From 19917791fd1433b79d20b26440ec321d71dcfc01 Mon Sep 17 00:00:00 2001 From: Peter Maydell Date: Sat, 4 Jan 2014 22:15:45 +0000 Subject: [PATCH 01/61] target-arm: Widen thread-local register state fields to 64 bits The common pattern for system registers in a 64-bit capable ARM CPU is that when in AArch32 the cp15 register is a view of the bottom 32 bits of the 64-bit AArch64 system register; writes in AArch32 leave the top half unchanged. The most natural way to model this is to have the state field in the CPU struct be a 64 bit value, and simply have the AArch32 TCG code operate on a pointer to its lower half. For aarch64-linux-user the only registers we need to share like this are the thread-local-storage ones. Widen their fields to 64 bits and provide the 64 bit reginfo struct to make them visible in AArch64 state. Note that minor cleanup of the AArch64 system register encoding space means We can share the TPIDR_EL1 reginfo but need split encodings for TPIDR_EL0 and TPIDRRO_EL0. Since we're touching almost every line in QEMU that uses the c13_tls* fields in this patch anyway, we take the opportunity to rename them in line with the standard ARM architectural names for these registers. Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- linux-user/aarch64/target_cpu.h | 5 ++++- linux-user/arm/target_cpu.h | 2 +- linux-user/main.c | 2 +- target-arm/cpu.h | 18 +++++++++++++++--- target-arm/helper.c | 22 +++++++++++++++------- 5 files changed, 36 insertions(+), 13 deletions(-) diff --git a/linux-user/aarch64/target_cpu.h b/linux-user/aarch64/target_cpu.h index 6f5539b50f..21560ef832 100644 --- a/linux-user/aarch64/target_cpu.h +++ b/linux-user/aarch64/target_cpu.h @@ -29,7 +29,10 @@ static inline void cpu_clone_regs(CPUARMState *env, target_ulong newsp) static inline void cpu_set_tls(CPUARMState *env, target_ulong newtls) { - env->sr.tpidr_el0 = newtls; + /* Note that AArch64 Linux keeps the TLS pointer in TPIDR; this is + * different from AArch32 Linux, which uses TPIDRRO. + */ + env->cp15.tpidr_el0 = newtls; } #endif diff --git a/linux-user/arm/target_cpu.h b/linux-user/arm/target_cpu.h index ed323c079d..39d65b692b 100644 --- a/linux-user/arm/target_cpu.h +++ b/linux-user/arm/target_cpu.h @@ -29,7 +29,7 @@ static inline void cpu_clone_regs(CPUARMState *env, target_ulong newsp) static inline void cpu_set_tls(CPUARMState *env, target_ulong newtls) { - env->cp15.c13_tls2 = newtls; + env->cp15.tpidrro_el0 = newtls; } #endif diff --git a/linux-user/main.c b/linux-user/main.c index 54f71fe8f6..c0df8b5cd4 100644 --- a/linux-user/main.c +++ b/linux-user/main.c @@ -566,7 +566,7 @@ do_kernel_trap(CPUARMState *env) end_exclusive(); break; case 0xffff0fe0: /* __kernel_get_tls */ - env->regs[0] = env->cp15.c13_tls2; + env->regs[0] = env->cp15.tpidrro_el0; break; case 0xffff0f60: /* __kernel_cmpxchg64 */ arm_kernel_cmpxchg64_helper(env); diff --git a/target-arm/cpu.h b/target-arm/cpu.h index ab8ef17862..fc36514d4b 100644 --- a/target-arm/cpu.h +++ b/target-arm/cpu.h @@ -66,6 +66,18 @@ /* ARM-specific interrupt pending bits. */ #define CPU_INTERRUPT_FIQ CPU_INTERRUPT_TGT_EXT_1 +/* The usual mapping for an AArch64 system register to its AArch32 + * counterpart is for the 32 bit world to have access to the lower + * half only (with writes leaving the upper half untouched). It's + * therefore useful to be able to pass TCG the offset of the least + * significant half of a uint64_t struct member. + */ +#ifdef HOST_WORDS_BIGENDIAN +#define offsetoflow32(S, M) (offsetof(S, M + sizeof(uint32_t)) +#else +#define offsetoflow32(S, M) offsetof(S, M) +#endif + /* Meanings of the ARMCPU object's two inbound GPIO lines */ #define ARM_CPU_IRQ 0 #define ARM_CPU_FIQ 1 @@ -188,9 +200,9 @@ typedef struct CPUARMState { uint32_t c12_vbar; /* vector base address register */ uint32_t c13_fcse; /* FCSE PID. */ uint32_t c13_context; /* Context ID. */ - uint32_t c13_tls1; /* User RW Thread register. */ - uint32_t c13_tls2; /* User RO Thread register. */ - uint32_t c13_tls3; /* Privileged Thread register. */ + uint64_t tpidr_el0; /* User RW Thread register. */ + uint64_t tpidrro_el0; /* User RO Thread register. */ + uint64_t tpidr_el1; /* Privileged Thread register. */ uint32_t c14_cntfrq; /* Counter Frequency register */ uint32_t c14_cntkctl; /* Timer Control register */ ARMGenericTimer c14_timer[NUM_GTIMERS]; diff --git a/target-arm/helper.c b/target-arm/helper.c index 3efe8b4302..f91e3fd35d 100644 --- a/target-arm/helper.c +++ b/target-arm/helper.c @@ -740,18 +740,26 @@ static const ARMCPRegInfo t2ee_cp_reginfo[] = { }; static const ARMCPRegInfo v6k_cp_reginfo[] = { + { .name = "TPIDR_EL0", .state = ARM_CP_STATE_AA64, + .opc0 = 3, .opc1 = 3, .opc2 = 2, .crn = 13, .crm = 0, + .access = PL0_RW, + .fieldoffset = offsetof(CPUARMState, cp15.tpidr_el0), .resetvalue = 0 }, { .name = "TPIDRURW", .cp = 15, .crn = 13, .crm = 0, .opc1 = 0, .opc2 = 2, .access = PL0_RW, - .fieldoffset = offsetof(CPUARMState, cp15.c13_tls1), - .resetvalue = 0 }, + .fieldoffset = offsetoflow32(CPUARMState, cp15.tpidr_el0), + .resetfn = arm_cp_reset_ignore }, + { .name = "TPIDRRO_EL0", .state = ARM_CP_STATE_AA64, + .opc0 = 3, .opc1 = 3, .opc2 = 3, .crn = 13, .crm = 0, + .access = PL0_R|PL1_W, + .fieldoffset = offsetof(CPUARMState, cp15.tpidrro_el0), .resetvalue = 0 }, { .name = "TPIDRURO", .cp = 15, .crn = 13, .crm = 0, .opc1 = 0, .opc2 = 3, .access = PL0_R|PL1_W, - .fieldoffset = offsetof(CPUARMState, cp15.c13_tls2), - .resetvalue = 0 }, - { .name = "TPIDRPRW", .cp = 15, .crn = 13, .crm = 0, .opc1 = 0, .opc2 = 4, + .fieldoffset = offsetoflow32(CPUARMState, cp15.tpidrro_el0), + .resetfn = arm_cp_reset_ignore }, + { .name = "TPIDR_EL1", .state = ARM_CP_STATE_BOTH, + .opc0 = 3, .opc1 = 0, .opc2 = 4, .crn = 13, .crm = 0, .access = PL1_RW, - .fieldoffset = offsetof(CPUARMState, cp15.c13_tls3), - .resetvalue = 0 }, + .fieldoffset = offsetof(CPUARMState, cp15.tpidr_el1), .resetvalue = 0 }, REGINFO_SENTINEL }; From 03468633b16e96176ffbe7fd1943f7014e63ccde Mon Sep 17 00:00:00 2001 From: Claudio Fontana Date: Sat, 4 Jan 2014 22:15:46 +0000 Subject: [PATCH 02/61] target-arm: A64: add support for add/sub with carry This patch adds support for C3.5.3 Add/subtract (with carry): instructions ADC, ADCS, SBC, SBCS. Signed-off-by: Claudio Fontana Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- target-arm/translate-a64.c | 105 ++++++++++++++++++++++++++++++++++++- 1 file changed, 103 insertions(+), 2 deletions(-) diff --git a/target-arm/translate-a64.c b/target-arm/translate-a64.c index c8ed799834..9f508b9e91 100644 --- a/target-arm/translate-a64.c +++ b/target-arm/translate-a64.c @@ -393,6 +393,71 @@ static void gen_sub_CC(int sf, TCGv_i64 dest, TCGv_i64 t0, TCGv_i64 t1) } } +/* dest = T0 + T1 + CF; do not compute flags. */ +static void gen_adc(int sf, TCGv_i64 dest, TCGv_i64 t0, TCGv_i64 t1) +{ + TCGv_i64 flag = tcg_temp_new_i64(); + tcg_gen_extu_i32_i64(flag, cpu_CF); + tcg_gen_add_i64(dest, t0, t1); + tcg_gen_add_i64(dest, dest, flag); + tcg_temp_free_i64(flag); + + if (!sf) { + tcg_gen_ext32u_i64(dest, dest); + } +} + +/* dest = T0 + T1 + CF; compute C, N, V and Z flags. */ +static void gen_adc_CC(int sf, TCGv_i64 dest, TCGv_i64 t0, TCGv_i64 t1) +{ + if (sf) { + TCGv_i64 result, cf_64, vf_64, tmp; + result = tcg_temp_new_i64(); + cf_64 = tcg_temp_new_i64(); + vf_64 = tcg_temp_new_i64(); + tmp = tcg_const_i64(0); + + tcg_gen_extu_i32_i64(cf_64, cpu_CF); + tcg_gen_add2_i64(result, cf_64, t0, tmp, cf_64, tmp); + tcg_gen_add2_i64(result, cf_64, result, cf_64, t1, tmp); + tcg_gen_trunc_i64_i32(cpu_CF, cf_64); + gen_set_NZ64(result); + + tcg_gen_xor_i64(vf_64, result, t0); + tcg_gen_xor_i64(tmp, t0, t1); + tcg_gen_andc_i64(vf_64, vf_64, tmp); + tcg_gen_shri_i64(vf_64, vf_64, 32); + tcg_gen_trunc_i64_i32(cpu_VF, vf_64); + + tcg_gen_mov_i64(dest, result); + + tcg_temp_free_i64(tmp); + tcg_temp_free_i64(vf_64); + tcg_temp_free_i64(cf_64); + tcg_temp_free_i64(result); + } else { + TCGv_i32 t0_32, t1_32, tmp; + t0_32 = tcg_temp_new_i32(); + t1_32 = tcg_temp_new_i32(); + tmp = tcg_const_i32(0); + + tcg_gen_trunc_i64_i32(t0_32, t0); + tcg_gen_trunc_i64_i32(t1_32, t1); + tcg_gen_add2_i32(cpu_NF, cpu_CF, t0_32, tmp, cpu_CF, tmp); + tcg_gen_add2_i32(cpu_NF, cpu_CF, cpu_NF, cpu_CF, t1_32, tmp); + + tcg_gen_mov_i32(cpu_ZF, cpu_NF); + tcg_gen_xor_i32(cpu_VF, cpu_NF, t0_32); + tcg_gen_xor_i32(tmp, t0_32, t1_32); + tcg_gen_andc_i32(cpu_VF, cpu_VF, tmp); + tcg_gen_extu_i32_i64(dest, cpu_NF); + + tcg_temp_free_i32(tmp); + tcg_temp_free_i32(t1_32); + tcg_temp_free_i32(t0_32); + } +} + /* * Load/Store generators */ @@ -2376,10 +2441,46 @@ static void disas_data_proc_3src(DisasContext *s, uint32_t insn) tcg_temp_free_i64(tcg_tmp); } -/* Add/subtract (with carry) */ +/* C3.5.3 - Add/subtract (with carry) + * 31 30 29 28 27 26 25 24 23 22 21 20 16 15 10 9 5 4 0 + * +--+--+--+------------------------+------+---------+------+-----+ + * |sf|op| S| 1 1 0 1 0 0 0 0 | rm | opcode2 | Rn | Rd | + * +--+--+--+------------------------+------+---------+------+-----+ + * [000000] + */ + static void disas_adc_sbc(DisasContext *s, uint32_t insn) { - unsupported_encoding(s, insn); + unsigned int sf, op, setflags, rm, rn, rd; + TCGv_i64 tcg_y, tcg_rn, tcg_rd; + + if (extract32(insn, 10, 6) != 0) { + unallocated_encoding(s); + return; + } + + sf = extract32(insn, 31, 1); + op = extract32(insn, 30, 1); + setflags = extract32(insn, 29, 1); + rm = extract32(insn, 16, 5); + rn = extract32(insn, 5, 5); + rd = extract32(insn, 0, 5); + + tcg_rd = cpu_reg(s, rd); + tcg_rn = cpu_reg(s, rn); + + if (op) { + tcg_y = new_tmp_a64(s); + tcg_gen_not_i64(tcg_y, cpu_reg(s, rm)); + } else { + tcg_y = cpu_reg(s, rm); + } + + if (setflags) { + gen_adc_CC(sf, tcg_rd, tcg_rn, tcg_y); + } else { + gen_adc(sf, tcg_rd, tcg_rn, tcg_y); + } } /* Conditional compare (immediate) */ From fc514b094e8f5aa211a36925b62348ab52243762 Mon Sep 17 00:00:00 2001 From: Claudio Fontana Date: Sat, 4 Jan 2014 22:15:46 +0000 Subject: [PATCH 03/61] target-arm: A64: add support for conditional compare insns this patch adds support for C3.5.4 - C3.5.5 Conditional compare (both immediate and register) Signed-off-by: Claudio Fontana Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- target-arm/translate-a64.c | 73 +++++++++++++++++++++++++++++++------- 1 file changed, 60 insertions(+), 13 deletions(-) diff --git a/target-arm/translate-a64.c b/target-arm/translate-a64.c index 9f508b9e91..538d69e634 100644 --- a/target-arm/translate-a64.c +++ b/target-arm/translate-a64.c @@ -2483,16 +2483,67 @@ static void disas_adc_sbc(DisasContext *s, uint32_t insn) } } -/* Conditional compare (immediate) */ -static void disas_cc_imm(DisasContext *s, uint32_t insn) +/* C3.5.4 - C3.5.5 Conditional compare (immediate / register) + * 31 30 29 28 27 26 25 24 23 22 21 20 16 15 12 11 10 9 5 4 3 0 + * +--+--+--+------------------------+--------+------+----+--+------+--+-----+ + * |sf|op| S| 1 1 0 1 0 0 1 0 |imm5/rm | cond |i/r |o2| Rn |o3|nzcv | + * +--+--+--+------------------------+--------+------+----+--+------+--+-----+ + * [1] y [0] [0] + */ +static void disas_cc(DisasContext *s, uint32_t insn) { - unsupported_encoding(s, insn); -} + unsigned int sf, op, y, cond, rn, nzcv, is_imm; + int label_continue = -1; + TCGv_i64 tcg_tmp, tcg_y, tcg_rn; -/* Conditional compare (register) */ -static void disas_cc_reg(DisasContext *s, uint32_t insn) -{ - unsupported_encoding(s, insn); + if (!extract32(insn, 29, 1)) { + unallocated_encoding(s); + return; + } + if (insn & (1 << 10 | 1 << 4)) { + unallocated_encoding(s); + return; + } + sf = extract32(insn, 31, 1); + op = extract32(insn, 30, 1); + is_imm = extract32(insn, 11, 1); + y = extract32(insn, 16, 5); /* y = rm (reg) or imm5 (imm) */ + cond = extract32(insn, 12, 4); + rn = extract32(insn, 5, 5); + nzcv = extract32(insn, 0, 4); + + if (cond < 0x0e) { /* not always */ + int label_match = gen_new_label(); + label_continue = gen_new_label(); + arm_gen_test_cc(cond, label_match); + /* nomatch: */ + tcg_tmp = tcg_temp_new_i64(); + tcg_gen_movi_i64(tcg_tmp, nzcv << 28); + gen_set_nzcv(tcg_tmp); + tcg_temp_free_i64(tcg_tmp); + tcg_gen_br(label_continue); + gen_set_label(label_match); + } + /* match, or condition is always */ + if (is_imm) { + tcg_y = new_tmp_a64(s); + tcg_gen_movi_i64(tcg_y, y); + } else { + tcg_y = cpu_reg(s, y); + } + tcg_rn = cpu_reg(s, rn); + + tcg_tmp = tcg_temp_new_i64(); + if (op) { + gen_sub_CC(sf, tcg_tmp, tcg_rn, tcg_y); + } else { + gen_add_CC(sf, tcg_tmp, tcg_rn, tcg_y); + } + tcg_temp_free_i64(tcg_tmp); + + if (cond < 0x0e) { /* continue */ + gen_set_label(label_continue); + } } /* C3.5.6 Conditional select @@ -2846,11 +2897,7 @@ static void disas_data_proc_reg(DisasContext *s, uint32_t insn) disas_adc_sbc(s, insn); break; case 0x2: /* Conditional compare */ - if (insn & (1 << 11)) { /* (immediate) */ - disas_cc_imm(s, insn); - } else { /* (register) */ - disas_cc_reg(s, insn); - } + disas_cc(s, insn); /* both imm and reg forms */ break; case 0x4: /* Conditional select */ disas_cond_select(s, insn); From 66759c063c3472fe70df0c8489ecc6fd7dc1a843 Mon Sep 17 00:00:00 2001 From: Alexander Graf Date: Sat, 4 Jan 2014 22:15:46 +0000 Subject: [PATCH 04/61] target-arm: aarch64: add support for ld lit MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Adds support for Load Register (literal), both normal and SIMD/FP forms. Signed-off-by: Alexander Graf Signed-off-by: Alex Bennée Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- target-arm/translate-a64.c | 47 ++++++++++++++++++++++++++++++++++++-- 1 file changed, 45 insertions(+), 2 deletions(-) diff --git a/target-arm/translate-a64.c b/target-arm/translate-a64.c index 538d69e634..61974419cd 100644 --- a/target-arm/translate-a64.c +++ b/target-arm/translate-a64.c @@ -1112,10 +1112,53 @@ static void disas_ldst_excl(DisasContext *s, uint32_t insn) unsupported_encoding(s, insn); } -/* Load register (literal) */ +/* + * C3.3.5 Load register (literal) + * + * 31 30 29 27 26 25 24 23 5 4 0 + * +-----+-------+---+-----+-------------------+-------+ + * | opc | 0 1 1 | V | 0 0 | imm19 | Rt | + * +-----+-------+---+-----+-------------------+-------+ + * + * V: 1 -> vector (simd/fp) + * opc (non-vector): 00 -> 32 bit, 01 -> 64 bit, + * 10-> 32 bit signed, 11 -> prefetch + * opc (vector): 00 -> 32 bit, 01 -> 64 bit, 10 -> 128 bit (11 unallocated) + */ static void disas_ld_lit(DisasContext *s, uint32_t insn) { - unsupported_encoding(s, insn); + int rt = extract32(insn, 0, 5); + int64_t imm = sextract32(insn, 5, 19) << 2; + bool is_vector = extract32(insn, 26, 1); + int opc = extract32(insn, 30, 2); + bool is_signed = false; + int size = 2; + TCGv_i64 tcg_rt, tcg_addr; + + if (is_vector) { + if (opc == 3) { + unallocated_encoding(s); + return; + } + size = 2 + opc; + } else { + if (opc == 3) { + /* PRFM (literal) : prefetch */ + return; + } + size = 2 + extract32(opc, 0, 1); + is_signed = extract32(opc, 1, 1); + } + + tcg_rt = cpu_reg(s, rt); + + tcg_addr = tcg_const_i64((s->pc - 4) + imm); + if (is_vector) { + do_fp_ld(s, rt, tcg_addr, size); + } else { + do_gpr_ld(s, tcg_rt, tcg_addr, size, is_signed, false); + } + tcg_temp_free_i64(tcg_addr); } /* From 90ba562c61b15f418f284751faa81aca25de2d95 Mon Sep 17 00:00:00 2001 From: Peter Maydell Date: Sat, 4 Jan 2014 22:15:47 +0000 Subject: [PATCH 05/61] target-arm: Widen exclusive-access support struct fields to 64 bits In preparation for adding support for A64 load/store exclusive instructions, widen the fields in the CPU state struct that deal with address and data values for exclusives from 32 to 64 bits. Although in practice AArch64 and AArch32 exclusive accesses will be generally separate there are some odd theoretical corner cases (eg you should be able to do the exclusive load in AArch32, take an exception to AArch64 and successfully do the store exclusive there), and it's also easier to reason about. The changes in semantics for the variables are: exclusive_addr -> extended to 64 bits; -1ULL for "monitor lost", otherwise always < 2^32 for AArch32 exclusive_val -> extended to 64 bits. 64 bit exclusives in AArch32 now use the high half of exclusive_val instead of a separate exclusive_high exclusive_high -> is no longer used in AArch32; extended to 64 bits as it will be needed for AArch64's pair-of-64-bit-values exclusives. exclusive_test -> extended to 64 bits, as it is an address. Since this is a linux-user-only field, in arm-linux-user it will always have the top 32 bits zero. exclusive_info -> stays 32 bits, as it is neither data nor address, but simply holds register indexes etc. AArch64 will be able to fit all its information into 32 bits as well. Note that the refactoring of gen_store_exclusive() coincidentally fixes a minor bug where ldrexd would incorrectly update the first CPU register even if the load for the second register faulted. Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- linux-user/main.c | 25 +++++++++------- target-arm/cpu.h | 8 +++--- target-arm/machine.c | 12 ++++---- target-arm/translate.c | 65 +++++++++++++++++++++++++----------------- 4 files changed, 64 insertions(+), 46 deletions(-) diff --git a/linux-user/main.c b/linux-user/main.c index c0df8b5cd4..20f9832d38 100644 --- a/linux-user/main.c +++ b/linux-user/main.c @@ -589,16 +589,21 @@ do_kernel_trap(CPUARMState *env) static int do_strex(CPUARMState *env) { - uint32_t val; + uint64_t val; int size; int rc = 1; int segv = 0; uint32_t addr; start_exclusive(); - addr = env->exclusive_addr; - if (addr != env->exclusive_test) { + if (env->exclusive_addr != env->exclusive_test) { goto fail; } + /* We know we're always AArch32 so the address is in uint32_t range + * unless it was the -1 exclusive-monitor-lost value (which won't + * match exclusive_test above). + */ + assert(extract64(env->exclusive_addr, 32, 32) == 0); + addr = env->exclusive_addr; size = env->exclusive_info & 0xf; switch (size) { case 0: @@ -618,19 +623,19 @@ static int do_strex(CPUARMState *env) env->cp15.c6_data = addr; goto done; } - if (val != env->exclusive_val) { - goto fail; - } if (size == 3) { - segv = get_user_u32(val, addr + 4); + uint32_t valhi; + segv = get_user_u32(valhi, addr + 4); if (segv) { env->cp15.c6_data = addr + 4; goto done; } - if (val != env->exclusive_high) { - goto fail; - } + val = deposit64(val, 32, 32, valhi); } + if (val != env->exclusive_val) { + goto fail; + } + val = env->regs[(env->exclusive_info >> 8) & 0xf]; switch (size) { case 0: diff --git a/target-arm/cpu.h b/target-arm/cpu.h index fc36514d4b..7084a74bdb 100644 --- a/target-arm/cpu.h +++ b/target-arm/cpu.h @@ -278,11 +278,11 @@ typedef struct CPUARMState { float_status fp_status; float_status standard_fp_status; } vfp; - uint32_t exclusive_addr; - uint32_t exclusive_val; - uint32_t exclusive_high; + uint64_t exclusive_addr; + uint64_t exclusive_val; + uint64_t exclusive_high; #if defined(CONFIG_USER_ONLY) - uint32_t exclusive_test; + uint64_t exclusive_test; uint32_t exclusive_info; #endif diff --git a/target-arm/machine.c b/target-arm/machine.c index 74f010f637..8f9e7d4d28 100644 --- a/target-arm/machine.c +++ b/target-arm/machine.c @@ -222,9 +222,9 @@ static int cpu_post_load(void *opaque, int version_id) const VMStateDescription vmstate_arm_cpu = { .name = "cpu", - .version_id = 13, - .minimum_version_id = 13, - .minimum_version_id_old = 13, + .version_id = 14, + .minimum_version_id = 14, + .minimum_version_id_old = 14, .pre_save = cpu_pre_save, .post_load = cpu_post_load, .fields = (VMStateField[]) { @@ -253,9 +253,9 @@ const VMStateDescription vmstate_arm_cpu = { VMSTATE_VARRAY_INT32(cpreg_vmstate_values, ARMCPU, cpreg_vmstate_array_len, 0, vmstate_info_uint64, uint64_t), - VMSTATE_UINT32(env.exclusive_addr, ARMCPU), - VMSTATE_UINT32(env.exclusive_val, ARMCPU), - VMSTATE_UINT32(env.exclusive_high, ARMCPU), + VMSTATE_UINT64(env.exclusive_addr, ARMCPU), + VMSTATE_UINT64(env.exclusive_val, ARMCPU), + VMSTATE_UINT64(env.exclusive_high, ARMCPU), VMSTATE_UINT64(env.features, ARMCPU), VMSTATE_TIMER(gt_timer[GTIMER_PHYS], ARMCPU), VMSTATE_TIMER(gt_timer[GTIMER_VIRT], ARMCPU), diff --git a/target-arm/translate.c b/target-arm/translate.c index 8bfe95007e..4387547a9c 100644 --- a/target-arm/translate.c +++ b/target-arm/translate.c @@ -61,11 +61,10 @@ TCGv_ptr cpu_env; static TCGv_i64 cpu_V0, cpu_V1, cpu_M0; static TCGv_i32 cpu_R[16]; static TCGv_i32 cpu_CF, cpu_NF, cpu_VF, cpu_ZF; -static TCGv_i32 cpu_exclusive_addr; -static TCGv_i32 cpu_exclusive_val; -static TCGv_i32 cpu_exclusive_high; +static TCGv_i64 cpu_exclusive_addr; +static TCGv_i64 cpu_exclusive_val; #ifdef CONFIG_USER_ONLY -static TCGv_i32 cpu_exclusive_test; +static TCGv_i64 cpu_exclusive_test; static TCGv_i32 cpu_exclusive_info; #endif @@ -96,14 +95,12 @@ void arm_translate_init(void) cpu_VF = tcg_global_mem_new_i32(TCG_AREG0, offsetof(CPUARMState, VF), "VF"); cpu_ZF = tcg_global_mem_new_i32(TCG_AREG0, offsetof(CPUARMState, ZF), "ZF"); - cpu_exclusive_addr = tcg_global_mem_new_i32(TCG_AREG0, + cpu_exclusive_addr = tcg_global_mem_new_i64(TCG_AREG0, offsetof(CPUARMState, exclusive_addr), "exclusive_addr"); - cpu_exclusive_val = tcg_global_mem_new_i32(TCG_AREG0, + cpu_exclusive_val = tcg_global_mem_new_i64(TCG_AREG0, offsetof(CPUARMState, exclusive_val), "exclusive_val"); - cpu_exclusive_high = tcg_global_mem_new_i32(TCG_AREG0, - offsetof(CPUARMState, exclusive_high), "exclusive_high"); #ifdef CONFIG_USER_ONLY - cpu_exclusive_test = tcg_global_mem_new_i32(TCG_AREG0, + cpu_exclusive_test = tcg_global_mem_new_i64(TCG_AREG0, offsetof(CPUARMState, exclusive_test), "exclusive_test"); cpu_exclusive_info = tcg_global_mem_new_i32(TCG_AREG0, offsetof(CPUARMState, exclusive_info), "exclusive_info"); @@ -6758,30 +6755,34 @@ static void gen_load_exclusive(DisasContext *s, int rt, int rt2, default: abort(); } - tcg_gen_mov_i32(cpu_exclusive_val, tmp); - store_reg(s, rt, tmp); + if (size == 3) { TCGv_i32 tmp2 = tcg_temp_new_i32(); + TCGv_i32 tmp3 = tcg_temp_new_i32(); + tcg_gen_addi_i32(tmp2, addr, 4); - tmp = tcg_temp_new_i32(); - gen_aa32_ld32u(tmp, tmp2, IS_USER(s)); + gen_aa32_ld32u(tmp3, tmp2, IS_USER(s)); tcg_temp_free_i32(tmp2); - tcg_gen_mov_i32(cpu_exclusive_high, tmp); - store_reg(s, rt2, tmp); + tcg_gen_concat_i32_i64(cpu_exclusive_val, tmp, tmp3); + store_reg(s, rt2, tmp3); + } else { + tcg_gen_extu_i32_i64(cpu_exclusive_val, tmp); } - tcg_gen_mov_i32(cpu_exclusive_addr, addr); + + store_reg(s, rt, tmp); + tcg_gen_extu_i32_i64(cpu_exclusive_addr, addr); } static void gen_clrex(DisasContext *s) { - tcg_gen_movi_i32(cpu_exclusive_addr, -1); + tcg_gen_movi_i64(cpu_exclusive_addr, -1); } #ifdef CONFIG_USER_ONLY static void gen_store_exclusive(DisasContext *s, int rd, int rt, int rt2, TCGv_i32 addr, int size) { - tcg_gen_mov_i32(cpu_exclusive_test, addr); + tcg_gen_extu_i32_i64(cpu_exclusive_test, addr); tcg_gen_movi_i32(cpu_exclusive_info, size | (rd << 4) | (rt << 8) | (rt2 << 12)); gen_exception_insn(s, 4, EXCP_STREX); @@ -6791,6 +6792,7 @@ static void gen_store_exclusive(DisasContext *s, int rd, int rt, int rt2, TCGv_i32 addr, int size) { TCGv_i32 tmp; + TCGv_i64 val64, extaddr; int done_label; int fail_label; @@ -6802,7 +6804,11 @@ static void gen_store_exclusive(DisasContext *s, int rd, int rt, int rt2, } */ fail_label = gen_new_label(); done_label = gen_new_label(); - tcg_gen_brcond_i32(TCG_COND_NE, addr, cpu_exclusive_addr, fail_label); + extaddr = tcg_temp_new_i64(); + tcg_gen_extu_i32_i64(extaddr, addr); + tcg_gen_brcond_i64(TCG_COND_NE, extaddr, cpu_exclusive_addr, fail_label); + tcg_temp_free_i64(extaddr); + tmp = tcg_temp_new_i32(); switch (size) { case 0: @@ -6818,17 +6824,24 @@ static void gen_store_exclusive(DisasContext *s, int rd, int rt, int rt2, default: abort(); } - tcg_gen_brcond_i32(TCG_COND_NE, tmp, cpu_exclusive_val, fail_label); - tcg_temp_free_i32(tmp); + + val64 = tcg_temp_new_i64(); if (size == 3) { TCGv_i32 tmp2 = tcg_temp_new_i32(); + TCGv_i32 tmp3 = tcg_temp_new_i32(); tcg_gen_addi_i32(tmp2, addr, 4); - tmp = tcg_temp_new_i32(); - gen_aa32_ld32u(tmp, tmp2, IS_USER(s)); + gen_aa32_ld32u(tmp3, tmp2, IS_USER(s)); tcg_temp_free_i32(tmp2); - tcg_gen_brcond_i32(TCG_COND_NE, tmp, cpu_exclusive_high, fail_label); - tcg_temp_free_i32(tmp); + tcg_gen_concat_i32_i64(val64, tmp, tmp3); + tcg_temp_free_i32(tmp3); + } else { + tcg_gen_extu_i32_i64(val64, tmp); } + tcg_temp_free_i32(tmp); + + tcg_gen_brcond_i64(TCG_COND_NE, val64, cpu_exclusive_val, fail_label); + tcg_temp_free_i64(val64); + tmp = load_reg(s, rt); switch (size) { case 0: @@ -6856,7 +6869,7 @@ static void gen_store_exclusive(DisasContext *s, int rd, int rt, int rt2, gen_set_label(fail_label); tcg_gen_movi_i32(cpu_R[rd], 1); gen_set_label(done_label); - tcg_gen_movi_i32(cpu_exclusive_addr, -1); + tcg_gen_movi_i64(cpu_exclusive_addr, -1); } #endif From 2df7566445c8b0934030627dff2c2417ec6d8e27 Mon Sep 17 00:00:00 2001 From: Michael Matz Date: Sat, 4 Jan 2014 22:15:47 +0000 Subject: [PATCH 06/61] target-arm: A64: support for ld/st/cl exclusive MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit This implement exclusive loads/stores for aarch64 along the lines of arm32 and ppc implementations. The exclusive load remembers the address and loaded value. The exclusive store throws an an exception which uses those values to check for equality in a proper exclusive region. This is not actually the architecture mandated semantics (for either AArch32 or AArch64) but it is close enough for typical guest code sequences to work correctly, and saves us from having to monitor all guest stores. It's fairly easy to come up with test cases where we don't behave like hardware - we don't for example model cache line behaviour. However in the common patterns this works, and the existing 32 bit ARM exclusive access implementation has the same limitations. AArch64 also implements new acquire/release loads/stores (which may be either exclusive or non-exclusive). These imposes extra ordering constraints on memory operations (ie they act as if they have an implicit barrier built into them). As TCG is single-threaded all our barriers are no-ops, so these just behave like normal loads and stores. Signed-off-by: Michael Matz Signed-off-by: Alex Bennée Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- linux-user/main.c | 127 +++++++++++++++++++++++++++++- target-arm/translate-a64.c | 156 ++++++++++++++++++++++++++++++++++++- 2 files changed, 277 insertions(+), 6 deletions(-) diff --git a/linux-user/main.c b/linux-user/main.c index 20f9832d38..cabc9e1a0e 100644 --- a/linux-user/main.c +++ b/linux-user/main.c @@ -585,8 +585,8 @@ do_kernel_trap(CPUARMState *env) return 0; } -#endif +/* Store exclusive handling for AArch32 */ static int do_strex(CPUARMState *env) { uint64_t val; @@ -670,7 +670,6 @@ done: return segv; } -#ifdef TARGET_ABI32 void cpu_loop(CPUARMState *env) { CPUState *cs = CPU(arm_env_get_cpu(env)); @@ -885,6 +884,122 @@ void cpu_loop(CPUARMState *env) #else +/* + * Handle AArch64 store-release exclusive + * + * rs = gets the status result of store exclusive + * rt = is the register that is stored + * rt2 = is the second register store (in STP) + * + */ +static int do_strex_a64(CPUARMState *env) +{ + uint64_t val; + int size; + bool is_pair; + int rc = 1; + int segv = 0; + uint64_t addr; + int rs, rt, rt2; + + start_exclusive(); + /* size | is_pair << 2 | (rs << 4) | (rt << 9) | (rt2 << 14)); */ + size = extract32(env->exclusive_info, 0, 2); + is_pair = extract32(env->exclusive_info, 2, 1); + rs = extract32(env->exclusive_info, 4, 5); + rt = extract32(env->exclusive_info, 9, 5); + rt2 = extract32(env->exclusive_info, 14, 5); + + addr = env->exclusive_addr; + + if (addr != env->exclusive_test) { + goto finish; + } + + switch (size) { + case 0: + segv = get_user_u8(val, addr); + break; + case 1: + segv = get_user_u16(val, addr); + break; + case 2: + segv = get_user_u32(val, addr); + break; + case 3: + segv = get_user_u64(val, addr); + break; + default: + abort(); + } + if (segv) { + env->cp15.c6_data = addr; + goto error; + } + if (val != env->exclusive_val) { + goto finish; + } + if (is_pair) { + if (size == 2) { + segv = get_user_u32(val, addr + 4); + } else { + segv = get_user_u64(val, addr + 8); + } + if (segv) { + env->cp15.c6_data = addr + (size == 2 ? 4 : 8); + goto error; + } + if (val != env->exclusive_high) { + goto finish; + } + } + val = env->xregs[rt]; + switch (size) { + case 0: + segv = put_user_u8(val, addr); + break; + case 1: + segv = put_user_u16(val, addr); + break; + case 2: + segv = put_user_u32(val, addr); + break; + case 3: + segv = put_user_u64(val, addr); + break; + } + if (segv) { + goto error; + } + if (is_pair) { + val = env->xregs[rt2]; + if (size == 2) { + segv = put_user_u32(val, addr + 4); + } else { + segv = put_user_u64(val, addr + 8); + } + if (segv) { + env->cp15.c6_data = addr + (size == 2 ? 4 : 8); + goto error; + } + } + rc = 0; +finish: + env->pc += 4; + /* rs == 31 encodes a write to the ZR, thus throwing away + * the status return. This is rather silly but valid. + */ + if (rs < 31) { + env->xregs[rs] = rc; + } +error: + /* instruction faulted, PC does not advance */ + /* either way a strex releases any exclusive lock we have */ + env->exclusive_addr = -1; + end_exclusive(); + return segv; +} + /* AArch64 main loop */ void cpu_loop(CPUARMState *env) { @@ -944,7 +1059,7 @@ void cpu_loop(CPUARMState *env) } break; case EXCP_STREX: - if (do_strex(env)) { + if (do_strex_a64(env)) { addr = env->cp15.c6_data; goto do_segv; } @@ -956,6 +1071,12 @@ void cpu_loop(CPUARMState *env) abort(); } process_pending_signals(env); + /* Exception return on AArch64 always clears the exclusive monitor, + * so any return to running guest code implies this. + * A strex (successful or otherwise) also clears the monitor, so + * we don't need to specialcase EXCP_STREX. + */ + env->exclusive_addr = -1; } } #endif /* ndef TARGET_ABI32 */ diff --git a/target-arm/translate-a64.c b/target-arm/translate-a64.c index 61974419cd..40c6fc4a9b 100644 --- a/target-arm/translate-a64.c +++ b/target-arm/translate-a64.c @@ -38,6 +38,15 @@ static TCGv_i64 cpu_X[32]; static TCGv_i64 cpu_pc; static TCGv_i32 cpu_NF, cpu_ZF, cpu_CF, cpu_VF; +/* Load/store exclusive handling */ +static TCGv_i64 cpu_exclusive_addr; +static TCGv_i64 cpu_exclusive_val; +static TCGv_i64 cpu_exclusive_high; +#ifdef CONFIG_USER_ONLY +static TCGv_i64 cpu_exclusive_test; +static TCGv_i32 cpu_exclusive_info; +#endif + static const char *regnames[] = { "x0", "x1", "x2", "x3", "x4", "x5", "x6", "x7", "x8", "x9", "x10", "x11", "x12", "x13", "x14", "x15", @@ -70,6 +79,19 @@ void a64_translate_init(void) cpu_ZF = tcg_global_mem_new_i32(TCG_AREG0, offsetof(CPUARMState, ZF), "ZF"); cpu_CF = tcg_global_mem_new_i32(TCG_AREG0, offsetof(CPUARMState, CF), "CF"); cpu_VF = tcg_global_mem_new_i32(TCG_AREG0, offsetof(CPUARMState, VF), "VF"); + + cpu_exclusive_addr = tcg_global_mem_new_i64(TCG_AREG0, + offsetof(CPUARMState, exclusive_addr), "exclusive_addr"); + cpu_exclusive_val = tcg_global_mem_new_i64(TCG_AREG0, + offsetof(CPUARMState, exclusive_val), "exclusive_val"); + cpu_exclusive_high = tcg_global_mem_new_i64(TCG_AREG0, + offsetof(CPUARMState, exclusive_high), "exclusive_high"); +#ifdef CONFIG_USER_ONLY + cpu_exclusive_test = tcg_global_mem_new_i64(TCG_AREG0, + offsetof(CPUARMState, exclusive_test), "exclusive_test"); + cpu_exclusive_info = tcg_global_mem_new_i32(TCG_AREG0, + offsetof(CPUARMState, exclusive_info), "exclusive_info"); +#endif } void aarch64_cpu_dump_state(CPUState *cs, FILE *f, @@ -767,6 +789,11 @@ static void handle_hint(DisasContext *s, uint32_t insn, } } +static void gen_clrex(DisasContext *s, uint32_t insn) +{ + tcg_gen_movi_i64(cpu_exclusive_addr, -1); +} + /* CLREX, DSB, DMB, ISB */ static void handle_sync(DisasContext *s, uint32_t insn, unsigned int op1, unsigned int op2, unsigned int crm) @@ -778,7 +805,7 @@ static void handle_sync(DisasContext *s, uint32_t insn, switch (op2) { case 2: /* CLREX */ - unsupported_encoding(s, insn); + gen_clrex(s, insn); return; case 4: /* DSB */ case 5: /* DMB */ @@ -1106,10 +1133,133 @@ static void disas_b_exc_sys(DisasContext *s, uint32_t insn) } } -/* Load/store exclusive */ +/* + * Load/Store exclusive instructions are implemented by remembering + * the value/address loaded, and seeing if these are the same + * when the store is performed. This is not actually the architecturally + * mandated semantics, but it works for typical guest code sequences + * and avoids having to monitor regular stores. + * + * In system emulation mode only one CPU will be running at once, so + * this sequence is effectively atomic. In user emulation mode we + * throw an exception and handle the atomic operation elsewhere. + */ +static void gen_load_exclusive(DisasContext *s, int rt, int rt2, + TCGv_i64 addr, int size, bool is_pair) +{ + TCGv_i64 tmp = tcg_temp_new_i64(); + TCGMemOp memop = MO_TE + size; + + g_assert(size <= 3); + tcg_gen_qemu_ld_i64(tmp, addr, get_mem_index(s), memop); + + if (is_pair) { + TCGv_i64 addr2 = tcg_temp_new_i64(); + TCGv_i64 hitmp = tcg_temp_new_i64(); + + g_assert(size >= 2); + tcg_gen_addi_i64(addr2, addr, 1 << size); + tcg_gen_qemu_ld_i64(hitmp, addr2, get_mem_index(s), memop); + tcg_temp_free_i64(addr2); + tcg_gen_mov_i64(cpu_exclusive_high, hitmp); + tcg_gen_mov_i64(cpu_reg(s, rt2), hitmp); + tcg_temp_free_i64(hitmp); + } + + tcg_gen_mov_i64(cpu_exclusive_val, tmp); + tcg_gen_mov_i64(cpu_reg(s, rt), tmp); + + tcg_temp_free_i64(tmp); + tcg_gen_mov_i64(cpu_exclusive_addr, addr); +} + +#ifdef CONFIG_USER_ONLY +static void gen_store_exclusive(DisasContext *s, int rd, int rt, int rt2, + TCGv_i64 addr, int size, int is_pair) +{ + tcg_gen_mov_i64(cpu_exclusive_test, addr); + tcg_gen_movi_i32(cpu_exclusive_info, + size | is_pair << 2 | (rd << 4) | (rt << 9) | (rt2 << 14)); + gen_exception_insn(s, 4, EXCP_STREX); +} +#else +static void gen_store_exclusive(DisasContext *s, int rd, int rt, int rt2, + TCGv_i64 addr, int size, int is_pair) +{ + qemu_log_mask(LOG_UNIMP, + "%s:%d: system mode store_exclusive unsupported " + "at pc=%016" PRIx64 "\n", + __FILE__, __LINE__, s->pc - 4); +} +#endif + +/* C3.3.6 Load/store exclusive + * + * 31 30 29 24 23 22 21 20 16 15 14 10 9 5 4 0 + * +-----+-------------+----+---+----+------+----+-------+------+------+ + * | sz | 0 0 1 0 0 0 | o2 | L | o1 | Rs | o0 | Rt2 | Rn | Rt | + * +-----+-------------+----+---+----+------+----+-------+------+------+ + * + * sz: 00 -> 8 bit, 01 -> 16 bit, 10 -> 32 bit, 11 -> 64 bit + * L: 0 -> store, 1 -> load + * o2: 0 -> exclusive, 1 -> not + * o1: 0 -> single register, 1 -> register pair + * o0: 1 -> load-acquire/store-release, 0 -> not + * + * o0 == 0 AND o2 == 1 is un-allocated + * o1 == 1 is un-allocated except for 32 and 64 bit sizes + */ static void disas_ldst_excl(DisasContext *s, uint32_t insn) { - unsupported_encoding(s, insn); + int rt = extract32(insn, 0, 5); + int rn = extract32(insn, 5, 5); + int rt2 = extract32(insn, 10, 5); + int is_lasr = extract32(insn, 15, 1); + int rs = extract32(insn, 16, 5); + int is_pair = extract32(insn, 21, 1); + int is_store = !extract32(insn, 22, 1); + int is_excl = !extract32(insn, 23, 1); + int size = extract32(insn, 30, 2); + TCGv_i64 tcg_addr; + + if ((!is_excl && !is_lasr) || + (is_pair && size < 2)) { + unallocated_encoding(s); + return; + } + + if (rn == 31) { + gen_check_sp_alignment(s); + } + tcg_addr = read_cpu_reg_sp(s, rn, 1); + + /* Note that since TCG is single threaded load-acquire/store-release + * semantics require no extra if (is_lasr) { ... } handling. + */ + + if (is_excl) { + if (!is_store) { + gen_load_exclusive(s, rt, rt2, tcg_addr, size, is_pair); + } else { + gen_store_exclusive(s, rs, rt, rt2, tcg_addr, size, is_pair); + } + } else { + TCGv_i64 tcg_rt = cpu_reg(s, rt); + if (is_store) { + do_gpr_st(s, tcg_rt, tcg_addr, size); + } else { + do_gpr_ld(s, tcg_rt, tcg_addr, size, false, false); + } + if (is_pair) { + TCGv_i64 tcg_rt2 = cpu_reg(s, rt); + tcg_gen_addi_i64(tcg_addr, tcg_addr, 1 << size); + if (is_store) { + do_gpr_st(s, tcg_rt2, tcg_addr, size); + } else { + do_gpr_ld(s, tcg_rt2, tcg_addr, size, false, false); + } + } + } } /* From 749c7970d00cdb2be71ce66d65b9eb2ade668832 Mon Sep 17 00:00:00 2001 From: Claudio Fontana Date: Sat, 4 Jan 2014 22:15:47 +0000 Subject: [PATCH 07/61] linux-user: AArch64: define TARGET_CLONE_BACKWARDS The AArch64 linux-user support was written before but merged after commit 4ce6243dc621 which cleaned up the handling of the clone() syscall argument order, so we failed to notice that AArch64 also needs TARGET_CLONE_BACKWARDS to be defined. Add this define so that clone and fork syscalls work correctly. Signed-off-by: Claudio Fontana Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- linux-user/aarch64/syscall.h | 1 + 1 file changed, 1 insertion(+) diff --git a/linux-user/aarch64/syscall.h b/linux-user/aarch64/syscall.h index aef419efeb..18f44a8a40 100644 --- a/linux-user/aarch64/syscall.h +++ b/linux-user/aarch64/syscall.h @@ -7,3 +7,4 @@ struct target_pt_regs { #define UNAME_MACHINE "aarch64" #define UNAME_MINIMUM_RELEASE "3.8.0" +#define TARGET_CLONE_BACKWARDS From e0350d0c17d31b276812e659d8bdf9fc48a59951 Mon Sep 17 00:00:00 2001 From: Will Newton Date: Sat, 4 Jan 2014 22:15:48 +0000 Subject: [PATCH 08/61] linux-user: AArch64: Use correct values for FPSR/FPCR in sigcontext Use the helpers provided for getting the correct FPSR and FPCR values for the signal context. Signed-off-by: Will Newton Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- linux-user/signal.c | 10 +++++++--- 1 file changed, 7 insertions(+), 3 deletions(-) diff --git a/linux-user/signal.c b/linux-user/signal.c index 4e7148a2d6..6c74b18196 100644 --- a/linux-user/signal.c +++ b/linux-user/signal.c @@ -1189,8 +1189,8 @@ static int target_setup_sigframe(struct target_rt_sigframe *sf, __put_user(env->vfp.regs[i * 2 + 1], &aux->fpsimd.vregs[i * 2 + 1]); #endif } - __put_user(/*env->fpsr*/0, &aux->fpsimd.fpsr); - __put_user(/*env->fpcr*/0, &aux->fpsimd.fpcr); + __put_user(vfp_get_fpsr(env), &aux->fpsimd.fpsr); + __put_user(vfp_get_fpcr(env), &aux->fpsimd.fpcr); __put_user(TARGET_FPSIMD_MAGIC, &aux->fpsimd.head.magic); __put_user(sizeof(struct target_fpsimd_context), &aux->fpsimd.head.size); @@ -1209,7 +1209,7 @@ static int target_restore_sigframe(CPUARMState *env, int i; struct target_aux_context *aux = (struct target_aux_context *)sf->uc.tuc_mcontext.__reserved; - uint32_t magic, size; + uint32_t magic, size, fpsr, fpcr; uint64_t pstate; target_to_host_sigset(&set, &sf->uc.tuc_sigmask); @@ -1235,6 +1235,10 @@ static int target_restore_sigframe(CPUARMState *env, for (i = 0; i < 32 * 2; i++) { __get_user(env->vfp.regs[i], &aux->fpsimd.vregs[i]); } + __get_user(fpsr, &aux->fpsimd.fpsr); + vfp_set_fpsr(env, fpsr); + __get_user(fpcr, &aux->fpsimd.fpcr); + vfp_set_fpcr(env, fpcr); return 0; } From 2d123ad256fb47bac33c38f90dafb4280879aee5 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Alex=20Benn=C3=A9e?= Date: Sat, 4 Jan 2014 22:15:48 +0000 Subject: [PATCH 09/61] .travis.yml: Add aarch64-* targets MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Now the AArch64 targets are in mainline we can include them in our Travis test matrix. Signed-off-by: Alex Bennée Signed-off-by: Peter Maydell --- .travis.yml | 1 + 1 file changed, 1 insertion(+) diff --git a/.travis.yml b/.travis.yml index 90f167630a..c7ff4da29c 100644 --- a/.travis.yml +++ b/.travis.yml @@ -16,6 +16,7 @@ env: matrix: - TARGETS=alpha-softmmu,alpha-linux-user - TARGETS=arm-softmmu,arm-linux-user + - TARGETS=aarch64-softmmu,aarch64-linux-user - TARGETS=cris-softmmu - TARGETS=i386-softmmu,x86_64-softmmu - TARGETS=lm32-softmmu From 0f64fdaf7bd11a9165cc4b7461fb5a316819d257 Mon Sep 17 00:00:00 2001 From: Peter Maydell Date: Sat, 4 Jan 2014 22:15:48 +0000 Subject: [PATCH 10/61] default-configs: Add config for aarch64-linux-user Add a config for aarch64-linux-user, thereby enabling it as a valid target. Signed-off-by: Peter Maydell Signed-off-by: Alexander Graf Reviewed-by: Richard Henderson --- default-configs/aarch64-linux-user.mak | 3 +++ 1 file changed, 3 insertions(+) create mode 100644 default-configs/aarch64-linux-user.mak diff --git a/default-configs/aarch64-linux-user.mak b/default-configs/aarch64-linux-user.mak new file mode 100644 index 0000000000..3df7de5b8f --- /dev/null +++ b/default-configs/aarch64-linux-user.mak @@ -0,0 +1,3 @@ +# Default configuration for aarch64-linux-user + +CONFIG_GDBSTUB_XML=y From 83d210fb670862e9df5ca19d260fd7628cd67843 Mon Sep 17 00:00:00 2001 From: Alexander Graf Date: Sat, 4 Jan 2014 22:15:49 +0000 Subject: [PATCH 11/61] target-arm: A64: Add support for dumping AArch64 VFP register state When dumping the current CPU state, we can also get a request to dump the FPU state along with the CPU's integer state. Add support to dump the VFP state when that flag is set, so that we can properly debug code that modifies floating point registers. Signed-off-by: Alexander Graf [WN: Commit message tweak, rebased. Output all registers, two per-line.] Signed-off-by: Will Newton Reviewed-by: Peter Maydell Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- target-arm/translate-a64.c | 16 ++++++++++++++++ 1 file changed, 16 insertions(+) diff --git a/target-arm/translate-a64.c b/target-arm/translate-a64.c index 40c6fc4a9b..326f36d589 100644 --- a/target-arm/translate-a64.c +++ b/target-arm/translate-a64.c @@ -119,6 +119,22 @@ void aarch64_cpu_dump_state(CPUState *cs, FILE *f, psr & PSTATE_C ? 'C' : '-', psr & PSTATE_V ? 'V' : '-'); cpu_fprintf(f, "\n"); + + if (flags & CPU_DUMP_FPU) { + int numvfpregs = 32; + for (i = 0; i < numvfpregs; i += 2) { + uint64_t vlo = float64_val(env->vfp.regs[i * 2]); + uint64_t vhi = float64_val(env->vfp.regs[(i * 2) + 1]); + cpu_fprintf(f, "q%02d=%016" PRIx64 ":%016" PRIx64 " ", + i, vhi, vlo); + vlo = float64_val(env->vfp.regs[(i + 1) * 2]); + vhi = float64_val(env->vfp.regs[((i + 1) * 2) + 1]); + cpu_fprintf(f, "q%02d=%016" PRIx64 ":%016" PRIx64 "\n", + i + 1, vhi, vlo); + } + cpu_fprintf(f, "FPCR: %08x FPSR: %08x\n", + vfp_get_fpcr(env), vfp_get_fpsr(env)); + } } static int get_mem_index(DisasContext *s) From 76afb0d5967ebee59efbf3bfc6a2a042f3cb06f6 Mon Sep 17 00:00:00 2001 From: Peter Maydell Date: Sat, 4 Jan 2014 22:15:49 +0000 Subject: [PATCH 12/61] target-arm: A64: Fix vector register access on bigendian hosts The A64 128 bit vector registers are stored as a pair of uint64_t values in the register array. This means that if we're directly loading or storing a value of size less than 64 bits we must adjust the offset appropriately to account for whether the host is bigendian or not. Provide utility functions to abstract away the offsetof() calculations for the FP registers. For do_fp_st() we can sidestep most of the issues for 64 bit and smaller reg-to-mem transfers by always doing a 64 bit load from the register and writing just the piece we need to memory. Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- target-arm/translate-a64.c | 69 +++++++++++++++++++------------------- 1 file changed, 35 insertions(+), 34 deletions(-) diff --git a/target-arm/translate-a64.c b/target-arm/translate-a64.c index 326f36d589..ba9573a57e 100644 --- a/target-arm/translate-a64.c +++ b/target-arm/translate-a64.c @@ -308,6 +308,26 @@ static TCGv_i64 read_cpu_reg_sp(DisasContext *s, int reg, int sf) return v; } +/* Return the offset into CPUARMState of a slice (from + * the least significant end) of FP register Qn (ie + * Dn, Sn, Hn or Bn). + * (Note that this is not the same mapping as for A32; see cpu.h) + */ +static inline int fp_reg_offset(int regno, TCGMemOp size) +{ + int offs = offsetof(CPUARMState, vfp.regs[regno * 2]); +#ifdef HOST_WORDS_BIGENDIAN + offs += (8 - (1 << size)); +#endif + return offs; +} + +/* Offset of the high half of the 128 bit vector Qn */ +static inline int fp_reg_hi_offset(int regno) +{ + return offsetof(CPUARMState, vfp.regs[regno * 2 + 1]); +} + /* Set ZF and NF based on a 64 bit result. This is alas fiddlier * than the 32 bit equivalent. */ @@ -538,31 +558,15 @@ static void do_gpr_ld(DisasContext *s, TCGv_i64 dest, TCGv_i64 tcg_addr, static void do_fp_st(DisasContext *s, int srcidx, TCGv_i64 tcg_addr, int size) { /* This writes the bottom N bits of a 128 bit wide vector to memory */ - int freg_offs = offsetof(CPUARMState, vfp.regs[srcidx * 2]); TCGv_i64 tmp = tcg_temp_new_i64(); - + tcg_gen_ld_i64(tmp, cpu_env, fp_reg_offset(srcidx, MO_64)); if (size < 4) { - switch (size) { - case 0: - tcg_gen_ld8u_i64(tmp, cpu_env, freg_offs); - break; - case 1: - tcg_gen_ld16u_i64(tmp, cpu_env, freg_offs); - break; - case 2: - tcg_gen_ld32u_i64(tmp, cpu_env, freg_offs); - break; - case 3: - tcg_gen_ld_i64(tmp, cpu_env, freg_offs); - break; - } tcg_gen_qemu_st_i64(tmp, tcg_addr, get_mem_index(s), MO_TE + size); } else { TCGv_i64 tcg_hiaddr = tcg_temp_new_i64(); - tcg_gen_ld_i64(tmp, cpu_env, freg_offs); tcg_gen_qemu_st_i64(tmp, tcg_addr, get_mem_index(s), MO_TEQ); tcg_gen_qemu_st64(tmp, tcg_addr, get_mem_index(s)); - tcg_gen_ld_i64(tmp, cpu_env, freg_offs + sizeof(float64)); + tcg_gen_ld_i64(tmp, cpu_env, fp_reg_hi_offset(srcidx)); tcg_gen_addi_i64(tcg_hiaddr, tcg_addr, 8); tcg_gen_qemu_st_i64(tmp, tcg_hiaddr, get_mem_index(s), MO_TEQ); tcg_temp_free_i64(tcg_hiaddr); @@ -577,7 +581,6 @@ static void do_fp_st(DisasContext *s, int srcidx, TCGv_i64 tcg_addr, int size) static void do_fp_ld(DisasContext *s, int destidx, TCGv_i64 tcg_addr, int size) { /* This always zero-extends and writes to a full 128 bit wide vector */ - int freg_offs = offsetof(CPUARMState, vfp.regs[destidx * 2]); TCGv_i64 tmplo = tcg_temp_new_i64(); TCGv_i64 tmphi; @@ -596,8 +599,8 @@ static void do_fp_ld(DisasContext *s, int destidx, TCGv_i64 tcg_addr, int size) tcg_temp_free_i64(tcg_hiaddr); } - tcg_gen_st_i64(tmplo, cpu_env, freg_offs); - tcg_gen_st_i64(tmphi, cpu_env, freg_offs + sizeof(float64)); + tcg_gen_st_i64(tmplo, cpu_env, fp_reg_offset(destidx, MO_64)); + tcg_gen_st_i64(tmphi, cpu_env, fp_reg_hi_offset(destidx)); tcg_temp_free_i64(tmplo); tcg_temp_free_i64(tmphi); @@ -3224,7 +3227,6 @@ static void handle_fmov(DisasContext *s, int rd, int rn, int type, bool itof) */ if (itof) { - int freg_offs = offsetof(CPUARMState, vfp.regs[rd * 2]); TCGv_i64 tcg_rn = cpu_reg(s, rn); switch (type) { @@ -3233,9 +3235,9 @@ static void handle_fmov(DisasContext *s, int rd, int rn, int type, bool itof) /* 32 bit */ TCGv_i64 tmp = tcg_temp_new_i64(); tcg_gen_ext32u_i64(tmp, tcg_rn); - tcg_gen_st_i64(tmp, cpu_env, freg_offs); + tcg_gen_st_i64(tmp, cpu_env, fp_reg_offset(rd, MO_64)); tcg_gen_movi_i64(tmp, 0); - tcg_gen_st_i64(tmp, cpu_env, freg_offs + sizeof(float64)); + tcg_gen_st_i64(tmp, cpu_env, fp_reg_hi_offset(rd)); tcg_temp_free_i64(tmp); break; } @@ -3243,32 +3245,31 @@ static void handle_fmov(DisasContext *s, int rd, int rn, int type, bool itof) { /* 64 bit */ TCGv_i64 tmp = tcg_const_i64(0); - tcg_gen_st_i64(tcg_rn, cpu_env, freg_offs); - tcg_gen_st_i64(tmp, cpu_env, freg_offs + sizeof(float64)); + tcg_gen_st_i64(tcg_rn, cpu_env, fp_reg_offset(rd, MO_64)); + tcg_gen_st_i64(tmp, cpu_env, fp_reg_hi_offset(rd)); tcg_temp_free_i64(tmp); break; } case 2: /* 64 bit to top half. */ - tcg_gen_st_i64(tcg_rn, cpu_env, freg_offs + sizeof(float64)); + tcg_gen_st_i64(tcg_rn, cpu_env, fp_reg_hi_offset(rd)); break; } } else { - int freg_offs = offsetof(CPUARMState, vfp.regs[rn * 2]); TCGv_i64 tcg_rd = cpu_reg(s, rd); switch (type) { case 0: /* 32 bit */ - tcg_gen_ld32u_i64(tcg_rd, cpu_env, freg_offs); + tcg_gen_ld32u_i64(tcg_rd, cpu_env, fp_reg_offset(rn, MO_32)); + break; + case 1: + /* 64 bit */ + tcg_gen_ld_i64(tcg_rd, cpu_env, fp_reg_offset(rn, MO_64)); break; case 2: /* 64 bits from top half */ - freg_offs += sizeof(float64); - /* fall through */ - case 1: - /* 64 bit */ - tcg_gen_ld_i64(tcg_rd, cpu_env, freg_offs); + tcg_gen_ld_i64(tcg_rd, cpu_env, fp_reg_hi_offset(rn)); break; } } From 286cb8b9cb04048b940bd9f385404781a42c6cff Mon Sep 17 00:00:00 2001 From: Peter Maydell Date: Sat, 4 Jan 2014 22:15:49 +0000 Subject: [PATCH 13/61] target-arm: Use VFP_BINOP macro for min, max, minnum, maxnum Use the VFP_BINOP macro to provide helpers for min, max, minnum and maxnum, rather than hand-rolling them. (The float64 max version is not used by A32 but will be needed for A64.) Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- target-arm/helper.c | 29 ++++------------------------- target-arm/helper.h | 15 ++++++++------- target-arm/neon_helper.c | 12 ------------ target-arm/translate.c | 16 ++++++++-------- 4 files changed, 20 insertions(+), 52 deletions(-) diff --git a/target-arm/helper.c b/target-arm/helper.c index f91e3fd35d..d0b7c27749 100644 --- a/target-arm/helper.c +++ b/target-arm/helper.c @@ -3864,6 +3864,10 @@ VFP_BINOP(add) VFP_BINOP(sub) VFP_BINOP(mul) VFP_BINOP(div) +VFP_BINOP(min) +VFP_BINOP(max) +VFP_BINOP(minnum) +VFP_BINOP(maxnum) #undef VFP_BINOP float32 VFP_HELPER(neg, s)(float32 a) @@ -4317,28 +4321,3 @@ float64 VFP_HELPER(muladd, d)(float64 a, float64 b, float64 c, void *fpstp) float_status *fpst = fpstp; return float64_muladd(a, b, c, 0, fpst); } - -/* ARMv8 VMAXNM/VMINNM */ -float32 VFP_HELPER(maxnm, s)(float32 a, float32 b, void *fpstp) -{ - float_status *fpst = fpstp; - return float32_maxnum(a, b, fpst); -} - -float64 VFP_HELPER(maxnm, d)(float64 a, float64 b, void *fpstp) -{ - float_status *fpst = fpstp; - return float64_maxnum(a, b, fpst); -} - -float32 VFP_HELPER(minnm, s)(float32 a, float32 b, void *fpstp) -{ - float_status *fpst = fpstp; - return float32_minnum(a, b, fpst); -} - -float64 VFP_HELPER(minnm, d)(float64 a, float64 b, void *fpstp) -{ - float_status *fpst = fpstp; - return float64_minnum(a, b, fpst); -} diff --git a/target-arm/helper.h b/target-arm/helper.h index 73d67dcc17..dd1160ed61 100644 --- a/target-arm/helper.h +++ b/target-arm/helper.h @@ -79,6 +79,14 @@ DEF_HELPER_3(vfp_muls, f32, f32, f32, ptr) DEF_HELPER_3(vfp_muld, f64, f64, f64, ptr) DEF_HELPER_3(vfp_divs, f32, f32, f32, ptr) DEF_HELPER_3(vfp_divd, f64, f64, f64, ptr) +DEF_HELPER_3(vfp_maxs, f32, f32, f32, ptr) +DEF_HELPER_3(vfp_maxd, f64, f64, f64, ptr) +DEF_HELPER_3(vfp_mins, f32, f32, f32, ptr) +DEF_HELPER_3(vfp_mind, f64, f64, f64, ptr) +DEF_HELPER_3(vfp_maxnums, f32, f32, f32, ptr) +DEF_HELPER_3(vfp_maxnumd, f64, f64, f64, ptr) +DEF_HELPER_3(vfp_minnums, f32, f32, f32, ptr) +DEF_HELPER_3(vfp_minnumd, f64, f64, f64, ptr) DEF_HELPER_1(vfp_negs, f32, f32) DEF_HELPER_1(vfp_negd, f64, f64) DEF_HELPER_1(vfp_abss, f32, f32) @@ -132,11 +140,6 @@ DEF_HELPER_2(neon_fcvt_f32_to_f16, i32, f32, env) DEF_HELPER_4(vfp_muladdd, f64, f64, f64, f64, ptr) DEF_HELPER_4(vfp_muladds, f32, f32, f32, f32, ptr) -DEF_HELPER_3(vfp_maxnmd, f64, f64, f64, ptr) -DEF_HELPER_3(vfp_maxnms, f32, f32, f32, ptr) -DEF_HELPER_3(vfp_minnmd, f64, f64, f64, ptr) -DEF_HELPER_3(vfp_minnms, f32, f32, f32, ptr) - DEF_HELPER_3(recps_f32, f32, f32, f32, env) DEF_HELPER_3(rsqrts_f32, f32, f32, f32, env) DEF_HELPER_2(recpe_f32, f32, f32, env) @@ -346,8 +349,6 @@ DEF_HELPER_2(neon_qneg_s8, i32, env, i32) DEF_HELPER_2(neon_qneg_s16, i32, env, i32) DEF_HELPER_2(neon_qneg_s32, i32, env, i32) -DEF_HELPER_3(neon_min_f32, i32, i32, i32, ptr) -DEF_HELPER_3(neon_max_f32, i32, i32, i32, ptr) DEF_HELPER_3(neon_abd_f32, i32, i32, i32, ptr) DEF_HELPER_3(neon_ceq_f32, i32, i32, i32, ptr) DEF_HELPER_3(neon_cge_f32, i32, i32, i32, ptr) diff --git a/target-arm/neon_helper.c b/target-arm/neon_helper.c index b028cc2c93..be6fbd997e 100644 --- a/target-arm/neon_helper.c +++ b/target-arm/neon_helper.c @@ -1765,18 +1765,6 @@ uint32_t HELPER(neon_qneg_s32)(CPUARMState *env, uint32_t x) } /* NEON Float helpers. */ -uint32_t HELPER(neon_min_f32)(uint32_t a, uint32_t b, void *fpstp) -{ - float_status *fpst = fpstp; - return float32_val(float32_min(make_float32(a), make_float32(b), fpst)); -} - -uint32_t HELPER(neon_max_f32)(uint32_t a, uint32_t b, void *fpstp) -{ - float_status *fpst = fpstp; - return float32_val(float32_max(make_float32(a), make_float32(b), fpst)); -} - uint32_t HELPER(neon_abd_f32)(uint32_t a, uint32_t b, void *fpstp) { float_status *fpst = fpstp; diff --git a/target-arm/translate.c b/target-arm/translate.c index 4387547a9c..d04fc9ff41 100644 --- a/target-arm/translate.c +++ b/target-arm/translate.c @@ -2725,9 +2725,9 @@ static int handle_vminmaxnm(uint32_t insn, uint32_t rd, uint32_t rn, tcg_gen_ld_f64(frn, cpu_env, vfp_reg_offset(dp, rn)); tcg_gen_ld_f64(frm, cpu_env, vfp_reg_offset(dp, rm)); if (vmin) { - gen_helper_vfp_minnmd(dest, frn, frm, fpst); + gen_helper_vfp_minnumd(dest, frn, frm, fpst); } else { - gen_helper_vfp_maxnmd(dest, frn, frm, fpst); + gen_helper_vfp_maxnumd(dest, frn, frm, fpst); } tcg_gen_st_f64(dest, cpu_env, vfp_reg_offset(dp, rd)); tcg_temp_free_i64(frn); @@ -2743,9 +2743,9 @@ static int handle_vminmaxnm(uint32_t insn, uint32_t rd, uint32_t rn, tcg_gen_ld_f32(frn, cpu_env, vfp_reg_offset(dp, rn)); tcg_gen_ld_f32(frm, cpu_env, vfp_reg_offset(dp, rm)); if (vmin) { - gen_helper_vfp_minnms(dest, frn, frm, fpst); + gen_helper_vfp_minnums(dest, frn, frm, fpst); } else { - gen_helper_vfp_maxnms(dest, frn, frm, fpst); + gen_helper_vfp_maxnums(dest, frn, frm, fpst); } tcg_gen_st_f32(dest, cpu_env, vfp_reg_offset(dp, rd)); tcg_temp_free_i32(frn); @@ -5121,9 +5121,9 @@ static int disas_neon_data_insn(CPUARMState * env, DisasContext *s, uint32_t ins { TCGv_ptr fpstatus = get_fpstatus_ptr(1); if (size == 0) { - gen_helper_neon_max_f32(tmp, tmp, tmp2, fpstatus); + gen_helper_vfp_maxs(tmp, tmp, tmp2, fpstatus); } else { - gen_helper_neon_min_f32(tmp, tmp, tmp2, fpstatus); + gen_helper_vfp_mins(tmp, tmp, tmp2, fpstatus); } tcg_temp_free_ptr(fpstatus); break; @@ -5133,9 +5133,9 @@ static int disas_neon_data_insn(CPUARMState * env, DisasContext *s, uint32_t ins /* VMAXNM/VMINNM */ TCGv_ptr fpstatus = get_fpstatus_ptr(1); if (size == 0) { - gen_helper_vfp_maxnms(tmp, tmp, tmp2, fpstatus); + gen_helper_vfp_maxnums(tmp, tmp, tmp2, fpstatus); } else { - gen_helper_vfp_minnms(tmp, tmp, tmp2, fpstatus); + gen_helper_vfp_minnums(tmp, tmp, tmp2, fpstatus); } tcg_temp_free_ptr(fpstatus); } else { From 459cc34290b2823403f9820345c8a83f66495e1d Mon Sep 17 00:00:00 2001 From: Alexander Graf Date: Sat, 4 Jan 2014 22:15:50 +0000 Subject: [PATCH 14/61] target-arm: A64: Add "Floating-point data-processing (2 source)" insns This patch adds emulation for the "Floating-point data-processing (2 source)" group of instructions. Signed-off-by: Alexander Graf [WN: Commit message tweak, merge single and double precision patches. Rebase and update to new infrastructure. Incorporate FMIN/FMAX support patch by Michael Matz.] Signed-off-by: Will Newton [PMM: * added convenience accessors for FP s and d regs * pulled the field decode and opcode validity check up a level] Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- target-arm/translate-a64.c | 182 ++++++++++++++++++++++++++++++++++++- 1 file changed, 181 insertions(+), 1 deletion(-) diff --git a/target-arm/translate-a64.c b/target-arm/translate-a64.c index ba9573a57e..c406e2a6fb 100644 --- a/target-arm/translate-a64.c +++ b/target-arm/translate-a64.c @@ -328,6 +328,60 @@ static inline int fp_reg_hi_offset(int regno) return offsetof(CPUARMState, vfp.regs[regno * 2 + 1]); } +/* Convenience accessors for reading and writing single and double + * FP registers. Writing clears the upper parts of the associated + * 128 bit vector register, as required by the architecture. + * Note that unlike the GP register accessors, the values returned + * by the read functions must be manually freed. + */ +static TCGv_i64 read_fp_dreg(DisasContext *s, int reg) +{ + TCGv_i64 v = tcg_temp_new_i64(); + + tcg_gen_ld_i64(v, cpu_env, fp_reg_offset(reg, MO_64)); + return v; +} + +static TCGv_i32 read_fp_sreg(DisasContext *s, int reg) +{ + TCGv_i32 v = tcg_temp_new_i32(); + + tcg_gen_ld_i32(v, cpu_env, fp_reg_offset(reg, MO_32)); + return v; +} + +static void write_fp_dreg(DisasContext *s, int reg, TCGv_i64 v) +{ + TCGv_i64 tcg_zero = tcg_const_i64(0); + + tcg_gen_st_i64(v, cpu_env, fp_reg_offset(reg, MO_64)); + tcg_gen_st_i64(tcg_zero, cpu_env, fp_reg_hi_offset(reg)); + tcg_temp_free_i64(tcg_zero); +} + +static void write_fp_sreg(DisasContext *s, int reg, TCGv_i32 v) +{ + TCGv_i64 tmp = tcg_temp_new_i64(); + + tcg_gen_extu_i32_i64(tmp, v); + write_fp_dreg(s, reg, tmp); + tcg_temp_free_i64(tmp); +} + +static TCGv_ptr get_fpstatus_ptr(void) +{ + TCGv_ptr statusptr = tcg_temp_new_ptr(); + int offset; + + /* In A64 all instructions (both FP and Neon) use the FPCR; + * there is no equivalent of the A32 Neon "standard FPSCR value" + * and all operations use vfp.fp_status. + */ + offset = offsetof(CPUARMState, vfp.fp_status); + tcg_gen_addi_ptr(statusptr, cpu_env, offset); + return statusptr; +} + /* Set ZF and NF based on a 64 bit result. This is alas fiddlier * than the 32 bit equivalent. */ @@ -3176,6 +3230,112 @@ static void disas_fp_1src(DisasContext *s, uint32_t insn) unsupported_encoding(s, insn); } +/* C3.6.26 Floating-point data-processing (2 source) - single precision */ +static void handle_fp_2src_single(DisasContext *s, int opcode, + int rd, int rn, int rm) +{ + TCGv_i32 tcg_op1; + TCGv_i32 tcg_op2; + TCGv_i32 tcg_res; + TCGv_ptr fpst; + + tcg_res = tcg_temp_new_i32(); + fpst = get_fpstatus_ptr(); + tcg_op1 = read_fp_sreg(s, rn); + tcg_op2 = read_fp_sreg(s, rm); + + switch (opcode) { + case 0x0: /* FMUL */ + gen_helper_vfp_muls(tcg_res, tcg_op1, tcg_op2, fpst); + break; + case 0x1: /* FDIV */ + gen_helper_vfp_divs(tcg_res, tcg_op1, tcg_op2, fpst); + break; + case 0x2: /* FADD */ + gen_helper_vfp_adds(tcg_res, tcg_op1, tcg_op2, fpst); + break; + case 0x3: /* FSUB */ + gen_helper_vfp_subs(tcg_res, tcg_op1, tcg_op2, fpst); + break; + case 0x4: /* FMAX */ + gen_helper_vfp_maxs(tcg_res, tcg_op1, tcg_op2, fpst); + break; + case 0x5: /* FMIN */ + gen_helper_vfp_mins(tcg_res, tcg_op1, tcg_op2, fpst); + break; + case 0x6: /* FMAXNM */ + gen_helper_vfp_maxnums(tcg_res, tcg_op1, tcg_op2, fpst); + break; + case 0x7: /* FMINNM */ + gen_helper_vfp_minnums(tcg_res, tcg_op1, tcg_op2, fpst); + break; + case 0x8: /* FNMUL */ + gen_helper_vfp_muls(tcg_res, tcg_op1, tcg_op2, fpst); + gen_helper_vfp_negs(tcg_res, tcg_res); + break; + } + + write_fp_sreg(s, rd, tcg_res); + + tcg_temp_free_ptr(fpst); + tcg_temp_free_i32(tcg_op1); + tcg_temp_free_i32(tcg_op2); + tcg_temp_free_i32(tcg_res); +} + +/* C3.6.26 Floating-point data-processing (2 source) - double precision */ +static void handle_fp_2src_double(DisasContext *s, int opcode, + int rd, int rn, int rm) +{ + TCGv_i64 tcg_op1; + TCGv_i64 tcg_op2; + TCGv_i64 tcg_res; + TCGv_ptr fpst; + + tcg_res = tcg_temp_new_i64(); + fpst = get_fpstatus_ptr(); + tcg_op1 = read_fp_dreg(s, rn); + tcg_op2 = read_fp_dreg(s, rm); + + switch (opcode) { + case 0x0: /* FMUL */ + gen_helper_vfp_muld(tcg_res, tcg_op1, tcg_op2, fpst); + break; + case 0x1: /* FDIV */ + gen_helper_vfp_divd(tcg_res, tcg_op1, tcg_op2, fpst); + break; + case 0x2: /* FADD */ + gen_helper_vfp_addd(tcg_res, tcg_op1, tcg_op2, fpst); + break; + case 0x3: /* FSUB */ + gen_helper_vfp_subd(tcg_res, tcg_op1, tcg_op2, fpst); + break; + case 0x4: /* FMAX */ + gen_helper_vfp_maxd(tcg_res, tcg_op1, tcg_op2, fpst); + break; + case 0x5: /* FMIN */ + gen_helper_vfp_mind(tcg_res, tcg_op1, tcg_op2, fpst); + break; + case 0x6: /* FMAXNM */ + gen_helper_vfp_maxnumd(tcg_res, tcg_op1, tcg_op2, fpst); + break; + case 0x7: /* FMINNM */ + gen_helper_vfp_minnumd(tcg_res, tcg_op1, tcg_op2, fpst); + break; + case 0x8: /* FNMUL */ + gen_helper_vfp_muld(tcg_res, tcg_op1, tcg_op2, fpst); + gen_helper_vfp_negd(tcg_res, tcg_res); + break; + } + + write_fp_dreg(s, rd, tcg_res); + + tcg_temp_free_ptr(fpst); + tcg_temp_free_i64(tcg_op1); + tcg_temp_free_i64(tcg_op2); + tcg_temp_free_i64(tcg_res); +} + /* C3.6.26 Floating point data-processing (2 source) * 31 30 29 28 24 23 22 21 20 16 15 12 11 10 9 5 4 0 * +---+---+---+-----------+------+---+------+--------+-----+------+------+ @@ -3184,7 +3344,27 @@ static void disas_fp_1src(DisasContext *s, uint32_t insn) */ static void disas_fp_2src(DisasContext *s, uint32_t insn) { - unsupported_encoding(s, insn); + int type = extract32(insn, 22, 2); + int rd = extract32(insn, 0, 5); + int rn = extract32(insn, 5, 5); + int rm = extract32(insn, 16, 5); + int opcode = extract32(insn, 12, 4); + + if (opcode > 8) { + unallocated_encoding(s); + return; + } + + switch (type) { + case 0: + handle_fp_2src_single(s, opcode, rd, rn, rm); + break; + case 1: + handle_fp_2src_double(s, opcode, rd, rn, rm); + break; + default: + unallocated_encoding(s); + } } /* C3.6.27 Floating point data-processing (3 source) From cd54be5fdbded3346a94fcf71f4a1d9aed340d40 Mon Sep 17 00:00:00 2001 From: Alexander Graf Date: Sat, 4 Jan 2014 22:15:50 +0000 Subject: [PATCH 15/61] target-arm: A64: Add "Floating-point data-processing (3 source)" insns This patch adds emulation for the "Floating-point data-processing (3 source)" group of instructions. Signed-off-by: Alexander Graf [WN: Commit message tweak, merged single and double precision patches. Implement using muladd as suggested by Richard Henderson.] Signed-off-by: Will Newton [PMM: pull field decode up a level, use register accessors] Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- target-arm/translate-a64.c | 95 +++++++++++++++++++++++++++++++++++++- 1 file changed, 94 insertions(+), 1 deletion(-) diff --git a/target-arm/translate-a64.c b/target-arm/translate-a64.c index c406e2a6fb..84497dc308 100644 --- a/target-arm/translate-a64.c +++ b/target-arm/translate-a64.c @@ -3367,6 +3367,82 @@ static void disas_fp_2src(DisasContext *s, uint32_t insn) } } +/* C3.6.27 Floating-point data-processing (3 source) - single precision */ +static void handle_fp_3src_single(DisasContext *s, bool o0, bool o1, + int rd, int rn, int rm, int ra) +{ + TCGv_i32 tcg_op1, tcg_op2, tcg_op3; + TCGv_i32 tcg_res = tcg_temp_new_i32(); + TCGv_ptr fpst = get_fpstatus_ptr(); + + tcg_op1 = read_fp_sreg(s, rn); + tcg_op2 = read_fp_sreg(s, rm); + tcg_op3 = read_fp_sreg(s, ra); + + /* These are fused multiply-add, and must be done as one + * floating point operation with no rounding between the + * multiplication and addition steps. + * NB that doing the negations here as separate steps is + * correct : an input NaN should come out with its sign bit + * flipped if it is a negated-input. + */ + if (o1 == true) { + gen_helper_vfp_negs(tcg_op3, tcg_op3); + } + + if (o0 != o1) { + gen_helper_vfp_negs(tcg_op1, tcg_op1); + } + + gen_helper_vfp_muladds(tcg_res, tcg_op1, tcg_op2, tcg_op3, fpst); + + write_fp_sreg(s, rd, tcg_res); + + tcg_temp_free_ptr(fpst); + tcg_temp_free_i32(tcg_op1); + tcg_temp_free_i32(tcg_op2); + tcg_temp_free_i32(tcg_op3); + tcg_temp_free_i32(tcg_res); +} + +/* C3.6.27 Floating-point data-processing (3 source) - double precision */ +static void handle_fp_3src_double(DisasContext *s, bool o0, bool o1, + int rd, int rn, int rm, int ra) +{ + TCGv_i64 tcg_op1, tcg_op2, tcg_op3; + TCGv_i64 tcg_res = tcg_temp_new_i64(); + TCGv_ptr fpst = get_fpstatus_ptr(); + + tcg_op1 = read_fp_dreg(s, rn); + tcg_op2 = read_fp_dreg(s, rm); + tcg_op3 = read_fp_dreg(s, ra); + + /* These are fused multiply-add, and must be done as one + * floating point operation with no rounding between the + * multiplication and addition steps. + * NB that doing the negations here as separate steps is + * correct : an input NaN should come out with its sign bit + * flipped if it is a negated-input. + */ + if (o1 == true) { + gen_helper_vfp_negd(tcg_op3, tcg_op3); + } + + if (o0 != o1) { + gen_helper_vfp_negd(tcg_op1, tcg_op1); + } + + gen_helper_vfp_muladdd(tcg_res, tcg_op1, tcg_op2, tcg_op3, fpst); + + write_fp_dreg(s, rd, tcg_res); + + tcg_temp_free_ptr(fpst); + tcg_temp_free_i64(tcg_op1); + tcg_temp_free_i64(tcg_op2); + tcg_temp_free_i64(tcg_op3); + tcg_temp_free_i64(tcg_res); +} + /* C3.6.27 Floating point data-processing (3 source) * 31 30 29 28 24 23 22 21 20 16 15 14 10 9 5 4 0 * +---+---+---+-----------+------+----+------+----+------+------+------+ @@ -3375,7 +3451,24 @@ static void disas_fp_2src(DisasContext *s, uint32_t insn) */ static void disas_fp_3src(DisasContext *s, uint32_t insn) { - unsupported_encoding(s, insn); + int type = extract32(insn, 22, 2); + int rd = extract32(insn, 0, 5); + int rn = extract32(insn, 5, 5); + int ra = extract32(insn, 10, 5); + int rm = extract32(insn, 16, 5); + bool o0 = extract32(insn, 15, 1); + bool o1 = extract32(insn, 21, 1); + + switch (type) { + case 0: + handle_fp_3src_single(s, o0, o1, rd, rn, rm, ra); + break; + case 1: + handle_fp_3src_double(s, o0, o1, rd, rn, rm, ra); + break; + default: + unallocated_encoding(s); + } } /* C3.6.28 Floating point immediate From 624490c33ccd803b730d3600d943472bfcc5fa59 Mon Sep 17 00:00:00 2001 From: Alexander Graf Date: Sat, 4 Jan 2014 22:15:50 +0000 Subject: [PATCH 16/61] target-arm: A64: Add fmov (scalar, immediate) instruction This patch adds emulation for the fmov instruction working on scalars with an immediate payload. Signed-off-by: Alexander Graf [WN: Commit message tweak, rebase and use new infrastructure.] Signed-off-by: Will Newton Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- target-arm/translate-a64.c | 32 +++++++++++++++++++++++++++++++- 1 file changed, 31 insertions(+), 1 deletion(-) diff --git a/target-arm/translate-a64.c b/target-arm/translate-a64.c index 84497dc308..bb36a66955 100644 --- a/target-arm/translate-a64.c +++ b/target-arm/translate-a64.c @@ -3479,7 +3479,37 @@ static void disas_fp_3src(DisasContext *s, uint32_t insn) */ static void disas_fp_imm(DisasContext *s, uint32_t insn) { - unsupported_encoding(s, insn); + int rd = extract32(insn, 0, 5); + int imm8 = extract32(insn, 13, 8); + int is_double = extract32(insn, 22, 2); + uint64_t imm; + TCGv_i64 tcg_res; + + if (is_double > 1) { + unallocated_encoding(s); + return; + } + + /* The imm8 encodes the sign bit, enough bits to represent + * an exponent in the range 01....1xx to 10....0xx, + * and the most significant 4 bits of the mantissa; see + * VFPExpandImm() in the v8 ARM ARM. + */ + if (is_double) { + imm = (extract32(imm8, 7, 1) ? 0x8000 : 0) | + (extract32(imm8, 6, 1) ? 0x3fc0 : 0x4000) | + extract32(imm8, 0, 6); + imm <<= 48; + } else { + imm = (extract32(imm8, 7, 1) ? 0x8000 : 0) | + (extract32(imm8, 6, 1) ? 0x3e00 : 0x4000) | + (extract32(imm8, 0, 6) << 3); + imm <<= 16; + } + + tcg_res = tcg_const_i64(imm); + write_fp_dreg(s, rd, tcg_res); + tcg_temp_free_i64(tcg_res); } /* C3.6.29 Floating point <-> fixed point conversions From 244976576d47543901c8d5ccb7f1f191f74b46f5 Mon Sep 17 00:00:00 2001 From: Claudio Fontana Date: Sat, 4 Jan 2014 22:15:50 +0000 Subject: [PATCH 17/61] target-arm: A64: Add support for floating point compare Add decoding support for C3.6.22 Floating-point compare. Signed-off-by: Claudio Fontana Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- target-arm/helper-a64.c | 45 ++++++++++++++++++++++++++ target-arm/helper-a64.h | 4 +++ target-arm/translate-a64.c | 65 +++++++++++++++++++++++++++++++++++++- 3 files changed, 113 insertions(+), 1 deletion(-) diff --git a/target-arm/helper-a64.c b/target-arm/helper-a64.c index d3f706754f..4ce0d01a85 100644 --- a/target-arm/helper-a64.c +++ b/target-arm/helper-a64.c @@ -77,3 +77,48 @@ uint64_t HELPER(rbit64)(uint64_t x) return x; } + +/* Convert a softfloat float_relation_ (as returned by + * the float*_compare functions) to the correct ARM + * NZCV flag state. + */ +static inline uint32_t float_rel_to_flags(int res) +{ + uint64_t flags; + switch (res) { + case float_relation_equal: + flags = PSTATE_Z | PSTATE_C; + break; + case float_relation_less: + flags = PSTATE_N; + break; + case float_relation_greater: + flags = PSTATE_C; + break; + case float_relation_unordered: + default: + flags = PSTATE_C | PSTATE_V; + break; + } + return flags; +} + +uint64_t HELPER(vfp_cmps_a64)(float32 x, float32 y, void *fp_status) +{ + return float_rel_to_flags(float32_compare_quiet(x, y, fp_status)); +} + +uint64_t HELPER(vfp_cmpes_a64)(float32 x, float32 y, void *fp_status) +{ + return float_rel_to_flags(float32_compare(x, y, fp_status)); +} + +uint64_t HELPER(vfp_cmpd_a64)(float64 x, float64 y, void *fp_status) +{ + return float_rel_to_flags(float64_compare_quiet(x, y, fp_status)); +} + +uint64_t HELPER(vfp_cmped_a64)(float64 x, float64 y, void *fp_status) +{ + return float_rel_to_flags(float64_compare(x, y, fp_status)); +} diff --git a/target-arm/helper-a64.h b/target-arm/helper-a64.h index a163a94322..bca19f3dea 100644 --- a/target-arm/helper-a64.h +++ b/target-arm/helper-a64.h @@ -22,3 +22,7 @@ DEF_HELPER_FLAGS_1(clz64, TCG_CALL_NO_RWG_SE, i64, i64) DEF_HELPER_FLAGS_1(cls64, TCG_CALL_NO_RWG_SE, i64, i64) DEF_HELPER_FLAGS_1(cls32, TCG_CALL_NO_RWG_SE, i32, i32) DEF_HELPER_FLAGS_1(rbit64, TCG_CALL_NO_RWG_SE, i64, i64) +DEF_HELPER_3(vfp_cmps_a64, i64, f32, f32, ptr) +DEF_HELPER_3(vfp_cmpes_a64, i64, f32, f32, ptr) +DEF_HELPER_3(vfp_cmpd_a64, i64, f64, f64, ptr) +DEF_HELPER_3(vfp_cmped_a64, i64, f64, f64, ptr) diff --git a/target-arm/translate-a64.c b/target-arm/translate-a64.c index bb36a66955..dc9cc1499f 100644 --- a/target-arm/translate-a64.c +++ b/target-arm/translate-a64.c @@ -3186,6 +3186,54 @@ static void disas_data_proc_reg(DisasContext *s, uint32_t insn) } } +static void handle_fp_compare(DisasContext *s, bool is_double, + unsigned int rn, unsigned int rm, + bool cmp_with_zero, bool signal_all_nans) +{ + TCGv_i64 tcg_flags = tcg_temp_new_i64(); + TCGv_ptr fpst = get_fpstatus_ptr(); + + if (is_double) { + TCGv_i64 tcg_vn, tcg_vm; + + tcg_vn = read_fp_dreg(s, rn); + if (cmp_with_zero) { + tcg_vm = tcg_const_i64(0); + } else { + tcg_vm = read_fp_dreg(s, rm); + } + if (signal_all_nans) { + gen_helper_vfp_cmped_a64(tcg_flags, tcg_vn, tcg_vm, fpst); + } else { + gen_helper_vfp_cmpd_a64(tcg_flags, tcg_vn, tcg_vm, fpst); + } + tcg_temp_free_i64(tcg_vn); + tcg_temp_free_i64(tcg_vm); + } else { + TCGv_i32 tcg_vn, tcg_vm; + + tcg_vn = read_fp_sreg(s, rn); + if (cmp_with_zero) { + tcg_vm = tcg_const_i32(0); + } else { + tcg_vm = read_fp_sreg(s, rm); + } + if (signal_all_nans) { + gen_helper_vfp_cmpes_a64(tcg_flags, tcg_vn, tcg_vm, fpst); + } else { + gen_helper_vfp_cmps_a64(tcg_flags, tcg_vn, tcg_vm, fpst); + } + tcg_temp_free_i32(tcg_vn); + tcg_temp_free_i32(tcg_vm); + } + + tcg_temp_free_ptr(fpst); + + gen_set_nzcv(tcg_flags); + + tcg_temp_free_i64(tcg_flags); +} + /* C3.6.22 Floating point compare * 31 30 29 28 24 23 22 21 20 16 15 14 13 10 9 5 4 0 * +---+---+---+-----------+------+---+------+-----+---------+------+-------+ @@ -3194,7 +3242,22 @@ static void disas_data_proc_reg(DisasContext *s, uint32_t insn) */ static void disas_fp_compare(DisasContext *s, uint32_t insn) { - unsupported_encoding(s, insn); + unsigned int mos, type, rm, op, rn, opc, op2r; + + mos = extract32(insn, 29, 3); + type = extract32(insn, 22, 2); /* 0 = single, 1 = double */ + rm = extract32(insn, 16, 5); + op = extract32(insn, 14, 2); + rn = extract32(insn, 5, 5); + opc = extract32(insn, 3, 2); + op2r = extract32(insn, 0, 3); + + if (mos || op || op2r || type > 1) { + unallocated_encoding(s); + return; + } + + handle_fp_compare(s, type, rn, rm, opc & 1, opc & 2); } /* C3.6.23 Floating point conditional compare From 32ca04a60fa1ab1b08bcdb9474af3ecf036f8f0f Mon Sep 17 00:00:00 2001 From: Claudio Fontana Date: Sat, 4 Jan 2014 22:15:51 +0000 Subject: [PATCH 18/61] target-arm: A64: Add support for floating point conditional compare This adds decoding support for C3.6.23 FP Conditional Compare. Signed-off-by: Claudio Fontana Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- target-arm/translate-a64.c | 35 ++++++++++++++++++++++++++++++++++- 1 file changed, 34 insertions(+), 1 deletion(-) diff --git a/target-arm/translate-a64.c b/target-arm/translate-a64.c index dc9cc1499f..44d8a09c4d 100644 --- a/target-arm/translate-a64.c +++ b/target-arm/translate-a64.c @@ -3268,7 +3268,40 @@ static void disas_fp_compare(DisasContext *s, uint32_t insn) */ static void disas_fp_ccomp(DisasContext *s, uint32_t insn) { - unsupported_encoding(s, insn); + unsigned int mos, type, rm, cond, rn, op, nzcv; + TCGv_i64 tcg_flags; + int label_continue = -1; + + mos = extract32(insn, 29, 3); + type = extract32(insn, 22, 2); /* 0 = single, 1 = double */ + rm = extract32(insn, 16, 5); + cond = extract32(insn, 12, 4); + rn = extract32(insn, 5, 5); + op = extract32(insn, 4, 1); + nzcv = extract32(insn, 0, 4); + + if (mos || type > 1) { + unallocated_encoding(s); + return; + } + + if (cond < 0x0e) { /* not always */ + int label_match = gen_new_label(); + label_continue = gen_new_label(); + arm_gen_test_cc(cond, label_match); + /* nomatch: */ + tcg_flags = tcg_const_i64(nzcv << 28); + gen_set_nzcv(tcg_flags); + tcg_temp_free_i64(tcg_flags); + tcg_gen_br(label_continue); + gen_set_label(label_match); + } + + handle_fp_compare(s, type, rn, rm, false, op); + + if (cond < 0x0e) { + gen_set_label(label_continue); + } } /* C3.6.24 Floating point conditional select From 7b2440e4384e59a3d5228cbd53b76089075e9e71 Mon Sep 17 00:00:00 2001 From: Claudio Fontana Date: Sat, 4 Jan 2014 22:15:51 +0000 Subject: [PATCH 19/61] target-arm: A64: Add support for floating point cond select This adds decoding support for C3.6.24 FP conditional select. Signed-off-by: Claudio Fontana Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- target-arm/translate-a64.c | 45 +++++++++++++++++++++++++++++++++++++- 1 file changed, 44 insertions(+), 1 deletion(-) diff --git a/target-arm/translate-a64.c b/target-arm/translate-a64.c index 44d8a09c4d..c9fbf0f0f7 100644 --- a/target-arm/translate-a64.c +++ b/target-arm/translate-a64.c @@ -3304,6 +3304,20 @@ static void disas_fp_ccomp(DisasContext *s, uint32_t insn) } } +/* copy src FP register to dst FP register; type specifies single or double */ +static void gen_mov_fp2fp(DisasContext *s, int type, int dst, int src) +{ + if (type) { + TCGv_i64 v = read_fp_dreg(s, src); + write_fp_dreg(s, dst, v); + tcg_temp_free_i64(v); + } else { + TCGv_i32 v = read_fp_sreg(s, src); + write_fp_sreg(s, dst, v); + tcg_temp_free_i32(v); + } +} + /* C3.6.24 Floating point conditional select * 31 30 29 28 24 23 22 21 20 16 15 12 11 10 9 5 4 0 * +---+---+---+-----------+------+---+------+------+-----+------+------+ @@ -3312,7 +3326,36 @@ static void disas_fp_ccomp(DisasContext *s, uint32_t insn) */ static void disas_fp_csel(DisasContext *s, uint32_t insn) { - unsupported_encoding(s, insn); + unsigned int mos, type, rm, cond, rn, rd; + int label_continue = -1; + + mos = extract32(insn, 29, 3); + type = extract32(insn, 22, 2); /* 0 = single, 1 = double */ + rm = extract32(insn, 16, 5); + cond = extract32(insn, 12, 4); + rn = extract32(insn, 5, 5); + rd = extract32(insn, 0, 5); + + if (mos || type > 1) { + unallocated_encoding(s); + return; + } + + if (cond < 0x0e) { /* not always */ + int label_match = gen_new_label(); + label_continue = gen_new_label(); + arm_gen_test_cc(cond, label_match); + /* nomatch: */ + gen_mov_fp2fp(s, type, rd, rm); + tcg_gen_br(label_continue); + gen_set_label(label_match); + } + + gen_mov_fp2fp(s, type, rd, rn); + + if (cond < 0x0e) { /* continue */ + gen_set_label(label_continue); + } } /* C3.6.25 Floating point data-processing (1 source) From f9dfa55857f4f172d4c23011d18f260c875e4bfb Mon Sep 17 00:00:00 2001 From: Alexander Graf Date: Sat, 4 Jan 2014 22:15:51 +0000 Subject: [PATCH 20/61] target-arm: Give the FPSCR rounding modes names When setting rounding modes we currently just hardcode the numeric values for rounding modes in a big switch statement. With AArch64 support coming, we will need to refer to these rounding modes at different places throughout the code though, so let's better give them names so we don't get confused by accident. Signed-off-by: Alexander Graf [WN: Commit message tweak, use names from ARM ARM.] Signed-off-by: Will Newton Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- target-arm/cpu.h | 9 +++++++++ target-arm/helper.c | 8 ++++---- 2 files changed, 13 insertions(+), 4 deletions(-) diff --git a/target-arm/cpu.h b/target-arm/cpu.h index 7084a74bdb..43ca5722e7 100644 --- a/target-arm/cpu.h +++ b/target-arm/cpu.h @@ -487,6 +487,15 @@ static inline void vfp_set_fpcr(CPUARMState *env, uint32_t val) vfp_set_fpscr(env, new_fpscr); } +enum arm_fprounding { + FPROUNDING_TIEEVEN, + FPROUNDING_POSINF, + FPROUNDING_NEGINF, + FPROUNDING_ZERO, + FPROUNDING_TIEAWAY, + FPROUNDING_ODD +}; + enum arm_cpu_mode { ARM_CPU_MODE_USR = 0x10, ARM_CPU_MODE_FIQ = 0x11, diff --git a/target-arm/helper.c b/target-arm/helper.c index d0b7c27749..9afec28a1f 100644 --- a/target-arm/helper.c +++ b/target-arm/helper.c @@ -3815,16 +3815,16 @@ void HELPER(vfp_set_fpscr)(CPUARMState *env, uint32_t val) if (changed & (3 << 22)) { i = (val >> 22) & 3; switch (i) { - case 0: + case FPROUNDING_TIEEVEN: i = float_round_nearest_even; break; - case 1: + case FPROUNDING_POSINF: i = float_round_up; break; - case 2: + case FPROUNDING_NEGINF: i = float_round_down; break; - case 3: + case FPROUNDING_ZERO: i = float_round_to_zero; break; } From 2c7fd04c9348abada0be331b04f9e5cbeeb2ac18 Mon Sep 17 00:00:00 2001 From: Peter Crosthwaite Date: Mon, 6 Jan 2014 10:16:37 +0000 Subject: [PATCH 21/61] char/cadence_uart: Mark struct fields as public/private As per current QOM conventions. Signed-off-by: Peter Crosthwaite Message-id: a1e31bd62e9709ffb9b3efc6c120f83f30b7a660.1388626249.git.peter.crosthwaite@xilinx.com Signed-off-by: Peter Maydell --- hw/char/cadence_uart.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/hw/char/cadence_uart.c b/hw/char/cadence_uart.c index f18db53bca..a7b2f2148c 100644 --- a/hw/char/cadence_uart.c +++ b/hw/char/cadence_uart.c @@ -110,7 +110,9 @@ #define CADENCE_UART(obj) OBJECT_CHECK(UartState, (obj), TYPE_CADENCE_UART) typedef struct { + /*< private >*/ SysBusDevice parent_obj; + /*< public >*/ MemoryRegion iomem; uint32_t r[R_MAX]; From 362f656878345e9e50b6f7c676a1961d9ecaa355 Mon Sep 17 00:00:00 2001 From: Peter Crosthwaite Date: Mon, 6 Jan 2014 10:16:37 +0000 Subject: [PATCH 22/61] char/cadence_uart: Add missing uart_update_state This should be rechecked on bus write accesses as such accesses may change the underlying state that generates the interrupt. Particular relevant for when the guest touches the interrupt status or mask. Signed-off-by: Peter Crosthwaite Message-id: 1c250cd61b7b8de492fbc8b79b8370958a56d83b.1388626249.git.peter.crosthwaite@xilinx.com Signed-off-by: Peter Maydell --- hw/char/cadence_uart.c | 1 + 1 file changed, 1 insertion(+) diff --git a/hw/char/cadence_uart.c b/hw/char/cadence_uart.c index a7b2f2148c..fb9db89ee9 100644 --- a/hw/char/cadence_uart.c +++ b/hw/char/cadence_uart.c @@ -403,6 +403,7 @@ static void uart_write(void *opaque, hwaddr offset, uart_parameters_setup(s); break; } + uart_update_status(s); } static uint64_t uart_read(void *opaque, hwaddr offset, From b56fab8847e5746085c46a5bf6b5903b3b9fe957 Mon Sep 17 00:00:00 2001 From: Peter Crosthwaite Date: Mon, 6 Jan 2014 10:16:38 +0000 Subject: [PATCH 23/61] char/cadence_uart: Fix reset. Don't reset the uart as an init step. Register the reset function as a proper reset fn instead. Signed-off-by: Peter Crosthwaite Message-id: d82cd2e65e5a6f8b6deeecb6cced61f0bf3f8c89.1388626249.git.peter.crosthwaite@xilinx.com Signed-off-by: Peter Maydell --- hw/char/cadence_uart.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) diff --git a/hw/char/cadence_uart.c b/hw/char/cadence_uart.c index fb9db89ee9..7edc1196f3 100644 --- a/hw/char/cadence_uart.c +++ b/hw/char/cadence_uart.c @@ -431,8 +431,10 @@ static const MemoryRegionOps uart_ops = { .endianness = DEVICE_NATIVE_ENDIAN, }; -static void cadence_uart_reset(UartState *s) +static void cadence_uart_reset(DeviceState *dev) { + UartState *s = CADENCE_UART(dev); + s->r[R_CR] = 0x00000128; s->r[R_IMR] = 0; s->r[R_CISR] = 0; @@ -465,8 +467,6 @@ static int cadence_uart_init(SysBusDevice *dev) s->chr = qemu_char_get_next_serial(); - cadence_uart_reset(s); - if (s->chr) { qemu_chr_add_handlers(s->chr, uart_can_receive, uart_receive, uart_event, s); @@ -508,6 +508,7 @@ static void cadence_uart_class_init(ObjectClass *klass, void *data) sdc->init = cadence_uart_init; dc->vmsd = &vmstate_cadence_uart; + dc->reset = cadence_uart_reset; } static const TypeInfo cadence_uart_info = { From 8998a3c0ceb174d1c6089361efe870bb4f40a27c Mon Sep 17 00:00:00 2001 From: Peter Crosthwaite Date: Mon, 6 Jan 2014 10:16:38 +0000 Subject: [PATCH 24/61] char/cadence_uart: s/r_fifo/rx_fifo Rename this field to match the many other uses of "rx". Xilinx docmentation (UG585) also refers to this as "RxFIFO". Signed-off-by: Peter Crosthwaite Message-id: 7386d7cee0ea175f7e53ed5ff045265528d34e32.1388626249.git.peter.crosthwaite@xilinx.com Signed-off-by: Peter Maydell --- hw/char/cadence_uart.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/hw/char/cadence_uart.c b/hw/char/cadence_uart.c index 7edc1196f3..d6abc5b985 100644 --- a/hw/char/cadence_uart.c +++ b/hw/char/cadence_uart.c @@ -116,7 +116,7 @@ typedef struct { MemoryRegion iomem; uint32_t r[R_MAX]; - uint8_t r_fifo[RX_FIFO_SIZE]; + uint8_t rx_fifo[RX_FIFO_SIZE]; uint32_t rx_wpos; uint32_t rx_count; uint64_t char_tx_time; @@ -280,7 +280,7 @@ static void uart_write_rx_fifo(void *opaque, const uint8_t *buf, int size) s->r[R_CISR] |= UART_INTR_ROVR; } else { for (i = 0; i < size; i++) { - s->r_fifo[s->rx_wpos] = buf[i]; + s->rx_fifo[s->rx_wpos] = buf[i]; s->rx_wpos = (s->rx_wpos + 1) % RX_FIFO_SIZE; s->rx_count++; @@ -344,7 +344,7 @@ static void uart_read_rx_fifo(UartState *s, uint32_t *c) if (s->rx_count) { uint32_t rx_rpos = (RX_FIFO_SIZE + s->rx_wpos - s->rx_count) % RX_FIFO_SIZE; - *c = s->r_fifo[rx_rpos]; + *c = s->rx_fifo[rx_rpos]; s->rx_count--; if (!s->rx_count) { @@ -492,7 +492,7 @@ static const VMStateDescription vmstate_cadence_uart = { .post_load = cadence_uart_post_load, .fields = (VMStateField[]) { VMSTATE_UINT32_ARRAY(r, UartState, R_MAX), - VMSTATE_UINT8_ARRAY(r_fifo, UartState, RX_FIFO_SIZE), + VMSTATE_UINT8_ARRAY(rx_fifo, UartState, RX_FIFO_SIZE), VMSTATE_UINT32(rx_count, UartState), VMSTATE_UINT32(rx_wpos, UartState), VMSTATE_TIMER(fifo_trigger_handle, UartState), From a24234cad03d70612e30494f21062fcee5eb9f0e Mon Sep 17 00:00:00 2001 From: Peter Crosthwaite Date: Mon, 6 Jan 2014 10:16:38 +0000 Subject: [PATCH 25/61] char/cadence_uart: Simplify status generation The status register bits are always pure functions of other device state. Move the generation of these bits to the update_status() function to simplify. Makes developing much easier as theres now no need to recheck status bits on all the changes to rx/tx fifo state. Signed-off-by: Peter Crosthwaite Message-id: 321994929f789096975104f99c55732774be4cae.1388626249.git.peter.crosthwaite@xilinx.com Signed-off-by: Peter Maydell --- hw/char/cadence_uart.c | 33 ++++++++------------------------- 1 file changed, 8 insertions(+), 25 deletions(-) diff --git a/hw/char/cadence_uart.c b/hw/char/cadence_uart.c index d6abc5b985..ddd7267259 100644 --- a/hw/char/cadence_uart.c +++ b/hw/char/cadence_uart.c @@ -128,6 +128,13 @@ typedef struct { static void uart_update_status(UartState *s) { + s->r[R_SR] = 0; + + s->r[R_SR] |= s->rx_count == RX_FIFO_SIZE ? UART_SR_INTR_RFUL : 0; + s->r[R_SR] |= !s->rx_count ? UART_SR_INTR_REMPTY : 0; + s->r[R_SR] |= s->rx_count >= s->r[R_RTRIG] ? UART_SR_INTR_RTRIG : 0; + + s->r[R_SR] |= UART_SR_INTR_TEMPTY; s->r[R_CISR] |= s->r[R_SR] & UART_SR_TO_CISR_MASK; qemu_set_irq(s->irq, !!(s->r[R_IMR] & s->r[R_CISR])); } @@ -166,15 +173,10 @@ static void uart_rx_reset(UartState *s) if (s->chr) { qemu_chr_accept_input(s->chr); } - - s->r[R_SR] |= UART_SR_INTR_REMPTY; - s->r[R_SR] &= ~UART_SR_INTR_RFUL; } static void uart_tx_reset(UartState *s) { - s->r[R_SR] |= UART_SR_INTR_TEMPTY; - s->r[R_SR] &= ~UART_SR_INTR_TFUL; } static void uart_send_breaks(UartState *s) @@ -274,8 +276,6 @@ static void uart_write_rx_fifo(void *opaque, const uint8_t *buf, int size) return; } - s->r[R_SR] &= ~UART_SR_INTR_REMPTY; - if (s->rx_count == RX_FIFO_SIZE) { s->r[R_CISR] |= UART_INTR_ROVR; } else { @@ -283,15 +283,6 @@ static void uart_write_rx_fifo(void *opaque, const uint8_t *buf, int size) s->rx_fifo[s->rx_wpos] = buf[i]; s->rx_wpos = (s->rx_wpos + 1) % RX_FIFO_SIZE; s->rx_count++; - - if (s->rx_count == RX_FIFO_SIZE) { - s->r[R_SR] |= UART_SR_INTR_RFUL; - break; - } - - if (s->rx_count >= s->r[R_RTRIG]) { - s->r[R_SR] |= UART_SR_INTR_RTRIG; - } } timer_mod(s->fifo_trigger_handle, new_rx_time + (s->char_tx_time * 4)); @@ -339,26 +330,17 @@ static void uart_read_rx_fifo(UartState *s, uint32_t *c) return; } - s->r[R_SR] &= ~UART_SR_INTR_RFUL; - if (s->rx_count) { uint32_t rx_rpos = (RX_FIFO_SIZE + s->rx_wpos - s->rx_count) % RX_FIFO_SIZE; *c = s->rx_fifo[rx_rpos]; s->rx_count--; - if (!s->rx_count) { - s->r[R_SR] |= UART_SR_INTR_REMPTY; - } qemu_chr_accept_input(s->chr); } else { *c = 0; - s->r[R_SR] |= UART_SR_INTR_REMPTY; } - if (s->rx_count < s->r[R_RTRIG]) { - s->r[R_SR] &= ~UART_SR_INTR_RTRIG; - } uart_update_status(s); } @@ -447,6 +429,7 @@ static void cadence_uart_reset(DeviceState *dev) s->rx_count = 0; s->rx_wpos = 0; + uart_update_status(s); } static int cadence_uart_init(SysBusDevice *dev) From 2c628d98989fa224dc3c07033ba4d89562a9bb5f Mon Sep 17 00:00:00 2001 From: Peter Crosthwaite Date: Mon, 6 Jan 2014 10:16:39 +0000 Subject: [PATCH 26/61] char/cadence_uart: Define Missing SR/ISR fields Some (interrupt) status register bits relating to the TxFIFO path were not defined. Define them. This prepares support for proper Tx data path flow control. Signed-off-by: Peter Crosthwaite Message-id: 2068b963f0af8cc834c353944e9fa816d950b163.1388626249.git.peter.crosthwaite@xilinx.com Signed-off-by: Peter Maydell --- hw/char/cadence_uart.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/hw/char/cadence_uart.c b/hw/char/cadence_uart.c index ddd7267259..216eed7442 100644 --- a/hw/char/cadence_uart.c +++ b/hw/char/cadence_uart.c @@ -34,6 +34,9 @@ #define UART_SR_INTR_RFUL 0x00000004 #define UART_SR_INTR_TEMPTY 0x00000008 #define UART_SR_INTR_TFUL 0x00000010 +/* somewhat awkwardly, TTRIG is misaligned between SR and ISR */ +#define UART_SR_TTRIG 0x00002000 +#define UART_INTR_TTRIG 0x00000400 /* bits fields in CSR that correlate to CISR. If any of these bits are set in * SR, then the same bit in CISR is set high too */ #define UART_SR_TO_CISR_MASK 0x0000001F @@ -43,6 +46,7 @@ #define UART_INTR_PARE 0x00000080 #define UART_INTR_TIMEOUT 0x00000100 #define UART_INTR_DMSI 0x00000200 +#define UART_INTR_TOVR 0x00001000 #define UART_SR_RACTIVE 0x00000400 #define UART_SR_TACTIVE 0x00000800 From faa79359cdb28916b102024e49916af212b344bf Mon Sep 17 00:00:00 2001 From: Peter Crosthwaite Date: Mon, 6 Jan 2014 10:16:39 +0000 Subject: [PATCH 27/61] char/cadence_uart: Remove TX timer & add TX FIFO state This tx timer implementation is flawed. Despite the controller attempting to time the guest visable assertion of the TX-empty status bit (and corresponding interrupt) the controller is still transmitting characters instantaneously. There is also no sense of multiple character delay. The only side effect of this timer is assertion of tx-empty status. So just remove the timer completely and hold tx-empty as permanently asserted (its reset status). This matches the actual behaviour of instantaneous transmission. While we are VMSD version bumping, add the tx_fifo as device state to prepare for upcomming TxFIFO flow control. Implement the interrupt generation logic for the TxFIFO occupancy. Signed-off-by: Peter Crosthwaite Message-id: 7a208a7eb8d79d6429fe28b1396c3104371807b2.1388626249.git.peter.crosthwaite@xilinx.com Signed-off-by: Peter Maydell --- hw/char/cadence_uart.c | 44 +++++++++++++----------------------------- 1 file changed, 13 insertions(+), 31 deletions(-) diff --git a/hw/char/cadence_uart.c b/hw/char/cadence_uart.c index 216eed7442..3eeadb18fd 100644 --- a/hw/char/cadence_uart.c +++ b/hw/char/cadence_uart.c @@ -121,13 +121,14 @@ typedef struct { MemoryRegion iomem; uint32_t r[R_MAX]; uint8_t rx_fifo[RX_FIFO_SIZE]; + uint8_t tx_fifo[TX_FIFO_SIZE]; uint32_t rx_wpos; uint32_t rx_count; + uint32_t tx_count; uint64_t char_tx_time; CharDriverState *chr; qemu_irq irq; QEMUTimer *fifo_trigger_handle; - QEMUTimer *tx_time_handle; } UartState; static void uart_update_status(UartState *s) @@ -138,8 +139,12 @@ static void uart_update_status(UartState *s) s->r[R_SR] |= !s->rx_count ? UART_SR_INTR_REMPTY : 0; s->r[R_SR] |= s->rx_count >= s->r[R_RTRIG] ? UART_SR_INTR_RTRIG : 0; - s->r[R_SR] |= UART_SR_INTR_TEMPTY; + s->r[R_SR] |= s->tx_count == TX_FIFO_SIZE ? UART_SR_INTR_TFUL : 0; + s->r[R_SR] |= !s->tx_count ? UART_SR_INTR_TEMPTY : 0; + s->r[R_SR] |= s->tx_count >= s->r[R_TTRIG] ? UART_SR_TTRIG : 0; + s->r[R_CISR] |= s->r[R_SR] & UART_SR_TO_CISR_MASK; + s->r[R_CISR] |= s->r[R_SR] & UART_SR_TTRIG ? UART_INTR_TTRIG : 0; qemu_set_irq(s->irq, !!(s->r[R_IMR] & s->r[R_CISR])); } @@ -152,24 +157,6 @@ static void fifo_trigger_update(void *opaque) uart_update_status(s); } -static void uart_tx_redo(UartState *s) -{ - uint64_t new_tx_time = qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL); - - timer_mod(s->tx_time_handle, new_tx_time + s->char_tx_time); - - s->r[R_SR] |= UART_SR_INTR_TEMPTY; - - uart_update_status(s); -} - -static void uart_tx_write(void *opaque) -{ - UartState *s = (UartState *)opaque; - - uart_tx_redo(s); -} - static void uart_rx_reset(UartState *s) { s->rx_wpos = 0; @@ -181,6 +168,7 @@ static void uart_rx_reset(UartState *s) static void uart_tx_reset(UartState *s) { + s->tx_count = 0; } static void uart_send_breaks(UartState *s) @@ -261,10 +249,6 @@ static void uart_ctrl_update(UartState *s) s->r[R_CR] &= ~(UART_CR_TXRST | UART_CR_RXRST); - if ((s->r[R_CR] & UART_CR_TX_EN) && !(s->r[R_CR] & UART_CR_TX_DIS)) { - uart_tx_redo(s); - } - if (s->r[R_CR] & UART_CR_STARTBRK && !(s->r[R_CR] & UART_CR_STOPBRK)) { uart_send_breaks(s); } @@ -447,9 +431,6 @@ static int cadence_uart_init(SysBusDevice *dev) s->fifo_trigger_handle = timer_new_ns(QEMU_CLOCK_VIRTUAL, (QEMUTimerCB *)fifo_trigger_update, s); - s->tx_time_handle = timer_new_ns(QEMU_CLOCK_VIRTUAL, - (QEMUTimerCB *)uart_tx_write, s); - s->char_tx_time = (get_ticks_per_sec() / 9600) * 10; s->chr = qemu_char_get_next_serial(); @@ -473,17 +454,18 @@ static int cadence_uart_post_load(void *opaque, int version_id) static const VMStateDescription vmstate_cadence_uart = { .name = "cadence_uart", - .version_id = 1, - .minimum_version_id = 1, - .minimum_version_id_old = 1, + .version_id = 2, + .minimum_version_id = 2, + .minimum_version_id_old = 2, .post_load = cadence_uart_post_load, .fields = (VMStateField[]) { VMSTATE_UINT32_ARRAY(r, UartState, R_MAX), VMSTATE_UINT8_ARRAY(rx_fifo, UartState, RX_FIFO_SIZE), + VMSTATE_UINT8_ARRAY(tx_fifo, UartState, RX_FIFO_SIZE), VMSTATE_UINT32(rx_count, UartState), + VMSTATE_UINT32(tx_count, UartState), VMSTATE_UINT32(rx_wpos, UartState), VMSTATE_TIMER(fifo_trigger_handle, UartState), - VMSTATE_TIMER(tx_time_handle, UartState), VMSTATE_END_OF_LIST() } }; From 61654c77413d1bccb69b384d891d0c688a31430b Mon Sep 17 00:00:00 2001 From: Peter Crosthwaite Date: Mon, 6 Jan 2014 10:16:39 +0000 Subject: [PATCH 28/61] char/cadence_uart: Fix can_receive logic The can_receive logic was only taking into account the RxFIFO occupancy. RxFIFO population is only used for the echo and normal modes however. Improve the logic to correctly return the true number of receivable characters based on the current mode: Normal mode: RxFIFO vacancy. Remote loopback: TxFIFO vacancy. Echo mode: The min of the TxFIFO and RxFIFO vacancies. Local Loopback: Return non-zero (to implement droppage) Signed-off-by: Peter Crosthwaite Message-id: 36a58440c9ca5080151e95765c2c81342de8a8df.1388626249.git.peter.crosthwaite@xilinx.com Signed-off-by: Peter Maydell --- hw/char/cadence_uart.c | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-) diff --git a/hw/char/cadence_uart.c b/hw/char/cadence_uart.c index 3eeadb18fd..3bcaf2905d 100644 --- a/hw/char/cadence_uart.c +++ b/hw/char/cadence_uart.c @@ -233,8 +233,16 @@ static void uart_parameters_setup(UartState *s) static int uart_can_receive(void *opaque) { UartState *s = (UartState *)opaque; + int ret = MAX(RX_FIFO_SIZE, TX_FIFO_SIZE); + uint32_t ch_mode = s->r[R_MR] & UART_MR_CHMODE; - return RX_FIFO_SIZE - s->rx_count; + if (ch_mode == NORMAL_MODE || ch_mode == ECHO_MODE) { + ret = MIN(ret, RX_FIFO_SIZE - s->rx_count); + } + if (ch_mode == REMOTE_LOOPBACK || ch_mode == ECHO_MODE) { + ret = MIN(ret, TX_FIFO_SIZE - s->tx_count); + } + return ret; } static void uart_ctrl_update(UartState *s) From 9e263f73912ba35510d7d7be4c40ba871d4b57e2 Mon Sep 17 00:00:00 2001 From: Peter Crosthwaite Date: Mon, 6 Jan 2014 10:16:40 +0000 Subject: [PATCH 29/61] char/cadence_uart: Use the TX fifo for transmission Populate the TxFIFO with the Tx data before sending. Prepares support for proper Tx flow control implementation. Signed-off-by: Peter Crosthwaite Message-id: bdf7f8af2ef02839bea18665701bc2612f7baa6f.1388626249.git.peter.crosthwaite@xilinx.com Signed-off-by: Peter Maydell --- hw/char/cadence_uart.c | 17 ++++++++++++++++- 1 file changed, 16 insertions(+), 1 deletion(-) diff --git a/hw/char/cadence_uart.c b/hw/char/cadence_uart.c index 3bcaf2905d..be32126bdc 100644 --- a/hw/char/cadence_uart.c +++ b/hw/char/cadence_uart.c @@ -292,7 +292,22 @@ static void uart_write_tx_fifo(UartState *s, const uint8_t *buf, int size) return; } - qemu_chr_fe_write_all(s->chr, buf, size); + if (size > TX_FIFO_SIZE - s->tx_count) { + size = TX_FIFO_SIZE - s->tx_count; + /* + * This can only be a guest error via a bad tx fifo register push, + * as can_receive() should stop remote loop and echo modes ever getting + * us to here. + */ + qemu_log_mask(LOG_GUEST_ERROR, "cadence_uart: TxFIFO overflow"); + s->r[R_CISR] |= UART_INTR_ROVR; + } + + memcpy(s->tx_fifo + s->tx_count, buf, size); + s->tx_count += size; + + qemu_chr_fe_write_all(s->chr, s->tx_fifo, s->tx_count); + s->tx_count = 0; } static void uart_receive(void *opaque, const uint8_t *buf, int size) From a33f37a1784691556ed6fd3f58badfce22b761ad Mon Sep 17 00:00:00 2001 From: Peter Crosthwaite Date: Mon, 6 Jan 2014 10:16:40 +0000 Subject: [PATCH 30/61] char/cadence_uart: Delete redundant rx rst logic uart_rx_reset() called immediately above already does this. Remove. Signed-off-by: Peter Crosthwaite Message-id: 05e30826496cf2579084ed801ac0b2c0d0a3071f.1388626249.git.peter.crosthwaite@xilinx.com Signed-off-by: Peter Maydell --- hw/char/cadence_uart.c | 2 -- 1 file changed, 2 deletions(-) diff --git a/hw/char/cadence_uart.c b/hw/char/cadence_uart.c index be32126bdc..8a9ef81d2c 100644 --- a/hw/char/cadence_uart.c +++ b/hw/char/cadence_uart.c @@ -438,8 +438,6 @@ static void cadence_uart_reset(DeviceState *dev) uart_rx_reset(s); uart_tx_reset(s); - s->rx_count = 0; - s->rx_wpos = 0; uart_update_status(s); } From 294cdac2a02d3172b69c84dec727d6ac6006bb59 Mon Sep 17 00:00:00 2001 From: Peter Crosthwaite Date: Mon, 6 Jan 2014 10:16:40 +0000 Subject: [PATCH 31/61] char/cadence_uart: Implement Tx flow control If the UART back-end blocks, buffer in the Tx FIFO to try again later. This stops the IO-thread busy waiting on char back-ends (which causes all sorts of performance problems). Signed-off-by: Peter Crosthwaite Message-id: 4bea048b3ab38425701d82ccc1ab92545c26b79c.1388626249.git.peter.crosthwaite@xilinx.com Signed-off-by: Peter Maydell --- hw/char/cadence_uart.c | 31 +++++++++++++++++++++++++++++-- 1 file changed, 29 insertions(+), 2 deletions(-) diff --git a/hw/char/cadence_uart.c b/hw/char/cadence_uart.c index 8a9ef81d2c..1012f1ad64 100644 --- a/hw/char/cadence_uart.c +++ b/hw/char/cadence_uart.c @@ -286,6 +286,34 @@ static void uart_write_rx_fifo(void *opaque, const uint8_t *buf, int size) uart_update_status(s); } +static gboolean cadence_uart_xmit(GIOChannel *chan, GIOCondition cond, + void *opaque) +{ + UartState *s = opaque; + int ret; + + /* instant drain the fifo when there's no back-end */ + if (!s->chr) { + s->tx_count = 0; + } + + if (!s->tx_count) { + return FALSE; + } + + ret = qemu_chr_fe_write(s->chr, s->tx_fifo, s->tx_count); + s->tx_count -= ret; + memmove(s->tx_fifo, s->tx_fifo + ret, s->tx_count); + + if (s->tx_count) { + int r = qemu_chr_fe_add_watch(s->chr, G_IO_OUT, cadence_uart_xmit, s); + assert(r); + } + + uart_update_status(s); + return FALSE; +} + static void uart_write_tx_fifo(UartState *s, const uint8_t *buf, int size) { if ((s->r[R_CR] & UART_CR_TX_DIS) || !(s->r[R_CR] & UART_CR_TX_EN)) { @@ -306,8 +334,7 @@ static void uart_write_tx_fifo(UartState *s, const uint8_t *buf, int size) memcpy(s->tx_fifo + s->tx_count, buf, size); s->tx_count += size; - qemu_chr_fe_write_all(s->chr, s->tx_fifo, s->tx_count); - s->tx_count = 0; + cadence_uart_xmit(NULL, G_IO_OUT, s); } static void uart_receive(void *opaque, const uint8_t *buf, int size) From d045815a5712afbeb8d2c76d8b624c91cfda777c Mon Sep 17 00:00:00 2001 From: Sergey Fedorov Date: Fri, 20 Dec 2013 10:33:11 +0400 Subject: [PATCH 32/61] target-arm: use c13_context field for CONTEXTIDR Use c13_context field instead of c13_fcse for CONTEXTIDR register definition. Signed-off-by: Sergey Fedorov Reviewed-by: Peter Crosthwaite Reviewed-by: Peter Maydell Message-id: 1387521191-15350-1-git-send-email-s.fedorov@samsung.com Signed-off-by: Peter Maydell --- target-arm/helper.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/target-arm/helper.c b/target-arm/helper.c index 9afec28a1f..be52c1f183 100644 --- a/target-arm/helper.c +++ b/target-arm/helper.c @@ -397,7 +397,7 @@ static const ARMCPRegInfo cp_reginfo[] = { .access = PL1_RW, .fieldoffset = offsetof(CPUARMState, cp15.c13_fcse), .resetvalue = 0, .writefn = fcse_write, .raw_writefn = raw_write, }, { .name = "CONTEXTIDR", .cp = 15, .crn = 13, .crm = 0, .opc1 = 0, .opc2 = 1, - .access = PL1_RW, .fieldoffset = offsetof(CPUARMState, cp15.c13_fcse), + .access = PL1_RW, .fieldoffset = offsetof(CPUARMState, cp15.c13_context), .resetvalue = 0, .writefn = contextidr_write, .raw_writefn = raw_write, }, /* ??? This covers not just the impdef TLB lockdown registers but also * some v7VMSA registers relating to TEX remap, so it is overly broad. From 8455bc76c76c5246694290a28c6ebd6797011c18 Mon Sep 17 00:00:00 2001 From: Peter Crosthwaite Date: Wed, 1 Jan 2014 23:58:20 -0800 Subject: [PATCH 33/61] target-arm: remove raw_read|write duplication There is an inline duplication of the raw_read and raw_write function bodies. Fix by just calling raw_read/raw_write instead. Signed-off-by: Peter Crosthwaite Reviewed-by: Peter Maydell Message-id: e69281b7e1462b346cb313cf0b89eedc0568125f.1388649290.git.peter.crosthwaite@xilinx.com Signed-off-by: Peter Maydell --- target-arm/helper.c | 12 ++---------- 1 file changed, 2 insertions(+), 10 deletions(-) diff --git a/target-arm/helper.c b/target-arm/helper.c index be52c1f183..6f629f36b0 100644 --- a/target-arm/helper.c +++ b/target-arm/helper.c @@ -142,11 +142,7 @@ static bool read_raw_cp_reg(CPUARMState *env, const ARMCPRegInfo *ri, } else if (ri->readfn) { return (ri->readfn(env, ri, v) == 0); } else { - if (ri->type & ARM_CP_64BIT) { - *v = CPREG_FIELD64(env, ri); - } else { - *v = CPREG_FIELD32(env, ri); - } + raw_read(env, ri, v); } return true; } @@ -167,11 +163,7 @@ static bool write_raw_cp_reg(CPUARMState *env, const ARMCPRegInfo *ri, } else if (ri->writefn) { return (ri->writefn(env, ri, v) == 0); } else { - if (ri->type & ARM_CP_64BIT) { - CPREG_FIELD64(env, ri) = v; - } else { - CPREG_FIELD32(env, ri) = v; - } + raw_write(env, ri, v); } return true; } From ec0e68ef1da316b3ead1943d8f607cc68b13e0d1 Mon Sep 17 00:00:00 2001 From: Peter Crosthwaite Date: Fri, 3 Jan 2014 17:58:43 -0800 Subject: [PATCH 34/61] arm/xilinx_zynq: Always instantiate the GEMs Don't conditionalise GEM instantiation on networking attachments. The device should always be present even if not attached to a network. This allows for probing of the device by expectant guests (such as OS's). This is needed because sysbus (or AXI in Xilinx's real hw case) is not self identifying so the guest has no dynamic way of detecting device absence. Also allows for testing of the GEM in loopback mode with -net none. Signed-off-by: Peter Crosthwaite Reviewed-by: Peter Maydell Message-id: 55649779a68ee3ff54b24c339b6fdbdccd1f0ed7.1388800598.git.peter.crosthwaite@xilinx.com Signed-off-by: Peter Maydell --- hw/arm/xilinx_zynq.c | 17 ++++++----------- 1 file changed, 6 insertions(+), 11 deletions(-) diff --git a/hw/arm/xilinx_zynq.c b/hw/arm/xilinx_zynq.c index 17251c7a65..98e0958a77 100644 --- a/hw/arm/xilinx_zynq.c +++ b/hw/arm/xilinx_zynq.c @@ -49,9 +49,11 @@ static void gem_init(NICInfo *nd, uint32_t base, qemu_irq irq) DeviceState *dev; SysBusDevice *s; - qemu_check_nic_model(nd, "cadence_gem"); dev = qdev_create(NULL, "cadence_gem"); - qdev_set_nic_properties(dev, nd); + if (nd->used) { + qemu_check_nic_model(nd, "cadence_gem"); + qdev_set_nic_properties(dev, nd); + } qdev_init_nofail(dev); s = SYS_BUS_DEVICE(dev); sysbus_mmio_map(s, 0, base); @@ -113,7 +115,6 @@ static void zynq_init(QEMUMachineInitArgs *args) DeviceState *dev; SysBusDevice *busdev; qemu_irq pic[64]; - NICInfo *nd; Error *err = NULL; int n; @@ -190,14 +191,8 @@ static void zynq_init(QEMUMachineInitArgs *args) sysbus_create_varargs("cadence_ttc", 0xF8002000, pic[69-IRQ_OFFSET], pic[70-IRQ_OFFSET], pic[71-IRQ_OFFSET], NULL); - for (n = 0; n < nb_nics; n++) { - nd = &nd_table[n]; - if (n == 0) { - gem_init(nd, 0xE000B000, pic[54-IRQ_OFFSET]); - } else if (n == 1) { - gem_init(nd, 0xE000C000, pic[77-IRQ_OFFSET]); - } - } + gem_init(&nd_table[0], 0xE000B000, pic[54-IRQ_OFFSET]); + gem_init(&nd_table[1], 0xE000C000, pic[77-IRQ_OFFSET]); dev = qdev_create(NULL, "generic-sdhci"); qdev_init_nofail(dev); From 455e1e14bbfaf72275efeddaa7bc5d6294253eff Mon Sep 17 00:00:00 2001 From: "Michael S. Tsirkin" Date: Mon, 23 Dec 2013 16:52:16 +0200 Subject: [PATCH 35/61] target-arm: fix build with gcc 4.8.2 MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit commit 5ce4f35781028ce1aee3341e6002f925fdc7aaf3 "target-arm: A64: add set_pc cpu method" introduces an array aarch64_cpus which is zero size if this code is built without CONFIG_USER_ONLY. In particular an attempt to iterate over this array produces a warning under gcc 4.8.2: CC aarch64-softmmu/target-arm/cpu64.o /scm/qemu/target-arm/cpu64.c: In function ‘aarch64_cpu_register_types’: /scm/qemu/target-arm/cpu64.c:124:5: error: comparison of unsigned expression < 0 is always false [-Werror=type-limits] for (i = 0; i < ARRAY_SIZE(aarch64_cpus); i++) { ^ cc1: all warnings being treated as errors This is the result of ARRAY_SIZE being an unsigned type, causing "i" to be promoted to unsigned int as well. As zero size arrays are a gcc extension, it seems cleanest to add a dummy element with NULL name, and test for it during registration. We'll be able to drop this when we add more CPUs. Cc: Alexander Graf Cc: Peter Maydell Cc: Richard Henderson Signed-off-by: Michael S. Tsirkin Reviewed-by: Stefan Weil Reviewed-by: Peter Maydell Message-id: 20131223145216.GA22663@redhat.com Signed-off-by: Peter Maydell --- target-arm/cpu64.c | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/target-arm/cpu64.c b/target-arm/cpu64.c index 04ce87951c..60acd24c0c 100644 --- a/target-arm/cpu64.c +++ b/target-arm/cpu64.c @@ -58,6 +58,7 @@ static const ARMCPUInfo aarch64_cpus[] = { #ifdef CONFIG_USER_ONLY { .name = "any", .initfn = aarch64_any_initfn }, #endif + { .name = NULL } /* TODO: drop when we support more CPUs */ }; static void aarch64_cpu_initfn(Object *obj) @@ -100,6 +101,11 @@ static void aarch64_cpu_register(const ARMCPUInfo *info) .class_init = info->class_init, }; + /* TODO: drop when we support more CPUs - all entries will have name set */ + if (!info->name) { + return; + } + type_info.name = g_strdup_printf("%s-" TYPE_ARM_CPU, info->name); type_register(&type_info); g_free((void *)type_info.name); From 07c935087a5541fb167fecf3bef810de2605f93f Mon Sep 17 00:00:00 2001 From: Christoffer Dall Date: Fri, 20 Dec 2013 22:09:32 -0800 Subject: [PATCH 36/61] arm_gic: Rename GIC_X_TRIGGER to GIC_X_EDGE_TRIGGER TRIGGER can really mean mean anything (e.g. was it triggered, is it level-triggered, is it edge-triggered, etc.). Rename to EDGE_TRIGGER to make the code comprehensible without looking up the data structure. Reviewed-by: Peter Maydell Signed-off-by: Christoffer Dall Message-id: 1387606179-22709-2-git-send-email-christoffer.dall@linaro.org Signed-off-by: Peter Maydell --- hw/intc/arm_gic.c | 12 ++++++------ hw/intc/arm_gic_common.c | 4 ++-- hw/intc/gic_internal.h | 6 +++--- include/hw/intc/arm_gic_common.h | 2 +- 4 files changed, 12 insertions(+), 12 deletions(-) diff --git a/hw/intc/arm_gic.c b/hw/intc/arm_gic.c index d431b7a881..27c258a6fd 100644 --- a/hw/intc/arm_gic.c +++ b/hw/intc/arm_gic.c @@ -128,7 +128,7 @@ static void gic_set_irq(void *opaque, int irq, int level) if (level) { GIC_SET_LEVEL(irq, cm); - if (GIC_TEST_TRIGGER(irq) || GIC_TEST_ENABLED(irq, cm)) { + if (GIC_TEST_EDGE_TRIGGER(irq) || GIC_TEST_ENABLED(irq, cm)) { DPRINTF("Set %d pending mask %x\n", irq, target); GIC_SET_PENDING(irq, target); } @@ -188,7 +188,7 @@ void gic_complete_irq(GICState *s, int cpu, int irq) return; /* No active IRQ. */ /* Mark level triggered interrupts as pending if they are still raised. */ - if (!GIC_TEST_TRIGGER(irq) && GIC_TEST_ENABLED(irq, cm) + if (!GIC_TEST_EDGE_TRIGGER(irq) && GIC_TEST_ENABLED(irq, cm) && GIC_TEST_LEVEL(irq, cm) && (GIC_TARGET(irq) & cm) != 0) { DPRINTF("Set %d pending mask %x\n", irq, cm); GIC_SET_PENDING(irq, cm); @@ -311,7 +311,7 @@ static uint32_t gic_dist_readb(void *opaque, hwaddr offset) for (i = 0; i < 4; i++) { if (GIC_TEST_MODEL(irq + i)) res |= (1 << (i * 2)); - if (GIC_TEST_TRIGGER(irq + i)) + if (GIC_TEST_EDGE_TRIGGER(irq + i)) res |= (2 << (i * 2)); } } else if (offset < 0xfe0) { @@ -386,7 +386,7 @@ static void gic_dist_writeb(void *opaque, hwaddr offset, /* If a raised level triggered IRQ enabled then mark is as pending. */ if (GIC_TEST_LEVEL(irq + i, mask) - && !GIC_TEST_TRIGGER(irq + i)) { + && !GIC_TEST_EDGE_TRIGGER(irq + i)) { DPRINTF("Set %d pending mask %x\n", irq + i, mask); GIC_SET_PENDING(irq + i, mask); } @@ -478,9 +478,9 @@ static void gic_dist_writeb(void *opaque, hwaddr offset, GIC_CLEAR_MODEL(irq + i); } if (value & (2 << (i * 2))) { - GIC_SET_TRIGGER(irq + i); + GIC_SET_EDGE_TRIGGER(irq + i); } else { - GIC_CLEAR_TRIGGER(irq + i); + GIC_CLEAR_EDGE_TRIGGER(irq + i); } } } else { diff --git a/hw/intc/arm_gic_common.c b/hw/intc/arm_gic_common.c index c7658508dd..710607b044 100644 --- a/hw/intc/arm_gic_common.c +++ b/hw/intc/arm_gic_common.c @@ -51,7 +51,7 @@ static const VMStateDescription vmstate_gic_irq_state = { VMSTATE_UINT8(active, gic_irq_state), VMSTATE_UINT8(level, gic_irq_state), VMSTATE_BOOL(model, gic_irq_state), - VMSTATE_BOOL(trigger, gic_irq_state), + VMSTATE_BOOL(edge_trigger, gic_irq_state), VMSTATE_END_OF_LIST() } }; @@ -126,7 +126,7 @@ static void arm_gic_common_reset(DeviceState *dev) } for (i = 0; i < 16; i++) { GIC_SET_ENABLED(i, ALL_CPU_MASK); - GIC_SET_TRIGGER(i); + GIC_SET_EDGE_TRIGGER(i); } if (s->num_cpu == 1) { /* For uniprocessor GICs all interrupts always target the sole CPU */ diff --git a/hw/intc/gic_internal.h b/hw/intc/gic_internal.h index 3989fd1bd5..efac78d44b 100644 --- a/hw/intc/gic_internal.h +++ b/hw/intc/gic_internal.h @@ -44,9 +44,9 @@ #define GIC_SET_LEVEL(irq, cm) s->irq_state[irq].level = (cm) #define GIC_CLEAR_LEVEL(irq, cm) s->irq_state[irq].level &= ~(cm) #define GIC_TEST_LEVEL(irq, cm) ((s->irq_state[irq].level & (cm)) != 0) -#define GIC_SET_TRIGGER(irq) s->irq_state[irq].trigger = true -#define GIC_CLEAR_TRIGGER(irq) s->irq_state[irq].trigger = false -#define GIC_TEST_TRIGGER(irq) s->irq_state[irq].trigger +#define GIC_SET_EDGE_TRIGGER(irq) s->irq_state[irq].edge_trigger = true +#define GIC_CLEAR_EDGE_TRIGGER(irq) s->irq_state[irq].edge_trigger = false +#define GIC_TEST_EDGE_TRIGGER(irq) (s->irq_state[irq].edge_trigger) #define GIC_GET_PRIORITY(irq, cpu) (((irq) < GIC_INTERNAL) ? \ s->priority1[irq][cpu] : \ s->priority2[(irq) - GIC_INTERNAL]) diff --git a/include/hw/intc/arm_gic_common.h b/include/hw/intc/arm_gic_common.h index 4f381bdce7..0d232dfb67 100644 --- a/include/hw/intc/arm_gic_common.h +++ b/include/hw/intc/arm_gic_common.h @@ -37,7 +37,7 @@ typedef struct gic_irq_state { uint8_t active; uint8_t level; bool model; /* 0 = N:N, 1 = 1:N */ - bool trigger; /* nonzero = edge triggered. */ + bool edge_trigger; /* true: edge-triggered, false: level-triggered */ } gic_irq_state; typedef struct GICState { From e3d8cc0536699b66f58b0e3a13a70ee007f61647 Mon Sep 17 00:00:00 2001 From: Christoffer Dall Date: Fri, 20 Dec 2013 22:09:33 -0800 Subject: [PATCH 37/61] hw: arm_gic: Introduce gic_set_priority function To make the code slightly cleaner to look at and make the save/restore code easier to understand, introduce this function to set the priority of interrupts. Reviewed-by: Peter Maydell Signed-off-by: Christoffer Dall Message-id: 1387606179-22709-3-git-send-email-christoffer.dall@linaro.org Signed-off-by: Peter Maydell --- hw/intc/arm_gic.c | 15 ++++++++++----- hw/intc/gic_internal.h | 1 + 2 files changed, 11 insertions(+), 5 deletions(-) diff --git a/hw/intc/arm_gic.c b/hw/intc/arm_gic.c index 27c258a6fd..6c5965093e 100644 --- a/hw/intc/arm_gic.c +++ b/hw/intc/arm_gic.c @@ -168,6 +168,15 @@ uint32_t gic_acknowledge_irq(GICState *s, int cpu) return new_irq; } +void gic_set_priority(GICState *s, int cpu, int irq, uint8_t val) +{ + if (irq < GIC_INTERNAL) { + s->priority1[irq][cpu] = val; + } else { + s->priority2[(irq) - GIC_INTERNAL] = val; + } +} + void gic_complete_irq(GICState *s, int cpu, int irq) { int update = 0; @@ -443,11 +452,7 @@ static void gic_dist_writeb(void *opaque, hwaddr offset, irq = (offset - 0x400) + GIC_BASE_IRQ; if (irq >= s->num_irq) goto bad_reg; - if (irq < GIC_INTERNAL) { - s->priority1[irq][cpu] = value; - } else { - s->priority2[irq - GIC_INTERNAL] = value; - } + gic_set_priority(s, cpu, irq, value); } else if (offset < 0xc00) { /* Interrupt CPU Target. RAZ/WI on uniprocessor GICs, with the * annoying exception of the 11MPCore's GIC. diff --git a/hw/intc/gic_internal.h b/hw/intc/gic_internal.h index efac78d44b..8c02d5888c 100644 --- a/hw/intc/gic_internal.h +++ b/hw/intc/gic_internal.h @@ -61,5 +61,6 @@ uint32_t gic_acknowledge_irq(GICState *s, int cpu); void gic_complete_irq(GICState *s, int cpu, int irq); void gic_update(GICState *s); void gic_init_irqs_and_distributor(GICState *s, int num_irq); +void gic_set_priority(GICState *s, int cpu, int irq, uint8_t val); #endif /* !QEMU_ARM_GIC_INTERNAL_H */ From 1bf6beec8a012cae4bb6fbc89c465127e9650c92 Mon Sep 17 00:00:00 2001 From: Peter Maydell Date: Mon, 6 Jan 2014 11:47:21 +0000 Subject: [PATCH 38/61] softfloat: Fix exception flag handling for float32_to_float16() Our float32 to float16 conversion routine was generating the correct numerical answers, but not always setting the right set of exception flags. Fix this, mostly by rearranging the code to more closely resemble RoundAndPackFloat*, and in particular: * non-IEEE halfprec always raises Invalid for input NaNs * we need to check for the overflow case before underflow * we weren't getting the tininess-detected-after-rounding case correct (somewhat academic since only ARM uses halfprec and it is always tininess-detected-before-rounding) * non-IEEE halfprec overflow raises only Invalid, not Invalid + Inexact * we weren't setting Inexact when we should Also add some clarifying comments about what the code is doing. Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- fpu/softfloat.c | 101 ++++++++++++++++++++++++++++++------------------ 1 file changed, 64 insertions(+), 37 deletions(-) diff --git a/fpu/softfloat.c b/fpu/softfloat.c index dbda61bc8e..6a6b6567d7 100644 --- a/fpu/softfloat.c +++ b/fpu/softfloat.c @@ -3046,6 +3046,10 @@ float16 float32_to_float16(float32 a, flag ieee STATUS_PARAM) uint32_t mask; uint32_t increment; int8 roundingMode; + int maxexp = ieee ? 15 : 16; + bool rounding_bumps_exp; + bool is_tiny = false; + a = float32_squash_input_denormal(a STATUS_VAR); aSig = extractFloat32Frac( a ); @@ -3054,11 +3058,12 @@ float16 float32_to_float16(float32 a, flag ieee STATUS_PARAM) if ( aExp == 0xFF ) { if (aSig) { /* Input is a NaN */ - float16 r = commonNaNToFloat16( float32ToCommonNaN( a STATUS_VAR ) STATUS_VAR ); if (!ieee) { + float_raise(float_flag_invalid STATUS_VAR); return packFloat16(aSign, 0, 0); } - return r; + return commonNaNToFloat16( + float32ToCommonNaN(a STATUS_VAR) STATUS_VAR); } /* Infinity */ if (!ieee) { @@ -3070,58 +3075,80 @@ float16 float32_to_float16(float32 a, flag ieee STATUS_PARAM) if (aExp == 0 && aSig == 0) { return packFloat16(aSign, 0, 0); } - /* Decimal point between bits 22 and 23. */ + /* Decimal point between bits 22 and 23. Note that we add the 1 bit + * even if the input is denormal; however this is harmless because + * the largest possible single-precision denormal is still smaller + * than the smallest representable half-precision denormal, and so we + * will end up ignoring aSig and returning via the "always return zero" + * codepath. + */ aSig |= 0x00800000; aExp -= 0x7f; + /* Calculate the mask of bits of the mantissa which are not + * representable in half-precision and will be lost. + */ if (aExp < -14) { + /* Will be denormal in halfprec */ mask = 0x00ffffff; if (aExp >= -24) { mask >>= 25 + aExp; } } else { + /* Normal number in halfprec */ mask = 0x00001fff; } - if (aSig & mask) { - float_raise( float_flag_underflow STATUS_VAR ); - roundingMode = STATUS(float_rounding_mode); - switch (roundingMode) { - case float_round_nearest_even: - increment = (mask + 1) >> 1; - if ((aSig & mask) == increment) { - increment = aSig & (increment << 1); - } - break; - case float_round_up: - increment = aSign ? 0 : mask; - break; - case float_round_down: - increment = aSign ? mask : 0; - break; - default: /* round_to_zero */ - increment = 0; - break; + + roundingMode = STATUS(float_rounding_mode); + switch (roundingMode) { + case float_round_nearest_even: + increment = (mask + 1) >> 1; + if ((aSig & mask) == increment) { + increment = aSig & (increment << 1); } - aSig += increment; - if (aSig >= 0x01000000) { - aSig >>= 1; - aExp++; - } - } else if (aExp < -14 - && STATUS(float_detect_tininess) == float_tininess_before_rounding) { - float_raise( float_flag_underflow STATUS_VAR); + break; + case float_round_up: + increment = aSign ? 0 : mask; + break; + case float_round_down: + increment = aSign ? mask : 0; + break; + default: /* round_to_zero */ + increment = 0; + break; } - if (ieee) { - if (aExp > 15) { - float_raise( float_flag_overflow | float_flag_inexact STATUS_VAR); + rounding_bumps_exp = (aSig + increment >= 0x01000000); + + if (aExp > maxexp || (aExp == maxexp && rounding_bumps_exp)) { + if (ieee) { + float_raise(float_flag_overflow | float_flag_inexact STATUS_VAR); return packFloat16(aSign, 0x1f, 0); - } - } else { - if (aExp > 16) { - float_raise(float_flag_invalid | float_flag_inexact STATUS_VAR); + } else { + float_raise(float_flag_invalid STATUS_VAR); return packFloat16(aSign, 0x1f, 0x3ff); } } + + if (aExp < -14) { + /* Note that flush-to-zero does not affect half-precision results */ + is_tiny = + (STATUS(float_detect_tininess) == float_tininess_before_rounding) + || (aExp < -15) + || (!rounding_bumps_exp); + } + if (aSig & mask) { + float_raise(float_flag_inexact STATUS_VAR); + if (is_tiny) { + float_raise(float_flag_underflow STATUS_VAR); + } + } + + aSig += increment; + if (rounding_bumps_exp) { + aSig >>= 1; + aExp++; + } + if (aExp < -24) { return packFloat16(aSign, 0, 0); } From e0a2db7128626a795c3657ffcaff40514e998e2f Mon Sep 17 00:00:00 2001 From: Will Newton Date: Tue, 7 Jan 2014 17:17:48 +0000 Subject: [PATCH 39/61] softfloat: Add float to 16bit integer conversions. ARMv8 requires support for converting 32 and 64bit floating point values to signed and unsigned 16bit integers. Signed-off-by: Will Newton [PMM: updated not to incorrectly set Inexact for Invalid inputs] Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- fpu/softfloat.c | 80 +++++++++++++++++++++++++++++++++++++++++ include/fpu/softfloat.h | 4 +++ 2 files changed, 84 insertions(+) diff --git a/fpu/softfloat.c b/fpu/softfloat.c index 6a6b6567d7..dbaa32c7e3 100644 --- a/fpu/softfloat.c +++ b/fpu/softfloat.c @@ -6491,6 +6491,46 @@ uint32 float32_to_uint32_round_to_zero( float32 a STATUS_PARAM ) return res; } +int_fast16_t float32_to_int16(float32 a STATUS_PARAM) +{ + int32_t v; + int_fast16_t res; + int old_exc_flags = get_float_exception_flags(status); + + v = float32_to_int32(a STATUS_VAR); + if (v < -0x8000) { + res = -0x8000; + } else if (v > 0x7fff) { + res = 0x7fff; + } else { + return v; + } + + set_float_exception_flags(old_exc_flags, status); + float_raise(float_flag_invalid STATUS_VAR); + return res; +} + +uint_fast16_t float32_to_uint16(float32 a STATUS_PARAM) +{ + int32_t v; + uint_fast16_t res; + int old_exc_flags = get_float_exception_flags(status); + + v = float32_to_int32(a STATUS_VAR); + if (v < 0) { + res = 0; + } else if (v > 0xffff) { + res = 0xffff; + } else { + return v; + } + + set_float_exception_flags(old_exc_flags, status); + float_raise(float_flag_invalid STATUS_VAR); + return res; +} + uint_fast16_t float32_to_uint16_round_to_zero(float32 a STATUS_PARAM) { int64_t v; @@ -6545,6 +6585,46 @@ uint32 float64_to_uint32_round_to_zero( float64 a STATUS_PARAM ) return res; } +int_fast16_t float64_to_int16(float64 a STATUS_PARAM) +{ + int64_t v; + int_fast16_t res; + int old_exc_flags = get_float_exception_flags(status); + + v = float64_to_int32(a STATUS_VAR); + if (v < -0x8000) { + res = -0x8000; + } else if (v > 0x7fff) { + res = 0x7fff; + } else { + return v; + } + + set_float_exception_flags(old_exc_flags, status); + float_raise(float_flag_invalid STATUS_VAR); + return res; +} + +uint_fast16_t float64_to_uint16(float64 a STATUS_PARAM) +{ + int64_t v; + uint_fast16_t res; + int old_exc_flags = get_float_exception_flags(status); + + v = float64_to_int32(a STATUS_VAR); + if (v < 0) { + res = 0; + } else if (v > 0xffff) { + res = 0xffff; + } else { + return v; + } + + set_float_exception_flags(old_exc_flags, status); + float_raise(float_flag_invalid STATUS_VAR); + return res; +} + uint_fast16_t float64_to_uint16_round_to_zero(float64 a STATUS_PARAM) { int64_t v; diff --git a/include/fpu/softfloat.h b/include/fpu/softfloat.h index 2365274daa..a9b8cd9950 100644 --- a/include/fpu/softfloat.h +++ b/include/fpu/softfloat.h @@ -265,6 +265,8 @@ extern const float16 float16_default_nan; /*---------------------------------------------------------------------------- | Software IEC/IEEE single-precision conversion routines. *----------------------------------------------------------------------------*/ +int_fast16_t float32_to_int16(float32 STATUS_PARAM); +uint_fast16_t float32_to_uint16(float32 STATUS_PARAM); int_fast16_t float32_to_int16_round_to_zero(float32 STATUS_PARAM); uint_fast16_t float32_to_uint16_round_to_zero(float32 STATUS_PARAM); int32 float32_to_int32( float32 STATUS_PARAM ); @@ -371,6 +373,8 @@ extern const float32 float32_default_nan; /*---------------------------------------------------------------------------- | Software IEC/IEEE double-precision conversion routines. *----------------------------------------------------------------------------*/ +int_fast16_t float64_to_int16(float64 STATUS_PARAM); +uint_fast16_t float64_to_uint16(float64 STATUS_PARAM); int_fast16_t float64_to_int16_round_to_zero(float64 STATUS_PARAM); uint_fast16_t float64_to_uint16_round_to_zero(float64 STATUS_PARAM); int32 float64_to_int32( float64 STATUS_PARAM ); From 44e09132cae159ce1e2f3d0aebd62c8298224a6d Mon Sep 17 00:00:00 2001 From: Peter Maydell Date: Tue, 7 Jan 2014 17:17:49 +0000 Subject: [PATCH 40/61] softfloat: Add 16 bit integer to float conversions Add the float to 16 bit integer conversion routines. These can be trivially implemented in terms of the int32_to_float* routines, but providing them makes our API more symmetrical and can simplify callers. Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- include/fpu/softfloat.h | 21 +++++++++++++++++++++ 1 file changed, 21 insertions(+) diff --git a/include/fpu/softfloat.h b/include/fpu/softfloat.h index a9b8cd9950..926d849252 100644 --- a/include/fpu/softfloat.h +++ b/include/fpu/softfloat.h @@ -239,6 +239,27 @@ floatx80 int64_to_floatx80( int64 STATUS_PARAM ); float128 int64_to_float128( int64 STATUS_PARAM ); float128 uint64_to_float128( uint64 STATUS_PARAM ); +/* We provide the int16 versions for symmetry of API with float-to-int */ +INLINE float32 int16_to_float32(int16_t v STATUS_PARAM) +{ + return int32_to_float32(v STATUS_VAR); +} + +INLINE float32 uint16_to_float32(uint16_t v STATUS_PARAM) +{ + return uint32_to_float32(v STATUS_VAR); +} + +INLINE float64 int16_to_float64(int16_t v STATUS_PARAM) +{ + return int32_to_float64(v STATUS_VAR); +} + +INLINE float64 uint16_to_float64(uint16_t v STATUS_PARAM) +{ + return uint32_to_float64(v STATUS_VAR); +} + /*---------------------------------------------------------------------------- | Software half-precision conversion routines. *----------------------------------------------------------------------------*/ From 564658dc8c5fabdc73702967590a5652bc45107f Mon Sep 17 00:00:00 2001 From: Peter Maydell Date: Tue, 7 Jan 2014 17:17:49 +0000 Subject: [PATCH 41/61] softfloat: Make the int-to-float functions take exact-width types Currently the int-to-float functions take types which are specified as "at least X bits wide", rather than "exactly X bits wide". This is confusing and unhelpful since it means that the callers have to include an explicit cast to [u]intXX_t to ensure the correct behaviour. Fix them all to take the exactly-X-bits-wide types instead. Note that this doesn't change behaviour at all since at the moment we happen to define the 'int32' and 'uint32' types as exactly 32 bits wide, and the 'int64' and 'uint64' types as exactly 64 bits wide. Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- fpu/softfloat.c | 26 +++++++++++++------------- include/fpu/softfloat.h | 26 +++++++++++++------------- 2 files changed, 26 insertions(+), 26 deletions(-) diff --git a/fpu/softfloat.c b/fpu/softfloat.c index dbaa32c7e3..3170b882b9 100644 --- a/fpu/softfloat.c +++ b/fpu/softfloat.c @@ -1121,7 +1121,7 @@ static float128 | according to the IEC/IEEE Standard for Binary Floating-Point Arithmetic. *----------------------------------------------------------------------------*/ -float32 int32_to_float32( int32 a STATUS_PARAM ) +float32 int32_to_float32(int32_t a STATUS_PARAM) { flag zSign; @@ -1138,7 +1138,7 @@ float32 int32_to_float32( int32 a STATUS_PARAM ) | according to the IEC/IEEE Standard for Binary Floating-Point Arithmetic. *----------------------------------------------------------------------------*/ -float64 int32_to_float64( int32 a STATUS_PARAM ) +float64 int32_to_float64(int32_t a STATUS_PARAM) { flag zSign; uint32 absA; @@ -1161,7 +1161,7 @@ float64 int32_to_float64( int32 a STATUS_PARAM ) | Arithmetic. *----------------------------------------------------------------------------*/ -floatx80 int32_to_floatx80( int32 a STATUS_PARAM ) +floatx80 int32_to_floatx80(int32_t a STATUS_PARAM) { flag zSign; uint32 absA; @@ -1183,7 +1183,7 @@ floatx80 int32_to_floatx80( int32 a STATUS_PARAM ) | according to the IEC/IEEE Standard for Binary Floating-Point Arithmetic. *----------------------------------------------------------------------------*/ -float128 int32_to_float128( int32 a STATUS_PARAM ) +float128 int32_to_float128(int32_t a STATUS_PARAM) { flag zSign; uint32 absA; @@ -1205,7 +1205,7 @@ float128 int32_to_float128( int32 a STATUS_PARAM ) | according to the IEC/IEEE Standard for Binary Floating-Point Arithmetic. *----------------------------------------------------------------------------*/ -float32 int64_to_float32( int64 a STATUS_PARAM ) +float32 int64_to_float32(int64_t a STATUS_PARAM) { flag zSign; uint64 absA; @@ -1231,7 +1231,7 @@ float32 int64_to_float32( int64 a STATUS_PARAM ) } -float32 uint64_to_float32( uint64 a STATUS_PARAM ) +float32 uint64_to_float32(uint64_t a STATUS_PARAM) { int8 shiftCount; @@ -1258,7 +1258,7 @@ float32 uint64_to_float32( uint64 a STATUS_PARAM ) | according to the IEC/IEEE Standard for Binary Floating-Point Arithmetic. *----------------------------------------------------------------------------*/ -float64 int64_to_float64( int64 a STATUS_PARAM ) +float64 int64_to_float64(int64_t a STATUS_PARAM) { flag zSign; @@ -1271,7 +1271,7 @@ float64 int64_to_float64( int64 a STATUS_PARAM ) } -float64 uint64_to_float64(uint64 a STATUS_PARAM) +float64 uint64_to_float64(uint64_t a STATUS_PARAM) { int exp = 0x43C; @@ -1292,7 +1292,7 @@ float64 uint64_to_float64(uint64 a STATUS_PARAM) | Arithmetic. *----------------------------------------------------------------------------*/ -floatx80 int64_to_floatx80( int64 a STATUS_PARAM ) +floatx80 int64_to_floatx80(int64_t a STATUS_PARAM) { flag zSign; uint64 absA; @@ -1312,7 +1312,7 @@ floatx80 int64_to_floatx80( int64 a STATUS_PARAM ) | according to the IEC/IEEE Standard for Binary Floating-Point Arithmetic. *----------------------------------------------------------------------------*/ -float128 int64_to_float128( int64 a STATUS_PARAM ) +float128 int64_to_float128(int64_t a STATUS_PARAM) { flag zSign; uint64 absA; @@ -1339,7 +1339,7 @@ float128 int64_to_float128( int64 a STATUS_PARAM ) } -float128 uint64_to_float128(uint64 a STATUS_PARAM) +float128 uint64_to_float128(uint64_t a STATUS_PARAM) { if (a == 0) { return float128_zero; @@ -6445,12 +6445,12 @@ int float128_unordered_quiet( float128 a, float128 b STATUS_PARAM ) } /* misc functions */ -float32 uint32_to_float32( uint32 a STATUS_PARAM ) +float32 uint32_to_float32(uint32_t a STATUS_PARAM) { return int64_to_float32(a STATUS_VAR); } -float64 uint32_to_float64( uint32 a STATUS_PARAM ) +float64 uint32_to_float64(uint32_t a STATUS_PARAM) { return int64_to_float64(a STATUS_VAR); } diff --git a/include/fpu/softfloat.h b/include/fpu/softfloat.h index 926d849252..78b165670d 100644 --- a/include/fpu/softfloat.h +++ b/include/fpu/softfloat.h @@ -225,19 +225,19 @@ enum { /*---------------------------------------------------------------------------- | Software IEC/IEEE integer-to-floating-point conversion routines. *----------------------------------------------------------------------------*/ -float32 int32_to_float32( int32 STATUS_PARAM ); -float64 int32_to_float64( int32 STATUS_PARAM ); -float32 uint32_to_float32( uint32 STATUS_PARAM ); -float64 uint32_to_float64( uint32 STATUS_PARAM ); -floatx80 int32_to_floatx80( int32 STATUS_PARAM ); -float128 int32_to_float128( int32 STATUS_PARAM ); -float32 int64_to_float32( int64 STATUS_PARAM ); -float32 uint64_to_float32( uint64 STATUS_PARAM ); -float64 int64_to_float64( int64 STATUS_PARAM ); -float64 uint64_to_float64( uint64 STATUS_PARAM ); -floatx80 int64_to_floatx80( int64 STATUS_PARAM ); -float128 int64_to_float128( int64 STATUS_PARAM ); -float128 uint64_to_float128( uint64 STATUS_PARAM ); +float32 int32_to_float32(int32_t STATUS_PARAM); +float64 int32_to_float64(int32_t STATUS_PARAM); +float32 uint32_to_float32(uint32_t STATUS_PARAM); +float64 uint32_to_float64(uint32_t STATUS_PARAM); +floatx80 int32_to_floatx80(int32_t STATUS_PARAM); +float128 int32_to_float128(int32_t STATUS_PARAM); +float32 int64_to_float32(int64_t STATUS_PARAM); +float32 uint64_to_float32(uint64_t STATUS_PARAM); +float64 int64_to_float64(int64_t STATUS_PARAM); +float64 uint64_to_float64(uint64_t STATUS_PARAM); +floatx80 int64_to_floatx80(int64_t STATUS_PARAM); +float128 int64_to_float128(int64_t STATUS_PARAM); +float128 uint64_to_float128(uint64_t STATUS_PARAM); /* We provide the int16 versions for symmetry of API with float-to-int */ INLINE float32 int16_to_float32(int16_t v STATUS_PARAM) From f883b3896ed5e180f234613aadca019fdbe7c49c Mon Sep 17 00:00:00 2001 From: Tom Musta Date: Tue, 7 Jan 2014 17:17:49 +0000 Subject: [PATCH 42/61] softfloat: Fix float64_to_uint64 The comment preceding the float64_to_uint64 routine suggests that the implementation is broken. And this is, indeed, the case. This patch properly implements the conversion of a 64-bit floating point number to an unsigned, 64 bit integer. This contribution can be licensed under either the softfloat-2a or -2b license. Signed-off-by: Tom Musta Reviewed-by: Peter Maydell Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- fpu/softfloat.c | 101 ++++++++++++++++++++++++++++++++++++++++++++---- 1 file changed, 93 insertions(+), 8 deletions(-) diff --git a/fpu/softfloat.c b/fpu/softfloat.c index 3170b882b9..2364513e0b 100644 --- a/fpu/softfloat.c +++ b/fpu/softfloat.c @@ -203,6 +203,56 @@ static int64 roundAndPackInt64( flag zSign, uint64_t absZ0, uint64_t absZ1 STATU } +/*---------------------------------------------------------------------------- +| Takes the 128-bit fixed-point value formed by concatenating `absZ0' and +| `absZ1', with binary point between bits 63 and 64 (between the input words), +| and returns the properly rounded 64-bit unsigned integer corresponding to the +| input. Ordinarily, the fixed-point input is simply rounded to an integer, +| with the inexact exception raised if the input cannot be represented exactly +| as an integer. However, if the fixed-point input is too large, the invalid +| exception is raised and the largest unsigned integer is returned. +*----------------------------------------------------------------------------*/ + +static int64 roundAndPackUint64(flag zSign, uint64_t absZ0, + uint64_t absZ1 STATUS_PARAM) +{ + int8 roundingMode; + flag roundNearestEven, increment; + + roundingMode = STATUS(float_rounding_mode); + roundNearestEven = (roundingMode == float_round_nearest_even); + increment = ((int64_t)absZ1 < 0); + if (!roundNearestEven) { + if (roundingMode == float_round_to_zero) { + increment = 0; + } else if (absZ1) { + if (zSign) { + increment = (roundingMode == float_round_down) && absZ1; + } else { + increment = (roundingMode == float_round_up) && absZ1; + } + } + } + if (increment) { + ++absZ0; + if (absZ0 == 0) { + float_raise(float_flag_invalid STATUS_VAR); + return LIT64(0xFFFFFFFFFFFFFFFF); + } + absZ0 &= ~(((uint64_t)(absZ1<<1) == 0) & roundNearestEven); + } + + if (zSign && absZ0) { + float_raise(float_flag_invalid STATUS_VAR); + return 0; + } + + if (absZ1) { + STATUS(float_exception_flags) |= float_flag_inexact; + } + return absZ0; +} + /*---------------------------------------------------------------------------- | Returns the fraction bits of the single-precision floating-point value `a'. *----------------------------------------------------------------------------*/ @@ -6643,16 +6693,51 @@ uint_fast16_t float64_to_uint16_round_to_zero(float64 a STATUS_PARAM) return res; } -/* FIXME: This looks broken. */ -uint64_t float64_to_uint64 (float64 a STATUS_PARAM) +/*---------------------------------------------------------------------------- +| Returns the result of converting the double-precision floating-point value +| `a' to the 64-bit unsigned integer format. The conversion is +| performed according to the IEC/IEEE Standard for Binary Floating-Point +| Arithmetic---which means in particular that the conversion is rounded +| according to the current rounding mode. If `a' is a NaN, the largest +| positive integer is returned. If the conversion overflows, the +| largest unsigned integer is returned. If 'a' is negative, the value is +| rounded and zero is returned; negative values that do not round to zero +| will raise the inexact exception. +*----------------------------------------------------------------------------*/ + +uint64_t float64_to_uint64(float64 a STATUS_PARAM) { - int64_t v; + flag aSign; + int_fast16_t aExp, shiftCount; + uint64_t aSig, aSigExtra; + a = float64_squash_input_denormal(a STATUS_VAR); - v = float64_val(int64_to_float64(INT64_MIN STATUS_VAR)); - v += float64_val(a); - v = float64_to_int64(make_float64(v) STATUS_VAR); - - return v - INT64_MIN; + aSig = extractFloat64Frac(a); + aExp = extractFloat64Exp(a); + aSign = extractFloat64Sign(a); + if (aSign && (aExp > 1022)) { + float_raise(float_flag_invalid STATUS_VAR); + if (float64_is_any_nan(a)) { + return LIT64(0xFFFFFFFFFFFFFFFF); + } else { + return 0; + } + } + if (aExp) { + aSig |= LIT64(0x0010000000000000); + } + shiftCount = 0x433 - aExp; + if (shiftCount <= 0) { + if (0x43E < aExp) { + float_raise(float_flag_invalid STATUS_VAR); + return LIT64(0xFFFFFFFFFFFFFFFF); + } + aSigExtra = 0; + aSig <<= -shiftCount; + } else { + shift64ExtraRightJamming(aSig, 0, shiftCount, &aSig, &aSigExtra); + } + return roundAndPackUint64(aSign, aSig, aSigExtra STATUS_VAR); } uint64_t float64_to_uint64_round_to_zero (float64 a STATUS_PARAM) From f2e933d20d5fd6c38bda227359b79bcc81654f99 Mon Sep 17 00:00:00 2001 From: Peter Maydell Date: Tue, 7 Jan 2014 17:17:49 +0000 Subject: [PATCH 43/61] softfloat: Only raise Invalid when conversions to int are out of range We implement a number of float-to-integer conversions using conversion to an integer type with a wider range and then a check against the narrower range we are actually converting to. If we find the result to be out of range we correctly raise the Invalid exception, but we must also suppress other exceptions which might have been raised by the conversion function we called. This won't throw away exceptions we should have preserved, because for the 'core' exception flags the IEEE spec mandates that the only valid combinations of exception that can be raised by a single operation are Inexact + Overflow and Inexact + Underflow. For the non-IEEE softfloat flag for input denormals, we can guarantee that that flag won't have been set for out of range float-to-int conversions because a squashed denormal by definition goes to plus or minus zero, which is always in range after conversion to integer zero. This bug has been fixed for some of the float-to-int conversion routines by previous patches; fix it for the remaining functions as well, so that they all restore the pre-conversion status flags prior to raising Invalid. Signed-off-by: Peter Maydell Reviewed-by: Aurelien Jarno Reviewed-by: Richard Henderson --- fpu/softfloat.c | 28 ++++++++++++++++------------ 1 file changed, 16 insertions(+), 12 deletions(-) diff --git a/fpu/softfloat.c b/fpu/softfloat.c index 2364513e0b..6312e0c66f 100644 --- a/fpu/softfloat.c +++ b/fpu/softfloat.c @@ -6509,17 +6509,18 @@ uint32 float32_to_uint32( float32 a STATUS_PARAM ) { int64_t v; uint32 res; + int old_exc_flags = get_float_exception_flags(status); v = float32_to_int64(a STATUS_VAR); if (v < 0) { res = 0; - float_raise( float_flag_invalid STATUS_VAR); } else if (v > 0xffffffff) { res = 0xffffffff; - float_raise( float_flag_invalid STATUS_VAR); } else { - res = v; + return v; } + set_float_exception_flags(old_exc_flags, status); + float_raise(float_flag_invalid STATUS_VAR); return res; } @@ -6527,17 +6528,18 @@ uint32 float32_to_uint32_round_to_zero( float32 a STATUS_PARAM ) { int64_t v; uint32 res; + int old_exc_flags = get_float_exception_flags(status); v = float32_to_int64_round_to_zero(a STATUS_VAR); if (v < 0) { res = 0; - float_raise( float_flag_invalid STATUS_VAR); } else if (v > 0xffffffff) { res = 0xffffffff; - float_raise( float_flag_invalid STATUS_VAR); } else { - res = v; + return v; } + set_float_exception_flags(old_exc_flags, status); + float_raise(float_flag_invalid STATUS_VAR); return res; } @@ -6585,17 +6587,18 @@ uint_fast16_t float32_to_uint16_round_to_zero(float32 a STATUS_PARAM) { int64_t v; uint_fast16_t res; + int old_exc_flags = get_float_exception_flags(status); v = float32_to_int64_round_to_zero(a STATUS_VAR); if (v < 0) { res = 0; - float_raise( float_flag_invalid STATUS_VAR); } else if (v > 0xffff) { res = 0xffff; - float_raise( float_flag_invalid STATUS_VAR); } else { - res = v; + return v; } + set_float_exception_flags(old_exc_flags, status); + float_raise(float_flag_invalid STATUS_VAR); return res; } @@ -6679,17 +6682,18 @@ uint_fast16_t float64_to_uint16_round_to_zero(float64 a STATUS_PARAM) { int64_t v; uint_fast16_t res; + int old_exc_flags = get_float_exception_flags(status); v = float64_to_int64_round_to_zero(a STATUS_VAR); if (v < 0) { res = 0; - float_raise( float_flag_invalid STATUS_VAR); } else if (v > 0xffff) { res = 0xffff; - float_raise( float_flag_invalid STATUS_VAR); } else { - res = v; + return v; } + set_float_exception_flags(old_exc_flags, status); + float_raise(float_flag_invalid STATUS_VAR); return res; } From 0ace25a5676af233a9b5abf48599dbcc45f92dcf Mon Sep 17 00:00:00 2001 From: Peter Maydell Date: Tue, 7 Jan 2014 17:17:50 +0000 Subject: [PATCH 44/61] softfloat: Fix factor 2 error for scalbn on denormal inputs If the input to float*_scalbn() is denormal then it represents a number 0.[mantissabits] * 2^(1-exponentbias) (and the actual exponent field is all zeroes). This means that when we convert it to our unpacked encoding the unpacked exponent must be one greater than for a normal number, which represents 1.[mantissabits] * 2^(e-exponentbias) for an exponent field e. This meant we were giving answers too small by a factor of 2 for all denormal inputs. Note that the float-to-int routines also have this behaviour of not adjusting the exponent for denormals; however there it is harmless because denormals will all convert to integer zero anyway. Signed-off-by: Peter Maydell Reviewed-by: Aurelien Jarno Reviewed-by: Richard Henderson --- fpu/softfloat.c | 29 +++++++++++++++++++++-------- 1 file changed, 21 insertions(+), 8 deletions(-) diff --git a/fpu/softfloat.c b/fpu/softfloat.c index 6312e0c66f..d2e909593f 100644 --- a/fpu/softfloat.c +++ b/fpu/softfloat.c @@ -6991,10 +6991,13 @@ float32 float32_scalbn( float32 a, int n STATUS_PARAM ) } return a; } - if ( aExp != 0 ) + if (aExp != 0) { aSig |= 0x00800000; - else if ( aSig == 0 ) + } else if (aSig == 0) { return a; + } else { + aExp++; + } if (n > 0x200) { n = 0x200; @@ -7024,10 +7027,13 @@ float64 float64_scalbn( float64 a, int n STATUS_PARAM ) } return a; } - if ( aExp != 0 ) + if (aExp != 0) { aSig |= LIT64( 0x0010000000000000 ); - else if ( aSig == 0 ) + } else if (aSig == 0) { return a; + } else { + aExp++; + } if (n > 0x1000) { n = 0x1000; @@ -7057,8 +7063,12 @@ floatx80 floatx80_scalbn( floatx80 a, int n STATUS_PARAM ) return a; } - if (aExp == 0 && aSig == 0) - return a; + if (aExp == 0) { + if (aSig == 0) { + return a; + } + aExp++; + } if (n > 0x10000) { n = 0x10000; @@ -7087,10 +7097,13 @@ float128 float128_scalbn( float128 a, int n STATUS_PARAM ) } return a; } - if ( aExp != 0 ) + if (aExp != 0) { aSig0 |= LIT64( 0x0001000000000000 ); - else if ( aSig0 == 0 && aSig1 == 0 ) + } else if (aSig0 == 0 && aSig1 == 0) { return a; + } else { + aExp++; + } if (n > 0x10000) { n = 0x10000; From 47a34e009de2f4a99862e71e1d15d178dac5f259 Mon Sep 17 00:00:00 2001 From: Tom Musta Date: Tue, 7 Jan 2014 17:17:50 +0000 Subject: [PATCH 45/61] softfloat: Add float32_to_uint64() This patch adds the float32_to_uint64() routine, which converts a 32-bit floating point number to an unsigned 64 bit number. This contribution can be licensed under either the softfloat-2a or -2b license. Signed-off-by: Tom Musta Reviewed-by: Peter Maydell [PMM: removed harmless but silly int64_t casts] Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- fpu/softfloat.c | 46 +++++++++++++++++++++++++++++++++++++++++ include/fpu/softfloat.h | 1 + 2 files changed, 47 insertions(+) diff --git a/fpu/softfloat.c b/fpu/softfloat.c index d2e909593f..c2c1a5650b 100644 --- a/fpu/softfloat.c +++ b/fpu/softfloat.c @@ -1557,6 +1557,52 @@ int64 float32_to_int64( float32 a STATUS_PARAM ) } +/*---------------------------------------------------------------------------- +| Returns the result of converting the single-precision floating-point value +| `a' to the 64-bit unsigned integer format. The conversion is +| performed according to the IEC/IEEE Standard for Binary Floating-Point +| Arithmetic---which means in particular that the conversion is rounded +| according to the current rounding mode. If `a' is a NaN, the largest +| unsigned integer is returned. Otherwise, if the conversion overflows, the +| largest unsigned integer is returned. If the 'a' is negative, the result +| is rounded and zero is returned; values that do not round to zero will +| raise the inexact exception flag. +*----------------------------------------------------------------------------*/ + +uint64 float32_to_uint64(float32 a STATUS_PARAM) +{ + flag aSign; + int_fast16_t aExp, shiftCount; + uint32_t aSig; + uint64_t aSig64, aSigExtra; + a = float32_squash_input_denormal(a STATUS_VAR); + + aSig = extractFloat32Frac(a); + aExp = extractFloat32Exp(a); + aSign = extractFloat32Sign(a); + if ((aSign) && (aExp > 126)) { + float_raise(float_flag_invalid STATUS_VAR); + if (float32_is_any_nan(a)) { + return LIT64(0xFFFFFFFFFFFFFFFF); + } else { + return 0; + } + } + shiftCount = 0xBE - aExp; + if (aExp) { + aSig |= 0x00800000; + } + if (shiftCount < 0) { + float_raise(float_flag_invalid STATUS_VAR); + return LIT64(0xFFFFFFFFFFFFFFFF); + } + + aSig64 = aSig; + aSig64 <<= 40; + shift64ExtraRightJamming(aSig64, 0, shiftCount, &aSig64, &aSigExtra); + return roundAndPackUint64(aSign, aSig64, aSigExtra STATUS_VAR); +} + /*---------------------------------------------------------------------------- | Returns the result of converting the single-precision floating-point value | `a' to the 64-bit two's complement integer format. The conversion is diff --git a/include/fpu/softfloat.h b/include/fpu/softfloat.h index 78b165670d..eb81c3b0cc 100644 --- a/include/fpu/softfloat.h +++ b/include/fpu/softfloat.h @@ -295,6 +295,7 @@ int32 float32_to_int32_round_to_zero( float32 STATUS_PARAM ); uint32 float32_to_uint32( float32 STATUS_PARAM ); uint32 float32_to_uint32_round_to_zero( float32 STATUS_PARAM ); int64 float32_to_int64( float32 STATUS_PARAM ); +uint64 float32_to_uint64(float32 STATUS_PARAM); int64 float32_to_int64_round_to_zero( float32 STATUS_PARAM ); float64 float32_to_float64( float32 STATUS_PARAM ); floatx80 float32_to_floatx80( float32 STATUS_PARAM ); From a757f1c2deb6c8eac4a75f8e817fa2d7c8cd2527 Mon Sep 17 00:00:00 2001 From: Tom Musta Date: Tue, 7 Jan 2014 17:17:50 +0000 Subject: [PATCH 46/61] softfloat: Fix float64_to_uint64_round_to_zero The float64_to_uint64_round_to_zero routine is incorrect. For example, the following test pattern: 46697351FF4AEC29 / 0x1.97351ff4aec29p+103 currently produces 8000000000000000 instead of FFFFFFFFFFFFFFFF. This patch re-implements the routine to temporarily force the rounding mode and use the float64_to_uint64 routine. This contribution can be licensed under either the softfloat-2a or -2b license. Signed-off-by: Tom Musta Message-id: 1387397961-4894-4-git-send-email-tommusta@gmail.com Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- fpu/softfloat.c | 12 +++++------- 1 file changed, 5 insertions(+), 7 deletions(-) diff --git a/fpu/softfloat.c b/fpu/softfloat.c index c2c1a5650b..f782c24ef2 100644 --- a/fpu/softfloat.c +++ b/fpu/softfloat.c @@ -6792,13 +6792,11 @@ uint64_t float64_to_uint64(float64 a STATUS_PARAM) uint64_t float64_to_uint64_round_to_zero (float64 a STATUS_PARAM) { - int64_t v; - - v = float64_val(int64_to_float64(INT64_MIN STATUS_VAR)); - v += float64_val(a); - v = float64_to_int64_round_to_zero(make_float64(v) STATUS_VAR); - - return v - INT64_MIN; + signed char current_rounding_mode = STATUS(float_rounding_mode); + set_float_rounding_mode(float_round_to_zero STATUS_VAR); + int64_t v = float64_to_uint64(a STATUS_VAR); + set_float_rounding_mode(current_rounding_mode STATUS_VAR); + return v; } #define COMPARE(s, nan_exp) \ From 8d551c988e96ac4cfe462e6b34d513bba2da572f Mon Sep 17 00:00:00 2001 From: Tom Musta Date: Tue, 7 Jan 2014 17:17:51 +0000 Subject: [PATCH 47/61] softfloat: Fix float64_to_uint32 The float64_to_uint32 has several flaws: - for numbers between 2**32 and 2**64, the inexact exception flag may get incorrectly set. In this case, only the invalid flag should be set. test pattern: 425F81378DC0CD1F / 0x1.f81378dc0cd1fp+38 - for numbers between 2**63 and 2**64, incorrect results may be produced: test pattern: 43EAAF73F1F0B8BD / 0x1.aaf73f1f0b8bdp+63 This patch re-implements float64_to_uint32 to re-use the float64_to_uint64 routine (instead of float64_to_int64). For the saturation case, we ignore any flags which the conversion routine has set and raise only the invalid flag. This contribution can be licensed under either the softfloat-2a or -2b license. Signed-off-by: Tom Musta Message-id: 1387397961-4894-5-git-send-email-tommusta@gmail.com Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- fpu/softfloat.c | 15 +++++++-------- 1 file changed, 7 insertions(+), 8 deletions(-) diff --git a/fpu/softfloat.c b/fpu/softfloat.c index f782c24ef2..e68a87b6da 100644 --- a/fpu/softfloat.c +++ b/fpu/softfloat.c @@ -6650,19 +6650,18 @@ uint_fast16_t float32_to_uint16_round_to_zero(float32 a STATUS_PARAM) uint32 float64_to_uint32( float64 a STATUS_PARAM ) { - int64_t v; + uint64_t v; uint32 res; + int old_exc_flags = get_float_exception_flags(status); - v = float64_to_int64(a STATUS_VAR); - if (v < 0) { - res = 0; - float_raise( float_flag_invalid STATUS_VAR); - } else if (v > 0xffffffff) { + v = float64_to_uint64(a STATUS_VAR); + if (v > 0xffffffff) { res = 0xffffffff; - float_raise( float_flag_invalid STATUS_VAR); } else { - res = v; + return v; } + set_float_exception_flags(old_exc_flags, status); + float_raise(float_flag_invalid STATUS_VAR); return res; } From 2958f14d75b9c2ef9a68f41a3101884af307e4f8 Mon Sep 17 00:00:00 2001 From: Tom Musta Date: Tue, 7 Jan 2014 17:17:51 +0000 Subject: [PATCH 48/61] softfloat: Fix float64_to_uint32_round_to_zero The float64_to_uint32_round_to_zero routine is incorrect. For example, the following test pattern: 425F81378DC0CD1F / 0x1.f81378dc0cd1fp+38 will erroneously set the inexact flag. This patch re-implements the routine to use the float64_to_uint64_round_to_zero routine. If saturation occurs we ignore any flags set by the conversion function and raise only Invalid. This contribution can be licensed under either the softfloat-2a or -2b license. Signed-off-by: Tom Musta Message-id: 1387397961-4894-6-git-send-email-tommusta@gmail.com Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- fpu/softfloat.c | 15 +++++++-------- 1 file changed, 7 insertions(+), 8 deletions(-) diff --git a/fpu/softfloat.c b/fpu/softfloat.c index e68a87b6da..3232ce2547 100644 --- a/fpu/softfloat.c +++ b/fpu/softfloat.c @@ -6667,19 +6667,18 @@ uint32 float64_to_uint32( float64 a STATUS_PARAM ) uint32 float64_to_uint32_round_to_zero( float64 a STATUS_PARAM ) { - int64_t v; + uint64_t v; uint32 res; + int old_exc_flags = get_float_exception_flags(status); - v = float64_to_int64_round_to_zero(a STATUS_VAR); - if (v < 0) { - res = 0; - float_raise( float_flag_invalid STATUS_VAR); - } else if (v > 0xffffffff) { + v = float64_to_uint64_round_to_zero(a STATUS_VAR); + if (v > 0xffffffff) { res = 0xffffffff; - float_raise( float_flag_invalid STATUS_VAR); } else { - res = v; + return v; } + set_float_exception_flags(old_exc_flags, status); + float_raise(float_flag_invalid STATUS_VAR); return res; } From e9ef30c079354db6a8eb5bd6574b948288a55b94 Mon Sep 17 00:00:00 2001 From: Peter Maydell Date: Tue, 7 Jan 2014 17:19:11 +0000 Subject: [PATCH 49/61] softfloat: Provide complete set of accessors for fp state Tidy up the get/set accessors for the fp state to add missing ones and make them all inline in softfloat.h rather than some inline and some not. Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- fpu/softfloat.c | 15 --------------- include/fpu/softfloat.h | 39 ++++++++++++++++++++++++++++++++++++--- 2 files changed, 36 insertions(+), 18 deletions(-) diff --git a/fpu/softfloat.c b/fpu/softfloat.c index 3232ce2547..4abcd36b15 100644 --- a/fpu/softfloat.c +++ b/fpu/softfloat.c @@ -59,21 +59,6 @@ these four paragraphs for those parts of this code that are retained. *----------------------------------------------------------------------------*/ #include "softfloat-specialize.h" -void set_float_rounding_mode(int val STATUS_PARAM) -{ - STATUS(float_rounding_mode) = val; -} - -void set_float_exception_flags(int val STATUS_PARAM) -{ - STATUS(float_exception_flags) = val; -} - -void set_floatx80_rounding_precision(int val STATUS_PARAM) -{ - STATUS(floatx80_rounding_precision) = val; -} - /*---------------------------------------------------------------------------- | Returns the fraction bits of the half-precision floating-point value `a'. *----------------------------------------------------------------------------*/ diff --git a/include/fpu/softfloat.h b/include/fpu/softfloat.h index eb81c3b0cc..a634a4ebff 100644 --- a/include/fpu/softfloat.h +++ b/include/fpu/softfloat.h @@ -180,12 +180,22 @@ typedef struct float_status { flag default_nan_mode; } float_status; -void set_float_rounding_mode(int val STATUS_PARAM); -void set_float_exception_flags(int val STATUS_PARAM); INLINE void set_float_detect_tininess(int val STATUS_PARAM) { STATUS(float_detect_tininess) = val; } +INLINE void set_float_rounding_mode(int val STATUS_PARAM) +{ + STATUS(float_rounding_mode) = val; +} +INLINE void set_float_exception_flags(int val STATUS_PARAM) +{ + STATUS(float_exception_flags) = val; +} +INLINE void set_floatx80_rounding_precision(int val STATUS_PARAM) +{ + STATUS(floatx80_rounding_precision) = val; +} INLINE void set_flush_to_zero(flag val STATUS_PARAM) { STATUS(flush_to_zero) = val; @@ -198,11 +208,34 @@ INLINE void set_default_nan_mode(flag val STATUS_PARAM) { STATUS(default_nan_mode) = val; } +INLINE int get_float_detect_tininess(float_status *status) +{ + return STATUS(float_detect_tininess); +} +INLINE int get_float_rounding_mode(float_status *status) +{ + return STATUS(float_rounding_mode); +} INLINE int get_float_exception_flags(float_status *status) { return STATUS(float_exception_flags); } -void set_floatx80_rounding_precision(int val STATUS_PARAM); +INLINE int get_floatx80_rounding_precision(float_status *status) +{ + return STATUS(floatx80_rounding_precision); +} +INLINE flag get_flush_to_zero(float_status *status) +{ + return STATUS(flush_to_zero); +} +INLINE flag get_flush_inputs_to_zero(float_status *status) +{ + return STATUS(flush_inputs_to_zero); +} +INLINE flag get_default_nan_mode(float_status *status) +{ + return STATUS(default_nan_mode); +} /*---------------------------------------------------------------------------- | Routine to raise any or all of the software IEC/IEEE floating-point From 84c6b75d8557a6e638dc72343428b0a78b66923f Mon Sep 17 00:00:00 2001 From: Peter Maydell Date: Tue, 7 Jan 2014 17:19:11 +0000 Subject: [PATCH 50/61] softfloat: Factor out RoundAndPackFloat16 and NormalizeFloat16Subnormal In preparation for adding conversions between float16 and float64, factor out code currently done inline in the float16<=>float32 conversion functions into functions RoundAndPackFloat16 and NormalizeFloat16Subnormal along the lines of the existing versions for the other float types. Note that we change the handling of zExp from the inline code to match the API of the other RoundAndPackFloat functions; however we leave the positioning of the binary point between bits 22 and 23 rather than shifting it up to the high end of the word. Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- fpu/softfloat.c | 209 +++++++++++++++++++++++++++++------------------- 1 file changed, 125 insertions(+), 84 deletions(-) diff --git a/fpu/softfloat.c b/fpu/softfloat.c index 4abcd36b15..c63e011ae8 100644 --- a/fpu/softfloat.c +++ b/fpu/softfloat.c @@ -3086,6 +3086,127 @@ static float16 packFloat16(flag zSign, int_fast16_t zExp, uint16_t zSig) (((uint32_t)zSign) << 15) + (((uint32_t)zExp) << 10) + zSig); } +/*---------------------------------------------------------------------------- +| Takes an abstract floating-point value having sign `zSign', exponent `zExp', +| and significand `zSig', and returns the proper half-precision floating- +| point value corresponding to the abstract input. Ordinarily, the abstract +| value is simply rounded and packed into the half-precision format, with +| the inexact exception raised if the abstract input cannot be represented +| exactly. However, if the abstract value is too large, the overflow and +| inexact exceptions are raised and an infinity or maximal finite value is +| returned. If the abstract value is too small, the input value is rounded to +| a subnormal number, and the underflow and inexact exceptions are raised if +| the abstract input cannot be represented exactly as a subnormal half- +| precision floating-point number. +| The `ieee' flag indicates whether to use IEEE standard half precision, or +| ARM-style "alternative representation", which omits the NaN and Inf +| encodings in order to raise the maximum representable exponent by one. +| The input significand `zSig' has its binary point between bits 22 +| and 23, which is 13 bits to the left of the usual location. This shifted +| significand must be normalized or smaller. If `zSig' is not normalized, +| `zExp' must be 0; in that case, the result returned is a subnormal number, +| and it must not require rounding. In the usual case that `zSig' is +| normalized, `zExp' must be 1 less than the ``true'' floating-point exponent. +| Note the slightly odd position of the binary point in zSig compared with the +| other roundAndPackFloat functions. This should probably be fixed if we +| need to implement more float16 routines than just conversion. +| The handling of underflow and overflow follows the IEC/IEEE Standard for +| Binary Floating-Point Arithmetic. +*----------------------------------------------------------------------------*/ + +static float32 roundAndPackFloat16(flag zSign, int_fast16_t zExp, + uint32_t zSig, flag ieee STATUS_PARAM) +{ + int maxexp = ieee ? 29 : 30; + uint32_t mask; + uint32_t increment; + int8 roundingMode; + bool rounding_bumps_exp; + bool is_tiny = false; + + /* Calculate the mask of bits of the mantissa which are not + * representable in half-precision and will be lost. + */ + if (zExp < 1) { + /* Will be denormal in halfprec */ + mask = 0x00ffffff; + if (zExp >= -11) { + mask >>= 11 + zExp; + } + } else { + /* Normal number in halfprec */ + mask = 0x00001fff; + } + + roundingMode = STATUS(float_rounding_mode); + switch (roundingMode) { + case float_round_nearest_even: + increment = (mask + 1) >> 1; + if ((zSig & mask) == increment) { + increment = zSig & (increment << 1); + } + break; + case float_round_up: + increment = zSign ? 0 : mask; + break; + case float_round_down: + increment = zSign ? mask : 0; + break; + default: /* round_to_zero */ + increment = 0; + break; + } + + rounding_bumps_exp = (zSig + increment >= 0x01000000); + + if (zExp > maxexp || (zExp == maxexp && rounding_bumps_exp)) { + if (ieee) { + float_raise(float_flag_overflow | float_flag_inexact STATUS_VAR); + return packFloat16(zSign, 0x1f, 0); + } else { + float_raise(float_flag_invalid STATUS_VAR); + return packFloat16(zSign, 0x1f, 0x3ff); + } + } + + if (zExp < 0) { + /* Note that flush-to-zero does not affect half-precision results */ + is_tiny = + (STATUS(float_detect_tininess) == float_tininess_before_rounding) + || (zExp < -1) + || (!rounding_bumps_exp); + } + if (zSig & mask) { + float_raise(float_flag_inexact STATUS_VAR); + if (is_tiny) { + float_raise(float_flag_underflow STATUS_VAR); + } + } + + zSig += increment; + if (rounding_bumps_exp) { + zSig >>= 1; + zExp++; + } + + if (zExp < -10) { + return packFloat16(zSign, 0, 0); + } + if (zExp < 0) { + zSig >>= -zExp; + zExp = 0; + } + return packFloat16(zSign, zExp, zSig >> 13); +} + +static void normalizeFloat16Subnormal(uint32_t aSig, int_fast16_t *zExpPtr, + uint32_t *zSigPtr) +{ + int8_t shiftCount = countLeadingZeros32(aSig) - 21; + *zSigPtr = aSig << shiftCount; + *zExpPtr = 1 - shiftCount; +} + /* Half precision floats come in two formats: standard IEEE and "ARM" format. The latter gains extra exponent range by omitting the NaN/Inf encodings. */ @@ -3106,15 +3227,12 @@ float32 float16_to_float32(float16 a, flag ieee STATUS_PARAM) return packFloat32(aSign, 0xff, 0); } if (aExp == 0) { - int8 shiftCount; - if (aSig == 0) { return packFloat32(aSign, 0, 0); } - shiftCount = countLeadingZeros32( aSig ) - 21; - aSig = aSig << shiftCount; - aExp = -shiftCount; + normalizeFloat16Subnormal(aSig, &aExp, &aSig); + aExp--; } return packFloat32( aSign, aExp + 0x70, aSig << 13); } @@ -3124,12 +3242,6 @@ float16 float32_to_float16(float32 a, flag ieee STATUS_PARAM) flag aSign; int_fast16_t aExp; uint32_t aSig; - uint32_t mask; - uint32_t increment; - int8 roundingMode; - int maxexp = ieee ? 15 : 16; - bool rounding_bumps_exp; - bool is_tiny = false; a = float32_squash_input_denormal(a STATUS_VAR); @@ -3164,80 +3276,9 @@ float16 float32_to_float16(float32 a, flag ieee STATUS_PARAM) * codepath. */ aSig |= 0x00800000; - aExp -= 0x7f; - /* Calculate the mask of bits of the mantissa which are not - * representable in half-precision and will be lost. - */ - if (aExp < -14) { - /* Will be denormal in halfprec */ - mask = 0x00ffffff; - if (aExp >= -24) { - mask >>= 25 + aExp; - } - } else { - /* Normal number in halfprec */ - mask = 0x00001fff; - } + aExp -= 0x71; - roundingMode = STATUS(float_rounding_mode); - switch (roundingMode) { - case float_round_nearest_even: - increment = (mask + 1) >> 1; - if ((aSig & mask) == increment) { - increment = aSig & (increment << 1); - } - break; - case float_round_up: - increment = aSign ? 0 : mask; - break; - case float_round_down: - increment = aSign ? mask : 0; - break; - default: /* round_to_zero */ - increment = 0; - break; - } - - rounding_bumps_exp = (aSig + increment >= 0x01000000); - - if (aExp > maxexp || (aExp == maxexp && rounding_bumps_exp)) { - if (ieee) { - float_raise(float_flag_overflow | float_flag_inexact STATUS_VAR); - return packFloat16(aSign, 0x1f, 0); - } else { - float_raise(float_flag_invalid STATUS_VAR); - return packFloat16(aSign, 0x1f, 0x3ff); - } - } - - if (aExp < -14) { - /* Note that flush-to-zero does not affect half-precision results */ - is_tiny = - (STATUS(float_detect_tininess) == float_tininess_before_rounding) - || (aExp < -15) - || (!rounding_bumps_exp); - } - if (aSig & mask) { - float_raise(float_flag_inexact STATUS_VAR); - if (is_tiny) { - float_raise(float_flag_underflow STATUS_VAR); - } - } - - aSig += increment; - if (rounding_bumps_exp) { - aSig >>= 1; - aExp++; - } - - if (aExp < -24) { - return packFloat16(aSign, 0, 0); - } - if (aExp < -14) { - aSig >>= -14 - aExp; - aExp = -14; - } - return packFloat16(aSign, aExp + 14, aSig >> 13); + return roundAndPackFloat16(aSign, aExp, aSig, ieee STATUS_VAR); } /*---------------------------------------------------------------------------- From 7b378119bd67609f25bd78b714d0fab922600f40 Mon Sep 17 00:00:00 2001 From: Peter Maydell Date: Tue, 7 Jan 2014 17:19:12 +0000 Subject: [PATCH 51/61] softfloat: Add float16 <=> float64 conversion functions Add the conversion functions float16_to_float64() and float64_to_float16(), which will be needed for the ARM A64 instruction set. Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- fpu/softfloat.c | 75 +++++++++++++++++++++++++++++++++++++++++ include/fpu/softfloat.h | 2 ++ 2 files changed, 77 insertions(+) diff --git a/fpu/softfloat.c b/fpu/softfloat.c index c63e011ae8..d4ca7cf7f1 100644 --- a/fpu/softfloat.c +++ b/fpu/softfloat.c @@ -3281,6 +3281,81 @@ float16 float32_to_float16(float32 a, flag ieee STATUS_PARAM) return roundAndPackFloat16(aSign, aExp, aSig, ieee STATUS_VAR); } +float64 float16_to_float64(float16 a, flag ieee STATUS_PARAM) +{ + flag aSign; + int_fast16_t aExp; + uint32_t aSig; + + aSign = extractFloat16Sign(a); + aExp = extractFloat16Exp(a); + aSig = extractFloat16Frac(a); + + if (aExp == 0x1f && ieee) { + if (aSig) { + return commonNaNToFloat64( + float16ToCommonNaN(a STATUS_VAR) STATUS_VAR); + } + return packFloat64(aSign, 0x7ff, 0); + } + if (aExp == 0) { + if (aSig == 0) { + return packFloat64(aSign, 0, 0); + } + + normalizeFloat16Subnormal(aSig, &aExp, &aSig); + aExp--; + } + return packFloat64(aSign, aExp + 0x3f0, ((uint64_t)aSig) << 42); +} + +float16 float64_to_float16(float64 a, flag ieee STATUS_PARAM) +{ + flag aSign; + int_fast16_t aExp; + uint64_t aSig; + uint32_t zSig; + + a = float64_squash_input_denormal(a STATUS_VAR); + + aSig = extractFloat64Frac(a); + aExp = extractFloat64Exp(a); + aSign = extractFloat64Sign(a); + if (aExp == 0x7FF) { + if (aSig) { + /* Input is a NaN */ + if (!ieee) { + float_raise(float_flag_invalid STATUS_VAR); + return packFloat16(aSign, 0, 0); + } + return commonNaNToFloat16( + float64ToCommonNaN(a STATUS_VAR) STATUS_VAR); + } + /* Infinity */ + if (!ieee) { + float_raise(float_flag_invalid STATUS_VAR); + return packFloat16(aSign, 0x1f, 0x3ff); + } + return packFloat16(aSign, 0x1f, 0); + } + shift64RightJamming(aSig, 29, &aSig); + zSig = aSig; + if (aExp == 0 && zSig == 0) { + return packFloat16(aSign, 0, 0); + } + /* Decimal point between bits 22 and 23. Note that we add the 1 bit + * even if the input is denormal; however this is harmless because + * the largest possible single-precision denormal is still smaller + * than the smallest representable half-precision denormal, and so we + * will end up ignoring aSig and returning via the "always return zero" + * codepath. + */ + zSig |= 0x00800000; + aExp -= 0x3F1; + + return roundAndPackFloat16(aSign, aExp, zSig, ieee STATUS_VAR); +} + /*---------------------------------------------------------------------------- | Returns the result of converting the double-precision floating-point value | `a' to the extended double-precision floating-point format. The conversion diff --git a/include/fpu/softfloat.h b/include/fpu/softfloat.h index a634a4ebff..83d324aca8 100644 --- a/include/fpu/softfloat.h +++ b/include/fpu/softfloat.h @@ -298,6 +298,8 @@ INLINE float64 uint16_to_float64(uint16_t v STATUS_PARAM) *----------------------------------------------------------------------------*/ float16 float32_to_float16( float32, flag STATUS_PARAM ); float32 float16_to_float32( float16, flag STATUS_PARAM ); +float16 float64_to_float16(float64 a, flag ieee STATUS_PARAM); +float64 float16_to_float64(float16 a, flag ieee STATUS_PARAM); /*---------------------------------------------------------------------------- | Software half-precision operations. From 01a0b83cdb1c29a0066cfee3445a34280b3ac03a Mon Sep 17 00:00:00 2001 From: Peter Maydell Date: Tue, 7 Jan 2014 17:19:12 +0000 Subject: [PATCH 52/61] softfloat: Refactor code handling various rounding modes Refactor the code in various functions which calculates rounding increments given the current rounding mode, so that instead of a set of nested if statements we have a simple switch statement. This will give us a clean place to add the case for the new tiesAway rounding mode. Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- fpu/softfloat.c | 405 ++++++++++++++++++++++++++++-------------------- 1 file changed, 241 insertions(+), 164 deletions(-) diff --git a/fpu/softfloat.c b/fpu/softfloat.c index d4ca7cf7f1..6c7a90a054 100644 --- a/fpu/softfloat.c +++ b/fpu/softfloat.c @@ -42,6 +42,9 @@ these four paragraphs for those parts of this code that are retained. #include "fpu/softfloat.h" +/* We only need stdlib for abort() */ +#include + /*---------------------------------------------------------------------------- | Primitive arithmetic functions, including multi-word arithmetic, and | division and square root approximations. (Can be specialized to target if @@ -106,20 +109,21 @@ static int32 roundAndPackInt32( flag zSign, uint64_t absZ STATUS_PARAM) roundingMode = STATUS(float_rounding_mode); roundNearestEven = ( roundingMode == float_round_nearest_even ); - roundIncrement = 0x40; - if ( ! roundNearestEven ) { - if ( roundingMode == float_round_to_zero ) { - roundIncrement = 0; - } - else { - roundIncrement = 0x7F; - if ( zSign ) { - if ( roundingMode == float_round_up ) roundIncrement = 0; - } - else { - if ( roundingMode == float_round_down ) roundIncrement = 0; - } - } + switch (roundingMode) { + case float_round_nearest_even: + roundIncrement = 0x40; + break; + case float_round_to_zero: + roundIncrement = 0; + break; + case float_round_up: + roundIncrement = zSign ? 0 : 0x7f; + break; + case float_round_down: + roundIncrement = zSign ? 0x7f : 0; + break; + default: + abort(); } roundBits = absZ & 0x7F; absZ = ( absZ + roundIncrement )>>7; @@ -155,19 +159,21 @@ static int64 roundAndPackInt64( flag zSign, uint64_t absZ0, uint64_t absZ1 STATU roundingMode = STATUS(float_rounding_mode); roundNearestEven = ( roundingMode == float_round_nearest_even ); - increment = ( (int64_t) absZ1 < 0 ); - if ( ! roundNearestEven ) { - if ( roundingMode == float_round_to_zero ) { - increment = 0; - } - else { - if ( zSign ) { - increment = ( roundingMode == float_round_down ) && absZ1; - } - else { - increment = ( roundingMode == float_round_up ) && absZ1; - } - } + switch (roundingMode) { + case float_round_nearest_even: + increment = ((int64_t) absZ1 < 0); + break; + case float_round_to_zero: + increment = 0; + break; + case float_round_up: + increment = !zSign && absZ1; + break; + case float_round_down: + increment = zSign && absZ1; + break; + default: + abort(); } if ( increment ) { ++absZ0; @@ -206,17 +212,21 @@ static int64 roundAndPackUint64(flag zSign, uint64_t absZ0, roundingMode = STATUS(float_rounding_mode); roundNearestEven = (roundingMode == float_round_nearest_even); - increment = ((int64_t)absZ1 < 0); - if (!roundNearestEven) { - if (roundingMode == float_round_to_zero) { - increment = 0; - } else if (absZ1) { - if (zSign) { - increment = (roundingMode == float_round_down) && absZ1; - } else { - increment = (roundingMode == float_round_up) && absZ1; - } - } + switch (roundingMode) { + case float_round_nearest_even: + increment = ((int64_t)absZ1 < 0); + break; + case float_round_to_zero: + increment = 0; + break; + case float_round_up: + increment = !zSign && absZ1; + break; + case float_round_down: + increment = zSign && absZ1; + break; + default: + abort(); } if (increment) { ++absZ0; @@ -354,20 +364,22 @@ static float32 roundAndPackFloat32(flag zSign, int_fast16_t zExp, uint32_t zSig roundingMode = STATUS(float_rounding_mode); roundNearestEven = ( roundingMode == float_round_nearest_even ); - roundIncrement = 0x40; - if ( ! roundNearestEven ) { - if ( roundingMode == float_round_to_zero ) { - roundIncrement = 0; - } - else { - roundIncrement = 0x7F; - if ( zSign ) { - if ( roundingMode == float_round_up ) roundIncrement = 0; - } - else { - if ( roundingMode == float_round_down ) roundIncrement = 0; - } - } + switch (roundingMode) { + case float_round_nearest_even: + roundIncrement = 0x40; + break; + case float_round_to_zero: + roundIncrement = 0; + break; + case float_round_up: + roundIncrement = zSign ? 0 : 0x7f; + break; + case float_round_down: + roundIncrement = zSign ? 0x7f : 0; + break; + default: + abort(); + break; } roundBits = zSig & 0x7F; if ( 0xFD <= (uint16_t) zExp ) { @@ -536,20 +548,21 @@ static float64 roundAndPackFloat64(flag zSign, int_fast16_t zExp, uint64_t zSig roundingMode = STATUS(float_rounding_mode); roundNearestEven = ( roundingMode == float_round_nearest_even ); - roundIncrement = 0x200; - if ( ! roundNearestEven ) { - if ( roundingMode == float_round_to_zero ) { - roundIncrement = 0; - } - else { - roundIncrement = 0x3FF; - if ( zSign ) { - if ( roundingMode == float_round_up ) roundIncrement = 0; - } - else { - if ( roundingMode == float_round_down ) roundIncrement = 0; - } - } + switch (roundingMode) { + case float_round_nearest_even: + roundIncrement = 0x200; + break; + case float_round_to_zero: + roundIncrement = 0; + break; + case float_round_up: + roundIncrement = zSign ? 0 : 0x3ff; + break; + case float_round_down: + roundIncrement = zSign ? 0x3ff : 0; + break; + default: + abort(); } roundBits = zSig & 0x3FF; if ( 0x7FD <= (uint16_t) zExp ) { @@ -719,19 +732,20 @@ static floatx80 goto precision80; } zSig0 |= ( zSig1 != 0 ); - if ( ! roundNearestEven ) { - if ( roundingMode == float_round_to_zero ) { - roundIncrement = 0; - } - else { - roundIncrement = roundMask; - if ( zSign ) { - if ( roundingMode == float_round_up ) roundIncrement = 0; - } - else { - if ( roundingMode == float_round_down ) roundIncrement = 0; - } - } + switch (roundingMode) { + case float_round_nearest_even: + break; + case float_round_to_zero: + roundIncrement = 0; + break; + case float_round_up: + roundIncrement = zSign ? 0 : roundMask; + break; + case float_round_down: + roundIncrement = zSign ? roundMask : 0; + break; + default: + abort(); } roundBits = zSig0 & roundMask; if ( 0x7FFD <= (uint32_t) ( zExp - 1 ) ) { @@ -778,19 +792,21 @@ static floatx80 if ( zSig0 == 0 ) zExp = 0; return packFloatx80( zSign, zExp, zSig0 ); precision80: - increment = ( (int64_t) zSig1 < 0 ); - if ( ! roundNearestEven ) { - if ( roundingMode == float_round_to_zero ) { - increment = 0; - } - else { - if ( zSign ) { - increment = ( roundingMode == float_round_down ) && zSig1; - } - else { - increment = ( roundingMode == float_round_up ) && zSig1; - } - } + switch (roundingMode) { + case float_round_nearest_even: + increment = ((int64_t)zSig1 < 0); + break; + case float_round_to_zero: + increment = 0; + break; + case float_round_up: + increment = !zSign && zSig1; + break; + case float_round_down: + increment = zSign && zSig1; + break; + default: + abort(); } if ( 0x7FFD <= (uint32_t) ( zExp - 1 ) ) { if ( ( 0x7FFE < zExp ) @@ -820,16 +836,21 @@ static floatx80 zExp = 0; if ( isTiny && zSig1 ) float_raise( float_flag_underflow STATUS_VAR); if ( zSig1 ) STATUS(float_exception_flags) |= float_flag_inexact; - if ( roundNearestEven ) { - increment = ( (int64_t) zSig1 < 0 ); - } - else { - if ( zSign ) { - increment = ( roundingMode == float_round_down ) && zSig1; - } - else { - increment = ( roundingMode == float_round_up ) && zSig1; - } + switch (roundingMode) { + case float_round_nearest_even: + increment = ((int64_t)zSig1 < 0); + break; + case float_round_to_zero: + increment = 0; + break; + case float_round_up: + increment = !zSign && zSig1; + break; + case float_round_down: + increment = zSign && zSig1; + break; + default: + abort(); } if ( increment ) { ++zSig0; @@ -1029,19 +1050,21 @@ static float128 roundingMode = STATUS(float_rounding_mode); roundNearestEven = ( roundingMode == float_round_nearest_even ); - increment = ( (int64_t) zSig2 < 0 ); - if ( ! roundNearestEven ) { - if ( roundingMode == float_round_to_zero ) { - increment = 0; - } - else { - if ( zSign ) { - increment = ( roundingMode == float_round_down ) && zSig2; - } - else { - increment = ( roundingMode == float_round_up ) && zSig2; - } - } + switch (roundingMode) { + case float_round_nearest_even: + increment = ((int64_t)zSig2 < 0); + break; + case float_round_to_zero: + increment = 0; + break; + case float_round_up: + increment = !zSign && zSig2; + break; + case float_round_down: + increment = zSign && zSig2; + break; + default: + abort(); } if ( 0x7FFD <= (uint32_t) zExp ) { if ( ( 0x7FFD < zExp ) @@ -1089,16 +1112,21 @@ static float128 zSig0, zSig1, zSig2, - zExp, &zSig0, &zSig1, &zSig2 ); zExp = 0; if ( isTiny && zSig2 ) float_raise( float_flag_underflow STATUS_VAR); - if ( roundNearestEven ) { - increment = ( (int64_t) zSig2 < 0 ); - } - else { - if ( zSign ) { - increment = ( roundingMode == float_round_down ) && zSig2; - } - else { - increment = ( roundingMode == float_round_up ) && zSig2; - } + switch (roundingMode) { + case float_round_nearest_even: + increment = ((int64_t)zSig2 < 0); + break; + case float_round_to_zero: + increment = 0; + break; + case float_round_up: + increment = !zSign && zSig2; + break; + case float_round_down: + increment = zSign && zSig2; + break; + default: + abort(); } } } @@ -1737,7 +1765,6 @@ float32 float32_round_to_int( float32 a STATUS_PARAM) flag aSign; int_fast16_t aExp; uint32_t lastBitMask, roundBitsMask; - int8 roundingMode; uint32_t z; a = float32_squash_input_denormal(a STATUS_VAR); @@ -1769,15 +1796,27 @@ float32 float32_round_to_int( float32 a STATUS_PARAM) lastBitMask <<= 0x96 - aExp; roundBitsMask = lastBitMask - 1; z = float32_val(a); - roundingMode = STATUS(float_rounding_mode); - if ( roundingMode == float_round_nearest_even ) { + switch (STATUS(float_rounding_mode)) { + case float_round_nearest_even: z += lastBitMask>>1; - if ( ( z & roundBitsMask ) == 0 ) z &= ~ lastBitMask; - } - else if ( roundingMode != float_round_to_zero ) { - if ( extractFloat32Sign( make_float32(z) ) ^ ( roundingMode == float_round_up ) ) { + if ((z & roundBitsMask) == 0) { + z &= ~lastBitMask; + } + break; + case float_round_to_zero: + break; + case float_round_up: + if (!extractFloat32Sign(make_float32(z))) { z += roundBitsMask; } + break; + case float_round_down: + if (extractFloat32Sign(make_float32(z))) { + z += roundBitsMask; + } + break; + default: + abort(); } z &= ~ roundBitsMask; if ( z != float32_val(a) ) STATUS(float_exception_flags) |= float_flag_inexact; @@ -3120,7 +3159,6 @@ static float32 roundAndPackFloat16(flag zSign, int_fast16_t zExp, int maxexp = ieee ? 29 : 30; uint32_t mask; uint32_t increment; - int8 roundingMode; bool rounding_bumps_exp; bool is_tiny = false; @@ -3138,8 +3176,7 @@ static float32 roundAndPackFloat16(flag zSign, int_fast16_t zExp, mask = 0x00001fff; } - roundingMode = STATUS(float_rounding_mode); - switch (roundingMode) { + switch (STATUS(float_rounding_mode)) { case float_round_nearest_even: increment = (mask + 1) >> 1; if ((zSig & mask) == increment) { @@ -3430,7 +3467,6 @@ float64 float64_round_to_int( float64 a STATUS_PARAM ) flag aSign; int_fast16_t aExp; uint64_t lastBitMask, roundBitsMask; - int8 roundingMode; uint64_t z; a = float64_squash_input_denormal(a STATUS_VAR); @@ -3463,15 +3499,27 @@ float64 float64_round_to_int( float64 a STATUS_PARAM ) lastBitMask <<= 0x433 - aExp; roundBitsMask = lastBitMask - 1; z = float64_val(a); - roundingMode = STATUS(float_rounding_mode); - if ( roundingMode == float_round_nearest_even ) { - z += lastBitMask>>1; - if ( ( z & roundBitsMask ) == 0 ) z &= ~ lastBitMask; - } - else if ( roundingMode != float_round_to_zero ) { - if ( extractFloat64Sign( make_float64(z) ) ^ ( roundingMode == float_round_up ) ) { + switch (STATUS(float_rounding_mode)) { + case float_round_nearest_even: + z += lastBitMask >> 1; + if ((z & roundBitsMask) == 0) { + z &= ~lastBitMask; + } + break; + case float_round_to_zero: + break; + case float_round_up: + if (!extractFloat64Sign(make_float64(z))) { z += roundBitsMask; } + break; + case float_round_down: + if (extractFloat64Sign(make_float64(z))) { + z += roundBitsMask; + } + break; + default: + abort(); } z &= ~ roundBitsMask; if ( z != float64_val(a) ) @@ -4699,7 +4747,6 @@ floatx80 floatx80_round_to_int( floatx80 a STATUS_PARAM ) flag aSign; int32 aExp; uint64_t lastBitMask, roundBitsMask; - int8 roundingMode; floatx80 z; aExp = extractFloatx80Exp( a ); @@ -4740,15 +4787,27 @@ floatx80 floatx80_round_to_int( floatx80 a STATUS_PARAM ) lastBitMask <<= 0x403E - aExp; roundBitsMask = lastBitMask - 1; z = a; - roundingMode = STATUS(float_rounding_mode); - if ( roundingMode == float_round_nearest_even ) { + switch (STATUS(float_rounding_mode)) { + case float_round_nearest_even: z.low += lastBitMask>>1; - if ( ( z.low & roundBitsMask ) == 0 ) z.low &= ~ lastBitMask; - } - else if ( roundingMode != float_round_to_zero ) { - if ( extractFloatx80Sign( z ) ^ ( roundingMode == float_round_up ) ) { + if ((z.low & roundBitsMask) == 0) { + z.low &= ~lastBitMask; + } + break; + case float_round_to_zero: + break; + case float_round_up: + if (!extractFloatx80Sign(z)) { z.low += roundBitsMask; } + break; + case float_round_down: + if (extractFloatx80Sign(z)) { + z.low += roundBitsMask; + } + break; + default: + abort(); } z.low &= ~ roundBitsMask; if ( z.low == 0 ) { @@ -5774,7 +5833,6 @@ float128 float128_round_to_int( float128 a STATUS_PARAM ) flag aSign; int32 aExp; uint64_t lastBitMask, roundBitsMask; - int8 roundingMode; float128 z; aExp = extractFloat128Exp( a ); @@ -5791,8 +5849,8 @@ float128 float128_round_to_int( float128 a STATUS_PARAM ) lastBitMask = ( lastBitMask<<( 0x406E - aExp ) )<<1; roundBitsMask = lastBitMask - 1; z = a; - roundingMode = STATUS(float_rounding_mode); - if ( roundingMode == float_round_nearest_even ) { + switch (STATUS(float_rounding_mode)) { + case float_round_nearest_even: if ( lastBitMask ) { add128( z.high, z.low, 0, lastBitMask>>1, &z.high, &z.low ); if ( ( z.low & roundBitsMask ) == 0 ) z.low &= ~ lastBitMask; @@ -5803,12 +5861,21 @@ float128 float128_round_to_int( float128 a STATUS_PARAM ) if ( (uint64_t) ( z.low<<1 ) == 0 ) z.high &= ~1; } } - } - else if ( roundingMode != float_round_to_zero ) { - if ( extractFloat128Sign( z ) - ^ ( roundingMode == float_round_up ) ) { - add128( z.high, z.low, 0, roundBitsMask, &z.high, &z.low ); + break; + case float_round_to_zero: + break; + case float_round_up: + if (!extractFloat128Sign(z)) { + add128(z.high, z.low, 0, roundBitsMask, &z.high, &z.low); } + break; + case float_round_down: + if (extractFloat128Sign(z)) { + add128(z.high, z.low, 0, roundBitsMask, &z.high, &z.low); + } + break; + default: + abort(); } z.low &= ~ roundBitsMask; } @@ -5842,19 +5909,29 @@ float128 float128_round_to_int( float128 a STATUS_PARAM ) roundBitsMask = lastBitMask - 1; z.low = 0; z.high = a.high; - roundingMode = STATUS(float_rounding_mode); - if ( roundingMode == float_round_nearest_even ) { + switch (STATUS(float_rounding_mode)) { + case float_round_nearest_even: z.high += lastBitMask>>1; if ( ( ( z.high & roundBitsMask ) | a.low ) == 0 ) { z.high &= ~ lastBitMask; } - } - else if ( roundingMode != float_round_to_zero ) { - if ( extractFloat128Sign( z ) - ^ ( roundingMode == float_round_up ) ) { + break; + case float_round_to_zero: + break; + case float_round_up: + if (!extractFloat128Sign(z)) { z.high |= ( a.low != 0 ); z.high += roundBitsMask; } + break; + case float_round_down: + if (extractFloat128Sign(z)) { + z.high |= (a.low != 0); + z.high += roundBitsMask; + } + break; + default: + abort(); } z.high &= ~ roundBitsMask; } From dcc69632089fece07a0bb4fd859b010659569fda Mon Sep 17 00:00:00 2001 From: Peter Maydell Date: Tue, 7 Jan 2014 17:19:12 +0000 Subject: [PATCH 53/61] softfloat: Add support for ties-away rounding IEEE754-2008 specifies a new rounding mode: "roundTiesToAway: the floating-point number nearest to the infinitely precise result shall be delivered; if the two nearest floating-point numbers bracketing an unrepresentable infinitely precise result are equally near, the one with larger magnitude shall be delivered." Implement this new mode (it is needed for ARM). The general principle is that the required code is exactly like the ties-to-even code, except that we do not need to do the "in case of exact tie clear LSB to round-to-even", because the rounding operation naturally causes the exact tie to round up in magnitude. Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- fpu/softfloat.c | 54 +++++++++++++++++++++++++++++++++++++++++ include/fpu/softfloat.h | 3 ++- 2 files changed, 56 insertions(+), 1 deletion(-) diff --git a/fpu/softfloat.c b/fpu/softfloat.c index 6c7a90a054..e0ea599769 100644 --- a/fpu/softfloat.c +++ b/fpu/softfloat.c @@ -111,6 +111,7 @@ static int32 roundAndPackInt32( flag zSign, uint64_t absZ STATUS_PARAM) roundNearestEven = ( roundingMode == float_round_nearest_even ); switch (roundingMode) { case float_round_nearest_even: + case float_round_ties_away: roundIncrement = 0x40; break; case float_round_to_zero: @@ -161,6 +162,7 @@ static int64 roundAndPackInt64( flag zSign, uint64_t absZ0, uint64_t absZ1 STATU roundNearestEven = ( roundingMode == float_round_nearest_even ); switch (roundingMode) { case float_round_nearest_even: + case float_round_ties_away: increment = ((int64_t) absZ1 < 0); break; case float_round_to_zero: @@ -214,6 +216,7 @@ static int64 roundAndPackUint64(flag zSign, uint64_t absZ0, roundNearestEven = (roundingMode == float_round_nearest_even); switch (roundingMode) { case float_round_nearest_even: + case float_round_ties_away: increment = ((int64_t)absZ1 < 0); break; case float_round_to_zero: @@ -366,6 +369,7 @@ static float32 roundAndPackFloat32(flag zSign, int_fast16_t zExp, uint32_t zSig roundNearestEven = ( roundingMode == float_round_nearest_even ); switch (roundingMode) { case float_round_nearest_even: + case float_round_ties_away: roundIncrement = 0x40; break; case float_round_to_zero: @@ -550,6 +554,7 @@ static float64 roundAndPackFloat64(flag zSign, int_fast16_t zExp, uint64_t zSig roundNearestEven = ( roundingMode == float_round_nearest_even ); switch (roundingMode) { case float_round_nearest_even: + case float_round_ties_away: roundIncrement = 0x200; break; case float_round_to_zero: @@ -734,6 +739,7 @@ static floatx80 zSig0 |= ( zSig1 != 0 ); switch (roundingMode) { case float_round_nearest_even: + case float_round_ties_away: break; case float_round_to_zero: roundIncrement = 0; @@ -794,6 +800,7 @@ static floatx80 precision80: switch (roundingMode) { case float_round_nearest_even: + case float_round_ties_away: increment = ((int64_t)zSig1 < 0); break; case float_round_to_zero: @@ -838,6 +845,7 @@ static floatx80 if ( zSig1 ) STATUS(float_exception_flags) |= float_flag_inexact; switch (roundingMode) { case float_round_nearest_even: + case float_round_ties_away: increment = ((int64_t)zSig1 < 0); break; case float_round_to_zero: @@ -1052,6 +1060,7 @@ static float128 roundNearestEven = ( roundingMode == float_round_nearest_even ); switch (roundingMode) { case float_round_nearest_even: + case float_round_ties_away: increment = ((int64_t)zSig2 < 0); break; case float_round_to_zero: @@ -1114,6 +1123,7 @@ static float128 if ( isTiny && zSig2 ) float_raise( float_flag_underflow STATUS_VAR); switch (roundingMode) { case float_round_nearest_even: + case float_round_ties_away: increment = ((int64_t)zSig2 < 0); break; case float_round_to_zero: @@ -1785,6 +1795,11 @@ float32 float32_round_to_int( float32 a STATUS_PARAM) return packFloat32( aSign, 0x7F, 0 ); } break; + case float_round_ties_away: + if (aExp == 0x7E) { + return packFloat32(aSign, 0x7F, 0); + } + break; case float_round_down: return make_float32(aSign ? 0xBF800000 : 0); case float_round_up: @@ -1803,6 +1818,9 @@ float32 float32_round_to_int( float32 a STATUS_PARAM) z &= ~lastBitMask; } break; + case float_round_ties_away: + z += lastBitMask >> 1; + break; case float_round_to_zero: break; case float_round_up: @@ -3183,6 +3201,9 @@ static float32 roundAndPackFloat16(flag zSign, int_fast16_t zExp, increment = zSig & (increment << 1); } break; + case float_round_ties_away: + increment = (mask + 1) >> 1; + break; case float_round_up: increment = zSign ? 0 : mask; break; @@ -3487,6 +3508,11 @@ float64 float64_round_to_int( float64 a STATUS_PARAM ) return packFloat64( aSign, 0x3FF, 0 ); } break; + case float_round_ties_away: + if (aExp == 0x3FE) { + return packFloat64(aSign, 0x3ff, 0); + } + break; case float_round_down: return make_float64(aSign ? LIT64( 0xBFF0000000000000 ) : 0); case float_round_up: @@ -3506,6 +3532,9 @@ float64 float64_round_to_int( float64 a STATUS_PARAM ) z &= ~lastBitMask; } break; + case float_round_ties_away: + z += lastBitMask >> 1; + break; case float_round_to_zero: break; case float_round_up: @@ -4771,6 +4800,11 @@ floatx80 floatx80_round_to_int( floatx80 a STATUS_PARAM ) packFloatx80( aSign, 0x3FFF, LIT64( 0x8000000000000000 ) ); } break; + case float_round_ties_away: + if (aExp == 0x3FFE) { + return packFloatx80(aSign, 0x3FFF, LIT64(0x8000000000000000)); + } + break; case float_round_down: return aSign ? @@ -4794,6 +4828,9 @@ floatx80 floatx80_round_to_int( floatx80 a STATUS_PARAM ) z.low &= ~lastBitMask; } break; + case float_round_ties_away: + z.low += lastBitMask >> 1; + break; case float_round_to_zero: break; case float_round_up: @@ -5862,6 +5899,15 @@ float128 float128_round_to_int( float128 a STATUS_PARAM ) } } break; + case float_round_ties_away: + if (lastBitMask) { + add128(z.high, z.low, 0, lastBitMask >> 1, &z.high, &z.low); + } else { + if ((int64_t) z.low < 0) { + ++z.high; + } + } + break; case float_round_to_zero: break; case float_round_up: @@ -5893,6 +5939,11 @@ float128 float128_round_to_int( float128 a STATUS_PARAM ) return packFloat128( aSign, 0x3FFF, 0, 0 ); } break; + case float_round_ties_away: + if (aExp == 0x3FFE) { + return packFloat128(aSign, 0x3FFF, 0, 0); + } + break; case float_round_down: return aSign ? packFloat128( 1, 0x3FFF, 0, 0 ) @@ -5916,6 +5967,9 @@ float128 float128_round_to_int( float128 a STATUS_PARAM ) z.high &= ~ lastBitMask; } break; + case float_round_ties_away: + z.high += lastBitMask>>1; + break; case float_round_to_zero: break; case float_round_up: diff --git a/include/fpu/softfloat.h b/include/fpu/softfloat.h index 83d324aca8..806ae13780 100644 --- a/include/fpu/softfloat.h +++ b/include/fpu/softfloat.h @@ -152,7 +152,8 @@ enum { float_round_nearest_even = 0, float_round_down = 1, float_round_up = 2, - float_round_to_zero = 3 + float_round_to_zero = 3, + float_round_ties_away = 4, }; /*---------------------------------------------------------------------------- From 59af2dc86866dc14a42eec596f3d100dd3a24d1e Mon Sep 17 00:00:00 2001 From: Will Newton Date: Tue, 7 Jan 2014 17:19:13 +0000 Subject: [PATCH 54/61] target-arm: Prepare VFP_CONV_FIX helpers for A64 uses Make the VFP_CONV_FIX helpers a little more flexible in preparation for the A64 uses. This requires two changes: * use the correct softfloat conversion function based on itype rather than always the int32 one; this is possible now that softfloat provides int16 versions and necessary for the future conversion-to-int64 A64 variants. This also allows us to drop the awkward 'sign' macro argument. * split the 'fsz' argument which currently controls both width of the input float type and width of the output integer type into two; this will allow us to specify the A64 64-bit-int-to-single conversion function, where the two widths are different. We can also drop the (itype##_t) cast now that softfloat guarantees that all the itype##_to_float* functions take an integer argument of exactly the correct type. Signed-off-by: Will Newton Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- target-arm/helper.c | 28 ++++++++++++++-------------- 1 file changed, 14 insertions(+), 14 deletions(-) diff --git a/target-arm/helper.c b/target-arm/helper.c index 6f629f36b0..5579565d19 100644 --- a/target-arm/helper.c +++ b/target-arm/helper.c @@ -3976,17 +3976,17 @@ float32 VFP_HELPER(fcvts, d)(float64 x, CPUARMState *env) } /* VFP3 fixed point conversion. */ -#define VFP_CONV_FIX(name, p, fsz, itype, sign) \ -float##fsz HELPER(vfp_##name##to##p)(uint##fsz##_t x, uint32_t shift, \ - void *fpstp) \ +#define VFP_CONV_FIX(name, p, fsz, isz, itype) \ +float##fsz HELPER(vfp_##name##to##p)(uint##isz##_t x, uint32_t shift, \ + void *fpstp) \ { \ float_status *fpst = fpstp; \ float##fsz tmp; \ - tmp = sign##int32_to_##float##fsz((itype##_t)x, fpst); \ + tmp = itype##_to_##float##fsz(x, fpst); \ return float##fsz##_scalbn(tmp, -(int)shift, fpst); \ } \ -uint##fsz##_t HELPER(vfp_to##name##p)(float##fsz x, uint32_t shift, \ - void *fpstp) \ +uint##isz##_t HELPER(vfp_to##name##p)(float##fsz x, uint32_t shift, \ + void *fpstp) \ { \ float_status *fpst = fpstp; \ float##fsz tmp; \ @@ -3998,14 +3998,14 @@ uint##fsz##_t HELPER(vfp_to##name##p)(float##fsz x, uint32_t shift, \ return float##fsz##_to_##itype##_round_to_zero(tmp, fpst); \ } -VFP_CONV_FIX(sh, d, 64, int16, ) -VFP_CONV_FIX(sl, d, 64, int32, ) -VFP_CONV_FIX(uh, d, 64, uint16, u) -VFP_CONV_FIX(ul, d, 64, uint32, u) -VFP_CONV_FIX(sh, s, 32, int16, ) -VFP_CONV_FIX(sl, s, 32, int32, ) -VFP_CONV_FIX(uh, s, 32, uint16, u) -VFP_CONV_FIX(ul, s, 32, uint32, u) +VFP_CONV_FIX(sh, d, 64, 64, int16) +VFP_CONV_FIX(sl, d, 64, 64, int32) +VFP_CONV_FIX(uh, d, 64, 64, uint16) +VFP_CONV_FIX(ul, d, 64, 64, uint32) +VFP_CONV_FIX(sh, s, 32, 32, int16) +VFP_CONV_FIX(sl, s, 32, 32, int32) +VFP_CONV_FIX(uh, s, 32, 32, uint16) +VFP_CONV_FIX(ul, s, 32, 32, uint32) #undef VFP_CONV_FIX /* Half precision conversions. */ From 6ab748af130912b92decfb1fa411ebdfd7cf8d87 Mon Sep 17 00:00:00 2001 From: Will Newton Date: Tue, 7 Jan 2014 17:19:13 +0000 Subject: [PATCH 55/61] target-arm: Rename A32 VFP conversion helpers The VFP conversion helpers for A32 round to zero as this is the only rounding mode supported. Rename these helpers to make it clear that they round to zero and are not suitable for use in the AArch64 code. Signed-off-by: Will Newton Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- target-arm/helper.c | 19 ++++++++++++++----- target-arm/helper.h | 16 ++++++++-------- target-arm/translate.c | 24 +++++++++++++----------- 3 files changed, 35 insertions(+), 24 deletions(-) diff --git a/target-arm/helper.c b/target-arm/helper.c index 5579565d19..db4f516cb6 100644 --- a/target-arm/helper.c +++ b/target-arm/helper.c @@ -3976,7 +3976,7 @@ float32 VFP_HELPER(fcvts, d)(float64 x, CPUARMState *env) } /* VFP3 fixed point conversion. */ -#define VFP_CONV_FIX(name, p, fsz, isz, itype) \ +#define VFP_CONV_FIX_FLOAT(name, p, fsz, isz, itype) \ float##fsz HELPER(vfp_##name##to##p)(uint##isz##_t x, uint32_t shift, \ void *fpstp) \ { \ @@ -3984,9 +3984,12 @@ float##fsz HELPER(vfp_##name##to##p)(uint##isz##_t x, uint32_t shift, \ float##fsz tmp; \ tmp = itype##_to_##float##fsz(x, fpst); \ return float##fsz##_scalbn(tmp, -(int)shift, fpst); \ -} \ -uint##isz##_t HELPER(vfp_to##name##p)(float##fsz x, uint32_t shift, \ - void *fpstp) \ +} + +#define VFP_CONV_FLOAT_FIX_ROUND(name, p, fsz, isz, itype, round) \ +uint##isz##_t HELPER(vfp_to##name##p##round)(float##fsz x, \ + uint32_t shift, \ + void *fpstp) \ { \ float_status *fpst = fpstp; \ float##fsz tmp; \ @@ -3995,9 +3998,13 @@ uint##isz##_t HELPER(vfp_to##name##p)(float##fsz x, uint32_t shift, \ return 0; \ } \ tmp = float##fsz##_scalbn(x, shift, fpst); \ - return float##fsz##_to_##itype##_round_to_zero(tmp, fpst); \ + return float##fsz##_to_##itype##round(tmp, fpst); \ } +#define VFP_CONV_FIX(name, p, fsz, isz, itype) \ +VFP_CONV_FIX_FLOAT(name, p, fsz, isz, itype) \ +VFP_CONV_FLOAT_FIX_ROUND(name, p, fsz, isz, itype, _round_to_zero) + VFP_CONV_FIX(sh, d, 64, 64, int16) VFP_CONV_FIX(sl, d, 64, 64, int32) VFP_CONV_FIX(uh, d, 64, 64, uint16) @@ -4007,6 +4014,8 @@ VFP_CONV_FIX(sl, s, 32, 32, int32) VFP_CONV_FIX(uh, s, 32, 32, uint16) VFP_CONV_FIX(ul, s, 32, 32, uint32) #undef VFP_CONV_FIX +#undef VFP_CONV_FIX_FLOAT +#undef VFP_CONV_FLOAT_FIX_ROUND /* Half precision conversions. */ static float32 do_fcvt_f16_to_f32(uint32_t a, CPUARMState *env, float_status *s) diff --git a/target-arm/helper.h b/target-arm/helper.h index dd1160ed61..b785623e34 100644 --- a/target-arm/helper.h +++ b/target-arm/helper.h @@ -115,14 +115,14 @@ DEF_HELPER_2(vfp_tosid, i32, f64, ptr) DEF_HELPER_2(vfp_tosizs, i32, f32, ptr) DEF_HELPER_2(vfp_tosizd, i32, f64, ptr) -DEF_HELPER_3(vfp_toshs, i32, f32, i32, ptr) -DEF_HELPER_3(vfp_tosls, i32, f32, i32, ptr) -DEF_HELPER_3(vfp_touhs, i32, f32, i32, ptr) -DEF_HELPER_3(vfp_touls, i32, f32, i32, ptr) -DEF_HELPER_3(vfp_toshd, i64, f64, i32, ptr) -DEF_HELPER_3(vfp_tosld, i64, f64, i32, ptr) -DEF_HELPER_3(vfp_touhd, i64, f64, i32, ptr) -DEF_HELPER_3(vfp_tould, i64, f64, i32, ptr) +DEF_HELPER_3(vfp_toshs_round_to_zero, i32, f32, i32, ptr) +DEF_HELPER_3(vfp_tosls_round_to_zero, i32, f32, i32, ptr) +DEF_HELPER_3(vfp_touhs_round_to_zero, i32, f32, i32, ptr) +DEF_HELPER_3(vfp_touls_round_to_zero, i32, f32, i32, ptr) +DEF_HELPER_3(vfp_toshd_round_to_zero, i64, f64, i32, ptr) +DEF_HELPER_3(vfp_tosld_round_to_zero, i64, f64, i32, ptr) +DEF_HELPER_3(vfp_touhd_round_to_zero, i64, f64, i32, ptr) +DEF_HELPER_3(vfp_tould_round_to_zero, i64, f64, i32, ptr) DEF_HELPER_3(vfp_shtos, f32, i32, i32, ptr) DEF_HELPER_3(vfp_sltos, f32, i32, i32, ptr) DEF_HELPER_3(vfp_uhtos, f32, i32, i32, ptr) diff --git a/target-arm/translate.c b/target-arm/translate.c index d04fc9ff41..8d240e160d 100644 --- a/target-arm/translate.c +++ b/target-arm/translate.c @@ -1098,27 +1098,29 @@ VFP_GEN_FTOI(tosi) VFP_GEN_FTOI(tosiz) #undef VFP_GEN_FTOI -#define VFP_GEN_FIX(name) \ +#define VFP_GEN_FIX(name, round) \ static inline void gen_vfp_##name(int dp, int shift, int neon) \ { \ TCGv_i32 tmp_shift = tcg_const_i32(shift); \ TCGv_ptr statusptr = get_fpstatus_ptr(neon); \ if (dp) { \ - gen_helper_vfp_##name##d(cpu_F0d, cpu_F0d, tmp_shift, statusptr); \ + gen_helper_vfp_##name##d##round(cpu_F0d, cpu_F0d, tmp_shift, \ + statusptr); \ } else { \ - gen_helper_vfp_##name##s(cpu_F0s, cpu_F0s, tmp_shift, statusptr); \ + gen_helper_vfp_##name##s##round(cpu_F0s, cpu_F0s, tmp_shift, \ + statusptr); \ } \ tcg_temp_free_i32(tmp_shift); \ tcg_temp_free_ptr(statusptr); \ } -VFP_GEN_FIX(tosh) -VFP_GEN_FIX(tosl) -VFP_GEN_FIX(touh) -VFP_GEN_FIX(toul) -VFP_GEN_FIX(shto) -VFP_GEN_FIX(slto) -VFP_GEN_FIX(uhto) -VFP_GEN_FIX(ulto) +VFP_GEN_FIX(tosh, _round_to_zero) +VFP_GEN_FIX(tosl, _round_to_zero) +VFP_GEN_FIX(touh, _round_to_zero) +VFP_GEN_FIX(toul, _round_to_zero) +VFP_GEN_FIX(shto, ) +VFP_GEN_FIX(slto, ) +VFP_GEN_FIX(uhto, ) +VFP_GEN_FIX(ulto, ) #undef VFP_GEN_FIX static inline void gen_vfp_ld(DisasContext *s, int dp, TCGv_i32 addr) From 39905e53db75b6df0cfdb65644056a9f1cd10069 Mon Sep 17 00:00:00 2001 From: Peter Maydell Date: Tue, 7 Jan 2014 17:19:13 +0000 Subject: [PATCH 56/61] target-arm: Ignore most exceptions from scalbn when doing fixpoint conversion The VFP fixed point conversion helpers first call float_scalbn and then convert the result to an integer. This scalbn operation may set floating point exception flags for: * overflow & inexact (if it overflows to infinity) * input denormal squashed to zero * output denormal squashed to zero Of these, we only care about the input-denormal flag, since the output of the whole scale-and-convert operation will be an integer (so squashed-output-denormal and overflow don't apply). Suppress the others by saving the pre-scalb exception flags and only copying across a potential input-denormal flag. Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- target-arm/helper.c | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/target-arm/helper.c b/target-arm/helper.c index db4f516cb6..c995562d87 100644 --- a/target-arm/helper.c +++ b/target-arm/helper.c @@ -3986,18 +3986,27 @@ float##fsz HELPER(vfp_##name##to##p)(uint##isz##_t x, uint32_t shift, \ return float##fsz##_scalbn(tmp, -(int)shift, fpst); \ } +/* Notice that we want only input-denormal exception flags from the + * scalbn operation: the other possible flags (overflow+inexact if + * we overflow to infinity, output-denormal) aren't correct for the + * complete scale-and-convert operation. + */ #define VFP_CONV_FLOAT_FIX_ROUND(name, p, fsz, isz, itype, round) \ uint##isz##_t HELPER(vfp_to##name##p##round)(float##fsz x, \ uint32_t shift, \ void *fpstp) \ { \ float_status *fpst = fpstp; \ + int old_exc_flags = get_float_exception_flags(fpst); \ float##fsz tmp; \ if (float##fsz##_is_any_nan(x)) { \ float_raise(float_flag_invalid, fpst); \ return 0; \ } \ tmp = float##fsz##_scalbn(x, shift, fpst); \ + old_exc_flags |= get_float_exception_flags(fpst) \ + & float_flag_input_denormal; \ + set_float_exception_flags(old_exc_flags, fpst); \ return float##fsz##_to_##itype##round(tmp, fpst); \ } From 8fe2a36f33f44e9d1333c851762ac463ad528233 Mon Sep 17 00:00:00 2001 From: Will Newton Date: Tue, 7 Jan 2014 17:19:14 +0000 Subject: [PATCH 57/61] target-arm: A64: Add extra VFP fixed point conversion helpers Define the full set of floating point to fixed point conversion helpers required to support AArch64. Signed-off-by: Will Newton Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- target-arm/helper.c | 11 ++++++++++- target-arm/helper.h | 16 ++++++++++++++++ 2 files changed, 26 insertions(+), 1 deletion(-) diff --git a/target-arm/helper.c b/target-arm/helper.c index c995562d87..dcf6aa4712 100644 --- a/target-arm/helper.c +++ b/target-arm/helper.c @@ -4012,16 +4012,25 @@ uint##isz##_t HELPER(vfp_to##name##p##round)(float##fsz x, \ #define VFP_CONV_FIX(name, p, fsz, isz, itype) \ VFP_CONV_FIX_FLOAT(name, p, fsz, isz, itype) \ -VFP_CONV_FLOAT_FIX_ROUND(name, p, fsz, isz, itype, _round_to_zero) +VFP_CONV_FLOAT_FIX_ROUND(name, p, fsz, isz, itype, _round_to_zero) \ +VFP_CONV_FLOAT_FIX_ROUND(name, p, fsz, isz, itype, ) + +#define VFP_CONV_FIX_A64(name, p, fsz, isz, itype) \ +VFP_CONV_FIX_FLOAT(name, p, fsz, isz, itype) \ +VFP_CONV_FLOAT_FIX_ROUND(name, p, fsz, isz, itype, ) VFP_CONV_FIX(sh, d, 64, 64, int16) VFP_CONV_FIX(sl, d, 64, 64, int32) +VFP_CONV_FIX_A64(sq, d, 64, 64, int64) VFP_CONV_FIX(uh, d, 64, 64, uint16) VFP_CONV_FIX(ul, d, 64, 64, uint32) +VFP_CONV_FIX_A64(uq, d, 64, 64, uint64) VFP_CONV_FIX(sh, s, 32, 32, int16) VFP_CONV_FIX(sl, s, 32, 32, int32) +VFP_CONV_FIX_A64(sq, s, 32, 64, int64) VFP_CONV_FIX(uh, s, 32, 32, uint16) VFP_CONV_FIX(ul, s, 32, 32, uint32) +VFP_CONV_FIX_A64(uq, s, 32, 64, uint64) #undef VFP_CONV_FIX #undef VFP_CONV_FIX_FLOAT #undef VFP_CONV_FLOAT_FIX_ROUND diff --git a/target-arm/helper.h b/target-arm/helper.h index b785623e34..2e1af4671c 100644 --- a/target-arm/helper.h +++ b/target-arm/helper.h @@ -123,14 +123,30 @@ DEF_HELPER_3(vfp_toshd_round_to_zero, i64, f64, i32, ptr) DEF_HELPER_3(vfp_tosld_round_to_zero, i64, f64, i32, ptr) DEF_HELPER_3(vfp_touhd_round_to_zero, i64, f64, i32, ptr) DEF_HELPER_3(vfp_tould_round_to_zero, i64, f64, i32, ptr) +DEF_HELPER_3(vfp_toshs, i32, f32, i32, ptr) +DEF_HELPER_3(vfp_tosls, i32, f32, i32, ptr) +DEF_HELPER_3(vfp_tosqs, i64, f32, i32, ptr) +DEF_HELPER_3(vfp_touhs, i32, f32, i32, ptr) +DEF_HELPER_3(vfp_touls, i32, f32, i32, ptr) +DEF_HELPER_3(vfp_touqs, i64, f32, i32, ptr) +DEF_HELPER_3(vfp_toshd, i64, f64, i32, ptr) +DEF_HELPER_3(vfp_tosld, i64, f64, i32, ptr) +DEF_HELPER_3(vfp_tosqd, i64, f64, i32, ptr) +DEF_HELPER_3(vfp_touhd, i64, f64, i32, ptr) +DEF_HELPER_3(vfp_tould, i64, f64, i32, ptr) +DEF_HELPER_3(vfp_touqd, i64, f64, i32, ptr) DEF_HELPER_3(vfp_shtos, f32, i32, i32, ptr) DEF_HELPER_3(vfp_sltos, f32, i32, i32, ptr) +DEF_HELPER_3(vfp_sqtos, f32, i64, i32, ptr) DEF_HELPER_3(vfp_uhtos, f32, i32, i32, ptr) DEF_HELPER_3(vfp_ultos, f32, i32, i32, ptr) +DEF_HELPER_3(vfp_uqtos, f32, i64, i32, ptr) DEF_HELPER_3(vfp_shtod, f64, i64, i32, ptr) DEF_HELPER_3(vfp_sltod, f64, i64, i32, ptr) +DEF_HELPER_3(vfp_sqtod, f64, i64, i32, ptr) DEF_HELPER_3(vfp_uhtod, f64, i64, i32, ptr) DEF_HELPER_3(vfp_ultod, f64, i64, i32, ptr) +DEF_HELPER_3(vfp_uqtod, f64, i64, i32, ptr) DEF_HELPER_2(vfp_fcvt_f16_to_f32, f32, i32, env) DEF_HELPER_2(vfp_fcvt_f32_to_f16, i32, f32, env) From 3617f86ba87a0f54b74dbaed0d92da13beb14e45 Mon Sep 17 00:00:00 2001 From: Alexander Graf Date: Tue, 7 Jan 2014 17:19:14 +0000 Subject: [PATCH 58/61] target-arm: A64: Add floating-point<->fixed-point instructions This patch adds emulation for the instruction group labeled "Floating-point <-> fixed-point conversions" in the ARM ARM. Namely this includes the instructions SCVTF, UCVTF, FCVTZS, FCVTZU (scalar, fixed-point). Signed-off-by: Alexander Graf [WN: Commit message tweak, rebased, updated to new infrastructure. Applied bug fixes from Michael Matz and Janne Grunau.] Signed-off-by: Will Newton [PMM: significant cleanup] Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- target-arm/helper.c | 13 +++ target-arm/helper.h | 2 + target-arm/translate-a64.c | 186 ++++++++++++++++++++++++++++++++++++- 3 files changed, 200 insertions(+), 1 deletion(-) diff --git a/target-arm/helper.c b/target-arm/helper.c index dcf6aa4712..2e76d0b093 100644 --- a/target-arm/helper.c +++ b/target-arm/helper.c @@ -4035,6 +4035,19 @@ VFP_CONV_FIX_A64(uq, s, 32, 64, uint64) #undef VFP_CONV_FIX_FLOAT #undef VFP_CONV_FLOAT_FIX_ROUND +/* Set the current fp rounding mode and return the old one. + * The argument is a softfloat float_round_ value. + */ +uint32_t HELPER(set_rmode)(uint32_t rmode, CPUARMState *env) +{ + float_status *fp_status = &env->vfp.fp_status; + + uint32_t prev_rmode = get_float_rounding_mode(fp_status); + set_float_rounding_mode(rmode, fp_status); + + return prev_rmode; +} + /* Half precision conversions. */ static float32 do_fcvt_f16_to_f32(uint32_t a, CPUARMState *env, float_status *s) { diff --git a/target-arm/helper.h b/target-arm/helper.h index 2e1af4671c..25b6b4f493 100644 --- a/target-arm/helper.h +++ b/target-arm/helper.h @@ -148,6 +148,8 @@ DEF_HELPER_3(vfp_uhtod, f64, i64, i32, ptr) DEF_HELPER_3(vfp_ultod, f64, i64, i32, ptr) DEF_HELPER_3(vfp_uqtod, f64, i64, i32, ptr) +DEF_HELPER_FLAGS_2(set_rmode, TCG_CALL_NO_RWG, i32, i32, env) + DEF_HELPER_2(vfp_fcvt_f16_to_f32, f32, i32, env) DEF_HELPER_2(vfp_fcvt_f32_to_f16, i32, f32, env) DEF_HELPER_2(neon_fcvt_f16_to_f32, f32, i32, env) diff --git a/target-arm/translate-a64.c b/target-arm/translate-a64.c index c9fbf0f0f7..ec8abc708a 100644 --- a/target-arm/translate-a64.c +++ b/target-arm/translate-a64.c @@ -3186,6 +3186,34 @@ static void disas_data_proc_reg(DisasContext *s, uint32_t insn) } } +/* Convert ARM rounding mode to softfloat */ +static inline int arm_rmode_to_sf(int rmode) +{ + switch (rmode) { + case FPROUNDING_TIEAWAY: + rmode = float_round_ties_away; + break; + case FPROUNDING_ODD: + /* FIXME: add support for TIEAWAY and ODD */ + qemu_log_mask(LOG_UNIMP, "arm: unimplemented rounding mode: %d\n", + rmode); + case FPROUNDING_TIEEVEN: + default: + rmode = float_round_nearest_even; + break; + case FPROUNDING_POSINF: + rmode = float_round_up; + break; + case FPROUNDING_NEGINF: + rmode = float_round_down; + break; + case FPROUNDING_ZERO: + rmode = float_round_to_zero; + break; + } + return rmode; +} + static void handle_fp_compare(DisasContext *s, bool is_double, unsigned int rn, unsigned int rm, bool cmp_with_zero, bool signal_all_nans) @@ -3651,6 +3679,132 @@ static void disas_fp_imm(DisasContext *s, uint32_t insn) tcg_temp_free_i64(tcg_res); } +/* Handle floating point <=> fixed point conversions. Note that we can + * also deal with fp <=> integer conversions as a special case (scale == 64) + * OPTME: consider handling that special case specially or at least skipping + * the call to scalbn in the helpers for zero shifts. + */ +static void handle_fpfpcvt(DisasContext *s, int rd, int rn, int opcode, + bool itof, int rmode, int scale, int sf, int type) +{ + bool is_signed = !(opcode & 1); + bool is_double = type; + TCGv_ptr tcg_fpstatus; + TCGv_i32 tcg_shift; + + tcg_fpstatus = get_fpstatus_ptr(); + + tcg_shift = tcg_const_i32(64 - scale); + + if (itof) { + TCGv_i64 tcg_int = cpu_reg(s, rn); + if (!sf) { + TCGv_i64 tcg_extend = new_tmp_a64(s); + + if (is_signed) { + tcg_gen_ext32s_i64(tcg_extend, tcg_int); + } else { + tcg_gen_ext32u_i64(tcg_extend, tcg_int); + } + + tcg_int = tcg_extend; + } + + if (is_double) { + TCGv_i64 tcg_double = tcg_temp_new_i64(); + if (is_signed) { + gen_helper_vfp_sqtod(tcg_double, tcg_int, + tcg_shift, tcg_fpstatus); + } else { + gen_helper_vfp_uqtod(tcg_double, tcg_int, + tcg_shift, tcg_fpstatus); + } + write_fp_dreg(s, rd, tcg_double); + tcg_temp_free_i64(tcg_double); + } else { + TCGv_i32 tcg_single = tcg_temp_new_i32(); + if (is_signed) { + gen_helper_vfp_sqtos(tcg_single, tcg_int, + tcg_shift, tcg_fpstatus); + } else { + gen_helper_vfp_uqtos(tcg_single, tcg_int, + tcg_shift, tcg_fpstatus); + } + write_fp_sreg(s, rd, tcg_single); + tcg_temp_free_i32(tcg_single); + } + } else { + TCGv_i64 tcg_int = cpu_reg(s, rd); + TCGv_i32 tcg_rmode; + + if (extract32(opcode, 2, 1)) { + /* There are too many rounding modes to all fit into rmode, + * so FCVTA[US] is a special case. + */ + rmode = FPROUNDING_TIEAWAY; + } + + tcg_rmode = tcg_const_i32(arm_rmode_to_sf(rmode)); + + gen_helper_set_rmode(tcg_rmode, tcg_rmode, cpu_env); + + if (is_double) { + TCGv_i64 tcg_double = read_fp_dreg(s, rn); + if (is_signed) { + if (!sf) { + gen_helper_vfp_tosld(tcg_int, tcg_double, + tcg_shift, tcg_fpstatus); + } else { + gen_helper_vfp_tosqd(tcg_int, tcg_double, + tcg_shift, tcg_fpstatus); + } + } else { + if (!sf) { + gen_helper_vfp_tould(tcg_int, tcg_double, + tcg_shift, tcg_fpstatus); + } else { + gen_helper_vfp_touqd(tcg_int, tcg_double, + tcg_shift, tcg_fpstatus); + } + } + tcg_temp_free_i64(tcg_double); + } else { + TCGv_i32 tcg_single = read_fp_sreg(s, rn); + if (sf) { + if (is_signed) { + gen_helper_vfp_tosqs(tcg_int, tcg_single, + tcg_shift, tcg_fpstatus); + } else { + gen_helper_vfp_touqs(tcg_int, tcg_single, + tcg_shift, tcg_fpstatus); + } + } else { + TCGv_i32 tcg_dest = tcg_temp_new_i32(); + if (is_signed) { + gen_helper_vfp_tosls(tcg_dest, tcg_single, + tcg_shift, tcg_fpstatus); + } else { + gen_helper_vfp_touls(tcg_dest, tcg_single, + tcg_shift, tcg_fpstatus); + } + tcg_gen_extu_i32_i64(tcg_int, tcg_dest); + tcg_temp_free_i32(tcg_dest); + } + tcg_temp_free_i32(tcg_single); + } + + gen_helper_set_rmode(tcg_rmode, tcg_rmode, cpu_env); + tcg_temp_free_i32(tcg_rmode); + + if (!sf) { + tcg_gen_ext32u_i64(tcg_int, tcg_int); + } + } + + tcg_temp_free_ptr(tcg_fpstatus); + tcg_temp_free_i32(tcg_shift); +} + /* C3.6.29 Floating point <-> fixed point conversions * 31 30 29 28 24 23 22 21 20 19 18 16 15 10 9 5 4 0 * +----+---+---+-----------+------+---+-------+--------+-------+------+------+ @@ -3659,7 +3813,37 @@ static void disas_fp_imm(DisasContext *s, uint32_t insn) */ static void disas_fp_fixed_conv(DisasContext *s, uint32_t insn) { - unsupported_encoding(s, insn); + int rd = extract32(insn, 0, 5); + int rn = extract32(insn, 5, 5); + int scale = extract32(insn, 10, 6); + int opcode = extract32(insn, 16, 3); + int rmode = extract32(insn, 19, 2); + int type = extract32(insn, 22, 2); + bool sbit = extract32(insn, 29, 1); + bool sf = extract32(insn, 31, 1); + bool itof; + + if (sbit || (type > 1) + || (!sf && scale < 32)) { + unallocated_encoding(s); + return; + } + + switch ((rmode << 3) | opcode) { + case 0x2: /* SCVTF */ + case 0x3: /* UCVTF */ + itof = true; + break; + case 0x18: /* FCVTZS */ + case 0x19: /* FCVTZU */ + itof = false; + break; + default: + unallocated_encoding(s); + return; + } + + handle_fpfpcvt(s, rd, rn, opcode, itof, FPROUNDING_ZERO, scale, sf, type); } static void handle_fmov(DisasContext *s, int rd, int rn, int type, bool itof) From 6a7413385d731a97cbd45c7c4b1c2cbd73d94409 Mon Sep 17 00:00:00 2001 From: Will Newton Date: Tue, 7 Jan 2014 17:19:14 +0000 Subject: [PATCH 59/61] target-arm: A64: Add floating-point<->integer conversion instructions Add support for the AArch64 floating-point <-> integer conversion instructions to disas_fpintconv. In the process we can rearrange and simplify the detection of unallocated encodings a little. We also correct a typo in the instruction encoding diagram for this instruction group: bit 21 is 1, not 0. Signed-off-by: Will Newton Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- target-arm/translate-a64.c | 23 ++++++++++++++++++++--- 1 file changed, 20 insertions(+), 3 deletions(-) diff --git a/target-arm/translate-a64.c b/target-arm/translate-a64.c index ec8abc708a..9b23d37483 100644 --- a/target-arm/translate-a64.c +++ b/target-arm/translate-a64.c @@ -3904,7 +3904,7 @@ static void handle_fmov(DisasContext *s, int rd, int rn, int type, bool itof) /* C3.6.30 Floating point <-> integer conversions * 31 30 29 28 24 23 22 21 20 19 18 16 15 10 9 5 4 0 * +----+---+---+-----------+------+---+-------+-----+-------------+----+----+ - * | sf | 0 | S | 1 1 1 1 0 | type | 0 | rmode | opc | 0 0 0 0 0 0 | Rn | Rd | + * | sf | 0 | S | 1 1 1 1 0 | type | 1 | rmode | opc | 0 0 0 0 0 0 | Rn | Rd | * +----+---+---+-----------+------+---+-------+-----+-------------+----+----+ */ static void disas_fp_int_conv(DisasContext *s, uint32_t insn) @@ -3917,10 +3917,20 @@ static void disas_fp_int_conv(DisasContext *s, uint32_t insn) bool sbit = extract32(insn, 29, 1); bool sf = extract32(insn, 31, 1); - if (!sbit && (rmode < 2) && (opcode > 5)) { + if (sbit) { + unallocated_encoding(s); + return; + } + + if (opcode > 5) { /* FMOV */ bool itof = opcode & 1; + if (rmode >= 2) { + unallocated_encoding(s); + return; + } + switch (sf << 3 | type << 1 | rmode) { case 0x0: /* 32 bit */ case 0xa: /* 64 bit */ @@ -3935,7 +3945,14 @@ static void disas_fp_int_conv(DisasContext *s, uint32_t insn) handle_fmov(s, rd, rn, type, itof); } else { /* actual FP conversions */ - unsupported_encoding(s, insn); + bool itof = extract32(opcode, 1, 1); + + if (type > 1 || (rmode != 0 && opcode > 1)) { + unallocated_encoding(s); + return; + } + + handle_fpfpcvt(s, rd, rn, opcode, itof, rmode, 64, sf, type); } } From e97db91cbf7df456a35996d7243cd0f55d23ca1a Mon Sep 17 00:00:00 2001 From: Peter Maydell Date: Tue, 7 Jan 2014 17:19:14 +0000 Subject: [PATCH 60/61] target-arm: A64: Add 1-source 32-to-32 and 64-to-64 FP instructions This patch adds support for those instructions in the "Floating-point data-processing (1 source)" group which are simple 32-bit-to-32-bit or 64-bit-to-64-bit operations (ie everything except FCVT between single/double/half precision). We put the new round-to-int helpers in helper.c because they will also be used by the new ARMv8 A32/T32 rounding instructions. Signed-off-by: Alexander Graf [WN: Commit message tweak, merged single and double precision patches, updated to new infrastructure.] Signed-off-by: Will Newton [PMM: reworked decode, split FCVT out into their own patch] Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- target-arm/helper.c | 45 ++++++++++++ target-arm/helper.h | 5 ++ target-arm/translate-a64.c | 142 ++++++++++++++++++++++++++++++++++++- 3 files changed, 191 insertions(+), 1 deletion(-) diff --git a/target-arm/helper.c b/target-arm/helper.c index 2e76d0b093..c6a62818e4 100644 --- a/target-arm/helper.c +++ b/target-arm/helper.c @@ -4353,3 +4353,48 @@ float64 VFP_HELPER(muladd, d)(float64 a, float64 b, float64 c, void *fpstp) float_status *fpst = fpstp; return float64_muladd(a, b, c, 0, fpst); } + +/* ARMv8 round to integral */ +float32 HELPER(rints_exact)(float32 x, void *fp_status) +{ + return float32_round_to_int(x, fp_status); +} + +float64 HELPER(rintd_exact)(float64 x, void *fp_status) +{ + return float64_round_to_int(x, fp_status); +} + +float32 HELPER(rints)(float32 x, void *fp_status) +{ + int old_flags = get_float_exception_flags(fp_status), new_flags; + float32 ret; + + ret = float32_round_to_int(x, fp_status); + + /* Suppress any inexact exceptions the conversion produced */ + if (!(old_flags & float_flag_inexact)) { + new_flags = get_float_exception_flags(fp_status); + set_float_exception_flags(new_flags & ~float_flag_inexact, fp_status); + } + + return ret; +} + +float64 HELPER(rintd)(float64 x, void *fp_status) +{ + int old_flags = get_float_exception_flags(fp_status), new_flags; + float64 ret; + + ret = float64_round_to_int(x, fp_status); + + new_flags = get_float_exception_flags(fp_status); + + /* Suppress any inexact exceptions the conversion produced */ + if (!(old_flags & float_flag_inexact)) { + new_flags = get_float_exception_flags(fp_status); + set_float_exception_flags(new_flags & ~float_flag_inexact, fp_status); + } + + return ret; +} diff --git a/target-arm/helper.h b/target-arm/helper.h index 25b6b4f493..7532e29769 100644 --- a/target-arm/helper.h +++ b/target-arm/helper.h @@ -171,6 +171,11 @@ DEF_HELPER_3(shr_cc, i32, env, i32, i32) DEF_HELPER_3(sar_cc, i32, env, i32, i32) DEF_HELPER_3(ror_cc, i32, env, i32, i32) +DEF_HELPER_FLAGS_2(rints_exact, TCG_CALL_NO_RWG, f32, f32, ptr) +DEF_HELPER_FLAGS_2(rintd_exact, TCG_CALL_NO_RWG, f64, f64, ptr) +DEF_HELPER_FLAGS_2(rints, TCG_CALL_NO_RWG, f32, f32, ptr) +DEF_HELPER_FLAGS_2(rintd, TCG_CALL_NO_RWG, f64, f64, ptr) + /* neon_helper.c */ DEF_HELPER_3(neon_qadd_u8, i32, env, i32, i32) DEF_HELPER_3(neon_qadd_s8, i32, env, i32, i32) diff --git a/target-arm/translate-a64.c b/target-arm/translate-a64.c index 9b23d37483..345a47b004 100644 --- a/target-arm/translate-a64.c +++ b/target-arm/translate-a64.c @@ -3386,6 +3386,118 @@ static void disas_fp_csel(DisasContext *s, uint32_t insn) } } +/* C3.6.25 Floating-point data-processing (1 source) - single precision */ +static void handle_fp_1src_single(DisasContext *s, int opcode, int rd, int rn) +{ + TCGv_ptr fpst; + TCGv_i32 tcg_op; + TCGv_i32 tcg_res; + + fpst = get_fpstatus_ptr(); + tcg_op = read_fp_sreg(s, rn); + tcg_res = tcg_temp_new_i32(); + + switch (opcode) { + case 0x0: /* FMOV */ + tcg_gen_mov_i32(tcg_res, tcg_op); + break; + case 0x1: /* FABS */ + gen_helper_vfp_abss(tcg_res, tcg_op); + break; + case 0x2: /* FNEG */ + gen_helper_vfp_negs(tcg_res, tcg_op); + break; + case 0x3: /* FSQRT */ + gen_helper_vfp_sqrts(tcg_res, tcg_op, cpu_env); + break; + case 0x8: /* FRINTN */ + case 0x9: /* FRINTP */ + case 0xa: /* FRINTM */ + case 0xb: /* FRINTZ */ + case 0xc: /* FRINTA */ + { + TCGv_i32 tcg_rmode = tcg_const_i32(arm_rmode_to_sf(opcode & 7)); + + gen_helper_set_rmode(tcg_rmode, tcg_rmode, cpu_env); + gen_helper_rints(tcg_res, tcg_op, fpst); + + gen_helper_set_rmode(tcg_rmode, tcg_rmode, cpu_env); + tcg_temp_free_i32(tcg_rmode); + break; + } + case 0xe: /* FRINTX */ + gen_helper_rints_exact(tcg_res, tcg_op, fpst); + break; + case 0xf: /* FRINTI */ + gen_helper_rints(tcg_res, tcg_op, fpst); + break; + default: + abort(); + } + + write_fp_sreg(s, rd, tcg_res); + + tcg_temp_free_ptr(fpst); + tcg_temp_free_i32(tcg_op); + tcg_temp_free_i32(tcg_res); +} + +/* C3.6.25 Floating-point data-processing (1 source) - double precision */ +static void handle_fp_1src_double(DisasContext *s, int opcode, int rd, int rn) +{ + TCGv_ptr fpst; + TCGv_i64 tcg_op; + TCGv_i64 tcg_res; + + fpst = get_fpstatus_ptr(); + tcg_op = read_fp_dreg(s, rn); + tcg_res = tcg_temp_new_i64(); + + switch (opcode) { + case 0x0: /* FMOV */ + tcg_gen_mov_i64(tcg_res, tcg_op); + break; + case 0x1: /* FABS */ + gen_helper_vfp_absd(tcg_res, tcg_op); + break; + case 0x2: /* FNEG */ + gen_helper_vfp_negd(tcg_res, tcg_op); + break; + case 0x3: /* FSQRT */ + gen_helper_vfp_sqrtd(tcg_res, tcg_op, cpu_env); + break; + case 0x8: /* FRINTN */ + case 0x9: /* FRINTP */ + case 0xa: /* FRINTM */ + case 0xb: /* FRINTZ */ + case 0xc: /* FRINTA */ + { + TCGv_i32 tcg_rmode = tcg_const_i32(arm_rmode_to_sf(opcode & 7)); + + gen_helper_set_rmode(tcg_rmode, tcg_rmode, cpu_env); + gen_helper_rintd(tcg_res, tcg_op, fpst); + + gen_helper_set_rmode(tcg_rmode, tcg_rmode, cpu_env); + tcg_temp_free_i32(tcg_rmode); + break; + } + case 0xe: /* FRINTX */ + gen_helper_rintd_exact(tcg_res, tcg_op, fpst); + break; + case 0xf: /* FRINTI */ + gen_helper_rintd(tcg_res, tcg_op, fpst); + break; + default: + abort(); + } + + write_fp_dreg(s, rd, tcg_res); + + tcg_temp_free_ptr(fpst); + tcg_temp_free_i64(tcg_op); + tcg_temp_free_i64(tcg_res); +} + /* C3.6.25 Floating point data-processing (1 source) * 31 30 29 28 24 23 22 21 20 15 14 10 9 5 4 0 * +---+---+---+-----------+------+---+--------+-----------+------+------+ @@ -3394,7 +3506,35 @@ static void disas_fp_csel(DisasContext *s, uint32_t insn) */ static void disas_fp_1src(DisasContext *s, uint32_t insn) { - unsupported_encoding(s, insn); + int type = extract32(insn, 22, 2); + int opcode = extract32(insn, 15, 6); + int rn = extract32(insn, 5, 5); + int rd = extract32(insn, 0, 5); + + switch (opcode) { + case 0x4: case 0x5: case 0x7: + /* FCVT between half, single and double precision */ + unsupported_encoding(s, insn); + break; + case 0x0 ... 0x3: + case 0x8 ... 0xc: + case 0xe ... 0xf: + /* 32-to-32 and 64-to-64 ops */ + switch (type) { + case 0: + handle_fp_1src_single(s, opcode, rd, rn); + break; + case 1: + handle_fp_1src_double(s, opcode, rd, rn); + break; + default: + unallocated_encoding(s); + } + break; + default: + unallocated_encoding(s); + break; + } } /* C3.6.26 Floating-point data-processing (2 source) - single precision */ From 096fc768184d5e9c47ccc4dd95d9f983fa83b9e1 Mon Sep 17 00:00:00 2001 From: Peter Maydell Date: Tue, 7 Jan 2014 17:19:15 +0000 Subject: [PATCH 61/61] target-arm: A64: Add support for FCVT between half, single and double Add support for FCVT between half, single and double precision. Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- target-arm/helper.c | 20 ++++++++++ target-arm/helper.h | 2 + target-arm/translate-a64.c | 75 +++++++++++++++++++++++++++++++++++++- 3 files changed, 96 insertions(+), 1 deletion(-) diff --git a/target-arm/helper.c b/target-arm/helper.c index c6a62818e4..c708f15e27 100644 --- a/target-arm/helper.c +++ b/target-arm/helper.c @@ -4089,6 +4089,26 @@ uint32_t HELPER(vfp_fcvt_f32_to_f16)(float32 a, CPUARMState *env) return do_fcvt_f32_to_f16(a, env, &env->vfp.fp_status); } +float64 HELPER(vfp_fcvt_f16_to_f64)(uint32_t a, CPUARMState *env) +{ + int ieee = (env->vfp.xregs[ARM_VFP_FPSCR] & (1 << 26)) == 0; + float64 r = float16_to_float64(make_float16(a), ieee, &env->vfp.fp_status); + if (ieee) { + return float64_maybe_silence_nan(r); + } + return r; +} + +uint32_t HELPER(vfp_fcvt_f64_to_f16)(float64 a, CPUARMState *env) +{ + int ieee = (env->vfp.xregs[ARM_VFP_FPSCR] & (1 << 26)) == 0; + float16 r = float64_to_float16(a, ieee, &env->vfp.fp_status); + if (ieee) { + r = float16_maybe_silence_nan(r); + } + return float16_val(r); +} + #define float32_two make_float32(0x40000000) #define float32_three make_float32(0x40400000) #define float32_one_point_five make_float32(0x3fc00000) diff --git a/target-arm/helper.h b/target-arm/helper.h index 7532e29769..70872dffc6 100644 --- a/target-arm/helper.h +++ b/target-arm/helper.h @@ -154,6 +154,8 @@ DEF_HELPER_2(vfp_fcvt_f16_to_f32, f32, i32, env) DEF_HELPER_2(vfp_fcvt_f32_to_f16, i32, f32, env) DEF_HELPER_2(neon_fcvt_f16_to_f32, f32, i32, env) DEF_HELPER_2(neon_fcvt_f32_to_f16, i32, f32, env) +DEF_HELPER_FLAGS_2(vfp_fcvt_f16_to_f64, TCG_CALL_NO_RWG, f64, i32, env) +DEF_HELPER_FLAGS_2(vfp_fcvt_f64_to_f16, TCG_CALL_NO_RWG, i32, f64, env) DEF_HELPER_4(vfp_muladdd, f64, f64, f64, f64, ptr) DEF_HELPER_4(vfp_muladds, f32, f32, f32, f32, ptr) diff --git a/target-arm/translate-a64.c b/target-arm/translate-a64.c index 345a47b004..cf80c46b90 100644 --- a/target-arm/translate-a64.c +++ b/target-arm/translate-a64.c @@ -3498,6 +3498,72 @@ static void handle_fp_1src_double(DisasContext *s, int opcode, int rd, int rn) tcg_temp_free_i64(tcg_res); } +static void handle_fp_fcvt(DisasContext *s, int opcode, + int rd, int rn, int dtype, int ntype) +{ + switch (ntype) { + case 0x0: + { + TCGv_i32 tcg_rn = read_fp_sreg(s, rn); + if (dtype == 1) { + /* Single to double */ + TCGv_i64 tcg_rd = tcg_temp_new_i64(); + gen_helper_vfp_fcvtds(tcg_rd, tcg_rn, cpu_env); + write_fp_dreg(s, rd, tcg_rd); + tcg_temp_free_i64(tcg_rd); + } else { + /* Single to half */ + TCGv_i32 tcg_rd = tcg_temp_new_i32(); + gen_helper_vfp_fcvt_f32_to_f16(tcg_rd, tcg_rn, cpu_env); + /* write_fp_sreg is OK here because top half of tcg_rd is zero */ + write_fp_sreg(s, rd, tcg_rd); + tcg_temp_free_i32(tcg_rd); + } + tcg_temp_free_i32(tcg_rn); + break; + } + case 0x1: + { + TCGv_i64 tcg_rn = read_fp_dreg(s, rn); + TCGv_i32 tcg_rd = tcg_temp_new_i32(); + if (dtype == 0) { + /* Double to single */ + gen_helper_vfp_fcvtsd(tcg_rd, tcg_rn, cpu_env); + } else { + /* Double to half */ + gen_helper_vfp_fcvt_f64_to_f16(tcg_rd, tcg_rn, cpu_env); + /* write_fp_sreg is OK here because top half of tcg_rd is zero */ + } + write_fp_sreg(s, rd, tcg_rd); + tcg_temp_free_i32(tcg_rd); + tcg_temp_free_i64(tcg_rn); + break; + } + case 0x3: + { + TCGv_i32 tcg_rn = read_fp_sreg(s, rn); + tcg_gen_ext16u_i32(tcg_rn, tcg_rn); + if (dtype == 0) { + /* Half to single */ + TCGv_i32 tcg_rd = tcg_temp_new_i32(); + gen_helper_vfp_fcvt_f16_to_f32(tcg_rd, tcg_rn, cpu_env); + write_fp_sreg(s, rd, tcg_rd); + tcg_temp_free_i32(tcg_rd); + } else { + /* Half to double */ + TCGv_i64 tcg_rd = tcg_temp_new_i64(); + gen_helper_vfp_fcvt_f16_to_f64(tcg_rd, tcg_rn, cpu_env); + write_fp_dreg(s, rd, tcg_rd); + tcg_temp_free_i64(tcg_rd); + } + tcg_temp_free_i32(tcg_rn); + break; + } + default: + abort(); + } +} + /* C3.6.25 Floating point data-processing (1 source) * 31 30 29 28 24 23 22 21 20 15 14 10 9 5 4 0 * +---+---+---+-----------+------+---+--------+-----------+------+------+ @@ -3513,9 +3579,16 @@ static void disas_fp_1src(DisasContext *s, uint32_t insn) switch (opcode) { case 0x4: case 0x5: case 0x7: + { /* FCVT between half, single and double precision */ - unsupported_encoding(s, insn); + int dtype = extract32(opcode, 0, 2); + if (type == 2 || dtype == type) { + unallocated_encoding(s); + return; + } + handle_fp_fcvt(s, opcode, rd, rn, dtype, type); break; + } case 0x0 ... 0x3: case 0x8 ... 0xc: case 0xe ... 0xf: