_I64); } else { - tcg_out_qemu_ld(s, args[0], args[1], args[2], -1, + tcg_out_qemu_ld(s, args[0], args[1], args[2], args[3], TCG_TYPE_I64); } break; @@ -3326,14 +3298,13 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, TCGType type, break; case INDEX_op_qemu_st_i32: - tcg_out_qemu_st(s, args[0], -1, args[1], -1, args[2], TCG_TYPE_I32); + tcg_out_qemu_st(s, args[0], -1, args[1], args[2], TCG_TYPE_I32); break; case INDEX_op_qemu_st_i64: if (TCG_TARGET_REG_BITS == 64) { - tcg_out_qemu_st(s, args[0], -1, args[1], -1, - args[2], TCG_TYPE_I64); + tcg_out_qemu_st(s, args[0], -1, args[1], args[2], TCG_TYPE_I64); } else { - tcg_out_qemu_st(s, args[0], args[1], args[2], -1, + tcg_out_qemu_st(s, args[0], args[1], args[2], args[3], TCG_TYPE_I64); } break; From 0cd38379a8f47f3bbfb0b0c8419a7de28f8c9b8c Mon Sep 17 00:00:00 2001 From: Richard Henderson Date: Tue, 4 Feb 2025 18:08:19 -0800 Subject: [PATCH 09/28] tcg: Replace addr{lo,hi}_reg with addr_reg in TCGLabelQemuLdst MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit There is now always only one guest address register. Reviewed-by: Philippe Mathieu-Daudé Signed-off-by: Richard Henderson --- tcg/aarch64/tcg-target.c.inc | 4 ++-- tcg/arm/tcg-target.c.inc | 4 ++-- tcg/i386/tcg-target.c.inc | 4 ++-- tcg/loongarch64/tcg-target.c.inc | 4 ++-- tcg/mips/tcg-target.c.inc | 4 ++-- tcg/ppc/tcg-target.c.inc | 4 ++-- tcg/riscv/tcg-target.c.inc | 4 ++-- tcg/s390x/tcg-target.c.inc | 4 ++-- tcg/sparc64/tcg-target.c.inc | 4 ++-- tcg/tcg.c | 18 +++++++++--------- 10 files changed, 27 insertions(+), 27 deletions(-) diff --git a/tcg/aarch64/tcg-target.c.inc b/tcg/aarch64/tcg-target.c.inc index 45dc2c649b..6f383c1592 100644 --- a/tcg/aarch64/tcg-target.c.inc +++ b/tcg/aarch64/tcg-target.c.inc @@ -1775,7 +1775,7 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, ldst = new_ldst_label(s); ldst->is_ld = is_ld; ldst->oi = oi; - ldst->addrlo_reg = addr_reg; + ldst->addr_reg = addr_reg; mask_type = (s->page_bits + s->tlb_dyn_max_bits > 32 ? TCG_TYPE_I64 : TCG_TYPE_I32); @@ -1837,7 +1837,7 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, ldst->is_ld = is_ld; ldst->oi = oi; - ldst->addrlo_reg = addr_reg; + ldst->addr_reg = addr_reg; /* tst addr, #mask */ tcg_out_logicali(s, I3404_ANDSI, 0, TCG_REG_XZR, addr_reg, a_mask); diff --git a/tcg/arm/tcg-target.c.inc b/tcg/arm/tcg-target.c.inc index 93a3ccaf66..83f6eb6099 100644 --- a/tcg/arm/tcg-target.c.inc +++ b/tcg/arm/tcg-target.c.inc @@ -1485,7 +1485,7 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, ldst = new_ldst_label(s); ldst->is_ld = is_ld; ldst->oi = oi; - ldst->addrlo_reg = addr; + ldst->addr_reg = addr; /* Load cpu->neg.tlb.f[mmu_idx].{mask,table} into {r0,r1}. */ QEMU_BUILD_BUG_ON(offsetof(CPUTLBDescFast, mask) != 0); @@ -1552,7 +1552,7 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, ldst = new_ldst_label(s); ldst->is_ld = is_ld; ldst->oi = oi; - ldst->addrlo_reg = addr; + ldst->addr_reg = addr; /* We are expecting alignment to max out at 7 */ tcg_debug_assert(a_mask <= 0xff); diff --git a/tcg/i386/tcg-target.c.inc b/tcg/i386/tcg-target.c.inc index b33fe7fe23..cfea4c496d 100644 --- a/tcg/i386/tcg-target.c.inc +++ b/tcg/i386/tcg-target.c.inc @@ -2201,7 +2201,7 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, ldst = new_ldst_label(s); ldst->is_ld = is_ld; ldst->oi = oi; - ldst->addrlo_reg = addr; + ldst->addr_reg = addr; if (TCG_TARGET_REG_BITS == 64) { ttype = s->addr_type; @@ -2257,7 +2257,7 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, ldst = new_ldst_label(s); ldst->is_ld = is_ld; ldst->oi = oi; - ldst->addrlo_reg = addr; + ldst->addr_reg = addr; /* jne slow_path */ jcc = tcg_out_cmp(s, TCG_COND_TSTNE, addr, a_mask, true, false); diff --git a/tcg/loongarch64/tcg-target.c.inc b/tcg/loongarch64/tcg-target.c.inc index 4f32bf3e97..dd67e8f6bc 100644 --- a/tcg/loongarch64/tcg-target.c.inc +++ b/tcg/loongarch64/tcg-target.c.inc @@ -1010,7 +1010,7 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, ldst = new_ldst_label(s); ldst->is_ld = is_ld; ldst->oi = oi; - ldst->addrlo_reg = addr_reg; + ldst->addr_reg = addr_reg; tcg_out_ld(s, TCG_TYPE_PTR, TCG_REG_TMP0, TCG_AREG0, mask_ofs); tcg_out_ld(s, TCG_TYPE_PTR, TCG_REG_TMP1, TCG_AREG0, table_ofs); @@ -1055,7 +1055,7 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, ldst->is_ld = is_ld; ldst->oi = oi; - ldst->addrlo_reg = addr_reg; + ldst->addr_reg = addr_reg; /* * Without micro-architecture details, we don't know which of diff --git a/tcg/mips/tcg-target.c.inc b/tcg/mips/tcg-target.c.inc index 153ce1f3c3..d744b853cd 100644 --- a/tcg/mips/tcg-target.c.inc +++ b/tcg/mips/tcg-target.c.inc @@ -1244,7 +1244,7 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, ldst = new_ldst_label(s); ldst->is_ld = is_ld; ldst->oi = oi; - ldst->addrlo_reg = addr; + ldst->addr_reg = addr; /* Load tlb_mask[mmu_idx] and tlb_table[mmu_idx]. */ tcg_out_ld(s, TCG_TYPE_PTR, TCG_TMP0, TCG_AREG0, mask_off); @@ -1309,7 +1309,7 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, ldst->is_ld = is_ld; ldst->oi = oi; - ldst->addrlo_reg = addr; + ldst->addr_reg = addr; /* We are expecting a_bits to max out at 7, much lower than ANDI. */ tcg_debug_assert(a_bits < 16); diff --git a/tcg/ppc/tcg-target.c.inc b/tcg/ppc/tcg-target.c.inc index 74b93f4b57..2d16807ec7 100644 --- a/tcg/ppc/tcg-target.c.inc +++ b/tcg/ppc/tcg-target.c.inc @@ -2473,7 +2473,7 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, ldst = new_ldst_label(s); ldst->is_ld = is_ld; ldst->oi = oi; - ldst->addrlo_reg = addr; + ldst->addr_reg = addr; /* Load tlb_mask[mmu_idx] and tlb_table[mmu_idx]. */ tcg_out_ld(s, TCG_TYPE_PTR, TCG_REG_TMP1, TCG_AREG0, mask_off); @@ -2577,7 +2577,7 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, ldst = new_ldst_label(s); ldst->is_ld = is_ld; ldst->oi = oi; - ldst->addrlo_reg = addr; + ldst->addr_reg = addr; /* We are expecting a_bits to max out at 7, much lower than ANDI. */ tcg_debug_assert(a_bits < 16); diff --git a/tcg/riscv/tcg-target.c.inc b/tcg/riscv/tcg-target.c.inc index 55a3398712..689fbea0df 100644 --- a/tcg/riscv/tcg-target.c.inc +++ b/tcg/riscv/tcg-target.c.inc @@ -1727,7 +1727,7 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, TCGReg *pbase, ldst = new_ldst_label(s); ldst->is_ld = is_ld; ldst->oi = oi; - ldst->addrlo_reg = addr_reg; + ldst->addr_reg = addr_reg; init_setting_vtype(s); @@ -1790,7 +1790,7 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, TCGReg *pbase, ldst = new_ldst_label(s); ldst->is_ld = is_ld; ldst->oi = oi; - ldst->addrlo_reg = addr_reg; + ldst->addr_reg = addr_reg; init_setting_vtype(s); diff --git a/tcg/s390x/tcg-target.c.inc b/tcg/s390x/tcg-target.c.inc index 6786e7b316..b2e1cd60ff 100644 --- a/tcg/s390x/tcg-target.c.inc +++ b/tcg/s390x/tcg-target.c.inc @@ -1920,7 +1920,7 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, ldst = new_ldst_label(s); ldst->is_ld = is_ld; ldst->oi = oi; - ldst->addrlo_reg = addr_reg; + ldst->addr_reg = addr_reg; tcg_out_sh64(s, RSY_SRLG, TCG_TMP0, addr_reg, TCG_REG_NONE, s->page_bits - CPU_TLB_ENTRY_BITS); @@ -1974,7 +1974,7 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, ldst = new_ldst_label(s); ldst->is_ld = is_ld; ldst->oi = oi; - ldst->addrlo_reg = addr_reg; + ldst->addr_reg = addr_reg; tcg_debug_assert(a_mask <= 0xffff); tcg_out_insn(s, RI, TMLL, addr_reg, a_mask); diff --git a/tcg/sparc64/tcg-target.c.inc b/tcg/sparc64/tcg-target.c.inc index ea0a3b8692..527af5665d 100644 --- a/tcg/sparc64/tcg-target.c.inc +++ b/tcg/sparc64/tcg-target.c.inc @@ -1127,7 +1127,7 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, ldst = new_ldst_label(s); ldst->is_ld = is_ld; ldst->oi = oi; - ldst->addrlo_reg = addr_reg; + ldst->addr_reg = addr_reg; ldst->label_ptr[0] = s->code_ptr; /* bne,pn %[xi]cc, label0 */ @@ -1147,7 +1147,7 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, ldst = new_ldst_label(s); ldst->is_ld = is_ld; ldst->oi = oi; - ldst->addrlo_reg = addr_reg; + ldst->addr_reg = addr_reg; ldst->label_ptr[0] = s->code_ptr; /* bne,pn %icc, label0 */ diff --git a/tcg/tcg.c b/tcg/tcg.c index fef93b25ff..55cb9b3ac7 100644 --- a/tcg/tcg.c +++ b/tcg/tcg.c @@ -100,8 +100,7 @@ struct TCGLabelQemuLdst { bool is_ld; /* qemu_ld: true, qemu_st: false */ MemOpIdx oi; TCGType type; /* result type of a load */ - TCGReg addrlo_reg; /* reg index for low word of guest virtual addr */ - TCGReg addrhi_reg; /* reg index for high word of guest virtual addr */ + TCGReg addr_reg; /* reg index for guest virtual addr */ TCGReg datalo_reg; /* reg index for low word to be loaded or stored */ TCGReg datahi_reg; /* reg index for high word to be loaded or stored */ const tcg_insn_unit *raddr; /* addr of the next IR of qemu_ld/st IR */ @@ -6061,7 +6060,7 @@ static void tcg_out_ld_helper_args(TCGContext *s, const TCGLabelQemuLdst *ldst, */ tcg_out_helper_add_mov(mov, loc + HOST_BIG_ENDIAN, TCG_TYPE_I32, TCG_TYPE_I32, - ldst->addrlo_reg, -1); + ldst->addr_reg, -1); tcg_out_helper_load_slots(s, 1, mov, parm); tcg_out_helper_load_imm(s, loc[!HOST_BIG_ENDIAN].arg_slot, @@ -6069,7 +6068,7 @@ static void tcg_out_ld_helper_args(TCGContext *s, const TCGLabelQemuLdst *ldst, next_arg += 2; } else { nmov = tcg_out_helper_add_mov(mov, loc, TCG_TYPE_I64, s->addr_type, - ldst->addrlo_reg, ldst->addrhi_reg); + ldst->addr_reg, -1); tcg_out_helper_load_slots(s, nmov, mov, parm); next_arg += nmov; } @@ -6226,21 +6225,22 @@ static void tcg_out_st_helper_args(TCGContext *s, const TCGLabelQemuLdst *ldst, /* Handle addr argument. */ loc = &info->in[next_arg]; - if (TCG_TARGET_REG_BITS == 32 && s->addr_type == TCG_TYPE_I32) { + tcg_debug_assert(s->addr_type <= TCG_TYPE_REG); + if (TCG_TARGET_REG_BITS == 32) { /* - * 32-bit host with 32-bit guest: zero-extend the guest address + * 32-bit host (and thus 32-bit guest): zero-extend the guest address * to 64-bits for the helper by storing the low part. Later, * after we have processed the register inputs, we will load a * zero for the high part. */ tcg_out_helper_add_mov(mov, loc + HOST_BIG_ENDIAN, TCG_TYPE_I32, TCG_TYPE_I32, - ldst->addrlo_reg, -1); + ldst->addr_reg, -1); next_arg += 2; nmov += 1; } else { n = tcg_out_helper_add_mov(mov, loc, TCG_TYPE_I64, s->addr_type, - ldst->addrlo_reg, ldst->addrhi_reg); + ldst->addr_reg, -1); next_arg += n; nmov += n; } @@ -6288,7 +6288,7 @@ static void tcg_out_st_helper_args(TCGContext *s, const TCGLabelQemuLdst *ldst, g_assert_not_reached(); } - if (TCG_TARGET_REG_BITS == 32 && s->addr_type == TCG_TYPE_I32) { + if (TCG_TARGET_REG_BITS == 32) { /* Zero extend the address by loading a zero for the high part. */ loc = &info->in[1 + !HOST_BIG_ENDIAN]; tcg_out_helper_load_imm(s, loc->arg_slot, TCG_TYPE_I32, 0, parm); From 09ac62682b8d2a8bac36d068f63a31331cc6259a Mon Sep 17 00:00:00 2001 From: Richard Henderson Date: Tue, 4 Feb 2025 18:49:40 -0800 Subject: [PATCH 10/28] plugins: Fix qemu_plugin_read_memory_vaddr parameters MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The declaration uses uint64_t for addr. Fixes: 595cd9ce2ec ("plugins: add plugin API to read guest memory") Reviewed-by: Philippe Mathieu-Daudé Signed-off-by: Richard Henderson --- plugins/api.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/plugins/api.c b/plugins/api.c index 4110cfaa23..cf8cdf076a 100644 --- a/plugins/api.c +++ b/plugins/api.c @@ -561,7 +561,7 @@ GArray *qemu_plugin_get_registers(void) return create_register_handles(regs); } -bool qemu_plugin_read_memory_vaddr(vaddr addr, GByteArray *data, size_t len) +bool qemu_plugin_read_memory_vaddr(uint64_t addr, GByteArray *data, size_t len) { g_assert(current_cpu); From 252394c95baddf3d61f95cdd0c4697298de425d4 Mon Sep 17 00:00:00 2001 From: Richard Henderson Date: Tue, 4 Feb 2025 18:51:30 -0800 Subject: [PATCH 11/28] accel/tcg: Fix tlb_set_page_with_attrs, tlb_set_page The declarations use vaddr for size. Signed-off-by: Richard Henderson --- accel/tcg/cputlb.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/accel/tcg/cputlb.c b/accel/tcg/cputlb.c index 17e2251695..75d075d044 100644 --- a/accel/tcg/cputlb.c +++ b/accel/tcg/cputlb.c @@ -1193,7 +1193,7 @@ void tlb_set_page_full(CPUState *cpu, int mmu_idx, void tlb_set_page_with_attrs(CPUState *cpu, vaddr addr, hwaddr paddr, MemTxAttrs attrs, int prot, - int mmu_idx, uint64_t size) + int mmu_idx, vaddr size) { CPUTLBEntryFull full = { .phys_addr = paddr, @@ -1208,7 +1208,7 @@ void tlb_set_page_with_attrs(CPUState *cpu, vaddr addr, void tlb_set_page(CPUState *cpu, vaddr addr, hwaddr paddr, int prot, - int mmu_idx, uint64_t size) + int mmu_idx, vaddr size) { tlb_set_page_with_attrs(cpu, addr, paddr, MEMTXATTRS_UNSPECIFIED, prot, mmu_idx, size); From a0ea8654e525a2bd5c4c197948d3ba179ded90c7 Mon Sep 17 00:00:00 2001 From: Richard Henderson Date: Mon, 17 Feb 2025 08:21:36 -0800 Subject: [PATCH 12/28] target/loongarch: Use VADDR_PRIx for logging pc_next DisasContextBase.pc_next has type vaddr; use the correct log format. Fixes: 85c19af63e7 ("include/exec: Use vaddr in DisasContextBase for virtual addresses") Signed-off-by: Richard Henderson --- target/loongarch/tcg/insn_trans/trans_atomic.c.inc | 2 +- target/loongarch/tcg/translate.c | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/target/loongarch/tcg/insn_trans/trans_atomic.c.inc b/target/loongarch/tcg/insn_trans/trans_atomic.c.inc index 974bc2a70f..3d70d75941 100644 --- a/target/loongarch/tcg/insn_trans/trans_atomic.c.inc +++ b/target/loongarch/tcg/insn_trans/trans_atomic.c.inc @@ -56,7 +56,7 @@ static bool gen_am(DisasContext *ctx, arg_rrr *a, if (a->rd != 0 && (a->rj == a->rd || a->rk == a->rd)) { qemu_log_mask(LOG_GUEST_ERROR, "Warning: source register overlaps destination register" - "in atomic insn at pc=0x" TARGET_FMT_lx "\n", + "in atomic insn at pc=0x%" VADDR_PRIx "\n", ctx->base.pc_next - 4); return false; } diff --git a/target/loongarch/tcg/translate.c b/target/loongarch/tcg/translate.c index 3480f54c71..e59e4ed25b 100644 --- a/target/loongarch/tcg/translate.c +++ b/target/loongarch/tcg/translate.c @@ -289,7 +289,7 @@ static void loongarch_tr_translate_insn(DisasContextBase *dcbase, CPUState *cs) if (!decode(ctx, ctx->opcode)) { qemu_log_mask(LOG_UNIMP, "Error: unknown opcode. " - TARGET_FMT_lx ": 0x%x\n", + "0x%" VADDR_PRIx ": 0x%x\n", ctx->base.pc_next, ctx->opcode); generate_exception(ctx, EXCCODE_INE); } From a630055df39e1960275d0e273af036f794b15662 Mon Sep 17 00:00:00 2001 From: Richard Henderson Date: Tue, 18 Feb 2025 08:26:40 -0800 Subject: [PATCH 13/28] target/mips: Use VADDR_PRIx for logging pc_next DisasContextBase.pc_next has type vaddr; use the correct log format. Fixes: 85c19af63e7 ("include/exec: Use vaddr in DisasContextBase for virtual addresses") Signed-off-by: Richard Henderson --- target/mips/tcg/octeon_translate.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/target/mips/tcg/octeon_translate.c b/target/mips/tcg/octeon_translate.c index e25c4cbaa0..d9eb43716e 100644 --- a/target/mips/tcg/octeon_translate.c +++ b/target/mips/tcg/octeon_translate.c @@ -18,8 +18,8 @@ static bool trans_BBIT(DisasContext *ctx, arg_BBIT *a) TCGv p; if (ctx->hflags & MIPS_HFLAG_BMASK) { - LOG_DISAS("Branch in delay / forbidden slot at PC 0x" - TARGET_FMT_lx "\n", ctx->base.pc_next); + LOG_DISAS("Branch in delay / forbidden slot at PC 0x%" VADDR_PRIx "\n", + ctx->base.pc_next); generate_exception_end(ctx, EXCP_RI); return true; } From a70af12addd9060fdf8f3dbd42b42e3072c3914f Mon Sep 17 00:00:00 2001 From: Richard Henderson Date: Tue, 4 Feb 2025 18:56:07 -0800 Subject: [PATCH 14/28] include/exec: Change vaddr to uintptr_t MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Since we no longer support 64-bit guests on 32-bit hosts, we can use a 32-bit type on a 32-bit host. Reviewed-by: Philippe Mathieu-Daudé Signed-off-by: Richard Henderson --- include/exec/vaddr.h | 16 +++++++++------- 1 file changed, 9 insertions(+), 7 deletions(-) diff --git a/include/exec/vaddr.h b/include/exec/vaddr.h index b9844afc77..28bec632fb 100644 --- a/include/exec/vaddr.h +++ b/include/exec/vaddr.h @@ -6,13 +6,15 @@ /** * vaddr: * Type wide enough to contain any #target_ulong virtual address. + * We do not support 64-bit guest on 32-host and detect at configure time. + * Therefore, a host pointer width will always fit a guest pointer. */ -typedef uint64_t vaddr; -#define VADDR_PRId PRId64 -#define VADDR_PRIu PRIu64 -#define VADDR_PRIo PRIo64 -#define VADDR_PRIx PRIx64 -#define VADDR_PRIX PRIX64 -#define VADDR_MAX UINT64_MAX +typedef uintptr_t vaddr; +#define VADDR_PRId PRIdPTR +#define VADDR_PRIu PRIuPTR +#define VADDR_PRIo PRIoPTR +#define VADDR_PRIx PRIxPTR +#define VADDR_PRIX PRIXPTR +#define VADDR_MAX UINTPTR_MAX #endif From bf455ec50b6fea15b4d2493059365bf94c706273 Mon Sep 17 00:00:00 2001 From: Richard Henderson Date: Tue, 4 Feb 2025 19:34:51 -0800 Subject: [PATCH 15/28] include/exec: Use uintptr_t in CPUTLBEntry MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Since we no longer support 64-bit guests on 32-bit hosts, we can use a 32-bit type on a 32-bit host. This shrinks the size of the structure to 16 bytes on a 32-bit host. Reviewed-by: Philippe Mathieu-Daudé Signed-off-by: Richard Henderson --- accel/tcg/cputlb.c | 21 ++++----------------- include/exec/tlb-common.h | 10 +++++----- tcg/arm/tcg-target.c.inc | 1 - tcg/mips/tcg-target.c.inc | 12 +++++------- tcg/ppc/tcg-target.c.inc | 21 +++++---------------- 5 files changed, 19 insertions(+), 46 deletions(-) diff --git a/accel/tcg/cputlb.c b/accel/tcg/cputlb.c index 75d075d044..ad158050a1 100644 --- a/accel/tcg/cputlb.c +++ b/accel/tcg/cputlb.c @@ -104,22 +104,15 @@ static inline uint64_t tlb_read_idx(const CPUTLBEntry *entry, { /* Do not rearrange the CPUTLBEntry structure members. */ QEMU_BUILD_BUG_ON(offsetof(CPUTLBEntry, addr_read) != - MMU_DATA_LOAD * sizeof(uint64_t)); + MMU_DATA_LOAD * sizeof(uintptr_t)); QEMU_BUILD_BUG_ON(offsetof(CPUTLBEntry, addr_write) != - MMU_DATA_STORE * sizeof(uint64_t)); + MMU_DATA_STORE * sizeof(uintptr_t)); QEMU_BUILD_BUG_ON(offsetof(CPUTLBEntry, addr_code) != - MMU_INST_FETCH * sizeof(uint64_t)); + MMU_INST_FETCH * sizeof(uintptr_t)); -#if TARGET_LONG_BITS == 32 - /* Use qatomic_read, in case of addr_write; only care about low bits. */ - const uint32_t *ptr = (uint32_t *)&entry->addr_idx[access_type]; - ptr += HOST_BIG_ENDIAN; - return qatomic_read(ptr); -#else - const uint64_t *ptr = &entry->addr_idx[access_type]; + const uintptr_t *ptr = &entry->addr_idx[access_type]; /* ofs might correspond to .addr_write, so use qatomic_read */ return qatomic_read(ptr); -#endif } static inline uint64_t tlb_addr_write(const CPUTLBEntry *entry) @@ -899,14 +892,8 @@ static void tlb_reset_dirty_range_locked(CPUTLBEntry *tlb_entry, addr &= TARGET_PAGE_MASK; addr += tlb_entry->addend; if ((addr - start) < length) { -#if TARGET_LONG_BITS == 32 - uint32_t *ptr_write = (uint32_t *)&tlb_entry->addr_write; - ptr_write += HOST_BIG_ENDIAN; - qatomic_set(ptr_write, *ptr_write | TLB_NOTDIRTY); -#else qatomic_set(&tlb_entry->addr_write, tlb_entry->addr_write | TLB_NOTDIRTY); -#endif } } } diff --git a/include/exec/tlb-common.h b/include/exec/tlb-common.h index dc5a5faa0b..03b5a8ffc7 100644 --- a/include/exec/tlb-common.h +++ b/include/exec/tlb-common.h @@ -19,14 +19,14 @@ #ifndef EXEC_TLB_COMMON_H #define EXEC_TLB_COMMON_H 1 -#define CPU_TLB_ENTRY_BITS 5 +#define CPU_TLB_ENTRY_BITS (HOST_LONG_BITS == 32 ? 4 : 5) /* Minimalized TLB entry for use by TCG fast path. */ typedef union CPUTLBEntry { struct { - uint64_t addr_read; - uint64_t addr_write; - uint64_t addr_code; + uintptr_t addr_read; + uintptr_t addr_write; + uintptr_t addr_code; /* * Addend to virtual address to get host address. IO accesses * use the corresponding iotlb value. @@ -37,7 +37,7 @@ typedef union CPUTLBEntry { * Padding to get a power of two size, as well as index * access to addr_{read,write,code}. */ - uint64_t addr_idx[(1 << CPU_TLB_ENTRY_BITS) / sizeof(uint64_t)]; + uintptr_t addr_idx[(1 << CPU_TLB_ENTRY_BITS) / sizeof(uintptr_t)]; } CPUTLBEntry; QEMU_BUILD_BUG_ON(sizeof(CPUTLBEntry) != (1 << CPU_TLB_ENTRY_BITS)); diff --git a/tcg/arm/tcg-target.c.inc b/tcg/arm/tcg-target.c.inc index 83f6eb6099..cec3d761d4 100644 --- a/tcg/arm/tcg-target.c.inc +++ b/tcg/arm/tcg-target.c.inc @@ -1500,7 +1500,6 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, * Add the tlb_table pointer, creating the CPUTLBEntry address in R1. * Load the tlb comparator into R2 and the fast path addend into R1. */ - QEMU_BUILD_BUG_ON(HOST_BIG_ENDIAN); if (cmp_off == 0) { tcg_out_ld32_rwb(s, COND_AL, TCG_REG_R2, TCG_REG_R1, TCG_REG_R0); } else { diff --git a/tcg/mips/tcg-target.c.inc b/tcg/mips/tcg-target.c.inc index d744b853cd..14b3cb1eba 100644 --- a/tcg/mips/tcg-target.c.inc +++ b/tcg/mips/tcg-target.c.inc @@ -1262,18 +1262,16 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, /* Add the tlb_table pointer, creating the CPUTLBEntry address. */ tcg_out_opc_reg(s, ALIAS_PADD, TCG_TMP3, TCG_TMP3, TCG_TMP1); - if (TCG_TARGET_REG_BITS == 32 || addr_type == TCG_TYPE_I32) { - /* Load the (low half) tlb comparator. */ + /* Load the tlb comparator. */ + if (TCG_TARGET_REG_BITS == 64 && addr_type == TCG_TYPE_I32) { tcg_out_ld(s, TCG_TYPE_I32, TCG_TMP0, TCG_TMP3, cmp_off + HOST_BIG_ENDIAN * 4); } else { - tcg_out_ld(s, TCG_TYPE_I64, TCG_TMP0, TCG_TMP3, cmp_off); + tcg_out_ld(s, TCG_TYPE_REG, TCG_TMP0, TCG_TMP3, cmp_off); } - if (TCG_TARGET_REG_BITS == 64 || addr_type == TCG_TYPE_I32) { - /* Load the tlb addend for the fast path. */ - tcg_out_ld(s, TCG_TYPE_PTR, TCG_TMP3, TCG_TMP3, add_off); - } + /* Load the tlb addend for the fast path. */ + tcg_out_ld(s, TCG_TYPE_PTR, TCG_TMP3, TCG_TMP3, add_off); /* * Mask the page bits, keeping the alignment bits to compare against. diff --git a/tcg/ppc/tcg-target.c.inc b/tcg/ppc/tcg-target.c.inc index 2d16807ec7..822925a19b 100644 --- a/tcg/ppc/tcg-target.c.inc +++ b/tcg/ppc/tcg-target.c.inc @@ -2490,27 +2490,16 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, tcg_out32(s, AND | SAB(TCG_REG_TMP1, TCG_REG_TMP1, TCG_REG_R0)); /* - * Load the (low part) TLB comparator into TMP2. + * Load the TLB comparator into TMP2. * For 64-bit host, always load the entire 64-bit slot for simplicity. * We will ignore the high bits with tcg_out_cmp(..., addr_type). */ - if (TCG_TARGET_REG_BITS == 64) { - if (cmp_off == 0) { - tcg_out32(s, LDUX | TAB(TCG_REG_TMP2, - TCG_REG_TMP1, TCG_REG_TMP2)); - } else { - tcg_out32(s, ADD | TAB(TCG_REG_TMP1, - TCG_REG_TMP1, TCG_REG_TMP2)); - tcg_out_ld(s, TCG_TYPE_I64, TCG_REG_TMP2, - TCG_REG_TMP1, cmp_off); - } - } else if (cmp_off == 0 && !HOST_BIG_ENDIAN) { - tcg_out32(s, LWZUX | TAB(TCG_REG_TMP2, - TCG_REG_TMP1, TCG_REG_TMP2)); + if (cmp_off == 0) { + tcg_out32(s, (TCG_TARGET_REG_BITS == 64 ? LDUX : LWZUX) + | TAB(TCG_REG_TMP2, TCG_REG_TMP1, TCG_REG_TMP2)); } else { tcg_out32(s, ADD | TAB(TCG_REG_TMP1, TCG_REG_TMP1, TCG_REG_TMP2)); - tcg_out_ld(s, TCG_TYPE_I32, TCG_REG_TMP2, TCG_REG_TMP1, - cmp_off + 4 * HOST_BIG_ENDIAN); + tcg_out_ld(s, TCG_TYPE_PTR, TCG_REG_TMP2, TCG_REG_TMP1, cmp_off); } /* From 6b8abd244b9355bc840bc14182aae9043f86f2f6 Mon Sep 17 00:00:00 2001 From: Richard Henderson Date: Sun, 9 Feb 2025 16:01:38 -0800 Subject: [PATCH 16/28] tcg: Introduce the 'z' constraint for a hardware zero register MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit For loongarch, mips, riscv and sparc, a zero register is available all the time. For aarch64, register index 31 depends on context: sometimes it is the stack pointer, and sometimes it is the zero register. Introduce a new general-purpose constraint which maps 0 to TCG_REG_ZERO, if defined. This differs from existing constant constraints in that const_arg[*] is recorded as false, indicating that the value is in a register. Reviewed-by: Philippe Mathieu-Daudé Signed-off-by: Richard Henderson --- docs/devel/tcg-ops.rst | 4 +++- include/tcg/tcg.h | 3 ++- tcg/aarch64/tcg-target.h | 2 ++ tcg/loongarch64/tcg-target.h | 2 ++ tcg/mips/tcg-target.h | 2 ++ tcg/riscv/tcg-target.h | 2 ++ tcg/sparc64/tcg-target.h | 3 ++- tcg/tcg.c | 29 ++++++++++++++++++++++------- 8 files changed, 37 insertions(+), 10 deletions(-) diff --git a/docs/devel/tcg-ops.rst b/docs/devel/tcg-ops.rst index 6608a29376..688984fd39 100644 --- a/docs/devel/tcg-ops.rst +++ b/docs/devel/tcg-ops.rst @@ -927,7 +927,9 @@ operation uses a constant input constraint which does not allow all constants, it must also accept registers in order to have a fallback. The constraint '``i``' is defined generically to accept any constant. The constraint '``r``' is not defined generically, but is consistently -used by each backend to indicate all registers. +used by each backend to indicate all registers. If ``TCG_REG_ZERO`` +is defined by the backend, the constraint '``z``' is defined generically +to map constant 0 to the hardware zero register. The movi_i32 and movi_i64 operations must accept any constants. diff --git a/include/tcg/tcg.h b/include/tcg/tcg.h index 1d1d668f52..84d99508b6 100644 --- a/include/tcg/tcg.h +++ b/include/tcg/tcg.h @@ -713,7 +713,8 @@ void tb_target_set_jmp_target(const TranslationBlock *, int, void tcg_set_frame(TCGContext *s, TCGReg reg, intptr_t start, intptr_t size); -#define TCG_CT_CONST 1 /* any constant of register size */ +#define TCG_CT_CONST 1 /* any constant of register size */ +#define TCG_CT_REG_ZERO 2 /* zero, in TCG_REG_ZERO */ typedef struct TCGArgConstraint { unsigned ct : 16; diff --git a/tcg/aarch64/tcg-target.h b/tcg/aarch64/tcg-target.h index 0dd6e1f069..3f3df5176d 100644 --- a/tcg/aarch64/tcg-target.h +++ b/tcg/aarch64/tcg-target.h @@ -45,6 +45,8 @@ typedef enum { TCG_AREG0 = TCG_REG_X19, } TCGReg; +#define TCG_REG_ZERO TCG_REG_XZR + #define TCG_TARGET_NB_REGS 64 #endif /* AARCH64_TCG_TARGET_H */ diff --git a/tcg/loongarch64/tcg-target.h b/tcg/loongarch64/tcg-target.h index 8533284631..6a206fb97e 100644 --- a/tcg/loongarch64/tcg-target.h +++ b/tcg/loongarch64/tcg-target.h @@ -85,4 +85,6 @@ typedef enum { TCG_VEC_TMP0 = TCG_REG_V23, } TCGReg; +#define TCG_REG_ZERO TCG_REG_ZERO + #endif /* LOONGARCH_TCG_TARGET_H */ diff --git a/tcg/mips/tcg-target.h b/tcg/mips/tcg-target.h index 3090acc4f5..bd4ca5f852 100644 --- a/tcg/mips/tcg-target.h +++ b/tcg/mips/tcg-target.h @@ -70,4 +70,6 @@ typedef enum { TCG_AREG0 = TCG_REG_S8, } TCGReg; +#define TCG_REG_ZERO TCG_REG_ZERO + #endif diff --git a/tcg/riscv/tcg-target.h b/tcg/riscv/tcg-target.h index db5f3d8b72..6dc77d944b 100644 --- a/tcg/riscv/tcg-target.h +++ b/tcg/riscv/tcg-target.h @@ -57,4 +57,6 @@ typedef enum { TCG_REG_TMP2 = TCG_REG_T4, } TCGReg; +#define TCG_REG_ZERO TCG_REG_ZERO + #endif diff --git a/tcg/sparc64/tcg-target.h b/tcg/sparc64/tcg-target.h index f7d75d5806..1b9adccd85 100644 --- a/tcg/sparc64/tcg-target.h +++ b/tcg/sparc64/tcg-target.h @@ -64,6 +64,7 @@ typedef enum { TCG_REG_I7, } TCGReg; -#define TCG_AREG0 TCG_REG_I0 +#define TCG_AREG0 TCG_REG_I0 +#define TCG_REG_ZERO TCG_REG_G0 #endif diff --git a/tcg/tcg.c b/tcg/tcg.c index 55cb9b3ac7..e8950df2ad 100644 --- a/tcg/tcg.c +++ b/tcg/tcg.c @@ -3223,6 +3223,11 @@ static void process_constraint_sets(void) case 'i': args_ct[i].ct |= TCG_CT_CONST; break; +#ifdef TCG_REG_ZERO + case 'z': + args_ct[i].ct |= TCG_CT_REG_ZERO; + break; +#endif /* Include all of the target-specific constraints. */ @@ -5074,13 +5079,23 @@ static void tcg_reg_alloc_op(TCGContext *s, const TCGOp *op) arg_ct = &args_ct[i]; ts = arg_temp(arg); - if (ts->val_type == TEMP_VAL_CONST - && tcg_target_const_match(ts->val, arg_ct->ct, ts->type, - op_cond, TCGOP_VECE(op))) { - /* constant is OK for instruction */ - const_args[i] = 1; - new_args[i] = ts->val; - continue; + if (ts->val_type == TEMP_VAL_CONST) { +#ifdef TCG_REG_ZERO + if (ts->val == 0 && (arg_ct->ct & TCG_CT_REG_ZERO)) { + /* Hardware zero register: indicate register via non-const. */ + const_args[i] = 0; + new_args[i] = TCG_REG_ZERO; + continue; + } +#endif + + if (tcg_target_const_match(ts->val, arg_ct->ct, ts->type, + op_cond, TCGOP_VECE(op))) { + /* constant is OK for instruction */ + const_args[i] = 1; + new_args[i] = ts->val; + continue; + } } reg = ts->reg; From 3d5939e57f959a30eaf13a7897e56a2388121cc2 Mon Sep 17 00:00:00 2001 From: Richard Henderson Date: Sun, 9 Feb 2025 16:15:31 -0800 Subject: [PATCH 17/28] tcg/aarch64: Use 'z' constraint Note that 'Z' is still used for addsub2. Signed-off-by: Richard Henderson --- tcg/aarch64/tcg-target-con-set.h | 12 ++++----- tcg/aarch64/tcg-target.c.inc | 46 ++++++++++++++------------------ 2 files changed, 26 insertions(+), 32 deletions(-) diff --git a/tcg/aarch64/tcg-target-con-set.h b/tcg/aarch64/tcg-target-con-set.h index 44fcc1206e..1281e5efc0 100644 --- a/tcg/aarch64/tcg-target-con-set.h +++ b/tcg/aarch64/tcg-target-con-set.h @@ -11,27 +11,27 @@ */ C_O0_I1(r) C_O0_I2(r, rC) -C_O0_I2(rZ, r) +C_O0_I2(rz, r) C_O0_I2(w, r) -C_O0_I3(rZ, rZ, r) +C_O0_I3(rz, rz, r) C_O1_I1(r, r) C_O1_I1(w, r) C_O1_I1(w, w) C_O1_I1(w, wr) -C_O1_I2(r, 0, rZ) +C_O1_I2(r, 0, rz) C_O1_I2(r, r, r) C_O1_I2(r, r, rA) C_O1_I2(r, r, rAL) C_O1_I2(r, r, rC) C_O1_I2(r, r, ri) C_O1_I2(r, r, rL) -C_O1_I2(r, rZ, rZ) +C_O1_I2(r, rz, rz) C_O1_I2(w, 0, w) C_O1_I2(w, w, w) C_O1_I2(w, w, wN) C_O1_I2(w, w, wO) C_O1_I2(w, w, wZ) C_O1_I3(w, w, w, w) -C_O1_I4(r, r, rC, rZ, rZ) +C_O1_I4(r, r, rC, rz, rz) C_O2_I1(r, r, r) -C_O2_I4(r, r, rZ, rZ, rA, rMZ) +C_O2_I4(r, r, rz, rz, rA, rMZ) diff --git a/tcg/aarch64/tcg-target.c.inc b/tcg/aarch64/tcg-target.c.inc index 6f383c1592..4645242d85 100644 --- a/tcg/aarch64/tcg-target.c.inc +++ b/tcg/aarch64/tcg-target.c.inc @@ -2125,10 +2125,6 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, TCGType ext, TCGArg a2 = args[2]; int c2 = const_args[2]; - /* Some operands are defined with "rZ" constraint, a register or - the zero register. These need not actually test args[I] == 0. */ -#define REG0(I) (const_args[I] ? TCG_REG_XZR : (TCGReg)args[I]) - switch (opc) { case INDEX_op_goto_ptr: tcg_out_insn(s, 3207, BR, a0); @@ -2171,18 +2167,18 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, TCGType ext, case INDEX_op_st8_i32: case INDEX_op_st8_i64: - tcg_out_ldst(s, I3312_STRB, REG0(0), a1, a2, 0); + tcg_out_ldst(s, I3312_STRB, a0, a1, a2, 0); break; case INDEX_op_st16_i32: case INDEX_op_st16_i64: - tcg_out_ldst(s, I3312_STRH, REG0(0), a1, a2, 1); + tcg_out_ldst(s, I3312_STRH, a0, a1, a2, 1); break; case INDEX_op_st_i32: case INDEX_op_st32_i64: - tcg_out_ldst(s, I3312_STRW, REG0(0), a1, a2, 2); + tcg_out_ldst(s, I3312_STRW, a0, a1, a2, 2); break; case INDEX_op_st_i64: - tcg_out_ldst(s, I3312_STRX, REG0(0), a1, a2, 3); + tcg_out_ldst(s, I3312_STRX, a0, a1, a2, 3); break; case INDEX_op_add_i32: @@ -2395,7 +2391,7 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, TCGType ext, /* FALLTHRU */ case INDEX_op_movcond_i64: tcg_out_cmp(s, ext, args[5], a1, a2, c2); - tcg_out_insn(s, 3506, CSEL, ext, a0, REG0(3), REG0(4), args[5]); + tcg_out_insn(s, 3506, CSEL, ext, a0, args[3], args[4], args[5]); break; case INDEX_op_qemu_ld_i32: @@ -2404,13 +2400,13 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, TCGType ext, break; case INDEX_op_qemu_st_i32: case INDEX_op_qemu_st_i64: - tcg_out_qemu_st(s, REG0(0), a1, a2, ext); + tcg_out_qemu_st(s, a0, a1, a2, ext); break; case INDEX_op_qemu_ld_i128: tcg_out_qemu_ldst_i128(s, a0, a1, a2, args[3], true); break; case INDEX_op_qemu_st_i128: - tcg_out_qemu_ldst_i128(s, REG0(0), REG0(1), a2, args[3], false); + tcg_out_qemu_ldst_i128(s, a0, a1, a2, args[3], false); break; case INDEX_op_bswap64_i64: @@ -2439,7 +2435,7 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, TCGType ext, case INDEX_op_deposit_i64: case INDEX_op_deposit_i32: - tcg_out_dep(s, ext, a0, REG0(2), args[3], args[4]); + tcg_out_dep(s, ext, a0, a2, args[3], args[4]); break; case INDEX_op_extract_i64: @@ -2459,25 +2455,25 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, TCGType ext, case INDEX_op_extract2_i64: case INDEX_op_extract2_i32: - tcg_out_extr(s, ext, a0, REG0(2), REG0(1), args[3]); + tcg_out_extr(s, ext, a0, a2, a1, args[3]); break; case INDEX_op_add2_i32: - tcg_out_addsub2(s, TCG_TYPE_I32, a0, a1, REG0(2), REG0(3), + tcg_out_addsub2(s, TCG_TYPE_I32, a0, a1, a2, args[3], (int32_t)args[4], args[5], const_args[4], const_args[5], false); break; case INDEX_op_add2_i64: - tcg_out_addsub2(s, TCG_TYPE_I64, a0, a1, REG0(2), REG0(3), args[4], + tcg_out_addsub2(s, TCG_TYPE_I64, a0, a1, a2, args[3], args[4], args[5], const_args[4], const_args[5], false); break; case INDEX_op_sub2_i32: - tcg_out_addsub2(s, TCG_TYPE_I32, a0, a1, REG0(2), REG0(3), + tcg_out_addsub2(s, TCG_TYPE_I32, a0, a1, a2, args[3], (int32_t)args[4], args[5], const_args[4], const_args[5], true); break; case INDEX_op_sub2_i64: - tcg_out_addsub2(s, TCG_TYPE_I64, a0, a1, REG0(2), REG0(3), args[4], + tcg_out_addsub2(s, TCG_TYPE_I64, a0, a1, a2, args[3], args[4], args[5], const_args[4], const_args[5], true); break; @@ -2513,8 +2509,6 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, TCGType ext, default: g_assert_not_reached(); } - -#undef REG0 } static void tcg_out_vec_op(TCGContext *s, TCGOpcode opc, @@ -3010,7 +3004,7 @@ tcg_target_op_def(TCGOpcode op, TCGType type, unsigned flags) case INDEX_op_st16_i64: case INDEX_op_st32_i64: case INDEX_op_st_i64: - return C_O0_I2(rZ, r); + return C_O0_I2(rz, r); case INDEX_op_add_i32: case INDEX_op_add_i64: @@ -3076,7 +3070,7 @@ tcg_target_op_def(TCGOpcode op, TCGType type, unsigned flags) case INDEX_op_movcond_i32: case INDEX_op_movcond_i64: - return C_O1_I4(r, r, rC, rZ, rZ); + return C_O1_I4(r, r, rC, rz, rz); case INDEX_op_qemu_ld_i32: case INDEX_op_qemu_ld_i64: @@ -3085,23 +3079,23 @@ tcg_target_op_def(TCGOpcode op, TCGType type, unsigned flags) return C_O2_I1(r, r, r); case INDEX_op_qemu_st_i32: case INDEX_op_qemu_st_i64: - return C_O0_I2(rZ, r); + return C_O0_I2(rz, r); case INDEX_op_qemu_st_i128: - return C_O0_I3(rZ, rZ, r); + return C_O0_I3(rz, rz, r); case INDEX_op_deposit_i32: case INDEX_op_deposit_i64: - return C_O1_I2(r, 0, rZ); + return C_O1_I2(r, 0, rz); case INDEX_op_extract2_i32: case INDEX_op_extract2_i64: - return C_O1_I2(r, rZ, rZ); + return C_O1_I2(r, rz, rz); case INDEX_op_add2_i32: case INDEX_op_add2_i64: case INDEX_op_sub2_i32: case INDEX_op_sub2_i64: - return C_O2_I4(r, r, rZ, rZ, rA, rMZ); + return C_O2_I4(r, r, rz, rz, rA, rMZ); case INDEX_op_add_vec: case INDEX_op_sub_vec: From 305370e78d31f742d0b5db4809cf00ef3eeea2a2 Mon Sep 17 00:00:00 2001 From: Richard Henderson Date: Sun, 9 Feb 2025 17:01:29 -0800 Subject: [PATCH 18/28] tcg/loongarch64: Use 'z' constraint MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Replace target-specific 'Z' with generic 'z'. Reviewed-by: Philippe Mathieu-Daudé Signed-off-by: Richard Henderson --- tcg/loongarch64/tcg-target-con-set.h | 15 ++++++------- tcg/loongarch64/tcg-target-con-str.h | 1 - tcg/loongarch64/tcg-target.c.inc | 32 ++++++++++++---------------- 3 files changed, 21 insertions(+), 27 deletions(-) diff --git a/tcg/loongarch64/tcg-target-con-set.h b/tcg/loongarch64/tcg-target-con-set.h index cae6c2aad6..8afaee9476 100644 --- a/tcg/loongarch64/tcg-target-con-set.h +++ b/tcg/loongarch64/tcg-target-con-set.h @@ -15,8 +15,8 @@ * tcg-target-con-str.h; the constraint combination is inclusive or. */ C_O0_I1(r) -C_O0_I2(rZ, r) -C_O0_I2(rZ, rZ) +C_O0_I2(rz, r) +C_O0_I2(rz, rz) C_O0_I2(w, r) C_O0_I3(r, r, r) C_O1_I1(r, r) @@ -28,14 +28,13 @@ C_O1_I2(r, r, rI) C_O1_I2(r, r, rJ) C_O1_I2(r, r, rU) C_O1_I2(r, r, rW) -C_O1_I2(r, r, rZ) -C_O1_I2(r, 0, rZ) -C_O1_I2(r, rZ, ri) -C_O1_I2(r, rZ, rJ) -C_O1_I2(r, rZ, rZ) +C_O1_I2(r, 0, rz) +C_O1_I2(r, rz, ri) +C_O1_I2(r, rz, rJ) +C_O1_I2(r, rz, rz) C_O1_I2(w, w, w) C_O1_I2(w, w, wM) C_O1_I2(w, w, wA) C_O1_I3(w, w, w, w) -C_O1_I4(r, rZ, rJ, rZ, rZ) +C_O1_I4(r, rz, rJ, rz, rz) C_N2_I1(r, r, r) diff --git a/tcg/loongarch64/tcg-target-con-str.h b/tcg/loongarch64/tcg-target-con-str.h index 2ba9c135ac..99759120b4 100644 --- a/tcg/loongarch64/tcg-target-con-str.h +++ b/tcg/loongarch64/tcg-target-con-str.h @@ -23,7 +23,6 @@ REGS('w', ALL_VECTOR_REGS) CONST('I', TCG_CT_CONST_S12) CONST('J', TCG_CT_CONST_S32) CONST('U', TCG_CT_CONST_U12) -CONST('Z', TCG_CT_CONST_ZERO) CONST('C', TCG_CT_CONST_C12) CONST('W', TCG_CT_CONST_WSZ) CONST('M', TCG_CT_CONST_VCMP) diff --git a/tcg/loongarch64/tcg-target.c.inc b/tcg/loongarch64/tcg-target.c.inc index dd67e8f6bc..cbd7642b58 100644 --- a/tcg/loongarch64/tcg-target.c.inc +++ b/tcg/loongarch64/tcg-target.c.inc @@ -173,14 +173,13 @@ static TCGReg tcg_target_call_oarg_reg(TCGCallReturnKind kind, int slot) #define TCG_GUEST_BASE_REG TCG_REG_S1 -#define TCG_CT_CONST_ZERO 0x100 -#define TCG_CT_CONST_S12 0x200 -#define TCG_CT_CONST_S32 0x400 -#define TCG_CT_CONST_U12 0x800 -#define TCG_CT_CONST_C12 0x1000 -#define TCG_CT_CONST_WSZ 0x2000 -#define TCG_CT_CONST_VCMP 0x4000 -#define TCG_CT_CONST_VADD 0x8000 +#define TCG_CT_CONST_S12 0x100 +#define TCG_CT_CONST_S32 0x200 +#define TCG_CT_CONST_U12 0x400 +#define TCG_CT_CONST_C12 0x800 +#define TCG_CT_CONST_WSZ 0x1000 +#define TCG_CT_CONST_VCMP 0x2000 +#define TCG_CT_CONST_VADD 0x4000 #define ALL_GENERAL_REGS MAKE_64BIT_MASK(0, 32) #define ALL_VECTOR_REGS MAKE_64BIT_MASK(32, 32) @@ -197,9 +196,6 @@ static bool tcg_target_const_match(int64_t val, int ct, if (ct & TCG_CT_CONST) { return true; } - if ((ct & TCG_CT_CONST_ZERO) && val == 0) { - return true; - } if ((ct & TCG_CT_CONST_S12) && val == sextreg(val, 0, 12)) { return true; } @@ -2229,7 +2225,7 @@ tcg_target_op_def(TCGOpcode op, TCGType type, unsigned flags) case INDEX_op_st_i64: case INDEX_op_qemu_st_i32: case INDEX_op_qemu_st_i64: - return C_O0_I2(rZ, r); + return C_O0_I2(rz, r); case INDEX_op_qemu_ld_i128: return C_N2_I1(r, r, r); @@ -2239,7 +2235,7 @@ tcg_target_op_def(TCGOpcode op, TCGType type, unsigned flags) case INDEX_op_brcond_i32: case INDEX_op_brcond_i64: - return C_O0_I2(rZ, rZ); + return C_O0_I2(rz, rz); case INDEX_op_ext8s_i32: case INDEX_op_ext8s_i64: @@ -2332,14 +2328,14 @@ tcg_target_op_def(TCGOpcode op, TCGType type, unsigned flags) case INDEX_op_deposit_i32: case INDEX_op_deposit_i64: /* Must deposit into the same register as input */ - return C_O1_I2(r, 0, rZ); + return C_O1_I2(r, 0, rz); case INDEX_op_sub_i32: case INDEX_op_setcond_i32: - return C_O1_I2(r, rZ, ri); + return C_O1_I2(r, rz, ri); case INDEX_op_sub_i64: case INDEX_op_setcond_i64: - return C_O1_I2(r, rZ, rJ); + return C_O1_I2(r, rz, rJ); case INDEX_op_mul_i32: case INDEX_op_mul_i64: @@ -2355,11 +2351,11 @@ tcg_target_op_def(TCGOpcode op, TCGType type, unsigned flags) case INDEX_op_rem_i64: case INDEX_op_remu_i32: case INDEX_op_remu_i64: - return C_O1_I2(r, rZ, rZ); + return C_O1_I2(r, rz, rz); case INDEX_op_movcond_i32: case INDEX_op_movcond_i64: - return C_O1_I4(r, rZ, rJ, rZ, rZ); + return C_O1_I4(r, rz, rJ, rz, rz); case INDEX_op_ld_vec: case INDEX_op_dupm_vec: From 065c8f64161b4b1334e4d244dfb4e9d66551e4c2 Mon Sep 17 00:00:00 2001 From: Richard Henderson Date: Sun, 9 Feb 2025 17:21:48 -0800 Subject: [PATCH 19/28] tcg/mips: Use 'z' constraint MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Replace target-specific 'Z' with generic 'z'. Reviewed-by: Philippe Mathieu-Daudé Signed-off-by: Richard Henderson --- tcg/mips/tcg-target-con-set.h | 26 ++++++++++----------- tcg/mips/tcg-target-con-str.h | 1 - tcg/mips/tcg-target.c.inc | 44 ++++++++++++++--------------------- 3 files changed, 31 insertions(+), 40 deletions(-) diff --git a/tcg/mips/tcg-target-con-set.h b/tcg/mips/tcg-target-con-set.h index 864034f468..06ab04cc4d 100644 --- a/tcg/mips/tcg-target-con-set.h +++ b/tcg/mips/tcg-target-con-set.h @@ -10,24 +10,24 @@ * tcg-target-con-str.h; the constraint combination is inclusive or. */ C_O0_I1(r) -C_O0_I2(rZ, r) -C_O0_I2(rZ, rZ) -C_O0_I3(rZ, r, r) -C_O0_I3(rZ, rZ, r) -C_O0_I4(rZ, rZ, rZ, rZ) -C_O0_I4(rZ, rZ, r, r) +C_O0_I2(rz, r) +C_O0_I2(rz, rz) +C_O0_I3(rz, r, r) +C_O0_I3(rz, rz, r) +C_O0_I4(rz, rz, rz, rz) +C_O0_I4(rz, rz, r, r) C_O1_I1(r, r) -C_O1_I2(r, 0, rZ) +C_O1_I2(r, 0, rz) C_O1_I2(r, r, r) C_O1_I2(r, r, ri) C_O1_I2(r, r, rI) C_O1_I2(r, r, rIK) C_O1_I2(r, r, rJ) -C_O1_I2(r, r, rWZ) -C_O1_I2(r, rZ, rN) -C_O1_I2(r, rZ, rZ) -C_O1_I4(r, rZ, rZ, rZ, 0) -C_O1_I4(r, rZ, rZ, rZ, rZ) +C_O1_I2(r, r, rzW) +C_O1_I2(r, rz, rN) +C_O1_I2(r, rz, rz) +C_O1_I4(r, rz, rz, rz, 0) +C_O1_I4(r, rz, rz, rz, rz) C_O2_I1(r, r, r) C_O2_I2(r, r, r, r) -C_O2_I4(r, r, rZ, rZ, rN, rN) +C_O2_I4(r, r, rz, rz, rN, rN) diff --git a/tcg/mips/tcg-target-con-str.h b/tcg/mips/tcg-target-con-str.h index 413c280a7a..dfe2b156df 100644 --- a/tcg/mips/tcg-target-con-str.h +++ b/tcg/mips/tcg-target-con-str.h @@ -19,4 +19,3 @@ CONST('J', TCG_CT_CONST_S16) CONST('K', TCG_CT_CONST_P2M1) CONST('N', TCG_CT_CONST_N16) CONST('W', TCG_CT_CONST_WSZ) -CONST('Z', TCG_CT_CONST_ZERO) diff --git a/tcg/mips/tcg-target.c.inc b/tcg/mips/tcg-target.c.inc index 14b3cb1eba..f8c105ba37 100644 --- a/tcg/mips/tcg-target.c.inc +++ b/tcg/mips/tcg-target.c.inc @@ -184,12 +184,11 @@ static bool patch_reloc(tcg_insn_unit *code_ptr, int type, g_assert_not_reached(); } -#define TCG_CT_CONST_ZERO 0x100 -#define TCG_CT_CONST_U16 0x200 /* Unsigned 16-bit: 0 - 0xffff. */ -#define TCG_CT_CONST_S16 0x400 /* Signed 16-bit: -32768 - 32767 */ -#define TCG_CT_CONST_P2M1 0x800 /* Power of 2 minus 1. */ -#define TCG_CT_CONST_N16 0x1000 /* "Negatable" 16-bit: -32767 - 32767 */ -#define TCG_CT_CONST_WSZ 0x2000 /* word size */ +#define TCG_CT_CONST_U16 0x100 /* Unsigned 16-bit: 0 - 0xffff. */ +#define TCG_CT_CONST_S16 0x200 /* Signed 16-bit: -32768 - 32767 */ +#define TCG_CT_CONST_P2M1 0x400 /* Power of 2 minus 1. */ +#define TCG_CT_CONST_N16 0x800 /* "Negatable" 16-bit: -32767 - 32767 */ +#define TCG_CT_CONST_WSZ 0x1000 /* word size */ #define ALL_GENERAL_REGS 0xffffffffu @@ -204,8 +203,6 @@ static bool tcg_target_const_match(int64_t val, int ct, { if (ct & TCG_CT_CONST) { return 1; - } else if ((ct & TCG_CT_CONST_ZERO) && val == 0) { - return 1; } else if ((ct & TCG_CT_CONST_U16) && val == (uint16_t)val) { return 1; } else if ((ct & TCG_CT_CONST_S16) && val == (int16_t)val) { @@ -1666,11 +1663,6 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, TCGType type, TCGArg a0, a1, a2; int c2; - /* - * Note that many operands use the constraint set "rZ". - * We make use of the fact that 0 is the ZERO register, - * and hence such cases need not check for const_args. - */ a0 = args[0]; a1 = args[1]; a2 = args[2]; @@ -2181,14 +2173,14 @@ tcg_target_op_def(TCGOpcode op, TCGType type, unsigned flags) case INDEX_op_st16_i64: case INDEX_op_st32_i64: case INDEX_op_st_i64: - return C_O0_I2(rZ, r); + return C_O0_I2(rz, r); case INDEX_op_add_i32: case INDEX_op_add_i64: return C_O1_I2(r, r, rJ); case INDEX_op_sub_i32: case INDEX_op_sub_i64: - return C_O1_I2(r, rZ, rN); + return C_O1_I2(r, rz, rN); case INDEX_op_mul_i32: case INDEX_op_mulsh_i32: case INDEX_op_muluh_i32: @@ -2207,7 +2199,7 @@ tcg_target_op_def(TCGOpcode op, TCGType type, unsigned flags) case INDEX_op_remu_i64: case INDEX_op_nor_i64: case INDEX_op_setcond_i64: - return C_O1_I2(r, rZ, rZ); + return C_O1_I2(r, rz, rz); case INDEX_op_muls2_i32: case INDEX_op_mulu2_i32: case INDEX_op_muls2_i64: @@ -2234,35 +2226,35 @@ tcg_target_op_def(TCGOpcode op, TCGType type, unsigned flags) return C_O1_I2(r, r, ri); case INDEX_op_clz_i32: case INDEX_op_clz_i64: - return C_O1_I2(r, r, rWZ); + return C_O1_I2(r, r, rzW); case INDEX_op_deposit_i32: case INDEX_op_deposit_i64: - return C_O1_I2(r, 0, rZ); + return C_O1_I2(r, 0, rz); case INDEX_op_brcond_i32: case INDEX_op_brcond_i64: - return C_O0_I2(rZ, rZ); + return C_O0_I2(rz, rz); case INDEX_op_movcond_i32: case INDEX_op_movcond_i64: return (use_mips32r6_instructions - ? C_O1_I4(r, rZ, rZ, rZ, rZ) - : C_O1_I4(r, rZ, rZ, rZ, 0)); + ? C_O1_I4(r, rz, rz, rz, rz) + : C_O1_I4(r, rz, rz, rz, 0)); case INDEX_op_add2_i32: case INDEX_op_sub2_i32: - return C_O2_I4(r, r, rZ, rZ, rN, rN); + return C_O2_I4(r, r, rz, rz, rN, rN); case INDEX_op_setcond2_i32: - return C_O1_I4(r, rZ, rZ, rZ, rZ); + return C_O1_I4(r, rz, rz, rz, rz); case INDEX_op_brcond2_i32: - return C_O0_I4(rZ, rZ, rZ, rZ); + return C_O0_I4(rz, rz, rz, rz); case INDEX_op_qemu_ld_i32: return C_O1_I1(r, r); case INDEX_op_qemu_st_i32: - return C_O0_I2(rZ, r); + return C_O0_I2(rz, r); case INDEX_op_qemu_ld_i64: return TCG_TARGET_REG_BITS == 64 ? C_O1_I1(r, r) : C_O2_I1(r, r, r); case INDEX_op_qemu_st_i64: - return TCG_TARGET_REG_BITS == 64 ? C_O0_I2(rZ, r) : C_O0_I3(rZ, rZ, r); + return TCG_TARGET_REG_BITS == 64 ? C_O0_I2(rz, r) : C_O0_I3(rz, rz, r); default: return C_NotImplemented; From f466db1e27131e58d1cfac0d7ce2eb5b28ed22a3 Mon Sep 17 00:00:00 2001 From: Richard Henderson Date: Sun, 9 Feb 2025 17:26:28 -0800 Subject: [PATCH 20/28] tcg/riscv: Use 'z' constraint MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Replace target-specific 'Z' with generic 'z'. Reviewed-by: Philippe Mathieu-Daudé Signed-off-by: Richard Henderson --- tcg/riscv/tcg-target-con-set.h | 10 +++++----- tcg/riscv/tcg-target-con-str.h | 1 - tcg/riscv/tcg-target.c.inc | 28 ++++++++++++---------------- 3 files changed, 17 insertions(+), 22 deletions(-) diff --git a/tcg/riscv/tcg-target-con-set.h b/tcg/riscv/tcg-target-con-set.h index 3c4ef44eb0..e92e815491 100644 --- a/tcg/riscv/tcg-target-con-set.h +++ b/tcg/riscv/tcg-target-con-set.h @@ -10,17 +10,17 @@ * tcg-target-con-str.h; the constraint combination is inclusive or. */ C_O0_I1(r) -C_O0_I2(rZ, r) -C_O0_I2(rZ, rZ) +C_O0_I2(rz, r) +C_O0_I2(rz, rz) C_O1_I1(r, r) C_O1_I2(r, r, ri) C_O1_I2(r, r, rI) C_O1_I2(r, r, rJ) -C_O1_I2(r, rZ, rN) -C_O1_I2(r, rZ, rZ) +C_O1_I2(r, rz, rN) +C_O1_I2(r, rz, rz) C_N1_I2(r, r, rM) C_O1_I4(r, r, rI, rM, rM) -C_O2_I4(r, r, rZ, rZ, rM, rM) +C_O2_I4(r, r, rz, rz, rM, rM) C_O0_I2(v, r) C_O1_I1(v, r) C_O1_I1(v, v) diff --git a/tcg/riscv/tcg-target-con-str.h b/tcg/riscv/tcg-target-con-str.h index 089efe96ca..2f9700638c 100644 --- a/tcg/riscv/tcg-target-con-str.h +++ b/tcg/riscv/tcg-target-con-str.h @@ -21,4 +21,3 @@ CONST('K', TCG_CT_CONST_S5) CONST('L', TCG_CT_CONST_CMP_VI) CONST('N', TCG_CT_CONST_N12) CONST('M', TCG_CT_CONST_M12) -CONST('Z', TCG_CT_CONST_ZERO) diff --git a/tcg/riscv/tcg-target.c.inc b/tcg/riscv/tcg-target.c.inc index 689fbea0df..f7e1ca5a56 100644 --- a/tcg/riscv/tcg-target.c.inc +++ b/tcg/riscv/tcg-target.c.inc @@ -112,13 +112,12 @@ static TCGReg tcg_target_call_oarg_reg(TCGCallReturnKind kind, int slot) return TCG_REG_A0 + slot; } -#define TCG_CT_CONST_ZERO 0x100 -#define TCG_CT_CONST_S12 0x200 -#define TCG_CT_CONST_N12 0x400 -#define TCG_CT_CONST_M12 0x800 -#define TCG_CT_CONST_J12 0x1000 -#define TCG_CT_CONST_S5 0x2000 -#define TCG_CT_CONST_CMP_VI 0x4000 +#define TCG_CT_CONST_S12 0x100 +#define TCG_CT_CONST_N12 0x200 +#define TCG_CT_CONST_M12 0x400 +#define TCG_CT_CONST_J12 0x800 +#define TCG_CT_CONST_S5 0x1000 +#define TCG_CT_CONST_CMP_VI 0x2000 #define ALL_GENERAL_REGS MAKE_64BIT_MASK(0, 32) #define ALL_VECTOR_REGS MAKE_64BIT_MASK(32, 32) @@ -391,9 +390,6 @@ static bool tcg_target_const_match(int64_t val, int ct, if (ct & TCG_CT_CONST) { return 1; } - if ((ct & TCG_CT_CONST_ZERO) && val == 0) { - return 1; - } if (type >= TCG_TYPE_V64) { /* Val is replicated by VECE; extract the highest element. */ val >>= (-8 << vece) & 63; @@ -2681,7 +2677,7 @@ tcg_target_op_def(TCGOpcode op, TCGType type, unsigned flags) case INDEX_op_st16_i64: case INDEX_op_st32_i64: case INDEX_op_st_i64: - return C_O0_I2(rZ, r); + return C_O0_I2(rz, r); case INDEX_op_add_i32: case INDEX_op_and_i32: @@ -2707,7 +2703,7 @@ tcg_target_op_def(TCGOpcode op, TCGType type, unsigned flags) case INDEX_op_sub_i32: case INDEX_op_sub_i64: - return C_O1_I2(r, rZ, rN); + return C_O1_I2(r, rz, rN); case INDEX_op_mul_i32: case INDEX_op_mulsh_i32: @@ -2723,7 +2719,7 @@ tcg_target_op_def(TCGOpcode op, TCGType type, unsigned flags) case INDEX_op_divu_i64: case INDEX_op_rem_i64: case INDEX_op_remu_i64: - return C_O1_I2(r, rZ, rZ); + return C_O1_I2(r, rz, rz); case INDEX_op_shl_i32: case INDEX_op_shr_i32: @@ -2745,7 +2741,7 @@ tcg_target_op_def(TCGOpcode op, TCGType type, unsigned flags) case INDEX_op_brcond_i32: case INDEX_op_brcond_i64: - return C_O0_I2(rZ, rZ); + return C_O0_I2(rz, rz); case INDEX_op_movcond_i32: case INDEX_op_movcond_i64: @@ -2755,14 +2751,14 @@ tcg_target_op_def(TCGOpcode op, TCGType type, unsigned flags) case INDEX_op_add2_i64: case INDEX_op_sub2_i32: case INDEX_op_sub2_i64: - return C_O2_I4(r, r, rZ, rZ, rM, rM); + return C_O2_I4(r, r, rz, rz, rM, rM); case INDEX_op_qemu_ld_i32: case INDEX_op_qemu_ld_i64: return C_O1_I1(r, r); case INDEX_op_qemu_st_i32: case INDEX_op_qemu_st_i64: - return C_O0_I2(rZ, r); + return C_O0_I2(rz, r); case INDEX_op_st_vec: return C_O0_I2(v, r); From 1bbcae5adaad2d8f026194002f54913be5ee0933 Mon Sep 17 00:00:00 2001 From: Richard Henderson Date: Sun, 9 Feb 2025 18:11:20 -0800 Subject: [PATCH 21/28] tcg/sparc64: Use 'z' constraint MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Replace target-specific 'Z' with generic 'z'. Reviewed-by: Philippe Mathieu-Daudé Signed-off-by: Richard Henderson --- tcg/sparc64/tcg-target-con-set.h | 12 ++++++------ tcg/sparc64/tcg-target-con-str.h | 1 - tcg/sparc64/tcg-target.c.inc | 17 +++++++---------- 3 files changed, 13 insertions(+), 17 deletions(-) diff --git a/tcg/sparc64/tcg-target-con-set.h b/tcg/sparc64/tcg-target-con-set.h index 434bf25072..61f9fa3d9f 100644 --- a/tcg/sparc64/tcg-target-con-set.h +++ b/tcg/sparc64/tcg-target-con-set.h @@ -10,11 +10,11 @@ * tcg-target-con-str.h; the constraint combination is inclusive or. */ C_O0_I1(r) -C_O0_I2(rZ, r) -C_O0_I2(rZ, rJ) +C_O0_I2(rz, r) +C_O0_I2(rz, rJ) C_O1_I1(r, r) C_O1_I2(r, r, r) -C_O1_I2(r, rZ, rJ) -C_O1_I4(r, rZ, rJ, rI, 0) -C_O2_I2(r, r, rZ, rJ) -C_O2_I4(r, r, rZ, rZ, rJ, rJ) +C_O1_I2(r, rz, rJ) +C_O1_I4(r, rz, rJ, rI, 0) +C_O2_I2(r, r, rz, rJ) +C_O2_I4(r, r, rz, rz, rJ, rJ) diff --git a/tcg/sparc64/tcg-target-con-str.h b/tcg/sparc64/tcg-target-con-str.h index 0577ec4942..2f033b3ac2 100644 --- a/tcg/sparc64/tcg-target-con-str.h +++ b/tcg/sparc64/tcg-target-con-str.h @@ -16,4 +16,3 @@ REGS('r', ALL_GENERAL_REGS) */ CONST('I', TCG_CT_CONST_S11) CONST('J', TCG_CT_CONST_S13) -CONST('Z', TCG_CT_CONST_ZERO) diff --git a/tcg/sparc64/tcg-target.c.inc b/tcg/sparc64/tcg-target.c.inc index 527af5665d..7c722f59a8 100644 --- a/tcg/sparc64/tcg-target.c.inc +++ b/tcg/sparc64/tcg-target.c.inc @@ -76,7 +76,6 @@ static const char * const tcg_target_reg_names[TCG_TARGET_NB_REGS] = { #define TCG_CT_CONST_S11 0x100 #define TCG_CT_CONST_S13 0x200 -#define TCG_CT_CONST_ZERO 0x400 #define ALL_GENERAL_REGS MAKE_64BIT_MASK(0, 32) @@ -340,9 +339,7 @@ static bool tcg_target_const_match(int64_t val, int ct, val = (int32_t)val; } - if ((ct & TCG_CT_CONST_ZERO) && val == 0) { - return 1; - } else if ((ct & TCG_CT_CONST_S11) && check_fit_tl(val, 11)) { + if ((ct & TCG_CT_CONST_S11) && check_fit_tl(val, 11)) { return 1; } else if ((ct & TCG_CT_CONST_S13) && check_fit_tl(val, 13)) { return 1; @@ -1579,7 +1576,7 @@ tcg_target_op_def(TCGOpcode op, TCGType type, unsigned flags) case INDEX_op_st_i64: case INDEX_op_qemu_st_i32: case INDEX_op_qemu_st_i64: - return C_O0_I2(rZ, r); + return C_O0_I2(rz, r); case INDEX_op_add_i32: case INDEX_op_add_i64: @@ -1611,22 +1608,22 @@ tcg_target_op_def(TCGOpcode op, TCGType type, unsigned flags) case INDEX_op_setcond_i64: case INDEX_op_negsetcond_i32: case INDEX_op_negsetcond_i64: - return C_O1_I2(r, rZ, rJ); + return C_O1_I2(r, rz, rJ); case INDEX_op_brcond_i32: case INDEX_op_brcond_i64: - return C_O0_I2(rZ, rJ); + return C_O0_I2(rz, rJ); case INDEX_op_movcond_i32: case INDEX_op_movcond_i64: - return C_O1_I4(r, rZ, rJ, rI, 0); + return C_O1_I4(r, rz, rJ, rI, 0); case INDEX_op_add2_i32: case INDEX_op_add2_i64: case INDEX_op_sub2_i32: case INDEX_op_sub2_i64: - return C_O2_I4(r, r, rZ, rZ, rJ, rJ); + return C_O2_I4(r, r, rz, rz, rJ, rJ); case INDEX_op_mulu2_i32: case INDEX_op_muls2_i32: - return C_O2_I2(r, r, rZ, rJ); + return C_O2_I2(r, r, rz, rJ); case INDEX_op_muluh_i64: return C_O1_I2(r, r, r); From 4b7b20a3b72c5000ea71bef505c16e6e628268b6 Mon Sep 17 00:00:00 2001 From: Fabiano Rosas Date: Thu, 13 Feb 2025 11:35:58 -0300 Subject: [PATCH 22/28] elfload: Fix alignment when unmapping excess reservation When complying with the alignment requested in the ELF and unmapping the excess reservation, having align_end not aligned to the guest page causes the unmap to be rejected by the alignment check at target_munmap and later brk adjustments hit an EEXIST. Fix by aligning the start of region to be unmapped. Fixes: c81d1fafa6 ("linux-user: Honor elf alignment when placing images") Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1913 Signed-off-by: Fabiano Rosas [rth: Align load_end as well.] Signed-off-by: Richard Henderson Message-ID: <20250213143558.10504-1-farosas@suse.de> --- linux-user/elfload.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/linux-user/elfload.c b/linux-user/elfload.c index a2c152e5ad..8799e4ea27 100644 --- a/linux-user/elfload.c +++ b/linux-user/elfload.c @@ -3351,8 +3351,8 @@ static void load_elf_image(const char *image_name, const ImageSource *src, if (align_size != reserve_size) { abi_ulong align_addr = ROUND_UP(load_addr, align); - abi_ulong align_end = align_addr + reserve_size; - abi_ulong load_end = load_addr + align_size; + abi_ulong align_end = TARGET_PAGE_ALIGN(align_addr + reserve_size); + abi_ulong load_end = TARGET_PAGE_ALIGN(load_addr + align_size); if (align_addr != load_addr) { target_munmap(load_addr, align_addr - load_addr); From 513823e7521a09ed7ad1e32e6454bac3b2cbf52d Mon Sep 17 00:00:00 2001 From: Andreas Schwab Date: Mon, 10 Feb 2025 13:59:34 +0100 Subject: [PATCH 23/28] linux-user: Move TARGET_SA_RESTORER out of generic/signal.h MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit SA_RESTORER and the associated sa_restorer field of struct sigaction are an obsolete feature, not expected to be used by future architectures. They are also absent on RISC-V, LoongArch, Hexagon and OpenRISC, but defined due to their use of generic/signal.h. This leads to corrupted data and out-of-bounds accesses. Move the definition of TARGET_SA_RESTORER out of generic/signal.h into the target_signal.h files that need it. Note that m68k has the sa_restorer field, but does not use it and does not define SA_RESTORER. Reported-by: Thomas Weißschuh Signed-off-by: Andreas Schwab Reviewed-by: Thomas Weißschuh Reviewed-by: Richard Henderson Signed-off-by: Richard Henderson Message-ID: --- linux-user/aarch64/target_signal.h | 2 ++ linux-user/arm/target_signal.h | 2 ++ linux-user/generic/signal.h | 1 - linux-user/i386/target_signal.h | 2 ++ linux-user/m68k/target_signal.h | 1 + linux-user/microblaze/target_signal.h | 2 ++ linux-user/ppc/target_signal.h | 2 ++ linux-user/s390x/target_signal.h | 2 ++ linux-user/sh4/target_signal.h | 2 ++ linux-user/x86_64/target_signal.h | 2 ++ linux-user/xtensa/target_signal.h | 2 ++ 11 files changed, 19 insertions(+), 1 deletion(-) diff --git a/linux-user/aarch64/target_signal.h b/linux-user/aarch64/target_signal.h index 40e399d990..6f66a50bfd 100644 --- a/linux-user/aarch64/target_signal.h +++ b/linux-user/aarch64/target_signal.h @@ -3,6 +3,8 @@ #include "../generic/signal.h" +#define TARGET_SA_RESTORER 0x04000000 + #define TARGET_SEGV_MTEAERR 8 /* Asynchronous ARM MTE error */ #define TARGET_SEGV_MTESERR 9 /* Synchronous ARM MTE exception */ diff --git a/linux-user/arm/target_signal.h b/linux-user/arm/target_signal.h index 0e6351d9f7..ff1810b1fe 100644 --- a/linux-user/arm/target_signal.h +++ b/linux-user/arm/target_signal.h @@ -3,6 +3,8 @@ #include "../generic/signal.h" +#define TARGET_SA_RESTORER 0x04000000 + #define TARGET_ARCH_HAS_SETUP_FRAME #define TARGET_ARCH_HAS_SIGTRAMP_PAGE 1 diff --git a/linux-user/generic/signal.h b/linux-user/generic/signal.h index 6fd05b77bb..b34740258e 100644 --- a/linux-user/generic/signal.h +++ b/linux-user/generic/signal.h @@ -15,7 +15,6 @@ #define TARGET_SA_RESTART 0x10000000 #define TARGET_SA_NODEFER 0x40000000 #define TARGET_SA_RESETHAND 0x80000000 -#define TARGET_SA_RESTORER 0x04000000 #define TARGET_SIGHUP 1 #define TARGET_SIGINT 2 diff --git a/linux-user/i386/target_signal.h b/linux-user/i386/target_signal.h index 9315cba241..eee792ef63 100644 --- a/linux-user/i386/target_signal.h +++ b/linux-user/i386/target_signal.h @@ -3,6 +3,8 @@ #include "../generic/signal.h" +#define TARGET_SA_RESTORER 0x04000000 + #define TARGET_ARCH_HAS_SETUP_FRAME #define TARGET_ARCH_HAS_SIGTRAMP_PAGE 1 diff --git a/linux-user/m68k/target_signal.h b/linux-user/m68k/target_signal.h index 6e0f4b74e3..b05b9303b0 100644 --- a/linux-user/m68k/target_signal.h +++ b/linux-user/m68k/target_signal.h @@ -3,6 +3,7 @@ #include "../generic/signal.h" +#define TARGET_ARCH_HAS_SA_RESTORER 1 #define TARGET_ARCH_HAS_SETUP_FRAME #define TARGET_ARCH_HAS_SIGTRAMP_PAGE 1 diff --git a/linux-user/microblaze/target_signal.h b/linux-user/microblaze/target_signal.h index 7dc5c45f00..ffe4442213 100644 --- a/linux-user/microblaze/target_signal.h +++ b/linux-user/microblaze/target_signal.h @@ -3,6 +3,8 @@ #include "../generic/signal.h" +#define TARGET_SA_RESTORER 0x04000000 + #define TARGET_ARCH_HAS_SIGTRAMP_PAGE 1 #endif /* MICROBLAZE_TARGET_SIGNAL_H */ diff --git a/linux-user/ppc/target_signal.h b/linux-user/ppc/target_signal.h index 5be24e152b..53fae473f3 100644 --- a/linux-user/ppc/target_signal.h +++ b/linux-user/ppc/target_signal.h @@ -3,6 +3,8 @@ #include "../generic/signal.h" +#define TARGET_SA_RESTORER 0x04000000 + #if !defined(TARGET_PPC64) #define TARGET_ARCH_HAS_SETUP_FRAME #endif diff --git a/linux-user/s390x/target_signal.h b/linux-user/s390x/target_signal.h index 41e0e34a55..738e0673f4 100644 --- a/linux-user/s390x/target_signal.h +++ b/linux-user/s390x/target_signal.h @@ -3,6 +3,8 @@ #include "../generic/signal.h" +#define TARGET_SA_RESTORER 0x04000000 + #define TARGET_ARCH_HAS_SETUP_FRAME #define TARGET_ARCH_HAS_SIGTRAMP_PAGE 1 diff --git a/linux-user/sh4/target_signal.h b/linux-user/sh4/target_signal.h index eee6a1a7cd..0bde417fd1 100644 --- a/linux-user/sh4/target_signal.h +++ b/linux-user/sh4/target_signal.h @@ -3,6 +3,8 @@ #include "../generic/signal.h" +#define TARGET_SA_RESTORER 0x04000000 + #define TARGET_ARCH_HAS_SETUP_FRAME #define TARGET_ARCH_HAS_SIGTRAMP_PAGE 1 diff --git a/linux-user/x86_64/target_signal.h b/linux-user/x86_64/target_signal.h index 9d9717406f..0af100c661 100644 --- a/linux-user/x86_64/target_signal.h +++ b/linux-user/x86_64/target_signal.h @@ -3,6 +3,8 @@ #include "../generic/signal.h" +#define TARGET_SA_RESTORER 0x04000000 + /* For x86_64, use of SA_RESTORER is mandatory. */ #define TARGET_ARCH_HAS_SIGTRAMP_PAGE 0 diff --git a/linux-user/xtensa/target_signal.h b/linux-user/xtensa/target_signal.h index e4b1bea5cb..8a198bf8ac 100644 --- a/linux-user/xtensa/target_signal.h +++ b/linux-user/xtensa/target_signal.h @@ -3,6 +3,8 @@ #include "../generic/signal.h" +#define TARGET_SA_RESTORER 0x04000000 + #define TARGET_ARCH_HAS_SIGTRAMP_PAGE 1 #endif From 807c3ebd1e3fc2a1be6cdfc702ccea3fa0d2d9b2 Mon Sep 17 00:00:00 2001 From: Mikael Szreder Date: Wed, 5 Feb 2025 10:03:32 +0100 Subject: [PATCH 24/28] target/sparc: Fix register selection for all F*TOx and FxTO* instructions A bug was introduced in commit 0bba7572d40d which causes the fdtox and fqtox instructions to incorrectly select the destination registers. More information and a test program can be found in issue #2802. Cc: qemu-stable@nongnu.org Fixes: 0bba7572d40d ("target/sparc: Perform DFPREG/QFPREG in decodetree") Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2802 Signed-off-by: Mikael Szreder Acked-by: Artyom Tarasenko [rth: Squash patches together, since the second fixes a typo in the first.] Signed-off-by: Richard Henderson Message-ID: <20250205090333.19626-3-git@miszr.win> --- target/sparc/insns.decode | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) diff --git a/target/sparc/insns.decode b/target/sparc/insns.decode index 989c20b44a..cfcdf6690e 100644 --- a/target/sparc/insns.decode +++ b/target/sparc/insns.decode @@ -321,12 +321,12 @@ FdMULq 10 ..... 110100 ..... 0 0110 1110 ..... @q_d_d FNHADDs 10 ..... 110100 ..... 0 0111 0001 ..... @r_r_r FNHADDd 10 ..... 110100 ..... 0 0111 0010 ..... @d_d_d FNsMULd 10 ..... 110100 ..... 0 0111 1001 ..... @d_r_r -FsTOx 10 ..... 110100 00000 0 1000 0001 ..... @r_r2 -FdTOx 10 ..... 110100 00000 0 1000 0010 ..... @r_d2 -FqTOx 10 ..... 110100 00000 0 1000 0011 ..... @r_q2 -FxTOs 10 ..... 110100 00000 0 1000 0100 ..... @r_r2 -FxTOd 10 ..... 110100 00000 0 1000 1000 ..... @d_r2 -FxTOq 10 ..... 110100 00000 0 1000 1100 ..... @q_r2 +FsTOx 10 ..... 110100 00000 0 1000 0001 ..... @d_r2 +FdTOx 10 ..... 110100 00000 0 1000 0010 ..... @d_d2 +FqTOx 10 ..... 110100 00000 0 1000 0011 ..... @d_q2 +FxTOs 10 ..... 110100 00000 0 1000 0100 ..... @r_d2 +FxTOd 10 ..... 110100 00000 0 1000 1000 ..... @d_d2 +FxTOq 10 ..... 110100 00000 0 1000 1100 ..... @q_d2 FiTOs 10 ..... 110100 00000 0 1100 0100 ..... @r_r2 FdTOs 10 ..... 110100 00000 0 1100 0110 ..... @r_d2 FqTOs 10 ..... 110100 00000 0 1100 0111 ..... @r_q2 From 7a74e468089a58756b438d31a2a9a97f183780d7 Mon Sep 17 00:00:00 2001 From: Mikael Szreder Date: Fri, 14 Feb 2025 08:03:43 +0100 Subject: [PATCH 25/28] target/sparc: Fix gdbstub incorrectly handling registers f32-f62 The gdbstub implementation for the Sparc architecture would incorrectly calculate the the floating point register offset. This resulted in, for example, registers f32 and f34 to point to the same value. The issue was caused by the confusion between even register numbers and even register indexes. For example, the register index of f32 is 64 and f34 is 65. Cc: qemu-stable@nongnu.org Fixes: 30038fd81808 ("target-sparc: Change fpr representation to doubles.") Signed-off-by: Mikael Szreder Reviewed-by: Richard Henderson Signed-off-by: Richard Henderson Message-ID: <20250214070343.11501-1-git@miszr.win> --- target/sparc/gdbstub.c | 18 ++++++++++++++---- 1 file changed, 14 insertions(+), 4 deletions(-) diff --git a/target/sparc/gdbstub.c b/target/sparc/gdbstub.c index ec0036e9ef..134617fb23 100644 --- a/target/sparc/gdbstub.c +++ b/target/sparc/gdbstub.c @@ -79,8 +79,13 @@ int sparc_cpu_gdb_read_register(CPUState *cs, GByteArray *mem_buf, int n) } } if (n < 80) { - /* f32-f62 (double width, even numbers only) */ - return gdb_get_reg64(mem_buf, env->fpr[(n - 32) / 2].ll); + /* f32-f62 (16 double width registers, even register numbers only) + * n == 64: f32 : env->fpr[16] + * n == 65: f34 : env->fpr[17] + * etc... + * n == 79: f62 : env->fpr[31] + */ + return gdb_get_reg64(mem_buf, env->fpr[(n - 64) + 16].ll); } switch (n) { case 80: @@ -173,8 +178,13 @@ int sparc_cpu_gdb_write_register(CPUState *cs, uint8_t *mem_buf, int n) } return 4; } else if (n < 80) { - /* f32-f62 (double width, even numbers only) */ - env->fpr[(n - 32) / 2].ll = tmp; + /* f32-f62 (16 double width registers, even register numbers only) + * n == 64: f32 : env->fpr[16] + * n == 65: f34 : env->fpr[17] + * etc... + * n == 79: f62 : env->fpr[31] + */ + env->fpr[(n - 64) + 16].ll = tmp; } else { switch (n) { case 80: From 172e7644f336dd82cb8d56cfb964477731f34f43 Mon Sep 17 00:00:00 2001 From: Artyom Tarasenko Date: Sun, 9 Feb 2025 22:12:48 +0100 Subject: [PATCH 26/28] target/sparc: fake UltraSPARC T1 PCR and PIC registers Fake access to PCR Performance Control Register and PIC Performance Instrumentation Counter. Ignore writes in privileged mode, and return 0 on reads. This allows booting Tribblix, MilaX and v9os under Niagara target. Signed-off-by: Artyom Tarasenko Reviewed-by: Richard Henderson Signed-off-by: Richard Henderson Message-ID: <20250209211248.50383-1-atar4qemu@gmail.com> --- target/sparc/insns.decode | 7 ++++++- target/sparc/translate.c | 19 +++++++++++++++++++ 2 files changed, 25 insertions(+), 1 deletion(-) diff --git a/target/sparc/insns.decode b/target/sparc/insns.decode index cfcdf6690e..9e39d23273 100644 --- a/target/sparc/insns.decode +++ b/target/sparc/insns.decode @@ -96,7 +96,10 @@ CALL 01 i:s30 RDTICK 10 rd:5 101000 00100 0 0000000000000 RDPC 10 rd:5 101000 00101 0 0000000000000 RDFPRS 10 rd:5 101000 00110 0 0000000000000 - RDASR17 10 rd:5 101000 10001 0 0000000000000 + { + RDASR17 10 rd:5 101000 10001 0 0000000000000 + RDPIC 10 rd:5 101000 10001 0 0000000000000 + } RDGSR 10 rd:5 101000 10011 0 0000000000000 RDSOFTINT 10 rd:5 101000 10110 0 0000000000000 RDTICK_CMPR 10 rd:5 101000 10111 0 0000000000000 @@ -114,6 +117,8 @@ CALL 01 i:s30 WRCCR 10 00010 110000 ..... . ............. @n_r_ri WRASI 10 00011 110000 ..... . ............. @n_r_ri WRFPRS 10 00110 110000 ..... . ............. @n_r_ri + WRPCR 10 10000 110000 01000 0 0000000000000 + WRPIC 10 10001 110000 01000 0 0000000000000 { WRGSR 10 10011 110000 ..... . ............. @n_r_ri WRPOWERDOWN 10 10011 110000 ..... . ............. @n_r_ri diff --git a/target/sparc/translate.c b/target/sparc/translate.c index 7e5c7351cb..bfe63649db 100644 --- a/target/sparc/translate.c +++ b/target/sparc/translate.c @@ -2882,6 +2882,14 @@ static TCGv do_rd_leon3_config(DisasContext *dc, TCGv dst) TRANS(RDASR17, ASR17, do_rd_special, true, a->rd, do_rd_leon3_config) +static TCGv do_rdpic(DisasContext *dc, TCGv dst) +{ + return tcg_constant_tl(0); +} + +TRANS(RDPIC, HYPV, do_rd_special, supervisor(dc), a->rd, do_rdpic) + + static TCGv do_rdccr(DisasContext *dc, TCGv dst) { gen_helper_rdccr(dst, tcg_env); @@ -3315,6 +3323,17 @@ static void do_wrfprs(DisasContext *dc, TCGv src) TRANS(WRFPRS, 64, do_wr_special, a, true, do_wrfprs) +static bool do_priv_nop(DisasContext *dc, bool priv) +{ + if (!priv) { + return raise_priv(dc); + } + return advance_pc(dc); +} + +TRANS(WRPCR, HYPV, do_priv_nop, supervisor(dc)) +TRANS(WRPIC, HYPV, do_priv_nop, supervisor(dc)) + static void do_wrgsr(DisasContext *dc, TCGv src) { gen_trap_ifnofpu(dc); From 175aa36668d6e91157d5e5b092b441f96f46b05e Mon Sep 17 00:00:00 2001 From: Richard Henderson Date: Fri, 10 Jan 2025 10:08:20 -0800 Subject: [PATCH 27/28] tcg/i386: Use tcg_{high,unsigned}_cond in tcg_out_brcond2 MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Eliminate code repetition by using the appropriate helpers. Reviewed-by: Philippe Mathieu-Daudé Signed-off-by: Richard Henderson --- tcg/i386/tcg-target.c.inc | 65 +++++---------------------------------- 1 file changed, 8 insertions(+), 57 deletions(-) diff --git a/tcg/i386/tcg-target.c.inc b/tcg/i386/tcg-target.c.inc index cfea4c496d..33d303a123 100644 --- a/tcg/i386/tcg-target.c.inc +++ b/tcg/i386/tcg-target.c.inc @@ -1658,6 +1658,7 @@ static void tcg_out_brcond2(TCGContext *s, const TCGArg *args, tcg_out_brcond(s, 0, cond, args[1], args[3], const_args[3], label_this, small); break; + case TCG_COND_NE: case TCG_COND_TSTNE: tcg_out_brcond(s, 0, cond, args[0], args[2], const_args[2], @@ -1665,64 +1666,14 @@ static void tcg_out_brcond2(TCGContext *s, const TCGArg *args, tcg_out_brcond(s, 0, cond, args[1], args[3], const_args[3], label_this, small); break; - case TCG_COND_LT: - tcg_out_brcond(s, 0, TCG_COND_LT, args[1], args[3], const_args[3], - label_this, small); - tcg_out_jxx(s, JCC_JNE, label_next, 1); - tcg_out_brcond(s, 0, TCG_COND_LTU, args[0], args[2], const_args[2], - label_this, small); - break; - case TCG_COND_LE: - tcg_out_brcond(s, 0, TCG_COND_LT, args[1], args[3], const_args[3], - label_this, small); - tcg_out_jxx(s, JCC_JNE, label_next, 1); - tcg_out_brcond(s, 0, TCG_COND_LEU, args[0], args[2], const_args[2], - label_this, small); - break; - case TCG_COND_GT: - tcg_out_brcond(s, 0, TCG_COND_GT, args[1], args[3], const_args[3], - label_this, small); - tcg_out_jxx(s, JCC_JNE, label_next, 1); - tcg_out_brcond(s, 0, TCG_COND_GTU, args[0], args[2], const_args[2], - label_this, small); - break; - case TCG_COND_GE: - tcg_out_brcond(s, 0, TCG_COND_GT, args[1], args[3], const_args[3], - label_this, small); - tcg_out_jxx(s, JCC_JNE, label_next, 1); - tcg_out_brcond(s, 0, TCG_COND_GEU, args[0], args[2], const_args[2], - label_this, small); - break; - case TCG_COND_LTU: - tcg_out_brcond(s, 0, TCG_COND_LTU, args[1], args[3], const_args[3], - label_this, small); - tcg_out_jxx(s, JCC_JNE, label_next, 1); - tcg_out_brcond(s, 0, TCG_COND_LTU, args[0], args[2], const_args[2], - label_this, small); - break; - case TCG_COND_LEU: - tcg_out_brcond(s, 0, TCG_COND_LTU, args[1], args[3], const_args[3], - label_this, small); - tcg_out_jxx(s, JCC_JNE, label_next, 1); - tcg_out_brcond(s, 0, TCG_COND_LEU, args[0], args[2], const_args[2], - label_this, small); - break; - case TCG_COND_GTU: - tcg_out_brcond(s, 0, TCG_COND_GTU, args[1], args[3], const_args[3], - label_this, small); - tcg_out_jxx(s, JCC_JNE, label_next, 1); - tcg_out_brcond(s, 0, TCG_COND_GTU, args[0], args[2], const_args[2], - label_this, small); - break; - case TCG_COND_GEU: - tcg_out_brcond(s, 0, TCG_COND_GTU, args[1], args[3], const_args[3], - label_this, small); - tcg_out_jxx(s, JCC_JNE, label_next, 1); - tcg_out_brcond(s, 0, TCG_COND_GEU, args[0], args[2], const_args[2], - label_this, small); - break; + default: - g_assert_not_reached(); + tcg_out_brcond(s, 0, tcg_high_cond(cond), args[1], + args[3], const_args[3], label_this, small); + tcg_out_jxx(s, JCC_JNE, label_next, 1); + tcg_out_brcond(s, 0, tcg_unsigned_cond(cond), args[0], + args[2], const_args[2], label_this, small); + break; } tcg_out_label(s, label_next); } From e726f65867087d86436de05e9f372a86ec1381a6 Mon Sep 17 00:00:00 2001 From: Richard Henderson Date: Fri, 10 Jan 2025 15:01:43 -0800 Subject: [PATCH 28/28] tcg: Remove TCG_TARGET_HAS_{br,set}cond2 from riscv and loongarch64 MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit These defines never should have been added as they were never used. Only 32-bit hosts may have these opcodes and they have them unconditionally. Fixes: 6cb14e4de29 ("tcg/loongarch64: Add the tcg-target.h file") Fixes: fb1f70f3685 ("tcg/riscv: Add the tcg-target.h file") Acked-by: Alistair Francis Reviewed-by: Philippe Mathieu-Daudé Signed-off-by: Richard Henderson --- tcg/loongarch64/tcg-target-has.h | 2 -- tcg/riscv/tcg-target-has.h | 2 -- 2 files changed, 4 deletions(-) diff --git a/tcg/loongarch64/tcg-target-has.h b/tcg/loongarch64/tcg-target-has.h index ac88522eef..188b00799f 100644 --- a/tcg/loongarch64/tcg-target-has.h +++ b/tcg/loongarch64/tcg-target-has.h @@ -37,8 +37,6 @@ #define TCG_TARGET_HAS_clz_i32 1 #define TCG_TARGET_HAS_ctz_i32 1 #define TCG_TARGET_HAS_ctpop_i32 0 -#define TCG_TARGET_HAS_brcond2 0 -#define TCG_TARGET_HAS_setcond2 0 #define TCG_TARGET_HAS_qemu_st8_i32 0 /* 64-bit operations */ diff --git a/tcg/riscv/tcg-target-has.h b/tcg/riscv/tcg-target-has.h index f35f9b31f5..98081084f2 100644 --- a/tcg/riscv/tcg-target-has.h +++ b/tcg/riscv/tcg-target-has.h @@ -37,8 +37,6 @@ #define TCG_TARGET_HAS_clz_i32 (cpuinfo & CPUINFO_ZBB) #define TCG_TARGET_HAS_ctz_i32 (cpuinfo & CPUINFO_ZBB) #define TCG_TARGET_HAS_ctpop_i32 (cpuinfo & CPUINFO_ZBB) -#define TCG_TARGET_HAS_brcond2 1 -#define TCG_TARGET_HAS_setcond2 1 #define TCG_TARGET_HAS_qemu_st8_i32 0 #define TCG_TARGET_HAS_negsetcond_i64 1