diff --git a/configure b/configure index 67beb47aca..e9090a0d29 100755 --- a/configure +++ b/configure @@ -368,7 +368,7 @@ else fi ar="${AR-${cross_prefix}ar}" -as="${AS-${cross_prefix}as}" +ccas="${CCAS-$cc}" cpp="${CPP-$cc -E}" objcopy="${OBJCOPY-${cross_prefix}objcopy}" ld="${LD-${cross_prefix}ld}" @@ -4490,6 +4490,13 @@ if test "$fortify_source" != "no"; then fi fi +################################################# +# clang does not support the 16-bit assembly for roms + +if echo | $ccas -dM -E - | grep __clang__ > /dev/null 2>&1 ; then + ccas="$ccas -fno-integrated-as" +fi + ########################################## # check if struct fsxattr is available via linux/fs.h @@ -5508,7 +5515,7 @@ echo "CXX=$cxx" >> $config_host_mak echo "OBJCC=$objcc" >> $config_host_mak echo "AR=$ar" >> $config_host_mak echo "ARFLAGS=$ARFLAGS" >> $config_host_mak -echo "AS=$as" >> $config_host_mak +echo "CCAS=$ccas" >> $config_host_mak echo "CPP=$cpp" >> $config_host_mak echo "OBJCOPY=$objcopy" >> $config_host_mak echo "LD=$ld" >> $config_host_mak @@ -5981,7 +5988,7 @@ for rom in seabios vgabios ; do config_mak=roms/$rom/config.mak echo "# Automatically generated by configure - do not modify" > $config_mak echo "SRC_PATH=$source_path/roms/$rom" >> $config_mak - echo "AS=$as" >> $config_mak + echo "CCAS=$ccas" >> $config_mak echo "CC=$cc" >> $config_mak echo "BCC=bcc" >> $config_mak echo "CPP=$cpp" >> $config_mak diff --git a/include/exec/cpu-all.h b/include/exec/cpu-all.h index 6a6796d0c9..b6a705982f 100644 --- a/include/exec/cpu-all.h +++ b/include/exec/cpu-all.h @@ -288,14 +288,22 @@ CPUArchState *cpu_copy(CPUArchState *env); #if !defined(CONFIG_USER_ONLY) /* Flags stored in the low bits of the TLB virtual address. These are - defined so that fast path ram access is all zeros. */ + * defined so that fast path ram access is all zeros. + * The flags all must be between TARGET_PAGE_BITS and + * maximum address alignment bit. + */ /* Zero if TLB entry is valid. */ -#define TLB_INVALID_MASK (1 << 3) +#define TLB_INVALID_MASK (1 << (TARGET_PAGE_BITS - 1)) /* Set if TLB entry references a clean RAM page. The iotlb entry will contain the page physical address. */ -#define TLB_NOTDIRTY (1 << 4) +#define TLB_NOTDIRTY (1 << (TARGET_PAGE_BITS - 2)) /* Set if TLB entry is an IO callback. */ -#define TLB_MMIO (1 << 5) +#define TLB_MMIO (1 << (TARGET_PAGE_BITS - 3)) + +/* Use this mask to check interception with an alignment mask + * in a TCG backend. + */ +#define TLB_FLAGS_MASK (TLB_INVALID_MASK | TLB_NOTDIRTY | TLB_MMIO) void dump_exec_info(FILE *f, fprintf_function cpu_fprintf); void dump_opcount_info(FILE *f, fprintf_function cpu_fprintf); diff --git a/rules.mak b/rules.mak index 72c5955954..7d7d83b124 100644 --- a/rules.mak +++ b/rules.mak @@ -68,11 +68,8 @@ LINK = $(call quiet-command, $(LINKPROG) $(QEMU_CFLAGS) $(CFLAGS) $(LDFLAGS) -o $(call process-archive-undefs, $1) \ $(version-obj-y) $(call extract-libs,$1) $(LIBS)," LINK $(TARGET_DIR)$@") -%.asm: %.S - $(call quiet-command,$(CPP) $(QEMU_INCLUDES) $(QEMU_CFLAGS) $(QEMU_DGFLAGS) $(CFLAGS) -o $@ $<," CPP $(TARGET_DIR)$@") - -%.o: %.asm - $(call quiet-command,$(AS) $(ASFLAGS) -o $@ $<," AS $(TARGET_DIR)$@") +%.o: %.S + $(call quiet-command,$(CCAS) $(QEMU_INCLUDES) $(QEMU_CFLAGS) $(QEMU_DGFLAGS) $(CFLAGS) -c -o $@ $<," AS $(TARGET_DIR)$@") %.o: %.cc $(call quiet-command,$(CXX) $(QEMU_INCLUDES) $(QEMU_CXXFLAGS) $(QEMU_DGFLAGS) $(CFLAGS) $($@-cflags) -c -o $@ $<," CXX $(TARGET_DIR)$@") diff --git a/softmmu_template.h b/softmmu_template.h index 208f808f3e..4d378ca630 100644 --- a/softmmu_template.h +++ b/softmmu_template.h @@ -171,20 +171,21 @@ WORD_TYPE helper_le_ld_name(CPUArchState *env, target_ulong addr, unsigned mmu_idx = get_mmuidx(oi); int index = (addr >> TARGET_PAGE_BITS) & (CPU_TLB_SIZE - 1); target_ulong tlb_addr = env->tlb_table[mmu_idx][index].ADDR_READ; + int a_bits = get_alignment_bits(get_memop(oi)); uintptr_t haddr; DATA_TYPE res; /* Adjust the given return address. */ retaddr -= GETPC_ADJ; + if (a_bits > 0 && (addr & ((1 << a_bits) - 1)) != 0) { + cpu_unaligned_access(ENV_GET_CPU(env), addr, READ_ACCESS_TYPE, + mmu_idx, retaddr); + } + /* If the TLB entry is for a different page, reload and try again. */ if ((addr & TARGET_PAGE_MASK) != (tlb_addr & (TARGET_PAGE_MASK | TLB_INVALID_MASK))) { - if ((addr & (DATA_SIZE - 1)) != 0 - && (get_memop(oi) & MO_AMASK) == MO_ALIGN) { - cpu_unaligned_access(ENV_GET_CPU(env), addr, READ_ACCESS_TYPE, - mmu_idx, retaddr); - } if (!VICTIM_TLB_HIT(ADDR_READ)) { tlb_fill(ENV_GET_CPU(env), addr, READ_ACCESS_TYPE, mmu_idx, retaddr); @@ -215,10 +216,6 @@ WORD_TYPE helper_le_ld_name(CPUArchState *env, target_ulong addr, DATA_TYPE res1, res2; unsigned shift; do_unaligned_access: - if ((get_memop(oi) & MO_AMASK) == MO_ALIGN) { - cpu_unaligned_access(ENV_GET_CPU(env), addr, READ_ACCESS_TYPE, - mmu_idx, retaddr); - } addr1 = addr & ~(DATA_SIZE - 1); addr2 = addr1 + DATA_SIZE; /* Note the adjustment at the beginning of the function. @@ -232,13 +229,6 @@ WORD_TYPE helper_le_ld_name(CPUArchState *env, target_ulong addr, return res; } - /* Handle aligned access or unaligned access in the same page. */ - if ((addr & (DATA_SIZE - 1)) != 0 - && (get_memop(oi) & MO_AMASK) == MO_ALIGN) { - cpu_unaligned_access(ENV_GET_CPU(env), addr, READ_ACCESS_TYPE, - mmu_idx, retaddr); - } - haddr = addr + env->tlb_table[mmu_idx][index].addend; #if DATA_SIZE == 1 res = glue(glue(ld, LSUFFIX), _p)((uint8_t *)haddr); @@ -255,20 +245,21 @@ WORD_TYPE helper_be_ld_name(CPUArchState *env, target_ulong addr, unsigned mmu_idx = get_mmuidx(oi); int index = (addr >> TARGET_PAGE_BITS) & (CPU_TLB_SIZE - 1); target_ulong tlb_addr = env->tlb_table[mmu_idx][index].ADDR_READ; + int a_bits = get_alignment_bits(get_memop(oi)); uintptr_t haddr; DATA_TYPE res; /* Adjust the given return address. */ retaddr -= GETPC_ADJ; + if (a_bits > 0 && (addr & ((1 << a_bits) - 1)) != 0) { + cpu_unaligned_access(ENV_GET_CPU(env), addr, READ_ACCESS_TYPE, + mmu_idx, retaddr); + } + /* If the TLB entry is for a different page, reload and try again. */ if ((addr & TARGET_PAGE_MASK) != (tlb_addr & (TARGET_PAGE_MASK | TLB_INVALID_MASK))) { - if ((addr & (DATA_SIZE - 1)) != 0 - && (get_memop(oi) & MO_AMASK) == MO_ALIGN) { - cpu_unaligned_access(ENV_GET_CPU(env), addr, READ_ACCESS_TYPE, - mmu_idx, retaddr); - } if (!VICTIM_TLB_HIT(ADDR_READ)) { tlb_fill(ENV_GET_CPU(env), addr, READ_ACCESS_TYPE, mmu_idx, retaddr); @@ -299,10 +290,6 @@ WORD_TYPE helper_be_ld_name(CPUArchState *env, target_ulong addr, DATA_TYPE res1, res2; unsigned shift; do_unaligned_access: - if ((get_memop(oi) & MO_AMASK) == MO_ALIGN) { - cpu_unaligned_access(ENV_GET_CPU(env), addr, READ_ACCESS_TYPE, - mmu_idx, retaddr); - } addr1 = addr & ~(DATA_SIZE - 1); addr2 = addr1 + DATA_SIZE; /* Note the adjustment at the beginning of the function. @@ -316,13 +303,6 @@ WORD_TYPE helper_be_ld_name(CPUArchState *env, target_ulong addr, return res; } - /* Handle aligned access or unaligned access in the same page. */ - if ((addr & (DATA_SIZE - 1)) != 0 - && (get_memop(oi) & MO_AMASK) == MO_ALIGN) { - cpu_unaligned_access(ENV_GET_CPU(env), addr, READ_ACCESS_TYPE, - mmu_idx, retaddr); - } - haddr = addr + env->tlb_table[mmu_idx][index].addend; res = glue(glue(ld, LSUFFIX), _be_p)((uint8_t *)haddr); return res; @@ -376,19 +356,20 @@ void helper_le_st_name(CPUArchState *env, target_ulong addr, DATA_TYPE val, unsigned mmu_idx = get_mmuidx(oi); int index = (addr >> TARGET_PAGE_BITS) & (CPU_TLB_SIZE - 1); target_ulong tlb_addr = env->tlb_table[mmu_idx][index].addr_write; + int a_bits = get_alignment_bits(get_memop(oi)); uintptr_t haddr; /* Adjust the given return address. */ retaddr -= GETPC_ADJ; + if (a_bits > 0 && (addr & ((1 << a_bits) - 1)) != 0) { + cpu_unaligned_access(ENV_GET_CPU(env), addr, MMU_DATA_STORE, + mmu_idx, retaddr); + } + /* If the TLB entry is for a different page, reload and try again. */ if ((addr & TARGET_PAGE_MASK) != (tlb_addr & (TARGET_PAGE_MASK | TLB_INVALID_MASK))) { - if ((addr & (DATA_SIZE - 1)) != 0 - && (get_memop(oi) & MO_AMASK) == MO_ALIGN) { - cpu_unaligned_access(ENV_GET_CPU(env), addr, MMU_DATA_STORE, - mmu_idx, retaddr); - } if (!VICTIM_TLB_HIT(addr_write)) { tlb_fill(ENV_GET_CPU(env), addr, MMU_DATA_STORE, mmu_idx, retaddr); } @@ -416,10 +397,6 @@ void helper_le_st_name(CPUArchState *env, target_ulong addr, DATA_TYPE val, >= TARGET_PAGE_SIZE)) { int i; do_unaligned_access: - if ((get_memop(oi) & MO_AMASK) == MO_ALIGN) { - cpu_unaligned_access(ENV_GET_CPU(env), addr, MMU_DATA_STORE, - mmu_idx, retaddr); - } /* XXX: not efficient, but simple */ /* Note: relies on the fact that tlb_fill() does not remove the * previous page from the TLB cache. */ @@ -434,13 +411,6 @@ void helper_le_st_name(CPUArchState *env, target_ulong addr, DATA_TYPE val, return; } - /* Handle aligned access or unaligned access in the same page. */ - if ((addr & (DATA_SIZE - 1)) != 0 - && (get_memop(oi) & MO_AMASK) == MO_ALIGN) { - cpu_unaligned_access(ENV_GET_CPU(env), addr, MMU_DATA_STORE, - mmu_idx, retaddr); - } - haddr = addr + env->tlb_table[mmu_idx][index].addend; #if DATA_SIZE == 1 glue(glue(st, SUFFIX), _p)((uint8_t *)haddr, val); @@ -456,19 +426,20 @@ void helper_be_st_name(CPUArchState *env, target_ulong addr, DATA_TYPE val, unsigned mmu_idx = get_mmuidx(oi); int index = (addr >> TARGET_PAGE_BITS) & (CPU_TLB_SIZE - 1); target_ulong tlb_addr = env->tlb_table[mmu_idx][index].addr_write; + int a_bits = get_alignment_bits(get_memop(oi)); uintptr_t haddr; /* Adjust the given return address. */ retaddr -= GETPC_ADJ; + if (a_bits > 0 && (addr & ((1 << a_bits) - 1)) != 0) { + cpu_unaligned_access(ENV_GET_CPU(env), addr, MMU_DATA_STORE, + mmu_idx, retaddr); + } + /* If the TLB entry is for a different page, reload and try again. */ if ((addr & TARGET_PAGE_MASK) != (tlb_addr & (TARGET_PAGE_MASK | TLB_INVALID_MASK))) { - if ((addr & (DATA_SIZE - 1)) != 0 - && (get_memop(oi) & MO_AMASK) == MO_ALIGN) { - cpu_unaligned_access(ENV_GET_CPU(env), addr, MMU_DATA_STORE, - mmu_idx, retaddr); - } if (!VICTIM_TLB_HIT(addr_write)) { tlb_fill(ENV_GET_CPU(env), addr, MMU_DATA_STORE, mmu_idx, retaddr); } @@ -496,10 +467,6 @@ void helper_be_st_name(CPUArchState *env, target_ulong addr, DATA_TYPE val, >= TARGET_PAGE_SIZE)) { int i; do_unaligned_access: - if ((get_memop(oi) & MO_AMASK) == MO_ALIGN) { - cpu_unaligned_access(ENV_GET_CPU(env), addr, MMU_DATA_STORE, - mmu_idx, retaddr); - } /* XXX: not efficient, but simple */ /* Note: relies on the fact that tlb_fill() does not remove the * previous page from the TLB cache. */ @@ -514,13 +481,6 @@ void helper_be_st_name(CPUArchState *env, target_ulong addr, DATA_TYPE val, return; } - /* Handle aligned access or unaligned access in the same page. */ - if ((addr & (DATA_SIZE - 1)) != 0 - && (get_memop(oi) & MO_AMASK) == MO_ALIGN) { - cpu_unaligned_access(ENV_GET_CPU(env), addr, MMU_DATA_STORE, - mmu_idx, retaddr); - } - haddr = addr + env->tlb_table[mmu_idx][index].addend; glue(glue(st, SUFFIX), _be_p)((uint8_t *)haddr, val); } diff --git a/tcg/aarch64/tcg-target.inc.c b/tcg/aarch64/tcg-target.inc.c index 1447f7c216..08b2d031aa 100644 --- a/tcg/aarch64/tcg-target.inc.c +++ b/tcg/aarch64/tcg-target.inc.c @@ -716,6 +716,16 @@ static inline void tcg_out_st(TCGContext *s, TCGType type, TCGReg arg, arg, arg1, arg2); } +static inline bool tcg_out_sti(TCGContext *s, TCGType type, TCGArg val, + TCGReg base, intptr_t ofs) +{ + if (val == 0) { + tcg_out_st(s, type, TCG_REG_XZR, base, ofs); + return true; + } + return false; +} + static inline void tcg_out_bfm(TCGContext *s, TCGType ext, TCGReg rd, TCGReg rn, unsigned int a, unsigned int b) { @@ -1071,19 +1081,20 @@ static void tcg_out_tlb_read(TCGContext *s, TCGReg addr_reg, TCGMemOp opc, int tlb_offset = is_read ? offsetof(CPUArchState, tlb_table[mem_index][0].addr_read) : offsetof(CPUArchState, tlb_table[mem_index][0].addr_write); - int s_mask = (1 << (opc & MO_SIZE)) - 1; + int a_bits = get_alignment_bits(opc); TCGReg base = TCG_AREG0, x3; uint64_t tlb_mask; /* For aligned accesses, we check the first byte and include the alignment bits within the address. For unaligned access, we check that we don't cross pages using the address of the last byte of the access. */ - if ((opc & MO_AMASK) == MO_ALIGN || s_mask == 0) { - tlb_mask = TARGET_PAGE_MASK | s_mask; + if (a_bits >= 0) { + /* A byte access or an alignment check required */ + tlb_mask = TARGET_PAGE_MASK | ((1 << a_bits) - 1); x3 = addr_reg; } else { tcg_out_insn(s, 3401, ADDI, TARGET_LONG_BITS == 64, - TCG_REG_X3, addr_reg, s_mask); + TCG_REG_X3, addr_reg, (1 << (opc & MO_SIZE)) - 1); tlb_mask = TARGET_PAGE_MASK; x3 = TCG_REG_X3; } diff --git a/tcg/arm/tcg-target.inc.c b/tcg/arm/tcg-target.inc.c index f9f54c64c6..172febafdd 100644 --- a/tcg/arm/tcg-target.inc.c +++ b/tcg/arm/tcg-target.inc.c @@ -2046,6 +2046,12 @@ static inline void tcg_out_st(TCGContext *s, TCGType type, TCGReg arg, tcg_out_st32(s, COND_AL, arg, arg1, arg2); } +static inline bool tcg_out_sti(TCGContext *s, TCGType type, TCGArg val, + TCGReg base, intptr_t ofs) +{ + return false; +} + static inline void tcg_out_mov(TCGContext *s, TCGType type, TCGReg ret, TCGReg arg) { diff --git a/tcg/i386/tcg-target.inc.c b/tcg/i386/tcg-target.inc.c index 317484cb5d..6f8cdca756 100644 --- a/tcg/i386/tcg-target.inc.c +++ b/tcg/i386/tcg-target.inc.c @@ -710,12 +710,19 @@ static inline void tcg_out_st(TCGContext *s, TCGType type, TCGReg arg, tcg_out_modrm_offset(s, opc, arg, arg1, arg2); } -static inline void tcg_out_sti(TCGContext *s, TCGType type, TCGReg base, - tcg_target_long ofs, tcg_target_long val) +static bool tcg_out_sti(TCGContext *s, TCGType type, TCGArg val, + TCGReg base, intptr_t ofs) { - int opc = OPC_MOVL_EvIz + (type == TCG_TYPE_I64 ? P_REXW : 0); - tcg_out_modrm_offset(s, opc, 0, base, ofs); + int rexw = 0; + if (TCG_TARGET_REG_BITS == 64 && type == TCG_TYPE_I64) { + if (val != (int32_t)val) { + return false; + } + rexw = P_REXW; + } + tcg_out_modrm_offset(s, OPC_MOVL_EvIz | rexw, 0, base, ofs); tcg_out32(s, val); + return true; } static void tcg_out_shifti(TCGContext *s, int subopc, int reg, int count) @@ -1195,8 +1202,8 @@ static inline void tcg_out_tlb_load(TCGContext *s, TCGReg addrlo, TCGReg addrhi, TCGType ttype = TCG_TYPE_I32; TCGType tlbtype = TCG_TYPE_I32; int trexw = 0, hrexw = 0, tlbrexw = 0; - int s_mask = (1 << (opc & MO_SIZE)) - 1; - bool aligned = (opc & MO_AMASK) == MO_ALIGN || s_mask == 0; + int a_bits = get_alignment_bits(opc); + target_ulong tlb_mask; if (TCG_TARGET_REG_BITS == 64) { if (TARGET_LONG_BITS == 64) { @@ -1213,19 +1220,22 @@ static inline void tcg_out_tlb_load(TCGContext *s, TCGReg addrlo, TCGReg addrhi, } tcg_out_mov(s, tlbtype, r0, addrlo); - if (aligned) { + if (a_bits >= 0) { + /* A byte access or an alignment check required */ tcg_out_mov(s, ttype, r1, addrlo); + tlb_mask = TARGET_PAGE_MASK | ((1 << a_bits) - 1); } else { /* For unaligned access check that we don't cross pages using the page address of the last byte. */ - tcg_out_modrm_offset(s, OPC_LEA + trexw, r1, addrlo, s_mask); + tcg_out_modrm_offset(s, OPC_LEA + trexw, r1, addrlo, + (1 << (opc & MO_SIZE)) - 1); + tlb_mask = TARGET_PAGE_MASK; } tcg_out_shifti(s, SHIFT_SHR + tlbrexw, r0, TARGET_PAGE_BITS - CPU_TLB_ENTRY_BITS); - tgen_arithi(s, ARITH_AND + trexw, r1, - TARGET_PAGE_MASK | (aligned ? s_mask : 0), 0); + tgen_arithi(s, ARITH_AND + trexw, r1, tlb_mask, 0); tgen_arithi(s, ARITH_AND + tlbrexw, r0, (CPU_TLB_SIZE - 1) << CPU_TLB_ENTRY_BITS, 0); @@ -1321,10 +1331,10 @@ static void tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *l) ofs += 4; } - tcg_out_sti(s, TCG_TYPE_I32, TCG_REG_ESP, ofs, oi); + tcg_out_sti(s, TCG_TYPE_I32, oi, TCG_REG_ESP, ofs); ofs += 4; - tcg_out_sti(s, TCG_TYPE_PTR, TCG_REG_ESP, ofs, (uintptr_t)l->raddr); + tcg_out_sti(s, TCG_TYPE_PTR, (uintptr_t)l->raddr, TCG_REG_ESP, ofs); } else { tcg_out_mov(s, TCG_TYPE_PTR, tcg_target_call_iarg_regs[0], TCG_AREG0); /* The second argument is already loaded with addrlo. */ @@ -1413,7 +1423,7 @@ static void tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *l) ofs += 4; } - tcg_out_sti(s, TCG_TYPE_I32, TCG_REG_ESP, ofs, oi); + tcg_out_sti(s, TCG_TYPE_I32, oi, TCG_REG_ESP, ofs); ofs += 4; retaddr = TCG_REG_EAX; diff --git a/tcg/ia64/tcg-target.inc.c b/tcg/ia64/tcg-target.inc.c index 395223e340..c91f39281b 100644 --- a/tcg/ia64/tcg-target.inc.c +++ b/tcg/ia64/tcg-target.inc.c @@ -973,6 +973,16 @@ static inline void tcg_out_st(TCGContext *s, TCGType type, TCGReg arg, } } +static inline bool tcg_out_sti(TCGContext *s, TCGType type, TCGArg val, + TCGReg base, intptr_t ofs) +{ + if (val == 0) { + tcg_out_st(s, type, TCG_REG_R0, base, ofs); + return true; + } + return false; +} + static inline void tcg_out_alu(TCGContext *s, uint64_t opc_a1, uint64_t opc_a3, TCGReg ret, TCGArg arg1, int const_arg1, TCGArg arg2, int const_arg2) diff --git a/tcg/mips/tcg-target.inc.c b/tcg/mips/tcg-target.inc.c index 50e98ea63a..2f9be48139 100644 --- a/tcg/mips/tcg-target.inc.c +++ b/tcg/mips/tcg-target.inc.c @@ -576,6 +576,16 @@ static inline void tcg_out_st(TCGContext *s, TCGType type, TCGReg arg, tcg_out_ldst(s, OPC_SW, arg, arg1, arg2); } +static inline bool tcg_out_sti(TCGContext *s, TCGType type, TCGArg val, + TCGReg base, intptr_t ofs) +{ + if (val == 0) { + tcg_out_st(s, type, TCG_REG_ZERO, base, ofs); + return true; + } + return false; +} + static inline void tcg_out_addi(TCGContext *s, TCGReg reg, TCGArg val) { if (val == (int16_t)val) { diff --git a/tcg/ppc/tcg-target.inc.c b/tcg/ppc/tcg-target.inc.c index da100528ab..eaf1bd9bfd 100644 --- a/tcg/ppc/tcg-target.inc.c +++ b/tcg/ppc/tcg-target.inc.c @@ -857,6 +857,12 @@ static inline void tcg_out_st(TCGContext *s, TCGType type, TCGReg arg, tcg_out_mem_long(s, opi, opx, arg, arg1, arg2); } +static inline bool tcg_out_sti(TCGContext *s, TCGType type, TCGArg val, + TCGReg base, intptr_t ofs) +{ + return false; +} + static void tcg_out_cmp(TCGContext *s, int cond, TCGArg arg1, TCGArg arg2, int const_arg2, int cr, TCGType type) { @@ -1399,6 +1405,7 @@ static TCGReg tcg_out_tlb_read(TCGContext *s, TCGMemOp opc, int add_off = offsetof(CPUArchState, tlb_table[mem_index][0].addend); TCGReg base = TCG_AREG0; TCGMemOp s_bits = opc & MO_SIZE; + int a_bits = get_alignment_bits(opc); /* Extract the page index, shifted into place for tlb index. */ if (TCG_TARGET_REG_BITS == 64) { @@ -1456,14 +1463,17 @@ static TCGReg tcg_out_tlb_read(TCGContext *s, TCGMemOp opc, * the bottom bits and thus trigger a comparison failure on * unaligned accesses */ + if (a_bits < 0) { + a_bits = s_bits; + } tcg_out_rlw(s, RLWINM, TCG_REG_R0, addrlo, 0, - (32 - s_bits) & 31, 31 - TARGET_PAGE_BITS); - } else if (s_bits) { - /* > byte access, we need to handle alignment */ - if ((opc & MO_AMASK) == MO_ALIGN) { + (32 - a_bits) & 31, 31 - TARGET_PAGE_BITS); + } else if (a_bits) { + /* More than byte access, we need to handle alignment */ + if (a_bits > 0) { /* Alignment required by the front-end, same as 32-bits */ tcg_out_rld(s, RLDICL, TCG_REG_R0, addrlo, - 64 - TARGET_PAGE_BITS, TARGET_PAGE_BITS - s_bits); + 64 - TARGET_PAGE_BITS, TARGET_PAGE_BITS - a_bits); tcg_out_rld(s, RLDICL, TCG_REG_R0, TCG_REG_R0, TARGET_PAGE_BITS, 0); } else { /* We support unaligned accesses, we need to make sure we fail diff --git a/tcg/s390/tcg-target.inc.c b/tcg/s390/tcg-target.inc.c index e0a60e618c..5a7495b063 100644 --- a/tcg/s390/tcg-target.inc.c +++ b/tcg/s390/tcg-target.inc.c @@ -798,6 +798,12 @@ static inline void tcg_out_st(TCGContext *s, TCGType type, TCGReg data, } } +static inline bool tcg_out_sti(TCGContext *s, TCGType type, TCGArg val, + TCGReg base, intptr_t ofs) +{ + return false; +} + /* load data from an absolute host address */ static void tcg_out_ld_abs(TCGContext *s, TCGType type, TCGReg dest, void *abs) { @@ -1499,18 +1505,19 @@ QEMU_BUILD_BUG_ON(offsetof(CPUArchState, tlb_table[NB_MMU_MODES - 1][1]) static TCGReg tcg_out_tlb_read(TCGContext* s, TCGReg addr_reg, TCGMemOp opc, int mem_index, bool is_ld) { - int s_mask = (1 << (opc & MO_SIZE)) - 1; + int a_bits = get_alignment_bits(opc); int ofs, a_off; uint64_t tlb_mask; /* For aligned accesses, we check the first byte and include the alignment bits within the address. For unaligned access, we check that we don't cross pages using the address of the last byte of the access. */ - if ((opc & MO_AMASK) == MO_ALIGN || s_mask == 0) { + if (a_bits >= 0) { + /* A byte access or an alignment check required */ a_off = 0; - tlb_mask = TARGET_PAGE_MASK | s_mask; + tlb_mask = TARGET_PAGE_MASK | ((1 << a_bits) - 1); } else { - a_off = s_mask; + a_off = (1 << (opc & MO_SIZE)) - 1; tlb_mask = TARGET_PAGE_MASK; } diff --git a/tcg/sparc/tcg-target.inc.c b/tcg/sparc/tcg-target.inc.c index 9938a5085e..8e98172ca0 100644 --- a/tcg/sparc/tcg-target.inc.c +++ b/tcg/sparc/tcg-target.inc.c @@ -504,6 +504,16 @@ static inline void tcg_out_st(TCGContext *s, TCGType type, TCGReg arg, tcg_out_ldst(s, arg, arg1, arg2, (type == TCG_TYPE_I32 ? STW : STX)); } +static inline bool tcg_out_sti(TCGContext *s, TCGType type, TCGArg val, + TCGReg base, intptr_t ofs) +{ + if (val == 0) { + tcg_out_st(s, type, TCG_REG_G0, base, ofs); + return true; + } + return false; +} + static void tcg_out_ld_ptr(TCGContext *s, TCGReg ret, uintptr_t arg) { tcg_out_movi(s, TCG_TYPE_PTR, ret, arg & ~0x3ff); diff --git a/tcg/tcg-op.c b/tcg/tcg-op.c index 569cdc642e..293b854370 100644 --- a/tcg/tcg-op.c +++ b/tcg/tcg-op.c @@ -1851,6 +1851,9 @@ void tcg_gen_goto_tb(unsigned idx) static inline TCGMemOp tcg_canonicalize_memop(TCGMemOp op, bool is64, bool st) { + /* Trigger the asserts within as early as possible. */ + (void)get_alignment_bits(op); + switch (op & MO_SIZE) { case MO_8: op &= ~MO_BSWAP; diff --git a/tcg/tcg.c b/tcg/tcg.c index 254427b1b7..0c46c43cfa 100644 --- a/tcg/tcg.c +++ b/tcg/tcg.c @@ -108,6 +108,8 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, const TCGArg *args, const int *const_args); static void tcg_out_st(TCGContext *s, TCGType type, TCGReg arg, TCGReg arg1, intptr_t arg2); +static bool tcg_out_sti(TCGContext *s, TCGType type, TCGArg val, + TCGReg base, intptr_t ofs); static void tcg_out_call(TCGContext *s, tcg_insn_unit *target); static int tcg_target_const_match(tcg_target_long val, TCGType type, const TCGArgConstraint *arg_ct); @@ -557,7 +559,7 @@ int tcg_global_mem_new_internal(TCGType type, TCGv_ptr base, ts2->mem_offset = offset + (1 - bigendian) * 4; pstrcpy(buf, sizeof(buf), name); pstrcat(buf, sizeof(buf), "_1"); - ts->name = strdup(buf); + ts2->name = strdup(buf); } else { ts->base_type = type; ts->type = type; @@ -997,6 +999,22 @@ static const char * const ldst_name[] = [MO_BEQ] = "beq", }; +static const char * const alignment_name[(MO_AMASK >> MO_ASHIFT) + 1] = { +#ifdef ALIGNED_ONLY + [MO_UNALN >> MO_ASHIFT] = "un+", + [MO_ALIGN >> MO_ASHIFT] = "", +#else + [MO_UNALN >> MO_ASHIFT] = "", + [MO_ALIGN >> MO_ASHIFT] = "al+", +#endif + [MO_ALIGN_2 >> MO_ASHIFT] = "al2+", + [MO_ALIGN_4 >> MO_ASHIFT] = "al4+", + [MO_ALIGN_8 >> MO_ASHIFT] = "al8+", + [MO_ALIGN_16 >> MO_ASHIFT] = "al16+", + [MO_ALIGN_32 >> MO_ASHIFT] = "al32+", + [MO_ALIGN_64 >> MO_ASHIFT] = "al64+", +}; + void tcg_dump_ops(TCGContext *s) { char buf[128]; @@ -1098,14 +1116,8 @@ void tcg_dump_ops(TCGContext *s) if (op & ~(MO_AMASK | MO_BSWAP | MO_SSIZE)) { qemu_log(",$0x%x,%u", op, ix); } else { - const char *s_al = "", *s_op; - if (op & MO_AMASK) { - if ((op & MO_AMASK) == MO_ALIGN) { - s_al = "al+"; - } else { - s_al = "un+"; - } - } + const char *s_al, *s_op; + s_al = alignment_name[(op & MO_AMASK) >> MO_ASHIFT]; s_op = ldst_name[op & (MO_BSWAP | MO_SSIZE)]; qemu_log(",%s%s,%u", s_al, s_op, ix); } @@ -1680,35 +1692,89 @@ static void temp_allocate_frame(TCGContext *s, int temp) static void temp_load(TCGContext *, TCGTemp *, TCGRegSet, TCGRegSet); -/* sync register 'reg' by saving it to the corresponding temporary */ -static void tcg_reg_sync(TCGContext *s, TCGReg reg, TCGRegSet allocated_regs) +/* Mark a temporary as free or dead. If 'free_or_dead' is negative, + mark it free; otherwise mark it dead. */ +static void temp_free_or_dead(TCGContext *s, TCGTemp *ts, int free_or_dead) { - TCGTemp *ts = s->reg_to_temp[reg]; + if (ts->fixed_reg) { + return; + } + if (ts->val_type == TEMP_VAL_REG) { + s->reg_to_temp[ts->reg] = NULL; + } + ts->val_type = (free_or_dead < 0 + || ts->temp_local + || temp_idx(s, ts) < s->nb_globals + ? TEMP_VAL_MEM : TEMP_VAL_DEAD); +} - tcg_debug_assert(ts->val_type == TEMP_VAL_REG); - if (!ts->mem_coherent && !ts->fixed_reg) { +/* Mark a temporary as dead. */ +static inline void temp_dead(TCGContext *s, TCGTemp *ts) +{ + temp_free_or_dead(s, ts, 1); +} + +/* Sync a temporary to memory. 'allocated_regs' is used in case a temporary + registers needs to be allocated to store a constant. If 'free_or_dead' + is non-zero, subsequently release the temporary; if it is positive, the + temp is dead; if it is negative, the temp is free. */ +static void temp_sync(TCGContext *s, TCGTemp *ts, + TCGRegSet allocated_regs, int free_or_dead) +{ + if (ts->fixed_reg) { + return; + } + if (!ts->mem_coherent) { if (!ts->mem_allocated) { temp_allocate_frame(s, temp_idx(s, ts)); - } else if (ts->indirect_reg) { - tcg_regset_set_reg(allocated_regs, ts->reg); + } + if (ts->indirect_reg) { + if (ts->val_type == TEMP_VAL_REG) { + tcg_regset_set_reg(allocated_regs, ts->reg); + } temp_load(s, ts->mem_base, tcg_target_available_regs[TCG_TYPE_PTR], allocated_regs); } - tcg_out_st(s, ts->type, reg, ts->mem_base->reg, ts->mem_offset); + switch (ts->val_type) { + case TEMP_VAL_CONST: + /* If we're going to free the temp immediately, then we won't + require it later in a register, so attempt to store the + constant to memory directly. */ + if (free_or_dead + && tcg_out_sti(s, ts->type, ts->val, + ts->mem_base->reg, ts->mem_offset)) { + break; + } + temp_load(s, ts, tcg_target_available_regs[ts->type], + allocated_regs); + /* fallthrough */ + + case TEMP_VAL_REG: + tcg_out_st(s, ts->type, ts->reg, + ts->mem_base->reg, ts->mem_offset); + break; + + case TEMP_VAL_MEM: + break; + + case TEMP_VAL_DEAD: + default: + tcg_abort(); + } + ts->mem_coherent = 1; + } + if (free_or_dead) { + temp_free_or_dead(s, ts, free_or_dead); } - ts->mem_coherent = 1; } /* free register 'reg' by spilling the corresponding temporary if necessary */ static void tcg_reg_free(TCGContext *s, TCGReg reg, TCGRegSet allocated_regs) { TCGTemp *ts = s->reg_to_temp[reg]; - if (ts != NULL) { - tcg_reg_sync(s, reg, allocated_regs); - ts->val_type = TEMP_VAL_MEM; - s->reg_to_temp[reg] = NULL; + temp_sync(s, ts, allocated_regs, -1); } } @@ -1778,45 +1844,9 @@ static void temp_load(TCGContext *s, TCGTemp *ts, TCGRegSet desired_regs, s->reg_to_temp[reg] = ts; } -/* mark a temporary as dead. */ -static inline void temp_dead(TCGContext *s, TCGTemp *ts) -{ - if (ts->fixed_reg) { - return; - } - if (ts->val_type == TEMP_VAL_REG) { - s->reg_to_temp[ts->reg] = NULL; - } - ts->val_type = (temp_idx(s, ts) < s->nb_globals || ts->temp_local - ? TEMP_VAL_MEM : TEMP_VAL_DEAD); -} - -/* sync a temporary to memory. 'allocated_regs' is used in case a - temporary registers needs to be allocated to store a constant. */ -static void temp_sync(TCGContext *s, TCGTemp *ts, TCGRegSet allocated_regs) -{ - if (ts->fixed_reg) { - return; - } - switch (ts->val_type) { - case TEMP_VAL_CONST: - temp_load(s, ts, tcg_target_available_regs[ts->type], allocated_regs); - /* fallthrough */ - case TEMP_VAL_REG: - tcg_reg_sync(s, ts->reg, allocated_regs); - break; - case TEMP_VAL_DEAD: - case TEMP_VAL_MEM: - break; - default: - tcg_abort(); - } -} - -/* save a temporary to memory. 'allocated_regs' is used in case a - temporary registers needs to be allocated to store a constant. */ -static inline void temp_save(TCGContext *s, TCGTemp *ts, - TCGRegSet allocated_regs) +/* Save a temporary to memory. 'allocated_regs' is used in case a + temporary registers needs to be allocated to store a constant. */ +static void temp_save(TCGContext *s, TCGTemp *ts, TCGRegSet allocated_regs) { #ifdef USE_LIVENESS_ANALYSIS /* ??? Liveness does not yet incorporate indirect bases. */ @@ -1827,8 +1857,7 @@ static inline void temp_save(TCGContext *s, TCGTemp *ts, return; } #endif - temp_sync(s, ts, allocated_regs); - temp_dead(s, ts); + temp_sync(s, ts, allocated_regs, 1); } /* save globals to their canonical location and assume they can be @@ -1861,7 +1890,7 @@ static void sync_globals(TCGContext *s, TCGRegSet allocated_regs) continue; } #endif - temp_sync(s, ts, allocated_regs); + temp_sync(s, ts, allocated_regs, 0); } } @@ -1905,21 +1934,21 @@ static void tcg_reg_alloc_movi(TCGContext *s, const TCGArg *args, val = args[1]; if (ots->fixed_reg) { - /* for fixed registers, we do not do any constant - propagation */ + /* For fixed registers, we do not do any constant propagation. */ tcg_out_movi(s, ots->type, ots->reg, val); - } else { - /* The movi is not explicitly generated here */ - if (ots->val_type == TEMP_VAL_REG) { - s->reg_to_temp[ots->reg] = NULL; - } - ots->val_type = TEMP_VAL_CONST; - ots->val = val; + return; } + + /* The movi is not explicitly generated here. */ + if (ots->val_type == TEMP_VAL_REG) { + s->reg_to_temp[ots->reg] = NULL; + } + ots->val_type = TEMP_VAL_CONST; + ots->val = val; + ots->mem_coherent = 0; if (NEED_SYNC_ARG(0)) { - temp_sync(s, ots, s->reserved_regs); - } - if (IS_DEAD_ARG(0)) { + temp_sync(s, ots, s->reserved_regs, IS_DEAD_ARG(0)); + } else if (IS_DEAD_ARG(0)) { temp_dead(s, ots); } } @@ -2004,7 +2033,7 @@ static void tcg_reg_alloc_mov(TCGContext *s, const TCGOpDef *def, ots->mem_coherent = 0; s->reg_to_temp[ots->reg] = ots; if (NEED_SYNC_ARG(0)) { - tcg_reg_sync(s, ots->reg, allocated_regs); + temp_sync(s, ots, allocated_regs, 0); } } } @@ -2163,9 +2192,8 @@ static void tcg_reg_alloc_op(TCGContext *s, tcg_out_mov(s, ts->type, ts->reg, reg); } if (NEED_SYNC_ARG(i)) { - tcg_reg_sync(s, reg, allocated_regs); - } - if (IS_DEAD_ARG(i)) { + temp_sync(s, ts, allocated_regs, IS_DEAD_ARG(i)); + } else if (IS_DEAD_ARG(i)) { temp_dead(s, ts); } } @@ -2298,9 +2326,8 @@ static void tcg_reg_alloc_call(TCGContext *s, int nb_oargs, int nb_iargs, ts->mem_coherent = 0; s->reg_to_temp[reg] = ts; if (NEED_SYNC_ARG(i)) { - tcg_reg_sync(s, reg, allocated_regs); - } - if (IS_DEAD_ARG(i)) { + temp_sync(s, ts, allocated_regs, IS_DEAD_ARG(i)); + } else if (IS_DEAD_ARG(i)) { temp_dead(s, ts); } } diff --git a/tcg/tcg.h b/tcg/tcg.h index 66d7fc01c5..66ae0c76e6 100644 --- a/tcg/tcg.h +++ b/tcg/tcg.h @@ -191,6 +191,15 @@ typedef uint64_t tcg_insn_unit; #endif +#ifdef CONFIG_DEBUG_TCG +# define tcg_debug_assert(X) do { assert(X); } while (0) +#elif QEMU_GNUC_PREREQ(4, 5) +# define tcg_debug_assert(X) \ + do { if (!(X)) { __builtin_unreachable(); } } while (0) +#else +# define tcg_debug_assert(X) do { (void)(X); } while (0) +#endif + typedef struct TCGRelocation { struct TCGRelocation *next; int type; @@ -275,10 +284,26 @@ typedef enum TCGMemOp { #endif /* MO_UNALN accesses are never checked for alignment. - MO_ALIGN accesses will result in a call to the CPU's - do_unaligned_access hook if the guest address is not aligned. - The default depends on whether the target CPU defines ALIGNED_ONLY. */ - MO_AMASK = 16, + * MO_ALIGN accesses will result in a call to the CPU's + * do_unaligned_access hook if the guest address is not aligned. + * The default depends on whether the target CPU defines ALIGNED_ONLY. + * Some architectures (e.g. ARMv8) need the address which is aligned + * to a size more than the size of the memory access. + * To support such check it's enough the current costless alignment + * check implementation in QEMU, but we need to support + * an alignment size specifying. + * MO_ALIGN supposes a natural alignment + * (i.e. the alignment size is the size of a memory access). + * Note that an alignment size must be equal or greater + * than an access size. + * There are three options: + * - an alignment to the size of an access (MO_ALIGN); + * - an alignment to the specified size that is equal or greater than + * an access size (MO_ALIGN_x where 'x' is a size in bytes); + * - unaligned access permitted (MO_UNALN). + */ + MO_ASHIFT = 4, + MO_AMASK = 7 << MO_ASHIFT, #ifdef ALIGNED_ONLY MO_ALIGN = 0, MO_UNALN = MO_AMASK, @@ -286,6 +311,12 @@ typedef enum TCGMemOp { MO_ALIGN = MO_AMASK, MO_UNALN = 0, #endif + MO_ALIGN_2 = 1 << MO_ASHIFT, + MO_ALIGN_4 = 2 << MO_ASHIFT, + MO_ALIGN_8 = 3 << MO_ASHIFT, + MO_ALIGN_16 = 4 << MO_ASHIFT, + MO_ALIGN_32 = 5 << MO_ASHIFT, + MO_ALIGN_64 = 6 << MO_ASHIFT, /* Combinations of the above, for ease of use. */ MO_UB = MO_8, @@ -317,6 +348,45 @@ typedef enum TCGMemOp { MO_SSIZE = MO_SIZE | MO_SIGN, } TCGMemOp; +/** + * get_alignment_bits + * @memop: TCGMemOp value + * + * Extract the alignment size from the memop. + * + * Returns: 0 in case of byte access (which is always aligned); + * positive value - number of alignment bits; + * negative value if unaligned access enabled + * and this is not a byte access. + */ +static inline int get_alignment_bits(TCGMemOp memop) +{ + int a = memop & MO_AMASK; + int s = memop & MO_SIZE; + int r; + + if (a == MO_UNALN) { + /* Negative value if unaligned access enabled, + * or zero value in case of byte access. + */ + return -s; + } else if (a == MO_ALIGN) { + /* A natural alignment: return a number of access size bits */ + r = s; + } else { + /* Specific alignment size. It must be equal or greater + * than the access size. + */ + r = a >> MO_ASHIFT; + tcg_debug_assert(r >= s); + } +#if defined(CONFIG_SOFTMMU) + /* The requested alignment cannot overlap the TLB flags. */ + tcg_debug_assert((TLB_FLAGS_MASK & ((1 << r) - 1)) == 0); +#endif + return r; +} + typedef tcg_target_ulong TCGArg; /* Define a type and accessor macros for variables. Using pointer types @@ -790,15 +860,6 @@ do {\ abort();\ } while (0) -#ifdef CONFIG_DEBUG_TCG -# define tcg_debug_assert(X) do { assert(X); } while (0) -#elif QEMU_GNUC_PREREQ(4, 5) -# define tcg_debug_assert(X) \ - do { if (!(X)) { __builtin_unreachable(); } } while (0) -#else -# define tcg_debug_assert(X) do { (void)(X); } while (0) -#endif - void tcg_add_target_add_op_defs(const TCGTargetOpDef *tdefs); #if UINTPTR_MAX == UINT32_MAX diff --git a/tcg/tci/tcg-target.inc.c b/tcg/tci/tcg-target.inc.c index fa74d5278e..3c47ea7a9e 100644 --- a/tcg/tci/tcg-target.inc.c +++ b/tcg/tci/tcg-target.inc.c @@ -834,6 +834,12 @@ static void tcg_out_st(TCGContext *s, TCGType type, TCGReg arg, TCGReg arg1, old_code_ptr[1] = s->code_ptr - old_code_ptr; } +static inline bool tcg_out_sti(TCGContext *s, TCGType type, TCGArg val, + TCGReg base, intptr_t ofs) +{ + return false; +} + /* Test if a constant matches the constraint. */ static int tcg_target_const_match(tcg_target_long val, TCGType type, const TCGArgConstraint *arg_ct)